Journal of the American College of Surgeons Volume 200, Issue 5, Pages 641-816 (May 2005) Centennial Perspectives

1. Adhesions • MISCELLANEOUS Page 641

2. Intra-abdominal adhesions • MISCELLANEOUS Page 642

3. Intraabdominal and Postoperative Peritoneal Adhesions • DISCUSSION Pages 643-644 Harold Ellis

4. Etiology/treatment varicose • MISCELLANEOUS Page 645

5. Etiology and Treatment of Varicose Ulcer of the Leg • DISCUSSION Pages 646-647 Robert L. Kistner

Scudder Oration on Trauma

6. Changes in the Management of Injuries to the Liver and Spleen • ARTICLE Pages 648-669 J. David Richardson

Original Scientific Articles

7. Impact of Work-Hour Restrictions on Residents’ Operative Volume on a Subspecialty Surgical Service • ARTICLE Pages 670-676 Ariel U. Spencer and Daniel H. Teitelbaum

8. Does Surgeon Frustration and Satisfaction with the Operation Predict Outcomes of Open or Laparoscopic Inguinal Hernia Repair? • ARTICLE Pages 677-683 Haytham M.A. Kaafarani, Kamal M.F. Itani, Anita Giobbie-Hurder, John J. Gleysteen, Martin McCarthy, Jr, James Gibbs and Leigh Neumayer

9. Perceptions and Predictors of Surgeon Satisfaction: A Survey of Spouses of Academic Surgeons • ARTICLE Pages 684-690 Lillian S. Kao, Erik B. Wilson and Kimberly D. Anderson

10. Characterization of Human Nasal Septal Chondrocytes Cultured in Alginate • ARTICLE Pages 691-704 Stanley H. Chia, Mark R. Homicz, Barbara L. Schumacher, Eugene J.-M.A. Thonar, Koichi Masuda, Robert L. Sah and Deborah Watson

11. GATA-3 Expression as a Predictor of Hormone Response in Breast Cancer • ARTICLE Pages 705-710 Purvi Parikh, Juan P. Palazzo, Lewis J. Rose, Constantine Daskalakis and Ronald J. Weigel

12. Fusion of the Planes of the Liver: An Anatomic Entity Merging the Midplane and the Left Intersectional Plane • ARTICLE Pages 711-719 Eric Savier, Jacques Taboury, Olivier Lucidarme, Kumiko Kitajima, Mehdi Cadi, Jean-Christophe Vaillant and Laurent Hannoun

13. Do Bone Marrow Micrometastases Correlate with Sentinel Lymph Node Metastases in Breast Cancer Patients? • ARTICLE Pages 720-725 Susan M. Trocciola, Syed Hoda, Michael P. Osborne, Paul J. Christos, Heather Levin, Diana Martins, Joshua Carson, John Daly and Rache M. Simmons

14. Invited Commentary • DISCUSSION Pages 725-726 V. Suzanne Klimberg

15. Reply • CORRESPONDENCE Page 726 Susan M. Trocciola and Rache M. Simmons

16. Cystadenoma and Cystadenocarcinoma of the Liver: A Single Center Experience • REVIEW ARTICLE Pages 727-733 David P. Vogt, J. Michael Henderson and Elaine Chmielewski

17. Safety of Carotid Endarterectomy in 2,443 Elderly Patients: Lessons from Nonagenarians—Are We Pushing the Limit? • ARTICLE Pages 734-741 Desarom Teso, Randolph E. Edwards, Jared C. Frattini, Stanley J. Dudrick and Alan Dardik

Consensus Statement

18. Use of Magnetic Resonance Imaging in Breast Oncology • SHORT COMMUNICATION Page 742 Alan Dardik What's New in Surgery

19. What’s New in Cardiac Surgery • ARTICLE Pages 743-754 David A. Fullerton

20. What’s New in General Surgery: Critical Care and Trauma • ARTICLE Pages 755-765 Lena M. Napolitano

Symposium

21. Introduction: Symposium on Surgery in the Elderly Patient, Part 2 • SHORT COMMUNICATION Page 766 James C. Thompson

22. Postoperative Delirium in the Older Patient • ARTICLE Pages 767-773 Luis F. Amador and James S. Goodwin

23. Common Orthopaedic Problems in the Elderly Patient • ARTICLE Pages 774-783 Michael J. Grecula and Mabel E. Caban

24. Anesthetic Pitfalls in the Elderly Patient • ARTICLE Pages 784-794 Donald S. Prough

Collective Review

25. Schistosomal Portal Hyptertension • REVIEW ARTICLE Pages 795-806 Adeyemi O. Laosebikan, Sandie R. Thomson and Namasha M. Naidoo

Images for Surgeons

26. Fibromuscular Dysplasia of the Carotid Artery • SHORT COMMUNICATION Page 807 Wei Zhou, Ruth L. Bush, Peter L. Lin and Alan B. Lumsden

27. Solitary Arteriovenous Malformation of the Small Intestine • SHORT COMMUNICATION Pages 808-809 Adnan Z. Rizvi, John A. Kaufman, Pamela Smith and Mark L. Silen

Surgeon at Work

28. Percutaneous Cephalic Vein Cannulation (in the Deltopectoral Groove), with Ultrasound Guidance • ARTICLE Pages 810-811 Jack LeDonne Letters

29. Complete Esophageal Diversion: A Simplified, Easily Reversible Technique • CORRESPONDENCE Page 812 Paul D. Kiernan, John Rhee, Lucas Collazo, Vivian Hetrick, Betty Vaughan and Paula Graling

30. Reply • CORRESPONDENCE Pages 812-813 Kevin F. Staveley-O’Carroll, Seth A. Spector and Leonidas G. Koniaris

31. Voice Changes after Thyroidectomy Without Recurrent Laryngeal Nerve Injury • CORRESPONDENCE Page 813 Allen D. Hillel

32. Centennial Perspective on Burn Treatment • CORRESPONDENCE Page 814 Martin Allgöwer

33. Reply • CORRESPONDENCE Page 815 John F. Burke

34. The Role of Claude Bernard and Others in the Discovery of Horner’s Syndrome • CORRESPONDENCE Page 815 August L. Reader III

35. Reply • CORRESPONDENCE Page 815 Ian Ross

Evidence-Based Surgery

36. gfedc Evidence-based surgery • MISCELLANEOUS Pages A41-A44

Continuing Medical Education Program

37. JACS CME-1 featured articles, volume 200, May 2005 • DISCUSSION Pages A45-A47

Vol. 200, No. 5, May 2005 Centennial Perspectives 641 642 Centennial Perspectives J Am Coll Surg CENTENNIAL PERSPECTIVES

Intraabdominal and Postoperative Peritoneal Adhesions

Harold Ellis, CBE, FACS (Hon), FRCS London, England

Until the introduction of anesthesia and then antiseptic and patients usually do not report their groin hernias surgery allowed laparotomy to become a comparatively until they strangulate, the situation is reversed; strangu- common and comparatively safe procedure in the lated hernias are common, and adhesive obstruction is 1880s, intraabdominal adhesions were an uncommon rare. phenomenon and of little, if any, interest to surgeons. Since the beginning of the 20th century, truly enor- You will not find mention of them in the standard sur- mous attempts have been made to prevent the formation gical textbooks up to that time. of postoperative adhesions. Today, there are thousands In contrast, adhesions from inflammatory disease of references on both the clinical and laboratory aspects within the peritoneal cavity were well recognized at au- of this problem, which now represents a considerable topsy. For example, Thomas Hodgkin,1 in 1836, com- burden on our health services.5 mented on the matted bowel found at autopsy in pa- Attempts to prevent postoperative adhesions can be tients with tuberculous peritonitis and also on the classified into:4 tendency of adhesions to occur in the lower abdomen in A. Prevention of fibrin deposition, using citrate, hepa- patients dying of pelvic sepsis. rin (both topically and systemically), and dicumoral. With the advent of the era of abdominal surgery, Deaths from hemorrhage were reported in animals deaths from postoperative adhesive obstruction began to under laboratory conditions, and, more alarmingly, be reported. The first I have come across was by Thomas there were examples of bleeding and even deaths in 2 Bryant, of Guy’s Hospital, London, in 1872; he de- patients given intraperitoneal heparin. scribed a fatal case of small bowel obstruction from a B. Removal of fibrin exudates between damaged sur- band formed after removal of an ovarian cyst. The first faces. Attempts have been made to wash away or account of a laparotomy for adhesive obstruction I have dilute the fibrin using saline, hypertonic dextrose, traced was reported in the Lancet in 1883 by William and other solutions, or to digest or remove it with Battle,3 then a surgical registrar at St Thomas’s Hospital, pepsin, trypsin, streptokinase, and streptodornase. London. The patient, a 43-year-old woman, had bilat- Tissue plasminogen activator has been shown to be eral ovarian tumors removed 4 years earlier. highly effective in a rabbit model, but there have She was admitted with intestinal obstruction, was been no clinical reports of its use to date.6 found at laparotomy to have matted adhesions of termi- C. Separation of surfaces. A wide range of materials nal ileum to the region of the cecum, and had a terminal has been used to separate surgically traumatized vis- ileostomy performed. Sadly, she died 3 weeks later. cera. Each, in the past, had its advocates, proceeding Reading the careful notes, it appears likely that with even to clinical use, but most have later been shown to be noneffective, or even to increase the problem. modern intravenous nutrition, nasogastric suction, and Materials used in the past included saline, Ringer’s antibiotic administration, she might well have solution, dextran, gelatine, olive oil, paraffin, sili- recovered. cones, plasma, lanoline, polyvinyl pyrrolidine, and Up to the 1930s, strangulated hernias accounted for an amazing variety of membranes—amnion, fish the majority of small bowel obstructions. In more mod- bladder, carp peritoneum, calf peritoneum, oiled ern times, as elective repair of hernias becomes standard silk, silver or gold foil, and free grafts of omentum. treatment, and abdominal surgery so common, adhesive Of these barriers, only a membrane composed of obstruction accounts for about three-quarters of all cases hyaluronic acid and carboxymethylcellulose was of small bowel occlusion.4 Interestingly, in the Third shown to reduce adhesion formation in a clinical World, where abdominal surgery is fairly uncommon prospective randomized trial.7 More recently, a solu-

tion of icodextrin has reduced adhesion formation Although no figures were given, the authors claimed “the after laparoscopic gynecologic operations in a pilot results on the whole have been satisfactory.” study.8 In these days of statistical appraisal, prospective ran- D. Inhibition of fibroblastic proliferation. Attempts to domized double-blinded trials, anonymous reviewers, prevent the conversion of fibrinous adhesions into and so on, this is hardly the conclusion that would be established fibrous tissue have included studies of allowed by the editors of today’s Journal of the American antihistamines and steroids given topically and sys- College of Surgeons. temically. The conflict between the need for the surgical wound to heal and the prevention of unwanted fibrous adhesions was shown by studies using flu- REFERENCES orouracil in rats. With a high dosage of this drug, complete inhibition of adhesions to the crushed ce- 1. Hodgkin T. Lectures in the morbid anatomy of the serous and mucous membranes, vol 1. London: Simkin Marshall and Co; cum could be demonstrated. The abdominal wound 1836. failed to heal, and gangrene occurred at the site of the 2. Bryant T. Clinical lectures on intestinal obstructions. Med Times cecal crush. Gazette 1872;1:363–365. 3. Battle WH. Intestinal obstruction coming on 4 years after the These two articles from the 1920s, which are the sub- operation of ovariotomy. Lancet 1883;1:818–819. ject of this commentary, are quite typical of the compar- 4. Ellis H. Collective review: The cause and prevention of postop- atively early studies on adhesions. Deaver, in the 1923 erative intraperitoneal adhesions. Surg Gynecol Obstet 1971; 133:497–511. article, reviews the concept that it is peritoneal inflam- 5. Ellis H, Moran BJ,Thompson JN, et al. Adhesion related hospital mation and trauma that are the important etiologic fac- readmissions after abdominal and pelvic surgery; a retrospective tors in the formation of these structures. He wisely review. Lancet 1999;353:1476–1480. doubts the efficacy of the various membranes and solu- 6. Menzies D, Ellis H.The role of plasminogen activator in adhesion prevention. Surg Gynecol Obstet 1991;172:362–366. tions advocated at that time in the hope of separating 7. Becker JM, Beck DE, Dayton MT, et al. Prevention of postoper- effectively the traumatized bowel surfaces. ative abdominal adhesions by a sodium hyaluronate-based The Williamson and Mann article of 1922 is repre- bioresorbable membrane: a prospective, randomized, double- sentative of the vast number of articles reporting exper- blind multicenter study. J Am Coll Surg 1996;183:297–306. 8. DiZerega GS, Verco SJ, Young P,et al. A randomized, controlled imental studies of adhesion prophylaxis classified previ- pilot study of the safety and efficacy of 4% icodextrin solution in ously. Using dogs, they applied a mixture of gelatin and the reduction of adhesions following lapaproscopic gynaecologi- gum acacia to their model of gauze abrasion of the liver. cal surgery. Hum Reprod 2002;17:1031–1038. Vol. 200, No. 5, May 2005 Centennial Perspectives 645 CENTENNIAL PERSPECTIVES

Etiology and Treatment of Varicose Ulcer of the Leg

Robert L Kistner, MD Honolulu, HI

“The Etiology and Treatment of Varicose Ulcer of the the major categories of advanced venous disease that Leg” is a title that might well appear in the next issue of were identified during the development of the CEAP the Journal of the American College of Surgeons, but that (clinical, etiologic, anatomic, pathophysiologic) classifi- would be inappropriate because it was published long cation of chronic venous disease in 1995.7 These catego- ago—in 1917 by John Homans.1 This article was re- ries were codified during an international consensus markable considering the limited knowledge of chronic conference held as an ad hoc committee of the American venous disease that existed at that time and the limited Venous Forum, in Hawaii, in 1995.7 Homans actually diagnostic methods available for identification of differ- described aspects of the entire CEAP classification in his ent processes in venous ulcer cases. 1917 article, when he addressed the clinical, etiologic, In an article titled “John Homans’ Impact on Diseases of anatomic, and pathophysiologic processes that could be the Veins of the Lower Extremity, with Special Reference to used to differentiate the simple “ulcers of surface varix” Deep Thrombophlebitis and the Post-Thrombotic Syn- from the more complicated “ulcers of surface varix (non- drome with Ulceration,”2 Robert R Linton gave credit to inflammatory type) complicated by varicosity of the per- Homans for being the first to recognize the “pathologic forating veins,” and a third group called “ulcers of sur- condition and significance of these blood vessels which play face varix (postphlebitic type) complicated by varicosity so much a factor in the etiology of the postphlebitic syn- of the perforating veins.” He associated the deep vein drome and ulcerations.” He was referring to the perforator reflux that may follow postpartum thrombosis with per- veins of the leg. In the same article, Linton gives credit to forator vein reflux in the distal leg. Homans for also recognizing the importance of the deep Within the first two and one-half pages of the 1917 veins in cases of postphlebitic venous ulcer, a theme that article, Homans described ulceration from primary su- he developed and reported for many years.3,4 perficial disease, from superficial plus perforator disease, The basic techniques that Homans had available for and from post-thrombotic disease with superficial and diagnosis were expressed in 1916,5 when he described his perforator reflux. He described illustrative clinical cases use of the Trendelenburg test of raising and lowering the with varicose veins, skin changes, and ulceration; he dif- leg to observe rapid refill through the surface vessels, and ferentiated the etiologic syndromes of primary and post- the addition of a constriction test in which he blocked thrombotic disease and identified the anatomic involve- the veins in the proximal extremity to observe the venous ment of the superficial and perforator veins, and the filling time distally as a measure of perforator vein in- serious pathophysiologic implications of both axial and competence. It was with these tests that he was able to perforator venous reflux. In the case reports, he associated divine the routes of reflux into the distal leg and to clinical state of different venous ulcers with the two causes differentiate those with simple varicose veins as the cause of simple reflux (primary reflux) and postphlebitic reflux of ulceration from those with more complicated reflux (secondary reflux); he addressed anatomic involvement by through the perforating veins. Of course, this was done separating superficial from perforator and from deep vein without imaging studies because venography was not reflux; and addressed the pathophysiologic mechanism of described until 1938 by dos Santos6 and duplex scan- reflux as separate from obstruction. ning was not to appear until the late 1970s from the After describing the various ulcer types and their patho- laboratories of Eugene Strandness at the University of physiologic basis, he recommended that treatment of ulcer Washington in Seattle. associated with simple varicose veins include excision of the Many of the points in Homans’ 1917 article might varicose veins down to the ulcer. For the more complicated still be accepted for publication today. An intriguing cases with perforator involvement he performed a more aspect of this article is that it described elements of all of radical procedure with complete groin dissection of the

© 2005 by the American College of Surgeons ISSN 1072-7515/05/$30.00 Published by Elsevier Inc. 646 doi:10.1016/j.jamcollsurg.2004.10.025 Vol. 200, No. 5, May 2005 Kistner Etiology and Treatment of Varicose Ulcer of the Leg 647 saphenous vein and stripping in the thigh, removal of the REFERENCES veins in the calf down to the ulcer, and excision of the ulcer 1. Homans J. The etiology and treatment of varicose ulcer of the leg. down to normal deep tissue whether that be periosteum, Surg Gynecol Obstet 1917;24:300–311. muscle sheath, or fat. Incompetent large perforator veins in 2. Linton RR. John Homans’ impact on diseases of the veins of the the base of the ulcer were ligated. Skin grafting could be lower extremity, with special reference to deep thrombophlebitis immediate or delayed. and the post-thrombotic syndrome with ulceration. Surgery 1977;81:1–11. This article illustrates the piercing insight into patho- 3. Homans J. Exploration and division of the femoral and iliac veins logic processes gained by an especially perceptive and in the treatment of thrombophlebitis of the leg. N Engl J Med intuitive physician using only his clinical skills of careful 1941;224:179. physical diagnosis and analysis of collective experience 4. Homans J. The surgery of the veins of the legs: varicosity and some problems in thrombosis.RIMedJ1945;28:565–582. in the course of day-to-day practice. Homans’ article was 5. Homans J. The operative treatment of varicose veins and ulcers, quoted in most of the important literature on venous based upon a classification of these lesions. Surg Gynecol Obstet ulceration for 40 years after it appeared in Surgery, Gy- 1916;22:143–159. necology & Obstetrics, and remains a classic contribution 6. Dos Santos JC. La phlebographie directe. Conception, technique, premier resultants. J Int Chir 1938;3:625. with pertinent observations currently. It represents a true 7. Porter JM, Moneta GL. Reporting standards in venous disease: landmark in the advancement of our understanding of an update. International Consensus Committee on Chronic chronic venous disease. Venous Disease. J Vasc Surg 1995;21:635–645. SCUDDER ORATION ON TRAUMA

Changes in the Management of Injuries to the Liver and Spleen

J David Richardson, MD, FACS

At the completion of my residency training in 1976, of other publications on these injuries. Virtually every there were relatively few controversies in the treatment publication on hepatic and splenic trauma written in the of injuries to the liver and spleen. Diagnosis of these English language before 1950 was reviewed, as were sev- injuries usually could be made by physical examination, eral sentinel papers written in German and French. Most positive diagnostic peritoneal lavage (DPL), or at oper- of the major reports on these injuries in the latter half of ation mandated by a penetrating wound. If injury to the the century were also examined, but the prodigious spleen occurred, splenectomy was indicated. Diagnosis growth in publication on these topics precluded a com- of liver injury was likewise noncontroversial, and only plete literature review. Nonetheless, more than 500 re- the methods of treatment of the hepatic lesions engen- ports were reviewed for this article. dered debate. But by the end of the last century, virtually A comparison of the etiology of splenic rupture and none of the dogma believed to be unequivocally true 25 liver hemorrhage between the first and latter halves of years earlier was practiced. The majority of splenic inju- the 20th century is also instructive. From 1930 to 1940, ries were detected by CT scan and treated without op- citations on spontaneous splenic rupture greatly out- eration. Liver injuries that resulted in several liters of numbered those on traumatic injury. Splenic hemor- blood and bile in the abdomen were observed if the rhage secondary to malaria, typhoid, and mononucleosis blood pressure could be maintained with several units of were reported much more commonly than hemorrhage blood. A surgical Rip Van Winkle, who awoke in 2000 produced by trauma.The 1930s produced fewer than 30 after 25 years of slumber, would never have believed the citations on liver and spleen injuries in the English lan- radical changes that occurred in the treatment of injuries guage; the 1990s produced more than 1,300 reports on to these solid organs, whose diagnosis and management those topics. had once seemed so straightforward. The purpose of this discussion is to review the changes in the treatment of injuries to the liver and MECHANISM OF INJURY: spleen that occurred during the past century. It is hoped CHANGES IN PATTERNS that the exhaustive literature search that is the lynchpin The changes in mechanism of injury are illustrated (Fig. 1)in of this presentation will serve as an evidence-based re- three time intervals: the early, middle, and latter portions of view that either validates or challenges some of our cur- the last century. In the early period, nearly one-third of all rent concepts about solid organ injury management. splenic injuries and one-fifth of liver injuries reported were caused by a variety of mechanisms classified as miscellaneous. These include industrial and farm injuries, falls, and METHODS OF REVIEW mechanisms other than motorcycle or motor vehicle Full-length papers (not abstracts) pertaining to injuries crashes. Gunshot wounds were much less common than of the liver and spleen were reviewed. Citations were were reported later in the century. By midcentury, miscel- extracted from the Current List of Medical Literature and laneous injuries became a minor part of the reported cases its successor, Index Medicus, and from the bibliographies as a percentage because the number of liver and spleen injuries increased dramatically. Stab wounds as a mecha- Presented at the American College of Surgeons 90th Annual Clinical Con- gress, New Orleans, LA, October 2004. nism of penetrating trauma increased greatly, accounting for 40% and 15% of liver and spleen injuries, respectively. Received October 26, 2004; Accepted November 2, 2004. From the Department of Surgery, University of Louisville School of Medi- What is often underappreciated is the impact of cine, Louisville, KY. mechanism on mortality rates. The increase in stab Correspondence address: J David Richardson, MD, FACS, Department of Surgery, University of Louisville, 550 South Jackson Street, Louisville, KY wounds reported had a dramatic impact on improve- 40292. ments in mortality data, particularly for liver injuries.

© 2005 by the American College of Surgeons ISSN 1072-7515/05/$30.00 Published by Elsevier Inc. 648 doi:10.1016/j.jamcollsurg.2004.11.005 Vol. 200, No. 5, May 2005 Richardson Managing Liver and Spleen Injuries 649

total mortality rate. The first huge series of civilian liver Abbreviations and Acronyms injuries14 reviewed the experience from the Jefferson DPL ϭ diagnostic peritoneal lavage Davis Hospital in Houston, TX, from 1939 to 1961. ϭ EAST Eastern Association for the Surgery of Trauma The overall mortality rate from 640 consecutive patients NOM ϭ nonoperative management was 17.3%. Mortality rates from gunshot wounds and blunt trauma were 26% and 45%, respectively, but the 296 stab wounds had a mortality rate of only 3.4%. Before World War II, the mortality rate for liver injury Reports from the latter portion of the century show a was often reported at 30% to 40%.1-11 One of the larger major decline in stab and gunshot wounds, with a tre- articles on hepatic trauma published a few years after mendous increase in blunt injuries. Because blunt World War II reported a mortality rate of only 10%.12 trauma generally has the highest mortality rate, any im- Although the results of this report were hailed as a major provement in mortality is likely from improvement in advance, they may be somewhat deceiving. Of the eight management. blunt injuries, only three were severe. Thirty-four stab wounds occurred, and no patient with an isolated stab wound died, which greatly improved overall results. DIAGNOSTIC ERAS Amerson and Blair13 described 189 patients treated be- As with changes in mechanism, differences in diagnostic tween 1947 and 1958, with a total mortality rate of modalities have been dramatic over time; such changes 16.4%. Twenty-two patients with blunt injury had a in diagnosis have had a major impact on not only differ- mortality rate of more than 45%, and the 99 patients ences in management but in reported mortality rates. with gunshot wounds had a 21% death rate. But none of Although numerous iterations of diagnostic eras could the 63 stab wound victims died, greatly lowering the be conceived, four will be considered (Table 1).

Figure 1. The mechanism of spleen and liver injury from selected series in three periods of time. In the early period there were a large number of miscellaneous injuries, which accounted for less than 1% during later periods. In the mid-portion of the 20th century, stab wounds were commonly encountered in liver injuries. Blunt trauma is now the predominant mechanism of injury for both spleen and liver injury. GSW, gunshot wound. 650 Richardson Managing Liver and Spleen Injuries J Am Coll Surg

Table 1. Diagnostic Eras noted that its accuracy did not approach 100%.The in- 1) Operation or autopsy accuracy of these indirect means of diagnosis led to the 2) Primitive diagnostic efforts development of DPL. 3) Diagnostic peritoneal lavage Root and colleagues16 developed the technique in the 4) Focused imaging techniques laboratory by infusing blood into the peritoneum and then aspirating the contents after peritoneal lavage of the From the beginning of operative treatment of trau- cavity with saline. This technique proved highly sensi- th matic wounds in the late 19 century through World tive for blood in the peritoneal cavity and was the diag- War I, diagnosis was made definitively by either opera- nostic method of choice for detection of hemoperito- tion or autopsy. Although one could assert that is still neum for more than 30 years. Although DPL was very true, in that era, there were no major reports of other sensitive for blood, its high degree of accuracy made it diagnostic strategies. Operations were usually based on less useful at times. Virtually all surgeons who used DPL physical signs or on mechanism of injury, as in the case noted that in some patients, injuries that produced of penetrating wounds. Emphasis in the literature was bleeding had either ceased or the injuries were so incon- on physical findings that might suggest solid organ in- sequential that no therapy was required. Nonetheless, juries, given the lack of other diagnostic methods. Bal- DPL was a major advance because it greatly diminished lance’s sign described dullness to percussion or shifting the incidence of missed solid organ injuries in the era dullness in the left upper quadrant. Kehr’s sign, referred before accurate scanning was available. The impact of to pain to the left supraclavicular region, was believed to DPL on reported improvements in mortality rates for be an important sign of splenic injury. Tenderness and liver and spleen injury cannot be underestimated. shock were also clinical indications of the need for Prompt use of this technique decreased missed injuries operation.4,8-10,12 and allowed rapid operative planning in emergent cases. Not surprisingly perhaps, several reports based on au- Clearly, treatment improved during this era, but patients topsy studies concluded that nonoperative therapy for with minor injuries were always treated operatively and liver injuries was uniformly fatal. One early study re- rarely died. This inclusion of less severely injured pa- ported a 100% mortality rate for nonoperative treat- tients certainly enhanced reported mortality rates. ment, with a 50% mortality rate for patients who had The current diagnostic era relies primarily on imaging operations. Although several reports assumed that some injuries must heal without operation, the lack of precise techniques to aid in diagnosis of solid organ injury. The diagnostic studies made it difficult to accurately deter- use of abdominal ultrasound has proved extremely effi- mine the incidence of either liver or spleen injury or cacious for detecting blood in the peritoneal cavity. This their actual mortality rates. The next phase of attempts technology has the advantage of portability, and when at diagnosis involved an indirect effort to establish a done by the surgeon or emergency physician, can be- solid organ injury by use of radiographs or by attempts come a part of the physical examination. But abdominal to confirm the presence of hemoperitoneum by needle ultrasound lacks specificity in terms of predicting the puncture. Several papers in the 1940s emphasized the source of blood and cannot grade organ injuries. following radiologic features for splenic injuries: obliter- CT scanning has become the gold standard for diag- ation of the splenic shadow; indentation of the gastric nosis of solid organ injury. CT scanning allows reason- bubble; reflex distention of the stomach; and tenting of ably accurate grading of organ injuries and provides the left diaphragm. But there were few reports on radio- crude quantitation of the degree of hemoperitoneum. logic features of liver injuries.15 Additionally, the use of oral contrast material permits a Beginning in the 1930s, many surgeons attempted to degree of diagnostic accuracy in excluding visceral inju- aspirate blood from the peritoneal cavity of trauma vic- ries. CT scanning is mandatory for patients with blunt tims with suspected solid organ injury. Whether by para- trauma whose solid organ injuries are to be managed centesis or four-quadrant tap, confirmation of blood in nonoperatively. CT has also been useful for detecting the peritoneum mandated operation. The diagnostic ac- missile tracts in penetrating trauma patients. Such infor- curacy rate for blunt liver and spleen injuries ranged mation is imperative for surgeons who want to attempt from 30% to 70%, but all reports on this technique nonoperative management of penetrating wounds.17-20 Vol. 200, No. 5, May 2005 Richardson Managing Liver and Spleen Injuries 651

MANAGEMENT OF LIVER INJURIES Table 2. Advances in Operative Treatment of Liver Wounds th The management of hepatic trauma can be divided into from 19 Century Through World War II Date Surgeon Advance three eras: from the beginning of operative treatment to 1888 Langenbuch Mass ligature controlled liver bleeding World War II; the postwar period to the mid-1980s 1890 Clementi Introduced inflow occlusion by (characterized by aggressive operative treatment); and clamping portal triad from the late 1980s to the present. In this latter period, 1894 Ceccharelli and Tamponade bleeding liver with there have been several major philosophical changes in Bianchi decalcified bone sutured on superior treatment that represent radical departures from previ- and inferior liver surface 1896 Kousnetzoff Special liver needle to suture hepatic ously accepted standards of care. and Penski wounds 1902 Beck Rubber catheter suture around the Late 19th century through World War II liver wound for compression Some of the advances in the operative treatment of he- 1902 Kocher Gastrointestinal clamp left on liver to patic injuries developed from the late 19th century compress bleeding 1,3,20-26 1905 Gillette Mattress through liver and skin (two through World War II are outlined inTable2. One survivors) of the notable aspects of this historical list has been the 1906 Schroeder Perihepatic packing rediscovery of many of the concepts that were intro- 1908 Pringle Compression of artery and vein for duced about a century ago but failed to find a niche in inflow occlusion the treatment of liver injuries, or were attempted and 1910 Boljarski Omental pedicle sutured over liver then subsequently rejected. Examples of the former in- wound 1911 Cushing Topical striated muscle on bleeding cluded creation of an omental pedicle to be sutured into liver for hemostasis 22 a liver wound, which was described in 1910, but did 1913 Halstead Rubber under packing not gain popularity for another 75 years. Topical fibrin 1915 Grey Topical sheep fibrin was introduced in 1915 and did not gain acceptance. 1918 Harvey Topical beef fibrin Although it is occasionally used today, it is not regarded 1939 Seegers Topical thrombin as standard therapy. Packing was reported in the late 1943 Frantz Oxidized cellulose (Oxycel) 1800s in Europe and in the United States in 1906.25 It 1944 Ingraham Topical thrombin and Gelfoam was a mainstay of therapy in World War I and in the (absorbable sponges) 1945 Light and Gelfoam period between the world wars. During World War II, Prentice the value of direct repair of liver wounds along with drainage of bile was highlighted. The resultant dramatic improvement in mortality provoked condemnation of declined to 9.6% by 1945. Only a small percentage packing as a means of treating liver injuries. (about 5%) required hepatorrhaphy in either time pe- The most common treatment administered to nearly riod. Drainage was used in nearly 90% of patients by the 600 civilian patients treated from the mid-1930s to end of the war. So, several trends developed that were 1945 was drainage alone followed by suture repair of the extended into civilian practice after the war: abandon- liver (usually with catgut). The development of absorb- ment of gauze packing; suture control only for active able hemostatic agents led to an explosion in their use. bleeding; and nearly uniform drainage of all wounds. This collected group of patients suffered a 27% mortal- Impact of World War II ity rate. Although the death rate was high by current stan- The mortality rate reported for operative treatment of dards, it was a dramatic improvement from the two-thirds liver injuries in World War I was 66.2% when packing who died in World War I. Likewise, the authors noted only was used as the primary mode of therapy. After World a 9.7% mortality rate when the liver wound was isolated. War II, Madding and colleagues27 reported on 829 op- Interestingly, these authors gave little credit for improve- erations performed for liver injuries by the Second Aux- ment in mortality rates to general advances in perioperative iliary Surgical Group. Although gauze packing was used care. They asserted a direct link between two operative les- in 28% of the total group of treated patients, this method sons and mortality, ie, discontinuation of gauze packing was predominantly used early in the war years. In 1941, and routine use of drains.These lessons were widely applied more than 34% of patients were packed; that number in the civilian experiences for the next 40 years.28-31 652 Richardson Managing Liver and Spleen Injuries J Am Coll Surg

Operative treatment: 1945 to 1985 Table 3. Surgical Trends from 1945 to 1985: Period of After the end of World War II, there was a great in- Aggressive Surgical Treatment crease in hepatic trauma from both blunt and pene- No gauze packing trating injuries. Beginning in the 1950s, multiple Uniform drainage Increased use of liver debridement large series of liver injuries were reported.14,31-36 In Tractomy addition to the standard treatment of hepatorrhaphy Popularization of omental patch for bleeding, packing with absorbable hemostatic Brief rise of anatomic resection agents, and routine drainage of bile leaks, several ad- Selective hepatic artery ligation ditional strategies were developed (Table 3). In re- Atriocaval shunt for retrohepatic vena caval injury sponse to more serious injuries undoubtedly caused Rediscovery of temporary inflow occlusion by more gunshot wounds and high-speed motor vehicle crashes, additional treatment options were re- uniform lethality of retrohepatic vena caval injuries with quired. Better resuscitative techniques and improved attempt at direct repair led to the development of the blood banking made early death from shock less com- atriocaval shunt. This technique, developed by Schrock mon and increased the number of patients with oper- and associates,41 theoretically bypassed the caval injury able, severe liver injuries. So increasing numbers of and allowed direct suture repair of the cava itself and patients died of hemorrhage during this period. Al- main hepatic veins. The operation required opening the though most reports did not discuss the source of chest to expose the atria. This bicavitary exposure accel- hepatic hemorrhage, four types were noted: arterial erated hypothermia and coagulopathy in many patients. hemorrhage; major venous including the retrohepatic Consequently, the mortality rate remained high, but the caval injury; ooze from injured or devitalized tissue; concept of direct repair of this deadly injury was very and a combination of these. Operative treatments important. were often developed to treat specific patterns of Both previously mentioned bleeding problems often hemorrhage. were treated initially with temporary inflow occlusion by clamping the portal triad. The concept of inflow occlu- Treatment of arterial hemorrhage sion actually predated Pringle,26 but his work published Several specific modalities began to be used more often in 1908 was rediscovered and popularized in the 1960s to treat arterial bleeding. Hepatorrhaphy was used with after rarely being mentioned in the literature for more increased frequency. When the arterial bleeding occurred than 50 years. deep within the hepatic parenchyma, a tractomy was Diffuse bleeding from damaged or devitalized liver advocated to expouse and suture ligate the arterial flow. increasingly required surgical treatment. Reports on ci- But control of deep arterial bleeding was often techni- vilian liver injuries from the 1950s generally cautioned cally difficult to accomplish.31,34-36 against debridement of damaged liver for fear it would In response to futile attempts to directly suture ligate worsen preexisting hemorrhage. Absorbable gauze pack- arterial bleeding, Dr Aaron’s group performed ligation ing and drainage were mostly used for this problem. As of the hepatic artery.37 Initially performed at the Louis- the forces of injury increased, other techniques were re- ville General Hospital to control arterial hemorrhage quired. Resectioned debridement was increasingly used. from a ruptured hepatic adenoma, Mays found this There was a brief flurry of activity with use of major technique useful to control arterial bleeding in trauma anatomic resections, but the high mortality rate of this patients. A literal explosion in its use occurred in Louis- procedure led to discontinuing its use in most American ville, and surgeons there proposed it to prevent rebleed- centers.42,43 The omental pedicle described for liver in- ing.38,39 A high rate of infection led to reconsideration of jury in 1910 and mentioned occasionally through the its use, and it was subsequently used less frequently, al- years was reintroduced by Stone and Lamb44 and gained though it remained an operation that could occasionally widespread popularity. be life-saving.40 Viewed from the perspective of several decades re- Major venous bleeding was recognized as a major moved, this era produced relatively few techniques in source of mortality, particularly in patients who had widespread use today. Clearly, debridement of devi- been in high-speed motor vehicle crashes. The nearly talized hepatic tissue is a concept that remains valid Vol. 200, No. 5, May 2005 Richardson Managing Liver and Spleen Injuries 653 but seems to be required less frequently than in the Table 4. Treatment Trends: 1985 to Present past. Omental pedicle flaps are very useful in a variety Direct repair of perihepatic venous injuries of clinical scenarios in which operation is required. Perihepatic packing and damage control strategies The atriocaval shunt is still widely mentioned as a Arteriographic embolization of hepatic arterial hemorrhage procedure for retrohepatic caval injuries, but its very Nonoperative treatment high mortality rate and the technical tour de force required for its placement (particularly because it’s Direct repair of perihepatic venous injuries not often needed) dictates its infrequent use today. Injuries to the perihepatic veins remain an unsolved Likewise, selective hepatic artery ligation is rarely problem. Although major venous injuries are fairly un- used today,45 and tractomy is rarely needed. Hepatic common, they are, nonetheless, highly lethal. Atriocaval lobectomy for trauma is almost never done unless the shunting was generally regarded as the optimal means to injury itself performs the dissection. treat this problem, but a review of its results are bleak. Although some results appear reasonable given the seri- Advances in treatment: 1985 to present ousness of the injury, the combined mortality rate re- In the 40 years from World War II to 1985, there were ported in the literature is high. The results from San enormous advances in diagnosis and treatment that Francisco General Hospital from 1968 to 1987 using markedly reduced the lethality of hepatic injuries. The this technique on 27 patients produced a mortality rate advent of DPL reduced missed injuries, but was begin- of 55%.50 This was the best experience reported in the ning to be challenged by CT scanning. CT scanning literature. Cogbill and associates47 reported on a multi- allowed greater specificity of diagnosis and permitted a center experience that treated 38 patients, with only 4 range of therapeutic options based on the ability to de- survivors. No patient with blunt injury survived when a termine degree of hemoperitoneum and grade of injury shunt was placed. Extracting mortality data from dozens that had not been imagined previously. Concurrently, of papers on liver injuries disclosed more than 412 cases the high incidence of nontherapeutic operation per- of shunt placement, with an 88% mortality rate. Despite formed as a consequence of DPL was increasingly rec- an occasional favorable report on shunting,51,52 most re- ognized. Deaths from perihepatic infection had dimin- ports through the years eschewed the use of atriocaval ished, presumably because of better debridement of shunting in favor of direct methods of repair.53 Bethea54 devitalized tissue, better resuscitation, better use of drains, reported three cases of direct repair using inflow occlu- and perhaps better antibiotic therapy. Despite these ad- sion, with all patients surviving. A review reported from vances, patients were still dying in great numbers sec- Charity Hospital noted a 40% survival rate with perihe- ondary to hemorrhage from the liver, from both blunt patic venous injuries treated without a shunt, which was and penetrating trauma. Several major problems were comparable to the best reported results with a shunt. identified as causative factors in death from hemorrhage. Coln and colleagues55 treated four children successfully There were numerous deaths from the vicious cycle of using a direct repair technique. Frustrated by their re- hypothermia and coagulopathy produced by major he- sults with the atriocaval shunt, Pachter and coworkers56 patic bleeding.46 Clearly, in many patients, direct con- treated five patients without a shunt, and all survived. trol of bleeding using standard surgical therapy was This group noted four principles that were believed im- not efficacious. Likewise, techniques advocated for portant in their good outcomes: compression of the liver control of perihepatic venous injuries usually failed. wound while the patients were being resuscitated; early New strategies were once again needed to address old recognition of the venous injury; portal triad occlusion problems and newer ones created by more seriously with steroids and topical hypothermia for liver protec- injured patients.47-49 tion; and the use of finger fracture through hepatic pa- Several reports in the late 1980s and early 1990s fo- renchyma to allow access to the bleeding veins. Results cused on these severe liver injuries and the continuing from these authors have inspired more direct repairs of problem of death from hemorrhage. Four strategies perihepatic venous injuries.57,58 (Table 4) that appear to have significantly decreased the Our group in Louisville reported a small series of mortality rate associated with hemorrhage evolved over patients using a different technique of direct repair.59 the latter part of the century. Three patients were successfully treated by clamping 654 Richardson Managing Liver and Spleen Injuries J Am Coll Surg bleeding hepatic veins with fine vascular clamps to con- much sooner than had been practiced in the pre-World trol the hemorrhage. An omental flap was then created, War II era to avoid infection.71 Some advocated pack the liver was packed, and the abdomen covered with a removal as soon as associated coagulopathies could be temporary sheet closure. The clamps were removed corrected, but the majority of reports advocated leaving within 24 hours, and no patient had major bleeding after packs for 1 to 2 days, for a mechanical effect as well.69,72,73 the clamps were removed. Three additional patients The tremendous amount of resuscitation these pa- have now been successfully treated with this technique. tients received and the volume occupied by the packs Direct repairs appear to offer a better alternative to atrio- created another series of problems. Attempts at fascial caval shunting in patients who require operative treat- closure of the abdomen usually met with failure. Even ment for this problem. closure of the skin often led to an increased intraabdominal pressure and a constellation of clinical events labeled Perihepatic packing as “abdominal compartment syndrome.”74 Many prob- As previously noted, perihepatic packing was a concept lems associated with this condition could be ameliorated popularized early in the 20th century by Schroeder,25 but by leaving the abdomen open and covered with some it was actually used in the late 1800s. The high mortality type of temporary atraumatic material. The open abdo- rate before World War II appeared to cause its virtual men itself is responsible for the considerable late mor- abandonment. But several large series of hepatic wounds bidity rate in survivors. Intestinal fistulas and huge ab- continued to report sporadic use of this technique. In a dominal wall hernias have been among the most vexing. 1976 review of 625 patients with liver injury, Lucas and Recent efforts have focused on earlier attempts at ab- Ledgerwood34 noted that 3 patients survived after pack- dominal wall closure to prevent loss of domain and de- ing. In 1979, Calne and colleagues60 advanced the con- velopment of huge midline hernias. The use of vacuum- cept of packing before transfer to a center for advanced assisted closure devices appears to be a promising step in treatment, with all 4 patients treated by this method this direction. surviving. By 1981, Feliciano and associates61 called for a Even 20 years after the resurrection of packing as a reappraisal of this technique after 9 of their 10 seriously treatment alternative, it remains an important part of injured patients treated by packing survived. In rapid the armamentarium of surgeons in managing difficult succession, two additional reports62,63 had more than hepatic injuries. 80% survival rate of patients with difficult bleeding injuries using packing. The term damage control was used Angiography and transcatheter embolization to describe this strategy for those patients with multiple Another important advance of the last 20 years has been the injuries, coagulopathy, hypothermia, and diffuse oozing development of transcatheter embolization for bleeding ar- from the liver and other areas.64 terial injuries within the liver. Numerous reports75-78 on Because the indications for packing included most this technique have been published, and technical suc- seriously injured patients, the reported mortality rate cess rates are usually greater than 80%. rose. By 1988, Feliciano and associates65 reported on a Mohr and coworkers79 reported on 26 patients who followed group of 66 patients who were packed; 17 died underwent angiographic embolization either early or early in the operative or postoperative period. The mor- later in the course of treatment. Two patients bled again tality rate of those who survived 24 hours was 47%. and were successfully treated by a second angiogram. Successive reports66-70 from several institutions showed a These authors noted that the mortality rate of this group high mortality rate, but concluded that innumerable pa- of patients was low, but considerable morbidity oc- tients were being salvaged who previously would have curred. These complications included five patients with died. hepatic necrosis, four of whom had an infarcted gall- Several lessons were learned from these extensive ex- bladder. Gallbladder infarction has been noted in several periences with perihepatic packing.66,69 First, patients other experiences.80,81 must be packed before their survival is unlikely to occur Indications for this procedure vary to some degree under any circumstances. Packing was excellent for dif- among institutions, but often include the presence of a fuse ooze or from venous bleeding, but was not effective contrast blush on CT scan, particularly in patients who for arterial hemorrhage. Pack removal should occur have required blood transfusion. Some centers recom- Vol. 200, No. 5, May 2005 Richardson Managing Liver and Spleen Injuries 655 mended angiography for most severe liver injuries in deaths. Additionally, her group collected 21 reports on blunt trauma patients. Angiography may be helpful in nonoperative treatment including both children and conjunction with other treatment modalities such as adults. In the 286 combined patients, only 1 patient died packing or nonoperative treatment.82-85 after starting warfarin sodium therapy postinjury. Myriad reports have subsequently validated the safety and efficacy Nonoperative treatment of blunt liver trauma of this approach.91-100 Currently, up to 90% of patients with Nonoperative treatment, thought to be a novel concept blunt liver injuries are managed nonoperatively. in the past 15 years, also represents ideas once practiced Although most reports on nonoperative therapy have and then abandoned. Tellmans reported experimental focused on avoidance of operation as the primary bene- observations in 1879 in which wedges of liver were re- fit, our group101 has suggested that mortality rates may moved from animals and the abdomen closed. He noted be directly improved by this treatment as well. Despite most animals spontaneously ceased bleeding promptly the fact that the number of total liver injuries treated has and generally survived. Clinically, Hinton,86 in 1926, increased in the past 25 years, and the number of severe noted liver injuries were relatively common and advo- liver injuries as judged by CT scanning has slowly in- cated “conservative” or nonoperative treatment because, creased, the mortality rate has declined. Particularly, the as he noted, most bleeding from the liver spontaneously number of patients who required operative treatment for ceased. Numerous reports discussed a selective approach perihepatic venous injuries has declined because we have to the management of hepatic injuries based on clinical used observation as the primary treatment. In our review factors such as hemodynamic stability. of 25 years’ experience,101 we treated 2.7 juxtahepatic The countercurrent argument was that nonoperative venous injuries per year from 1975 to 1994, but now treatment was associated with virtually a 100% mortal- operate on only 1.5 patients per year. Because the num- ity rate, so all patients with suspected or diagnosed liver ber of patients and the grade of injury based on CT scan injuries must have an operation. Improved mortality have increased, it would seem that the number of venous rates during and after World War II assured the primacy injuries should have actually increased. It is our hypoth- of operative treatment. esis that many patients with venous injuries that are Three observations prompted the move toward non- low-pressure wounds cease bleeding spontaneously and operative treatment. Several reports of injuries to chil- heal if they are left alone. The anecdotal reflections of dren demonstrated the efficacy of nonoperative treatment many experienced trauma surgeons note that some pa- for blunt injuries.87,88 First, the practice of nonoperative tients who were stable preoperatively die during opera- treatment was initially advocated for splenic injuries and tion, after the venous injury is disturbed and profuse then extended to liver wounds.The success in children led bleeding is initiated. to attempts of nonoperative treatment in adults. Second, Current protocols for nonoperative management of the high rate of nontherapeutic operations in many pa- hepatic injuries are relatively standard in most trauma tients with blunt hepatic injuries was not in patients’ centers. The key feature in the ability to offer nonopera- best interest. Third, the advent of CT scanning greatly tive treatment is hemodynamic stability. Patients who facilitated both diagnosis and grading of injuries and are not stable must have operation or angiography (in gave some reassurance that the intestinal injuries had not selected patients). Most protocols will allow continued occurred. observation with up to4Uofblood transfusion for the Grading and stratification of both liver and spleen hepatic injury itself. CT scans should be performed to injuries represented a major advance89 and furthered the confirm the diagnosis and attempt to exclude other in- concept of selective treatment based on injury severity. juries, particularly to the hollow viscera. Patients are Initially, nonoperative treatment was used only in pa- usually admitted to a closely monitored unit and kept on tients who required no blood transfusion, and only pa- bed rest for several days. tients in the most stable condition received such treat- The success rate of nonoperative treatment has been ment. Success with the method led to liberalization of remarkably high. The necessity for operations for ongo- indications for observation alone. In 1990, Knudson ing hemorrhage has been reported to be from 5% to and colleagues90 reported on 52 adults treated in the 15%. There remains a concern over missed bowel inju- decade of the 1980s with no delayed hemorrhage or ries that have been reported from 1% to 3%.102-106 656 Richardson Managing Liver and Spleen Injuries J Am Coll Surg

Nonoperative treatment has created additional prob- developed abdominal compartment syndrome. Because lems less frequently encountered in the operative era. nearly a third of the patients assigned to nonoperative Hemobilia is seen not uncommonly and may require treatment failed, this result would not seem strongly angiographic embolization. Perihepatic fluid collections positive, but the authors believed it was a useful practice might need to be drained. Endoscopic retrograde cholan- in selected patients. Moore,112 in an accompanying edi- giopancreatography (ERCP) is occasionally needed for torial, questioned the wisdom of this form of manage- larger bile leaks. Our unit reported that a significant ment. At least one other experience on this treatment number of patients treated nonoperatively needed some strategy has been reported with reasonable success.113 form of intervention to treat secondary problems created The use of CT scanning potentially permits the mis- by the initial injury.107,108 Major bile collections should sile tract to be outlined, and if the tract is confined to the not be allowed to remain in the abdomen.109 We have liver and the patient is stable, operation may be avoided. practiced routine drainage of large collections through Our unit has had two patients with missed colon injuries laparoscopy and have treated more than 30 such patients in a small group of patients treated nonoperatively. This by this method. The procedure is performed several days led to discontinuation of the practice of nonoperative after admission using a gasless system. Laparoscopy per- treatment of hepatic gunshot wounds. Clearly, this is a mits thorough irrigation and suctioning of old blood technique that must be used with great caution in highly and bile, which we remove from around the liver, the selected patients. gutters, and the pelvis. A suction drain is usually placed as well. No patient has bled after the procedure, air em- Recommendations for treatment of liver injuries bolism has not occurred, and no technical complications The simple algorithm in Figure 2 outlines current rec- have been observed. The amount of fluid removed has ommendations for treatment of liver injuries. Obvi- ranged from 800 to 4,500 mL. In addition to subjective ously, clinical circumstances and capabilities within the improvement, many patients have decreases in heart treating institution will have an impact on the treatment rate, better respiratory mechanics, decreased leukocyto- of individual patients. sis, and decreased temperature after this procedure. There are numerous unresolved questions in the non- Changes in mortality from hepatic trauma operative management of these patients, including the A steady decline in mortality rate has occurred be- importance of bed rest, the timing of return to normal cause of the inception of treatment of liver injuries activities and exercise, and the role of followup scans. (Fig. 3).1,2,4,6,8,10,12,14,27-29,31-36,101,114-123 In 1987, Edler114 But nonoperative treatment seems to be a secure treat- collected 543 cases from the world’s literature, with a ment at this point. There are no firm recommendations mortality rate of 66.8%. Several collected reports before about followup scans.110 Our unit generally obtains such World War I disclosed mortality rates of 60% to 80%, scans, but we have no defined protocol to recommend although it appears many of the same patients were re- their timing. Likewise, we have patients avoid strenuous viewed by each author.115-118 The mortality rate during activity for several months, but this recommendation is World War I was 66.2%. intuitive rather than data-driven. Collected series between the world wars disclosed 416 patients, with a mortality range of 30% to 81% in these Nonoperative treatment for penetrating reports, and an overall mortality rate of 69%. Several hepatic injuries civilian experiences reported in the mid-1940s had a Nonoperative treatment of abdominal stab wounds has mortality rate of 55%. The classic paper by Madding been practiced successfully in numerous centers for and associates27 on their experiences during World War many years, and some patients with liver injury have II, reported a marked decline in mortality to 27% of 829 been so treated. Nonoperative treatment of gunshot patients treated. The experiences of several large civilian wounds has been more controversial. Demetriades and reports in the 1960s, 1970s, and 1980s show declining associates111 reported 36 patients with gunshot wounds total mortality rates of 12.7%, 8.7%, and 6.0%, respec- to the liver, of which 16 were initially treated without tively. Beginning in the 1970s, many reports began to operation. Five required delayed operation. Four pa- dissect liver-related mortality from total mortality. In the tients required operation because of bleeding and one last three decades of the 20th century, the liver mortality Vol. 200, No. 5, May 2005 Richardson Managing Liver and Spleen Injuries 657

Figure 2. The demarcation point for management of either blunt or penetrating injury is hemodynamic stability. The broken lines indicate treatment protocols advocated by some that generally are not recommended by most authors. rate was reported at 6.6%, 5.1%, and 4.2%, respectively, War II, in fact, most of them were collected cases that in these collected series. Richardson and colleagues101 generally analyzed the same reports and added a handful demonstrated a similar trend in a 25-year analysis of a of their own. Before the mid-1930s, it appears there were database of more than 1,800 patients with liver injury at less than 500 cases of ruptured spleen associated with the University of Louisville. trauma reported in the English literature; in fact, some The focus in the literature in the past 15 years has of these traumatic injuries appear to have damaged ab- been on the management of complex injuries and the normal spleens. The mortality rate for these patients was unsolved problems of patients with grades IV and V approximately 38%. injuries.124-128 Whether it is reasonable to expect further In a manual entitled Abdominal and Genitourinary improvements in hepatic-related mortality is difficult to Injuries, published by the National Research Council in predict. 1942, it states that splenectomy for trauma “in the ex-

INJURY TO THE SPLEEN Historical references yield conflicting information on the origins of splenectomy, although mention of the operation dates back to the mid-16th century.129 Bessel- Hagen reported 37 splenectomies for ruptured spleen in the German literature in 1900, although the causes of the ruptured spleens were not elucidated. In fact, reports of splenic rupture until the time of World War II were more likely from mononucleosis or malaria than trauma. Regardless of the primacy of splenectomy for trauma, the operation was performed with some frequency by th the early 20 century, with gradually improving results Figure 3. The mortality for liver injury from selected reports in the (Table 5). Although the total mortality rate is high in literature over the past century demonstrates the progress in im- many of these series, then, as now, most of the deaths proved mortality. Total mortality is reported in the 6% to 7% range. Mortality related to the liver injury itself has been reported from the were from associated injuries. Although there are few 1970s to the present and now is often less than 5% in many reports of a large number of splenectomies before World experiences. WW, World War. 658 Richardson Managing Liver and Spleen Injuries J Am Coll Surg

Table 5. Mortality Rates for Splenectomy tion to immediate treatment for bleeding, there was con- Year Author Patients, n Mortality, % cern about the potential for delayed bleeding. Numer- 1880 Russell 28 100 ous publications on delayed splenic rupture were 1900 Bessel-Hagen 37 47 reported.134,140-144 In 1943, Zabinski and Harkins145 1907 Berger (collected series) 135 38.7 published a paper on this subject, and reports continued 1908 Johnston 108 40 through each succeeding decade. In 1956, Bollinger and 1908 Lotch 138 37 134 1909 Brositter (collected series) 203 35 Fowler collected 258 cases of splenic trauma from pre- 1919 Willis — 28.8 vious reviews and noted a 21.5% incidence of delayed 1926 Beer 90 31 rupture or bleeding. At least a dozen additional articles 1930 Dretzka 27 33 reporting several cases of delayed rupture were published 1943 Roettig 11 9 before the era of nonoperative management, many re- 1946 Pugh 15 6 porting patients who died. Interestingly, delayed bleeding from splenic injuries continues to be a problem re- periences of the American Expeditionary Force in the ported in recent literature reviews. In 1990, Farhat and 143 war of 1917 to 1919, was associated with a mortality rate colleagues reported delayed splenic rupture in 75 pa- of practically 100%.” This manual, written as a guide on tients with splenic injury. One of these patients died. trauma for physicians in the military, was apparently These authors reviewed more than 30 reports outlining intended to discourage splenectomy. Nonetheless, as cases of delayed splenic rupture. In 1994, Kluger and 144 with liver injury, the experiences of World War II coworkers presented 3 patients with delayed rupture, brought dramatic improvement to the treatment of all of whom were initially admitted in a stable condition. splenic injury. Pugh129 reported that various experiences These authors collected 24 cases from the recent literature from that war had lowered the mortality rate to between (1985 to 1992) as well. What seems clear in a literature 10% and 20% and he personally reported a death rate of review that spans from the 1930s to the end of the century only 6%. The one death in his series was from a head is an incidence rate of at least 1% to 2% of patients who injury. developed major delayed hemorrhage. Surgeons in the splenectomy era achieved remarkable Eras in the management of splenic injury results in the treatment of potentially fatal bleeding with Although the management of liver injury was divided extremely low mortality rates because of isolated splenic into several eras, the discussion on splenic injury will be injury. divided into two phases: the period in which splenectomy was the treatment for virtually all spleen injuries; Shift from emphasis on hemorrhage to and the era of splenic preservation. This latter treatment postsplenectomy infection phase has had several iterations. Recognition of the spleen’s role in the resistance to infection was known for most of the 20th century. In 1919, The era of splenectomy Morris and Bullock146 found an increased death rate in After World War II, numerous series of splenic injuries splenectomized rats injected with a strain of bacteria were published.130-139 Techniques for performing sple- causing rat plague. Several other studies in experimental nectomy for injury were relatively uniform, and results animals indicated the importance of the spleen in resist- were fairly comparable. Although mortality rates in ing various infections. The classic study that raised clin- these reports were often high, deaths were generally re- ical awareness of this problem was published in 1952 by lated to associated injury, and a patient’s demise from an King and Shumacker.147 isolated splenic injury was relatively uncommon. Al- Although this reference is widely quoted, its details though complication rates for general issues such as at- are almost never mentioned. The authors reported five electasis were high, specific technical misadventures cases of congenital hemolytic anemia treated by splenec- such as gastric fistulas or pancreatic injuries occurred tomy. Remarkably, two pairs of the five patients were infrequently. siblings who tragically died. These five infants had sple- Surgeons in these eras were most intent on avoidance nectomy at 4 weeks, 3 weeks, 2 weeks, 6 months, and of major morbidity or death from hemorrhage. In addi- 25 months of age, respectively. Singer148 reviewed 2,795 Vol. 200, No. 5, May 2005 Richardson Managing Liver and Spleen Injuries 659 asplenic patients collected from 23 series in addition to 6 Table 6. Postoperative Infection Rates after Splenectomy patients from the Texas Children’s Hospital. These 6 for Trauma* Overwhelming patients had splenectomy at 4, 5, 10, and 17 months, postsplenectomy and 2 and 3 years of age. The 2-year-old was a trauma Author Sepsis Mortality infection patient who died 3 years later of a Haemophilus influenza Schwartz 3.3 0 0 infection. Singer found 688 trauma patients, including Malangoni 0.25 0 0 388 children, with 4 deaths from sepsis. The death rate Sekikawa 0.57 0 0 in children was 0.58% and total death rate was 0.01%. Green 1.91 0.14 0.14 Eraklis and Filler149 reviewed 1,413 collected patients Cullingford 0.21 0.03 0.04 Total 0.59 0.03 0.04 and, not surprisingly, found similar results because there were considerable overlays in reports studied. But these *Incidence per 100 years of patient exposure. reports appeared to include only studies with reported infection and ignored those without reported sepsis. It is lowup appears to be available than in the United States. not clear whether infants undergoing splenectomy for In 1991, Cullingford and associates152 reported followup hematologic disorders behave similarly to older children on 1,490 patients undergoing splenectomy in Western with splenectomy for trauma and, even more uncertain, Australia, of which 628 were for trauma. In 3,922 whether these results on infection could be generalized person-years of exposure, 8 infections developed. Only to patients older than 55 years with injuries needing one had overwhelming postsplenectomy infection, and splenectomies. The issue of postsplenectomy infection only one death occurred. In 2001, a British group162 appears settled in children,147-154 but the data in adults reported a questionnaire study of microbiologists who are less clear. had data on overwhelming postsplenectomy infection. In 1969, Whitaker155 described an adult who devel- They found that 24 cases had occurred in trauma pa- oped infection postsplenectomy, although his spleen was tients undergoing splenectomy. The mortality rate was not removed for an injury. O’Neal and McDonald156 46%. This is similar to the 50% mortality rate collected noted 7 cases of fatal sepsis in 256 asplenic patients and by Zarrabi and Rosner165 in the 34 adult trauma patients calculated a mortality rate of 7.3 per 1,000 person-years reported in the world’s literature. Because these reports of followup. all contain tremendous overlap, it is difficult to ascertain Several reports have now documented the hazard of how many cases have been reported; it seems that with postsplenectomy infection in adults.157-162 One problem the addition of patients from Australia and England with reports on postsplenectomy sepsis is that they often there are less than 70 cases worldwide, with a death rate review the same patients, which leads to the conclusion of about 30%. that the problem occurs more frequently than actually Our unit obtained longterm followup data on 414 may be the case. Forty-five articles in the trauma litera- splenectomy patients in the late 1980s (unpublished ture devoted to this subject were reviewed, and there data), reflecting 2,167 patient-years. The major prob- appeared to be nearly as many articles as cases of sepsis in lems our patients faced in order of prevalence were se- asplenic adult trauma patients. quelae of the injury in which the splenic trauma oc- DiCataldo and colleagues163 from Italy reviewed the curred, alcoholism and drug dependence, and trauma world’s literature to 1987 and found 12 deaths from recidivism or other injuries occurring later. One patient overwhelming postsplenectomy sepsis (a rate of 0.66%). developed pneumonia that may have been from pneu- None of their personal 148 patients developed sepsis mococcus 6 years postinjury and was successfully treated. problems over a several-year followup. Luna and Del- An alcoholic patient suffered aspiration and polymicro- linger151 reviewed most of the same patients, and, not bial lung abscess. One patient with an intraventricular surprisingly, found 11 deaths. Pachter and associates164 shunt developed meninigitis believed to be related to mentioned three cases in a discussion of another paper problems with the shunt itself. with two deaths. Table 6 reviews the incidence of infec- Despite this low rate of postsplenectomy infection in tion problems in trauma patients for whom years of adults, it was one of several factors that were used as the followup are available. rationale for a shift from splenectomy as the primary The best reports are from countries where better fol- treatment of splenic injury to splenic conservation. Addi- 660 Richardson Managing Liver and Spleen Injuries J Am Coll Surg tionally, the mortality rate appears higher in patients under- the spleen, including wrapping the injured spleen.180-183 going splenectomy than in a normal population.166,167 As It is unclear what the penetrance of operative splenic with liver injuries, DPL led to celiotomy in many pa- salvage was into the care of trauma patients not treated tients with minimal injury where the spleen was not in trauma centers. Several studies show considerable severely injured. But the major driving force behind variations in the rate of attempts at splenic salvage splenic preservation was the observation by surgeons between trauma centers and nontrauma centers.95,184 caring for injured children that the spleen could be saved Nonetheless, operative splenic salvage was a concept that by operative means such as splenorrhaphy or through had a high rate of success in many centers. When non- avoidance of operation altogether.168-171 operative management came to the fore, splenorrhaphy and other forms of operative splenic salvage began to The era of splenic preservation decline, although they are still useful when operation is Efforts at splenic preservation could be divided into required. But most nonoperative failures are treated by three different areas: operative attempts at maintaining splenectomy. splenic function; embolization of the splenic artery and its branches; and nonoperative management of splenic Embolization of splenic artery hemorrhage injury. In 1973, embolization of the splenic artery was described to decrease the splenic mass in a patient with Operative attempts at splenic salvage hypersplenism. This procedure did not gain widespread With the increased awareness of the immunologic im- popularity because of reports of splenic rupture and ab- portance of the spleen, efforts at operative splenic salvage scess.185 In 1984, the transplant group186 at the Univer- began to appear in the 1980s, led by pediatric surgeons. sity of Illinois reported a prospective randomized trial of Splenic autotransplantation was advocated by several splenectomy versus partial splenic embolism to decrease groups, but, eventually, placement of these implants was the functional splenic mass. Several additional publica- shown to be ineffective.171-175 Splenorrhaphy was de- tions on the use of splenic artery embolization to described in children and, within a few years, several series crease splenic function and size were subsequently re- of splenic salvage by suturing the spleen had been reported. In 1995, Scalfani and coauthors187 introduced ported. Overall success rates in children were reported to the concept of embolization of splenic artery injuries, be high. With the exception of hilar injuries, most and 150 patients with all grades of splenic injuries un- splenic lacerations were amenable to repair. Initially, it derwent diagnostic arteriography on admission. Ninety was believed that properties of the child’s spleen made patients had negative angiograms and were observed this possible, but that suturing an adult’s spleen was not only; 60 had embolization of splenic vascular lesions. feasible. But within a few years, splenorrhaphy was being The total salvage rate was reported to be 98.5%, which is practiced in adults with a reasonable success rate. On an the highest success rate reported in the literature. interesting historic note, William Mayo176 described a The technique of embolization has also been a matter patient managed by splenorrhaphy in 1906. Feliciano of debate: ie, main artery coil, which may render the and colleagues177 and Pachter and associates164 reported entire spleen ischemic and obviate the value of “preserv- extensive experiences, with rebleeding rates of 1.5% and ing” the spleen versus distal embolization for active 1.8%, respectively. Most articles on splenorrhaphy were bleeding areas. A problem with this latter approach is the positive, but Beal and Spisso178 mentioned rebleeding as potential of rebleeding, because the vessels may be in a problem and noted the higher risks of blood transfu- spasm at the time of the initial angiogram. sions with splenorrhaphy. A study from Memphis demonstrated that vascular Some splenic injuries were devitalizing or shattered lesions identified by a repeat CT after resuscitation were the lower pole such that splenorrhaphy was not feasible. not present on an admission CT scan. Concern also Partial resection of the spleen with oversewing of the existed about embolization of multiple arteries that capsule was practiced with reasonable success in experi- could affect most of the functioning spleen. enced hands. An upper pole artery is present in the ma- Several studies on splenic embolization for trauma are jority of spleens, which facilitates lower pole resection.179 now present in the literature, including two published Multiple ingenious methods were devised to tamponade within the past year.188-192 Haan and associates188 pre- Vol. 200, No. 5, May 2005 Richardson Managing Liver and Spleen Injuries 661 sented data from four institutions that performed The success rate of NOM appears in adults to vary splenic embolization from 1997 to 2002. Indications for widely from 60% to 98%.199-217 The multiinstitutional tri- embolization were based on CT findings and included als committee of the Eastern Association for the Surgery of significant hemoperitoneum (outside the perisplenic Trauma(EAST) examined the results of treatment of 1,488 area), contrast extravasation, splenic artery pseudoaneu- adults contributed by 27 trauma centers.200 Grade of rysm, and arteriovenous fistula, with failure rates of splenic injury, degree of hemoperitoneum, and presence 10%, 17%, 12%, and 40%, respectively, depending on of associated injury were important determinants of out- the injury treated. The overall mortality rate was 5%, comes of NOM. A failure rate of 11% occurred in those although none appeared related to the injured spleen. managed nonoperatively. A second study by that group The failure rate was 13.5% and complications were nu- showed that patients older than age 55 had a higher rate merous (20%), including hemorrhage (13%), missed of failure of NOM than those younger. The mortality injuries (3%), and infection (4%). Fourteen patients un- rate was increased for those with successful NOM (8% derwent repeat angiography, and six developed splenic versus 4%) in patients older than 55 years versus those abscesses. younger than 55 years, which is not surprising. The Thirteen of 168 patients underwent splenic emboli- death rate in NOM failures was 29% versus 12% in zation in a series from San Antonio.193 Twelve were those over age 55 and in the less than age 55 groups, deemed a success. No mention was made of complica- respectively.218 Another recent study has also confirmed tions or outcomes other than splenic salvage. These an increased mortality rate in older patients in whom studies augment data from Memphis, where 26 patients NOM is not successful. out of a population of 526 with splenic injury had suc- Several predictors of failure of NOM have been exam- cessful embolization. These authors reported a high rate ined, although there are no universally accepted recom- of success, and no mention was made of complications. mendations for management of patients who might be at The total mortality rate for all splenic injured patients higher risk for failure. In 1995, the Memphis group219 was 10.5%, but included no deaths in those treated by noted a “contrast blush” in two-thirds of the patients with angiography. failed operative management. Several subsequent reports Although splenic embolization has been espoused to demonstrated vascular lesions within the spleen in patients preserve the spleen, no studies document its effect on the with failed NOM.220-222 These studies emphasize the im- organ’s immunologic function. Because the perfusion of portance of the CT scan in detection of these abnormali- the sinusoids of the spleen are driven by arterial pressure, ties. Several issues with CT scan reliability have been raised: it is unclear if thrombosis of the artery will alter normal Do early CT scans accurately depict vascular lesions or splenic function. should a second scan be done after resuscitation? Issues of interrater reliability for detecting these lesions have been Nonoperative management of splenic injury raised as well.223 It does seem likely these abnormalities Nonoperative management (NOM) was initially prac- should heighten awareness of potential failure of NOM. ticed in children with splenic injury with excellent re- Table 7 shows the results of NOM in several studies. sults. Virtually all studies in children have been positive, Success rates vary from 52% to 98% even though the and NOM clearly is the treatment of choice in this pop- criteria for inclusion were relatively uniform. With pas- ulation. As in hepatic injuries, the progress in children sage of time, surgeons appear to be accepting this treat- paved the way for trials in adults. The criteria for NOM ment strategy with greater frequency for patients with a may vary somewhat among institutions but generally higher degree of hemoperitoneum and a higher grade of include hemodynamic stability and lack of evidence of splenic injuries. Some now take patients in unstable con- visceral injuries.194,195 Initially, it was believed that patients dition to the angiogram suite instead of the operating older than 55 years should be treated operatively,196 but room. It is interesting that most articles published on some subsequent studies refute that assertion.197,198 Ab- splenic injuries focused on the ability to successfully sence of head injury was initially considered a contrain- avoid operation as the optimal end point. Several recent dication of NOM, but that criterion is not used in many publications made no mention of the mortality or mor- centers. Some centers are more likely to operate on bidity incurred with operative or nonoperative treat- higher-grade injuries or those with a vascular blush. ment, or the fate of the spleen after NOM failure. 662 Richardson Managing Liver and Spleen Injuries J Am Coll Surg

Table 7. Outcomes of Nonoperative Management of Splenic Injuries in Selected Series % Immediate % Successful nonoperative Author Year No. of patients operation management Cogbill* 1989 832 87 88 Shackford† 1990 1,254 — 69 Pachter 1990 193 87 96 Schurr 1995 309 58 87 Smith 1996 173 58 97 Gooley 1996 46 — 52 Davis 1998 524 39 94 Pachter 1998 190 47 98 Konstantakos 1999 267 33 83 Cocanour 1999 461 42 86 Myers 2000 233 31 94 Nix 2001 542 25 92 EAST Group‡ 2001 1,488 39 89 Dent 2004 168 17 98 *Experience of six centers. †Collected series from literature. ‡Experience of 27 centers.

The time of failure of NOM was reviewed in several rates remain at 6% to 7% or higher in many series be- studies where such data were available. The EAST study cause of the presence of associated injuries. On the other confirmed that nonoperative management that failed hand, the mortality rate of isolated splenic injuries has usually did so within 96 hours, but it failed in 7 patients effectively been at 0% for nearly 40 years (Fig. 4).224-229 It after day 9.200 Cogbill and colleagues195 noted failure of is difficult to find any report where more than an occa- NOM at days 6, 9, 10, 13, 19, and 36. A report from sional patient died of treatment of the splenic injury Cleveland showed that NOM failed in 31 adults at an itself. This creates even greater pressure on surgeons to average of 71 hours, but 1 patient suffered failure at ensure that patients not die in their attempt to finesse a 20 days.180 A review of 14 series, in which data on time of salvageable splenic injury by NOM, embolization, or failure were present, disclosed at least 30 patients suf- any treatment other than splenectomy. fered NOM failure after Day 7. One was smoking a Although numerous reports of deaths associated with cigarette outside the hospital on Day 8 when major NOM in adults assert that the deaths are not related to bleeding occurred!211 the splenic injury itself, several list multiple system organ There may be a natural reluctance to report untoward events associated with failure of nonoperative management. Our unit has had at least 4 deaths from delayed bleeding in NOM patients over a 10-year period (unpublished data). Two older patients died in the ICU on Day 6 and Day 10; 1 died on Day 9 after transfer to the orthopaedic service, and 1 died on Day 16 at Walmart. We had another patient with an anoxic brain injury after an arrest from splenic bleeding. None of these had a drifting decline in hemoglobin but, rather, catastrophic bleeding reminiscent of reports on delayed rupture from an earlier era.

Mortality and morbidity of splenic trauma Figure 4. Total deaths in patients with splenic injury is usually Unlike liver trauma, in which the mortality rate has de- about 6% to 7%. *Mortality from the splenic injury itself has been clined considerably in the past 20 years, total mortality less than 1% for more than 50 years. Vol. 200, No. 5, May 2005 Richardson Managing Liver and Spleen Injuries 663 failure as a frequent cause of death. In such patients, the splenic injuries has not changed for 40 years. By the additive effect of hemorrhage from the spleen is difficult 1960s, deaths from isolated spleen injuries approached to assess. A few patients on our unit have developed zero in large collected series, and a review of numerous multiple system organ failure without obvious cause. It series demonstrated a mortality rate of less than 1% for remains unclear whether or not the occult blood loss the past several decades. The total mortality rate varied from the spleen could have been a cofactor in their among series, depending on the severity of associated deaths. injuries, but was 13.8% in the EAST study that reflected Patients able to be successfully treated by NOM in- the experience of 27 trauma centers. Because relatively variably have decreased intensive care unit (ICU) stay, few patients die of isolated splenic injury after reaching decreased total length of hospital stay, and decreased the hospital alive, we must by careful that our manage- blood usage when compared with patients who have ment does not imperil the patient. operations. On the other hand, the operative group usu- I admit a certain dismay over the current manage- ally had a higher injury severity score and higher grade of ment of splenic injuries in some centers. I believe that splenic injury. the balance between concerns with bleeding and infection has shifted illogically to favor infection. Splenic DISCUSSION preservation has been granted a position of “political Progress in outcomes of hepatic injuries in the last 20 correctness” that must be balanced against the fact that years has been dramatic. Major improvements in mor- occasionally a shattered spleen must be removed. In fact, tality rates appear to be related to declining death rates many patients are still receiving splenectomy, but the from hemorrhage. The strategies of perihepatic packing, emphasis on NOM as a laudable end point in and of better management of major juxtahepatic venous inju- itself is worrisome to me. When I hear our residents ries, use of angiographic embolization for hepatic artery apologize for removing a bleeding spleen and saving a bleeding, and perhaps even nonoperative management life, I become concerned. We must not appear to be itself appear to have improved the mortality rate. Sur- cavalier about patients with high-grade splenic injuries geons caring for liver injuries must be prepared to use a or a large hemoperitoneum. The manner in which variety of operative and interventional maneuvers to NOM success rates are reported is also deceiving. Nu- treat those hepatic injuries that require more than non- merous reports of 95% success made little mention of operative management. The decision not to operate may the 17% to 45% of patients who must have urgent sple- be fraught with uncertainly not only for issues involving nectomy. Several series also immediately discount the the liver but because of concerns over a missed visceral deaths from their computation. injury. Surgeons must not, in my opinion, adopt a I am also puzzled about what is termed “failure of mindset in which an operation or other interventional nonoperative management.” When NOM fails at Day procedure is viewed as a defeat. 14 or Day 29, as has frequently been reported, what Surgeons in World War II learned that drainage of bile happens to the patient? The literature implies that pais very important, which is a lesson that is often forgot- tients are in a situation in which they can promptly ten by those providing nonoperative treatment of liver receive a splenectomy. Surely, some of these patients, injuries today. Patients who have major liver injuries who are many days from injury, must be placed at risk with a great deal of bile and old blood in the abdomen from these “failures.” It is incumbent on those who re- need to have them removed. If there is a major fluid port on these experiences to provide more data on the collection in the right gutter and pelvis, as is often the potential risk of offering treatment other than operation. case, percutaneous drainage will not remove this fluid. Much more information is needed on postsplenec- Our unit has advocated the use of laparoscopy a few days tomy sepsis in adults. Large studies from multiple postinjury to accomplish these goals. The results have trauma centers need to be organized, using the multiin- been excellent, and we certainly recommend this proce- stitutional study committees of American Association dure to surgeons as a useful adjunct to nonoperative for the Surgery of Trauma, EAST, or the Western management. Trauma Association. Those of us who care for adult The treatment of splenic injuries presents a different trauma patients with splenic rupture should cease refer- set of issues. Unlike liver injuries, the mortality rate from ring to the studies by King and Shumacker and Singer as 664 Richardson Managing Liver and Spleen Injuries J Am Coll Surg rationale for avoidance of splenectomy. We should vitally needed data on several issues, it is time to pay organize studies to determine the incidence of infections tribute to those caring for the injured and the remark- in posttraumatic asplenia. Equally importantly, data are able strides that have been made in treating liver and needed on delayed bleeding. There appear to have been spleen injuries. Not only has the care for patients greatly more “failures” of NOM occurring after Day 7 reported improved, but our organizations have served us well. in the past decade than total cases of overwhelming The learned societies in trauma have organized studies to postsplenectomy infection ever reported in the world’s elucidate solutions to problem areas and have promul- literature. We need reassurance that these “failures” are gated grading systems for liver and spleen injuries that not being harmed. Anecdotally, I have been impressed in are now routinely applied. Most importantly, our Amer- private discussions about deaths or “near misses” from ican College of Surgeons has provided leadership in the bleeding occurring in NOM failures. These are rarely care of the injured. Through the auspices of the Com- reported in the literature. Additionally, many reports list mittee on Trauma, we have organized trauma care, veri- multiple organ failure as a leading cause of death. Does fied centers for provisions of care to the most critically unrecognized shock play a role in these deaths? injured, and provided a wealth of educational opportu- The concerns I have about splenic preservation apply nities since the founding of the College, for all surgeons even more strongly to embolization of the spleen. Sev- who care for the injured. The improvements made in eral algorithms recommend taking patients in unstable treatment of the solid intraperitoneal organs should condition to angiography suites rather than to the oper- stand as one of our best achievements. ating room. A recent positive report on splenic embolization showed a 13.5% failure rate and a 20% complication rate, including 6 patients with splenic abscess. Fourteen REFERENCES patients required repeat angiography to achieve these re- 1. Papen GW, Mikal S. Liver trauma. Review Gastroenter 1950; 17:633–642. sults. Quite frankly, if a surgical series reported these 2. Shedden WM, Johnston F.Traumarupture of the liver. N Engl results for splenic operation, the surgeon would un- J Med 1935;213:961–965. doubtedly lose his operative privileges! We now have 3. Frantz VK, Clarke HT, Lattes R. Hemostasis with absorbable reports of unstable patients being taken to angiography gauze (oxidized cellulose). Ann Surg 1943;118:116–126. 4. Martin JD. Wounds of the liver. Ann Surg 1947;125:756–767. instead of to the operating suite, and then, when they 5. Boone DC, Federle M, Billiar TC, et al. Evolution of manage- develop a splenic abscess, there is an attempt at percuta- ment of major hepatic trauma: identification of patterns of neous drainage of that process. The logic involved in injury. J Trauma 1995;39:344–350. 6. Pilcher LS. Massive rupture of the liver. Ann Surg 1942;116: some of these decisions might make sense if it was being 827–832. advocated by groups other than surgeons. 7. Babcock WW. Temporary occlusion of the portal vein and My real concern is not what highly skilled trauma hepatic artery. Ann Surg 1942;116:853–856. 8. Krieg EC. Hepatic trauma. Arch Surg 1936;32:902–914. surgeons and angiographers do in Level I trauma centers 9. Robertson DE, Graham RR. Rupture of the liver without tear to try to push the envelope of splenic preservation fur- of the capsule. Ann Surg 1933;98:899–916. ther and further. My apprehension is the mindset that it 10. Branch CD. Injury of the liver. Ann Surg 1938;107:475–477. creates in those who care for ruptured spleens in venues 11. Christopher F. Rupture of the liver. Ann Surg 1936;103:461– 468. other than Level I trauma centers and the uncertain 12. Sparkman RS, Fogelman MJ. Wounds of the liver. Ann Surg message it sends to residents. Those of us who train 1954;138:690–703. residents in the care of trauma patients must ensure 13. Amerson JR, Blair HD. Traumatic liver injuries. Am Surg 1959;25:648–653. those residents can safely remove a spleen after they are 14. Crosthwait RW, Allen JE, Murga F,et al. The surgical manage- in practice. Arresting splenic hemorrhage can be a life- ment of 640 consecutive liver injuries in civilian practice. Surg saving procedure, and surgeons must not abrogate that Gynecol Obstet 1962;115:650–654. 15. Wang CC, Robbins LL. Roentgenologic diagnosis of ruptured responsibility to interventionalists who, almost cer- spleen. N Engl J Med 1956;254:445–449. tainly, will not be available in all the hospitals in which a 16. Root HD, Hauser LW, McKinley CR, et al. Diagnostic perito- splenectomy is urgently needed. It should also be noted neal lavage. Surgery 1965;57:633–639. that the efficacy of this procedure in preserving immu- 17. Meredith JW, Trunkey DD. CT scanning in acute abdominal injuries. Surg Clin North Am 1988;68:255–268. nologic function remains unproved. 18. Federle MP, Jeffrey RB Jr. Hemoperitoneum studied by com- Having challenged our trauma community to acquire puted tomography. Radiology 1983;148:187–192. Vol. 200, No. 5, May 2005 Richardson Managing Liver and Spleen Injuries 665

19. Federle MP, Goldberg HI, Kaiser JA, et al. Evaluation of ab- enous pack for control of hemorrhage in major injuries of the dominal trauma by computed tomography. Radiology 1981; liver. Surg Gynecol Obstet 1975;141:92–94. 138:637–644. 45. Lucas CE, Ledgerwood AM. Liver necrosis following hepatic 20. Toombs BD, Lester RC, Ben-Menachem Y, et al. Computed artery transection due to trauma. Arch Surg 1978;113:1107. tomography in blunt trauma. Rad Clin North Am 1981;19: 46. Clagett GP,Olsen WR. Non-mechanical hemorrhage in severe 17–35. liver injury. Ann Surg 1978;187:369–374. 21. Gillette WJ. Surgery of liver. Surg Gynecol Obstet 1905;1: 47. Cogbill TH, Moore EE, Jurkovich GJ, et al. Severe hepatic 361–364. trauma: a multi-institutional experience with liver injuries. 22. Boljarski N. Uber Leberverletzungen in Klinischer und experi- J Trauma 1988;28:1433–1438. mentellar hinsicht, unter besonder Berucksichtigung der Isoli- 48. Beal SL. Fatal hepatic hemorrhage: an unresolved problem of etern Netzplastik. Arch f Klin Chir 1910;93:507–547. complex liver injuries. J Trauma 1990;30:163–169. 23. Grey EG. Fibrin as a hemostatic agent in cerebral surgery. Surg 49. Fabian TC, Croce MA, Stranford GG, et al. Factors affecting Gyn Obstet 1915;21:452–454. morbidity from hepatic trauma. Ann Surg 1981;213:540–548. 24. Bailey OT, Ingraham FD. Human fibrin foam with thrombin 50. Ciresi KF, Lim RC Jr. Hepatic vein and retrohepatic vena cava as hemostatic agent in general surgery: experimental study and injury. World J Surg 1990;14:472–477. clinical use. Surg 1945;18:347–369. 51. Rovito PF. Atrial caval shunting in blunt hepatic vascular in- 25. Schroeder WE.The progress of liver hemastosis – report of uses jury. Ann Surg 1987;205:318–321. (resections, sutures, etc). Surg Gynecol Obstet 1906;2:52–61. 52. Burch JM, Feliciano DV, Mattox KL. The atriocaval shunt. 26. Pringle JH. Notes on the arrest of hemorrhage due to trauma. Ann Surg 1988;207:555–568. Ann Surg 1908;48:546–566. 53. Evans S, Jackson RJ, Smith SD. Successful repair of major 27. Madding GF,Lawrence KR, Kennedy PA. War wounds of liver. retrohepatic vascular injuries without the use of shunt or ster- Bull US Army Med Dept 1946;5:579–589. notomy. J Pediatr Surg 1993;28:317–320. 28. Cohn R. The treatment of traumatic injuries of the liver. Stan- 54. Bethea MC. A simplified approach to hepatic vein injuries. ford Med Bull 1947;5:120–124. Surg Gynecol Obstet 1977;145:78–80. 29. Wright LT, Prigot A, Hill LM Jr. Traumatic rupture of the liver 55. Coln D, Crighton J, Schorn L. Successful management of he- without penetrating wounds. Arch Surg 1947;54:613–632. patic vein injury from blunt trauma in children. Amer J Surg 30. McAllister AJ, Hicken NF, Clark P. Practical problems in the 1980;140:858–864. 56. Pachter HL, Spencer FC, Hoffstetter SR. The management of treatment of hepatic trauma. Amer Surg 1961;27:529–532. juxtahepatic venous injuries without an atriocaval shunt: pre- 31. McInnis WD, Richardson JD, Aust JB. Hepatic trauma 1977; liminary clinical observations. Surgery 1986;99:569–575. 112:157–161. 57. Chen RJ, Fang JF, Lin BC, et al. Surgical management of 32. Hellstrom G. Closed injury of the liver. Acta Chir Scand 1961; juxtahepatic venous injury. J Trauma 1995;30:886–890. 122:490–501. 58. Buechter KJ, Sereda D, Gomez G, et al. Retrohepatic vein 33. Pachter HL, Spencer FC, Hoffstetter SR, et al. Experiences injuries: experience with 20 cases. J Trauma 1989;29:1698– with the finger fracture technique to achieve intra-hepatic he- 1704. mostasis in 75 patients with severe liver injuries. Ann Surg 59. Carrillo EH, Spain DA, Miller F, et al. Intrahepatic vascular 1983;197:771–778. clamping in complex hepatic vein injuries. J Trauma 1997;43: 34. Lucas CE, Ledgerwood AM. Prospective evaluation of hemo- 131–133. static techniques for liver injuries. JTrauma1976;16:442–451. 60. Calne RY, McMaster P, Pentlow BD. The treatment of major 35. Feliciano DV, Mattox KL, Jordan GL Jr, et al. The manage- liver trauma by primary packing with transfer of the patient for ment of 1000 consecutive cases of hepatic trauma. Ann Surg definitive treatment. Brit J Surg 1979;66:338–339. 1988;204:438–495. 61. Feliciano DV, Mattox KL, Jordan GL Jr. Intra-abdominal 36. Trunkey DD, Shires GT, McClellan R. Management of liver packing for control of hepatic hemorrhage: a reappraisal. trauma in 811 consecutive patients. Ann Surg 1974;179:722– J Trauma 1981;21:285–290. 728. 62. Carmona RH, Peck D, Lim RC. The role of packing and 37. Aaron WS, Fulton RL, Mays ET. Selective ligation of the here-operation in severe hepatic trauma. J Trauma 1984;24:779– patic artery for trauma of the liver. Surg Gynecol Obstet 1975; 784. 141:187–189. 63. Svoboda JA, Peter ET,Dang CV.Severe liver trauma in the face 38. Mays ET, Conti S, Fallahzadkh H, et al. Hepatic artery liga- of coagulopathy: a case for packing and early re-exploration. tion. Surgery 1979;86:536–543. Amer J Surg 1982;144:717–721. 39. Mays ET. Hepatic trauma. Curr Prob Surg 1976;13:1–86. 64. Rotondo MF, Schwab CW, McGonigal MD, et al. “Damage 40. Flint LM, Polk HC. Selective hepatic artery ligation: limita- control”: an approach for improved survival in exsanguinating tions and failures. J Trauma 1979;19:319–323. penetrating injury. J Trauma 1993;35:375–380. 41. Schrock T, Blaisdell FW, Mathewson C Jr, et al. Management 65. Feliciano DV, Mattox KL, Burch JM, et al. Packing for control of blunt trauma to the liver and hepatic veins. Arch Surg 1968; of hepatic hemorrhage: 58 consecutive patients. J Trauma 96:698–704. 1986;26:738–743. 42. Moore FA, Moore EE, Seagraves A. Non-resectional manage- 66. Cue JI, Cryer HG, Miller FB, et al. Packing and planned re- ment of major hepatic trauma. Amer J Surg 1985;150:725– exploration for hepatic and retroperitoneal hemorrhage critical 729. refinements of a useful technique. J Trauma 1990;30:1007– 43. Walt AJ. The mythology of hepatic trauma or babel revisited. 1014. Amer J Surg 1978;135:1218. 67. Caruso DM, Battistella FD, Owings JT,et al. Perihepatic pack- 44. Stone HH, Lamb JM. Use of pedicled omentum as an autog- ing of major liver injuries. Arch Surg 1999;134:958–963. 666 Richardson Managing Liver and Spleen Injuries J Am Coll Surg

68. Pachter HL, Spencer FC, Hoffstetter JR, et al. Significant scaling: spleen and liver (1994 revision). J Trauma 1995; trends in treatment of hepatic trauma. Ann Surg 1992;216: 38:323–324. 492–502. 90. Knudson MM, Lim RC, Oakes DD, et al. Non-operative man- 69. Garrison JR, Richardson JD, Hilakos AS, et al. Predicting the agement of blunt liver injuries in adults: the need for continued need to pack early for severe injury-abdominal hemorrhage. surveillance. J Trauma 1990;30:1494–1500. J Trauma 1996;40:923–929. 91. Carrillo EH, Wohltmann C, Richardson JD, et al. Evolution in 70. Brasel KJ, DeLisle CM, Olson CT, et al. Trends in the man- the management of complex liver injuries. Curr Probl Surg agement of hepatic injury. Am J Surg 1987;174:674–677. 2001;38:1–60. 71. Ivatury RR, Nallathambi I, Gunduz Y, et al. Liver packing for 92. Hiatt JR, Harrier HD, Koenig BV, et al. Non-operative man- uncontrollable hemorrhage: a reappraisal. J Trauma 1986;26: agement of major blunt liver injury with hemo-peritoneum. 744–753. Arch Surg 1990;125:101–106. 72. Baracco-Gandolfo V, Vidarte O, Baracco-Miller V. Prolonged 93. Meyers AA, Crass RA, Lim RA, et al. Selective non-operative closed liver packing in severe hepatic trauma: experience with management of blunt liver injury using computed tomogra- 36 patients. J Trauma 1986;26:754–756. phy. Arch Surg 1985;120:550–554. 73. Krige JET, Bornman PC, Terblanche J. Therapeutic perihepatic 94. Sherman HF,Savage BA, Jones LM, et al. Non-operative man- packing in complex liver trauma. Br J Surg 1992;79:43–48. agement of blunt hepatic injury: safe at any grade. J Trauma 74. Morris JA, Eddy VA, Blinman TA, et al. The staged celiotomy 1994;37:616–621. for trauma: issues in unpacking and reconstruction. Ann Surg 95. Rutledge R, Hunt JP,Lentz CW, et al. A statewide population- 1993;217:576–586. based time-series analysis of the increasing frequency of non- 75. Scalfani SJA, Shaftan GW, McAuley J, et al. Interventional operative management of abdominal solid organ injury. Ann radiology in the management of hepatic trauma. J Trauma Surg 1995;222:311–326. 1984;24:256–262. 96. Hellins TE, Morse G, McNabney WK. Treatment of liver in- 76. Ciraudo DL, Luk S, Palter M, et al. Selective hepatic arterial juries at level I and II centers in a multi-institutional metropol- embolization for Grade IV and V blunt hepatic injuries: An itan trauma system. J Trauma 1997;42:1091–1096. extension of resuscitation in the non-operative management of 97. Meredith JW, Young JS, Bowling J, et al. Non-operative man- traumatic hepatic injuries. J Trauma 1998;45:353–359. agement of blunt hepatic trauma: the exception or the rule. 77. Velmahos GC, Demetriads D, Chattwan S, et al. Angiographic J Trauma 1994;36:529–535. embolization for arrest of bleeding after penetrating trauma to 98. Carrillo EH, Platz A, Miller FB, et al. Non-operative manage- the abdomen. Am J Surg 1999;178:367–373. ment of blunt hepatic trauma. Br J Surg 1998;85:461–468. 78. Wahl WL, Ahrns KS, Brandt MM, et al. The need for early 99. Croce MA, Fabian TC, Menke PG, et al. Non-operative man- angiographic embolization in blunt liver injuries. J Trauma agement of blunt hepatic trauma is the treatment of choice for 2002;52:1097–1101. hemodynamically stable patients. Ann Surg 1995;221:744– 79. Mohr AM, Lavery RF, Barone A, et al. Angiographic emboli- 755. zation for liver injuries: Low mortality, high morbidity. 100. Pachter HL, Knudson MM, Esrig B, et al. Status of non- J Trauma 2003;55:1077–1082. operative management of blunt hepatic injuries in 1995: A 80. Takayasu K, Moriyama N, Nuramajsu Y, et al. Gallbladder multi-center experience with 404 patients. J Trauma 1996;40: infarction after hepatic artery embolization. 1985;144:135– 31–38. 138. 101. Richardson JD, Franklin GA, Lukan JK, et al. Evolution in the 81. Carrillo EH, Richardson JD. The current management of he- management of hepatic trauma: a 25-year perspective. Ann patic trauma. Adv Surg 2001;35:39–59. Surg 2000;232:324–330. 82. Johnson JW, Graciac VH, Gupta R, et al. Hepatic angiography 102. Marx JA, Moore EE, Jordan RC, et al. Limitations of com- in patients undergoing damage control laparotomy. J Trauma puted tomography in the evaluation of acute abdominal 2002;52:1102–1106. trauma—prospective randomized study. J Trauma 1985;25: 83. Denton JR, Moore EE, Coldwell DM. Multimodality treat- 933–938. ment for grade V hepatic injuries: perihepatic packing, arterial 103. Buckman RF, Piano G, Dunham CM, et al. Major bowel and embolization and venous stenting. J Trauma 1997;42:964– diaphragmatic injuries associated with blunt spleen or liver 967. rupture. J Trauma 1988;28:1317–1321. 84. DeToma G, Mingoli A, Modini C, et al. The value of angiog- 104. Fischer RP, Miller-Crotchet P, Reed RL. The hazards of non- raphy and selective hepatic artery embolization for continuous operative management of adults with blunt abdominal injury. bleeding after surgery in liver trauma: case reports. J Trauma J Trauma 1988;28:1445–1449. 1994;37:508–511. 105. Kemmeter PR, Hoedema RE, Foote JA, et al. Concomitant 85. Franklin RH, Bloom WI, Schoffstal RO. Angiographic embo- blunt enteric injuries with injuries of the liver and spleen: a lization as a definite treatment post-traumatic hemobilia. dilemma for trauma surgeons. Am Surg 2001;267:221–226. J Trauma 1980;20:702–705. 106. Sherk JP, Oakes DD. Intestinal injuries missed by computed 86. Hinton J. Injuries to abdominal viscera: their relative fre- tomography. J Trauma 1990;30:1–7. quency and their management. Ann Surg 1926;90:351–356. 107. Carrillo EH, Richardson JD. Delayed surgery and interven- 87. Richie JP,Fonkalsrud EN. Subcapsular hematoma of the liver. tional procedures in complex liver injuries. J Trauma 1999;46: Arch Surg 1972;104:781–784. 978–979. 88. Karp MP, Cooney DR, Pros GA, et al. The non-operative 108. Carrillo EH, Spain DA, Wolhtmann CD, et al. Interventional management of pediatric hepatic trauma. J Pediatr Surg 1983; techniques are useful adjuncts in nonoperative management of 18:512–518. hepatic injuries. J Trauma 1999;46:619–624. 89. Moore EE, Cogbill TH, Jurkovich GJ, et al. Organ injury 109. Ackerman NB, Sillin LF,Suresh K. Consequences of intraperi- Vol. 200, No. 5, May 2005 Richardson Managing Liver and Spleen Injuries 667

toneal bile: bile ascites versus bile peritonitis. Am J Surg special reference to delayed splenic rupture. Am J Surg 1985;149:244–248. 1956;91:561–570. 110. Allins A, Ho T, Nguyen TH, et al. Limited value of routine 135. Parsons L,Thompson JE.Traumaticrupture of the spleen from follow-up CT scans in non-operative management of blunt non-penetrating injuries Ann Surg 1958;147:214–220. liver and splenic injury. Am Surg 1996;62:883–886. 136. Economy D, Koucky C, Novack RL. Non-penetrating injuries 111. Demetriades D, Gomez H, Chahwan S, et al. Gunshot wounds to the spleen. Am J Surg 1960;99:646–650. to the liver: the role of selective nonoperative management. 137. Iung OS, Cammack KV, Dodds M, et al. Traumatic rupture of J Am Coll Surg 1998;188:343–348. the spleen. Am J Surg 1961;101:357–365. 112. Moore EG. When is nonoperative management of a gunshot 138. Knopp LM, Harkins HN. Traumatic rupture of the normal wound to the liver appropriate (editorial)? J Am Coll Surg spleen. Surgery 1954;35:493–500. 1998;188:427–428. 139. Dennehy T, Lamphier TA, Wickman W, et al. Traumatic rup- 113. Renz BM, Feliciano DV. Gunshot wounds to the right thora- ture of the normal spleen. Am J Surg 1961;102:58–64. coabdomen: A retrospective study of non-operative manage- 140. Fultz CT, Altemeier WA. Delayed rupture of the spleen after ment. J Trauma 1994;37:732–740. trauma. Surg 1955;38:414–422. 114. Edler L. Die traumatischen verletzungen der parenchymatasen 141. Bailey HA, Schreiber SL. Delayed rupture of the spleen. Am J Unterleibsergane. Arch f Klin Chir 1987;34:343–356. Surg 1944;66:4–14. 115. Thole FWH. Verletzungen der Leber und der Gallenwege. 142. Fey DW, Turaw IL. Traumatic rupture of normal spleen with Neve Deutsch Chirurgie 1912;4:218. delayed rupture. Am J Surg 1941;53:363–366. 116. Tilton BJ. Consideration regarding wounds of the liver. Ann 143. Farhat GA, Abdu RA, Vanek VW.Delayed splenic rupture: real Surg 1905;98:27–41. or imaginary? Am Surg 1992;58:340–345. 117. Hitzrot JM. Subcutaneous injuries of the liver. Ann Surg 1917; 144. Kluger Y, Paul DB, Raves JJ, et al. Delayed rupture of the 60:50–55. spleen–-myths, facts, and their importance. Case report and 118. Fraenkel F.Subcutaneous rupture of the liver – management by literature review. J Trauma 1994;36:568–571. primary laparotomy. Beitr Z Klin Chir 1901;30:418–499. 145. Zabinski EJ, Harkins HH. Delayed splenic rupture–-a clinical 119. Thorlakson PHT, Hay AWS. Rupture of the liver. Canad Med syndrome following trauma. Arch Surg 1923;76:186–213. J 1929;20:593–595. 146. Morris DH, Bullock FD. The importance of the spleen in the 120. Knudson MM, Lim RC, Olcott CW. Morbidity and mortality resistance to infection. Ann Surg 1919;70:153–158. 147. King H, Shumacker HB Jr. Splenic studies: susceptibility to following major penetrating liver injuries. Arch Surg 1994; infection after splenectomy performed in infancy. Ann Surg 129:256–261. 1952;136:239–242. 121. Federico JA, Horner WR, Clark DG, et al. Blunt hepatic 148. Singer DB. Post-splenectomy sepsis. Perspectives in Ped Path trauma. Arch Surg 1990;125:905–909. 1973;1:285–311. 122. Chen RJ, Frang JF, Lin BC, et al. Factors that influence the 149. Eraklis AJ, Filler RM. Splenectomy in childhood: a review of operative mortality after blunt hepatic injuries. Eur J Surg 1413 cases. J Pediatr Surg 1972;4:382–388. 1995;161:811–817. 150. Horan M, Colebatch JH. Relation between splenectomy and 123. Sanders GB, Macguire CH. Massive rupture of the liver. Amer subsequent infection. Arch Dis Child 1962;37:398–414. J Surg 1949;78:699–704. 151. Luna GK, Dellinger EP.Nonoperative observations therapy for 124. Cachecho R, Clas D, Gerskin K, et al. Evolution in the man- splenic injury: a safe therapeutic option? Am J Surg 1987;153: agement of complex liver injury at a Level I trauma center. 462–468. J Trauma 1998;45:79–82. 152. Cullingford GL, Watkins DN, Watts AD, et al. Severe late 125. Pachter HL, Feliciano DV. Complex hepatic injury. Surg Clin post-splenectomy infection. Br J Surg 1991;78:716–721. North Am 1996;76:763–783. 153. Huntley CC. Infection following splenectomy in infants and 126. Cogbill TH, Moore EE, Jurkovich GJ, et al. Severe hepatic children. A review of the experience at Duke Hospital in in- trauma: a multi-center experience with 1335 liver injuries. fants and children during a 22-year period (1933–1954). Am J J Trauma 1988;28:1433–1438. Dis Child 1958;95:477–481. 127. Asensio JA, Demetriades D, Chahwan S, et al. Approach to the 154. Horn M, Colebatch JH. Relation between splenectomy and management of complex hepatic injuries. J Trauma 2000;48: subsequent infections: a clinical study. Arch Dis Child 1962; 66–69. 37:398–415. 128. Menegaux F, Langlois P,Chigot JP.Severe blunt trauma of the 155. Whitaker AN. Infection and the spleen: association between liver: study of mortality factors. J Trauma 1993;35:865–869. hyposplenism pneumonococcal infection, and disseminated 129. Pugh HL. Splenectomy with special reference to its historical intravascular coagulation. Med J Aust 1969;1:1213–1219. background. Internet Abstr Surg 1946;83:209–224. 156. O’Neal BJ, McDonald FC. The risk of sepsis in the asplenic 130. Byrne RV. Splenectomy for traumatic rupture with intra- adult. Ann Surg 1981;194:775–778. abdominal hemorrhage. Arch Surg 1950;61:263–268. 157. Green JB, Shackford SR, Sise MJ, et al. Late septic complica- 131. Averbook BD, Pearson S. Traumatic rupture of the spleen. Am tions in adults following splenectomy for trauma: a prospective Surg 1959;25:900–903. analysis in 144 patients. J Trauma 1986;26:999–1004. 132. Cloutier LC, Zaepfel FM. Traumatic rupture of the spleen. 158. Gopal V, Bisno AL. Fulminant pneumococcal infections in Surg Gynecol Obstet 1958;107:749–756. normal splenic hosts. Arch Intern Med 1977;137:1526–1530. 133. Terry JH, Self MM, Howard JM. Injuries of the spleen; report 159. Leonard AS. The overwhelming post-splenectomy sepsis prob- of the 102 patients and review of literature. Surgery 1956;40: lem. World J Surg 1980;4:423–432. 615–622. 160. Standage BA, Goss JC. Outcome and sepsis after splenectomy 134. Bollinger JA, Fowler EF.Traumatic rupture of the spleen, with in adults. Am J Surg 1982;143:545–548. 668 Richardson Managing Liver and Spleen Injuries J Am Coll Surg

161. Schwartz PE, Sterioff S, Mucha P,et al. Post-splenectomy sepsis ence with the use of absorbable mesh splenorrhaphy. J Trauma and mortality in adults. JAMA 1982;248:2279–2284. 1985;25:909–913. 162. Waghorn DJ. Overwhelming infection in asplenic patients: 184. Mooney DP, Forbes PW. Variation in the management of pe- current best practice preventive measures are not being fol- diatric splenic injuries in New England. J Trauma 2004;56: lowed. J Clin Path 2001;54:214–218. 328–333. 163. DiCataldo A, Puelo S, LiDestri G, et al. Splenic trauma and 185. Wholey MH, Chamorro HA, Rao G, et al. Splenic infection overwhelming post-splenectomy infection. Br J Surg 1987;74: and spontaneous rupture of the spleen after therapeutic embo- 343–345. lization. Cardiovasc Radiol 1978;1:249–253. 164. Pachter HL, Spencer FC, Hoffstetter SR, et al. Experiences 186. Mozes MF, Spigos DG, Pollak R, et al. Partial splenic emboli- with selective operative and nonoperative treatment of splenic zation, an alternative to splenectomy. Results of a prospective, injuries in 193 patients. Ann Surg 1989;211:583–591. randomized study. Surgery 1984;96:694–702. 165. Zarrabi NH, Rosner F. Serious infection in adults following 187. Scalfani SJA, Shaftan GW, Scalea TM, et al. Nonoperative splenectomy for trauma. Arch Intern Med 1984;144:1421– salvage of computed tomography-diagnosed splenic injuries: 1424. utilization of angiography for triage and embolization for he- 166. Linet MS, Nyren O, Gridley G, et al. Cases of death among mostasis. J Trauma 1995;38:818–825. patients surviving at least one year following splenectomy. 188. Haan JM, Biffl W, Knudson MM, et al. Splenic embolization Am J Surg 1996;172:320–323. revisited: a multicenter review. J Trauma 2004;56:542–547. 167. Pimpl E, Dapaunt O, Kaindl H, et al. Incidence of 189. Davis KA, Fabian TC, Croce MA, et al. Improved success in thromboembolic-related deaths after splenectomy in adults. Br J nonoperative management of blunt splenic injuries: emboliza- Surg 1989;76:517–521. tion of splenic artery pseudoaneurysms. J Trauma 1998;44: 168. Bond SJ, Eichelberger MR, Gotschall CS, et al. Nonoperative 1008–1015. management of blunt hepatic and splenic injury in children. 190. Haan JM, Scott J, Boyd-Kranis RL, et al. Admission angiogra- Ann Surg 1996;223:286–289. phy for blunt splenic injury: advantages and pitfalls. J Trauma 169. Pearl RH, Wesson PE, Spence LJ, et al. Splenic injury: a 5-year 2001;51:1161–1165. update with improved results and changes for conservative 191. Salis A, Pais O, Scalea T, et al. Superselective embolization of a management. J Ped Surg 1989;24:428–431. traumatic intra-splenic arteriovenous fistula. J Trauma 1999; 170. Powell M, Courcoulas A, Gardner M, et al. Management of 46:186–188. blunt splenic trauma: significant differences between adults 192. Hagiwara A, Yukioka M, Ohra S, et al. Nonsurgical manage- and children. Surgery 1997;122:654–660. ment of patients with blunt splenic injury: efficacy of trans- 171. Kakkasseril JB, Steward D, Cox JA, et al. Changing manage- catheter arterial embolization. Am J Roentgenol 1996;167: ment of pediatric splenic trauma. Arch Surg 1982;117:758– 151–166. 759. 193. Dent D, Alsabrook G, Erickson BA, et al. Blunt splenic injury: 172. Cooney DR, Swanson SE, Dearth JE, et al. Splenic auto- high nonoperative management rates can be achieved with transplantation in prevention of overwhelming post- selective embolization. J Trauma 2004;56:1063–1067. splenectomy infection. J Pediatric Surg 1979;14:336–342. 194. Longo WE, Baker CC, McMillen MA, et al. Nonoperative 173. Vega A, Howell C, Krasna I, et al. Splenic autotransplantation: management of adult blunt splenic trauma–-criteria for suc- optimal functional factors. J Pediatr Surg 1981;16:898–904. cessful outcome. Ann Surg 1989;211:626–629. 174. Moore GE, Stevens RE, Moore EE, et al. Failure of splenic 195. Cogbill TH, Moore EE, Jurkovich GJ, et al. Nonoperative implants to protect against fatal post-splenectomy infection. management of blunt splenic trauma: a multicenter experi- Am J Surg 1983;146:413–414. ence. J Trauma 1989;29:1312–1317. 175. Traub A, Giebnik GS, Smith C, et al. Splenic reticuloendothe- 196. Godley CD, Warren RL, Sheridan FL, et al. Nonoperative lial function after splenectomy, spleen repair and splenic auto- management of blunt splenic injury in adults: age over 55 years transplantation. N Engl J Med 1987;317:1559–1564. as a powerful predictor of failure. J Am Coll Surg 1996;183: 176. Mayo WJ. Principles underlying surgery of the spleen, with a 133–139. report of ten splenectomies. JAMA 1910;54:14–18. 197. Barone JE, Burns G, Svehlak SA, et al. Management of blunt 177. Feliciano DV, Bitondo CG, Mattox KL, et al. A four year splenic trauma in patients older than 55 years. J Trauma 1999; experience with splenectomy versus splenorrhaphy. Ann Surg 46:87–90. 1985;201:568–575. 198. Cocanour CS, Moore FA, Ware ON, et al. Age should not be a 178. Beal SL, Spisso JM.The risk of splenorrhaphy. Arch Surg 1988; determinant for nonoperative management of blunt splenic 123:1158–1163. injuries. J Trauma 2000;48:606–612. 179. Whitesell FB Jr. A clinical and surgical anatomic study of rup- 199. Bee TK, Croce MA, Miller PR, et al. Failure of splenic nonop- ture of the spleen due to blunt trauma. Surg Gynecol Obstet erative management: is the glass half empty or half full? 1960;110:750–754. J Trauma 2001;50:230–236. 180. Malangoni MA, Levine AW, Droege EA, et al. Management of 200. Peitzman AB, Heil B, Rivera L, et al. Blunt splenic injury in injury to the spleen in adults: results of early operation and adults: a multi-institutional study of Eastern Association for observation. Ann Surg 1984;200:702–705. the Surgery of Trauma. J Trauma 2000;49:177–189. 181. Delaney HM, Porreca F, Mitsudo S, et al. Splenic capping: an 201. Shackford SR, Molina M. Management of splenic injuries. experimental study of a new technique for splenorrhaphy using Surg Clin North Am 1990;70:595–620. woven polyglycolic acid mesh. Ann Surg 1982;196:187–193. 202. Jalovec LM, Boe BS, Wyffels PL, et al. The advantage of early 182. King DR, Lobe TE, Haase GM, et al. Selective management of operation with splenorrhaphy versus nonoperative manage- injured spleen. Surgery 1981;90:677–682. ment for the blunt splenic trauma patient. Am Surg 1993;59: 183. Delaney HM, Rudavsky AZ, Lan S. Preliminary clinical experi- 698–705. Vol. 200, No. 5, May 2005 Richardson Managing Liver and Spleen Injuries 669

203. Schweizer W,Bohlen L, Dennison A, et al. Prospective study in 217. Morgenstern L, Uyeda RY. Non-operative management of in- adults of splenic preservation after traumatic rupture. Br J Surg juries to the spleen in adults. Surg Gynecol Obstet 1983;157: 1992;79:1330–1333. 513–518. 204. Mucha P, Daly RC, Farnell MB. Selective management of 218. Harbrecht BG, Peitzman AB, Rivera L, et al. Contribution of blunt splenic trauma. J Trauma 1986;26:970–979. age and gender to outcome of blunt splenic injury in adults: 205. Malangoni MA, Cue JI, Fallat ME, et al. Evaluation of splenic multicenter study of the Eastern Association for the Surgery of injury by computed tomography and its impact on treatment. Trauma. J Trauma 2001;51:887–895. Ann Surg 1990;211:592–599. 219. Schurr MJ, Fabian TC, Gauant M, et al. Management of blunt 206. Brasel KL, DeLisle CM, Olson CJ, et al. Splenic injuries: splenic trauma: computed tomographic contrast blush predicts trends in evaluation and management. JTrauma1998;44:283– failure of nonoperative management. J Trauma 1995;39:507– 286. 513. 207. Pachter HL, Guth AA, Hofstetter SR, et al. Changing patterns 220. Federle MP, Courcoulas AP, Powell M, et al. Blunt splenic in the management of splenic trauma. Ann Surg 1998;227: injury in adults: clinical and CT criteria for management, with 708–719. emphasis on active extravasation. Radiology 1998;206:137– 208. Konstantakos AK, Barnoski AL, Plaisier BR, et al. Optimizing 142. the management of blunt splenic injury in adults and children. 221. Omert LA, Salyer D, Dunham CM, et al. Implications of the Surgery 1999;126:805–813. “contrast blush” finding on computed tomographic scan of the 209. Zucker KA. Nonoperative of adult blunt splenic trauma: cri- spleen in trauma. J Trauma 2001;51:272–278. teria for successful management. Ann Surg 1989;210:626– 222. Digiacoma JC, McGonigal MD, Haskal ZJ, et al. Arterial 629. bleeding diagnosed by CT in hemodynamically stable victims 210. Nix JA, Costanza M, Daley BJ, et al. Outcome of the current of blunt trauma. J Trauma 1996;40:249–252. management of splenic injuries. J Trauma 2001;50:835–842. 211. Smith SJ, Cooney RN, Mucha P. Nonoperative management 223. Barquist ES, Pizano LR, Feuer W, et al. Inter- and intra-rater of the ruptured spleen: a revalidation of criteria. Surgery 1996; reliability in computed axial tomographic grading of splenic 120:745–751. injury: why so many grading scales. J Trauma 2004;56:334– 212. Lucas CE. Splenic trauma: choice of management. Ann Surg 338. 1991;213:98–112. 224. Buxton JT. Traumatic rupture of the spleen. Am J Surg 1922; 213. Koury HI, Peschiera JL, Welling RE. Non-operative manage- 36:312–316. ment of blunt splenic trauma: a 10-year experience. Injury 225. Foster JM, Prey D. Rupture of the spleen: an analysis of twenty 1991;22:349–352. cases. Am J Surg 1940;47:487–492. 214. Cathy KL, Brady JS. Blunt splenic trauma characteristics of 226. Mansfield RD. Traumatic rupture of the normal spleen. Am J patients requiring urgent laparotomy. Am Surg 1998;64:450– Surg 1955;89:759–768. 454. 227. Donhauser JL, Locke DJ. Traumatic rupture of the spleen. 215. Sartorelli KH, Frumiento C, Rogers FB, et al. Nonoperative Arch Surg 1960;153:1013–1018. management of hepatic splenic and renal injuries in adults with 228. Foster JN Jr, Prey D. Rupture of the spleen. An analysis of multiple injuries. J Trauma 2000;49:56–62. twenty cases. Am J Surg 1940;47:487–492. 216. Myers JG, Dent DL, Stewart RM, et al. Blunt splenic injuries: 229. Nallathambi MN, Ivatury RR, Wapnir I, et al. Non-operative dedicated trauma surgeons can achieve a high rate of nonoperative management versus early operation for blunt splenic trauma in success in patients of all ages. J Trauma 2000;48:801–806. adults. Surg Gynecol Obstet 1988;166:252–258. ORIGINAL SCIENTIFIC ARTICLES

Impact of Work-Hour Restrictions on Residents’ Operative Volume on a Subspecialty Surgical Service

Ariel U Spencer, MD, Daniel H Teitelbaum, MD, FACS

BACKGROUND: Whether the 80 hours per week limit on surgical residents’ work hours has reduced the number or variety of cases performed by residents is unknown. STUDY DESIGN: We quantified residents’ operative experience, by case category, on a pediatric surgical service. The number of senior and junior residents’ cases were compared between residents from the year before (n ϭ 47) and after (n ϭ 44) the 80-hour limit. Residents also completed a questionnaire about their operative and educational experience. As an additional dimension of the educational experience, resident participation in clinic was assessed. Student’s t-test was used. RESULTS: Total number of cases performed either by senior (before, 1.58 Ϯ 0.42 versus after, 1.84 Ϯ 0.82 cases/day) or junior (before, 0.70 Ϯ 0.21 versus after, 0.71 Ϯ 0.15) residents has not changed (p ϭ NS). Senior residents’ vascular access and endoscopy rate increased; other categories remained stable. Residents’ perception of their experience was unchanged. But residents’ participation in outpatient clinic was significantly decreased (before, 66.0% Ϯ 14.7% versus after, 17.0% Ϯ 19.9% of clinics covered, p Ͻ 0.005). CONCLUSIONS: The 80-hour limit has had minimal impact on residents’ operative experience, in case number and variety, and residents’ perceptions of their educational experience. Residents’ reduction in duty hours may have been achieved at the expense of outpatient clinic experiences. (J Am Coll Surg 2005;200:670–676. © 2005 by the American College of Surgeons)

On July 1, 2003, a new set of regulations from the Ac- Although the effect of the new ACGME rules on creditation Council for Graduate Medical Education number of surgical procedures performed by residents is (ACGME) took effect. These regulations mandated a still unknown, data from New York provide insight into reduction in the number of hours that house officers the issue of work-hour limitations. In 1989, New York work, limiting surgical programs (indeed, all specialties) State legislated regulations quite similar to the ACGME’s to a maximum of 80 duty hours per week per resident, rules.4 Enforcement was initially lax, but the state of New averaged over a 4-week period.1 Regulations also stipu- York eventually did compel residency programs to meet late that in-house call cannot be more frequent than the specified work-hour limits.5 Hassett and colleagues6 every third night. The intent of these new regulations is reported the case volume of graduating chief residents at to improve educational quality of residency training2 a large academic New York general surgery program in and reduce risk of medical errors, ostensibly because of compliance with the 80-hour limit, and did not find a resident fatigue.3 Significant reorganization of residency reduction in total number of cases performed by those training has been necessary to comply with the new residents. Another report from New York indicated that regulations. chief residents’ total number of cases actually increased after the work-hour legislation was passed, although in- Competing interests declared: None. creased surgical volume in the hospital studied makes Abstract presented at the American College of Surgeons 90th Annual Clinical interpretation of these figures more difficult.7 In con- Congress, Surgical Forum, New Orleans, LA, October 2004. trast, an extensive study by Whang and colleagues,8 Received October 26, 2004; Revised December 28, 2004; Accepted January based on questionnaires administered to surgical resi- 11, 2005. From the Department of Surgery, Section of Pediatric Surgery, University of dents in New York, found that slightly more than 50% Michigan, and the CS Mott Children’s Hospital, Ann Arbor, MI. of these residents believed that the number of operations Correspondence address: Daniel H Teitelbaum, MD, FACS, F3970 Mott Children’s Hospital, University of Michigan Hospitals, 1500 E Medical Cen- in which they were participating was “somewhat fewer” ter Dr, Ann Arbor, MI 48109-0245. or “much fewer” because of work-hour limits. The New

© 2005 by the American College of Surgeons ISSN 1072-7515/05/$30.00 Published by Elsevier Inc. 670 doi:10.1016/j.jamcollsurg.2005.01.008 Vol. 200, No. 5, May 2005 Spencer and Teitelbaum Eighty Hours and Residents’ Surgical Experience 671

York experience has generated conflicting reports, dem- Table 1. Case Categories onstrating the need for further study of the duty-hour Category Representative examples limit issue. Although studies suggest that work-hour I Esophagoscopy with dilation 9-11 Flexible sigmoidoscopy limits may improve residents’ quality of life, very few Broviac catheter placement studies have examined the impact of these limits on sur- Percutaneous endoscopic gastrostomy gical residents’ operative experience. For this reason, we Bronchoscopy examined the effect of the ACGME regulations on the II Open appendectomy Inguinal or umbilical herniorrhaphy operative experience of surgical residents. We chose a Skin/soft tissue procedures such as cyst/lipoma excision closely controlled setting to eliminate potential con- Implant/foreign-body removal founding factors. Our presumptive hypothesis was that III Any procedure (other than appendectomy) involving reduction in residents’ working hours to 80 per week laparotomy, eg, total abdominal colectomy, open fundoplication would result in a quantifiable reduction in their total Any procedure involving thoracotomy operative experience and case variety. Any other major operation, eg, major tumor excision IV All minimally invasive procedures, eg, METHODS Laparoscopic cholecystectomy Thoracoscopic wedge resection of lung Setting Laparoscopic fundoplication We quantified the operative experience of all residents Thoracoscopic resection of mediastinal tumor rotating through our pediatric surgery section from July 1, 2002, through June 30, 2004. This service is a major rotation at an academic medical center for first-, second-, open surgical procedures. Major open procedures— third-, and fourth-year surgical residents. In our study, both abdominal and thoracic—were placed into Cate- we classified first- and second-year residents as “junior” gory III. Category IV included all minimally invasive residents, and third- and fourth-year as “senior” resi- procedures. dents. Each month, there were three junior residents and Actual number of cases each resident performed dur- one senior resident on service (exceptions are noted) and ing his rotation on the service was determined. To cor- two pediatric surgery fellows. The pediatric surgery fel- rect for the difference in number of days between lows’ operative experience was not included in the months, and to correct for occasional vacation leave, the analysis. number of days that a given resident was actually on service was used in the analysis. For each resident, total Study design number of operative cases per day (including all four The Institutional Review Board of the University of case categories) equals the number of cases performed by Michigan approved the study design and exempted the that resident divided by the number of days he was on study from continuing review. We compared the average the service. Similarly, within each case category, the operative experience of senior residents for the 12 con- number of cases in that category performed by each secutive months before July 1, 2003, with the average resident was divided by the number of days that the operative experience of senior residents for the 12 resident was on the service. Residents’ average experi- months subsequent to July 1, 2003, and, likewise, com- ence (cases per day) before implementation of the 80- pared the operative experience of junior residents for the hour week was then compared with their average expe- 12 months before and after that date, when the 80-hour rience after the 80-hour week (seniors before versus week took effect. All cases performed by attending sur- seniors after; juniors before versus juniors after). geons during these 24 months, regardless of resident To assess residents’ perceptions of their operative and participation, were recorded in a database according to educational experience, a 10-item questionnaire, with type of procedure and date of operation. Residents were each item graded on a Likert-type scale, was adminis- then recorded as participating in the case if their role was tered to residents after they had rotated on the service. surgeon junior, first assistant, or teaching assistant. Residents’ participation was voluntary and their privacy Cases were divided into four subcategories (Table 1). was protected by a confidential response system. All res- Category I included all endoscopic procedures and vas- idents (both before and after the 80-hour limit) received cular access procedures. Category II included simple, the same questionnaire (see Table 2). 672 Spencer and Teitelbaum Eighty Hours and Residents’ Surgical Experience J Am Coll Surg

In each comparison, two-tailed Student’s t-test was Residents’ operative experience used. Data shown are mean Ϯ SD. Total number of cases actually performed by residents was then stratified according to level of training (senior or junior). As shown in Figure 1, the total number of RESULTS cases per day performed by senior residents did not Residency and service characteristics change significantly (before, 1.58 Ϯ 0.42 versus after, During the 12 consecutive months before July 2003, a 1.84 Ϯ 0.82 total cases per day, p ϭ 0.37). Likewise, the total of 47 residents rotated on this service (33 men and number of cases per day for junior residents, before the 14 women). There were 3 juniors and 1 senior on service work-hour limit, was virtually identical to their case each month, with the exception of 1 month when there number per day after the limit (before, 0.70 Ϯ 0.21 Ϯ ϭ were only 3 juniors (no senior on service). For the 12 versus after, 0.71 0.15, p 0.91). These data suggest months beginning July 1, 2003, a total of 44 residents that the 80-hour limit has had essentially no impact on rotated on the service (35 men and 9 women). As in the total number of cases performed by residents. preceding year, there were 3 juniors and 1 senior on To assess whether the work-hour limit may have di- service each month, with the notable exception of 4 out minished residents’ participation in longer (and more of the 12 months, when there were only 3 juniors. This complex) cases, we then analyzed the number of cases restructuring of the service was attributable to reshuf- performed within each of the four case categories (see Table 1). By case category, there was very little difference fling of residents, which became necessary in response to in the average senior’s operative experience before versus the 80-hour limits. after the 80-hour limit, with one notable exception (Fig. 2). Actual number of cases performed by all attending Endoscopic and vascular access procedures were per- surgeons on the service per month is shown in Table 3. formed nearly twice as frequently by seniors after the There was no significant difference in average monthly 80-hour limit (before, 0.43 Ϯ 0.25 versus after, 0.80 Ϯ total number of cases performed on the service before 0.41 Category I cases per day, p ϭ 0.03).This increase in versus after implementation of the 80-hour week (Table 3). Category I cases entirely accounts for the slight (nonsig- In addition, this total number of cases each month, nificant) increase noted in seniors’ total number of cases without exception, always exceeded the total number of (Fig. 1). Otherwise, average number of cases performed cases actually performed by residents (Table 3), demon- by seniors within each of the other categories did not strating that residents’ education was not handicapped change, suggesting that participation in longer, more by a “shortage” of available operative cases.Totalnumber complex operations was unaffected (Fig. 2). The same of cases performed by attendings could not have ob- analysis, by case category, was performed for junior res- scured a potential effect of the 80-hour limit on resi- idents (Fig. 3). Before and after the 80-hour week, junior dents’ operative experience. residents’ operative experience did not change signifi-

Table 2. Questionnaire for Residents’ Assessment of Surgical Education Experience Before 80-hour week After 80-hour week p Value (27 ؍ n) (31 ؍ Questionnaire item description (n Educational quality of weekly conferences 3.2 Ϯ 1.3 2.9 Ϯ 1.2 0.42 Quality of teaching received from surgical attendings 2.9 Ϯ 1.1 3.1 Ϯ 1.1 0.48 Quality of teaching received from pediatric surgery fellows 3.7 Ϯ 1.1 3.8 Ϯ 1.3 0.86 Amount of time available for reading 2.7 Ϯ 1.0 2.5 Ϯ 1.2 0.56 Number of cases they had participated in 3.7 Ϯ 1.0 3.6 Ϯ 1.3 0.74 Degree permitted to participate in cases 3.0 Ϯ 1.1 3.2 Ϯ 1.1 0.62 Variety of cases exposed to during the month 2.8 Ϯ 1.3 3.1 Ϯ 1.2 0.37 Rate the amount of independent decision making entrusted to them 2.5 Ϯ 1.1 2.1 Ϯ 1.3 0.23 Degree to which residents had been involved in patient management 2.7 Ϯ 1.2 2.6 Ϯ 1.2 0.87 Overall, qualitative assessment of the educational value of rotation 3.3 Ϯ 1.1 3.1 Ϯ 1.2 0.48 The questionnaire was confidential to encourage candid responses. All items were answered on a Likert-type scale, where higher numbers indicate greater satisfaction, greater number of cases, or greater participation. For example: 1 ϭ “very unsatisfied” to 5 ϭ “very satisfied.” Results are given as mean Ϯ SD. Vol. 200, No. 5, May 2005 Spencer and Teitelbaum Eighty Hours and Residents’ Surgical Experience 673

Table 3. Average Total Case Numbers Overall July 1, 2002–June 30, 2003 July 1, 2003–June 30, 2004 Mean ؎ SD Range Mean ؎ SD Range p Value A. Total cases performed per month by all attendings on service 207 Ϯ 28 157–241 202 Ϯ 23 178–249 0.61 B. Total cases performed per month by all residents (seniors ϩ juniors) 102 Ϯ 18 71–130 95 Ϯ 29 55*–142 0.51 C. Percentage of (A), ie, the percentage of cases performed by a resident 49 Ϯ 7 38–66 47 Ϯ 13 24†–64 0.66 Operative experience of residents was not limited by availability of cases. There were always more than enough cases available for residents to participate in; residents performed Ͻ50% of total cases each month on average. This proportion did not differ before or after implementation of the 80-hour week. However, the 80-hour week did require removal of the senior resident from the service on 4 different months. *There were 2 months during which the total number of cases performed by residents was only 55; during both of these months, no senior was on service. †No senior was on service during this month. cantly within any of the 4 case categories. Both before Residents’ perceptions and after July 1, 2003, the majority of cases performed Although data indicated that operative experience had by junior residents fell into Categories I and II, as not significantly changed, we questioned whether resi- expected. dents’ perceptions would reflect this fact. The 10-item To determine whether junior residents’ operative ex- questionnaire allowed residents to assess the educational perience was skewed by presence or absence of a senior experience during their month on service; responses are resident on the service, months lacking a senior resident shown inTable2. Response rate to the questionnaire was were compared with those when a senior was present. similar for both time periods (before, 31 responses Junior residents’ operative experience was not signifi- [66%]; after, 27 responses [61%]). There was no signif- cantly different when there was no senior (0.81 Ϯ 0.14 icant difference in any of the 10 questionnaire items cases per day) versus when a senior was present (0.68 Ϯ between residents before versus after implementation of 0.18 cases per day, p ϭ 0.15). Junior residents main- the 80-hour week (Table 2). tained their operative experience under the 80-hour limit, independently of the fact that there was no senior Involvement in outpatient clinic resident on the service for 4 months, later in the study. After several months had passed since implementation of the 80-hour limit, it gradually became apparent that

Figure 2. Seniors’ operative experience by category. *p ϭ 0.03, Figure 1. Total number of cases per day, on average, performed by before versus after. There was no signiﬁcant change in any of the residents was stratiﬁed by level of training (senior or junior). The other three categories after implementation of the 80-hour week. 80-hour per week limit on residents’ duty hours did not reduce total Case categories: I ϭ vascular access and endoscopic procedures; number of cases they performed. *p ϭ 0.37, seniors before versus II ϭ minor surgical procedures; III ϭ major open thoracic or abdom- seniors after; †p ϭ 0.91, juniors before versus juniors after. inal procedures; IV ϭ minimally invasive procedures. 674 Spencer and Teitelbaum Eighty Hours and Residents’ Surgical Experience J Am Coll Surg

subjected to statistical analysis, other factors in the surgical educational experience are much more difficult to assess, such as technical skill, professionalism, and surgical judgment. It will be important in the future to continue to monitor the actual number of cases performed by residents. Although we have reported data here from 1 complete year under the 80-hour system, changes in the structure of many residency programs continue to occur.6,8,11-14 Continued monitoring will be important, both for residents and for surgery as a profession, because the volume of cases performed by residents may yet be affected in the future.

Figure 3. Juniors’ operative experience by category. The 80-hour Before the ACGME mandate, residents in our pro- week did not signiﬁcantly affect juniors’ experience in any of the 4 gram regularly worked well beyond 80 hours per week, a case categories. See Figure 2 for case categories. documented characteristic of many surgical residency programs.8,14-16 The dramatic changes necessitated by resident participation in outpatient clinic was being cur- the 80-hour limit were unprecedented in our residency tailed in efforts to maintain inpatient care on the service. program. To comply with the requirements, after July 1, Although the outpatient aspect of the educational expe- 2003, duty hours were closely monitored, and junior rience had not been included in the original study de- residents were required to leave the hospital at 10:00 AM sign, we retrospectively sought to quantify residents’ on the morning postcall. Even with this change, compli- clinic experience to determine whether it had actually ance with the 80-hour limit was not always achieved; been reduced. A survey was distributed to all attending and within a few months, the postcall residents had to be surgeons who had held outpatient clinics during both relieved of duty after 8:00 AM. With this system, com- time periods (both before and after the 80-hour limit), pliance with the 80-hour limit was achieved. We were to assess clinic coverage by residents (Table 4). Response surprised to find that the actual number of residents’ rate was 100% (5 of 5 attendings) and responses were operative cases did not decrease. It may be that independent of each other. There was a significant de- residents—to avoid a loss of operative experience— crease in average number of residents attending clinic, consciously made extra efforts to scrub into cases during and a decrease in percentage of outpatient clinics cov- their allotted time in the hospital, after the 80-hour limit ered by residents, after implementation of the 80-hour took effect. Although Sawyer and colleagues17 have dem- week (Table 4). onstrated that resident participation in surgery decreases with increasing frequency of call (every other night, ver- DISCUSSION sus every third or fourth night), frequency of call was We did not find that the 80-hour limit has reduced unchanged in our study. Both before and after the 80- resident participation in surgery. The total number of hour week, residents took call every third night. cases performed by residents, and participation of resi- Although our original hypothesis—that a reduction dents in longer and more complex cases, has not dimin- in work hours would lead to a reduction in operative ished. Our data suggest that the effects of work-hour experience—had proved false, we found that residents’ limits may be complex. Although case load is readily participation in outpatient surgical clinic had signifi-

Table 4. Resident Participation in Outpatient Clinic Before 80-hour limit After 80-hour limit p Value Average number of residents participating in each clinic 1.3 Ϯ 0.4 0.2 Ϯ 0.4 Ͻ 0.005 Percentage of clinics covered by residents 66.0 Ϯ 14.7 17.0 Ϯ 19.9 Ͻ 0.005 Data gathered retrospectively. Responses (n ϭ 5) from all attending surgeons were independent and were unanimous in trend: all reported a decrease in resident clinic participation after the 80-hour limit. Values are mean Ϯ SD. Vol. 200, No. 5, May 2005 Spencer and Teitelbaum Eighty Hours and Residents’ Surgical Experience 675 cantly decreased. Unfortunately, clinic participation had cases in Category I are likely to be less highly sought after not been included in the original study design, so these by upper-level residents than cases in the other catego- data could only be gathered retrospectively. This finding ries. Our data suggest that a minor shifting of junior seems to be real, and was unanimously reported, inde- responsibilities to seniors may have occurred. pendently, by all attending surgeons (Table 4). Before We found no significant difference in any of the 10 the 80-hour limit, all residents were heavily involved in questionnaire items between residents before versus outpatient surgical clinic. With limitation on hours, it after the 80-hour week, which correlated with the appears that clinic experiences have suffered. Although finding that there was no actual difference in number maintaining residents’ intraoperative experience was an of cases performed by residents. This consistency of encouraging result, clinic experiences are also clearly vi- residents’ responses validates use of our question- tal for residents. Our program continues to modify com- naire. Although residents under the 80-hour limit pliance efforts to avoid compromising resident educa- might have been expected to respond differently than tion. Achieving this goal requires identification and residents from the previous year, the fact that they elimination of noneducational activities, which have responded similarly (in light of a similar operative been shown to comprise a large percentage of residents’ experience) suggests that residents’ own assessment of time.18 the surgical educational experience is an important Another point to be mentioned is that restructuring and accurate gauge of at least some major components of the service took place in the last few months of the of that experience. study, with removal of the senior from the service, leav- Our study has measured residents’ operative experi- ing only junior residents under the supervision of the ence before and after implementation of the 80-hour pediatric surgical fellows. This restructuring was a direct week, both in terms of residents’ total number of cases result of compliance with ACGME regulations. In the and their case variety, in a tightly controlled setting. We future, it is certainly conceivable that this type of restruc- did not attempt to measure patient outcomes, although turing of resident rotations may result in an overall de- other studies suggest that the 80-hour week ultimately crease in number of cases performed by a senior-level may not enhance patient outcomes, and may have ad- resident on some subspecialty services. Data describing verse effects in some settings.19-21 Additional studies are the operative experience of general surgery residents on clearly needed to assess the impact of reduced working surgical subspecialties such as pediatric surgery (for hours on patient care. which the ACGME requires residents to log a specified In conclusion, after 1 year of compliance with the number of cases) are very scarce. ACGME 80-hour weekly limit, we did not find a signif- It is possible that the lack of changes observed on our icant reduction in number or variety of cases performed pediatric surgical service may not reflect operative expe- by surgical residents on a major subspecialty service. riences of residents on other surgical rotations. Our ser- This finding is encouraging and demonstrates that com- vice provided a well-controlled setting because residents pliance with the 80-hour limit can be achieved while within our program from one year to the next provided essentially maintaining surgical residents’ operative ex- a fairly homogenous study group, and because we were perience. We also did find evidence of some negative able to exclude potential confounding factors, such as impacts of the 80-hour limit on the educational changes in total number of cases performed by attending process—in particular, clinic experience—and conclude surgeons, or changes in frequency of call schedules. that it will be necessary to continue to monitor and We found that seniors performed significantly more modify the process of compliance with new regulations. vascular access and endoscopic cases after the 80-hour Studies of this nature may help to clarify the most ap- week was in effect. The explanation for this is not im- propriate ways to achieve compliance and improve qual- mediately apparent, but several previous studies have ity of residency training at the same time. indicated that senior residents may be affected by work- hour limits more than juniors.8,9 In New York, 35% of residents reported that duties had been shifted from jun- REFERENCES 8 ior to senior residents to comply with work-hour limits. 1. Accreditation Council for Graduate Medical Education. Com- Our finding may represent this type of phenomenon, as mon program requirements; section VI: resident duty hours and 676 Spencer and Teitelbaum Eighty Hours and Residents’ Surgical Experience J Am Coll Surg

the working environment, 7Ϫ8. Available at: http:// resident duty-hour regulations. J Am Coll Surg 2004;198: www.acgme.org/DutyHours/dutyHoursCommonPR.pdf. Ac- 989–993. cessed October 7, 2004. 12. Mendoza KA, Mendoza B, Britt LD. A template for change and 2. Accreditation Council for Graduate Medical Education. Second response to work hour restrictions. Am J Surg 2003;186:89–96. report of the subcommittee on duty hours: advising ACGME on 13. Curet MJ, McAdams TR. Improving resident work environ- the implementation and monitoring of the duty hour standards. ment: evaluation of a novel cooperative program. Surgery 2003; June 2004. Available at: http://www.acgme.org/DutyHours/dh_ 134:158–163. subcomreport0604.pdf. Accessed October 19, 2004. 14. Neumayer LA, Cochran A, Melby S, et al. The state of general 3. Gaba DM, Howard SK. Fatigue among clinicians and the safety surgery residency in the United States: program director per- of patients. N Engl J Med 2002;347:1249–1255. spectives, 2001. Arch Surg 2002;137:1262–1265. 4. New York State Department of Health, Medical Staff in New 15. Niederee MJ, Knudtson JL, Byrnes MC, et al. A survey of resi- York Codes, Rules and Regulations, Title (1998);10.§405.4. dents and faculty regarding work hour limitations in surgical 5. DeBuono BA, Osten WM. The medical resident workload: the training programs. Arch Surg 2003;138:663–671. case of New York state. JAMA 1998;280:1882–1883. 16. Cockerham WT,Cofer JB, Lewis PL, et al. Resident work hours: 6. Hassett JM, Nawotniak R, Cummiskey D, et al. Maintaining can we meet the ACGME requirements? Am Surg 2004;70: outcomes in a surgical residency while complying with resident 687–690. working hour regulations. Surgery 2002;132:635–641. 7. Barden CB, Specht MC, McCarter MD, et al. Effects of limited 17. Sawyer RG, Tribble CG, Newberg DS, et al. Intern call sched- work hours on surgical training. J Am Coll Surg 2002;195:531– ules and their relationship to sleep, operating room participa- 538. tion, stress, and satisfaction. Surgery 1999;126:337–342. 8. Whang EE, Mello MM, Ashley SW, Zinner MJ. Implementing 18. Brasel KJ, Pierre AL, Weigelt JA. Resident work hours: what resident work hour limitations: lessons from the New York state they are really doing. Arch Surg 2004;139:490–494. experience. Ann Surg 2003;237:449–455. 19. Haynes DF, Schewelder M, Dyslin DC, et al. Are postoperative 9. Whang EE, Perez A, Ito H, et al. Work hours reform: percep- complications related to resident sleep deprivation? South Med J tions and desires of contemporary surgical residents. J Am Coll 1995;88:283–289. Surg 2003;197:624–630. 20. Ellman PI, Law MG, Tache-Leon C, et al. Sleep deprivation 10. Ruby ST, Allen L, Fielding LP, Deckers PJ. Survey of residents’ does not affect operative results in cardiac surgery. Ann Thorac attitudes toward reform of work hours. Arch Surg 1990;125: Surg 2004;78:906–911. 764–768. 21. Laine C, Goldman L, Soukoup JR, Hayes JG. The impact of a 11. Underwood W, Boyd AJ, Fletcher KE, et al. Viewpoints from regulation restricting medical house staff working hours on the generation X: a survey of candidate and associate viewpoints on quality of patient care. JAMA 1993;269:374–378.

ANNOUNCEMENT All manuscripts will be submitted on the electronic system ONLY. SEE GUIDELINES ON www.journalacs.org Does Surgeon Frustration and Satisfaction with the Operation Predict Outcomes of Open or Laparoscopic Inguinal Hernia Repair?

Haytham MA Kaafarani, MD, Kamal MF Itani, MD, FACS, Anita Giobbie-Hurder, MS, John J Gleysteen, MD, FACS, Martin McCarthy Jr, PhD, James Gibbs, PhD, Leigh Neumayer, MD, MS, FACS

BACKGROUND: A surgeon’s level of frustration when performing an operation and level of satisfaction at completion may be correlated with patients’ outcomes. We evaluated the relationship between the attending surgeons’ frustration and satisfaction and recurrence and complications of open and laparoscopic inguinal hernia repair. STUDY DESIGN: Men with detectable inguinal hernias were randomized to undergo open or laparoscopic herniorrhaphy at 14 Veterans Affairs hospitals. After completion of the procedure, surgeons were asked to assess their level of frustration during the operation and their overall satisfaction with the operative result. Two subjective scales ranging from 1 (not frustrated/not satisfied) to 5 (very frustrated/very satisfied) were used to independently assess both parameters. Reasons for surgeon frustration were evaluated. Patients were followed for 2 years for recurrence and complications. RESULTS: Of 1,983 patients who underwent hernia repair, 1,622 were available for analysis; 808 had open repair and 813 had laparoscopic repair. Surgeons reported less frustration and more satisfaction with open than with laparoscopic repair (p ϭ 0.0001 and 0.0001, respectively). Frustration was associated with a higher rate of hernia recurrence at 2 years (adjusted odds ratio, 2.01, 95% CI, 1.15Ϫ3.51) in open repair, and a higher overall rate of postoperative complications (adjusted odds ratio, 1.27, 95% CI, 1.03Ϫ1.56) in both open and laparoscopic hernia repair. Satisfaction was not correlated with recurrence or complications. CONCLUSIONS: The level of a surgeon’s frustration during performance of an inguinal herniorrhaphy was a better predictor of outcomes of the operation than was satisfaction with the procedure. Sources of intraoperative frustration should be controlled to improve outcomes. (J Am Coll Surg 2005; 200:677–683. © 2005 by the American College of Surgeons)

Outcomes of a surgical procedure are related to a num- clude preoperative risk status of the patient,1,2 risk asso- ber of factors involving the patient, the surgeon, the ciated with the procedure itself,1-3 operative experience procedure, and the operative environment. These in- of the surgeon,4-7 volume of patients and quality of care in a specific institution,6-8 and team coordination of the Competing interests declared: None. surgical staff.9 Supported by a grant from the Cooperative Studies Program of the Depart- Because the patient is of paramount importance in ment of Veterans Affairs Office of Research and Development (CSP# 456), Washington, DC. any delivery of care, particular emphasis has been given Received November 11, 2004; Accepted November 17, 2004. to patient-dependent measurements of outcomes, such From the Michael E DeBakey Veterans Affairs Medical Center and Department as patient satisfaction, health-related quality of life, and of Surgery, Baylor College of Medicine, Houston,TX (Kaafarani); Department of 10,11 Surgery, Veterans Affairs Boston Health Care System and Boston Medical Center, disease-related quality of life. Boston, MA (Itani); Veterans Affairs Cooperative Studies Program Coordinating Most studies evaluating “the surgeon factor” have cen- Center, Hines, IL (Giobbie-Hurder); Birmingham Veterans Affairs Medical Center and Department of Surgery, University of Alabama at Birmingham, tered on variables relating to technical skill and amount Birmingham, AL (Gleysteen); Department of Preventive Medicine, The of operative experience.5,12 Some have compared the sur- Feinberg School of Medicine (McCarthy) and Institute for Health Services Research and Policy Studies (Gibbs), Northwestern University, Chicago, IL; geon’s level of satisfaction with that of the patient and and Department of Surgery, Salt Lake City Veterans Affairs Medical Center with the outcomes of the operation, especially for total and University of Utah, Salt Lake City, UT (Neumayer). Correspondence address: Leigh Neumayer, MD, MS, Surgical Service, 500 hip arthroplasty procedures. Discrepancies have been Foothill Dr, Salt Lake City, UT 84148. found between surgeons’ and patients’ satisfaction with

© 2005 by the American College of Surgeons ISSN 1072-7515/05/$30.00 Published by Elsevier Inc. 677 doi:10.1016/j.jamcollsurg.2004.11.018 678 Kaafarani et al Operative Frustration and Satisfaction J Am Coll Surg outcomes of the operation.13-15 Literature concerning a Surgeon frustration and satisfaction surgeon’s sense of frustration during the operation is At the conclusion of each operation, the research nurse even more limited.16-18 provided the attending surgeons with two evaluation This study was conducted as part of a larger clinical sheets. One sheet requested a subjective evaluation of trial comparing open Lichtenstein to laparoscopic her- frustration experienced during the operation, using a nia repair. Levels of surgeon frustration and surgeon sat- scale of 1 (no frustration) to 5 (very frustrated). If any isfaction with open and laparoscopic procedures were level of frustration above “none” was chosen, the sur- evaluated and compared prospectively. Then, for each geon was asked to assign it to one or more causes: type of operation, we determined whether frustration equipment or instruments, technical difficulties, per- and satisfaction were related to or could predict postop- sonnel in the operating room, or anatomy of the patient. erative complications, hernia recurrence, and operative The second evaluation sheet asked for the surgeon’s time. overall satisfaction at the conclusion of the procedure using a 1 (not satisfied) to 5 (very satisfied) scale. Cross- sectional and comparative analyses of percentages of METHODS these ratings were performed comparatively during the Fourteen Veterans Affairs medical centers participated in active enrollment phase of the trial. a Veterans Affairs Cooperative Study (CSP 456), which recruited male veterans with detectable inguinal hernias Outcomes between January 1999 and December 2001. The design Outcomes measured in this study were complications and principal results of this larger trial have been re- and hernia recurrence. ported previously.19,20 In addition to listing intraoperative complications, the research nurse recorded short-term postoperative complications at 1 to 2 weeks after hernia repair and Clinical trial longterm postoperative complications at 3 months to 2 After completion of a screening process, consenting pa- years after operation. For the aggregate analysis of total tients with unilateral or bilateral, primary or recurrent in- complications, a patient was counted as having a com- guinal hernias were randomized to undergo open tension- plication if he had one or more of an intraoperative, free mesh (Lichtenstein) or a laparoscopic tension-free short-term postoperative, or longterm postoperative mesh (extraperitoneal or transabdominal) repair of the her- complication. A patient was counted only once in any nia. Patients were followed for 2 years. analysis. Death was not included. Although surgical housestaff physicians were allowed Hernia recurrence within 2 years was also recorded. to participate as primary surgeons during each opera- Recurrence was defined as a reducible bulge at the site of tion, the protocol required an attending staff physician hernia repair. with earlier surgical experience of at least 25 procedures of the type to be performed to be present and scrubbed Operative time “from initial incision to onset of skin closure.” Before the Time from incision to skin closure was recorded for each cooperative study began, specific technique protocols operation. were agreed on by the participants, and were supplemented by videos. Surgeon members of the trial’s Exec- Analysis utive Committee visited each participating center early Because the operative technique and skills required for in the study on days of scheduled operations to assure open and laparoscopic hernia repair are distinct, we first familiarity with and compliance with the standardized compared the levels of surgeon frustration and satisfac- techniques. tion between the two techniques. A research nurse was appointed for each participating We assumed that surgeon frustration during an oper- institution. These nurses or their representatives at- ative procedure and satisfaction with the results by the tended each operative procedure, recorded technical and end of the procedure are related but not necessarily the anatomic specifics of the operation, and reported any same; we evaluated the correlation between levels of intraoperative problems. frustration satisfaction. We then investigated the corre- Vol. 200, No. 5, May 2005 Kaafarani et al Operative Frustration and Satisfaction 679

Table 1. Comparison of the Two Treatment Groups on Baseline Characteristics Open repair Laparoscopic p Value (813 ؍ repair (n (808 ؍ Characteristic (n Days between randomization and operation, meanϮSD (%) 41.1 (81.3) 35.4 (52.6) 0.10 Age (y), meanϮSD (%) 58.0 (12.5) 58.3 (12.9) 0.61 Height (in), meanϮSD (%) 69.9 (2.7) 69.8 (2.8) 0.71 Weight (pounds), meanϮSD (%) 177.8 (28.7) 178.5 (30.6) 0.63 Highest grade completed, meanϮSD (%) 12.7 (2.4) 12.7 (2.4) 0.78 Race (%) 0.57 Caucasian 75.8 73.5 African American 20.6 23.2 Asian 0.3 0.1 Multiracial 3.4 3.1 Duration of hernia (%) 0.10 Ͻ 6 wk 10.6 9.7 6 wk to 1 y 49.9 56.0 Ͼ 1 y 36.2 30.4 Unknown 3.3 3.9 Hernia (%) Unilateral 81.8 82.5 Bilateral 18.2 17.5 0.70 Primary 91.8 90.5 Recurrent 8.2 9.5 0.36 Coexisting conditions (%) Congestive heart failure 0.1 0.6 0.22 Prior myocardial infarction 0.4 0.1 0.37 Hypertension 35.8 35.1 0.76 COPD 5.0 4.9 0.98 Chronic cough 7.2 9.2 0.13 Prostate 16.8 17.7 0.64 Diabetes 4.6 6.0 0.38 Smoke cigarettes 43.2 40.3 0.24 Alcohol consumption (Ͼ 2 drinks/day) 15.5 14.2 0.45 American Society of Anesthesiologists class (%) 0.41 I 32.7 35.8 II 48.8 46.1 III 18.6 18.1 lation between surgeon frustration and satisfaction and Because study enrollment was stratified by hospital, postoperative outcomes listed earlier. Last, we examined hernia site (unilateral or bilateral), and hernia type (pri- the causes of frustration (levels 2 to 5) in open and mary or recurrent), it was assumed a priori that there laparoscopic hernia repairs. would be differences in outcomes related to these vari- Statistical models were constructed for operative ables. Hence, all models included these factors as covari- time, recurrence, overall complications, intraoperative ates. Treatment group (open or laparoscopic) was also complications, short-term complications, and long- included in each model as an independent variable. term complications. Recurrence and complication rates were modeled using logistic regression. All models used RESULTS generalized estimating equations to accommodate cor- Patient population relations introduced by an attending who performed A total of 1,983 patients had a hernia repair. In 1,621 multiple hernia repairs. The computer program used patients (808 randomized to open repair and 813 ran- was PROC GENMOD from SAS. domized to laparoscopic repair), data with regard to sur- 680 Kaafarani et al Operative Frustration and Satisfaction J Am Coll Surg

Figure 1. Surgeon frustration in open versus laparoscopic hernia repair showing that surgeons report signiﬁcantly more frustration in the laparoscopic technique (p ϭ 0.0009). White, no frustration; gray, some frustration. geons’ frustration intraoperatively and their satisfac- Satisfaction tion with the operation postoperatively were complete Surgeons reported higher levels of satisfaction with open (81.7%). The two treatment groups were similar in hernia repair. They were “satisfied” (levels 4 to 5) in terms of baseline characteristics (Table 1). 93.7% of the open herniorrhaphies, but in only 89.1% of the laparoscopic repairs (p Ͻ 0.0001) (Fig. 2). Surgeon frustration and satisfaction in open versus laparoscopic hernia repair Frustration versus satisfaction Reclassiﬁcation of levels of surgeon frustration and In both open and laparoscopic hernia repairs, an inverse surgeon satisfaction relationship between levels of frustration and those of Distribution of levels of frustration or satisfaction was satisfaction was evident, with Spearman rank order cor- skewed; most answers were classified as “No frustration” relation coefficient being Ϫ0.24 and Ϫ0.29, respec- (level 1) and “Very satisfied” (level 5). Levels of frustra- tively (Table 2). tion were grouped into “No frustration” (level 1; 58%) or “Some frustration” (levels 2 to 5; 42%). The levels of Surgeon frustration versus hernia recurrence at surgeon satisfaction were also grouped as “No/little sat- 2 years isfaction” (levels 1 to 3; 9%) or “Satisfied” (levels 4 to In the 1,621 patients with complete frustration/satis- 5; 91%). faction data there were 111 (6.8%) recurrences (37 Frustration [4.6%] in the open group, and 74 [9.1%] in the laparo- In open hernia repairs, surgeons reported “No frustration” scopic group). The level of surgeon frustration corre- (level 1) in 70.3% of patients; frustration (levels 2 to 5) was lated with hernia recurrence only in the open repair reported in 29.7% of the procedures (Fig. 1). In com- group (adjusted odds ratio, 2.01; CI, 1.15Ϫ3.51) (Table parison, surgeons reported frustration (levels 2 to 5) 3); in other words, an open herniorrhaphy procedure in with 52.5% of laparoscopic herniorrhaphies. Differ- which a surgeon expressed any level of frustration (levels ences in levels of frustration between the open and lapa- 2 to 5) was twice as likely to recur within 2 years than the roscopic groups were significant (p ϭ 0.0009). procedure in which the surgeon had no frustration. Re- Vol. 200, No. 5, May 2005 Kaafarani et al Operative Frustration and Satisfaction 681

Figure 2. Surgeon satisfaction in open versus laparoscopic hernia repair showing that surgeons report significantly more satisfaction in the open technique (p Ͻ 0.0001). White, no/little satisfaction; gray, satisfied. currence rate was similar after laparoscopic repair be- odds ratio [AOR], 1.3; 95% CI, 1.03Ϫ1.56, Table 3). tween frustration and no frustration with the procedure. The adjusted odds ratios were the same for open and laparoscopic repairs. Surgeon frustration versus A total of 53 patients had intraoperative complica- postoperative complications tions. Procedures in which surgeons expressed frustra- The frustration level of the attending surgeon signifi- tion were 2.9 times more likely to be accompanied by an cantly influenced the overall complication rate. Proce- intraoperative complication than those in which the sur- dures for which the surgeon expressed frustration were geons had no frustration (95% CI, 1.62Ϫ5.11; Table 3). accompanied by an overall risk of complications that was approximately 1.3 times higher than the procedures for which the surgeon expressed no frustration (adjusted Table 3. Surgical Outcomes (Recurrence, Complications) as a Function of Surgeon Frustration (Some Frustration Versus Table 2. Correlation Between Surgeon Frustration and Sur- No Frustration) geon Satisfaction with Hernia Repair 95% Odds Standard confidence Little/no Spearman † ratio error interval satisfaction* Satisfied rank order n%n%correlation Hernia recurrence at 2 y Open repair (n ϭ 808) Laparoscopic repair 0.93 0.31 (0.49–1.79) No frustration‡ 14 2.5 554 97.4 Open repair 2.01 0.57 (1.15–3.51)* Ϫ0.24 Frustration§ 37 15.4 203 84.6 Postoperative complications (open and laparoscopic repair) Laparoscopic repair (n ϭ 813) IOC 2.88 0.84 (1.62–5.11)* No frustration 5 1.3 381 98.7 ST-POC 1.22 0.15 (0.96–1.56) Ϫ0.29 Frustration 84 19.7 383 80.3 LT-POC 1.14 0.15 (0.88–1.49) Total 1.27 0.13 (1.03–1.56)* *Little/no satisfaction, levels 1–3. †Satisfied, levels 4–5. *Statistically significant. ‡No frustration, level 1. IOC, intraoperative complications; LT-POC, longterm postoperative com- §Frustration, levels 2–5. plications; ST-POC, short-term postoperative complications. 682 Kaafarani et al Operative Frustration and Satisfaction J Am Coll Surg

Table 4. Operative Time as a Function of Surgeon satisfaction at the end was recorded had significantly Frustration/Satisfaction shorter operative times (30 minutes shorter in laparoscopic Average time repair, 20 minutes shorter in the open repair) (Table 4). difference Standard 95% conﬁdence (min) error interval DISCUSSION Surgeon frustration (some frustration–no frustration) In this study, our primary goal was to determine if sur- Laparoscopic repair 30.2 2.6 (25.1–35.3)* geon frustration and satisfaction during and after hernia Open repair 18.1 2.4 (13.5–22.8)* Surgeon satisfaction (satisfied–no/little satisfaction) repair were correlated with postoperative complications, Laparoscopic repair -30.5 5.1 (-40.5–-20.6)* hernia recurrence, and operative time. We first com- Open repair -20.3 4.7 (-29.5–-11.1)* pared levels of frustration between open and laparo- *Statistically significant. scopic repairs. Surgeons were significantly more frustrated and less satisfied with the laparoscopic technique. The adjusted odds ratios were not different between the We then compared frustration with satisfaction. We open and the laparoscopic techniques. assumed a negative relationship between the two, ie, There were 344 patients with short-term and 271 frustration was accompanied by less satisfaction. The Ͻ patients with longterm postoperative complications. relationship was not strong ( -0.3), suggesting that sur- Whether or not the surgeon was frustrated during the geons perceive frustration and satisfaction as related but repair did not affect the rate of either of these catego- not binding, eg, the source of frustration may have re- ries of complications (Table 3). solved during the case with a satisfying result at the end. This concept was further strengthened when we com- Source of surgeon frustration pared frustration and satisfaction in terms of their pre- The major reason for frustration (levels 2 to 5) was the dictability of surgical outcomes. anatomy of the patient; this was reported as a cause in Frustration was strongly related to the incidence of 60% (143 of 240) of the procedures in the open repair postoperative complications, especially intraoperative and 45% (191 of 427) of procedures in the laparoscopic complications. A herniorrhaphy (open or laparoscopic) repair (p ϭ 0.0002). The anatomy of the patient was the during which the surgeon was frustrated was one of sole cause of frustration in 41.7% of the procedures higher risk of complications and these were most often (278 of 667). Other causes of frustration were personnel intraoperative. When surgeons were asked to indicate (32.5%, 217 of 667), equipment (17.4%, 116 of 667), the cause of their frustration, most reported the anatomy instruments or technical issues (such as unexpected of the patient as the principal cause, rather than the complications) in 13.9% (93 of 667). occurrence of a technical problem, which was the least frequent source of frustration.The second most frequent Satisfaction and outcomes source was personnel (not further defined), and one Surgeon satisfaction with the hernia repair was not cor- might imagine a close fit with anatomy, reflecting the related with incidence of hernia recurrence at 2 years for capability of an assistant during a difficult dissection. either open or laparoscopic repairs. Whether a surgeon Surgeon frustration was a significant predictor of hernia reported being satisfied did not affect the rate of occur- recurrence within 2 years if the procedure was an open rence of any of the postoperative complications for ei- repair. A procedure in which the surgeon reported being ther the open or the laparoscopic hernia repairs. frustrated had double the risk of recurrence within 2 years than a procedure in which the surgeon reported no frustra- Operative time tion. This association was not found for laparoscopic re- Average operating time was 77 minutes with a median of pairs. The reason may reflect familiarity and comfort with 70 minutes (range 10 to 320 minutes). In 19 of 1,621 the open repair, and some difficulty in recognition of an operations, the time was longer than 180 minutes. insecure repair during a laparoscopic procedure. Procedures in which the surgeons were “frustrated” were A surgeon’s satisfaction with the hernia repair was not significantly associated with a longer operative time (30 predictive of postoperative complications or recurrence minutes longer in laparoscopic repair, 18 minutes longer in in either an open or a laparoscopic repair. This suggests the open repair) (Table 4). Procedures in which surgeon that surgeons are trained to resolve intraoperative diffi- Vol. 200, No. 5, May 2005 Kaafarani et al Operative Frustration and Satisfaction 683 culties and inconsistencies before completing an opera- 2. Khuri SF,Daley J, Henderson WG. The comparative assessment tion. This also explains the longer operative time when and improvement of quality of surgical care in the Department of Veterans Affairs. Arch Surg 2002;137:20–27. the surgeon was frustrated and dissatisfied. It might also 3. Fink AS, Campbell DA Jr, Mentzer RM Jr, et al. The National indicate that surgeons’ evaluation of their own work and Surgical Quality Improvement Program in non-veterans admin- perception of the quality of the procedure is irrelevant to istration hospitals: initial demonstration of feasibility. Ann Surg 2002;236:344–353; discussion 353Ϫ354. postoperative outcomes as measured by the incidence of 4. Barrat C, Cueto Rozon R, Catheline JM, et al. The effect of the complications or recurrence. learning curve and the experience on the technique and the Lieberman introduced the concepts of surgeon satis- outcome of laparoscopic treatment for gastroesophageal reflux. Chirurgia (Bucur) 2000;95:325–333. faction and surgeons’ evaluation of outcomes of surgical 5. Bencini L, Sanchez LJ. Learning curve for laparoscopic ventral 13 procedures in 1996. This concept has been evaluated hernia repair. Am J Surg 2004;187:378–382. in several subsequent studies in relation to patient satis- 6. Flood AB, Scott WR, Ewy W. Does practice make perfect? Part faction, specifically in total hip arthroplasty and knee I: the relation between volume and outcomes and other hospital 14,15,21,22 characteristics. Med Care 1984;22:98–114. replacement operations. Most of these studies 7. Flood AB, Scott WR, Ewy W. Does practice make perfect? Part suggested discrepancies between surgeons’ satisfaction II: the relation between hospital volume and outcomes for se- and patients’ satisfaction with outcomes of operation. lected diagnostic categories. Med Care 1984;22:115–125. 8. Hughes RG, Hunt SS, Luft HS. Effects of surgeon volume and Otherwise, surgeon satisfaction has not previously been hospital volume on quality of care in hospitals. Med Care 1987; assessed as a predictor of surgical outcomes. 25:489–503. Both measurements we report here are subjective and 9. Young GJ, Charns MP, Desai K, et al. Patterns of coordination and clinical outcomes: a study of surgical services. Health Serv largely dependent on each individual surgeon’s ability for Res 1998;33:1211–1236. self-criticism. The difference between frustration and satis- 10. Daley J, Khuri SF, Henderson W, et al. Risk adjustment of the faction at completion of the procedure has many interest- postoperative morbidity rate for the comparative assessment of ing implications for the practice and art of surgery. the quality of surgical care: results of the National Veterans Af- fairs Surgical Risk Study. J Am Coll Surg 1997;185:328–340. 11. Wright JG. Outcomes research: what to measure. World J Surg Author Contributions 1999;23:1224–1226. Study conception and design: Kaafarani, Itani, Neumayer 12. Edwards CC 2nd, Bailey RW. Laparoscopic hernia repair: the learning curve. Surg Laparosc Endosc Percutan Tech 2000;10:149–153. Acquisition of data: Itani, Giobbie-Hurder, Gleysteen, 13. Lieberman JR, Dorey F, Shekelle P, et al. Differences between Neumayer patients’ and physicians’ evaluations of outcome after total hip Analysis and interpretation of data: Kaafarani, Itani, Giobbie- arthroplasty. J Bone Joint Surg Am 1996;78:835–838. 14. McGee MA, Howie DW, Ryan P,et al. Comparison of patient and Hurder, Gleysteen, McCarthy, Gibbs, Neumayer doctor responses to a total hip arthroplasty clinical evaluation ques- Drafting of manuscript: Kaafarani, Itani tionnaire. J Bone Joint Surg Am 2002;84-A:1745Ϫ1752. Critical revision: Kaafarani, Gleysteen, McCarthy, Gibb, 15. Brokelman RB, van Loon CJ, Rijnberg WJ. Patient versus sur- Neumayer geon satisfaction after total hip arthroplasty. J Bone Joint Surg Br 2003;85:495–498. Statistical expertise: Giobbie-Hurder 16. Anonymous. Frustration in the surgery. Probe (Lond) 1968;9: Obtaining funding: Neumayer 83Ϫ84. Supervision: Itani 17. Patkin M, Isabel L. Ergonomics, engineering and surgery of endosurgical dissection. J R Coll Surg Edinb 1995;40:120–132. 18. Olmi S, Croce E. Scissor-knot-pusher: an instrument for simplified Acknowledgment: We thank Olga Jonasson, MD, FACS, for laparoscopic extracorporeal knotting. JSLS 2003;7:281–284. 19. Neumayer L, Jonasson O, Fitzgibbons R, et al. Tension-free ingui- her guidance throughout the study and her critical appraisal of nal hernia repair: the design of a trial to compare open and laparo- the manuscript, and Domenic Reda, PhD, for his guidance scopic surgical techniques. J Am Coll Surg 2003;196:743–752. throughout the study. 20. Neumayer L, Giobbie-Hurder A, Jonasson O, et al. Veterans Affairs Cooperative Studies Program 456 Investigators. Open mesh versus laparoscopic mesh repair of inguinal hernia. N Engl J Med 2004;350:1819–1827. REFERENCES 21. Ragab AA. Validity of self-assessment outcome questionnaires: 1. Khuri SF, Daley J, Henderson W, et al. The Department of patient-physician discrepancy in outcome interpretation. Biomed Veterans Affairs’ NSQIP: the first national, validated, outcome- Sci Instrum 2003;39:579–584. based, risk-adjusted, and peer-controlled program for the mea- 22. Moran M, Khan A, Sochart DH, Andrew G. Expect the best, surement and enhancement of the quality of surgical care. prepare for the worst: surgeon and patient expectation of the National VA Surgical Quality Improvement Program. Ann Surg outcome of primary total hip and knee replacement. Ann R Coll 1998;228:491–507. Surg Engl 2003;85:204–206. Perceptions and Predictors of Surgeon Satisfaction: A Survey of Spouses of Academic Surgeons

Lillian S Kao, MD, FACS, Erik B Wilson, MD, FACS, Kimberly D Anderson, PhD

BACKGROUND: Although several studies have addressed the role of spouses in physicians’ career choices, there is limited data about spousal perception of surgeons’ careers after training. This study examined satisfaction with surgeons’ careers and potential contributing factors from a spousal standpoint. STUDY DESIGN: A survey of spouses of academic surgeons in 38 participating departments was conducted. Questions included demographic information and perceptions of career satisfaction and contributing factors. Data were analyzed using descriptive statistics, Student’s t-test or Mann- Whitney U test, and chi-square or Fisher’s exact test. RESULTS: Three hundred seventy-nine surveys (27%) were returned, with the majority (81%) perceiving their surgeon spouses to be satisfied. Contributing factors to surgeon dissatisfaction included: work hours/call (42%); practice limitations (18%); reimbursement/income (12%); and malpractice/insurance (7%). Dissatisfied surgeons, as perceived by their spouses, did not differ from satisfied surgeons in terms of work hours, income expectations, geographic desirability, or home involvement. On the other hand, predictors of dissatisfied spouses included lack of input into career decisions, less satisfaction with location, and lack of their surgeon spouses’ participation in household and child-care activities. CONCLUSIONS: There are many positive and negative aspects to being an academic surgeon. Yet, despite the time commitment, work hours per se do not appear to contribute to either surgeon or spousal satisfaction. Spousal satisfaction is dependent on surgeon contribution to household and child- care activities. Despite the multiple potential detractors from an academic surgical career, most surgeons, as perceived by their spouses, would not have chosen differently and are satisfied with their career choices. (J Am Coll Surg 2005;200:684–690. © 2005 by the American College of Surgeons)

Multiple studies have demonstrated that factors associ- tant to identify so as to be able to maintain our current ated with relationships outside the hospital environment pool of surgeons. impact job satisfaction and the rate of burnout.1,2 In fact, Consider these quotations: studies have demonstrated that spouses have an impact on career choice,3 but there is a paucity of data about Example 1 spousal impact on or perception of surgeon satisfaction. His life and our life as a couple have not been our As fewer and fewer medical graduates decide on a surgi- own throughout his career. He was on call basically cal career because of lifestyle issues4 and an increasing all of the time.This resulted in very few vacations and number of practicing surgeons choose to retire early,5 numerous cancellations on our part for social and family events. Even though he now has a significant factors contributing to surgeon satisfaction are impor- administrative/leadership role, does not take call, and does less surgery, he still works the long hours. Competing interests declared: None. How do I feel about this after all these years? My Presented as a poster at the Association for Surgical Education, Houston, TX, main frustration is that he has not had much time for April 2004. his personal life. At times he feels guilty about miss- Received November 11, 2004; Revised January 5, 2005; Accepted January 11, ing family events. I adjusted to all of this a long time 2005. From the Department of Surgery, University of Texas Health Science Center ago and so did our children. Fortunately, because of at Houston, Houston, TX. his income, it was not necessary for me to work out- Correspondence address: Lillian S Kao, MD, FACS, Department of Surgery, University of Texas Health Science Center at Houston, 5656 Kelley St, Suite side the home and I was able to be there for the 30S 62008, Houston, TX 77026. children and take care of all of the details of our life

© 2005 by the American College of Surgeons ISSN 1072-7515/05/$30.00 Published by Elsevier Inc. 684 doi:10.1016/j.jamcollsurg.2005.01.009 Vol. 200, No. 5, May 2005 Kao et al Spousal Perceptions of Surgeon Satisfaction 685

and family. It was a lifestyle that came with being a study, and 38 chairpersons (26.6%) agreed to partici- surgeon and I don’t think either of us regrets it. The pate. By agreeing to participate, chairpersons facilitated bottom line is that despite the more difficult lifestyle, the distribution of surveys to faculty spouses. A total of my husband enjoys operating and taking care of pa- 1,380 mailings were sent to the collective departments tients more than anything. I cannot imagine him in of surgery. Administrative assistants in each of the de- any other career. partments addressed the envelopes to spouses at their Example 2 home addresses. Surveys were sent directly to the au- I feel strongly that spouses can make or break a doc- thors, and all questionnaires were coded to ensure tor’s career. I encourage spouses to be involved and confidentiality. help, not whine about the long hours and poor pay. Respondents completed a 49-question survey requir- Too many spouses don’t realize that the patients have ing responses consisting of yes and no answers, a 5-point to come first. Once the spouse accepts that fact he or Likert scale (5 indicating strong agreement), or free re- she can be proud and not resentful. A lot falls on the sponse. These questions related to demographics, career, spouse, but that’s just the way it is. The worst thing and household decision making, spousal perceptions of about my spouse’s career is the negative impact it had career satisfaction, contributing factors to satisfaction on my own career. I needed to be the one involved and dissatisfaction, career expectations, and personal with the children when they were young. As they got satisfaction with surgeons’ lives outside of work. With older and their needs changed he became involved regard to demographics, baseline characteristics about with them. surgeons’ earlier marriages, number of years married or These quotations are representative of common per- committed in and out of training, marriage status at ceptions of spouses of surgeons today in terms of lifestyle time of career choice, family support, number of chil- issues, particularly time management and balancing dren, and spousal employment were obtained.The ques- work and family life. Although there has been much tionnaire was piloted with spouses of faculty at the Uni- recent emphasis on quality-of-life issues during training versity of Texas at Houston, and revisions were made with the advent of the 80-hour work week, little has based on input received. been written about the impact of work hours and job Statistical analysis was performed using a statistical stressors on personal satisfaction after training, espe- software program (NCSS Statistical Software). Descrip- cially from a spouse’s perspective. Given that spouses tive statistics were reported as mean with standard devi- often provide both emotional and financial support durations for normally distributed data and medians with ing and after surgical training, the factors contributing interquartile ranges for nonnormally distributed data. to spousal satisfaction and the impact on surgeon career Univariate analysis of categorical variables was per- satisfaction are important to elucidate. Such informa- formed using the chi-square or Fisher’s exact probability tion is essential for counseling of medical students and test, and continuous variables were analyzed using the Ͻ their partners in terms of future expectations. Using data Student’s t-test or Mann-Whitney U test.Apvalue gathered from a national survey of academic surgeons’ 0.05 was considered statistically significant. spouses, this study aims to define the factors contribut- RESULTS ing to surgeon satisfaction after training from a spouse’s standpoint and to elucidate the role that spouses and General characteristics partners play in surgeon career satisfaction. Completed surveys were received from 379 spouses for a response rate of 27%. Average age of the respondents was 46 Ϯ 10 years, and 90% of the respondents were METHODS women. Forty-eight percent of respondents were mar- A national survey of spouses of academic surgeons cur- ried or committed when their spouses made their career rently used in a department of surgery in the United choice, and of those, 45% agreed or strongly agreed with States or Canada was conducted. The study was re- surgery as a career choice for their spouses. Fifty-four viewed and approved by our Institutional Review Board. percent of respondents worked outside the home, and Subsequently, all surgery chairpersons (n ϭ 143) were 17% were physicians as well. Ninety percent of respon- contacted about their willingness to participate in the dents had children. 686 Kao et al Spousal Perceptions of Surgeon Satisfaction J Am Coll Surg

Eighty-one percent of respondents were satisfied with and income did not appear to be a determinant of spou- their spouse’s career choice, but given the opportunity, sal or surgeon satisfaction. 13% of respondents would have encouraged their spouse to choose a different career knowing what they Positive comments know now. When asked what the single most important Despite negative comments, 88% of respondents agreed factor detracting from their spouses’ careers was, respon- that their spouses were satisfied with their careers and dents most frequently cited: work hours or call-related 81% of respondents were satisfied with their surgeon issues (42%); practice limitations and issues (18%); re- spouses’ careers. Only 12% of respondents felt that their imbursement policies and income (12%); and malprac- spouses would have chosen a different career in hind- tice or insurance issues (7%). Other negative factors sight, and, similarly, 13% of respondents would have listed included hospital politics, academic or administra- urged their spouses to choose differently. When asked to tive responsibilities, relationships with colleagues or an- respond freely about their perceptions as to the best as- cillary staff, and stress. pect of their spouses’ careers, respondents listed: personal satisfaction or love of surgery; the challenging na- Work hours ture of the job; altruism; financial security; lifestyle Overall, when questioned about their perceptions on issues in terms of flexibility or travel; and teaching resi- their spouses’ work hours, 60% of respondents per- dents and medical students. ceived that their spouses work more than 80 hours per week, and 64% of respondents overall perceived their spouses to work too many hours. On the whole, 65% of Surgeon satisfaction respondents agreed that they would be happier if their Three hundred thirty-four spouses (88%) perceived that spouses worked less and 53% perceived that their their surgeon partners were satisfied with their career spouses would be happier. Less than half of respondents choice, as indicated by a Likert score of 4 or 5 in response agreed or strongly agreed with government mandates to the statement, “Overall, my spouse is satisfied with restricting resident work hours (47%) and about work- his/her career choice.” Twenty-nine spouses (8%) dis- hour limitations for practicing surgeons (41%). Only agreed with this statement. There were no statistically 26% of respondents perceived that their spouses work significant differences between the two groups, although more now than during training. there was a trend that surgeons who were perceived to be dissatisfied were more likely to have been married previ- Personal and family issues ously (p ϭ 0.11) and to have a spouse who worked more Overall, 74% of respondents agreed (Likert scale 4 or than 40 hours per week outside the home (p ϭ 0.12) or 5) that they made the majority of domestic decisions to be in the health-care profession (p ϭ 0.13). in the household. Less than half of respondents felt In terms of career decisions, surgeons perceived to be that their surgeon spouses contributed adequately to dissatisfied were more likely to have spouses who were household responsibilities (48%) or child care (37%). less involved in career decisions (Table 1). Although re- Thirty-eight percent of respondents felt that their spondents of dissatisfied surgeons disagreed less strongly, spouses spent adequate time with their children and overall most respondents did not perceive that their sur- 38% agreed that their spouses spent adequate time geon spouses would be happier in a different career. with them. A majority (69%) agreed that their There were no differences between spousal perceptions spouses made an effort to spend time on activities about work hours or income expectations between sur- outside of work-related obligations. geons who were perceived as satisfied versus those perceived as dissatisfied. There were no differences in per- Practice issues, malpractice, and income ceived contributions of surgeon spouses to household or The remainder of negative comments about surgeon and child-care issues and no differences in quality time spent spousal satisfaction dealt with practice limitations, in- with their partners or children. Respondents of dissatis- surance, malpractice, and income. With regard to the fied surgeons were much more likely to agree that both latter, almost three-quarters of respondents (74%) they and their spouses would have chosen a career dif- agreed that their spouses’ income met their expectations, ferently had they known what they know now. Vol. 200, No. 5, May 2005 Kao et al Spousal Perceptions of Surgeon Satisfaction 687

Table 1. Perceptions of Satisfied (4 or 5) Versus Dissatisfied (1 or 2) Spouses and Surgeons (as Perceived by Their Spouses) Spouses Surgeons Satisfied Dissatisfied Satisfied Dissatisfied (20 ؍ n) (334 ؍ n) (29 ؍ n) (309 ؍ n) Career choice Surgery was my spouse’s first choice for a career choice. 4.6 Ϯ 1.0 4.9 Ϯ 0.3 4.6 Ϯ 1.0 5.0 Ϯ 0.2 Surgery was my first choice for my spouse’s career choice. 3.7 Ϯ 1.2 3.5 Ϯ 1.1 3.6 Ϯ 1.3 3.5 Ϯ 1.0 My spouse’s career specialty choice was a joint decision. 2.7 Ϯ 1.5 2.6 Ϯ 1.4 2.6 Ϯ 1.5 2.4 Ϯ 1.1 My spouse and I usually agree about decisions regarding his or her career. 4.2 Ϯ 0.9* 3.6 Ϯ 1.2* 4.1 Ϯ 1.0† 3.6 Ϯ 1.0† My spouse and I usually make decisions about his/her career together. 3.9 Ϯ 1.1* 2.9 Ϯ 1.5* 3.8 Ϯ 1.2† 3.1 Ϯ 1.3† My spouse would be happier in a different career. 1.6 Ϯ 0.9* 2.1 Ϯ 1.2* 1.5 Ϯ 0.8‡ 2.5 Ϯ 1.3‡ Work hours My spouse works on average more than 80 hours per week. 3.6 Ϯ 1.4 3.7 Ϯ 1.5 3.7 Ϯ 1.4 3.7 Ϯ 1.4 My spouse works too many hours. 3.8 Ϯ 1.1 4.0 Ϯ 1.0 3.8 Ϯ 1.1 3.8 Ϯ 1.1 I would be happier if my spouse worked fewer hours. 3.8 Ϯ 1.0 3.8 Ϯ 1.1 3.9 Ϯ 1.0 3.6 Ϯ 0.9 My spouse would be happier if he or she worked fewer hours. 3.5 Ϯ 1.2 3.4 Ϯ 1.3 3.5 Ϯ 1.2 3.4 Ϯ 1.2 My spouse’s time commitment to his or her work is more than what I originally expected. 111/302 (37) 13/28 (46) 133/328 (41) 7/20 (35) My spouse currently works more than when he or she was in training. 2.5 Ϯ 1.3 2.8 Ϯ 1.7 2.6 Ϯ 1.3 2.7 Ϯ 1.6 Income My spouse’s career meets our financial expectations. 4.1 Ϯ 2.4 3.7 Ϯ 1.5 4.1 Ϯ 2.3 3.7 Ϯ 1.3 Geography and support Geographic location was an important factor in choosing my spouse’s current position. 3.5 Ϯ 1.3 3.3 Ϯ 1.6 3.4 Ϯ 1.3 3.5 Ϯ 1.5 I like the location (geographic) where we live. 4.2 Ϯ 1.0 3.7 Ϯ 1.3 4.1 Ϯ 1.1 3.8 Ϯ 1.2 I have a strong support network in the city/location where we currently live. 3.8 Ϯ 1.3 3.8 Ϯ 1.3 3.7 Ϯ 1.3† 4.3 Ϯ 1.1† I have family members in the city/location where we currently live. 2.6 Ϯ 1.8 2.6 Ϯ 1.7 2.6 Ϯ 1.7 2.4 Ϯ 1.6 Household and family I make the majority of domestic decisions in our household. 4.0 Ϯ 1.1 4.1 Ϯ 1.1 4.0 Ϯ 1.1 3.9 Ϯ 1.2 My spouse contributes sufficiently to household responsibilities. 3.3 Ϯ 1.2§ 2.7 Ϯ 1.5§ 3.2 Ϯ 1.2 2.8 Ϯ 1.3 My spouse participates adequately in child care. 3.3 Ϯ 1.3§ 2.7 Ϯ 1.5§ 3.3 Ϯ 1.3 3.1 Ϯ 1.0 My spouse spends enough quality time with our children. 3.2 Ϯ 1.2§ 2.6 Ϯ 1.5§ 3.1 Ϯ 1.3 2.7 Ϯ 1.3 My spouse spends enough quality time with me. 3.1 Ϯ 1.2 2.7 Ϯ 1.4 3.0 Ϯ 1.1 2.8 Ϯ 1.1 My spouse makes an effort to spend time on activities outside job requirements/obligations. 3.8 Ϯ 1.1 3.4 Ϯ 1.4 3.7 Ϯ 1.1 3.6 Ϯ 1.3 If I knew then what I know now, I would have urged my spouse to choose a different career. 1.9 Ϯ 1.0* 3.1 Ϯ 1.6* 2.0 Ϯ 1.1‡ 3.3 Ϯ 1.6‡ If my spouse knew then what he or she knows now, he or she would have chosen a different career specialty. 1.8 Ϯ 1.0* 2.6 Ϯ 1.6* 1.7 Ϯ 0.9‡ 3.3 Ϯ 1.6‡ Values are mean Ϯ SD unless otherwise noted. Continuous variables (representing the Likert responses on a scale from 1 to 5) are reported as mean Ϯ SD. Categoric variables are reported as proportions answering “yes” (and percentages). *p Value Ͻ0.01 between satisfied and dissatisfied spouses. †p Value Յ0.05 between satisfied and dissatisfied surgeons. ‡p Value Ͻ0.01 between satisfied and dissatisfied surgeons. §p Value Յ0.05 between satisfied and dissatisfied spouses. 688 Kao et al Spousal Perceptions of Surgeon Satisfaction J Am Coll Surg

Spousal (respondent) satisfaction spouses, there were no differences in baseline character- In terms of the 307 respondents (81%) who agreed or istics. There were no significant differences in work strongly agreed (4 or 5 on the Likert scale) with the hours, family involvement, or financial expectations. statement, “Overall, I am satisfied with my spouse’s ca- The only difference was in level of perceived spousal reer choice” and the 29 (7%) who disagreed or strongly impact on decision making. disagreed (1 or 2 on the Likert scale), there were no Despite the fact that work hours was cited by respon- statistically significant differences in age or gender. Sat- dents to our survey as the largest job-related detractor isfied and dissatisfied spouses were similar in terms of from surgeon satisfaction, there was no difference be- years of marriage or commitment, number of children, tween respondents of perceived satisfied versus dissatis- and level of employment. fied surgeons with regard to time-related issues. Issue of In general, respondents desired that their surgeon work hours has previously been addressed in another spouses work less, but work hours did not appear to spousal survey. In the largest study of physician spouses, influence respondents’ satisfaction. Although initial ca- Sotile and Sotile12 conducted a survey of 603 members’ reer specialty choice did not appear to be a point of wives of the American Medical Association Alliance. In contention between satisfied and dissatisfied spouses, this study, 28 different medical specialties were repre- the latter were significantly less likely to agree with their sented. Average number of hours worked per week by surgeon spouses’ career decisions and less likely to be physician spouses was 59.7, and there was noted to be a involved in the decision-making process currently. In correlation between number of hours worked and level addition, dissatisfied spouses were less likely to like the of satisfaction of the wives with the marriage and with geographic location where they lived, although this dif- their life in medicine. The authors suggested that the ference did not seem to be related to a lack of a support success of a medical marriage was more a function of the system in that location. Last, significant differences ex- quality of spousal interactions than of hours worked by isted in terms of surgeon involvement with household the physician husbands.12 and child-care responsibilities—dissatisfied respondents Other studies have demonstrated a lack of correlation were much more likely to have less-involved spouses. between hours worked and marital satisfaction as well. Gabbard and colleagues10 performed a survey of 134 DISCUSSION physicians and 125 physicians’ spouses in which they For medical graduates, work hours and lifestyle issues are determined that number of hours worked (40 to 50 becoming an increasing deterrent to choosing a surgical hours per week versus 60 or more hours) did not relate to specialty,6,7 a trend that is reflected in the declining number marital satisfaction. In another study of 415 married of applicants for surgical residencies over the past decade.8 physicians with children, work hours per se were not These concerns about time commitment and lifestyle are correlated to parental or marital satisfaction.2 The au- also shared by spouses and have an impact on career selec- thors postulated instead that role conflict, defined as tion among medical students.3 The additional influence of “the perceived frustration resulting from the competing spouses on career satisfaction is reflected by the fact that demands of career, marriage, and family,” was an inter- marital satisfaction has been linked to physician work sat- vening variable between multiple factors that included isfaction.1 Although there have been several studies that work hours and marital satisfaction.2 have solicited spousal opinions about physician marital and In our study, 60% of respondents agreed that their career satisfaction,1,9-11 few studies have focused on spousal spouses work more than 80 hours a week. Although over perceptions about work satisfaction as pertaining to sur- 50% of respondents agreed that either they or their geons specifically. spouses would be happier with fewer weekly work hours, According to our survey, satisfied and dissatisfied there was no difference between satisfied and dissatisfied spouses did not differ in baseline characteristics. A dis- spouses with their responses to these statements. There satisfied spouse was less likely to have input with regard was a significant influence of surgeon spouses’ degree of to career decision making and more likely to be unhappy involvement in decision making and participation in with the level of involvement of their surgeon spouse household and child-care matters on respondent satis- with regard to household and child-care activities. As far faction. Results of our survey agree with those from pre- as satisfied and dissatisfied surgeons, as perceived by the vious studies demonstrating that quality of spousal in- Vol. 200, No. 5, May 2005 Kao et al Spousal Perceptions of Surgeon Satisfaction 689 teraction and involvement is more important than work of the larger groups of surgeon spouses to have been hours in determining spousal satisfaction. studied, and the results will provide useful preliminary Although the issue of malpractice was listed as the data for future studies about spousal issues and surgeon main detractor in an academic surgical practice by only a satisfaction. minority of respondents (7%), medical-legal issues are In conclusion, although there are a number of detrac- an increasing problem faced by physicians today and can tors associated with a career in academic surgery, most have a significant impact on physician satisfaction. The surgeons as perceived by their spouses are satisfied with results of a recent survey by Mello and colleagues13 of their careers. Given the body of the literature that sup- 824 Pennsylvania physicians suggested that the liability ports the influence of a supportive spouse and of marital crisis in that state is having a negative impact on physi- satisfaction on physician work satisfaction, the predic- cian satisfaction. Compared with results of a national tors of spousal satisfaction are important to determine. sampling, 40% of Pennsylvania surgeons sampled in In keeping with the results of previous studies, work 2003 were either somewhat or very dissatisfied with hours per se are not a major determinant of spousal or their career as compared with 19% and 20% nationally surgeon satisfaction, but rather the level of spousal infor 1999 and 2001, respectively.13 In Mello’s survey, volvement in decision making and the level of physician 86% of physicians had been named in a malpractice suit participation in household and child-care activities im- at least once during their careers, which is comparable to pact spousal satisfaction the most. Medical students in- the national rate for surgeons. In a survey of academic terested in surgery as a career should be encouraged to do and private surgeons by Schroen and colleagues,14 more so with realistic expectations but without fear of the than 75% of surgeons had been named in a malpractice impact of time commitment on their family lives. suit at least once in their careers; there was no significant difference in malpractice experiences between primary Author Contributions and academic surgeons. Because the survey instrument Study conception and design: Kao, Wilson, Anderson did not specifically look at malpractice as a predictor of Acquisition of data: Wilson, Anderson surgeon or spousal satisfaction, the impact of medical- Analysis and interpretation of data: Kao, Wilson legal pressures on surgeon satisfaction could not be Drafting of manuscript: Kao assessed. Critical revision: Kao, Wilson, Anderson One limitation of this study is that only surgeons’ Statistical expertise: Kao spouses or partners were surveyed. No correlation was Obtaining funding: Anderson made between spouses’ and the surgeons’ perceptions. Supervision: Anderson And there may be a selection bias in terms of only spouses who were still currently married or whose partners were still in academic surgery were queried; so the REFERENCES study may overestimate the percentage of surgeons and 1. Lewis JM, Barnhart FD, Nace EP, et al. Marital satisfaction in spouses who are satisfied. Second, only spouses of aca- the lives of physicians. Bull Menninger Clin 1993;57:458–465. demic surgeons were surveyed. A recent study compared 2. Warde CM, Moonesinghe K, Allen W, Gelberg L. Marital and parental satisfaction of married physicians with children. J Gen academic and private surgeons with regard to their career Intern Med 1999;14:157–165. satisfaction and did not find a significant difference be- 3. Valente J, Rappaport W, Neumayer L, et al. Influence of spousal tween the two.14 Third, a large limitation of the study opinions on residency selection. Am J Surg 1992;163:596–598. 4. Dorsey ER, Jarjoura D, Rutecki GW. Influence of controllable was the low response rate of 27%. Difficulty in identi- lifestyle on recent trends in specialty choice by US medical stu- fying and contacting physician spouses and in perform- dents. JAMA 2003;290:1173–1178. ing followup reminders about survey completion given 5. Harder F.I would like to be a surgeon, but. Ann Surg 2002;236: 699–702. the costs and practical challenges may have contributed 6. Marschall JG, Karimuddin AA. Decline in popularity of general to this low response rate. Last, there may be other factors surgery as a career choice in North America: review of postgrad- that were not measured by the survey instrument that uate residency training selection in Canada, 1996Ϫ2001. World may impact on spousal and surgeon satisfaction. None- J Surg 2003;27:249–252. 7. Bland KI, Isaacs G. Contemporary trends in student selection of theless, despite the limitations, the number of respon- medical specialties: the potential impact on general surgery. dents to the survey was significant in terms of being one Arch Surg 2002;137:259–267. 690 Kao et al Spousal Perceptions of Surgeon Satisfaction J Am Coll Surg

8. Newton DA, Grayson MS. Trends in career choice by US med- 12. Sotile WM, Sotile MO. Physicians’ wives evaluate their mar- ical school graduates. JAMA 2003;290:1179–1182. riages, their husbands, and life in medicine: results of the AMA– 9. Spendlove DC, Reed BD, Whitman N, et al. Marital adjust- Alliance Medical Marriage Survey. Bull Menninger Clin 2004; ment among housestaff and new attorneys. Acad Med 1990;65: 68:39–59. 599–603. 13. Mello MM, Studdert DM, DesRoches CM, et al. Caring for 10. Gabbard GO, Menninger RW, Coyne L. Sources of conflict in patients in a malpractice crisis: physician satisfaction and quality the medical marriage. Am J Psychiatry 1987;144:567–572. of care. Health Aff (Millwood) 2004;23:42–53. 11. Fabri PJ, McDaniel MD, Gaskill HV 3rd, et al. Great expec- 14. Schroen AT, Brownstein MR, Sheldon GF. Comparison of pri- tations: stress and the medical family. 1987 Committee on vate versus academic practice for general surgeons: a guide for Issues, Association for Academic Surgery. J Surg Res 1989; medical students and residents. J Am Coll Surg 2003;197: 47:379–382. 1000–1011. Characterization of Human Nasal Septal Chondrocytes Cultured in Alginate

Stanley H Chia, MD, Mark R Homicz, MD, Barbara L Schumacher, BS, Eugene J-MA Thonar, PhD, Koichi Masuda, MD, Robert L Sah, MD, SCD, Deborah Watson, MD, FACS

BACKGROUND: After serial passages in monolayer, chondrocytes dedifferentiate into a fibroblast-like phenotype. Our objective was to determine if culture in alginate affects the phenotype of dedifferentiated human nasal septal chondrocytes. STUDY DESIGN: Human nasal septal chondrocytes were seeded at low density and passaged in monolayer culture. At passages (P) 1, 2, and 3 a portion of cells were cultured in alginate. Collagen, glycosaminoglycan (GAG), and DNA production were assessed. RESULTS: Chondrocytes in alginate proliferated less yet produced higher levels of GAG and collagen than those in monolayer culture. Alginate encapsulated P1 chondrocytes stained strongly for GAG and collagen type II, and minimally for collagen type I. Monolayer cells at P0 and P1 stained positively for collagen type II. All monolayer passages stained positive for collagen type I with minimal GAG staining. CONCLUSIONS: Compared with monolayer culture, alginate stimulates deposition of GAG and collagen type II, and supports the chondrocyte phenotype through P1, but does not promote redifferentiation. (J Am Coll Surg 2005;200:691–704. © 2005 by the American College of Surgeons)

Nasal septal cartilage is widely used in facial plastic and the focus of recent research because of its possible appli- reconstructive surgery. Advantages over other cartilage cations in craniofacial reconstruction and facial plastic donor sites (eg, costal, auricular, articular) include ease surgery. Through expansion of cell numbers in culture, and low morbidity of harvest, and relative mechanical there is theoretically no limit to the amount of cartilage stability of septal cartilage. Its use is limited by the rela- that can be grown to reconstruct even very large cranio- tively small size of the nasal septum, particularly in sec- facial defects. In addition, by growing cells on biode- ondary procedures where the septum has been harvested gradable scaffolds, the neocartilage can be potentially previously. grown in any desired shape and size. Tissue engineering of nasal septal cartilage has been Cell populations are typically expanded in vitro in monolayer culture. Previous studies have demonstrated that multiplication of articular chondrocytes in mono- Competing interests declared: None. Portions of this project were supported by grants from the AAFPRS Research layer leads to dedifferentiation, in which cells lose their Foundation, the Arthritis Foundation, NASA, NIH (AG07996, AR44058, chondrocyte phenotype.1,2 These dedifferentiated cells AR46555, AG04736, AR39239, AR48152), and the National Science Foun- dation. assume a fibroblast-like appearance and produce fewer Presented at the Annual Meeting of the American Academy of glycosaminoglycans. The cells also produce less cartilage- OtolaryngologyϪHead and Neck Surgery Foundation, San Diego, CA, Sep- specific collagen type II, and greater amounts of collagen tember 2002. type I. Received October 4, 2004; Revised January 11, 2005; Accepted January 11, 2005. Studies involving articular chondrocytes have demon- From the Division of Head and Neck Surgery, University of California, San strated that when dedifferentiated cells are suspended in Diego, and San Diego Veterans Affairs Healthcare System, San Diego, CA (Chia, Homicz, Watson); Department of Bioengineering and Whitaker In- agarose gel, they reexpress differentiated chondrocyte stitute of Biomedical Engineering, University of California, San Diego, San phenotype. This is likely because of the maintenance of Diego, CA (Schumacher, Sah); Departments of Biochemistry, Internal Med- icine, and Orthopedic Surgery, Rush Medical College, Chicago, IL (Thonar, a spherical shape in agarose, as opposed to the flattened Masuda). shape in monolayer culture.3 Excellent results have been Correspondence address: Deborah Watson, MD, FACS, Division of Head and Neck Surgery, Veterans Affairs Hospital, 3350 La Jolla Village Dr, 112C, reported recently using articular chondrocytes encapsu- San Diego, CA 92161. lated in a three-dimensional matrix within alginate

the University of California at San Diego and the San Abbreviations and Acronyms Diego Veterans Affairs Health care System. Four volun- EDTA ϭ ethylenediaminetetraacetic acid teers (mean age Ϯ SD 43.7 Ϯ 13.65 years) consented to ϭ GAG glycosaminoglycan donate their septal cartilage for research use after its re- P ϭ passage moval during routine septoplasty. Cartilage was placed in normal saline solution at 4°C until ready for use. beads.4,5 Alginate is a copolymer of L-guluronic acid and Separate chondrocyte cultures were established from D-mannuronic acid that polymerizes to form a gel in the each donor cartilage specimen. A schema for the exper- presence of calcium. Advantage of alginate over agarose imental design is seen in Figure 1. is that alginate can be depolymerized easily with addition of a calcium chelator such as sodium citrate or eth- Cartilage digestion and chondrocyte isolation ylenediaminetetraacetic acid (EDTA). This allows re- Within 48 hours of procurement, each cartilage specimen lease of the chondrocytes and their associated matrix. was dissected free of perichondrium and diced into pieces ϳ 3 Agarose, in contrast, must be treated with dissociative ( 1mm). Digestion of these fragments was achieved by solvents to release the majority of proteoglycans. incubation at 37°C in 0.2% Pronase type XIV (Sigma, Previously, we demonstrated superior glycosamino- P-5417) in medium (DMEM [Dulbecco’s Modified Eagle glycan (GAG) accumulation by human nasal chondro- Medium; HyClone]/F-12, 5% fetal bovine serum, 0.4 cytes when encapsulated in alginate immediately after mmol/L L-proline, 2 mmol/L L-glutamine, 0.1 mmol/L chondrocyte isolation, compared with culture in polyg- nonessential amino acids, 10 mmol/L HEPES buffer, 100 ␮ lycolic acid scaffolds or in monolayer.6 No previous stud- U/mL penicillin G, 100 g/mL streptomycin sulfate, 0.25 ␮ ies have been performed with human nasal septal chon- g/mL amphotericin B) for 60 minutes followed by incu- drocytes to characterize their ability to redifferentiate in bation with 0.025% collagenase P (Roche Diagnostics, alginate after successive passages in monolayer. The ob- #1249-002) in medium for 12 to 15 hours. ␮ jective of this study is to determine if culture in alginate Suspensions of digested cartilage were filtered (70 m), after serial monolayer passages affects the phenotype of then washed and centrifuged to isolate the chondrocytes. human nasal septal chondrocytes. Cells were resuspended in cell culture medium (DMEM [low glucose], 10% fetal bovine serum, 25 ␮g/mL ascor- METHODS bate, 0.4 mmol/L L-proline, 2 mmol/L L-glutamine, Approval for harvest and use of human nasal septal car- 0.1 mmol/L nonessential amino acids, 10 mmol/L HEPES tilage was granted by the Institutional Review Boards of buffer, 100 U/mL penicillin G, 100 ␮g/mL streptomycin

Figure 1. Chondrocytes are harvested from human septal cartilage, then expanded in monolayer with serial passages. Cells from each passage are subsequently cultured either in monolayer or alginate for 14 days, then terminated for analysis. Vol. 200, No. 5, May 2005 Chia et al Human Nasal Septal Chondrocytes 693 sulfate, 0.25 ␮g/mL amphotericin B). Number of chon- acids, 100 U/mL penicillin G, 100 ␮g/mL streptomycin drocytes isolated was determined by hemocytometer sulfate, 0.25 ␮g/mL amphotericin B) was changed daily. counting after trypan blue exclusion. Passage 2: seeding of monolayer and alginate Passage 0: seeding of monolayer cultures for cell At confluency of the T-75 monolayer culture (P1), cells population expansion were trypsinized and counted as described. Cells were For each patient, isolated chondrocytes were seeded at seeded into a T-75 flask, 2 T-25 flasks, 4 8-well chamber low density (10,000 cells per cm2 surface area) into a slides, and alginate beads (designated P2) as stated in the T-75 flask, 2 T-25 flasks, and each well of 4 8-well description of P1. A portion of alginate beads was ter- Laboratory-Tekchamber slides (Nalge Nunc) for mono- minated at day 0, and culture of 1 T-25 flask was termi- layer culture growth at 37°C. These chondrocytes were nated at day 1 for quantitative DNA analysis. Culture of designated passage zero (P0) cells. Culture medium was the remaining T-25 flask, chamber slides, and alginate changed every other day. Culture of 1 T-25 flask and beads was carried out to 14 days. Culture of the T-75 chamber slides was carried out for 14 days. Culture of flask was carried out to confluency. the remaining T-25 flask was terminated at day 1 for quantitative DNA analysis. Passage 3: seeding of monolayer and alginate The T-75 flask was cultured until cells grew to con- At confluency of the T-75 monolayer culture (P2), cells fluency (6 to 8 days). Confluent cells were released from were trypsinized and counted. Cells were seeded into 2 monolayer using 0.05% trypsin (Life Technologies, T-25 flasks, 4 8-well chamber slides, and alginate beads #25095-019). After centrifugation and removal of the (P3) as stated in the description of P1. A portion of alginate beads was terminated at day 0 and 1 T-25 was supernatant, chondrocytes were counted using a hemo- terminated at day 1 for quantitative DNA analysis. Re- cytometer. These cells were designated passage 1 (P1). maining cultures were carried out to 14 days.

Passage 1: seeding of monolayer and alginate Termination of monolayer cultures The chondrocytes designated P1 after a single passage in At day 1, designated monolayer cells in T-25 flasks were monolayer were then either seeded back into monolayer digested with 0.5 mg/mL Proteinase K (Roche Diagnos- culture or encapsulated in alginate as described here. tics) in phosphate buffered ethylenediaminetetraacetic A portion of the chondrocytes was seeded into 2 T-25 acid for 1 hour at 37°C to release cells from the flask.The flasks, a T-75 flask, and each well of 4 8-well chamber solution was then heated at 60°C for 16 hours to com- 2 slides at cell density 10,000 cells per cm for monolayer plete digestion. Appropriate T-25 flasks were terminated culture. Cell culture in the T-75 flask was carried out to at day 14 in an identical manner. confluency, while culture of 1 T-25 flask and the cham- At culture day 14, media from monolayer cultures in ber slides was carried out to 14 days. Culture of the chamber slides were removed and cells were washed with remaining T-25 flask was terminated at day 1 for quan- 1ϫ PBS. Cells were then fixed by addition of 4% para- titative DNA analysis. Culture medium was changed formaldehyde for 10 minutes. Slides were washed in 1ϫ every other day. PBS and stored hydrated in 1ϫ PBS at 4°C until ready P1 chondrocytes from each patient were also sus- for use. pended at a density of 2 ϫ 106 cells/mL in a solution of 1.2% low viscosity alginate (Kelco LV) in 0.15 M NaCl. Termination of alginate cultures Droplets of the alginate/chondrocyte solution were po- Alginate beads were prepared for DNA assay by treatment lymerized in 102 mmol/L CaCl2 to form solid beads with 0.02% sodium dodecyl sulfate at 100°C for 10 min- (approximate volume ϭ 10 mm3 per bead). A small utes, followed by digestion with 0.5 mg/mL Proteinase K in portion of the beads was removed for DNA analysis. phosphate buffered ethylenediaminetetraacetic acid and Remaining beads were cultured in 6-well plates for 14 depolymerization of alginate (55 mmol/L sodium citrate, days. Culture medium (DMEM/F-12, 10% fetal bovine 30 mmol/L EDTA, 0.15 mmol/L NaCl). serum, 25 ␮g/mL ascorbate, 0.4 mmol/L L-proline, Beads were prepared for the GAG and hydroxyproline 2 mmol/L L-glutamine, 0.1 mmol/L nonessential amino assays by depolymerization (as described previously), 694 Chia et al Human Nasal Septal Chondrocytes J Am Coll Surg followed by centrifugation and separation of the super- Quantitative assay for collagen (hydroxyproline) natant (containing far-removed matrix) and pellet (con- Collagen content in monolayer samples was estimated taining cell-associated matrix). The pellet was then di- from analysis of hydroxyproline.10 Digests were hydrolyzed gested with 0.5 mg/mL Proteinase K. in 12 N HCl at 110°C for 18 hours. HCl was then evapo- In preparation for immunohistologic analysis, algi- rated and samples reacted with chloramine T reagent and nate beads were depolymerized as described, followed by p-dimethylaminobenzaldehyde at 60°C for 20 minutes. centrifugation and removal of the supernatant. Cells Absorbance at 560 nm of assay solutions was measured and were washed and resuspended in 1 mL 1ϫ PBS. Fifty calibrated with standards of L-4-hydroxyproline (Sigma, microliters of cell suspension was placed onto poly-L- H-6002) to determine hydroxyproline content. lysineϪcoated slides and allowed to dry overnight. Collagen content in far-removed and cell-associated Slides were then fixed with 4% paraformaldehyde and fractions of alginate was determined using a modification stored in 1ϫ PBS at 4°C until ready for use. of the previously mentioned technique.11 Content of hydroxyproline was measured by phenylthiocarbamyl deriva- Quantitative assay for DNA tization and isocratic reverse-phase high-performance liq- Quantitative analysis of DNA content in monolayer and uid chromatography. Samples were hydrolyzed in 6 M alginate samples was performed using Hoechst 33258.7 hydrochloric acid for 16 hours at 120°C, evaporated to Fluorescence (excitation 365 nm/emission 458 nm) of dryness, and the residue dissolved in methanol:water:tri- mixtures of 100 ␮L digested samples with 1.0 mL ethanolamine (2:2:1) and redried twice. After derivatiza- Hoechst 33258 dye solution (0.1 ␮g/mL Hoechst tion by phenylisothiocyanate, samples were separated iso- 33258 dye [Sigma, B-2883] in 10 mmol/L Tris, 1 mmol/L cratically using a reverse-phase C18 octadecylsilane column EDTA, and 200 mmol/L NaCl, pH 7.5) was determined (25 cm ϫ 4.6-mm internal diameter; Beckman Ultrasphere) in duplicate. Fluorescence values were converted to and monitored on an absorbance detector at 254 nm. The DNA equivalents using standards of DNA from calf mobile phase was composed of acetonitrile:water:140 thymus (Sigma, D-1501) in the appropriate buffer solu- mmol/L CH3COONa buffer (6:4:90, v/v/v), pH 6.4 with tion. DNA content of digests was converted to cellular- 0.5 mL/L triethylamine. Concentrations of hydroxypro- ity using the estimated value of 7.7 pg DNA per line are reported as equivalents of external standards of chondrocyte.7 hydroxyproline. Collagen content was estimated using this relation: Quantitative assay for GAG 1 g hydroxyproline present in 7.1 g collagen.12-14 Colla- GAG content in monolayer culture and the cell- gen content in all samples was normalized to DNA (␮g associated compartment of alginate beads was deter- collagen/␮g DNA). mined using dimethylϪmethylene blue reaction.8 GAG content of the far-removed and cell-associated matrix Live/dead assay on alginate compartments of alginate beads was independently de- At the end of 14 days culture in alginate, several beads termined by dimethylϪmethylene blue reaction and were placed in solution containing 1% calcein (Molec- then added to obtain total GAG values. In the far- ular Probes) and 1% ethidium homodimer-1 (Molecu- removed compartment, alginate created interference lar Probes) in 1ϫ PBS for 40 minutes to assess cell via- when measuring spectrophotometric absorbance. To bility. In this assay, viable cells demonstrate green amplify the signal attributable to GAG and reduce noise fluorescence and dead cells demonstrate red fluores- caused by alginate in samples from this compartment, cence. Approximate cell viability in beads was assessed the trough in absorbance (at 590 nm) was subtracted under fluorescent microscopy. from the peak in absorbance (at 520 nm) to determine the GAG values in the far-removed compartment.9 Histochemistry Comparison of all absorbance values to those of stan- Immunohistochemistry of sections from the alginate dards made from shark chondroitin sulfate type C and monolayer cultures was performed using the Vectastain (Sigma, C-4384) yielded GAG content. GAG content Elite ABC kit (Vector Laboratories), a peroxidase-based de- was then normalized to DNA content (␮gGAG/␮g tection system. A mouse monoclonal anticollagen type I DNA). antibody (Sigma, #C2456) was used at 1:2,000 dilution Vol. 200, No. 5, May 2005 Chia et al Human Nasal Septal Chondrocytes 695

Figure 2. Passage 1 cells (a) encapsulated in alginate at culture day 14. Well-developed cell-associated matrix is visible surrounding chondrocytes (arrow). Passage 2 cells (b) encapsulated in alginate at culture day 14 exhibit rounded shape, but minimal cell-associated matrix.

for detection of collagen type I. A mouse monoclonal layers became hyperconfluent, with multiple layers of antitype II collagen antibody cocktail (Chondrex) was cell growth. On encapsulation in alginate, all cells re- used at 1:1,500 dilution for detection of collagen type II. gained a rounded phenotype that was maintained for the Slides were counterstained with methyl green nuclear duration of cell culture. After 14 days of culture, P1 stain (Vector Laboratories). For histochemical localiza- beads demonstrated cells with a well-defined cell- tion of GAG, slides from each sample group were associated matrix (Fig. 2a), although P2 (Fig. 2b) and P3 stained with 0.1% alcian blue powder (Sigma) in buffer beads lacked this characteristic. After 14 days of culture, (0.4 M MgCl2, 0.025 M NaAcetate, 2.5% glutaralde- all P1 beads demonstrated Ͼ95% cell viability, and P2 hyde, pH 5.6) overnight, and destained with 3% acetic and P3 beads demonstrated Ͼ90% viability. acid until clear. Slides were then observed under light Quantitative analysis of DNA by the Hoechst 33258 microscopy. assay revealed that cells encapsulated in alginate prolif- Statistical analysis erated less than those cultured in monolayer (Fig. 3). Results are expressed as fold increase in DNA content All data are expressed as mean Ϯ standard error of the (amount DNA at day 14 divided by initial DNA). After mean. Statistical analysis was performed with Systat 5.2 Ϯ (Systat, Inc). Differences between monolayer and algi- 14 days, cells in monolayer culture (P0: 77.34 15.15, P1: 52.90 Ϯ 6.30, P2: 59.27 Ϯ 6.62, P3: 54.90 Ϯ 7.12) nate culture at P1, P2, and P3 were assessed by a two- Ͻ way repeated measures ANOVA with two repeated fac- proliferated significantly more (p 0.01) than cells in Ϯ Ϯ tors of culture method (alginate versus monolayer) and alginate gel (P1: 1.90 0.16, P2: 1.49 0.11, P3: Ϯ passage (P1, P2, P3), followed by planned comparisons 1.25 0.14). P2 and P3 chondrocytes in alginate pro- Ͻ between groups. Differences between monolayers P0, liferated less than their P1 counterparts (p 0.01). P1, P2, and P3 were assessed by a one-way ANOVA with Values from quantitative GAG analysis were normalized ␮ ␮ repeated measures, followed by planned comparisons to DNA content ( gGAG/g DNA) (Fig. 4). The between passages. amount of GAG per DNA in monolayer cultures remained relatively stable through multiple passages (P0: 0.24 Ϯ RESULTS 0.03 ␮g/␮g, P1: 0.20 Ϯ 0.04 ␮g/␮g, P2: 0.19 Ϯ 0.02 ␮g/ Cells cultured in monolayer developed a spindle-shaped, ␮g, P3: 0.29 Ϯ 0.05 ␮g/␮g). In alginate beads, it decreased fibroblast-like phenotype. By 14 days of culture, mono- with successive passages (P1: 17.98 Ϯ 1.07 ␮g/␮g, P2: 696 Chia et al Human Nasal Septal Chondrocytes J Am Coll Surg

Figure 3. Comparison of cell proliferation between monolayer and alginate beads. Monolayer cells demonstrated significantly greater Figure 4. Average glycosaminoglycan (GAG) per DNA in monolayer proliferation than cells cultured in alginate at all passages (P). and alginate beads. Overall GAG accumulation was significantly Significant differences between groups and between passages are greater in alginate than in monolayer (p Ͻ 0.01). Significant differ- indicated with solid bars at the top of the graph. ences in values between groups and between passages (P) are indicated with black bars at the top of the graph. 5.96 Ϯ 3.05 ␮g/␮g, P3: 3.95 Ϯ 1.76 ␮g/␮g). GAG accu- Ϯ mulation in alginate was significantly greater than in accumulation with each successive passage (P0: 0.26 ␮ ␮ Ϯ ␮ ␮ Ϯ monolayer (p Ͻ 0.01). Decrease in GAG accumulation in 0.12 g/ g, P1: 1.07 1.59 g/ g, P2: 8.48 4.79 ␮ ␮ Ϯ ␮ ␮ alginate cultures from P1 to P3 was significant (p Ͻ 0.001), g/ g, 1.12 5.54 g/ g). Alginate cultures had sig- but there was a trend toward decrease from P1 to P2 (p ϭ nificantly greater collagen accumulation than mono- Ͻ 0.051). Distribution of GAG content in the far-removed layer overall (p 0.01). Figure 9 demonstrates the and cell-associated matrix compartments within the algi- breakdown of collagen in the far-removed and cell- nate gel is seen in Figure 5. GAG accumulation in both the associated portions of alginate. Collagen in the far- Ϯ ␮ ␮ Ϯ ␮ removed portion of alginate (P1: 2.91 Ϯ 0.52 ␮g/␮g, far-removed (P1: 7.29 1.38 g/ g, P2: 1.30 0.32 g/ Ϯ ␮ ␮ Ϯ ␮ ␮ ␮g, P3: 0.91 Ϯ 0.40 ␮g/␮g) and cell-associated compart- P2: 2.35 1.09 g/ g, P3: 0.92 0.17 g/ g) re- ments (P1: 10.69 Ϯ 1.75 ␮g/␮g, P2: 4.66 Ϯ 2.83 ␮g/␮g, P3: 3.04 Ϯ 1.46 ␮g/␮g) decreased from P1 to P2. Minimal change was noted from P2 to P3. Alcian blue staining of alginate bead sections is seen in Figure 6. At P1, significant staining is noted, indicating large amounts of GAG accumulation in the matrix surrounding chondrocytes. At P2 and P3 minimal staining is noted, indicating minimal GAG accumulation. Alcian blue staining in monolayer culture is seen in Figure 7. Minimal staining is noted at all passages, reflecting low levels of GAG accumulation. Results from quantitative analysis of collagen (through hydroxyproline assay) were similar to those of GAG. Col- lagen accumulation, expressed as ␮g collagen/␮g DNA, is shown in Figure 8. In alginate beads, collagen accumulation decreased with successive passage (P1: 3.66 Ϯ 0.88 ␮g/␮g, P2: 2.96 Ϯ 1.23 ␮g/␮g, P3: 1.39 Ϯ Figure 5. Glycosaminoglycan (GAG) per DNA in alginate beads, ␮ ␮ separated by portion. GAG accumulation in the far-removed and 0.21 g/ g). In monolayer cultures, on the other hand, cell-associated portions of the extracellular matrix decreased from it increased from P0 to P1, then had similar collagen passage (P) 1 to P2. Vol. 200, No. 5, May 2005 Chia et al Human Nasal Septal Chondrocytes 697

Figure 6. Alcian blue staining of cells released from alginate beads after 14 days culture. Passage 1 cells (a) demonstrate strong staining of cell-associated matrix, indicating large amounts of glycosaminoglycan (GAG) accumulation. Passage 2 (b) and passage 3 (c) cells show minimal staining, indicating little GAG accumulation. mained relatively stable from P1 to P2, but decreased by Staining for collagen type I (Fig. 11) revealed slightly P3. Collagen in the cell-associated portion (P1: 1.49 Ϯ positive staining at P1 in the pericellular region and 0.51 ␮g/␮g, P2: 0.61 Ϯ 0.17 ␮g/␮g, P3: 0.48 Ϯ slightly positive staining of the cells at P2 or P3. 0.1 ␮g/␮g) decreased from P1 to P2, but had little Immunohistochemical staining of monolayers is seen change from P2 to P3. in Figures 12 and 13. Collagen type II staining (Fig. 12) Immunohistochemical staining of cells released revealed positive staining at P0 in a fibrous pattern. The from alginate beads is seen in Figures 10 and 11. P1 monolayer demonstrated slightly positive staining, Staining for collagen type II (Fig. 10) revealed signif- although no staining was noted at P2 or P3. Collagen icant pericellular matrix staining at P1. Few type II- type I staining (Fig. 13) revealed positive staining in a positive cells are noted at P2, and none is noted at P3. fibrous pattern at all passages. 698 Chia et al Human Nasal Septal Chondrocytes J Am Coll Surg

Figure 8. Average collagen normalized to DNA in monolayer and alginate beads. Overall collagen accumulation was greater in alginate than in monolayer (p Ͻ 0.01). Signiﬁcant differences in values between groups and between passages (P) are indicated with solid bars at the top of the graph (p Ͻ 0.01) and dashed bars (p Ͻ 0.05). Figure 7. Alcian blue stain of passage 2 monolayer after 14 days culture. Minimal staining is noted, indicating little glycosaminoglycan accumulation. Similar ﬁndings are noted at all cell passages. aggrecan (a GAG produced by chondrocytes). In contrast, chondrocytes in monolayer have only been shown DISCUSSION to maintain their phenotype for a few weeks.19 Here we Tissue engineering of human cartilage invariably re- have described the culture characteristics of human nasal quires expansion of cells before neocartilage formation. chondrocytes in alginate beads after serial passage in These cell populations are typically expanded in mono- monolayer. layer culture. Previous studies1,2 have demonstrated that At P1, cells cultured in alginate exhibit characteristics articular chondrocytes lose their chondrocyte phenotype typical of chondrocytes. Grossly, these cells demonstrate after successive monolayer passages. These “dedifferentiated” cells produce lower amounts of GAGs and cartilage-specific collagen type II, and greater amounts of collagen type I. We have described similar results with neocartilage created from human nasal septal chondrocytes cultured in monolayer.15 In this study, septal chondrocytes passaged in monolayer were seeded onto polyglycolic acid scaffolds. With serial passages, cells exhibited progressively greater fibroblastic characteristics. Recently, alginate has been described as an effective culture medium for inducing redifferentiation of the chondrocyte phenotype in dedifferentiated articular chondrocytes.4,5 Articular chondrocytes that have dedifferentiated in monolayer produce high levels of collagen II and low levels of collagen I when encapsulated in alginate, even after several passages.16 These chondrocytes cultured in alginate have been shown to secrete a matrix that is similar to that seen in native human car- Figure 9. Collagen per DNA in alginate beads, separated by portion. Collagen in the far-removed portion of alginate remained relatively 17 tilage. Cells remain phenotypically stable for at least 8 stable from passage (P) 1 to P2, but decreased by P3. Collagen in months18 and continue to synthesize collagen type II and the cell-associated portion decreased from P1 to P2. Vol. 200, No. 5, May 2005 Chia et al Human Nasal Septal Chondrocytes 699

Figure 10. Collagen type II staining in cells released from alginate beads. Cells from passage 1 (a) strongly demonstrate positive staining and large cell-associated matrix. Cells from passages 2 (b) and 3 (c) demonstrate minimal staining and minimal cell-associated matrix. a rounded shape and produce a substantial pericellular remains tightly associated with the cell. A significant matrix similar to that seen in articular chondrocytes in portion of GAG and collagen diffuses into the far- alginate.4 Histologic and biochemical analysis indicates removed matrix, as well (Figs. and5 9). On release of that this matrix contains a large amount of GAG and cells from the alginate, only the cell-associated matrix is collagen type II, with small amounts of collagen type I. retained. Proteins in this pericellular, or cell-associated, matrix By P2 and P3, alginate-encapsulated cells lose these become firmly bound to the cell. On release of cells from chondrocytic characteristics. Although cells still retain a alginate by sodium citrate or EDTA addition, the matrix rounded shape because of three-dimensional encapsula- 700 Chia et al Human Nasal Septal Chondrocytes J Am Coll Surg

Figure 11. Collagen type I staining in cells released from alginate beads. Passage 1 cells (a) demonstrate large cell-associated matrix and mildly positive staining. Passage 2 (b) and passage 3 (c) cells demonstrate mildly positive staining. tion by alginate, there is little to no cell-associated matrix Monolayer cells at P0 and P1 have the gross appear- formation, and cells exhibit biochemical and immuno- ance of fibroblasts, given their flattened and spindle histologic characteristics of fibroblasts. These cells accu- shape. As expected, there is no significant pericellular mulate little GAG and collagen, and immunohisto- matrix formation, although immunohistochemical stain- chemistry reveals that collagen accumulation is ing indicates a fibrous network of collagen overlying the primarily that of collagen type I. These results demon- monolayer. Biochemical tests suggest that these cells are strate that cells cultured in alginate fail to express the fibroblastic, as they accumulate little GAG and collagen. chondrocyte phenotype by the second passage in On immunohistochemical analysis, these cells demon- monolayer. strate both collagen type I and type II production. Al- Vol. 200, No. 5, May 2005 Chia et al Human Nasal Septal Chondrocytes 701

Figure 12. Collagen type II staining of monolayers. Positive staining in a ﬁbrous pattern is noted at passage 0 (a), with decreased staining at passage 1 (b). No staining at passages 2 (c) and 3 (d) is noted. though gross appearance of these P0 and P1 monolayers produced by alginate-encapsulated cells are held in place indicates a fibroblastic phenotype, these cells in fact re- largely by the surrounding alginate polymer. tain chondrocyte phenotype as indicated by type II col- Additional passage in monolayer leads to complete lagen production. Presence of substantial amounts of dedifferentiation of these cells into a fibroblastic pheno- collagen type I on immunohistochemistry may indicate type. Monolayer cells at P2 and 3 have the gross, bio- that these cells are beginning to dedifferentiate into a chemical, and histologic characteristics of fibroblasts. fibroblastic phenotype. Although little GAG or collagen These cells demonstrate little GAG accumulation and is accumulated in the pericellular area, it is likely that no collagen type II production. Strongly positive colla- these proteins are being produced, but are lost to the gen type I staining is noted on immunohistochemistry. surrounding medium. In contrast, GAG and collagen These results indicate that encapsulation of dedifferenti- 702 Chia et al Human Nasal Septal Chondrocytes J Am Coll Surg

Figure 13. Collagen type I staining in monolayers: passage (P) 0 (a), P1 (b), P2 (c), P3 (d). Positive staining is noted in all passages. ated human nasal septal chondrocytes in alginate beads the ability of articular and septal chondrocytes to rediffer- after serial passage in monolayer does not lead to redifferentiate in a three-dimensional culture medium. Our cur- entiation in these cells. Although earlier studies indicate rent study also uses adult tissue only. The ability of human success with alginate encapsulation of dedifferentiated ar- nasal septal chondrocytes to redifferentiate may vary with ticular chondrocytes,4,5,16-18 our study does not demonstrate age of tissue donor. similar findings. Lack of redifferentiation of septal cells de- Alteration of cell shape alone may not be sufficient to spite rounded phenotype contradicts Benya and Shaffer’s stimulate reversion to the chondrocyte phenotype in hypothesis3 that dedifferentiated chondrocytes respond to dedifferentiated nasal septal chondrocytes. Recent stud- suspension in an anchorage-independent system (such as ies have demonstrated the ability of specific growth fac- agarose or alginate) by switching to the differentiated phe- tors (eg, fibroblastic growth factor, transforming growth notype. This may be because of innate differences between factor-␤, insulin-like growth factor) to promote reex- Vol. 200, No. 5, May 2005 Chia et al Human Nasal Septal Chondrocytes 703 pression20,21 of the cartilage phenotype in articular chon- Drafting of manuscript: Chia, Homicz, Schumacher, drocytes passaged in monolayers. One study has also Thonar, Masuda demonstrated the ability of articular chondrocytes to main- Critical revision: Chia, Schumacher, Thonar, Masuda, tain chondrocyte phenotype during monolayer passage in Sah, Watson the presence of fibroblastic growth factor.22 Addition of Statistical expertise: Chia, Sah these factors during monolayer passage and alginate cul- Obtaining funding: Sah, Watson ture may help human nasal septal chondrocytes main- Supervision: Sah, Watson tain chondrocyte phenotype in monolayer, or may facilitate redifferentiation when encapsulated in alginate. Acknowledgment: We wish to thank Won Bae for his assis- Studies indicate that osteogenic protein-1 can stimulate tance in statistical analysis. synthesis of proteoglycans and collagens in articular chondrocytes.23 Similar studies with human nasal chondrocytes may reveal a means to generate large amounts of REFERENCES matrix formation in cultured cells and neocartilage. 1. Takigawa M, Shirai E, Fukuo K, et al. Chondrocytes dediffer- GAG and collagen accumulation in cells retaining chon- entiated by serial monolayer culture form cartilage nodules in drocyte phenotype is clearly superior with alginate culture nude mice. Bone Miner 1987;2:449–462. 2. von der Mark K, Gauss V, von der Mark H, Muller P.Relation- compared with monolayer. As evidenced by the lower pro- ship between cell shape and type of collagen synthesised as chon- liferation rate of cells in alginate compared with cells in drocytes lose their cartilage phenotype in culture. Nature 1977; monolayer, alginate is not an appropriate medium for ex- 267:531–532. 3. Benya PD, Shaffer JD. Dedifferentiated chondrocytes reexpress pansion of cell numbers. It does have great potential as an the differentiated collagen phenotype when cultured in agarose intermediate culture medium in which cells with chondro- gels. Cell 1982;30:215–224. cyte phenotype can accumulate large amounts of GAG and 4. Hauselmann HJ, Aydelotte MB, Schumacher BL, et al. Synthesis and turnover of proteoglycans by human and bovine adult articular chon- collagen in their pericellular matrix. These cells could be drocytes cultured in alginate beads. Matrix 1992;12:116–129. released from alginate after a period of matrix accumulation 5. Guo JF, Jourdian GW, MacCallum DK. Culture and growth for use in neocartilage construct formation. Once expres- characteristics of chondrocytes encapsulated in alginate beads. sion of the chondrocyte phenotype in monolayer passaged Connect Tissue Res 1989;19:277–297. 6. Homicz MR, Chia SH, Schumacher BL, et al. Human septal cells can be reliably maintained, alginate may be used to chondrocyte redifferentiation in alginate, PGA scaffold, and generate tremendous quantities of matrix critical for neo- monolayer culture. Laryngoscope 2003;113:25–32. cartilage formation. 7. Kim YJ, Sah RL, Doong JY, Grodzinsky AJ. Fluorometric assay of DNA in cartilage explants using Hoechst 33258. Anal Bio- In contrast to studies with articular cartilage, encap- chem 1988;174:168–176. sulation in alginate alone does not promote redifferen- 8. Farndale RW, Buttle DJ, Barrett AJ. Improved quantitation tiation of human nasal septal chondrocytes. Culture in and discrimination of sulphated glycosaminoglycans by use of dimethylmethylene blue. Biochim Biophys Acta 1986; alginate does stimulate deposition of GAG and collagen 883:173–177. type II in cells retaining chondrocyte phenotype. Nasal 9. D’Souza AL, Masuda K, Otten LM, et al. Differential effects of chondrocyte culture in alginate beads has potential ap- interleukin-1 on hyaluronan and proteoglycan metabolism in two plication as an intermediate step in promoting cartilage compartments of the matrix formed by articular chondrocytes maintained in alginate. Arch Biochem Biophys 2000;374:59–65. matrix production for cartilage tissue engineering. Ad- 10. Woessner JF.The determination of hydroxyproline in tissue and ditional studies are required to develop a technique for protein samples containing small proportions of this imino acid. stimulating redifferentiation of cells expanded in mono- Arch Biochem Biophys 1961;93:440–447. 11. Chiba K, Andersson GB, Masuda K, Thonar EJ. Metabolism of layer, or maintaining chondrocyte phenotype during the extracellular matrix formed by intervertebral disc cells cul- monolayer expansion. tured in alginate. Spine 1997;22:2885–2893. 12. Herbage D, Bouillet J, Bernengo JC. Biochemical and physio- Author Contributions chemical characterization of pepsin-solubilized type-II collagen from bovine articular cartilage. Biochem J 1977;161:303–312. Study conception and design: Chia, Schumacher, Sah, 13. Jackson DS, Cleary EG. The determination of collagen and Watson elastin. Methods Biochem Anal 1967;15:25–76. Acquisition of data: Chia, Schumacher 14. Pal S, Tang LH, Choi H, et al. Structural changes during development in bovine fetal epiphyseal cartilage. Coll Relat Res 1981; Analysis and interpretation of data: Chia, Schumacher, 1:151–176. Sah, Watson 15. Homicz MR, Schumacher BL, Sah RL, Watson D. Effects of 704 Chia et al Human Nasal Septal Chondrocytes J Am Coll Surg

serial expansion of septal chondrocytes on tissue engineered during the expansion and redifferentiation of adult human neocartilage composition. Otolaryngol Head Neck Surg articular chondrocytes enhance chondrogenesis and cartilagi- 2002;127:398–408. nous tissue formation in vitro. J Cell Biochem 2001;81:368– 16. Bonaventure J, Kadhom N, Cohen-Solal L, et al. Reexpression 377. of cartilage-specific genes by dedifferentiated human articular 21. Yaeger PC, Masi TL, de Ortiz JL, et al. Synergistic action of chondrocytes cultured in alginate beads. Exp Cell Res 1994; transforming growth factor-beta and insulin-like growth 212:97–104. factor-I induces expression of type II collagen and aggrecan 17. Hauselmann HJ, Masuda K, Hunziker EB, et al. Adult human genes in adult human articular chondrocytes. Exp Cell Res chondrocytes cultured in alginate form a matrix similar to native 1997;237:318–325. human articular cartilage. Am J Physiol 1996;271:C742–C752. 22. Martin I, Vunjak-Novakovic G, Yang J, et al. Mammalian chon- 18. Hauselmann HJ, Fernandes RJ, Mok SS, et al. Phenotypic sta- drocytes expanded in the presence of fibroblast growth factor 2 bility of bovine articular chondrocytes after long-term culture in maintain the ability to differentiate and regenerate three- alginate beads. J Cell Sci 1994;107:17–27. dimensional cartilaginous tissue. Exp Cell Res 1999;253:681– 19. Thonar EJ, Buckwalter JA, Kuettner KE. Maturation-related 688. differences in the structure and composition of proteoglycans 23. Flechtenmacher J, Huch K, Thonar EJ, et al. Recombinant hu- synthesized by chondrocytes from bovine articular cartilage. man osteogenic protein 1 is a potent stimulator of the synthesis J Biol Chem 1986;261:2467–2474. of cartilage proteoglycans and collagens by human articular 20. Jakob M, Demarteau O, Schafer D, et al. Specific growth factors chondrocytes. Arthritis Rheum 1996;39:1896–1904. GATA-3 Expression as a Predictor of Hormone Response in Breast Cancer

Purvi Parikh, MD, Juan P Palazzo, MD, Lewis J Rose, MD, Constantine Daskalakis, SCD, Ronald J Weigel, MD, PhD, FACS

BACKGROUND: Expression of estrogen receptor-␣ (ER␣) as determined by immunohistochemistry of tumor tissue is currently the most clinically useful test to predict hormone responsiveness of breast cancer. Thirty percent of ER␣-positive breast cancers do not respond to hormonal therapy. GATA-3 is a transcription factor that is expressed in association with ER␣ and there is evidence that GATA factors influence response to estrogen. In this pilot study, we investigated whether GATA-3 expression is associated with hormone response in breast cancer. STUDY DESIGN: Breast cancer tissue was stained for GATA-3 expression by immunohistochemistry in ER␣- positive cancers from 28 patients, 14 of whom were defined as hormone unresponsive (cases) and 14 of whom were age-matched controls with hormone-responsive, ER␣-positive cancers (controls). RESULTS: Comparing cases and controls, there were no differences in expression of ER␣; progesterone receptor, ErbB2; or tumor grade. Using 20% nuclear staining to characterize tumors as GATA-3 positive or GATA-3 negative, 6 of 14 (43%) cancers in the hormone-unresponsive group and none of the controls were classified as GATA-3 negative (odds ratio, 8.2; 95% confidence interval, 1.2Ϫϱ; p ϭ 0.031). Using different cut points to characterize GATA-3 positivity yielded very similar results, indicating a positive association between lack of GATA-3 expression and lack of response to hormonal therapy. CONCLUSIONS: The study suggests that analyzing ER␣-positive breast tumors for GATA-3 using immunohistochemistry might improve prediction of hormone responsiveness. The association between GATA-3 expression and hormone response suggests that GATA-3 may play a role in mechanisms controlling response to estrogen. (J Am Coll Surg 2005;200:705–710. © 2005 by the American College of Surgeons)

Breast cancer is a heterogeneous disease and current treat- often occur in postmenopausal patients,2 haveanim- ment options are individualized to the extent possible based proved prognosis,3 and are more likely to respond to on a determination of the biologic character of each tumor. hormonal therapy4 compared with tumors with low or For example, treatment with Trastuzumab (Herceptin) is negligible expression of the receptor. Despite recent only indicated for patients with tumors that demonstrate work using genomics, proteomics, and other diagnostic amplification of the protooncogene HER-2/neu (c-ErbB-2).1 tests, expression of ER␣ remains the most clinically use- Hormonal therapy has been an important treatment op- ful method to predict hormone response in breast cancer tion for many breast cancer patients. Breast cancers that patients.5 For many years, expression of ER␣ had been overexpress the estrogen receptor-␣ (ER␣) protein most determined by measuring the binding of radiolabeled estrogen to cytosolic proteins using a dextran-coated 6 Competing interests declared: None. charcoal assay. The dextran-coated charcoal method re- Supported in part by the Kristen Olewine Milke Breast Cancer Research quired that fresh frozen tumor be saved at the time of Fund. resection and the technique was sensitive to the amount Received September 2, 2004; Revised December 29, 2004; Accepted Decem- and quality of the saved tumor sample. Over the past 20 ber 30, 2004. From the Departments of Surgery (Parikh, Weigel), Pathology (Palazzo), and years, the dextran-coated charcoal method has been re- Department of Medicine, Division of Medical Oncology (Rose) and Biosta- placed by immunohistochemistry in which ER␣ protein tistics Section (Daskalakis), Thomas Jefferson University, Philadelphia, PA. Correspondence address: Ronald J Weigel, MD, PhD, FACS, Department of is detected in tumor cells using a monoclonal antibody. Surgery, 1025 Walnut St, Ste 605, Philadelphia, PA 19107. Immunohistochemistry has become the preferred

© 2005 by the American College of Surgeons ISSN 1072-7515/05/$30.00 Published by Elsevier Inc. 705 doi:10.1016/j.jamcollsurg.2004.12.025 706 Parikh et al GATA-3 Predicts Hormone Response J Am Coll Surg method to determine ER␣ expression because it can be tamoxifen or aromatase inhibitor therapy. All patients performed on formalin-fixed, archival specimens; re- had breast tumors previously determined to be ER␣ pos- quires minimal tumor tissue; and is based on techniques itive by immunohistochemistry. A control group, con- that can be applied to a wide range of antigens.7 sisting of patients confirmed to be hormone responsive, Although expression of ER␣ is the best available predic- was individually matched to the case group by age and tor of hormone response, the association between ER␣ ex- ER␣ status (ie, all ER␣-positive cancers). pression and hormone response is not perfect. There is un- certainty about the optimal cutoff for the percent staining GATA-3 immunohistochemistry used to categorize a tumor as ER␣-positive versus ER␣- Archival tumor tissue was collected by the Department negative. Most pathologists use a value between 10% and of Pathology and stained using a 1:250 dilution of 20% as the cutoff for positive staining.8-10 Approximately GATA-3 monoclonal antibody HG3-31X (Santa Cruz 70% of ER␣-positive/progesterone receptor (PR)-positive Biotechnology) with microwave antigen retrieval pre- breast cancers respond to hormonal therapy.6 Response treatment in citrate buffer as described previously.12 rates for ER-positive/PR-negative cancers are in the range GATA-3 expression was characterized as to intensity and of 30% to 35% and Ͻ10% for ER-negative/PR-negative percent of nuclear staining. Evaluation of GATA-3 stain- cancers.6,11 Use of additional markers to refine the predic- ing was done blinded to clinical information. tion of hormone response could improve the treatment plan for a significant number of patients. Statistical analysis GATA-3 transcription factor has a pattern of expres- Preliminary comparisons of cases and controls on sion in breast cancer that has a striking association with GATA-3 expression, and other biomarkers, were con- ER␣.12-14 Approximately 96% of ER␣-positive breast ducted through Wilcoxon’s signed rank test for paired cancers express GATA-3, but only 22% of ER␣-negative data. In the main analyses, GATA-3 positivity was de- cancers express the GATA-3 protein. Studies examining fined using an a priori value of 20% nuclear staining, clinical outcomes have concluded that GATA-3 expres- which is the same cut point used to determine ER␣ sion, as determined by microarray analysis, can distin- positivity. Data were also analyzed using different cutoff guish tumor subclasses with differences in disease-free values for the percent nuclear staining for GATA-3 by survival and overall survival.15 It has been shown that a immunohistochemistry (10% through 50%). Because related GATA factor, GATA-1, is responsible for of the small sample size, exact statistical methods were estrogen-mediated repression of erythropoiesis, and this used. The proportion of GATA-3Ϫnegative patients study also demonstrated that ER␣ binds to GATA-1 in a among cases and controls was compared using the exact region with extensive homology to GATA-3.16 These version of McNemar’s test for matched data. Odds ratios findings have led us to hypothesize that GATA-3 may be (comparing the odds of GATA-3 negatives among cases involved in cellular response to estrogen and expression versus controls) and associated 95% confidence intervals of GATA-3 might be predictive of hormone response in were computed through exact logistic regression. Statis- breast cancer. The purpose of this pilot study was to tical analyses were carried out with StatXact 6.1 and determine if expression of GATA-3 is associated with LogXact 5 (Cytel Software Corp).17 hormone response in ER␣-positive breast cancers. RESULTS METHODS The population of this pilot study consisted of 14 patients Patient selection with ER␣-positive breast cancers who failed hormonal We conducted a pilot matched case-control study to therapy (cases), and an equal number of age-matched pa- evaluate the association between expression of GATA-3 tients with ER␣-positive cancers who exhibited hormone and hormone responsiveness of breast cancer. The study response (controls). Table 1 presents data for ER␣, PR, and was carried out at Thomas Jefferson University and all ErbB2 and reports GATA-3 expression based on percent of protocols were IRB approved. Patients were identified positive tumor cells as determined by immunohistochem- by the medical oncology service and were determined to istry. Patients were between 37 and 73 years of age (mean be hormone unresponsive on the basis of clinical char- 57 years) and because of matching, cases and controls were acteristics of progression of disease or early recurrence on very similar with respect to age (all matched within 2 years). Vol. 200, No. 5, May 2005 Parikh et al GATA-3 Predicts Hormone Response 707

Table 1. Characteristics (Including GATA-3 Expression) of Cases (Hormone Unresponsive) and Age-Matched Controls (Hor- mone Responsive) Cases Controls Pair ID Age (y) ER␣ (%) PR (%) ErbB2 GATA-3 (%) GATA-3 notes Age (y) ER␣ (%) PR (%) ErbB2 GATA-3 (%) 1618590Ϫ 60 — 60 90 80 Ϫ 50 2 60 90 90 N/A 20 Weak 60 90 0 Ϫ 95 3569595ϩ 95 — 56 80 60 Ϫ 60 4508050Ϫ 5 Weak 48 70 85 ϩ 75 5639070Ϫ 90 — 63 80 20 Ϫ 80 6 73 90 90 N/A 80 — 73 100 35 Ϫ 90 7 37 80 5 N/A 20 Hetero 37 70 80 Ϫ 40 868605Ϫ 70 — 68 100 75 Ϫ 90 966325Ϫ 80 Hetero 66 90 95 Ϫ 90 10 48 95 80 Ϫ 5 Weak 48 90 90 ϩ 60 11 51 70 90 N/A 0 — 51 90 95 Ϫ 80 12 51 90 10 Ϫ 20 Weak 53 90 50 ϩ 95 13 52 90 85 Ϫ 30 — 52 100 5 Ϫ 80 14 57 95 2 ϩ 80 — 55 90 65 ϩ 50 ER␣, estrogen receptor-␣; Hetero, heterogeneity of staining pattern; N/A, not available; PR, progesterone receptor; Weak, staining was weak nuclear.

All breast cancers were determined to be ER␣ positive (1 hormone-unresponsive group were classified as negative case was 32% ER␣ positive, but the remaining cases and for GATA-3 expression, and no cancers in the control controls ranged between 60% and 100%; mean 85). Be- group were classified as negative. This corresponds to an cause all cancers were selected to be ER␣ positive, cases and odds ratio of 8.2 (ie, GATA-3Ϫnegative tumors have an controls were similar with regard to ER␣ positivity (mean eightfold higher risk of being hormone unresponsive expression of ER␣ was 82% for cases and 88% for controls, compared with GATA-3Ϫpositive tumors), with an as- p ϭ 0.57). Cases and controls were also similar with respect sociated p value of 0.031 (Table 2). to PR expression (mean 55% for cases, 60% for controls, When introducing a new test using immunohisto- p ϭ 0.77). Finally, no major differences were found with chemistry, the cutoff point to characterize a tumor as ErbB2 expression or tumor grade among the case-control positive or negative needs to be established. We evalu- pairs for which such information was available. ated data using a range of cutoff values (from 10% to Percent of nuclei staining with GATA-3 ranged between 50%, see Table 2). Loss of GATA-3 expression was asso- 0% and 95% and there was a significant difference when ciated with increased risk of hormone unresponsiveness, comparing cases and controls. Average percent nuclear irrespective of criterion (ie, odds ratios were consistently staining was 47% in the hormone-unresponsive group and Ͼ1). Cut points between 20% and 40% performed sim- 74% in the age-matched hormone-responsive control ilarly, although 30% seemed to yield the strongest result. group (p ϭ 0.039). All cancers that had GATA-3 nuclear staining of Յ 20% were also noted to have weak staining, possibly suggesting low levels of protein expression in the DISCUSSION nuclei of these cells. Figure 1 provides representative images Approximately 200,000 women are diagnosed annu- of GATA-3 staining. ally in the US with invasive breast cancer and three- Clinical evaluation with ER␣ does not consider stain- quarters of these cancers will be ER␣ positive.18 Be- ing as a continuous variable, but establishes a cutoff used cause 30% of ER␣ positive cancers will not be to characterize tumors as positive or negative. The stan- hormone responsive, it is estimated that 45,000 dard ER␣ evaluation by immunohistochemistry at women will be treated with hormone therapy on an Thomas Jefferson University’s Department of Pathology annual basis but will not respond to this form of has established that Յ 20% tumor cell staining is con- therapy. Improving the ability to predict hormone sidered negative. Applying the same criterion to response will have a significant impact on breast can- GATA-3 staining, 6 of 14 (43%) of tumors in the cer treatment by avoiding costs related to ineffective 708 Parikh et al GATA-3 Predicts Hormone Response J Am Coll Surg

staining of Յ 20% to define negative GATA-3 expression, we would estimate that 43% of hormone- unresponsive, ER␣-positive breast cancers could be correctly predicted to be hormone insensitive based on the lack of GATA-3 expression (sensitivity). Approximately 19,350 women who are diagnosed with an ER␣-positive invasive breast cancer annually and who are currently offered hormonal therapy could be correctly defined as hormone unresponsive using expression of GATA-3. The fact that none of the age-matched controls was classified as GATA-3 negative suggests a low probability of incorrectly predicting that an ER␣-positive tumor is hormone unresponsive (high specificity). These findings should be viewed with caution because of the study’s limitations as a result of its small sample size. First, the strength of the association between GATA-3 expression and hormone response, and the sensitivity and specificity of this marker, were estimated with very limited precision. Second, no adjustment for possible differences between cases and controls on other markers of hormone response was possible. And, third, it was not possible to formally determine an optimal GATA-3Ϫpositivity cut point. Despite these limitations, the study’s findings can now guide the design of a larger, more definitive study that can address these issues. A number of methods have been pursued as a means to improve the prediction of hormone response in breast cancer. One recent study attempted to use methylation of the ER␣ gene as a clinical test to predict hormone response and suggested an improvement over hormone receptor evaluation.19 Attempts to use expression of ER␣-regulated genes to improve the prediction of estrogen response have been reported but results have not consistently indicated that this approach improves the prediction of endocrine responsiveness beyond hormone receptor status.20-22 Use of a radiolabeled ER␣ li-

Figure 1. Immunohistochemistry of GATA-3 staining. Examples of gand has demonstrated a promising method to predict 23 three different invasive ductal carcinomas stained with the GATA-3 hormone responsiveness of metastatic breast cancers. antibody. (A) Shows 95% of tumor cells with strong nuclear staining; Expression of ER␤ has been evaluated but there does not (B) 20% of tumor cells staining; and (C) Ͻ5% of the tumor cells appear to be a clear correlation between ER␤ expression staining (immunohistochemistry, ϫ150 and ϫ250). and hormone response of breast cancer.24,25 A recent study suggested that expression of PR isoforms might be treatment and by eliminating the delay of initiating used to identify a subgroup of patients with ER␣- more-effective therapy. positive tumors that are less likely to respond to hor- Results from this pilot study suggest that GATA-3 monal therapy.26 Another study suggested that expres- expression may be associated with hormone response in sion of the cdk inhibitor p27Kip1 might be predictive of ER␣-positive breast cancers. Using a cutoff for nuclear response to antiestrogens.27 Although there is active in- Vol. 200, No. 5, May 2005 Parikh et al GATA-3 Predicts Hormone Response 709

Table 2. Comparison of GATA-3 Expression in Cases (Hormone Unresponsive) and Age-Matched Controls (Hormone Respon- sive), Using Various Deﬁnitions of GATA-3 Positivity (Cut Points) Negative Positive cut point Negative cases controls for GATA-3؉ n%n% OR 95% CI p Value Ͼ10% 3 21 0 0 3.85 0.41, ϱ 0.250 Ͼ20% 6 43 0 0 8.17 1.18, ϱ 0.031 Ͼ30% 7 50 0 0 9.61 1.44, ϱ 0.016 Ͼ40% 7 50 1 7 8.17 1.18, ϱ 0.031 Ͼ50% 7 50 3 21 3.00 0.54, 30.39 0.289 95% CI, 95% confidence interval; OR, odds ratio. terest in this area of investigation, none of these studies breast cancer will be needed to define mechanisms by has been definitive and expression of ER␣ by immuno- which this transcription factor may alter the clinical histochemistry remains the most clinically useful predic- course of breast cancers. tor of hormone response. Many of the tests being developed use polymerase chain reaction amplification, Author Contributions genome array hybridization, or more complex molecular Study conception and design: Rose, Daskalakis, Weigel tests that require the use of fresh frozen tissue. A predic- Acquisition of data: Parikh, Palazzo, Weigel tive test that is based on immunohistochemistry would Analysis and interpretation of data: Parikh, Palazzo, have the advantage of being rapidly adopted in the clin- Daskalakis, Weigel ical setting. We hope that a future study will be able to Drafting of manuscript: Parikh, Weigel confirm the strength of the predictive utility of GATA-3 Critical revision: Palazzo, Rose, Daskalakis, Weigel expression in patients with ER␣-positive cancers and to Statistical expertise: Daskalakis evaluate the optimal positivity cut point for such a Obtaining funding: Weigel GATA-3 test. Supervision: Palazzo, Rose, Weigel Expression of GATA-3 has been shown to predict a breast cancer subtype, defined as luminal A, which carries an improved disease-free survival and overall Acknowledgment: We wish to thank Kathy Califano and Mag- survival when compared with tumors that do not ex- dalena Potoczek for performing the immunohistochemical press GATA-3.15 The method used to define a breast stains. cancer as hormone responsive or unresponsive based on clinical outcomes is open to debate. If a patient has a recurrence or develops metastatic disease while on REFERENCES hormonal therapy, we tend to classify the cancer as 1. Pegram MD, Pienkowski T, Northfelt DW, et al. Results of two open-label, multicenter phase II studies of docetaxel, platinum hormone unresponsive. On the other hand, if a pa- salts, and trastuzumab in HER2-positive advanced breast can- tient remains disease free on hormonal therapy, we cer. J Natl Cancer Inst 2004;96:759–769. tend to classify her as hormone responsive. But the 2. Fisher ER, Redmond CK, Liu H, et al. Correlation of estrogen clinical course of the patient may not be directly at- receptor and pathologic characteristics of invasive breast cancer. Cancer 1980;45:349–353. tributable to the hormone responsiveness of the tu- 3. Knight WA, Livingston RB, Gregory EJ, McGuire WL. Estro- mor. Hence, GATA-3 may be a marker for a cancer gen receptor as an independent prognostic factor for early recur- with a more favorable biology independent of hor- rence in breast cancer. Cancer Res 1977;37:4669–4671. mone response and may control differentiation of the 4. Heuson JC, Longeval E, Mattheiem WH, et al. Significance of quantitative assessment of estrogen receptors for endocrine ther- tumor as suggested by a recent study examining apy in advanced breast cancer. Cancer 1977;39:1971–1977. GATA-3 mutations.28 An additional question that re- 5. Buzdar AU, Vergote I, Sainsbury R. The impact of hormone mains to be addressed is whether GATA-3 expression receptor status on the clinical efficacy of the new-generation may also be used to predict hormone response or out- aromatase inhibitors: a review of data from first-line metastatic ␣ disease trials in postmenopausal women. Breast J 2004;10:211– comes among patients with ER -negative breast can- 217. cers. Additional work on the action of GATA-3 in 6. Jordan VC, Wolf MF, Mirecki DM, et al. Hormone receptor 710 Parikh et al GATA-3 Predicts Hormone Response J Am Coll Surg

assays: clinical usefulness in the management of carcinoma of 18. Ghafoor A, Jemal A, Ward E, et al. Trends in breast cancer by the breast. Crit Rev Clin Lab Sci 1988;26:97–152. race and ethnicity. CA Cancer J Clin 2003;53:342–355. 7. Harvey JM, Clark GM, Osborne CK, Allred DC. Estrogen recep- 19. Widschwendter M, Siegmund KD, Muller HM, et al. Associa- tor status by immunohistochemistry is superior to the ligand- tion of breast cancer DNA methylation profiles with hormone binding assay for predicting response to adjuvant endocrine therapy receptor status and response to tamoxifen. Cancer Res 2004;64: in breast cancer. J Clin Oncol 1999;17:1474–1481. 3807–3813. 8. Diaz LK, Sahin A, Sneige N. Interobserver agreement for estro- 20. Hayashi S. Prediction of hormone sensitivity by DNA microar- gen receptor immunohistochemical analysis in breast cancer: a ray. Biomed Pharmacother 2004;58:1–9. comparison of manual and computer-assisted scoring methods. 21. Yoshida N, Omoto Y, Inoue A, et al. Prediction of prognosis of Ann Diagn Pathol 2004;8:23–27. estrogen receptor-positive breast cancer with combination of 9. Regitnig P, Reiner A, Dinges HP, et al. Quality assurance for selected estrogen-regulated genes. Cancer Sci 2004;95:496– detection of estrogen and progesterone receptors by immuno- 502. histochemistry in Austrian pathology laboratories. Virchows 22. Luqmani YA, Ricketts D, Ryall G, et al. Prediction of response Arch 2002;441:328–34. Epub September 17, 2002. to endocrine therapy in breast cancer using immunocytochem- 10. Wishart GC, Gaston M, Poultsidis AA, Purushotham AD. Hor- ical assays for pS2, oestrogen receptor and progesterone receptor. mone receptor status in primary breast cancer—time for a con- Int J Cancer 1993;54:619–623. sensus? Eur J Cancer 2002;38:1201–1203. 23. Bennink RJ, van Tienhoven G, Rijks LJ, et al. In vivo prediction 11. McGuire WL, Horwitz KB, Pearson OH, Segaloff A. Current of response to antiestrogen treatment in estrogen receptor- status of estrogen and progesterone receptors in breast cancer. positive breast cancer. J Nucl Med 2004;45:1–7. Cancer 1977;39:2934–2947. 24. de Cremoux P,Tran-Perennou C, Elie C, et al. Quantitation of 12. Hoch RV, Thompson DA, Baker RJ, Weigel RJ. GATA-3 is expressed in association with estrogen receptor in breast cancer. estradiol receptors alpha and beta and progesterone receptors in Int J Cancer 1999;84:122–128. human breast tumors by real-time reverse transcription- 13. Schaner ME, Ross DT, Ciaravino G, et al. Gene expression polymerase chain reaction. Correlation with protein assays. Bio- patterns in ovarian carcinomas. Mol Biol Cell 5,2003;14:4376– chem Pharmacol 2002;64:507–515. 4386. Epub September 5, 2003. 25. Iwase H, Zhang Z, Omoto Y, et al. Clinical significance of the 14. Yang GP, Ross DT, Kuang WW, et al. Combining SSH and expression of estrogen receptors alpha and beta for endocrine cDNA microarrays for rapid identification of differentially ex- therapy of breast cancer. Cancer Chemother Pharmacol 2003; pressed genes. Nucleic Acids Res 1999;27:1517–1523. 52[Suppl 1]:S34–S38. Epub June 19, 2003. 15. SorlieT,Perou CM,Tibshirani R, et al. Gene expression patterns 26. Hopp TA, Weiss HL, Hilsenbeck SG, et al. Breast cancer pa- of breast carcinomas distinguish tumor subclasses with clinical tients with progesterone receptor PR-A-Rich tumors have implications. Proc Natl Acad Sci USA 2001;98:10869–10874. poorer disease-free survival rates. Clin Cancer Res 2004;10: 16. Blobel GA, Sieff CA, Orkin SH. Ligand-dependent repression 2751–2760. of the erythroid transcription factor GATA-1 by the estrogen 27. Arteaga CL. Cdk inhibitor p27Kip1 and hormone dependence receptor. Mol Cell Biol 1995;15:3147–3153. in breast cancer. Clin Cancer Res 2004;10:368S–371S. 17. Hirji KF, Mehta CR, Patel NR. Exact inference for matched 28. Usary J, Llaca V, Karaca G, et al. Mutation of GATA3 in human case-control studies. Biometrics 1988;44:803–814. breast tumors. Oncogene 2004;23:7669–7678. Fusion of the Planes of the Liver: An Anatomic Entity Merging the Midplane and the Left Intersectional Plane

Eric Savier, MD, Jacques Taboury, MD, Olivier Lucidarme, MD, PhD, Kumiko Kitajima, MD, Mehdi Cadi, MD, Jean-Christophe Vaillant, MD, Laurent Hannoun, MD

BACKGROUND: Alignment of the gallbladder fossa and the round ligament may be associated with an almost unknown portal vein branching anomaly. STUDY DESIGN: Ultrasonographic imaging allowed detection of this anomaly, which we characterized as fusion of the planes of the liver. When appropriate, additional specific radiologic examinations were performed (CT scanner supplemented with a three-dimensional reconstruction, a biliary car- tography, or an angiography). Surgical consequences were studied from this series and from the literature. RESULTS: Seven patients (0.5%) had the following criteria: 1) round ligament, gallbladder fossa, and termination of the portal vein occurring in the same plane; 2) typical portal vein branching, including a right posterior branch, left branches, and a main medial branch terminated by the Rex’s recessus; 3) two main hepatic veins without a significant middle hepatic vein; and 4) absence of the horizontal part of the left hepatic duct. Fusion of the planes may have been involved in two cases of iatrogenic bile duct injury and contraindicated a tumor resection and a right-liver donation. A review of the literature revealed that lack of recognition of the fusion of the planes led to a high proportion of surgical iatrogenic injury. Fusion of the planes could result from incomplete development of the central part of the liver, in agreement with embryologic knowledge. CONCLUSIONS: Knowledge of the fusion of the planes by hepato-biliary surgeons is important. This anomaly may lead to serious complications if it remains undetected during liver resection or bile duct surgery. (J Am Coll Surg 2005;200:711–719. © 2005 by the American College of Surgeons)

Portal, arterial, or biliary anatomic variations are often (eg, the round ligament, the Rex-Cantlie’s line), and observed during liver or biliary surgery.1 Among these operative ultrasonography or cholangiography to obtain anomalies, intrahepatic abnormalities of the portal vein optimal knowledge of any particular anatomy. In some or of the biliary tree are the most dangerous pitfalls that patients, the appearance of the liver may indicate a portal can lead to operative accidents if ignored.2-4 So, in addi- vein anomaly, for instance, if the gallbladder is located in tion to the theoretic background,5 the surgeon uses pre- the plane of the round ligament (the left intersectional operative imaging, external anatomic marks of the liver plane). In such cases, the bifurcation of the portal vein in two main branches may not exist. The first reported cases 6 Competing interests declared: None. were thought to be published by Hochstetter in 1886. These data have been presented in part as posters at the Journées Francoph- More recently, several cases were discovered, mainly in ones de Pathologie Digestive, Paris, France, April 2003, and at the European Asian countries, either by ultrasonography7-11 or during ex- Congress of Radiology, Vienna, Austria, March 2004. amination of the liver graft for a split procedure.1,12-14 For Received June 10, 2004; Revised December 6, 2004; Accepted December 20, 2004. several years, we have noticed that this anatomic anomaly From the Service de Chirurgie Digestive et Transplantation Hépatique was well highlighted by ultrasonography and have prospec- (Savier, Taboury, Kitajima,Vaillant, Hannoun) and the Service de Radiologie (Lucidarme, Cadi), Groupe hospitalier Pitié-Salpêtrière, Paris, France. tively looked for it. Here, we report seven patients and Correspondence address: Dr Eric Savier, Service de chirurgie digestive et discuss their surgical consequences. Because this variation transplantation hépatique (Pr L Hannoun), Groupe hospitalier Pitié- Salpêtrière, 47-83 Boulevard de l’Hôpital, 75651 Paris Cedex 13, France. merges the midplane and the left intersectional plane, we email: [email protected] called it fusion of the planes of the liver.

© 2005 by the American College of Surgeons ISSN 1072-7515/05/$30.00 Published by Elsevier Inc. 711 doi:10.1016/j.jamcollsurg.2004.12.017 712 Savier et al Fusion of the Planes of the Liver J Am Coll Surg

METHODS these 2 angles were similar or did not differ by more than Patients 5 degrees. If indicated, cholangiography or arteriogra- Between September 1, 1998, and February 28, 2003, phy was performed using standard techniques. 1,678 patients had ultrasonographic examinations of their livers for the first time in our center. Each examination was RESULTS performed by one operator (JT). Any anatomic anomaly had to be confirmed by CT scanner or magnetic resonance Seven cases of fusion of the planes of the liver were prospec- cholangiopancreatography (MRCP). The anatomic and tively discovered among 1,678 patients (0.4%). In each surgical terminology used was the Brisbane 2000 interna- patient, the following criteria were found: 1) juxtaposition tional terminology.15 of the round ligament and the gallbladder (or gallbladder fossa); 2) portal trunk giving on its right side a posterior Ultrasonography branch giving an ascending principal branch for segment 7 Ultrasonography with a standard 3.5-MHz probe and 1 or 2 parallel descending branches for segment 6; 3) (Hitachi Model EUB 525) placed under the right costal portal trunk giving on its left side two main branches: 1 margin visualized the indentation of the round ligament anterior for segment 3, 1 posterior for segment 2, and pos- between segments 3 and 4. In cases of fusion of the two terior accessory branches for segment 1; 4) in the axis of the planes, a displacement of the probe to the right was not portal vein, the Rex’s recessus prolonged by the round necessary to visualize the gallbladder, which was visible ligament in front of the gallbladder fossa; 5) small in line with the round ligament (Fig. 1A). Through an branches arising from the right side of the Rex’s recessus intercostal window, it was then possible to visualize the vascularazing the top of the liver; 6) usually, two main gallbladder, the round ligament, and the termination of hepatic veins, one draining the right part of the liver and the portal vein in the same plane (Fig. 1B). the other draining the left part, the latter lying on the left side of the midplane; 7) normal inferior vena cava and Imaging techniques segment 1. These anomalies, recorded by ultrasonogra- Three-dimensional postprocessing techniques using multi- phy cine-loop, were then confirmed by other radiologic detector row CT (MDCT) were used. Scans were per- examinations. The characteristics of each patient are de- formed with a GE Lightspeed 16 (GEMS) after iodinated scribed in the following text. contrast injection (slice thickness, 2.5 mm). Magnetic resonance cholangiopancreatographies (MRCP) were Patient 1 performed on a Philips 1.5 tesla Gyroscan Intera (Philips A 42-year-old man, originally from Cameroon, had Medical Systems). Standard CT scan images allowed morphologic examinations of his liver in view of a living- comparison between the left intersectional plane and related donor. The intended recipient was his homozy- midplane angles (Fig. 2). In cases of fusion of the planes, gote twin (with no hepatic malformation), who had hep-

Figure 1. Doppler sonography in a case of fusion of the planes of the liver. (A) Subcostal window (patient on his left side). (B) Intercostal window. The gallbladder, the round ligament, and termination of the portal vein (Rex’s recessus) were located in one plane. G, gallbladder; pv, portal vein; arrow, round ligament (ligamentum teres). Vol. 200, No. 5, May 2005 Savier et al Fusion of the Planes of the Liver 713

Figure 2. Left intersectional plane and midplane on CT scanner images. Inclination of the left intersectional plane was determined by the angle between a vertical axis and a line in the axis of the ascending part of the left portal branch and joining the Rex’s recessus (R). Inclination of the midplane was determined by the angle between a vertical axis and a line in the axis of the gallbladder fossa crossing the thin posterior extremity of the gallbladder plate (g). In panels A (19 degrees) and B (21 degrees), anomaly is present. In panels C (18 degrees) and D (45 degrees), anomaly is absent. atocellular carcinoma associated with human B virus later, the patient became jaundiced and underwent op- hepatitis. Ultrasonography showed a fusion of the eration for biliary peritonitis. The operative report planes. The MDCT reconstruction confirmed this re- described a section of the right hepatic duct; the left sult. Coelio-mesenteric arteriography demonstrated a hepatic duct appeared necrotic. Both ducts were anasto- hepatic artery that initially gave a left branch, which mosed with a Roux-en-Y jejunal loop. After subsequent divided into two branches for segments 2 and 3. The cholangitis attacks, a percutaneous cholangiography was right hepatic artery gave a vertical branch, then the cystic performed that showed independent right and left bili- artery, then two branches for the right liver. Portography ary trees. During 20 months, 9 consecutive percutane- demonstrated the absence of the left portal branch and ous biliary procedures failed to prevent the recurrence of of the right anterior sectorial branch. The bile duct anat- cholangitis. The patient was then referred to our depart- omy was documented by MRCP (Fig. 3), which dem- ment. Ultrasound examination discovered the fusion of onstrated the absence of the left hepatic duct. Taken the two planes, which was confirmed by MDCT. together, these anatomic variations contraindicated the The cholangiographies were revisited. On the right liver donation. side, the segmental bile ducts met in only one very short duct. On the left, four biliary territories ended in a com- Patient 2 mon and vertical duct. Surgical exploration confirmed A 57-year-old woman had a laparoscopic cholecystec- that the round ligament and the gallbladder fossa planes tomy for gallstones in another hospital. The surgeon matched one another. Also, it showed that the left noticed an additional bile duct close to the gallbladder hepatico-jejunal anastomosis was patent and did not re- fossa; this bile duct was clipped and divided. One month quire modification and that the right anastomosis had 714 Savier et al Fusion of the Planes of the Liver J Am Coll Surg

Figure 3. Magnetic resonance cholangiopancreatography in a case Figure 4. Operative view of the liver. In a case of fusion of the two of fusion of the planes. (A) transversal reconstruction; (B) oblique planes, the round ligament and the gallbladder are in a line. reconstruction. On the left side, bile ducts from segments 2 (b2) and 3 (b3) are connected separately to the common bile duct. On the right side, a bulky right posterior (rp) duct drained two posterior round ligament (Figs. 4 and 5). After an uneventful cho- segments of the right liver. On the midplane, are located the gall- lecystectomy, an operative cholangiography was per- bladder (G) and small ducts for the right anterior (ra) part of the liver formed by inserting a cannula into the cystic duct. A (authors’ interpretation, see discussion and Fig. 8). metallic stylet placed against the falciform ligament was superimposed on the common hepatic duct, demon- been replaced by a 10-mm long fibrous stenosis, which strating the absence of the left hepatic duct and the was removed. A new anastomosis was performed on fusion of the planes (Fig. 6). The common hepatic duct three healthy ducts of three posterior bile ducts. One received the bile duct from segments 2 and 3 on its left year later, the patient was asymptomatic, without any and a bulky duct on its right, which drained the major biliary drainage. Hepatic blood tests and ultrasono- part of the right liver. A pancreaticoduodenectomy was graphic examination of the liver were normal. then performed without technical problems. Patient 3 A 26-year-old woman had a laparoscopic cholecystec- Patient 5 tomy for acute cholecystitis in another hospital. Postop- A 46-year-old woman from Mauritania was investigated eratively, a bilious peritoneal effusion was treated by for acute cholecystitis. Our preoperative ultrasonogra- endoscopic prosthesis and iterative percutaneous drain- ages. Two months later, she was referred to our department for persistence of the peritoneal collection with sepsis. The ultrasonography visualized the prosthesis in the common bile duct without intrahepatic bile duct dilatation. The portal anatomy was characteristic of the fusion of the two planes, as detailed previously. Laparot- omy was required to drain an encapsulated peritonitis. No more bile leakage was observed, and postoperative recovery was uneventful. The endoscopic prosthesis was removed 6 months later. After 1 year, biologic tests and MRCP ruled out any biliary dilatation or stenosis.

Patient 4 A 71-year-old man had an ampullary neoplasm. Jaun- Figure 5. Surgical exposure of the gallbladder and of the hepatoduodenal ligament. To expose the hepatoduodenal ligament and dice was first treated endoscopically with a prosthesis. At the gallbladder neck, the round ligament, the gallbladder body, and laparotomy, the gallbladder was in the plane of the the left liver should be lifted, as indicated by arrows. P, pylorus. Vol. 200, No. 5, May 2005 Savier et al Fusion of the Planes of the Liver 715

toneal nodule was located in front of segment 3. Despite an initial response to chemotherapy, the portal vein anomaly and the tumor extension precluded the planned hepatectomy.

DISCUSSION Among a large number of published hepatic anatomic variations, we focused on a particular variation, which can be integrated within a homogeneous entity. It associates a portal malformation and a malposition of the gallbladder fossa. It is distinct from anomalies of the portal vein bifurcation2,5,16,17 or from simple variations of the position of the gallbladder.18 Among 41 patients with left-sided gallbladder reported in the Japanese literature, only 5 had an anomaly of the intrahepatic portal 19 Figure 6. Cholangiography in a case of fusion of the planes. Oper- branches. Although the anatomic variation has been ative cholangiography through a cannula inserted into the cystic described elsewhere as “left-sided gallbladder” or “right- duct. Two metallic stylets were placed: a sagittal against the left sided round ligament,” we propose, based on our seven side of the falciform ligament and a transversal at the superior level of the pedicle. b2 and b3, bile duct for segments 2 and 3; ra, right patients, a new concept for this anatomic entity: two anterior bile duct; rp, right posterior bile duct (authors’ interpreta- liver lobes separated by a unique plane containing the tion, see discussion and Fig. 8). portal axis, without true bifurcation. Because the warning sign is the impossibility of distinguishing the mid- phy discovered the fusion of the planes, which was con- plane from the left intersectional plane, we propose the firmed by the MDCT. Her history of previous abdomi- term fusion of the planes, which summarizes the architec- nal surgery combined with the anatomic variation ture of the liver. Anatomic considerations are discussed justified an open cholecystectomy. The macroscopical in the following text. appearance of the liver confirmed the alignment of the round ligament and the gallbladder. A cholangiography Incidence and discovery ruled out any stone in the common bile duct. Postoper- From 1,678 patients, our 7 patients gave a ratio of 0.4%, ative followup was uneventful. which included Caucasian and African races. Asonuma and colleagues13 reported an incidence of 4 in 379 Patient 6 (1.1%), and Nagai and associates,14 an incidence of 3 in A 63-year-old man had hepatitis C–related cirrhosis 1,621 (0.2%). Yamasaki and coworkers11 listed 5 pa- with an A5 Child-Pugh score. The ultrasonographic tients among 2,210 patients studied by ultrasonography monitoring discovered a 15 ϫ 15 ϫ 13-mm single nod- (0.2%); Maetani and coauthors,12 4 in 327 (1.2%); and ule in segment 3 corresponding to a hepatocellular car- Kuwayama and colleagues,20 7 in 1,025 patients (0.7%). cinoma and fusion of the planes of the liver, which was Interestingly, Yoshida and coworkers21 reported a 0.5% confirmed by MDCT (Fig. 7). This tumor was success- ratio of cholangiograms without the left hepatic duct. fully destroyed by percutaneous radiofrequency Surprisingly, fusion of the planes is rarely reported, ablation. despite an incidence of about 0.5%, progress in hepatic imaging, and a very old original description.6 Interest- Patient 7 ingly, ultrasonography allowed us to discover fusion of A 50-year-old man had a histologically proved cholan- the planes in all our patients, and in almost all cases from giocarcinoma. Ultrasonography showed a 7 ϫ 5 ϫ 5-cm the literature.3,4,7-11,13,14,16,17,19,22-26 Yamasaki and associ- tumor involving most of the right anterior part of the ates11 reported a case in which the CT, after a hepatic liver. The main portal branch and the pedicles for the trauma, failed to detect the fusion of the planes, which central part of the liver were compressed by the tumor. A was subsequently recognized by an ultrasonographic ex- 1.5-cm nodule was described in segment 3, and a peri- amination. This highlighted the need for a three- 716 Savier et al Fusion of the Planes of the Liver J Am Coll Surg

Figure 7. Three-dimensional reconstruction in a case of fusion of the planes. Inferior view of the liver and successive three-dimensional volume rendering reconstruction of (A) the portal vein, (B) portal with hepatic veins, (C) veins with gallbladder, and (D) veins with gallbladder and liver. G, gallbladder; lhv and lhv’, two left hepatic veins; ivc, inferior vena cava; pv2, pv3, and pv4, portal vein for segments 2, 3, and 4; rasv, right anterior sectorial vein; R, Rex’s recessus; rhv, right hepatic vein; rpsv, right posterior sectorial vein. dimensional image of the architecture of the liver, which great variability of cystic junction with the bile duct.19,27 is far from being performed routinely. But in two- Several authors have reported laparoscopic cholecystec- dimensional planar images, fusion of the planes may be tomy in left-sided gallbladders.10,26,27 Based on their ex- suggested by measurement of respective inclinations of perience, it is recommended to recognize the anomaly the gallbladder fossa and of the Rex’s recessus (Fig. 2). immediately, lift the round ligament to improve the exposure, and move the position of the gallbladder- Surgical consequences retracting port to the right and the right-hand operating Treatment of cholelithiasis in the presence of port.27 Other recommendations are to keep the dissec- fusion of the planes tion near the neck of the gallbladder to avoid aligning During data collection, seven patients with iatrogenic the cystic and main bile ducts by lifting the gallbladder, injury of the biliary tree were referred to our department. to perform an anterograde cholecystectomy to pull the Among these, two patients had a laparoscopic cholecys- gallbladder to the right, and to perform a laparatomy in tectomy associated with a fusion of the planes of the liver case of difficulties.28 (patients 2 and 3).This high proportion of injury should stress the potential danger of this association to the bil- Hepatectomy in the presence of fusion of the iary tract. During laparoscopic cholecystectomy, the planes main problem seems to be surgical exposure because of During a hepatectomy, the risk of a major vascular mis- the vicinity of several bile ducts in a limited space under take is increased. The involved mechanisms could be the round ligament, including the hepatoduodenal liga- related to the absence of the left hepatic pedicle, with the ment just behind the gallbladder neck (Fig. 5); the ab- following consequences: tie the bile ducts too closely to sence of the quadrate lobe that can be lifted; and the the bile convergence during a hepatectomy on the left Vol. 200, No. 5, May 2005 Savier et al Fusion of the Planes of the Liver 717

Table 1. Operative Consequences of Fusion of the Planes of the Liver Preoperative First author (year) Gender Age (y) Pathology recognition Operation Consequence Ozeki (1989)8 F 66 Acute cholecystitis No Cholecystectomy None Yamasaki, (1991)11 F 68 Gastric ulcer and biliary No Gastrectomy, None stone percutaneous bile duct drainage, cholecystectomy, choledocotomy Ikoma (1992)3 F 66 Cholangiocarcinoma No Left lateral sectionectomy* Clamping of the “left branch” devascularized 3/4 of the liver Regimbeau (2003)4 M 26 B-cell lymphoma No Left lateral sectionectomy* Bile duct injury Present series F 26 Biliary stone No Laparoscopic Bile duct injury cholecystectomy Present series F 57 Biliary stone No Laparoscopic Bile duct injury cholecystectomy Shirono (1990)25 F 47 Sigmoid cancer with Yes Colectomy and left lateral None synchronous metastasis sectionectomy* and nonanatomic resection in segment 5 Yamasaki (1991)11 M 53 Cirrhosis, Yes Segmentectomy 6 None hepatocellular carcinoma in segment 6, preduodenal portal vein, annular pancreas Uesaka (1995)24 F 68 Postoperative bile duct Yes Percutaneous transhepatic None stricture after distal biliary drainage gastrectomy for cancer Uesaka (1995)24 M 53 Metachrones hepatic Yes Right hepatectomy None metastases from a distal bile duct cancer Nishio (1995)10 F 57 Biliary stone Yes Laparoscopic None cholecystectomy Nagai (1997)14 M 67 Hepatocellular Yes Anterosuperior None carcinoma hepatectomy ϩ part of the caudate lobe† Nagai (1997)14 F 32 Acalculous cholecystitis Yes Cholecystectomy None Nagai (1997)14 M 67 Cholangiocarcinoma of Yes Anterior segmentectomy None the main bile duct ϩ caudate lobe resection† ϩ pancreatectomy Chung (1997)26 F 16 Epigastric pain Yes Laparoscopic None cholecystectomy Asonuma (1999)13 M 48 Candidate for left liver Yes Left lateral section None donation donation Asonuma (1999)13 M 29 Candidate for left liver Yes Left lateral section None donation donation Asonuma (1999)13 F 35 Candidate for left liver Yes Left lateral section Graft portal vein donation donation interposition Present series F 46 Biliary stone Yes Cholecystectomy None Present series M 63 Hepatocellular Yes Percutaneous None carcinoma, cirrhosis radiofrequency ablation Present series M 71 Pancreatic cancer Yes Cholecystectomy and None pancreatectomy *Or bisegmentectomy 2, 3. †Terminology used in cited reference. 718 Savier et al Fusion of the Planes of the Liver J Am Coll Surg

Figure 8. Relation between fusion of the planes of the liver and normal anatomy. Fusion of the planes (A) is assumed to correspond to incomplete development of the central part of the liver. In contrast, development of the central region separates the medial plane (MP) and the left intersectional planes (LIP) (B). Also developing in the central area are the horizontal part of the left portal branch (C) and the right anterior branch, with its known variations (C), resulting in normal hepatic architecture (D). (Taboury’s hypothesis.) side of the falciform ligament;4 and mistake a left portal mental anatomy, the immediate interpretation, from the branch and the vertical axis of the portal trunk.3 Conse- external appearance of the liver, would be that segment 4 quently, in the transplantation field, fusion of the planes was atrophic. In fact, the square lobe was never observed is an absolute contraindication of harvesting the right and no parenchyma was visible between the Rex- liver from a living donor1,13 (Patient 1). Harvesting the Cantlie’s line and the left part of the falciform ligament. left lobe has been described, but transplantation is a But study of the vascularization revealed the presence of demanding procedure because it requires complex bili- small venous branches from the top of the Rex’s recessus ary or portal vein anastomosis.13 (Fig. 7A) and the presence of a small hepatic vein in the From our experience and from the literature, preop- median position (Fig. 7B). Consequently, isolated as- erative recognition appeared essential to prevent opera- sumption of an agenesis of segment 4 seems to us insuf- tive accidents. From detailed cases, we found six cases of ficient to describe the anomaly. From anatomy and em- fusion of the planes, in which operations were per- bryology, we know that portal and hepatic veins of 20- to formed without preoperative recognition, and there 24-mm embryos resemble fusion of the planes;29 that the were four serious consequences. In contrast, no technical central portion of the liver develops secondarily between problems were reported for 15 other patients who had two primitive lateral lobes;5 and that there is variability hepatobiliary procedures (Table 1). in the position of the anterior sectorial branch.5 So we hypothesize that the origin of fusion of the planes, as Anatomic and embryogenetic assumptions observed in adults, would correspond to an embryologic To explain the anomaly, Hochstetter6 and later Matsu- step of liver development. In the case of incomplete de- moto23 referred to the persistence of the right umbilical velopment of the central part of the liver, the left inter- vein instead of the left.7 Considering the hepatic seg- sectional plane resembles the midplane, which could not Vol. 200, No. 5, May 2005 Savier et al Fusion of the Planes of the Liver 719 yet be clearly identified between the outlines of the por- sided umbilical portion). Jpn J Med Ultrasonics 1991; tal branches for the future segment 4 and the future right 18:436–443. 12. Maetani Y, Itoh K, Kojima N, et al. Portal vein anomaly associ- sector. The absence of growth of segment 4 prevented ated with deviation of the ligamentum teres to the right and shift of the round ligament toward the left, the forma- malposition of the gallbladder. Radiology 1998;207:723–728. tion of the horizontal part of the left bile duct, the for- 13. Asonuma K, Shapiro AM, Inomata Y, et al. Living related liver transplantation from donors with the left-sided gallbladder/ mation of the quadrate lobe between the round ligament portal vein anomaly. Transplantation 1999;68:1610–1612. and the gallbladder, and the curving of its anterior 14. Nagai M, Kubota K, Kawasaki S, et al. Are left-sided gallblad- branch30 (Fig. 8). Additional embryologic studies are ders really located on the left side? Ann Surg 1997;225:274– 280. needed to refine this hypothesis. 15. Strasberg SM, Belghiti J, Clavien PA, et al. The Brisbane 2000 In conclusion, alignment of the gallbladder and terminology of liver anatomy and resections. HBP 2000;2:333– round ligament is easily recognized by preoperative ul- 339. 16. Hiramatsu K, Nagino M, Kamiya J, Nimura Y. Anomaly of the trasonography. It is easily suspected during operation. portal vein with an anomalous hepatic vein–the first case report. Because this anomaly is associated with an absence of Hepatogastroenterology 2001;48:1142–1144. portal bifurcation, detection is essential if biliary surgery 17. Chevallier P,Oddo F, Baldini E, et al. Agenesis of the horizontal segment of the left portal vein demonstrated by magnetic reso- or hepatectomy is considered. nance imaging including phase-contrast magnetic resonance venography. Eur Radiol 2000;10:365–367. 18. Gross RE. Congenital anomalies of the gallbladder. Arch Surg Acknowledgment: We thank Christina Quaiser for Ger- 1936;32:131–162. manic translation and Anne Du Verdier for reviewing this 19. Ogawa T, Ohwada S, Ikeya T, et al. Left-sided gallbladder with article. anomalies of the intrahepatic portal vein and anomalous junction of the pancreaticobiliary ductal system: a case report. Hepa- togastroenterology 1995;42:645–649. 20. Kuwayama M, Takeuchi K, Tsuruoka N, et al. Ultrasonic eval- REFERENCES uation of the different branching patterns of the portal vein in 1. Deshpande RR, Heaton ND, Rela M. Surgical anatomy of seg- the hepatic hilum. Jpn J Med Ultrasonics 1989;16:346–353. mental liver transplantation. Br J Surg 2002;89:1078–1088. 21. Yoshida J, Chijiiwa K, Yamaguchi K, et al. Practical classification 2. Hardy KJ, Jones RM. Failure of the portal vein to bifurcate. of the branching types of the biliary tree: an analysis of 1,094 Surgery 1997;121:226–228. consecutive direct cholangiograms. J Am Coll Surg 1996;182: 3. Ikoma A, Tanaka K, Hamada N, et al. Left-sided gallbladder 37–40. with accessory liver accompanied by intrahepatic cholangiocar- 22. Matsusue S, Kashihara S, Koizumi S. Pancreatectomy for carcinoma of the head of the pancreas associated with multiple cinoma. (In Japanese). Nippon Geka Gakkai Zasshi 1992;93: anomalies including preduodenal portal vein. Jpn J Surg 1984; 434–436. 14:394–398. 4. Regimbeau JM, Panis Y, Couinaud C, et al. Sinistroposition of 23. Matsumoto H. A newer concept of the segmentation of the liver. the gallbladder and the common bile duct. Hepatogastroentero- Jpn J Med Ultrasonics 1986;13:551–552. logy 2003;50:60–61. 24. Uesaka K, Yasui K, Morimoto T, et al. Left-sided gallbladder 5. Couinaud C. Surgical anatomy of the liver revisited. Paris: with intrahepatic portal venous anomalies. J Hepatobiliary Pan- Couinaud, 1989. creat Surg 1995;2:425–430. 6. Hochstetter F. Anomalies of the portal vein and of the umbilical 25. Shirono K, Kounosu H, Kubo H, et al. Case of left-sided gall- vein related to a defect or to a left-sided gallbladder. (In German). bladder accompanying the anomalous intrahepatic portal and Arch Anat Entwick 1886;369–384. venous branching. Jpn J Gastroenterol Surg 1990;23:1477. 7. Moriyama S, Mori T, Umegae S, et al. Two cases of left-sided 26. Chung CC, Leung KL, Lau WY, Li AK. Ectopic gallbladder gallbladder with intrahepatic portal vein anomaly. Jpn J Gastro- revisited, laparoscopically: a case report. Can J Surg 1997;40: enterol Surg 1990;23:2813–2817. 464–466. 8. Ozeki Y, Onitsuka A, Hino A. Anomalous branching of intra- 27. Idu M, Jakimowicz J, Iuppa A, Cuschieri A. Hepatobiliary anat- hepatic portal vein associated with anomalous position of round omy in patients with transposition of the gallbladder: implica- ligament. Kanzou 1989;30:372–378. tions for safe laparoscopic cholecystectomy. Br J Surg 1996;83: 9. Ozeki Y, Onitsuka A, Hayashi M, Sasaki E. Left-sided gallblad- 1442–1443. der: report of a case and study of 26 cases in Japan. (In Japanese). 28. Strasberg SM. Avoidance of biliary injury during laparoscopic Nippon Geka Gakkai Zasshi 1987;88:1644–1650. cholecystectomy. J Hepatobiliary Pancreat Surg 2002;9:543– 10. Nishio H, Hasegawa H, Ogiso S, et al. A case of the left-sided 547. gallbladder with gallstones treated with laparoscopic cholecys- 29. Mall FP. A study of the structural unit of the liver. Am J Anat tectomy. Tan to Sui 1995;16:697–701. 1906;5:227–308. 11. Yamasaki H, Miyasaki S, Tanemura M, et al. Clinical signifi- 30. Sales JP, Hannoun L, Sichez JP, et al. Surgical anatomy of liver cance of anomalous branching of intrahepatic portal vein (right- segment IV. Anat Clin 1984;6:295–304. Do Bone Marrow Micrometastases Correlate with Sentinel Lymph Node Metastases in Breast Cancer Patients?

Susan M Trocciola, MD, Syed Hoda, MD, Michael P Osborne, MD, FACS, Paul J Christos, MPH, MS, Heather Levin, BS, Diana Martins, PA-C, Joshua Carson, BS, John Daly, MD, FACS, Rache M Simmons, MD, FACS

BACKGROUND: Sentinel lymph node biopsies (SLNB) are used to detect axillary metastases as an important prognostic indicator for breast cancer patients. Bone marrow micrometastases (BMM) have also been shown to predict prognosis. This study examines whether SLNB and BMM are associated. STUDY DESIGN: A retrospective analysis was performed on 124 stages I to III breast cancer patients treated with mastectomy or lumpectomy, SLNB, and bone marrow aspiration between 1997 and 2003. SLNB were examined for the presence of metastases by hematoxylin and eosin (H&E) stains and also by immunohistochemistry (IHC) for lymph nodes negative by H&E. The kappa statistic was used to evaluate the association (agreement) between SLNB and BMM. RESULTS: In this study population, 36 patients (29%) had micrometastases detected in their bone marrow, and 51 patients (41%) had positive sentinel lymph nodes. Of the patients with positive BMM (n ϭ 36), 53% (19 of 36) had positive SLNB (14 of 19 by H&E and 5 of 19 by IHC). In patients with negative BMM (n ϭ 88), 36% (32 of 88) had a positive SLNB (27 of 32 by H&E and 5 of 32 by IHC). The kappa statistic and associated 95% confidence interval indicated poor agreement between SLNB and BMM (kappa ϭ 0.15; 95% CI ϭϪ0.03, 0.32). CONCLUSIONS: There was poor agreement between axillary metastases and micrometastases detected in the bone marrow. This study suggests that BMM and axillary metastases are not concordant findings in most patients. (J Am Coll Surg 2005;200:720–726. © 2005 by the American College of Surgeons)

Sentinel lymph node biopsy (SLNB) is used to detect diagnosis. Studies have suggested that bone marrow mi- axillary metastases as an important prognostic indicator crometastases (BMM) are predictive of prognosis and for breast cancer patients. But approximately 30% of associated with decreased survival.1-3 This study exam- patients with negative axillary lymph nodes by histologic ines the association between SLNB and BMM in breast analysis will have a systemic recurrence. Research is cur- cancer patients. rently focused on identifying other criteria that can help METHODS differentiate patients likely to suffer recurrence and to Patients offer them appropriate adjuvant therapy at the time of A retrospective analysis was performed on 124 stage I to III breast cancer patients treated with mastectomy or Competing interests declared: None. lumpectomy, SLNB, and bone marrow aspiration be- Funding through the Manhasset Women’s Coalition Against Breast Cancer, tween 1997 and 2003 at the New York Presbyterian Lahiff, Pratesi, Darrow, Scranton,Villa, Levy, Russell, Tennis Against Breast Hospital/Weill Medical College of Cornell University Cancer, and the Faicco and Finnell Research Grants. Presented at the American College of Surgeons 89th Annual Clinical Con- and the Weill-Cornell Breast Center. Patients under- gress, Chicago, IL, October 2003. went bone marrow aspiration from both iliac crests and Received October 22, 2003; Revised October 19, 2004; Accepted December SLNB at the time of the primary breast cancer surgery. 15, 2004. From The Breast Center, New York Presbyterian Hospital/Weill Medical Lymph node analysis College of Cornell University, New York, NY. Correspondence address: Rache M Simmons, MD, FACS,The Breast Center, SLNB were evaluated by frozen and permanent section 425 East 61st St, 8th Floor, New York, NY 10021. (hematoxylin and eosin [H&E] stain) and, when appro-

© 2005 by the American College of Surgeons ISSN 1072-7515/05/$30.00 Published by Elsevier Inc. 720 doi:10.1016/j.jamcollsurg.2004.12.029 Vol. 200, No. 5, May 2005 Trocciola et al Bone Marrow Micrometatases and Breast Cancer 721

Statistical analysis Abbreviations and Acronyms The kappa statistic was used to quantify the degree of BMM ϭ bone marrow micrometastases agreement between bone marrow micrometastases and DFS ϭ disease-free survival Ͼ ϭ sentinel lymph node metastases. Kappa values 0.75 ER estrogen receptors Յ H&E ϭ hematoxylin and eosin and 1 indicate excellent agreement, between 0.4 and IHC ϭ immunohistochemistry 0.75 indicate good agreement, and Ͻ 0.4 indicate mar- OS ϭ overall survival ginal or poor agreement. A kappa value of 0.0 indicates PR ϭ progesterone receptors SLNB ϭ sentinel lymph node biopsy no agreement better than chance expected agreement, and a negative value for kappa indicates a trend toward disagreement on the two measures. Our sample size was powered at 80% to detect an overall agreement (kappa) priate, by immunohistochemistry (IHC, cytokeratin AE between BMM and SLNB metastases Ն 0.45 (indicat- 1/3). All patients who had histologically negative axilla ing kappa values ranging from good agreement to excel- by H&E (frozen and permanent) underwent IHC anal- lent agreement). ysis. If a frozen section was performed, one section was Descriptive frequencies and percentages were calcu- taken through each lymph node and stained with H&E. lated to characterize the study population with respect to Permanent sections were examined for all sentinel demographics and tumor characteristics. Chi-square lymph nodes, which included three sections through the analysis was used to compare subgroups of patients (de- lymph node, stained with H&E. fined by tumor characterisics) with micrometastases to For H&E staining, a minimum of one metastatic cell, their bone marrow or with positive sentinel lymph node when present, was considered positive. All cells regarded biopsy. Subgroup analyses were exploratory and should as positive were morphologically consistent with a cell be interpreted with caution because of the issue of mul- extrinsic to the lymph node. Cells less than 0.2 mm in tiple comparisons. Because 12 subgroup analyses were size were difficult to identify on H&E slides and were performed, the significance level used to evaluate statis- often found only on IHC analysis. tical significance for these analyses was set at alpha ϭ For this study, a positive SLNB was defined as positive 0.05 / 12 ϭ 0.004. by frozen section, permanent section, or IHC. But for staging purposes, an SLNB was defined as positive only RESULTS if it was positive by H&E (permanent or frozen). Pa- Bone marrow aspirates were obtained from 124 patients tients who had positive SLNB (frozen or permanent) diagnosed with breast cancer. Table 1 shows the demo- underwent a complete axillary dissection either at the graphics of the patients. All but 1 patient were women, time of the original operation or later. In addition, com- with a mean age of 54 years (range 33 to 83 years). The pletion axillary dissection was performed on selected pa- majority of patients had infiltrating ductal carcinoma (n ϭ tients who were positive only on IHC. 107, 86%), with the remainder having infiltrating lobular carcinoma (n ϭ 14, 11%) or pure mucinous carcinoma (n ϭ Bone marrow analysis 3, 2%).The majority of patients (57%) had stage I breast Bone marrow aspirates were obtained from bilateral an- cancer at the time of primary operation. Median fol- terior iliac crests at the time of operation, as previously lowup was 18 months (range 0.1 to 5.6 years). described.4 A small skin incision was made and the aspi- In this study, 29% of patients (n ϭ 36) had microme- ration needle introduced to yield 5 mL of aspirate from tastases detected in their bone marrow; 41% (n ϭ 51) each side. Cytospins were prepared and stained for anal- had positive sentinel lymph nodes (Table 2). In patients ysis. A minimum of 1 million cells were screened for the with a positive BMM, 53% (19 of 36) had a positive presence of micrometastases using cytokeratin AE 1/3. SLNB (14 of 19 by H&E and 5 of 19 by IHC). In MCF7 (human breast cancer cell line) cells were used as patients with a negative BMM, 36% (32 of 88) had a positive controls; negative controls were from bone mar- positive SLNB (27 of 32 by H&E and 5 of 32 by IHC). row of patients with no known malignancy. All cells The mean number of positive cells in the bone marrow regarded as positive were morphologically consistent was 4 (range 1 to 33). with a cell extrinsic to the bone marrow. Table 2 shows the overall characteristics of the tumors of 722 Trocciola et al Bone Marrow Micrometatases and Breast Cancer J Am Coll Surg

Table 1. Demographics of 124 Patients Studied with lymph vessel invasion (p Ͻ 0.0001). Subgroup Characteristic n % analyses were not defined a priori and were considered Gender only exploratory. As a result, p values should be inter- Female 123 99 preted with caution because of the issue of multiple Male 1 1 comparisons. A value Ͻ 0.004 was considered statisti- Type of cancer cally significant for these analyses. Infiltrating ductal carcinoma 107 86 Infiltrating lobular carcinoma 14 11 Table 3 shows the relationship between findings in Mucinous carcinoma 3 2 the bone marrow and in the sentinel lymph nodes. The Stage* majority of patients in the study had both a negative I7157SLNB and a negative BMM (45%); 26% had a positive II 48 39 SLNB with a negative BMM. Approximately equal III 5 4 numbers had a negative SLNB with a positive BMM *Staged according to 6th edition of American Joint Committee on Cancer. (14%) or a positive SLNB with a positive BMM (15%). Table 3 further examines the association between the the patients in the study. The majority of patients in our findings in the sentinel lymph nodes and the bone mar- study had breast cancer involving small tumors (Յ 2 cm) row in different subgroups of patients. with positive estrogen receptors (ER) or progesterone Table 4 shows the poor association seen between receptors (PR), which had a high nuclear grade (II/III) BMM and SLNB as demonstrated by a kappa value of and did not involve lymphatic invasion. The data were 0.15 (95% CI ϭϪ0.03, 0.32). SLNB that were positive examined to see if any subgroup of patients was more only on H&E compared with BMM also showed a poor likely to have either a positive BMM or a positive SLNB. association (kappa ϭ 0.08, 95% CI ϭϪ0.10, 0.26). In None of these subgroups was more likely to have a pos- addition, there was a poor association between micro- itive BMM. Positive SLNB was significantly more com- metastases in the SLNB (IHC positive only) compared mon in patients with large tumors (p ϭ 0.0008) or those with BMM (kappa ϭ 0.18, 95%, CI ϭϪ0.05, 0.40).

Table 2. Tumor Characteristics Bone marrow micrometastases Sentinel lymph node biopsy Negative Positive Negative Positive Characteristics n%n%p Value* n%n%p Value* Total, n ϭ124 88 71 36 29 73 59 51 41 Tumor size† Յ2 cm, nϭ 96 68 71 28 29 64 67 32 33 Ͼ2 cm, nϭ24 17 71 7 29 Ͼ0.99 7 29 17 71 0.0008 Lymphatic invasion† No, nϭ 95 70 74 25 26 65 68 30 32 Yes, nϭ 26 15 58 11 42 0.11 7 27 19 73 Ͻ0.0001 Tumor grade* I, nϭ 36 28 78 8 22 28 78 8 22 II/III, nϭ 78 51 65 27 35 0.18 41 53 37 47 0.01 Nuclear grade* I, nϭ 11 8 73 3 27 6 55 5 45 II/III, nϭ 105 74 70 31 30 0.88 62 59 43 41 0.77 Estrogen receptor Negative, nϭ 201260840 10501050 Positive, n ϭ104 76 73 28 27 0.24 63 61 41 39 0.38 Progesterone receptor Negative, nϭ31 20 65 11 35 16 52 15 48 Positive, nϭ93 68 73 25 27 0.36 57 61 36 39 0.34 *p value calculated by a chi-square test. †Some subgroups do not add up to 124 because pathologic data were unavailable. Vol. 200, No. 5, May 2005 Trocciola et al Bone Marrow Micrometatases and Breast Cancer 723

Table 3. Comparison of Results in Sentinel Lymph Nodes and Bone Marrow Aspirations for 124 Patients and in Subgroups of Patients, Based on Tumor Characteristics SLNB؊ BMM؊ SLNB؉ BMM؊ SLNB؊ BMM؉ SLNB؉ BMM؉ Characteristics n row % n row % n row % n row % Total, n ϭ 124 56 45 32 26 17 14 19 15 Tumor size* Յ2 cm, n ϭ 96 48 50 20 21 16 17 12 13 Ͼ2 cm, n ϭ 24625114614625 Lymphatic invasion* No, n ϭ 95 50 53 20 21 15 16 10 11 Yes, n ϭ 26519103828935 Tumor grade* I, n ϭ 36 22 61 6 17 6 17 2 6 II/III, n ϭ 78 30 38 21 27 11 14 16 21 Nuclear grade* I, n ϭ 1132754532700 II/III, n ϭ 105 48 46 26 25 14 13 17 16 Estrogen receptor Negative, n ϭ 20 7 35 5 25 3 15 5 25 Positive, n ϭ 104 49 47 27 26 14 13 14 13 Progesterone receptor Negative, n ϭ 31 11 35 9 29 5 16 6 19 Positive, n ϭ 93 45 48 23 25 12 13 13 14 *Some subgroups do not add up to 124 because pathologic data were not available. BMM, bone marrow micrometastases; SLNB, sentinel lymph node biopsy.

The association of BMM and SLNB did not improve Identifying patients who are at risk for systemic recur- when a positive bone marrow was defined differently. rence is critical for clinicians to appropriately recom- For example, when a positive BMM was considered mend adjuvant therapies after breast cancer surgery. greater than 5 cells, kappa was 0.07 (95% CI ϭϪ0.04, Current prognostic factors, such as lymph node involve- 0.17), indicating a poor agreement. The same conclu- ment, tumor size, and hormone receptor status have sion held when the threshold for BMM was raised to 10 been used with limited success. cells (kappa ϭ 0.04, 95% CI ϭϪ0.05, 0.12). Technical aspects DISCUSSION Sloane and colleagues5 at the Ludwig Institute for Can- In this study, we found that there was poor or no agree- cer Research described the use of the polyclonal antigen ment between BMM and sentinel lymph node metasta- termed epithelial membrane antigen (EMA) for detection sis. BMM were more common in patients with ER neg- of metastatic cells in the bone marrow of breast cancer ative, PR negative, or high-grade tumors, although none patients. Epithelial membrane antigen is found against of these factors significantly predicted a positive BMM. human milk fat globule membrane, which was later Axillary metastases, measured by SLNB, were more found to have low specificity because it reacts with lym- common in patients with large tumors (p ϭ 0.0008) or phoid cells, especially plasma cells.6 those showing lymphatic invasion (p Ͻ 0.0001). The use of monoclonal antibodies was then devel-

Table 4. Association (Agreement) Between Sentinel Lymph Node Biopsy and Bone Marrow Micrometastases Kappa BMM versus 95% confidence Kappa BMM versus 95% confidence Kappa BMM versus 95% confidence SLNB (all) interval SLNB (H&E*) interval SLNB (IHC†) interval 0.15 (Ϫ0.03, 0.32) 0.08 (Ϫ0.10, 0.26) 0.18 (Ϫ0.05, 0.40) *Hematoxylin and eosin (H&E). †Immunohistochemistry (IHC). BMM, bone marrow micrometastases; SLNB, sentinel lymph node biopsy. 724 Trocciola et al Bone Marrow Micrometatases and Breast Cancer J Am Coll Surg oped to detect micrometastases in the bone marrow.7 method. Diel and colleagues2 published a prospective study BMM have been detected by a variety of different anti- involving 727 patients with primary breast cancer who had bodies.8 Osborne and associates9,10 looked at the use of undergone both bone marrow aspiration and axillary dis- immunofluorescene and immunocytochemical detec- section. These data demonstrated that the presence of tion of monoclonal antibodies, including anticytokera- BMM was an independent predictor of both disease-free tin monoclonal antibodies, for breast cancer cells in survival (DFS) and overall survival (OS). bone marrow. Both methods were found to have the Braun and associates3 also showed a similar result in a ability to detect 1 cancer cell in 1 million normal nucle- prospective study of 552 patients with stages I, II, and III ated bone marrow cells. The advantage of the immuno- breast cancer. In a median followup of 38 months, they cytochemical method is that it allows the morphology of showed that BMM was associated with distant metastases the cells to be examined and also provides a permanent and death from cancer-related causes (p Ͻ 0.001), but was record.10 Morphology is an important method for con- not associated with a higher occurrence of locoregional re- firming that positively stained cells represent cancer cells lapse (p ϭ 0.77). and helps to minimize false-positive results.11 A study by Mansi and coauthors12 with a longer fol- Other current techniques used to detect BMM in- lowup period expanded on these results. The study, involv- clude polymerase chain reaction and flow cytometry. ing 350 women with a followup of 12.5 years, showed that Limitations of the polymerase chain reaction technique although the presence of BMM was associated with de- include the inability to examine microscopic cells, the creased DFS and OS, it was not an independent prognostic possibility of false positives from cytokeratin pseudo- factor. genes, DNA rearrangements, and a relatively high cost.4 Flow cytometry is associated with a high cost and false Bone marrow micrometastases and axillary positives that may result from cytokeratin debris in the lymph nodes macrophages. Flow cytometry is also associated with a Braun and colleagues13 also looked at the relationship of lower sensitivity, in that it can detect 1 cancer cell per BMM to the axillary lymph node. In a study of 150 axillary 104 cells studied.4 node–negative (H&E) women, BMM were associated One of the more significant problems in comparing dif- with a reduced DFS and OS (p ϭ 0.032 and p ϭ 0.014, ferent studies involving BMM is the lack of standardiza- respectively), an association that was not seen with micro- tion. Prevalence rates for BMM vary from 4% to 48%, with metastases (IHC) in the lymph nodes. This study also much of this variation explained by lack of standardization found that in axillary node–negative women, there was no in terms of number of sites aspirated, number of cells ex- association between BMM and lymph node micrometas- amined, and type of antibodies used.8 tases, with only 2 in150 patients having micrometastases in both areas. Prognostic significance There have been other studies that have shown BMM Many studies have looked at the association of BMM with does not predict DFS or OS. In a study by Gebauer and other prognostic factors. A number of groups have found associates,14 212 women with node–negative axilla (H&E) that a higher rate of BMM is associated with increasing were studied to see if micrometastases to the lymph size1,2,12 of tumor. For example, Diel and coworkers2 found nodes or bone marrow predicted survival. After a 10- BMM in 30%, 42%, 62%, and 75% ofT1,T2,T3, andT4 year followup, they found that immunohistochemical tumors, respectively. In our study, we did not find a higher analysis of either the lymph nodes or the bone marrow rate of BMM in patients with larger tumors. This may have did not add prognostic information, and instead advo- been because of the smaller sample size in this subgroup or cated additional sectioning of axillary lymph nodes.14 it could be an artifact of multiple subgroup comparisons. The American College of Surgeons Oncology Group Additionally, it could have been from the fact that the ma- (ACOSOG) Z0010 trial is a large, prospective multicenter jority of the tumors in our study were Յ 2 cm (T1). trial to evaluate the presence and prognostic value of micro- The relationship of lymph node status measured by ax- metastases in the bone marrow and sentinel lymph nodes of illary dissection stained with H&E and BMM has been breast cancer patients. This trial has recently closed and the studied by several investigators. Most of these studies have results are currently being analyzed. focused on the difference in the prognostic value of each Our study is important because it shows that there is Vol. 200, No. 5, May 2005 Trocciola et al Bone Marrow Micrometatases and Breast Cancer 725 poor or no evidence of agreement between sentinel lymph 8. Osborne MP,Rosen PP.Detection and management of bone mar- node status and BMM. This lack of agreement was present row micrometastases in breast cancer. Oncology 1994;8:25–31. 9. Osborne MP,Asina S, Cote RJ, et al. Immunofluorescent mono- both for sentinel lymph nodes examined with H&E and clonal antibody detection of breast cancer in bone marrow: sen- those with only micrometastases. Previous studies have sitivity in a model system. Cancer Res 1989;49:2510–2513. looked at micrometastases in patients who underwent a 10. Osborne MP,Wong GY, Cote RJ, et al. Sensitivity of immunocytochemical detection of breast cancer cells in human bone complete axillary dissection. Now that sentinel lymph marrow. Cancer Res 1991;51:2706–2709. nodes have become the standard of care, we believe that our 11. Borgen E, Beiske K, Trachsel S, et al. Immunocytochemical comparison is more pertinent. Validation of these data detection of isolated epithelial cells in bone marrow: non- awaits such studies as the American College of Surgeons specific staining and contribution by plasma cells directly reac- tive to alkaline phosphatase. J Pathol 1998;185:427–434. Oncology Group Z0010, and the use of BMM remains 12. Mansi JL, Gogas H, Bliss JM, et al. Outcome of primary-breast investigational. cancer patients with micrometastases: a long-term follow-up. Our followup interval in this study is too short to Lancet 1999;354:197–202. 13. Braun S, Cevatli BS, Assemi C, et al. Comparative analysis of analyze BMM as a prognostic marker. More information micrometastasis to the bone marrow and lymph nodes of node- will be obtained from current and future large multi- negative breast cancer patients receiving no adjuvant therapy. center trials to confirm or contradict the prognostic J Clin Oncol 2001;19:1468–1475. 14. Gebauer G, Fehm T, Merkle E, et al. Micrometastases in axillary value of BMM implied in historical studies. lymph nodes and bone marrow of lymph node-negative breast cancer patients—prognostic relevance after 10 years. Anticancer Author Contributions Res 2003;23:4319–4324. Study conception and design: Simmons Acquisition of data: Trocciola, Hoda, Levin, Martins, Carson Invited Commentary Analysis and interpretation of data: Trocciola, Osborne, Christos, Simmons V Suzanne Klimberg, MD, FACS Drafting of manuscript: Trocciola, Simmons Little Rock, AR Critical revision: Trocciola, Osborne, Christos, Daly, Simmons For more than a decade, we have been wondering what the Statistical expertise: Christos implication of bone marrow micrometastasis (BMM) is in the prognosis of breast cancer. Studies have been incon- sistent and small. Trocciola and colleagues1 point out REFERENCES that the now closed American College of Surgeons On- 1. Cote RJ, Rosen PP,Lesser ML, et al. Prediction of early relapse in cology Group Z-10 Trial may answer this question. The patients with operable breast cancer by detection of occult bone marrow micrometastases. J Clin Oncol 1991;10:1749–1756. flaws continue, secondary to probable overcall of BMM 2. Diel IJ, Kaufman M, Costa SD, et al. Micrometastastic breast with only a single stain. Relying simply on the patholo- cancer cells in bone marrow at primary surgery: prognostic value gist to distinguish between epithelial cells, plasma cells, in comparison with nodal status. J Natl Cancer Inst 1996; and tumor cells is not enough. Available triple stains are 69:566–571. 3. Braun S, Pantel K, Muller P, et al. Cytokeratin-positive cells in required to tell the difference, and few studies have used the bone marrow and survival of patients with Stage I, II or III these. We are also plagued with the same question of breast cancer. N Engl J Med 2000;342:525–533. how many cells will constitute a positive sentinel lymph 4. Simmons RM, Hoda S, Osborne M. Bone marrow micrometastases in breast cancer patients. Am J Surg 2000;180:309–312. node (SLN) and now a positive bone marrow. The def- 5. Sloane JP, Ormerod MG, Neville AM. Potential pathological inition of Trocciola and colleagues was one cell for bone application of immunocytochemical methods to the detection marrow, but do we know what one cell means? By their of micrometastases. Can Res 1980;40:3079–3081. 6. Delsol G, Gatter KC, Stein H, et al. Human lymphoid cells results, not much. Did their study even have the power express epithelial membrane antigen: Implications for diagnosis to distinguish this? It would have been nice to see a of human neoplasms. Lancet 1984;1124–1128. breakdown of size of metastases and correlation. 7. Cote RJ, Rosen, PP,Osborne MP,et al. Monoclonal antibodies I worry about the lack of definition of a positive detect occult breast carcinoma metastasis in the bone marrow of patients with early stage disease. Am J Surg Path 1988;12:333– SLN; how many cells have to be present? How did so 340. many positive nodes be positive by frozen section (FS) 726 Trocciola et al Bone Marrow Micrometatases and Breast Cancer J Am Coll Surg and not be positive by permanent section? It would cells by permanent staining. In our study, 25 patients have been nice to see a breakdown of correlation of had positive sentinel lymph nodes on frozen section, an frozen section positive only and BMM, permanent additional 16 patients had positive sentinel lymph nodes only and BMM and so forth. on permanent section, and 10 were positive by immu- This also leaves us with who should we routinely nohistochemistry only. get bone marrow aspirations on in our patients? How After originally presenting our data at the Ameri- do we select these patients to get bone marrow aspi- can College of Surgeons meeting, we examined the rations? I don’t think we are quite ready to use this association between bone marrow micrometastases clinically. and sentinel lymph node metastases defined by method of detection (frozen, permanent, or immun- histochemistry). A portion of these data is shown in REFERENCE Table 4 in the article. There continued to be poor 1. Trocciola SM, Hoda S, Osborne MP, et al. Do bone marrow micrometastases correlate with sentinel lymph node metastases in agreement between findings in the sentinel lymph breast cancer patients? J Am Coll Surg 2005;200:720–725. and bone marrow even when we looked at sentinel lymph nodes that were positive by frozen section, permanent section, or immunohistochemstry. The presence of one positive cell by either hematoxy- Reply lin and eosin or immunohistochemistry was diagnosed as a positive sentinel lymph node. But cells that were Susan M Trocciola, MD, Ͻ 0.2 mm in size were hard to diagnose on hematoxylin Rache M Simmons, MD, FACS and eosin and often needed confirmation with immu- New York, NY nohistochemistry. We examined the association between Thank you for allowing us to share our data. In our size of sentinel lymph node metastases (number of pos- study, a positive sentinel lymph node was identified by itive cells) and presence of a positive bone marrow and the histologic technique in which it was first identified found there was no agreement. (ie, frozen section, permanent section [hematoxylin and At our institution, two of our breast surgeons rou- eosin staining], or immunohistochemistry). If a patient tinely do bone marrow aspirations on all patients with had malignant cells initially on frozen section, she would invasive breast cancer at the time of operation, regardless be categorized as having a positive sentinel lymph node of the size of the primary tumor. Our study also included by frozen (not permanent) section, even though the per- patients from other surgeons, at our institute, who par- manent sections would have also shown malignancy. In ticipated in the American College of Surgeons Oncology reality, all sentinel lymph nodes that were positive by Group Z-10 trial and did bone marrow aspirations as frozen section were then confirmed to have metastatic part of this trial. Cystadenoma and Cystadenocarcinoma of the Liver: A Single Center Experience

David P Vogt, MD, FACS, J Michael Henderson, MD, FRCS, Elaine Chmielewski, RN

BACKGROUND: Biliary cystadenomas and cystadenocarcinomas comprise 5% of cystic lesions in the liver. Cystadenomas are often incorrectly diagnosed as simple cysts, which results in inadequate therapy. Recurrence and possible malignant transformation are consequences of incomplete excision. Cystadenocarcinomas are very rare tumors that are felt to be biologically indolent. STUDY DESIGN: A retrospective review of 18 cystadenomas and 4 cystadenocarcinomas treated at the Cleveland Clinic from July 1985 to November 2002. RESULTS: All 18 patients with cystadenomas were women; mean age was 48 years. The majority (16 of 18) were symptomatic. Preoperative CT scans demonstrated cyst(s) with septations in all patients. Fifty-five percent had undergone prior intervention(s) to treat the cyst. Thirteen patients had complete excision of the cystadenoma, either by enucleation or liver resection. None of the patients developed recurrent cystadenomas (mean followup 37 months). Of 4 patients with cystadenocarcinoma, 3 were women; mean age was 60 years. All were symptomatic. Preopera- tive CT scans demonstrated masses with both cystic and solid components. No patient had undergone prior intervention. All had a liver resection. Two patients died of metastatic disease at 6 and 12 months, respectively. One patient is alive and disease-free at 16 years; 1 is alive with metastatic disease 10 years after the liver resection. CONCLUSIONS: Cystadenomas are uncommon tumors that are often incorrectly diagnosed as simple cysts. Preoperative imaging that demonstrates the presence of internal septations highly suggests the diagnosis of cystadenoma. Intraoperative biopsy and frozen section(s) are essential, although they are not 100% accurate. Cystadenomas require complete excision to prevent recurrence and the possibility of malignant transformation. Cystadenocarcinomas are very rare. Despite complete resection, cystadenocarcinomas can recur in a short period of time. The biologic behavior of these tumors can vary widely. (J Am Coll Surg 2005;200:727–733. © 2005 by the American College of Surgeons)

Cystic lesions of the liver are being discovered more fre- structure with internal septations. Cystadenomas are of- quently because of advances in abdominal imaging over ten incorrectly diagnosed as simple cysts and are treated the past several years. The majority of cystic lesions are with either aspiration or incomplete excision. The result benign, simple cysts that are rarely symptomatic. Cystic is recurrence or persistence of the cyst and its associated neoplasms are estimated to comprise approximately 5% symptoms. The potential for malignant transformation of liver cysts.1 Cystadenomas occur predominantly in into a cystadenocarcinoma is documented.2,3 Several pa- the liver, although they have been reported in the biliary tients have been referred to our institution over the years tree (including the gallbladder) as well. Cystadenomas because of recurrence of cystic mass(es) in the liver. The occur predominately in women and are usually symp- focus of this article is to review our experience, particu- tomatic at the time of diagnosis. Imaging studies, pri- larly longterm followup, of patients who underwent op- marily CT and ultrasonography, demonstrate a cystic erations for cystadenoma or cystadenocarcinoma at our institution in an effort to formulate proper management Competing interests declared: None. strategy. Received November 4, 2004; Revised January 10, 2005; Accepted January 10, 2005. From the Department of General Surgery, Cleveland Clinic Foundation, METHODS Cleveland, OH. Correspondence address: David P Vogt, MD, Department of General Sur- This is a retrospective review of all patients who under- gery, Cleveland Clinic Foundation, 9500 Euclid Ave, Cleveland, OH 44195. went surgical procedures at the Cleveland Clinic Foun-

© 2005 by the American College of Surgeons ISSN 1072-7515/05/$30.00 Published by Elsevier Inc. 727 doi:10.1016/j.jamcollsurg.2005.01.005 728 Vogt et al Cystadenocarcinoma of the Liver J Am Coll Surg dation for biliary cystadenoma or cystadenocarcinoma from July 1985 to September 2002. Preoperative data extracted from the patients’ charts included gender, age, symptoms, physical findings, and results of imaging studies. For patients who were referred for recurrence after having a surgical procedure elsewhere, both the operative notes and pathology slides were reviewed, if they were available. Perioperative data extracted included the operative note, pathology report, and hospital summary. Followup was conducted by clinic visits or phone reports, or both. Many patients had followup imaging studies.

Figure 1. A large cystadenoma arising in the left lobe of the liver. RESULTS Note the internal septations. Cystadenomas Demographics, symptoms, signs, and imaging Surgical procedures All 18 patients were women with a mean age of 48 years. All Thirteen patients had complete removal of the cystade- but two patients had symptoms of upper abdominal or noma, including four left lateral sectionectomies, one right upper quadrant pain or discomfort for a mean of 16 left hepatectomy, one right hepatectomy, seven open months (range 0.3 to 108 months). Only two patients had enucleations, and one laparoscopic enucleation. One of a palpable mass in the upper abdomen. Computed tomog- the large, centrally located lesions was enucleated with raphy had been performed in all patients; ultrasonography the aid of total vascular exclusion (Fig. 2). Five patients or MRI scanning, or both, was done in 75%. CT scans had partial resection of the cystadenoma. In one patient, demonstrated septations in the cyst(s) of all patients. Mean the intraoperative frozen section was interpreted as a maximum diameter of the cysts was 12.5 cm (range 7 to simple cyst, which was adherent to the right hepatic duct 22 cm) (Fig. 1). Twelve of the cysts were in the left lobe, for several centimeters. In another patient, the posterior three in the right, one involved both lobes, and two were in wall of the tumor involved the IVC, precluding safe the caudate lobe. complete excision. In the third patient, six frozen sections taken at the time of laparoscopic unroofing were all Prior interventions interpreted as a simple cyst. The final pathologic diag- Ten of the 18 patients (55%) had undergone 17 earlier nosis was read as a cystadenoma. That patient is sched- interventions, including percutaneous aspiration (4 pa- uled to have open enucleation performed in the future. tients), aspiration and instillation of ethanol (1 patient), Early in our experience, two patients underwent partial laparoscopic unroofing (3 patients), open unroofing excisions for what appeared grossly to be simple benign with or without placing omentum into the cyst cavity (4 cysts. Frozen sections were not performed on either pa- patients), decompression into a Roux limb (1 patient), tient. Both were cystadenomas on final pathology and no further surgical procedures were performed. There and partial excision combined with external drainage (1 were no postoperative deaths. patient) (Table 1). Four patients had more than one intervention. Two of these patients had earlier proce- Pathology dures performed at our center: one laparoscopic unroof- All pathology reports were reviewed. In 10 of 18 pa- ing and one partial excision combined with external tients, lesions were specifically classified as cystadeno- drainage. Six frozen-section reports on the patient who mas with mesenchymal stroma. None of the specimens had laparoscopic unroofing were interpreted as a benign, was categorized as serous cystadenomas. simple cyst, but the cyst recurred and was enucleated 3 Followup months later. The second patient had the cystadenoma Followup information was available on all 13 patients enucleated 5 months later (Fig. 2). who had complete cystadenoma excision, ranging from Vol. 200, No. 5, May 2005 Vogt et al Cystadenocarcinoma of the Liver 729

Table 1. Prior Procedures for Cystadenomas Age (y) Date; prior procedure Date; CCF operation Followup 44 1988; partial excision, omentum in cavity 5/94; partial excision 11/94; asymptomatic 5-cm cyst 45 1986; partial excision 1/90; partial excision Not available 46 1996; aspiration 2/01; complete laparscopic CT 10/01; no recurrence 1999; aspiration excision 1/00; Roux limb 52 2/00; aspiration (X2) 12/00; right hepatectomy US 7/01; no recurrence 6/00; partial excision (laparoscopic) 9/00; partial excision, omentum in cavity; open 38 1/00; aspiration, ETOH instillation 3/00; enucleation US 4/01; no recurrence 51 12/99; partial excision (laparoscopic) 8/00; left lateral sectionectomy US 11/01; no recurrence 65 10/88; partial excision 6/90; enucleation US 7/01; no recurrence 2/89; partial excision 2/90; partial excision, external drainage 90 7/85; aspiration 7/85; left lateral sectionectomy Not available 56 1/01; aspiration 4/01; left lateral sectionectomy US 9/01; no recurrence 72 8/01; partial excision (laparoscopic) 11/01; enucleation Not available CCF, Cleveland Clinic Foundation; US, ultrasonography.

1 month to 11 years (mean 37 months). All but 1 pa- had CT scans that demonstrated cystic lesions with solid tient, who underwent enucleation of a segment 1 cyst- components (Figs. 3, 4). The male patient had a mag- adenoma only 1 month before this analysis, had either netic resonance cholangiopancreatography in addition followup ultrasonography (seven patients) or CT scan to a CT scan. No patient had undergone an earlier sur- (five patients). One patient, who had undergone a left gical procedure. lateral sectionectomy, had a recurrent cystic lesion identified 3 years postoperatively. The excised recurrent cys- Surgical procedures, pathology tic mass was a fibrous pseudocyst from the earlier resec- The surgical procedures performed included 1 right tion of the left lateral lobe of the liver. There was no hepatectomy, 1 extended right hepatectomy (Fig. 4), 1 evidence of a recurrent cystadenoma. left lateral sectionectomy, and 1 enucleation from the Only two of the five patients who had partial excision medial portion of the left lobe (Fig. 3). There were no of the cystadenoma at our institution have followup perioperative deaths. Two patients had papillary cyst- data. Three patients could not be located. The patient adenocarcinomas (Figs. 3, 4). One was arising in a cyst- who had 6 negative frozen sections at the time of laparoscopic unroofing had a followup CT scan 13 months later because of recurrent symptoms. The CT demonstrated a 9 ϫ 6-cm cyst in the right lobe of the liver. She is scheduled to have complete excision of the cystadenoma in the future. The other patient has had percutaneous aspirations of recurrent cysts at another institution, 10 and 16 years after her initial operation at the Cleveland Clinic.

Cystadenocarcinoma Demographics, symptoms, signs, and imaging Four patients in this series had cystadenocarcinomas. The 3 women and 1 man had a mean age of 60 years (range 55 to 78 years). All had either right upper quad- Figure 2. A central cystadenoma that had been partially excised in the past. An external drain is visible. The tumor was successfully rant or upper abdominal pain or discomfort. Two pa- enucleated using total vascular exclusion and there was no recur- tients had palpable masses in the upper abdomen. All rence after 10 years. 730 Vogt et al Cystadenocarcinoma of the Liver J Am Coll Surg

and Edmondson7 (17 cases), and Lewis and colleagues8 (15 cases). Our series is the largest single institution report to date with 22 patients; 18 cystadenomas and 4 cystadenocarcinomas. More than 85% of cystadenomas are reported in women.7,9,10 All cystadenomas in our series were in women. The mean age of our patients was 48 years, which falls into the range of 41 to 53 years from other reports.4,9,10 The cause of cystadenomas is not known. These cystic neoplasms, which involve primarily the hepatic parenchyma and occasionally the biliary tree including the gallbladder, may originate from a congenitally aberrant Figure 3. Cystadenocarcinoma. Note solid and cystic areas. The bile duct or directly from a primitive hepatobiliary stem tumor was enucleated and there was no recurrence after 16 years. cell.4,6-8,11 Evidence does not support that cystadenomas are derived from ectopic ovarian tissue.7 The almost ex- adenoma with mesenchymal stroma and was low grade, clusive female predominance suggests a strong hormonal and the second had a spectrum of papillary changes, influence. ranging from benign to severe dysplasranging from be- Grossly, cystadenomas are lobulated, multiloculated, nignia and invasive adenocarcinoma. The male patient and contain clear to mucinous fluid of various col- had a mucinous biliary cystadenocarcinoma. The fourth ors.7,9,12 The internal lining is generally smooth, al- specimen was classified no further than a cystadenocar- though it can be trabeculated or contain polypoid cysts cinoma; mesenchymal stroma was not described. that may project into the lumen.7 Histologically, cystadenomas can be divided into two subgroups that are Followup distinguished by presence or absence of mesenchymal Followup was available for all patients. The patient who (“ovarian-like”) stroma between an inner epithelial lin- had the low-grade papillary cystadenocarcinoma enucle- ing and an outer connective tissue capsule.12 Wheeler ated remains alive and disease free 16 years later. The and colleagues7 were the first to describe cystadenomas other patient with a papillary cancer died 6 months after with mesenchymal stroma (CMS), which occur exclu- resection from metastatic disease. The male patient with sively in women. CMS consist of three distinct layers of the mucinous biliary cystadenocarcinoma was dead at tissue: an epithelial layer of mucin producing columnar 12 months from carcinomatosis. The fourth patient is to cuboidal cells that line the cysts; a layer of undiffer- alive with metastases in the liver and spleen 10 years after entiated mesenchymal cells; and an outer layer of collag- resection (Fig. 5).

DISCUSSION Cystic neoplasms of the liver, both cystadenomas and cystadenocarcinomas are rare, but these cystic neoplasms are being discovered with increasing frequency because of advances in abdominal imaging, particularly ultrasonography, CT, and MRI, and because of an increased awareness of these entities. Less than 200 cases of cystadenomas and a little more than half as many cystadenocarcinomas have been reported in the literature.4,5 It is estimated that cystic neoplasms constitute approximately 5% of liver cysts.1 Most earlier series consist of a few cases and a review of the literature. Large series in- Figure 4. Cystadenocarcinoma. An extended right hepatectomy was clude those of Ishak and colleagues6 (14 cases), Wheeler performed. Patient died of metastatic disease 6 months later. Vol. 200, No. 5, May 2005 Vogt et al Cystadenocarcinoma of the Liver 731

Figure 5. (A) Cystadenocarcinoma. A right hepatectomy was performed. (B) Followup CT scan in same patient as in (A), 8 years later. Extensive metastatic disease in both the liver and spleen. The patient remains alive 10 years after the liver resection with recurrent disease. enous connective tissue. Cystadenomas without mesen- mation concerning the nature of the fluid in the cyst, chymal stroma occur in both men and women.Ten of 18 blood versus mucin. CT scans had been performed in all pathology reports in our series specifically mention our patients, 75% of whom had also undergone ultra- CMS. sonography or MRI, or both. Clinical presentation of cystadenomas can vary Cystadenomas require complete excision. Partial ex- widely. Occasionally, a patient may have the cystade- cision results in persistence or recurrence of the cyst and noma found incidentally, as was the case in two of our associated symptoms.6-8 Although rare, transformation patients. The majority of patients present with upper into a cystadenocarcinoma has been reported.2,3,11 Fifty- pain or discomfort, including 88% (16 of 18) of those in five percent10 of the patients in our series had undergone our article. Unusual presentations, which were not 17 earlier interventions, of which 12 were partial resec- present in any of our patients, include jaundice, cholan- tions. All these patients developed recurrent or persistent gitis, IP rupture, intracystic hemorrhage, compression of cystadenomas and symptoms a mean of 21 months the portal vein or vena cava, or both, which can result in (range 1 month to 4 years) postoperatively (Table 1). ascites formation and peripheral edema, respectively, Proper surgical treatment of cystadenomas begins with a and stone formation.4,8,9,13 The most frequent finding on high index of suspicion based on preoperative imaging physical examination is a palpable upper abdominal that demonstrates a cystic mass with septa. An intraop- mass, which was present in two of our patients. Labora- erative frozen section of a generous biopsy of the cyst tory liver studies are usually normal unless the biliary wall is also important in directing the appropriate surgi- tree is compressed, which results in elevation of the al- cal procedure. We recommend a laparoscopic frozen- kaline phosphatase and perhaps bilirubin. Carbohydrate section biopsy of the cyst wall. If the frozen section is antigen 19-9 may be elevated, but the CEA and the consistent with a simple, benign cyst, laparoscopic par- ␣-fetoprotein are usually normal.4,9 Although none of tial excision is adequate. If the biopsy demonstrates that these tumor markers was measured in our patients, they the mass is a cystadenoma, complete excision is neces- should be considered in patients suspected of having a sary, almost always as an open procedure. One of our cystadenocarcinoma. patients successfully underwent excision of a small, pe- Preoperative imaging studies are of key importance in ripheral cystadenoma laparoscopically. evaluation of a patient with a cystic mass in the liver. Frozen-section biopsies are not always accurate. One Ultrasonography demonstrates anechoic lesions with of our patients had six laparoscopic frozen-section biop- septations and hyperechoic areas that represent cyst wall sies that were read as a simple, benign cyst. The final fibrosis and papillary projections. A cystic mass with report stated that the cystic mass was a cystadenoma and water density attenuation, septa, and mural nodules is this patient was advised to have complete excision at a seen on CT scanning. MRI may provide further infor- second procedure. None of the 13 patients in our series 732 Vogt et al Cystadenocarcinoma of the Liver J Am Coll Surg whose cystadenoma was completely excised has had a grade. Cystadenocarcinomas not associated with an under- recurrence. Both patients in our series who had partial lying CMS have been reported primarily in men, although excision and followup information available required they have been reported in women as well.7,9,10,15 Neither further intervention for recurrence. the man nor the remaining woman in our series had an Excision of cystadenomas can be accomplished by ei- associated CMS. ther a liver resection or enucleation. Reports supporting The only potentially curative treatment for cystade- resection cite the low associated mortality of the proce- nocarcinomas is complete removal, usually by a major dure and the permanent relief of symptoms.13 Most of liver resection with 1-cm margins. Reported survival our resections were performed for cystadenomas in the rates for cystadenocarcinomas range from 25% to 100% left lateral segment, which is a procedure that is not (87% disease free) at 5 years.5-7,11 Cystadenocarcinomas difficult and has very low morbidity and mortality. The arising in women with CMS are felt to be relatively majority of cystadenomas can be completely and safely indolent, but those not associated with CMS, particu- excised by enucleation, including those that are centrally larly in men, have a worse prognosis.7,10,15 Our longest 6,7,8,11,14 located. Once the cyst has been decompressed survivor had a low-grade lesion associated with CMS and the proper plane identified, enucleation can proceed enucleated over 16 years ago. The other woman with without significant blood loss. If major venous vascular both CMS and cystadenocarcinoma had an extensive, structures present the possibility of hemorrhage, enucle- high-grade cancer that resulted in death from distant ation can be completed with either inflow occlusion metastases at 6 months. The man in our series also died (Pringle maneuver) or total vascular exclusion (Fig. 2). from carcinomatosis at about 12 months. The fourth Performing a hepatectomy for a large, central lesion may patient (no CMS) is alive, but with extensive metastases be hazardous because of injury that may occur to vascu- in both the liver and spleen (Fig. 5). lar or biliary, or both, structures as a result of displace- In summary, cystadenomas and cystadenocarcinomas ment by the cyst. are uncommon liver tumors. Cystadenomas occur pre- Cystadenocarcinomas are very rare tumors. An article dominately in women. The majority are symptomatic. in the European Journal of Cancer from 1998 reported 1 Preoperative imaging studies that demonstrate a com- patient and a review of 112 previously published cases.5 plex cystic structure with internal septae should be pre- In contrast to the female predominance in cyst- sumed to be a cystic neoplasm that requires complete adenomas, 38% to 44% of cystadenocarcinomas have excision to prevent recurrence and persistence or malig- occurred in men with an older mean age of 56 to 59 years nant degeneration. Cystadenocarcinomas in men, which versus 41 to 53 years.5,9,10 One of our 4 patients was a are not associated with a CMS, have a worse prognosis, man and mean age was 60 years. Virtually all patients even after complete excision. with cystadenocarcinomas are symptomatic and many have palpable upper abdominal masses. Preoperative imaging studies demonstrate cystic lesions with septa, areas REFERENCES of solid components, and papillary projections along the 3 1. Walt AJ. Cysts and benign tumors of the liver. Surg Clin North wall of the cyst(s). Am 1977;57:449–464. Cystadenocarcinomas can arise from congenital liver 2. Woods GL. Biliary cystadenocarcinoma: case report of hepatic cysts, bile ducts, and CMS.9 In a series of 18 cases of cysta- malignancy origination in benign cystadenoma. Cancer 1981; 47:2936–2940. denocarcinomas reported from the Armed Forces Institute 3. Matsuoka Y, Hayashi K, Yano M. Case report: malignant trans- of Pathology in Washington, DC, one-third arose in female formation of biliary cystadenoma with mesenchymal stroma: patients with CMS.10 In this group of patients, a papillary documentation by CT. Clin Radiol 1997;52:318–321. adenocarcinoma, which invades the underlying stroma, de- 4. Duchini A. Hepatic cystadenomas. eMedicine Journal 2001. 6,7 Available at: http://www.emedicine.com/med/topic492.htm. velops in the epithelial layer. Malignant degeneration of 5. Lauffer JM, Maurer CA, Stoupis C, et al. Biliary cystadenocar- CMS into a cystadenocarcinoma occurs over the course of cinoma of the liver: the need for complete resection. Eur J Can- several years, as reported by Woods,2 and Matsuoka and cer 1998;34:1845–1851. colleagues.3 Two of the women in our series had papillary 6. Ishak KG, Willis GW, Cummins SD, et al. Biliary cystadenoma and cystadenocarcinoma. Report of 14 cases and review of the adenocarcinomas associated with CMS. One was a low- literature. Cancer 1977;38:322–338. grade lesion and the other was more extensive and high 7. Wheeler DA, Edmondson HA. Cystadenoma with mesenchy- Vol. 200, No. 5, May 2005 Vogt et al Cystadenocarcinoma of the Liver 733

mal stroma (CMS) in the liver and bile ducts. Cancer denoma with mesenchymal stroma. Ann Surg 1990;211:18– 1985;56:1434–1445. 27. 8. Lewis WD, Jenkins RL, Rossi RL, et al. Surgical treatment of 12. Siren J, Karkkainen P,Luukkonen P,et al. A case report of biliary biliary cystadenoma. A report of 15 cases. Arch Surg 1988;123: cystadenoma and cystadenocarcinoma. Hepatogastroenterology 563–568. 1998;45:83–89. 9. Coumbari R, Tsui WM. Biliary tumors of the liver. Semin Liver 13. Madariaga JR, Iwatsuki S, Starzl TE, et al. Hepatic resection for Dis 1995;15:402–413. cystic lesions of the liver. Ann Surg 1993;218:610–614. 10. Devaney K, Goodman ZD, Ishak KG. Hepatobiliary cystade- 14. Pinson CW, Munson JL, Rossi RL, et al. Enucleation of intrahe- noma and cystadenocarcinoma. A light microscopic and immu- patic cystadenomas. Surg Gynecol Obstet 1989;168:535–537. nohistochemical study of 70 patients. Am J Pathol 1994;18: 15. Asahara TI, Katayama K, Nakahara H, et al. A case of biliary 1078–1091. cystadenocarcinoma of the liver. Hiroshima J Med Sci 1999;48: 11. Akwar OE, Tucker A, Seigler HF, et al. Hepatobiliary cysta- 45–48. Safety of Carotid Endarterectomy in 2,443 Elderly Patients: Lessons from Nonagenarians—Are We Pushing the Limit? Desarom Teso, MD, Randolph E Edwards, MD, Jared C Frattini, MD, Stanley J Dudrick, MD, FACS, Alan Dardik, MD, PhD, FACS

BACKGROUND: Elderly patients are a rapidly expanding segment of the population. Recent studies suggest that octogenarians have mortality and morbidity after carotid endarterectomy (CEA) similar to that in their younger cohort. Outcomes of CEA performed in nonagenarians have not been commonly reported; this study seeks to determine the safety of CEA in nonagenarians in general practice. STUDY DESIGN: All patients in nonfederal Connecticut hospitals undergoing CEA between 1990 and 2002 were identified using the state discharge database (Chime Inc; www.cthosp.org). RESULTS: A total of 14,679 procedures were performed during the 12 study years. Sixty-four patients were nonagenarians (0.4%). Perioperative mortality was higher among nonagenarians (3.1%) compared with younger patients, including the 2,379 octogenarians (0.6%; p ϭ 0.008, chi-square; odds ratio ϭ 9.1, p ϭ 0.006). No statistically significant difference was noted in perioperative stroke rates between nonagenarians (3.1%) and octogenarians (1.2%; p ϭ 0.35, chi-square; odds ratio 2.3, p ϭ 0.28). Nonagenarians had longer hospital lengths of stay (7.3 days, p Ͻ 0.0001), intensive care unit lengths of stay (1.2 days, p ϭ 0.0013), and greater hospital charges ($17,967 Ϯ $1,907, p Ͻ 0.0001) than younger patients. Nonagenarians underwent operative procedures more frequently in an emergent setting (22%) compared with octogenarians (11%, p Ͻ 0.001) and had a greater percentage of symptomatic presentations (stroke: 14% versus 11%, p ϭ 0.04; transient ischemic attack: 8% versus 5%, p ϭ 0.04, respectively). All perioperative deaths and strokes occurred in symptomatic nonagenarians (15% versus 0%, p ϭ 0.038; 15% versus 0%, p ϭ 0.038; respectively). CONCLUSIONS: Carotid endarterectomy is performed in nonagenarians, as a group, with greater rates of perioperative mortality and morbidity than in younger patients, including octogenarians. But nonagenarians have a greater rate of symptomatic and emergent presentations than younger patients, which may account for their increased mortality, morbidity, length of stay, and incurred charges. Asymptomatic nonagenarians have similar outcomes after carotid endarterectomy compared with younger patients, including octogenarians, with low rates of mortality and morbidity. (J Am Coll Surg 2005;200:734–741. © 2005 by the American College of Surgeons)

Carotid endarterectomy (CEA) is the current gold standard for treatment of both symptomatic and Competing interests declared: None. asymptomatic carotid artery disease.1-2 But because of Supported by the Bell Fund, St Mary’s Hospital, Waterbury, CT, and The Dennis W Jahnigen Career Development Scholarship program administered the perceived risks of the operation, the presumed by the American Geriatrics Society through an initiative funded by The John short life span of elderly patients, and the presumed A Hartford Foundation of New York City and The Atlantic Philanthropies similar stroke rate of elderly and younger patients, (AD). Presented at the 32nd annual meeting of the Society for Clinical Vascular patients greater than 80 years of age were excluded Surgery, Rancho Mirage, CA, March 2004. from these landmark studies. Octogenarians are the Received November 6, 2004; Revised December 11, 2004; Accepted Decem- fastest growing segment of the population in the ber 15, 2004. United States and are likely to account for a substan- From the Departments of Surgery, St Mary’s Hospital, Waterbury, CT (Teso, Edwards, Dudrick, Dardik) and Yale University School of Medicine, New tial caseload of vascular surgical procedures in the Haven, CT (Frattini, Dudrick, Dardik). next decade.3 Recent literature suggests that CEA can Correspondence address: Alan Dardik, MD, PhD, Yale University School of Medicine, Boyer Center for Molecular Medicine, 295 Congress Ave, Room be performed safely and effectively in patients over 436, New Haven, CT 06519. the age of 80. Studies also suggest that many of these

© 2005 by the American College of Surgeons ISSN 1072-7515/05/$30.00 Published by Elsevier Inc. 734 doi:10.1016/j.jamcollsurg.2004.12.012 Vol. 200, No. 5, May 2005 Teso et al Carotid Endarterectomy in Elderly Patients 735

416.8 to 416.9, 491 to 494, or 496), renal disease (di- Abbreviations and Acronyms agnostic codes 582 to 583, 585 to 588, V42.0, V45.1, CEA ϭ carotid endarterectomy V56), history of stroke (diagnostic codes 342 or 438), ϭ HTN hypertension transient ischemic attack (TIA; diagnostic codes 435, LOS ϭ length of stay TIA ϭ transient ischemic attack 437.1, or 781.4), and amaurosis fugax (diagnostic codes 362.34 or 368.12).14,15 Octogenarians are patients 80 to 89 years old and nonagenarians are patients 90 years old and greater. patients continue their lives symptom free both short- and Outcomes studied included in-hospital mortality, longterm after CEA.4-11 The safety of CEA in the more perioperative stroke (diagnostic code 997.0: nervous sys- elderly, ie, nonagenarians, has not been well defined. tem complications, anoxic brain damage, cerebral hyp- The Connecticut state discharge database was studied oxia, iatrogenic cerebrovascular infarction or hemor- to evaluate contemporary outcomes after CEA; in par- rhage), perioperative cardiac complications (diagnostic ticular, factors associated with age, which could poten- code 997.1: cardiac arrest, insufficiency, cardiorespiratory tially influence the rates of perioperative mortality and failure, heart failure, and arrhythmias during or resulting stroke, were examined in nonagenarians in comparison to from a procedure), perioperative bleeding complications younger octogenarians. such as hematoma or hemorrhage (diagnostic code 998.1x: hemorrhage, hematoma, or seroma complicat- METHODS ing a procedure), hospital and ICU length of stay (LOS), A database consisting of patient discharge records from and total hospital charges. Hospitals were stratified as all acute-care, nonfederal Connecticut hospitals is main- low-, medium-, or high-volume institutions, defined as tained by Chime, Inc (www.cthosp.org). The Connecticut performing Յ 10, 10 to 49, or Ն 50 procedures annu- Hospital Association Chime Data Program has estab- ally, as previously described.12,13 lished and maintains a proprietary health-care informa- Results are reported as mean Ϯ SEM. Categoric vari- tion system that incorporates statewide clinical, finan- ables were analyzed by Pearson’s chi-square or Fisher’s cial, and patient demographic data dating back to 1980. exact test. Continuous variables were analyzed by the Reports containing selected variables are available on a Mann-Whitney U test or Kruskal-Wallis test. The ef- fee-for-service basis. A previously published algorithm fects of patient-associated risk factors on outcomes were was used to select all CEA procedures from the Chime analyzed by multivariable logistic regression. All tests database.12,13 In brief, discharge records were selected if were two-tailed and p values Յ 0.05 were considered they contained the DRG 5 (extracranial vascular proce- statistically significant (Statview 5.0; SAS Institute). dures), the ICD-9-CM principal procedure code of 38.12 (head/neck endarterectomy), and the ICD-9-CM RESULTS principal diagnosis code of 433.xx (occlusion and steno- All CEA procedures performed in nonfederal hospitals sis of precerebral arteries). Records noted to be urgent or in the state of Connecticut between 1991 and 2002 were emergent CEA cases were included, as were elective identified. During this period, 14,679 procedures were cases. The accuracy of this algorithm in the Chime da- performed, of which 64 were in nonagenarians (0.4%), tabase was reviewed by comparing data produced by the and 2,379 were in octogenarians (16%). Compared with algorithm with the medical records of all CEA proce- octogenarians, nonagenarians had a higher rate of symp- dures performed at St Mary’s Hospital between 1991 tomatic presentation, with both a higher previous stroke and 2001; the algorithm correctly identified all proce- rate (14% versus 11%, p ϭ 0.04) and TIA rate (8% dures performed at our institution.13 versus 5%, p ϭ 0.04). But nonagenarians had lower Demographic variables for each patient were ana- incidences of diabetes (p Ͻ 0.0001) and cardiac disease lyzed, including age, gender, race, hypertension (HTN; (p ϭ 0.01) compared with their younger cohort (Table 1). diagnostic codes 401 to 405), diabetes (diagnostic code Nonagenarians also had a higher rate of symptomatic 250), cardiac disease (diagnostic codes 391, 394 to 398, and emergent presentation compared with octogenari- 402, 404, 411 to 414, 416, or 425), chronic obstructive ans (11% versus 8%, p Ͻ 0.001, Table 2 ). Nonagenar- pulmonary disease (COPD; diagnostic codes 415.0, ians were admitted to low-volume hospitals at a similar 736 Teso et al Carotid Endarterectomy in Elderly Patients J Am Coll Surg

Table 1. Demographic and Risk Factors Age, y <70 70-79 80-89 >90 Total p Value* n 5,825 (40) 6,411 (44) 2,379 (16) 64 (0.4) 14,679 (100) Male 3,572 (61) 3,702 (58) 1,289 (54) 36 (56) 8,599 (59) Ͻ0.0001 Caucasian 5,524 (95) 6,179 (96) 2,302 (97) 62 (97) 14,067 (96) Ͻ0.0001 HTN 3,732 (64) 4,357 (68) 1,709 (71) 46 (72) 9,844 (67) Ͻ0.0001 Diabetes 1,548 (27) 1,625 (25) 444 (19) 5 (8) 3,622 (25) Ͻ0.0001 Cardiac disease 975 (17) 1,219 (19) 421 (18) 11 (17) 2,626 (18) 0.01 COPD 636 (11) 841 (13) 273 (11) 9 (14) 1,759 (12) 0.002 Renal disease 66 (1) 65 (1) 23 (1) 0 (0) 154 (1) 0.73 Prior stroke 619 (11) 596 (9) 252 (11) 9 (14) 1,476 (10) 0.04 TIA 232 (4) 227 (4) 112 (5) 5 (8) 576 (4) 0.03 Amaurosis fugax 240 (4) 216 (3) 64 (3) 1 (2) 521 (4) 0.007 Numbers in parentheses are percentages. *Chi-square analysis. HTN, hypertension; TIA, transient ischemic attack. rate as younger patients, but were admitted to high- the analysis. Age greater than 90 years was a significant volume hospitals less frequently (Table 2). independent risk factor for perioperative mortality (odds There were 71 (0.5%) perioperative deaths. Nonage- ratio 9.1, p ϭ 0.006). Significantly higher mortality was narians had a perioperative death rate of 3.1%, the high- also seen in patients with either previous stroke (odds est among the age deciles, including the 0.6% death rate ratio 10.9, p Ͻ 0.0001), renal disease (odds ratio 7.5, p of the octogenarians (p ϭ 0.008, Table 3). Although Ͻ 0.0001), or diabetes mellitus (odds ratio 1.8, nonagenarians had a higher perioperative stroke rate p ϭ 0.02). Patients with hypertension incurred a lower than octogenarians (3.1% versus 1.2%), it was not sta- risk of perioperative mortality (Table 4). Age was not an tistically significant (p ϭ 0.35). Nonagenarians had independent risk factor for perioperative stroke com- fewer perioperative cardiac complications than their pared with earlier stroke, TIA, diabetes, and female gen- younger cohort (1.6% versus 3.0%, p ϭ 0.001), al- der (Table 5). though there was no difference between octogenarians and nonagenarians in subgroup analysis (p ϭ 0.51, chi- Effect of symptomatic presentation square). Nonagenarians had higher mean hospital and Although nonagenarians have a higher mean periopera- ICU lengths of stay (7.3 versus 3.7 days, p Ͻ 0.0001, tive mortality rate than octogenarians (Table 3), nona- and 1.2 versus 0.8 days, p ϭ 0.15, respectively), with genarians have higher rates of symptomatic presentation resultant increased mean hospital charges ($17,967 ver- (Table 1). The effect of symptomatic presentation on sus $12,700, p Ͻ 0.0001, Table 3). outcomes in nonagenarians was examined. All perioper- Multivariable logistic regression was used to deter- ative deaths, strokes, and cardiac complications occurred in mine the relative effect of risk factors on outcomes; all nonagenarians who presented symptomatically; whereas patient-associated variables examined were included in perioperative bleeding complications occurred with sim-

Table 2. Admission and Hospital Type Admission, n Age Elective Urgent Emergent Hospital volume, % group, y n % n % n % Low Medium High Ͻ70 4,357 85 262 5 509 10 3 44 53 70–79 5,008 87 269 5 489 8 4 45 51 80–89 1,903 84 102 5 254 11 3 46 51 Ն90 44 75 2 3 13 22 5 61 34 p Value 0.0002* 0.04* *Chi-square analysis. Vol. 200, No. 5, May 2005 Teso et al Carotid Endarterectomy in Elderly Patients 737

Table 3. Perioperative Outcomes Cardiac Bleeding, Age Died Stroke complications hematoma Hospital ICU LOS, group, y n%n% n %n%LOS, d‡ d‡ Charge, $ Ͻ70 21 0.4 67 1.1 112 1.9 159 2.7 3.2 Ϯ 0.1 0.7 Ϯ 0.02 11,137 Ϯ 98 70–79 34 0.5 89 1.3 172 2.7 239 3.7 3.5 Ϯ 0.1 0.8 Ϯ 0.02 11,975 Ϯ 128 80–89 14 0.6 28 1.2 71 3.0 122 5.1 3.7 Ϯ 0.1 0.8 Ϯ 0.03 12,700 Ϯ 205 Ն90 2 3.1 2 3.1 1 1.6 2 3.1 7.3 Ϯ 1.3 1.2 Ϯ 0.2 17,967 Ϯ 1,907 Total 71 0.5 186 1.2 356 2.4 522 3.6 3.4 Ϯ 0.04 0.8 Ϯ 0.01 11,786 Ϯ 76 p Value 0.008* 0.35* 0.001* Ͻ0.0001* Ͻ0.0001† 0.15† Ͻ0.0001† *Chi-square analysis. †Kruskal-Wallis. ‡Results are Ϯ SEM. LOS, length of stay. ilar frequency in symptomatic and asymptomatic pa- DISCUSSION tients (Table 6). The differences in perioperative mortal- Using a statewide discharge database, we have dem- ity and stroke between symptomatic and asymptomatic onstrated that CEA is performed safely in the state of nonagenarians were significant (Table 6). Of the two Connecticut, with overall low mortality and morbid- nonagenarians who died perioperatively, only one was ity. Octogenarians have perioperative mortality and associated with a perioperative stroke. Within the sub- stroke rates similar to their younger cohort. But nona- group of asymptomatic patients, multivariable analysis genarians have the highest morbidity and mortality rates of factors determining outcomes revealed that asymp- among all the age groups, with a combined mortality tomatic nonagenarians were not at increased risk of ei- and stroke rate of 6.2%, more than 3 times the com- ther perioperative mortality (p ϭ 0.9841) or stroke bined death and stroke rate of octogenarians. Nonage- (p ϭ 0.9810) compared with younger patients (data not narians also had a higher rate of symptomatic presenta- shown). tion and are admitted in more emergent and urgent Although nonagenarians had higher mean hospital settings than octogenarians. This higher rate of symp- and ICU LOS and hospital charges than younger pa- tomatic presentation could account for the higher mor- tients (Table 3), the increased LOS and charges were bidity and mortality in nonagenarians compared with incurred mostly by symptomatic nonagenarians (Fig. 1). younger patients, including octogenarians.

Table 4. Multivariable Analysis of Factors Affecting Periop- Table 5. Multivariable Analysis of Factors Affecting Periop- erative Mortality erative Stroke Odds 95% conﬁdence Odds 95% conﬁdence Factors ratio interval p Value Factors ratio interval p Value Race (Caucasian) 1.124 0.40–3.168 0.824 Race (Caucasian) 0.788 0.38–1.638 0.524 Gender (female) 0.986 0.607–1.603 0.960 Gender (female) 0.610 0.453–0.820 0.0011 Hypertension (no) 0.504 0.312–0.815 0.005 Hypertension (no) 0.814 0.594–1.114 0.199 Diabetes (no) 1.773 1.077–2.919 0.024 Diabetes (no) 1.425 1.038–1.956 0.03 Cardiac disease (no) 0.978 0.530–1.804 0.944 Cardiac disease (no) 0.862 0.577–1.286 0.466 Renal disease (no) 7.54 2.836–20.067 Ͻ0.0001 Renal disease (no) 1.588 0.486–5.187 0.444 COPD (no) 1.569 0.847–2.909 0.152 COPD (no) 1.343 0.896–2.013 0.153 Age, y Age, y Ͻ70 — — — Ͻ70 — — — 70–79 1.736 0.997–3.023 0.06 70–79 1.301 0.938–1.804 0.114 80–89 1.837 0.918–3.674 0.09 80–89 1.029 0.653–1.619 0.903 Ն90 9.139 1.892–44.143 0.006 Ն90 2.305 0.514–10.341 0.276 Prior stroke (no) 10.85 6.726–17.500 Ͻ0.0001 Prior stroke (no) 11.217 8.335–15.095 Ͻ0.0001 Transient ischemic Transient ischemic attack (no) 2.235 0.928–5.379 0.073 attack (no) 2.483 1.448–4.260 0.001 Amaurosis (no) 2.181 0.777–6.123 0.140 Amaurosis (no) 1.118 0.486–2.572 0.793 738 Teso et al Carotid Endarterectomy in Elderly Patients J Am Coll Surg

Table 6. Outcomes in Symptomatic and Asymptomatic Non- agenarians Cardiac Bleeding, n Death Stroke complication hematoma Symptomatic 13 2 2 1 1 Asymptomatic 51 0 0 0 1 Total 64 2 2 1 2 p Value* 0.038 0.038 0.99 0.37 *Fisher’s exact test.

The low perioperative mortality and morbidity of CEA in octogenarians are consistent with numerous other reports (0.0% to 2.4%).4-11 This analysis from the Connecticut state database is similar to other reports using administrative data.12 In particular, reports from the Maryland database documented a combined mortality and stroke rate of 2.6%, which is similar to the rate for octogenarians in Connecticut. These results confirm the safety of CEA in elderly patients in their ninth decade. Of additional importance, these figures are below the threshold for a positive benefit-to-risk ratio as determined by the Asymptomatic Carotid Atherosclerosis Study (ACAS),1 confirming the safety of CEA in octogenarians in contemporary practice. But we found that nonagenarians have higher perioperative mortality and stroke rates compared with younger patients. Most other reports included nonagenarians within the elderly group; for example, in the previously reported Maryland database, 20 nonagenarians were included in the 1,036 patients above 80 years old.12 This suggests that octogenarians may have results of CEA truly comparable with younger patients. On the other hand, the combined mortality and stroke rate of nonagenarians suggests that the appropriateness of performing elective CEA in patients in the tenth decade of life may not be a straightforward decision. The higher perioperative mortality rate of nonagenarians (Tables 3 and 4) might be secondary to their increased rate of symptomatic presentation before operation, because all perioperative mortality, strokes, and cardiac complications occurred in symptomatic nonagenarians (Table6). A history of symptomatic presentation with stroke has been shown to increase mortality and morbidity in most studies.13,16-19 For example, in a metaanalysis of 14,399 CEA, Bond and colleagues16 reported a Figure 1. Effect of symptoms on economic outcomes. (A) Effect of symptoms on hospital length of stay, by age group. (B) Effect of combined death and stroke risk approximately twice as symptoms on intensive care unit length of stay, by age group. (C) high in symptomatic patients (absolute risk 5.1%; 95% CI, Effect of symptoms on mean hospital charges, by age group. 4.6% to 5.6%) compared with asymptomatic patients Vol. 200, No. 5, May 2005 Teso et al Carotid Endarterectomy in Elderly Patients 739

(absolute risk 2.8%; 95% CI, 2.4% to 3.2%); CEA per- genarians in the state of Connecticut are admitted to formed urgently was associated with a combined death low-volume hospitals at a similar rate as younger pa- and stroke risk of 19.2% (95% CI, 10.7% to 27.8%). tients, but are admitted less frequently to high- Although Eckstein and colleagues20 noted that CEA volume hospitals (Table 2). Because elective admis- could be performed safely in an emergent situation with sion to high-volume hospitals is associated with less appropriate patient selection, their study notably did not mortality and stroke in the state of Connecticut,13 include nonagenarians. Our data suggest that asymp- increased referral of nonagenarians to high-volume tomatic nonagenarians have a low rate of perioperative hospitals might improve procedural safety and out- mortality and morbidity, although our small sample size comes, especially for symptomatic patients. Because may underestimate these exact risks. the Connecticut database does not report operating Our study suggests that CEA is performed safely in surgeon volume, we could not examine the effect of octogenarians, because reasonable results from the pro- surgeon volume on outcomes. But because high hos- cedure have been achieved in these patients, with low pital volume, high operative surgeon volume, and vas- morbidity and mortality, and elderly patients with ca- cular surgical specialization are all associated with im- 21,22 rotid stenosis are at high risk for stroke. Nevertheless, proved outcomes but reflect underlying processes of it is not apparent if the higher mortality and morbidity care, it is still imperative to examine the underlying of the procedure in nonagenarians as a group truly rep- factors in performance of CEA that will further im- resent relative contraindications in asymptomatic pa- prove its safety for nonagenarians and for all tients without critical stenosis, given that all periopera- patients.29 tive mortality and strokes in nonagenarians occurred in Limitations of this study include the use of retrospec- symptomatic patients (Table 4). Additionally, perioper- tive administrative data and the well-known potential to ative cardiac events occurred in nonagenarians at a rate underreport adverse events. For example, the incidence similar to that in younger patients (Table 3). of cardiac disease reported in this database may be lower So we believe that healthy elderly patients, including than expected for this population (Table 1). In addition, nonagenarians, who are being considered as candidates it is possible that some patients, especially those admit- for CEA, and are thought to be sufficiently healthy to ted urgently with a TIA or stroke, may not be included derive benefit from prevention of stroke, should not be by the case selection algorithm. But nonagenarians denied the procedure because of their advanced age. Ad- ditional studies that define the exact life expectancy and should have had the same coding and selection biases as risk of stroke in the nonagenarian would provide a more younger patients in this database. In addition, mortality precise calculation of the risk-benefit ratio of CEA. In is a variable that is usually accurately and well reported. addition, the effects of risk factors that correlate signifi- Although the sample size of our subgroup of nonagenar- ϭ cantly with outcomes in younger patients, such as ians is relatively small (n 64), and the number of insulin-dependent diabetes, renal disease, or history of complications and statistical power is accordingly small, previous stroke, may have different significance in non- this study is larger than several reported series, including 12 agenarians.23,24 Symptomatic nonagenarians require es- the Maryland database. The small sample size also pre- pecially careful and comprehensive discussions of the cluded subgroup analysis of patients presenting with a risks and benefits of the procedure preoperatively, be- stroke or TIA compared with those presenting with am- cause their risk-benefit ratio of CEA may be significantly aurosis fugax. different from those in asymptomatic patients. For ex- Carotid artery angioplasty and stenting is a newer ample, if future studies define the rate of stroke in non- option for the treatment of carotid artery stenosis. Its agenarians to be much higher than in younger patients, application to elderly patients should be cautious, with then the current outcomes of CEA in nonagenarians early results demonstrating inferior results compared may strongly favor performance of CEA, even with its with those in younger patients. Roubin and associates30 higher risk in symptomatic patients. demonstrated that the best predictor of periprocedural Several studies have related good outcome of CEA stroke and death was age 80 years or older and that to performance of the procedure in high-volume hos- longterm durability of the procedure was also reduced in pitals, by high-volume surgeons, or both.11,25-28 Nona- these patients. The University of Alabama team reported 740 Teso et al Carotid Endarterectomy in Elderly Patients J Am Coll Surg similar results, with a 25% neurologic event rate in pa- tomy in octogenarians: early results and late outcome. J Vasc tients 80 years or older.31 Similar results were recently Surg 1998;27:860–869. 10. Coyle KA, Smith RB 3rd, Salam AA, et al. Carotid endarterec- reported, even with the consistent use of cerebral protec- tomy in the octogenarian. Ann Vasc Surg 1994;8:417–420. tion.32 We believe that our data strongly suggest that 11. Favre JP, Guy JM, Frering V, et al. Carotid surgery in the octo- performance of CEA in octogenarians is safe and that genarian. Ann Vasc Surg 1994;8:421–426. 12. Perler BA, Dardik A, Burleyson GP, et al. Influence of age and this option should not be denied to the patient on the hospital volume on the results of carotid endarterectomy: a state- basis of age alone. We believe that asymptomatic nona- wide analysis of 9918 cases. J Vasc Surg 1998;27:25–33. genarians in relatively good health should also not be 13. Teso D, Edwards RE, Antezana JN, et al. Do vascular surgeons denied CEA on the basis of age. But performance of improve outcome of carotid endarterectomy? An analysis of 12,618 elective cases in the state of Connecticut. Vascular 2004;12:155– CEA for symptomatic nonagenarians requires a compre- 165. hensive preoperative evaluation of the individual’s 14. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new health profile and a carefully discussed informed opera- method of classifying prognostic comorbidity in longitudinal studies: Development and validation. J Chron Dis 1987;40: tive consent. 373–383. 15. Romano PS, Roos LL, Jollis JG. Adapting a clinical comorbidity Author Contributions index for use with ICD-9-CM administrative data: Differing perspectives. J Clin Epidemiol 1993;46:1075–1079. Study conception and design: Teso, Dardik 16. Bond R, Rerkasem K, Rothwell PM. Systematic review of the Acquisition of data: Teso, Dardik risks of carotid endarterectomy in relation to the clinical indica- Analysis and interpretation of data: Teso, Edwards, tion for and timing of surgery. Stroke 2003;34:2290–2301. 17. Kucey DS, Bowyer B, Iron K, et al. Determinants of outcome Frattini, Dudrick, Dardik after carotid endarterectomy. J Vasc Surg 1998;28:1051–1058. Drafting of manuscript: Teso, Dardik 18. Golledge J, Cuming R, Beattie DK, et al. Influence of patient- Critical revision: Teso, Edwards, Frattini, Dudrick, Dardik related variables on the outcome of carotid endarterectomy. J Vasc Surg 1996;24:120–126. Statistical expertise: Dudrick, Dardik 19. Goldstein LB, Moore WS, Robertson JT, Chaturvedi S. Com- Obtaining funding: Dudrick, Dardik plication rates for carotid endarterectomy. A call to action. Supervision: Dudrick, Dardik Stroke 1997;28:889–890. 20. Eckstein HH, Schumacher H, Klemm K, et al. Emergency carotid endarterectomy. Cerebrovasc Dis 1999;9:270–281. 21. Alamowitch S, Eliasziw M, Algra A, et al. Risk, causes and REFERENCES prevention of ischaemic stroke in elderly patients with symp- 1. ACAS collaborators. Endarterectomy for asymptomatic carotid tomatic internal-carotid artery stenosis. Lancet 2001;357: artery stenosis. Executive Committee for the Asymptomatic Ca- 1154–1160. rotid Atherosclerosis Study. JAMA 1995;273:1421–1428. 22. Schultz RD, Sterpetti AV, Feldhaus RJ. Carotid endarterectomy in 2. NASCET collaborators. Beneficial effect of carotid endarterec- octogenarians and nonagenarians. Surg Gynecol Obstet 1988; tomy in symptomatic patients with high-grade carotid stenosis. 166:245–251. North American Symptomatic Carotid Endarterectomy Trial 23. Rockman CB, Cappadona C, Riles TS, et al. Causes of the Collaborators. N Engl J Med 1991;325:445–453. increased stroke rate after carotid endarterectomy in patients 3. Etzioni DA, Liu JH, O’Connell JB, et al. Elderly patients in with previous strokes. Ann Vasc Surg 1997;11:28–34. surgical workloads: a population-based analysis. Am Surg 2003; 24. Axelrod DA, Upchurch GR Jr, DeMonner S, et al. Perioperative 69:961–965. cardiovascular risk stratification of patients with diabetes who 4. Pruner G, Castellano R, Jannello AM, et al. Carotid endarter- undergo elective major vascular surgery. J Vasc Surg 2002;35: ectomy in the octogenarian: outcomes of 345 procedures per- 894–901. formed from 1995–2000. Cardiovasc Surg 2003;11:105–112. 25. Dardik A, Bowman HM, Gordon TA, et al. Impact of race on 5. Ting ACW, Taylor DC, Salvian AJ, et al. Carotid endarterec- the outcome of carotid endarterectomy: a population-based tomy in octogenarians. Cardiovasc Surg 2000;8:441–445. analysis of 9,842 recent elective procedures. Ann Surg 2000; 6. Schneider JR, Droste JS, Schindler N, Golan JF. Carotid endar- 232:704–709. terectomy in octogenarians: comparison with patient character- 26. Pearce WH, Parker MA, Feinglass J, et al. The importance of istics and outcomes in younger patients. J Vasc Surg 2000;31: surgeon volume and training in outcomes for vascular surgical 927–935. procedures. J Vasc Surg 1999;29:768–778. 7. Maxwell JG, Taylor AJ, Maxwell BG, et al. Carotid endarterec- 27. Hannan EL, Popp AJ, Feustel P, et al. Association of surgical tomy in the community hospital in patients age 80 and older. specialty and processes of care with patient outcomes for carotid Ann Surg 2000;231:781–788. endarterectomy. Stroke 2001;32:2890–2897. 8. Hoballah JJ, Nazzal MM, Jacobovicz C, Sharp WJ, et al. Enter- 28. Cebul RD, Snow RJ, Pine R, et al. Indications, outcomes, and ing the ninth decade is not a contraindication for carotid end- provider volumes for carotid endarterectomy. JAMA 1998;279: arterectomy. Angiology 1998;49:275–278. 1282–1287. 9. O’Hara PJ, Hertzer NR, Mascha EJ, et al. Carotid endarterec- 29. Teso D, Frattini JC, Dardik A. Improved outcomes of carotid Vol. 200, No. 5, May 2005 Teso et al Carotid Endarterectomy in Elderly Patients 741

endarterectomy: the critical role of vascular surgeons. Semin upon complications of carotid artery stenting. UAB Vasc Surg 2004;17:214–218. Neurovascular Angioplasty Team.J Endovasc Surg 1999;6:217– 30. Roubin GS, New G, Iyer SS, et al. Immediate and late clinical 222. outcomes of carotid artery stenting in patients with symptom- 32. Kastrup A, Schulz JB, Raygrotzki S, et al. Comparison of angio- atic and asymptomatic carotid artery stenosis: a 5-year prospec- plasty and stenting with cerebral protection versus endarterec- tive analysis. Circulation 2001;103:532–537. tomy for treatment of internal carotid artery stenosis in elderly 31. Chastain HD 2nd, Gomez CR, Iyer S, et al. Influence of age patients. J Vasc Surg 2004;40:945–951. CONSENSUS STATEMENT*

Use of Magnetic Resonance Imaging in Breast Oncology

Magnetic Resonance Imaging (MRI) of the breast has physical examination, mammography, and US in the been used increasingly for the detection and evaluation following settings: of breast cancer since its approval by the FDA 13 years ago. Multiple studies comparing the results of breast 1. Axillary node metastasis from a suspected occult primary MRI with pathologic outcomes show that breast MRI is breast cancer. Breast MRI can aid the treating physician in locating the primary tumor. sensitive (identifying at least 95% of invasive cancers), 2. For determining ipsilateral tumor extent or the presence of but that specificity varies widely (30–90%), with fre- contralateral disease, in patients with a proven breast can- quent false-positive results because of evolving technol- cer (especially those with invasive lobular carcinoma) when ogy and variable interpretations. For this reason breast dense breast tissue precludes an accurate mammographic MRI may help guide the breast evaluation as indicated assessment. below. Breast MRI finding, however, may not be substi- 3. To monitor response to neoadjuvant hormonal therapy or tuted for histologic tissue diagnosis, especially when the chemotherapy. Pre- and post-treatment MRI can help patient and her surgeon are considering breast identify those patients who are candidates for breast con- conservation. servation, and assist in determining the extent of resection Breast MRI requires a high field system, a dedicated required. breast surface coil (breast images taken in a body scanner 4. As part of breast cancer screening for patients at very high are inadequate) and intravenous contrast. Breast MRI risk for developing breast cancer, especially those with suspected or proven deleterious mutations of BRCA 1/2. should be performed by a dedicated team, including 5. For the additional evaluation of suspicious clinical findings radiologists experienced in all three breast imaging mo- or imaging results that remain indeterminate after com- dalities (mammography, ultrasound [US] and MRI), plete mammographic and sonographic evaluations com- and in image-guided biopsy techniques. Focal MRI le- bined with a thorough physical examination. sions confirmed on US are amenable to US-core biopsy, but MRI-guided wire localization or core biopsy should be available for biopsy of lesions found only on MRI. SELECTED READINGS Breast MRI should not replace mammography for 1. Liberman L. Breast cancer screening with MRI—What are the data for patients at high risk? N Engl J Med 2004;351:497–500. yearly screening examination. While no prospective ran- 2. Kriege M, Brekelmans CTM, Boetes C, et al. Efficacy of MRI domized trials have studied the role of breast MRI for and mammography for breast cancer screening in women with a general breast cancer screening, or for the evaluation of familial or genetic predisposition. N Engl J Med 2004;351:427–37. patients with proven breast cancers, the reported clinical 3. Orel SG, Schnall MD. MR imaging of the breast for the detection, diagnosis and staging of breast cancer. Radiology 2001;220:13– experience with breast MRI is growing rapidly. Based on 30. a review of current studies, the American Society of 4. Hata T, Takashi H, Watanabe K, et al. Magnetic Resonance Breast Surgeons supports the addition of breast MRI to Imaging for Preoperative Evaluation of Breast Cancer: A comparative study with mammography and ultrasonography. J Am Coll Surg 2004;198:190–197. 5. Cheung Y, Wan Y, Lo Y. Preoperative magnetic resonance imaging evaluation for breast cancers after sonographically guided core- *This is a Consensus from the American Society of Breast Surgeons needle biopsy: a comparative study. Ann Surg Oncol 2004;11:756– (ASBS) only and does not represent policy or endorsement from any other 761. association, including the American College of Surgeons. The Journal is pleased to present this as information. 6. Warner E, Plewes, D, Hill K, et al. Surveillance of BRCA1 and BRCA2 mutation carriers with magnetic resonance imaging, ultra- Approved, September 21, 2004. sound, mammography, and clinical breast examination. JAMA Board of Directors, The American Society of Breast Surgeons. 2004;292:1317–1325.

What’s New in Cardiac Surgery

David A Fullerton, MD, FACS

“What’s New in Surgery” evolves from the contributions of leaders in each of the fields of surgery. In every instance the author has been designated by the appropriate Council from the American College of Surgeons’ Advisory Councils for the Surgical Specialties. This feature is now presented in issues of the Journal throughout the year.

Thoracic surgical education and 79 repairs, were performed by residents. During the Among the many changes in thoracic surgical education same period, staff surgeons performed 261 mitral valve is the implementation of the Accreditation Council for procedures. These two groups were compared and ana- Graduate Medical Education (ACGME) “Outcomes lyzed for operative mortality and morbidity, including Project.” This is an effort by the ACGME to incorporate reoperation for bleeding, myocardial infarction, infec- six core competencies into medical education. These six tion, stroke, or prolonged mechanical ventilation. The general core competencies include: patient care, medical operative mortality rate was approximately 5% in each knowledge, professionalism, systems-based practice, group for mitral valve replacement and was approxi- practice-based learning and improvement, and interper- mately 4% in each group for mitral valve repair. The sonal and communication skills. All graduate medical morbidity was 30% in the resident group and 35% in education programs must now include education and the attending group. The authors demonstrated that in evaluation of these competencies within a given resi- appropriately selected patients, residents may be the sur- dency program. Higgins and colleagues1 reported their geons of record in academic medical centers for the con- experience using a 360-degree evaluation of residents in duct of mitral valve procedures. the six competency areas. Critics of this tool cite the fact As the job market for cardiothoracic surgeons has that the evaluations are not provided by an individual’s tightened in recent years, many residents graduating peer group l and might not represent an accurate impres- from cardiothoracic surgery residency programs have re- sion of the resident’s performance. Higgins adminis- ported difficulty in securing desirable jobs. Salazar and tered a 360-degree evaluation of the residents in his pro- colleagues3 used the Thoracic Surgery Residency Associ- grams before and after resident exposure to educational ation to survey residents graduating in 2003. The results venues in the six competencies. Data from the 360- of the survey were both enlightening and disturbing. Of degree evaluation indicated that resident performance the estimated 140 graduates, 89 responded. Ninety-one increased in all six areas, and Higgins suggested that this percent of the respondents were men, 80% were mar- 360-degree tool might be used in resident evaluation. ried, and 61% had children. The average age was 36.2 One of the more difficult operations for academic years. Of the 89 respondents, 77 initially sought jobs cardiac surgeons to teach residents is a mitral valve pro- and 12 sought additional training. For residents seeking cedure, so many residents complete their surgical train- jobs, 19.5% received no offers, and 13 of these ulti- ing having experienced only a limited number of valve mately pursued additional training. Seventy-three per- operations. Baskett and colleagues2 reported the experi- cent of those securing jobs involved positions in general ence in their academic medical center of residents doing thoracic surgery; 53% of the secured jobs were in private mitral valve operations. Between 1998 and 2003, 165 practice and 47% were in academic practice. Every year mitral valve operations, which included 86 replacements since 1997, there have been fewer applicants for thoracic surgical residencies from US medical schools than positions available. Received February 2, 2005; Accepted February 3, 2005. From the University of Colorado Health Sciences Center, Denver, CO. Myocardial revascularization Correspondence address: David A Fullerton, MD, Cardiothoracic Surgery, University of Colorado Health Sciences Center, Box C-310, 4200 E Ninth It is clear that the treatment of coronary artery disease is Ave, Denver, CO 80262. changing, and as it changes, percutaneous coronary re-

Early enthusiasm for off-pump coronary bypass sur- Abbreviations and Acronyms gery seems to have waned in recent years. One of the CABG ϭ coronary artery bypass grafting factors contributing to this is the subjective impression ϭ LVAD left ventricular assist device among many surgeons that emergency conversion to MMP ϭ matrix metalloproteinase SVR ϭ surgical ventricular restoration cardiopulmonary bypass during an off-pump coronary TMR ϭ transmyocardial revascularization bypass procedure is associated with worse outcomes. This observation was confirmed by Patel and colleagues6 from Lenox Hill Hospital in New York. Evaluating their vascularization is often seen as a competitive therapy for experience with 1,678 consecutive coronary bypass op- coronary artery bypass grafting. Nonetheless, the groups erations from 1999 through 2002, performed with an of patients treated with these two modalities are often intention to avoid cardiopulmonary bypass, the authors incomparable. Mack and colleagues4 reviewed the data- examined their results among the 50 patients (3%) who base of a large hospital system, HCA, Inc, during the required urgent conversion to cardiopulmonary bypass. years 1999 through the first quarter of 2002. There were It was striking to find that the mortality rate was 1.4% in 148,396 consecutive patients in 69 hospitals who under- the group that did not require conversion, but rose went either percutaneous intervention or coronary ar- steeply to 12% in the group requiring urgent conver- tery bypass grafting (CABG). The review suggested that sion. Other complications that also increased signifi- now 65% of all coronary revascularization is by percu- cantly included the risk of stroke, which was 6% in the taneous intervention, with a 6.8% annual increase. On converted group versus 1% in the control group. Deep the other hand, CABG volume is declining by 1.2% per sternal wound infection was 1.54% in the control group year. Coronary artery bypass grafting continues to be versus 8% in the urgently converted group and respira- used primarily for multivessel disease. The unadjusted tory failure was 3.75% in the control group versus 28% mortality rate for percutaneous intervention was 1.25% in the urgently converted group. The authors indicated and for CABG was 2.63%. As the amount of surgical that they found it impossible to predict which patients treatment for coronary artery disease declines, these data might require urgent conversion to cardiopulmonary must continue to be tracked. bypass and when. Nonetheless, their data highlight the Although off-pump coronary artery bypass grafting significance of urgent conversion. procedures have become incorporated into most surgical To facilitate off-pump coronary bypass operations, a programs, controversy remains about the clinical effi- proximal anastomotic device manufactured by St Jude cacy of this technique. Puskas and colleagues5 reported a Medical, the Symmetry device, was introduced. In re- prospective randomized trial of off- versus on-pump cor- cent years, several anecdotal reports have appeared in the onary bypass surgery, with a focus on graft patency, the literature suggesting that use of this device was associ- cost of the procedure, and the quality of life after the ated with premature graft closure. Reuthebuch and col- procedure. A total of 200 patients were enrolled into leagues7 reported their experience using 77 connectors randomization, 100 in each group. Three patients were in 61 patients between June 2001 and April 2002. The withdrawn after randomization for one reason or an- surgeons in this group believed they were well trained in other, leaving 190 patients for followup at 30 days and the application of the device. Unfortunately, they found 185 patients at 1 year. Coronary angiography was per- its use to be associated with early graft failure. Eight formed before hospital discharge and at 1 year postop- patients (13% of this series) with 12 implanted connec- eratively. The authors found that graft patency was sim- tors became symptomatic between 1 day and 8 months ilar regardless of whether the procedure was performed after operation. Angiography demonstrated that 95% or on or off pump. In addition, there were no differences in greater occlusion of the proximal vein to aorta anasto- the incidence of death, stroke, myocardial infarction, mosis was present at the level of the anastomosis in all angina, or reintervention at 30 days or 1 year. The au- patients. thors concluded that this single surgeon trial suggests Additional experience with the Symmetry proximal that off-pump coronary bypass surgery may provide anastomotic device was reported by Bergsland and col- complete revascularization with durable and cost- leagues.8 In 46 patients undergoing coronary artery by- effective results. pass grafting off pump, 23 of 46 patients had proximal Vol. 200, No. 5, May 2005 Fullerton What’s New in Cardiac Surgery 745 anastomoses performed with the Symmetry connector; thelial cells had grown over the anastomotic region in a 23 patients had a hand-sewn anastomosis. In the oper- single layer. ating room, intraoperative assessment was done with flowmetry and angiography. Angiography was then re- Transmyocardial revascularization peated between 3 and 5 months postoperatively. There Cardiac surgeons are faced with more patients with non- were no differences between the two groups with intra- bypassable coronary arteries undergoing coronary bypass operative assessment. But followup angiography reoperations. Transmyocardial revascularization (TMR) vealed that only 50% of the Symmetry group vein grafts has become a useful adjunct to CABG. Allen and col- were patent, and 90% of the hand-sewn grafts were leagues11 reported results of a prospective randomized patent. An additional 25% of the Symmetry grafts had trial involving 263 patients in 13 centers undergoing a significant stenosis at the connector. The authors con- CABG who could not be completely revascularized by cluded with the recommendation against the use of the CABG alone. Patients were randomized to receive a Symmetry device for coronary bypass operations. CABG plus TMR. Between 1996 and 1998, these 13 Another new proximal anastomotic device is the centers randomized 218 patients who had either CABG Heartstring proximal anastomotic system (Guidant). alone or CABG/TMR. At 5-year followup, the CABG/ Unlike other systems for the creation of proximal con- TMR group had a lower mean angina score, with 0% of duit anastomoses, the Heartstring system basically en- patients in the CABG/TMR with Classes III to IV an- tails creation of a 4.5-mm hole in the ascending aorta gina, and 10% of patients with CABG alone with into which an occluder is introduced that minimizes the Classes III to IV angina. Eighty-eight percent of patients flow of blood through the aortotomy. In this manner, a in the CABG/TMR group were free of angina, compared with 63% in the CABG-alone group. The actuar- hand-sewn proximal anastomosis may be created with ial survival at 6 years was not different between the the aortic occluder in place in a fashion analogous to the CABG/TMR and CABG-alone patients (76% versus way small occluding stents are placed in the coronary 80%, respectively). The results of this randomized mul- artery during creation of a distal anastomosis. Medalion ticentered clinical trial showed that CABG/TMR pro- and colleagues9 reported their experience using this vides superior relief of angina when compared with Heartstring system in 12 patients with very diseased CABG alone in this very difficult group of patients. aortas who underwent off-pump coronary bypass Other studies have likewise shown that TMR as sole operations. therapy is an effective adjunct when compared with As efforts advance to perform coronary bypass graft- medical therapy alone. ing with a minimal access approach, perhaps even with Mühling and colleagues12 used a porcine model of robotics, the need to create innovative technologies for ischemic myocardium to examine whether or not treat- the creation of the distal anastomosis is paramount. Fil- ment with a carbon dioxide laser (Edwards LifeSciences) 10 soufi and colleagues reported a quite novel approach to improves ischemic myocardium. Chronic myocardial creating the distal anastomosis by the use of magnets. A ischemia was created in the circumflex coronary artery small elliptical magnet is placed within the conduit and distribution by placement of a hollow bead in the left the coronary artery and the two are brought into appo- circumflex artery. Animals underwent MRI before the sition with a magnetic vascular positioning system. The induction of chronic myocardial ischemia, and 8 weeks efficacy of this system seemed to be verified in a porcine after treatment with carbon dioxide laser TMR. The model. The magnets measure approximately 3 mm in control group underwent a sham surgical procedure. the minor axis and 7 mm in the major axis, with a The authors found a significant improvement in blood thickness of 0.4 mm. When brought into alignment, the supply assessed by MRI in the region treated with TMR. magnetic forces immediately create the anastomosis, re- This improvement in blood supply was associated with a quiring an average ischemic time of less than 60 seconds. significant improvement in function. These data lend Histologic examination of the animals revealed widely very strong support to the continued clinical use of patent anastomoses with no significant luminal obstruc- TMR with the carbon dioxide laser. tion as late as 6 months postoperatively. Immunohisto- Allen and colleagues13 reported results of a 9-center chemical staining for Factor VIII demonstrated endo- trial preparing 100 patients treated with TMR as sole 746 Fullerton What’s New in Cardiac Surgery J Am Coll Surg therapy or continued medical management in 112 con- Using univariate analysis, factors adversely affecting sur- trol patients. All patients had refractory Class IV angina vival included female gender, cause of heart failure, du- and were not candidates for conventional therapy. ration of LVAD support, and LVAD score. Using mul- Patients were followed for 5 years. The results confirm tivariate analysis, only LVAD score was a significant the clinical efficacy of TMR for treatment of severe predictor to survival for transplantation. ischemic heart disease. The median angina score for Dembitsky and colleagues15 reported data from the TMR patients was 4.0 at baseline and significantly im- REMATCH (randomized evaluation of mechanical as- proved to 1.5 at 1 year and 1.2 at 5 years. Eighty-eight sistance for the treatment of congestive heart failure) percent of patients treated by TMR had at least 2 classes trial. This trial demonstrated that LVAD implantation of improvement in angina compared with only 44% decreased 1-year mortality from 75% to 51% when treated with medical therapy. In addition, there was a compared with medical management. This trial also significant reduction in the risk of late death for TMR demonstrated that LVADs may fail after implantation. patients, with an annual mortality beyond 1 year of 8% Dembitsky’s group reported on the performance of versus 13% in the medically treated group. At 5 years, these devices after implantation. In this study, the 1-year the survival rate of patients treated with TMR was 65%; survival rate was 52% for LVAD patients versus 28% for it was 52% in the medically treated group. medical patients. Two-year survival was 29% for LVAD patients and 13% for medical patients. System failure Heart failure was approximately 0.13% per patient per year. The free- One of the largest experiences with bridging patients to dom from device replacement was 87% at 1 year and heart transplantation with ventricular assist devices was 37% at 2 years. Indications for pump replacement in- reported by the group from Columbia Presbyterian Hos- cluded sepsis, motor failure, inflow valve and compe- pital.14 Between 1990 and January 2003, 243 patients at tence, diaphragm failure, outflow regurgitation, or kink. Columbia underwent implantation of the Thoratec After device replacement, the actuarial 1-year survival HeartMate devices as a bridge to transplantation. Over was 41% and the 2-year actuarial survival was 33%. The this 12-year period, 52 patients (21%) had a pneumatic most common cause of death after device replacement device implanted, 17 patients had a dual-lead vented was sepsis. Acknowledging the observed incidents of electric device implanted, and 174 patients (72%) re- LVAD failures, the authors concluded that LVAD im- ceived a single-lead vented electric device. The pneu- plantation led to clinically significant improvement in matic devices were used early in the program and much survival when compared with medical management. of the learning curve was accomplished with the pneu- Given the growing disparity between the number of matic devices. Successful bridging to transplantation patients with end-stage heart disease and the number of was accomplished in 64% of patients with the pneu- organs available for transplantation, the quest for the matic device and 72% of patients with the single-lead totally implantable artificial heart continues. The first vented electric device. Perhaps most encouraging, the results of the AbioCor (Abiomed) implantable heart overall actuarial survival rates at 1, 3, 5, and 10 years were published in January 2004. Dowling and col- posttransplant were 91%, 85%, 70%, and 40%, respec- leagues16 from around the country reported the out- tively. Over the 12-year study period, 10 devices were comes in the first seven patients to undergo treatment explanted: 7 for infection and 3 because patients dem- with the AbioCor heart. All seven patients were men onstrated left ventricular recovery. The overall incidence with an age range of 51 to 79 years.The first implant was of infection was 17.7%. Seven percent of patients had a performed in July 2001 and the seventh was performed drive line infection, a pocket infection occurred in 6%, in April 2002. At the time the article was submitted for with pump infection occurring in 4%. Neurologic com- publication, five of the seven patients had died, but only plications were reasonably low, with a postoperative one was an intraoperative death (secondary to bleeding). stroke occurring in 7% of patients who had a pneumatic An additional early death was attributable to an aproti- device placed but only 4% of patients with a single-line nin reaction. There have been six significant periopera- vented electric device. The authors had previously pub- tive morbidities, primarily relating to preexisting ill- lished a scoring system to identify risk factors for death nesses. Five patients had prolonged intubation, two after left ventricular assist device (LVAD) implantation. patients had hepatic failure (in one patient the hepatic Vol. 200, No. 5, May 2005 Fullerton What’s New in Cardiac Surgery 747 failure resolved), four had renal failure (in three patients should be addressed at the time of the procedure. Most it resolved), and one patient had recurrent gastrointesti- authors concur with this recommendation and a mitral nal bleeding. There were three late deaths: one caused by valvuloplasty ring is the procedure of choice. The long- multiple organ failure on postoperative day 56, one term outcomes in this group of patients were excellent, caused by a cerebrovascular accident on postoperative with actuarial survival at 1, 5, and 10 years of 92%, 82%, day 142, and one caused by retroperitoneal bleeding that and 62%, respectively.These data confirm the efficacy of resulted in multiple organ failure on postoperative day offering surgical ventricular restoration to patients un- 151. Despite these comorbidities, the device worked dergoing cardiac operation who have had an earlier myo- well, with flows from 4 to 8 L per minute. Central ve- cardial infarction with attenuated myocardium, partic- nous and left atrial pressures were maintained between 5 ularly in the anterior left ventricular wall. and 15 mmHg. The device is powered through transcu- It is now well appreciated that increased left ventric- taneous energy transfer, eliminating the need for percu- ular volume is an independent predictor of diminished taneous lines. At autopsy, it was noteworthy that in four longterm survival. Although it is true that left ventricu- patients, thrombus had formed on struts, which were a lar ejection fraction frequently improves after myocar- component of the atrial cusp device. So the device has dial revascularization with an associated decrease in left been modified to eliminate these struts, which will hope- ventricular volume, it has been less clear whether surgi- fully lead to diminished morbidity with use of the de- cal ventricular restoration (SVR procedure) might opti- vice. The result in these seven patients confirmed that mize improvement in left ventricular volume, improving the AbioCor system has demonstrated excellent func- patient outcomes. Maxey and colleagues18 retrospectively tion of all its device components. The system allowed for examined their outcomes with 95 patients operated on be- excellent patient mobility, two patients being discharged tween 1998 and 2002. All patients had an enlarged left from the hospital. ventricular end diastolic dimension (Ն 6.0 cm). Thirty- The ongoing STITCH (surgical treatment for isch- nine patients underwent coronary artery bypass grafting emic heart failure) Trial is an international NIH-funded alone and 56 patients underwent SVR procedure with trial that includes an arm of therapy in which patients concomitant coronary artery bypass grafting. Both undergo surgical ventricular restoration. The procedure groups demonstrated improved postoperative ejection is a modification of the DOR procedure, with an em- fractions. But patients who underwent SVR plus coro- phasis on appropriate sizing of the left ventricular dia- nary bypass grafting had significantly greater improve- stolic volume and with particular attention paid to re- ment in left ventricular ejection fraction, which rose gaining the elliptoid shape of the left ventricle. from 22% to 33%. Patients undergoing coronary bypass Mickleborough and colleagues17 have one of the largest grafting alone showed an improvement from 26% to experiences with this type of procedure and reported 29%. There were no perioperative deaths, but late mor- their results with 285 patients. One hundred fifty-two of tality was greater in the coronary bypass–alone group at these patients had a preoperative ejection fraction of less 5.1%; it was only 1.8% in the SVR plus coronary bypass than 20%; an additional 116 patients had a preoperative group. left ventricular ejection fraction between 20% and 40%. Despite the high-risk nature of this population, Mickle- Valve surgery borough and colleagues had excellent surgical outcomes. Although risk stratification for coronary artery bypass The operating room mortality was 2.8%. Only 17% of grafting is a recently mature science, it is only in recent this group required an intraaortic balloon pump, and years that risk adjustment for valvular heart surgery has only 54% required inotropic drug support. During a been under investigation. Nowicki and colleagues,19 rep- mean followup of 63 Ϯ 48 months, 8 patients did go on resenting the Northern New England Cardiovascular to require cardiac transplantation, with a mean interval Study Group, examined patient-specific outcomes in of 49 Ϯ 41 months. Two additional patients required aortic and mitral valve operations. Covering the years mitral valve replacement. In regard to their approach to between 1991 and 2001, 8,943 heart valve operations the mitral valve, Mickleborough and colleagues17 recom- were performed. Of these operations, 5,793 were aortic mended that if the preoperative echocardiogram re- valve replacements and 3,152 were cases of mitral valve vealed anything above 2ϩ mitral regurgitation, the valve surgery (repair or replacement). Using multivariate anal- 748 Fullerton What’s New in Cardiac Surgery J Am Coll Surg ysis, 11 variables in the aortic model and 10 variables in origin. Of the 19 early survivors, 2 required subsequent the mitral model were identified as patient-specific risk valve replacements and 1 required heart transplantation factors. In the aortic model, older age, lower body sur- 3 years after valve repair. In four patients who did not face area, earlier cardiac operations, elevated creatinine, undergo reoperation, serial echocardiography revealed earlier stroke, New York Heart Association Class IV, progressive deterioration of the repaired valves. From congestive heart failure, atrial fibrillation, acuity, emer- their experience, the authors concluded that radiation- gency surgery, and concomitant coronary artery bypass associated valve disease is a significant problem and may grafting were all important variables. The 10 variables in exist in up to 60% of patients after mediastinal irradia- the mitral model were identified to be female gender, tion, even if the cause of the valve dysfunction is not older age, diabetes, coronary artery disease, earlier cere- radiation induced. The progressive effects of radiation bral vascular accident, elevated creatinine, New York will continue to act on the repaired valve, leading to Heart Association Class IV,congestive heart failure, acu- recurrent regurgitation. They concluded that valve re- ity, and valve replacement. The authors provided math- placement might be preferable. ematical models that were highly significant predictors Patients with regurgitant bicuspid aortic valves with of surgical outcomes in hospital mortality. This analysis aortic root dilatation are typically young adults. Aicher is particularly useful to the practicing surgeon. Virtually and colleagues22 reported their experience covering the all of these variables may be known to the surgeon pre- time period between 1995 and 2004, consisting of 60 operatively, permitting a reasonably accurate assessment patients who underwent a root remodeling for a bicus- of patient-specific risk. pid aortic valve and compared them with 130 patients The ideal prosthetic valve remains elusive, given the who underwent aortic root remodeling for a tricuspid limited durability of bioprosthetic devices and the need aortic valve. The patients were followed annually by for chronic anticoagulation with a mechanical prosthe- echocardiography. The surgical results were excellent. sis. Kabbani and colleagues20 reported their results using There were no hospital deaths in the bicuspid aortic a pulmonary autograft in the mitral position. In this valve group, and the operative mortality was 5% in the series of 80 patients operated on between 1997 and tricuspid aortic valve group. No patients developed sig- 2003, a pulmonary autograft was placed. nificant aortic stenosis in longitudinal followup; the typ- The surgical technique required first securing the pul- ical gradient across the aortic valve was a mean approx- monary autograft with a rigid Dacron (DuPont) tubing, imately 3 to 5 mmHg in both groups. Freedom from which was then sewn to the mitral annulus. The opera- aortic regurgitation of a grade greater than 2 was 96% in tive mortality rate was 5% and the late mortality rate, the bicuspid group after 5 years and 83% in the tricuspid which was clearly related to the procedure, was 6.25%. group. Likewise, the freedom from reoperation in 5 One patient developed noncritical mitral stenosis and a years was 98% in the bicuspid group and 98% in the second patient developed noncritical mitral stenosis and tricuspid group. The authors concluded that a valve- insufficiency at approximately 5 years postoperatively. sparing aortic root replacement with root remodeling Four additional patients had progression of mitral regur- can be safely applied to the regurgitant bicuspid aortic gitation from mild to moderate over a period of 8 valve. This group found no significant difference be- months to 3 years. Eighty-three percent of surviving tween hemodynamic function or valve stability of the patients remain in Class I status. repaired bicuspid valve or a tricuspid valve. As more and more patients with earlier mediastinal The appropriate treatment strategy for the patient irradiation come to have heart surgery, an important with functional ischemic mitral regurgitation remains clinical question concerns the appropriate management controversial. Some surgeons advocate aggressive mitral of mitral and tricuspid valve regurgitation in such pa- valve annuloplasty at the time of concomitant coronary tients. Crestanello and colleagues21 reported their series bypass operation; other surgeons advocate a strategy of of 22 patients who underwent mitral, tricuspid, or both, leaving the mitral valve alone in anticipation that the valve repairs. The results suggest that the valve repair mitral regurgitation will improve after revascularization. after mediastinal irradiation has limited durability.Their The strategy of the former position is based on the assump- early mortality included three deaths, and there were tion that an annuloplasty ring will eliminate the mitral re- seven late deaths, four of which were cardiovascular in gurgitation. McGee and colleagues23 reported a large series Vol. 200, No. 5, May 2005 Fullerton What’s New in Cardiac Surgery 749 of patients who underwent ring mitral valvuloplasty alone tor may offer an enticing therapeutic strategy for spinal for treatment of functional ischemic mitral regurgitation. cord protection. Using a swine model of spinal cord

The results suggest that a large percentage of patients ischemia, pretreatment with this adenosine A2A receptor develop recurrent mitral regurgitation soon after opera- activator (ATL-146e) before 30 minutes of ischemia fol- tion. Between 1985 and 2002, 585 patients underwent lowed by 3 hours of reperfusion provided a significant mitral annuloplasty alone for functional ischemic mitral reduction in spinal cord neutrophil accumulation (my- regurgitation. Ninety-five percent of these patients eloperoxidase activity) spinal cord tumor necrosis factor had concomitant coronary revascularization. Within 6 level (TNF-alpha) and significantly improved hind limb months after the repair, the proportion of patients with function as assessed by the Tarlov scale during the first 0or1ϩ mitral regurgitation fell from 71% to 41%. At 48 hours after reperfusion. Cellular level, neuronal viabil- the same time, the proportion with 3ϩ or 4ϩ regurgi- ity in the treatment group was significantly improved after tation increased from 13% to 28%. The results of this reperfusion, as was microtubule-associated protein-2 ex- study confirmed that annuloplasty alone does not guar- pression in the gray matter of the spinal cord. The study antee durable elimination of ischemic mitral regurgita- obviously needs clinical application before widespread tion. In an effort to identify factors associated with re- use, but it offers a very encouraging mechanistic insight current mitral regurgitation, the authors suggested that into spinal cord protection associated with aortic the presence of severe preoperative mitral regurgitation, surgery. particularly with a central jet, might indicate restriction In some centers, endovascular approaches to descend- of both anterior and posterior leaflets of the valve, sug- ing thoracic aortic aneurysms are becoming more com- gesting that future restoration of appropriate leaflet co- mon. The group at Stanford University reported the aptation may be less likely. feasibility of endovascular repair of descending thoracic aortic aneurysms in 1994. In 1998, they reported the Aortic surgery results of their initial 5-year experience with the first Many surgeons recommend routine use of left heart by- generation of these grafts in 103 consecutive patients. pass for repair of descending thoracic aortic aneurysms Demers and associates26 reported in 2004 the midterm with an aim toward reducing the incidence of perioper- results of this same patient cohort. Among these 103 ative paraplegia. Coselli and colleagues24 retrospectively patients, 62 patients (60%) were not believed to be can- reviewed 387 consecutive patients undergoing surgical didates for conventional open surgical repair. Overall repair of a descending thoracic aortic aneurysm over a survival rates at 1, 5, and 8 years were 82%, 49%, and 15-year period of time. Three hundred forty-one of 27%, respectively. Survival rates among patients who these patients had operations with the “clamp-sew” tech- would have been considered candidates for open surgical nique and 46 patients (12%) had left heart bypass. The repair were 93% and 78% at 1 and 5 years, respectively, authors reported that using multivariate analysis and but only 74% and 31%, respectively, in patients deemed propensity score analysis, patients operated on with left unsuitable for a conventional open repair. The actuarial heart bypass had a similar incidence of paraplegia (4%) freedom from aortic reintervention and treatment fail- compared with those operated on without left heart by- ure at 8 years were 70% and 39%, respectively. Paraple- pass (2.3%). The propensity score analysis suggested gia and cerebrovascular accident occurred in 3% and 7% that left heart bypass was not associated with reduced of patients, respectively. Surprisingly, perioperative mor- risk of paraplegia. Acknowledging that this analysis was tality rate was only 9%. One noteworthy finding of this a retrospective review over a 15-year period of time, the study was the relatively high incidence of late ruptures of authors concluded that the clamp-sew technique is an the treated aortic segment. This occurred only in pa- appropriate surgical strategy for treatment of descending tients with a documented endoleak. The estimated free- thoracic aortic aneurysms. dom from late aortic rupture was 91% at 5 years and Approaches to minimize this risk of paraplegia have 80% at 8 years. This should be contrasted with a rate of been based largely on rapid surgical technique, left heart virtually 0 among patients with conventional surgical bypass, and pharmacologic manipulation. Reece and repair. It should be noted that the data from this study colleagues25 provided evidence in a laboratory animal represent the first generation of thoracic aortic en- study suggesting that activation of adenosine A2A recep- dografts. The current endografts have significant modi- 750 Fullerton What’s New in Cardiac Surgery J Am Coll Surg fications in the technology of deployment, which should graft sewn onto the subclavian axillary artery. In these optimize the durability of the result. patients, the flow rate was approximately 10 mm/kg/ Iannelli and colleagues27 reported their experience minute to maintain a pressure between 40 and 60 with 22 patients with acute thoracic aortic emergency mmHg. treated with endovascular stent technology. Included In patients with chronic dissection or aneurysmal dis- were three with traumatic rupture, four with contained ease of the ascending, transverse proximal descending or free rupture of the thoracic aortic aneurysm, and eight thoracic aorta, a common strategy is included with a with an acute type B dissection that appeared to be two-stage approach. During the first stage, a median evolving to rupture. Fifteen of these patients were sternotomy is performed and an elephant trunk recon- treated with an endovascular stent. In the group treated struction of the ascending and transverse arch is per- with endovascular techniques, there were no preopera- formed using deep hypothermia and circulatory arrest. tive deaths and there were no open conversions. The Thereafter, a second surgical approach is required to average stay in the ICU was just under 2 days. One track the proximal descending thoracic aorta. Unfortu- 84-year-old patient with a thoracic aortic aneurysm died nately, this strategy is associated with a significant cumu- 78 days after endovascular stent placement of pneumo- lative mortality exceeding 20%, it requires two major nia. There was only one endoleak, in a patient who had procedures, and the morbidity associated with recovery a traumatic aortic tear at 3 months after the procedure. is that for two major procedures. Kouchoukos and col- This experience strongly suggests that endovascular leagues29 reported their experience with a single stage stents might play a very important role in the treatment approach to this disease process. Using an approach with of thoracic aortic emergencies. a bilateral anterior thoracotomy and a transverse sternot- Repair of aortic pathology requiring replacement of omy incision, they reported their experience with 46 the transverse aortic arch typically uses hypothermic cir- patients undergoing a single stage repair of extensive culatory arrest. Additionally, in patients with pathology thoracic aortic disease including replacement of the that extends into the descending thoracic aorta, a transverse aortic arch. In essence, their technique used planned second procedure may be necessary. To this right axillary artery cannulation for arterial inflow. After point, an elephant trunk procedure has been the usual creation of bilateral anterior thoracotomies through the surgical approach. Svensson and colleagues28 reported fourth intercostal space and a transverse sternotomy, car- their experience with 94 consecutive patients undergo- diopulmonary bypass was initiated and the aorta dis- ing an elephant trunk procedure between 1990 and sected out. In the initial 35 patients in this experience, 2003. The aortic dissection was present in 39% of pa- once the patient had been cooled to a temperature of 16° tients and Marfan’s syndrome was present in 7%; 24% to 18°C, a brief period of deep hypothermia and circu- of the procedures were reoperations. Concomitant pro- latory arrest was initiated. A cuff of aorta surrounding cedures were performed in many of these patients, with the brachiocephalic was sewn to the graft, which was coronary bypass grafting performed in 38% and aortic cannulated, and antegrade cerebral perfusion was pro- valve replacement in 59%. The 30-day mortality rate vided. Distal and proximal anastomoses were then con- was 2.1%, with an incidence of permanent stroke at structed. The technique was revised during the second 5.3%. Acknowledging that this operation is designed for part of this experience. For the last 11 patients, the axil- a second stage, 11 patients died before the second stage lary artery was used for arterial inflow. During a very procedure could be performed. Fifty-seven percent un- brief period of deep hypothermia and circulatory arrest, derwent successful second stage procedures, with 40 by a the three brachiocephalic arteries were amputated from left thoracotomy and 7 by stent graft insertion. Sixty- the aorta. Using a branched graft, individual anastomo- eight percent of patients had retrograde cerebral perfu- ses were performed to the left subclavian, left carotid, sion. The authors indicated that the perfusion pressure and the innominate artery. The flow was then reestab- in these patients was approximately 25 mmHg, with a lished through to the right axillary artery and after evac- flow rate of about 500 mL/minute. The average retro- uation of air from the three brachiocephalic vessels, they grade brain perfusion time was 37 minutes. Sixteen per- were clamped several centimeters from their origin, and cent of patients had antegrade brain perfusion per- perfusion of the brain through the right vertebral and formed through the right subclavian artery with a side carotid arteries was initiated. The graft was deaired and Vol. 200, No. 5, May 2005 Fullerton What’s New in Cardiac Surgery 751 then clamped proximally and distally, providing ante- diseases of the aortic root. Karck and his group31 re- grade cerebral flow while the proximal and distal anas- ported excellent use of aortic reimplantation techniques tomoses of the graft to the aorta were constructed. Flow in patients with Marfan’s syndrome. In this study, span- rate, which was initiated between 10 and 15 ml/kg/ ning the years 1979 to 2002, 119 patients with Marfan’s minute was adjusted to maintain a mixed venous oxygen syndrome underwent either composite graft replace- saturation measured in the right atrium of at least 80%. ment with a mechanical valve conduit (n ϭ 74) or aortic Additional cerebral oxygen saturation monitoring was valve–sparing reimplantation (n ϭ 45). Patients under- accomplished in six patients using a cerebral Oximetrix going aortic valve–sparing procedures typically were catheter. Hospital mortality was 6.5% (3 patients). The somewhat younger (28 versus 35 years) and required authors confirmed the efficacy of a single stage approach longer operations with longer aortic cross-clamp times. to extensive thoracic aortic disease. But the early operative mortality rate was 0% in the Bicuspid aortic valve is recognized to be frequently group undergoing aortic valve–sparing procedures com- associated with aneurysmal dilatation of the proximal pared with 6.8% in the valve-conduit group. Freedom ascending aorta. Boyum and colleagues30 examined ma- from reoperation at 5 years was 92% in the valve- trix metalloproteinase (MMP) activity in patients with conduit group and 84% in the valve-sparing group (p ϭ bicuspid and tricuspid aortic valves. Matrix metallopro- 0.31). Thromboembolic complications or postoperative teinases are endopeptidases synthesized and secreted by hemorrhage occurred in 17 patients undergoing com- the cellular components of the vessel wall and inflamma- posite grafting but in only 1 patient after an aortic valve– tory cells. Metalloproteinases function in the extracellu- sparing procedure. The actuarial 5-year survival in the lar matrix turnover by degrading matrix protein, such as valve-sparing group was 96% compared with 89% in the collagen, elastin, and proteoglycans. It is now recognized valve-conduit group. This report from Karck and col- that the walls of abdominal aortic aneurysms have an leagues31 does suggest that in appropriately selected pa- imbalance of this normal turnover mechanism, leading tients with Marfan’s syndrome, the aortic root surgery to significantly higher MMP activity in the walls of the may be successfully performed with preservation of the aneurysms and controls. This is especially true for patient’s native valve. So despite the continued contro- MMP-9. This imbalance favors elastin fragmentation, versy about this strategy, very fine early and intermediate which is thought to lead to aneurysm formation. Boyum term results are attainable. and colleagues30 found that MMP-2 activity was signif- Management of intramural hematoma of the aorta icantly greater in aneurysms associated with bicuspid remains confusing. This is now recognized to be basi- aortic valves as compared with those associated with tri- cally a disease of the media of the aortic wall and for that cuspid valves. Of note, total MMP-9 activity was also reason, some authors have advocated treating this lesion greater in the aneurysms associated with bicuspid valves. as an aortic dissection. Type A intramural hematoma is It should be noted that one of the principal limitations found in the ascending aorta and Type B intramural of this study was that tissue was obtained at an isolated hematoma is found in the descending thoracic aorta. point in time in the continuum of the disease, so a spe- With a paucity of data, most programs have recom- cific cause-and-effect relationship between MMP ex- mended immediate surgical therapy forTypeA intramu- pression and aneurysmal dilatation of the aorta should ral hematoma but medical therapy for Type B. Moizumi not be definitively concluded. Nonetheless, the findings and colleagues32 reported their experience with 94 cases do suggest that increased activity of MMP in the walls of of intramural hematoma (41 Type A and 53 Type B). aneurysms associated with a bicuspid aortic valve may Initial medical therapy (control of hypertension) was contribute to aneurysm formation. initiated in all patients. Immediate operations were per- The appropriate operation for aortic root surgery and formed only in the instance of cardiac tamponade, im- Marfan’s syndrome remains somewhat controversial. pending rupture, or rupture. Patients treated medically The gold standard has been a mechanical valve conduit, were converted to surgical therapy for the following rea- which is an effective therapy yet does obligate these pa- sons: recurrence of chest or back pain, suggesting im- tients to lifelong anticoagulation. Since the mid 1990s, pending rupture, progression to an overt Type A dissec- aortic valve–sparing operations following the technique tion, progressive aortic dilatation to more than 6 cm in of David have been more widely applied for acquired maximal diameter, or progressive enlargement of the 752 Fullerton What’s New in Cardiac Surgery J Am Coll Surg ulcer-like projection to more than 20 mm in diameter. [Pfizer]) in the amelioration of acute pulmonary hyper- Using these criteria, 27% of patients with a Type A he- tension with a porcine model. Sildenafil is a phosphodi- matoma and only 2% of patients with a Type B hema- esterase V inhibitor. In this study, acute pulmonary hy- toma underwent early surgical intervention. But during pertension was achieved by infusion of U46610, followup, 22% of patients with a Type A and 26% of standard model of acute pulmonary hypertension. Once patients with a Type B hematoma required operation. In in a steady state, the sildenafil analog UK343–64 was short, Moizumi and colleagues demonstrated that aorta- administered in a single dose over a 2-minute time in- related events occur with approximately equal frequency terval. In the treatment group, pulmonary vascular resis- in intramural hematomas of the ascending and descend- tance and pulmonary arterial pressure were significantly ing aorta. A therapeutic strategy in which either type is lowered by the administration of the sildenafil analog. initially managed medically seems warranted. The authors concluded that this therapeutic strategy Another aortic pathologic condition related to intra- might be valuable in treating postoperative cardiac sur- mural hematoma is a penetrating ulcer of the aorta. gical patients. Management of a penetrating atherosclerotic ulcer re- Pulmonary embolism continues to be a major cause of mains controversial. Although some authors advocate postoperative morbidity and mortality in surgical pa- aggressive intervention on diagnosis, others have recom- tients. Yalamanchili and colleagues35 reported a series of mended medical therapy. Cho and colleagues33 reviewed 19 patients with acute major pulmonary embolism who 105 patients with penetrating atherosclerotic ulcer over underwent pulmonary embolectomy. The procedures a 25-year period from 1977 through 2002. In nonoper- were performed with cardiopulmonary bypass with or ated patients in followup, the mean thickness of the without aortic cross-clamp. An inferior vena cava filter lesion decreased at 1 month in approximately 90% of was placed in all patients preoperatively except one. The patients and had completely resolved in 1 year in 85%. survival rate after this open pulmonary embolectomy Four percent of patients died within 30 days of diagnosis procedure was 92%. The patients who died presented among 76 patients treated medically, and 6 deaths with cardiopulmonary arrest before the procedure and (21%) occurred among 29 patients treated surgically. suffered hypoxic encephalopathy. The authors con- Acknowledging that this is a retrospective review and cluded that these data, combined with other studies of subject to the limitations of all retrospective analyses, it the literature, strongly support the use of open pulmo- is interesting to note that the early mortality rate among nary embolectomy as an effective therapy for treatment medically treated patients was only 3%, although the of acute massive pulmonary embolism. mortality rate among surgically treated patients was 40%. This likely reflects in large part the significant Atrial ﬁbrillation comorbidities in this patient population. The authors In recent years, surgical correction of atrial fibrillation emphasized the importance of longterm followup and has become more and more common, and several surgi- highlighted the possibility that sacular aneurysm forma- cal techniques and technologies have been applied. tion could occur and rupture with this group of patients. Chiappini and colleagues36 reported their experience Nonetheless, the authors were unable to identify any comparing a standard cut-and-sew Cox/Maze III proce- usual predictors of clinical behavior for this aortic pa- dure with radiofrequency ablation therapy of atrial thology. The authors concluded by acknowledging that fibrillation. Between 1995 and 2002, 70 patients under- their recommendation for expectant management is at went operations for atrial fibrillation by Chiappini and odds with other authors. colleagues. Thirty patients underwent a standard Cox/ Maze III procedure and 40 patients underwent radiofre- Pulmonary hypertension quency ablation of at least the left atrium. The results of Acute pulmonary hypertension continues to be a very this group demonstrated that there was no statistically vexing problem in the management of preoperative car- significant difference between the two surgical ap- diac surgical patients. Current therapeutic modalities in- proaches in the achievement of sinus rhythm after the clude inhaled nitric oxide, milrinone, and other intrave- procedure. It should be noted that in the longitudinal nously administered vasodilating agents. Bonnell and followup of up to 40 months, there was a significant loss colleagues34 reported special use of sildenafil (Viagra of sinus rhythm in both groups. Acknowledging that Vol. 200, No. 5, May 2005 Fullerton What’s New in Cardiac Surgery 753 this study is a retrospective and nonrandomized review to cardiopulmonary bypass during attempted off-pump of the authors’ experience, it is noteworthy that there revascularization results in increased morbidity and mortality. J Thorac Cardiovasc Surg 2004;128:655–661. was no early difference between these two techniques to 7. Reuthebuch O, Kadner A, Lachat M, et al. Early bypass occlu- eradicate atrial fibrillation. sion after deployment of Nitinol connector devices. J Thorac Romano and colleagues37 reported their experience Cardiovasc Surg 2004;127:1421–1426. 8. Bergsland J, Hol PK, Lingås PS, et al. Intraoperative and with 36 patients undergoing the Maze procedure with intermediate-term angiographic results of coronary artery by- bilateral atrial reduction plasty. A standard Cox/Maze III pass surgery with Symmetry proximal anastomotic device. procedure was performed with use of radiofrequency J Thorac Cardiovasc Surg 2004;128:718–723. ablation to completely encompasses the pulmonary 9. Medalion B, Meirson D, Hauptman E, et al. Initial experience with the Heartstring proximal anastomotic system. J Thorac veins and the posterior left atrium in the region of the Cardiovasc Surg 2004;128:273–277. mitral valve annulus. The surgical technique for atrial 10. Filsoufi F, Farivar RS, Aklog L, et al. Automated distal coronary reduction plasty included resection of both atrial ap- bypass with a novel magnetic coupler (MVP system). J Thorac Cardiovasc Surg 2004;185–192. pendages and biatrial reduction by use of resection of the 11. Allen KB, Dowling RD, Schuch DR, et al. Adjunctive left atrial posterior wall in the region from left to right transmyocardial revascularization: Five-year follow-up of a pro- pulmonary veins and from the inferior veins to the mi- spective, randomized trial. Ann Thorac Surg 2004;78:458–465. tral valve annuloplasty. In addition, removal of a signif- 12. Mühling OM, Wang Y, Jerosch-Herold M, et al. Improved myocardial function after transmyocardial laser revascularization ac- icant portion of the right atrial wall was included. The cording to cine magnetic resonance in a porcine model. J Thorac preoperative left atrium measured 66 Ϯ 16 mm. Only 3 Cardiovasc Surg 2004;128:391–395. of these 36 patients underwent an isolated Maze proce- 13. Allen KB, Dowling RD, Angell WW, et al. Transmyocardial dure and the remaining 33 had an associated procedure, revascularization: 5-year follow-up of a prospective, randomized multicenter trial. Ann Thorac Surg 2004;77:1228–1234. typically a mitral valve with or without a tricuspid valve 14. Morgan JA, John R, Rao V, et al. Bridging to transplant with the procedure. There were no perioperative deaths, and dur- HeartMate left ventricular assist device: The Columbia Presby- ing a mean followup time of 19 Ϯ 16 months, 32 of the terian 12-year experience. J Thorac Cardiovasc Surg 2004;127: 1309–1316. 36 patients were in sinus rhythm and were considered 15. Dembitsky WP,TectorAJ, Park S, et al. Left ventricular assist device New York Heart Association Class I. The authors con- performance with long-term circulatory support: Lessons from the cluded that an aggressive approach to the surgical REMATCH Trial. Ann Thorac Surg 2004;78:2123–2130. treatment of atrial fibrillation including a biatrial re- 16. Dowling RD, Gray LA, Etoch SW, et al. Initial experience with the AbioCor implantable replacement heart system. J Thorac duction plasty offers approximately 90% efficacy in Cardiovasc Surg 2004;127:131–141. eliminating atrial fibrillation even in this difficult 17. Mickleborough LL, Merchant N, Ivanov J, et al. Left ventricular group of patients. reconstruction: Early and late results. J Thorac Cardiovasc Surg 2004;128:27–37. 18. Maxey TS, Reece TB, Ellman PI, et al. Coronary artery bypass with ventricular restoration is superior to coronary artery bypass REFERENCES alone in patients with ischemic cardiomyopathy. J Thorac Car- 1. Higgins RSD, Bridges J, Burke JM, et al. Implementing the diovasc Surg 2004;127:428–434. ACGME general competencies in a cardiothoracic surgery resi- 19. Nowicki ER, Birkmeyer NJO, Weintraub RW, et al. Multivari- dency program using 360-degree feedback. Ann Thorac Surg able prediction of in-hospital mortality associated with aortic 2004;77:12–17. and mitral valve surgery in Northern New England. Ann Thorac 2. Baskett RJF,Kalavrousziotis D, Buth KJ, et al.Trainingresidents Surg 2004;77:1966–1977. in mitral valve surgery. Ann Thorac Surg 2004;78:1236–1240. 20. Kabbani SS, Jamil H, Hammoud A, et al. The mitral pulmonary 3. Salazar JD, Lee R, Wheatley GH III, Doty JR. Are there enough autograft: Assessment at midterm. Ann Thorac Surg 2004;78: jobs in cardiothoracic surgery? The Thoracic Surgery Residents 60–66. Association Job Placement Survey for finishing residents. Ann 21. Crestanello JA, McGregor CG, Danielson GK, et al. Mitral and Thorac Surg 2004;78:1523–1527. tricuspid valve repair in patients with previous mediastinal radi- 4. Mack MJ, Brown PP, Kugelmass AD, et al. Current status and ation therapy. Ann Thorac Surg 2004;78:826–831. outcomes of coronary revascularization 1999–2002: 148,396 22. Aicher D, Langer F, Kissinger A, et al. Valve-sparing aortic root surgical and percutaneous procedures. Ann Thorac Surg 2004; replacement in bicuspid aortic valves: A reasonable option? 77:761–768. J Thorac Cardiovasc Surg 2004;128:662–668. 5. Puskas JD, Williams WH, Mahoney EM, et al. Off-pump vs 23. McGee EC Jr, Gillinov AM, Blackstone EH, et al. Recurrent mitral conventional coronary artery bypass grafting: Early and 1-year regurgitation after annuloplasty for functional ischemic mitral re- graft patency, cost and quality-of-life outcomes. JAMA 2004; gurgitation. J Thorac Cardiovasc Surg 2004;128:916–924. 291:1841–1849. 24. Coselli JS, LeMaire SA, Conklin LD, Adams GJ. Left heart 6. Patel NC, Patel NU, Loulmet DF, et al. Emergency conversion bypass during descending thoracic aortic aneurysm repair does 754 Fullerton What’s New in Cardiac Surgery J Am Coll Surg

not reduce the incidence of paraplegia. Ann Thorac Surg 31. Karck M, Kallenbach K, Hagl C, et al. Aortic root surgery in 2004;77:1298–1303. Marfan syndrome: Comparison of aortic valve-sparing reim- 25. Reece TB, Okonkwo DO, Ellman PI, et al. The evolution of plantation versus composite grafting. J Thorac Cardiovasc Surg ischemic spinal cord injury in function, cytoarchitecure, and 2004;127:391–398.

inflammation and the effects of adenosine A2A receptor activa- 32. Moizumi Y, Komatsu T, Motoyoshi N, Tabayashi K. Clinical tion. J Thorac Cardiovasc Surg 2004;128:925–932. features and long-term outcome of type A and type B intramural 26. Demers P, Miller DC, Mitchell RS, et al. Midterm results of hematoma of the aorta. J Thorac Cardiovasc Surg 2004;127: endovascular repair of descending thoracic aortic aneurysms 421–428. with first-generation stent grafts. J Thorac Cardiovasc Surg 33. Cho KR, Stanson AW, Potter DD, et al. Penetrating atheroscle- 2004;664–673. rotic ulcer of the descending thoracic aorta and arch. J Thorac 27. Iannelli G, Piscione F, Di Tommaso L, et al. Thoracic aortic Cardiovasc Surg 2004;127:1393–1400. emergencies: Impact of endovascular surgery. Ann Thorac Surg 34. Bonnell MR, Urdaneta F, Kirby DS, et al. Effects of Sildenafil 2004;77:591–596. analogue UK 343–664 on a porcine model of acute pulmonary 28. Svensson LG, Kim K-H, Blackston EH, et al. Elephant trunk hypertension. Ann Thorac Surg 2004;77:238–242. procedure: Newer indications and uses. Ann Thorac Surg 2004; 35. Yalamanchili K, Gleisher AG, Lehrman SG, et al. Open pulmo- 78:109–116. nary embolectomy for treatment of major pulmonary embo- 29. Kouchoukos NT, Mauney MC, Masetti P, Castner CF. Single- lism. Ann Thorac Surg 2004;77:819–823. stage repair of extensive thoracic aortic aneurysms: Experience 36. Chiappini B, Martin-Suarez S, LoForte A, et al. Cox/Maze III with the arch-first technique and bilateral anterior thoracotomy. operation versus radiofrequency ablation for the surgical treat- J Thorac Cardiovasc Surg 2004;128:669–676. ment of atrial fibrillation: A comparative study. Ann Thorac 30. Boyum J, Fellinger EK, Schmoker JD, et al. Matrix metallopro- Surg 2004;77:87–92. teinase activity in thoracic aortic aneurysms associated with bi- 37. Romano MA, Bach DS, Pagani FD, et al. Atrial reduction plasty cuspid and tricuspid aortic valves. J Thorac Cardiovasc Surg Cox Maze procedure: Extended indications for atrial fibrillation 2004;127:686–691. surgery. Ann Thorac Surg 2004;77:1282–1287. What’s New in General Surgery: Critical Care and Trauma

Lena M Napolitano, MD, FACS, FCCP, FCCM

There have been a number of major developments in Clinical Centers and a Data and Coordinating Center critical care and trauma in 2004, ranging from new re- that will provide the necessary infrastructure to conduct search findings to critical care workforce issues and dis- multiple collaborative trials to aid rapid translation of cussions regarding optimal education and training in promising scientific and clinical advances to improve trauma and surgical critical care. We have made great resuscitation outcomes. The ongoing direct study com- advances in a number of areas, and additional initiatives mitment of the sponsors is for at least 5 years of funding to expand research efforts in trauma and critical care are at approximately $10 million per year. Information re- ongoing. garding the specific participating centers can be obtained on the ROC Web site.2 Others contributing sup- Research port to this initiative include the US Department of There are a number of obstacles to performing clinical Defense, other institutes within the NIH, the Institute of trials and outcomes studies involving critically ill and Circulatory and Respiratory Health from the Canadian In- injured patients with complex syndromes that currently stitutes of Health Research, the Canadian Defense Re- lack effective treatments. The Canadian Critical Care search and Development Program, and the National Insti- Trials Group and the Australia and New Zealand Inten- tute of Neurological Disorders and Stroke. sive Care Society Clinical Trials Group have established There is significant concern that it has become in- comprehensive research efforts in this regard. The Acute creasingly difficult to conduct clinical research in trauma Respiratory Distress Syndrome (ARDS) Clinical Net- and critical care, and that it may negatively impact on work (ARDSNet) was the first organized US national future efforts.3 Difficulties in obtaining patient or sur- effort to conduct multicenter clinical trials in critical rogate consent and in determining the standard of care care, established in 1994.1 for the control group are relevant issues.4 An important The US and Canada launched a major collaborative statement from the American Thoracic Society regard- research program in 2004. The leading federal health ing the ethical conduct of clinical research involving research agencies in the US and Canada—the National critically ill patients was published in 2004, providing Institutes of Health (NIH) and the Canadian Institutes some principles and guidance in this difficult area.5 of Health Research (CIHR)—formed a partnership to advance research in cardiovascular and respiratory Education and resident training diseases. The Guidelines for Critical Care Medicine Training and Three research programs were initiated in 2004, in- Continuing Medical Education were published this year cluding “Clinical Research Consortium to Improve Re- by the American College of Critical Care Medicine.6 suscitation Outcomes,” which will address novel strate- Guidelines for the continuum of education in critical gies to resuscitate heart attack and trauma patients. This care, from residency training through specialty training is the first organized multicenter effort to perform resus- and ongoing through practice, will facilitate standard- citation research in trauma in the US. The Resuscitation ization of physician education in critical care medicine. Outcomes Consortium (ROC) consists of 10 Regional The future of trauma care as a specialty and the optimal training paradigm for those interested in trauma and surgical critical care have come under great scrutiny. Received February 11, 2005; Accepted February 11, 2005. In 2004, a number of national organizations established From the Department of Surgery, University of Michigan School of Medi- cine, Ann Arbor, MI. an initiative to collaborate and coordinate with all of the Correspondence address: Lena M Napolitano, MD, FACS, FCCP, FCCM, professional societies that represent surgical critical care, University of Michigan School of Medicine, Department of Surgery, Section of General Surgery, University of Michigan Health System,1500 E Medical trauma, and acute surgery. This effort includes the Center Dr, Ann Arbor, MI 48109-0331. American Board of Surgery, American Association for

© 2005 by the American College of Surgeons ISSN 1072-7515/05/$30.00 Published by Elsevier Inc. 755 doi:10.1016/j.jamcollsurg.2005.02.018 756 Napolitano What’s New in Trauma and Critical Care J Am Coll Surg

Numerous studies demonstrate that critical care ser- Abbreviations and Acronyms vices directed by physicians formally trained in critical CCM ϭ critical care medicine care medicine reduce mortality in the ICU and reduce ϭ CRB catheter-related bacteremia health-care costs. To address the shortage, the critical ECG ϭ electrocardiographic HBOC ϭ hemoglobin-based oxygen carrier care professional societies recommend that steps be HR ϭ hazard ratio taken to improve the efficiency of critical care providers, NPPV ϭ noninvasive positive pressure ventilation to increase the number of critical care providers, and to NTDB ϭ National Trauma Data Bank PEEP ϭ positive end-expiratory pressure address the demand for critical care services. rFVIIa ϭ recombinant factor VIIa TBI ϭ traumatic brain injury Critical care use in the US TRISS ϭ Trauma and Injury Severity Score VAP ϭ ventilator-associated pneumonia A recent analysis of nonfederal acute care hospitals in the US documented that the number of hospitals offering critical care medicine (CCM) between 1985 and 2000 decreased by 13.7% (4,150 to 3,581). Also, total hospi- the Surgery of Trauma, the American College of Sur- tal beds decreased substantially, by 26.4% (889,600 to geons Committee on Trauma, the Eastern Association 654,400) in these hospitals. In contrast, CCM beds infor the Surgery of Trauma, the Surgical Section of the creased by 26.2% (69,300 to 87,400) and CCM bed Society of Critical Care Medicine, the American Burn costs per day increased by 126% ($1,185 to $2,674 US). Association, and the WesternTraumaAssociation.These Although CCM costs increased by 190.4% ($19.1 bil- organizations are examining options for linking the spe- lion to $55.5 billion US), the proportion of national cialties of general surgery, trauma surgery, and surgical health expenditures allocated to CCM decreased by critical care to create a career practice model for the 5.4%. In 2000, CCM costs represented 13.3% of hos- future. pital costs, 4.2% of national health expenditures, and 0.56% of the gross domestic product. These authors Critical care workforce and manpower issues concluded that CCM is increasingly used and promi- The US is currently facing an unprecedented, and nent in a shrinking US hospital system.11 largely unrecognized, shortage of physicians trained to The Leapfrog Group has recommended that imple- provide critical care services. This is one of the most mentation of the Intensive Care Unit Physician Staffing pressing issues affecting the future of our aging popula- be standard in all hospitals.12 The ICU Physician Staff- tion and American medicine. As initially described in a ing standard requires that intensive care units have a study by the Committee on Manpower for Pulmonary dedicated intensivist present during daytime hours. and Critical Care Societies (COMPACCS), future de- Outside of these hours, an intensivist must be immedi- mand for critical care services in the US will soon exceed ately available by pager, and a physician or “physician the capabilities of the current delivery system.7 The most extender” must be in the hospital and able to immedi- alarming problem is the anticipated shortage of health- ately reach intensive care unit patients. Research shows care professionals practicing critical care. that if the first three leaps (Computer Physician Order This past year, the four major critical care societies in Entry, Intensive Care Unit Physician Staffing, and the US (American Association of Critical Care Nurses, Evidence-Based Hospital Referral) were implemented in American College of Chest Physicians, American Tho- all urban hospitals in the US, we could save up to 65,341 racic Society, and Society of Critical Care Medicine) lives and prevent as many as 907,600 serious medication united in their efforts to address the shortage of health- errors each year. care providers who care for the critically ill.8 The Critical A recent study using financial modeling examined Care Workforce Partnership was established and the rec- hospital costs and savings over a 1-year period of imple- ommendations for actions were reported by the FOCCUS menting the ICU Physician Staffing standard compared (Framing Options for Critical Care in the United States) with the existing standard of nonintensivist staffing in Task Force.9 An additional publication outlined federal adult ICUs using published data for nonrural hospi- policy initiatives to address the shortage of critical care tals.13 Cost savings ranged from $510,000 to $3.3 mil- providers.10 lion US for 6- to 18-bed intensive care units. The best- Vol. 200, No. 5, May 2005 Napolitano What’s New in Trauma and Critical Care 757 case scenario demonstrated savings of $4.2 to $13 International consensus conference—pancreatitis million US. Under the worst-case scenario, there was a Acute pancreatitis represents a spectrum of disease rang- net cost of $890,000 to $1.3 million US. These eco- ing from a mild, self-limited course requiring only brief nomic findings must be interpreted in the context of hospitalization to a rapidly progressive, fulminant illness significant reductions in patient morbidity and mortal- resulting in multiple organ dysfunction syndrome, with ity rates also associated with intensivist staffing. The au- or without accompanying sepsis. An international con- thors concluded by stating that, given the magnitude of sensus conference was held in April 2004 to develop its clinical and financial impact, hospital leaders should recommendations for the management of the critically be asking “how to” rather than “whether to” implement ill patient with severe acute pancreatitis. Evidence-based The Leapfrog Group’s ICU Physician Staffing standard. recommendations were developed by a jury of 10 persons representing surgery, internal medicine, and critical care after conferring with experts and reviewing the per- Surviving sepsis guidelines tinent literature to address specific questions concerning In 2003, critical care and infectious disease experts the management of patients with severe acute pancreati- representing 11 international organizations developed tis. There were a total of 23 recommendations developed management guidelines for severe sepsis and septic to provide guidance to critical care clinicians caring for shock that would be of practical use for the bedside the patient with severe acute pancreatitis.16 clinician, under the auspices of the Surviving Sepsis These included a recommendation against the rou- Campaign, an international effort to increase awareness tine use of prophylactic systemic antibacterial or anti- and improve outcomes in severe sepsis. These guidelines fungal agents in patients with or without necrotizing 14 were published in 2004. The impact of these guidelines pancreatitis.The jury also recommended against pancre- will be formally tested and guidelines updated annually atic debridement or drainage for sterile necrosis, limiting and even more rapidly as important new knowledge be- debridement or drainage to those with infected pancre- comes available. atic necrosis or abscess confirmed by radiologic evidence An effort to translate these evidence-based recom- of gas or results of fine needle aspiration. The jury also mendations regarding many aspects of the acute man- recommended that whenever possible, operative necro- agement of sepsis and septic shock into improved out- sectomy or drainage be delayed at least 2 to 3 weeks to comes for the critically ill patient is under way by the use allow for demarcation of the necrotic pancreas. These of “sepsis bundles.” A bundle is a selected set of inter- recommendations differ in several ways from previous ventions or processes of care distilled from evidence- recommendations because of the release of recent data based practice guidelines that, when implemented as a concerning the management of these patients and also group, provides a more robust picture of the quality of because of the focus on the critically ill patient. There care provided. The Severe Sepsis Bundles are a distilla- were a number of important questions that could not be tion of the concepts and recommendations found in the answered using an evidence-based approach, and areas practice guidelines published by the Surviving Sepsis in need of further research were identified. Campaign in 2004. The Severe Sepsis Bundles are designed to allow teams Resuscitation to follow the timing, sequence, and goals of the individ- Two important clinical studies on fluid resuscitation ual elements of care, in order to achieve the goal of a 25% were published in 2004. The Saline versus Albumin reduction in mortality from severe sepsis. There are cur- Fluid Evaluation (SAFE) Study was a multicenter, ran- rently two sepsis bundles recommended for use: domized, double-blind trial comparing the effect of fluid resuscitation with albumin or saline on mortality in Sepsis Resuscitation Bundle: Tasks that should begin im- a heterogeneous population of patients in the ICU.17 mediately but must be done within 6 hours for patients This study was a collaboration of the Australian and with severe sepsis or septic shock. New Zealand Intensive Care Society Clinical Trials Sepsis Management Bundle: Tasks that should begin im- Group, the Australian Red Cross Blood Service, and the mediately but must be done within 24 hours for pa- George Institute for International Health. Patients were tients with severe sepsis or septic shock.15 randomized to receive either 4% albumin or normal 758 Napolitano What’s New in Trauma and Critical Care J Am Coll Surg saline for intravascular-fluid resuscitation during the 28 Respiratory failure days after ICU admission. Of the 6,997 patients who A number of studies have previously documented that underwent randomization, 3,497 were assigned to re- the use of noninvasive positive pressure ventilation ceive albumin and 3,500 to receive saline; the two (NPPV) for respiratory failure related to cardiogenic groups had similar baseline characteristics. There was no pulmonary edema and COPD exacerbation is associated difference in 28-day mortality identified (relative risk of with improved outcomes. But, does noninvasive positive death, 0.99; 95% CI 0.91 to 1.09; p ϭ 0.87). pressure ventilation improve outcomes in acute hypox- Interestingly, among the 1,186 patients with trauma, emia respiratory failure unrelated to these diagnoses?21 A albumin was associated with a trend toward increased systematic review of prospective randomized trials re- mortality (relative risk of death in the albumin group, cently addressed this question. This study confirmed 1.36; 95% CI 0.99 to 1.86), possibly explained by the that the addition of NPPV to standard care in the setting effect of trauma associated with brain injury (relative of acute hypoxemic respiratory failure reduced the rate risk of death, 1.62; 95% CI 1.12 to 2.34). In contrast, of endotracheal intubation (absolute risk reduction among 1,218 patients with severe sepsis, albumin was 23%, 95% CI 10% to 35%) and ICU length of stay associated with a trend toward reduced mortality (rel- (absolute reduction 2 days, 95% CI 1 to 3 days); ICU ative risk of death, 0.87; 95% CI 0.74 to 1.02). These mortality was less clear, and the heterogeneity found subgroup analyses should be interpreted with cau- among studies suggested that effectiveness varies among tion and will require validation in future different populations. As a result, the literature does not studies. support the routine use of NPPV in all patients with acute hypoxemic respiratory failure. A prehospital trial of hypertonic saline resuscitation In patients who require intubation for respiratory fail- in patients with hypotension (systolic blood pressure ure, there appears to be accumulating evidence in favor Ͻ 100 mmHg) and severe traumatic brain injury (Glasgow of early tracheostomy. A prospective randomized study Coma Scale score Ͻ 9, n ϭ 229) was conducted in in 120 patients projected to need mechanical ventilation Australia.18 Patients were randomly assigned to receive a for less than 14 days was randomized to early percuta- rapid intravenous infusion of either 250 mL of 7.5% neous tracheotomy within 48 hours or delayed trache- saline (n ϭ 114) or 250 mL of Ringer’s lactate solution ϭ otomy at days 14 to 16. The early tracheotomy group (n 115; controls) in addition to conventional intrave- had significantly less mortality (31.7% versus 61.7%), nous fluid and resuscitation protocols administered by pneumonia (5% versus 25%), and accidental extuba- paramedics. No significant differences in survival at tions (0 versus 6) compared with the prolonged transla- hospital discharge, 6-month survival, or functional ryngeal group. The early tracheotomy group spent less neurologic outcomes at 6 months after injury were time in the ICU (4.8 Ϯ 1.4 days versus 16.2 Ϯ 3.8 days) identified. and on mechanical ventilation (7.6 Ϯ 2.0 days versus There is still great interest in the use of hypertonic 17.4 Ϯ 5.3 days). There was also significantly more saline for resuscitation in shock, because of its ability to damage to the mouth and larynx in the prolonged trans- 19 modulate the inflammatory response and reduce fluid laryngeal intubation group.22 20 resuscitation volume requirements. In fact, two multi- We have now recognized the high incidence of delir- center clinical trials under consideration by the Resusci- ium as a common form of organ dysfunction in the ICU, tation Outcomes Consortium seek to determine the im- especially in patients requiring mechanical ventilation. pact of hypertonic resuscitation on survival in In a prospective cohort study of 224 mechanically ven- hypovolemic shock in trauma, and on longterm neuro- tilated patients, 81.5% developed delirium at some logic outcomes in blunt severe traumatic brain injury. point during the ICU stay.23 Patients who developed Both studies will be three arm, randomized, blinded delirium had higher 6-month mortality rates (34% ver- intervention trials comparing hypertonic saline/dextran sus 15%, p ϭ 0.03) and spent 10 days longer in the (7.5% saline, 6% dextran 70), hypertonic saline alone hospital than those who never developed delirium (p Ͻ (7.5% saline), and normal saline as the initial resuscita- 0.001). After adjusting for covariates (including age, se- tion fluid administered to these patients in the prehos- verity of illness, comorbid conditions, coma, and use of pital setting. sedatives or analgesic medications), delirium was inde- Vol. 200, No. 5, May 2005 Napolitano What’s New in Trauma and Critical Care 759 pendently associated with higher 6-month mortality Infections (adjusted hazard ratio [HR], 3.2; 95% CI, 1.4 to 7.7; Prevention of health care–associated infections contin- p ϭ 0.008), and longer hospital stay (adjusted HR, 2.0; ues to be a primary goal in the care of all critically ill and 95% CI, 1.4 to 3.0; p Ͻ 0.001). Delirium in the ICU injured patients. was also independently associated with a longer post- Health care–associated pneumonia ϭ ICU stay (adjusted HR, 1.6; 95% CI, 1.2 to 2.3; p Pneumonia continues to be a leading cause of health 0.009), fewer median days alive and without mechanical care–associated infection in the ICU, and all efforts to ventilation (19 days [interquartile range, 4 to 23 days] prevent pneumonia should be implemented. The CDC versus 24 days [interquartile range 19 to 26 days]; ad- Guidelines for the Prevention of Healthcare-Associated Pneu- ϭ justed p 0.03), and a higher incidence of cognitive monia were published in 2004.27 Most recently, the Guide- impairment at hospital discharge (adjusted HR, 9.1; lines for the Management of Adults with Hospital-Acquired, ϭ 95% CI, 2.3 to 35.3; p 0.002). Ventilator-Associated, and Healthcare-Associated Pneumonia The ARDSNet trial previously documented that in were published by the American Thoracic Society and the patients with acute lung injury and the acute respiratory Infectious Diseases Society of America.28 These two docu- distress syndrome, mechanical ventilation with a lower ments provide comprehensive and current information tidal volume (6 versus 12 mL/kg) resulted in decreased regarding evidence-based practices for prevention, diag- mortality and increased the number of days without nosis, and treatment of pneumonia in critically ill and ventilator use.24 In 2004, the ARDSNet group published injured patients. the trial comparing the effects of higher and lower pos- An important study examined the optimal duration itive end-expiratory pressure (PEEP) levels on clinical of antimicrobial therapy for the treatment of ventilator- outcomes in ARDS patients (n ϭ 549).25 This study associated pneumonia (VAP).29 A prospective, random- documented that in patients with acute lung injury and ized, double-blind clinical trial conducted in 51 French ARDS who received mechanical ventilation with a tidal- ICUs enrolled 401 patients diagnosed with VAP by volume goal of 6 mL/kg of predicted body weight and an quantitative culture of bronchoscopic specimens and end-inspiratory plateau-pressure limit of 30 cm of water, who had received initial appropriate empirical antimi- clinical outcomes were similar whether lower or higher crobial therapy. Patients were randomized to receive 8 or PEEP levels were used. 15 days of antibiotic therapy. This study documented Two multicenter, randomized, double-blind phase comparable clinical effectiveness against VAP with the trials of a protein-containing surfactant in adults with 8- and 15-day antibiotic treatment regimens, and de- ARDS from various causes (n ϭ 448) compared stan- creased antibiotic use in the 8-day group. Importantly, dard therapy alone with standard therapy plus up to four patients with VAP caused by nonfermenting gram- intratracheal doses of a recombinant surfactant protein negative bacilli, including Pseudomonas aeruginosa, did C–based surfactant given within a period of 24 hours.26 have a higher pulmonary infection-recurrence rate com- There was no significant difference between the groups pared with those receiving 15 days of treatment (40.6% in terms of mortality or the need for mechanical venti- versus 25.4%; difference, 15.2%, 90% CI, 3.9% to lation. Patients receiving surfactant had a significantly 26.6%). greater improvement in blood oxygenation during the Catheter-related bacteremia initial 24 hours of treatment than patients receiving The importance of education and process improvement standard therapy, according to both univariate and mul- in prevention of catheter-related bacteremia (CRB) has tivariate analyses. been documented in a number of studies. A new bench- Surfactant replacement therapy for ARDS in neo- mark has recently been set, with complete elimination of nates is standard of care. Future studies with surfactant CRB in the surgical ICU in a single institution study therapy in ARDS may consider alternative formulations over 5 years (1998 to 2002).30 To eliminate CRB, a qual- of recombinant surfactant, earlier initiation of surfactant ity improvement team implemented five interventions: therapy, studies in more homogenous patient groups, educating the staff; creating a catheter insertion cart; and different methods of delivery of surfactant to the asking providers daily whether catheters could be re- injured lung. moved; implementing a checklist to ensure adherence to 760 Napolitano What’s New in Trauma and Critical Care J Am Coll Surg evidence-based guidelines for preventing CRB; and em- catheter with standard care without the use of a powering nurses to stop the catheter insertion procedure pulmonary-artery catheter.34 The subjects were high-risk if a violation of the guidelines was observed. During the patients 60 years of age or older, with American Society intervention time period, the CRB rate decreased from of Anesthesiologists class III or IV risk, who were sched- 11.3 of 1,000 catheter days in the first quarter of 1998 to uled for urgent or elective major surgery, followed by a 0 of 1,000 catheter days in the fourth quarter of 2002. stay in an ICU. Outcomes were adjudicated by observers These interventions may have prevented 43 CRB, 8 who were unaware of the treatment group assignments. deaths, and $1,945,922 in additional costs per year. This Of 3,803 eligible patients, 1,994 (52.4%) underwent rate of zero CRB has been maintained over the next 2 randomization. No difference in hospital mortality rates years. Multifaceted interventions that helped to ensure was identified (7.7% in control group veresus 7.8% in adherence with evidence-based infection control guide- pulmonary-artery catheter group, 95% CI, Ϫ2.3 to lines to eliminate CRB should be implemented in all 2.5). Furthermore, no difference in 6- or 12-month sur- ICUs. vival rates was identified. This study found no benefit to therapy directed by pulmonary-artery catheter over Glucose control standard care in elderly, high-risk surgical patients re- Insulin resistance and hyperglycemia are common in quiring intensive care. But it also questions whether the critical illness. Increasing evidence has documented that physiologic end points used in the group that received a hyperglycemia is associated with adverse outcomes in pulmonary-artery catheter were appropriate. Clearly, both critically ill and injured patients.31 Unraveling the additional research is warranted in this important area. molecular mechanisms of glucose toxicity and the salu- The use of perioperative beta-blockers is standard for tary effects of insulin will provide new insights and open patients with cardiac risk. Animal studies suggest that avenues for novel therapeutic strategies.32 Efforts to es- the beta-blocker propranolol increases bone formation. tablish safe and effective protocols for tight glucose con- An interesting recent study documented that the use of trol in these patients are continuing. beta-blockers alone and in combination therapy with thiazides in humans is associated with a reduced risk of Cardiac issues fractures.35 Experts from the American Heart Association’s Coun- cils on Cardiovascular Nursing, Clinical Cardiology, and Blood transfusion Cardiovascular Disease in the Young and the Interna- Consistent evidence has emerged regarding the risks and tional Society of Computerized Electrocardiology pub- adverse outcomes associated with blood transfusion in lished a comprehensive consensus document to provide trauma and critical care. Published data regarding the recommendations for best practices in electrocardio- efficacy of red blood cell transfusion in the critically ill graphic (ECG) monitoring in the hospital setting.33 This confirms that transfusion does not improve tissue oxy- consensus document represents the first attempt in the gen consumption consistently in critically ill patients, literature to encompass all areas of hospital cardiac mon- either globally or at the level of the microcirculation. itoring, including arrhythmia, ischemia, and QT- Lack of efficacy of RBC transfusion likely is related to interval monitoring in both children and adults. This storage time, increased endothelial adherence of stored report focuses on real-time ECG monitoring, and em- RBCs, nitric oxide binding by free hemoglobin in stored phasis is on the information clinicians need to know to blood, donor leukocytes, host inflammatory response, monitor patients safely and effectively. Recommenda- and reduced red cell deformability.36 tions are made with regard to indications, time frames, Despite the frequent use of red cell transfusions, only and strategies to improve the diagnostic accuracy of car- one large randomized trial (Transfusion in Critical Care diac arrhythmia, ischemia, and QT-interval monitoring. trial) has examined red cell administration in the critical Currently available ECG lead systems are described, and care setting, documenting the safety of a restrictive trans- recommendations related to staffing, training, and meth- fusion strategy (transfuse only if hemoglobin Ͻ 7g/ ods to improve quality are provided. dL).37 A recent post hoc analysis of the trauma patient co- A randomized controlled trial recently compared hort (n ϭ 203) from the Transfusion in Critical Care trial goal-directed therapy guided by a pulmonary-artery documented that 30-day all-cause mortality rates in the Vol. 200, No. 5, May 2005 Napolitano What’s New in Trauma and Critical Care 761 restrictive group were 10%, as compared with 9% in the advisory committee for this protocol includes navy, liberal group (p ϭ 0.81). The presence of multiple organ army, and air force medical researchers and academic dysfunction, the changes in multiple organ dysfunction experts in trauma, emergency medicine, critical care, from baseline scores adjusted for death, and the length of and statistics. We anticipate that risks and benefits of stay in the ICU and hospital also were similar between the alternatives to red blood cell transfusion will be eluci- restrictive and liberal transfusion groups. This study con- dated in the coming years. firmed that a restrictive red blood cell transfusion strategy appears to be safe for critically ill multiple-trauma Recombinant factor VIIa for hemorrhage control patients.38 Recombinant FVIIa (rFVIIa) has been approved for The Surviving Sepsis Guidelines also concluded that treatment of bleeding in hemophilia patients with inhib- “In the absence of extenuating circumstances and fol- itors and in nonhemophilia patients with acquired anti- lowing resolution of tissue hypoperfusion, red blood cell bodies against FVIII (acquired hemophilia). Pharmaco- transfusion should be targeted to maintain hemoglobin logic doses of rFVIIa have been found to enhance at 7.0 g/dL or greater.”39 Continued efforts to reduce thrombin generation on activated platelets and may also blood transfusion in critically ill patients are required. likely be of benefit in providing hemostasis in other sit- These strategies will require education, unit and institu- uations characterized by profuse bleeding and impaired tional protocols, and reduction of phlebotomy for diag- thrombin generation, such as in patients with thrombo- nostic laboratory testing in the ICU. cytopenia and in those with functional platelet defects. We also have an increased understanding of the Additionally, it has been used successfully in a variety of pathophysiology of the anemia associated with critical less well-characterized bleeding situations and in pa- care, related to the inflammatory response, downregula- tients with impaired liver function and trauma.42 To tion of erythropoietin, and lack of iron availability from date, case reports, case series, anecdotal experience, and macrophage sequestration. Clinical trials are under way limited clinical trials describe these uses of rFVIIa; data to confirm the efficacy of recombinant erythropoietin in from randomized clinical trials are limited. the treatment of critically ill patients with anemia.40 A small double-blind, randomized placebo-controlled trial (n ϭ 36) documented that an intravenous bolus (20 Hemoglobin-based oxygen carriers or 40 ␮g/kg)) of rFVIIa significantly reduced perioper- Alternatives to blood transfusion in the form of ative median blood loss and reduced the need for allo- hemoglobin-based oxygen carriers (HBOC) continue to geneic blood transfusion in patients undergoing retro- undergo intense investigation.41 A prehospital trial has pubic prostatectomy, which is often associated with recently been initiated with the human HBOC Poly- major blood loss and need for transfusion. In fact, no Heme (Northfield Labs). Patient enrollment is under patients who received 40 ␮g/kg of rFVIIa needed blood way in the landmark phase III study designed to evaluate transfusion, and no associated adverse events were the safety and efficacy of PolyHeme when used to treat identified.43 patients in hemorrhagic shock following traumatic inju- The results of a phase II multicenter, multinational, pro- ries. Treatment begins at the scene of injury, continues in spective, randomized, double-blind, placebo-controlled the ambulance during transport, and for 12 hours trial (n ϭ 301) regarding the efficacy and safety of rFVIIa postinjury in the hospital. This is the first US trial of a as adjunctive therapy for control of severe traumatic blood substitute in which treatment begins at the scene hemorrhage were reported in 2004. Patients with blunt of injury. The trial will be conducted in approximately or penetrating trauma and blood loss, requiring transfu- 20 to 25 Level I trauma centers throughout the US and sion of 8 units of red blood cells, were randomized to 720 patients will be enrolled in the trial. receive 3 infusions of rFVIIa (200 ␮g/kg, 100 ␮g/kg, A prehospital clinical trial with the bovine HBOC- and 100 ␮g/kg) or placebo at entry, 1, and 3 hours in 201 Hemopure (Biopure) for resuscitation of patients addition to standard surgical treatment. The primary with severe hemorrhagic shock is also planned. Entitled study outcomes measure was total transfusion require- “Restore Effective Survival in Shock” (RESUS), the trial ment. In blunt trauma, there was a significant decrease is intended to support an indication for out-of-hospital in the number of transfusions required within 48 hours military and civilian trauma applications. The scientific and a trend to reduced organ failure and ARDS. No 762 Napolitano What’s New in Trauma and Critical Care J Am Coll Surg significant differences were identified in the penetrating longer be routinely used in patients with TBI.51 New trauma group. No safety issues were identified in either advances in both invasive52 and noninvasive53 monitor- group.44 ing devices for measurement of cerebral oxygenation54 It is important to recognize that efficacy of rFVIIa is and level of consciousness55 in TBI patients continue to reduced significantly with hypothermia and acidosis. emerge. In addition, significant advances in surgical The recommended dose of rFVIIa in patients with he- therapy of severe TBI, including decompressive craniec- mophilia is 90 ␮g/kg, and the optimal dosing of rFVIIa tomy, have been reported.56 Guidelines for the surgical in trauma and other ICU patients with coagulopathy management ofTBI are under development by the Brain and bleeding is not known at present. The optimal tim- Trauma Foundation. ing and redosing are also unclear, because the half-life of rFVIIa is approximately 2 hours. Discussions regarding Trauma outcomes—National Trauma Data initiation of a US trial of rFVIIa in trauma are under Bank (NTDB) way. The NTDB has provided us with an opportunity to critically examine patient outcomes using very large pa- Hemorrhage control—hemostatic agents tient cohorts. A number of clinical studies on a variety of Significant advances have been made in local hemor- interesting topics were published in 2004. The tradi- rhage control with the development of new hemostatic tional Trauma and Injury Severity Score (TRISS) re- agents for application in traumatic injury. The first mains a standard method for survival prediction and FDA-approved product is QuikClot (Z-Medica), which correction for severity in trauma outcomes analyses. In has documented efficacy in a number of preclinical tri- the NTDB, the traditional TRISS had limited ability to als.45 A number of other hemostatic products are under- predict survival after trauma. Accuracy of prediction was going both preclinical46,47 and clinical investigation.48 improved by recalculating the TRISS coefficients, but further improvements were not seen with models that Traumatic brain injury included information about comorbidities.57 A large de- Traumatic brain injury (TBI) is a leading cause of death scriptive study of outcomes of patients with combined and disability in trauma. Despite the publication and burn and trauma injuries documented increased mortal- dissemination of traumatic brain injury treatment ity despite similar total body surface area burned when guidelines in 1995, it has been documented that imple- compared with patients with burn injury as the sole mentation is infrequent, with only 16% of 433 surveyed mechanism.58 The largest study of motorcycle trauma trauma centers in full compliance with the Guidelines.49 performed using the NTDB documented that nonhel- A single-institution study documented that implemen- meted motorcyclists have worse outcomes than their tation of these Guidelines was associated with both im- helmeted counterparts independent of the use of alcohol proved patient outcomes and decreased hospital charg- or drugs. They monopolize more hospital resources, ines.50 These findings indicate the need to focus on cur higher hospital charges, and because nonhelmeted changing practice to provide Guideline-compliant care motorcyclists frequently do not have insurance, reim- and improve patient outcomes. bursement in this group of patients is poor. So the bur- In 2004 the BrainTraumaFoundation announced the den of caring for these patients is transmitted to society National Quality Improvement TBI Initiative, aimed at as a whole.59 increasing the quality of care provided to severe TBI patients. The Initiative combines the use of an Internet- Nutrition based database to collect real-time patient data, Web A prospective validation of the Canadian clinical prac- conferencing, and quality improvement strategies to help tice guidelines for nutrition support in mechanically trauma centers analyze their care processes and make im- ventilated, critically ill adult patients was reported from provements where necessary (www.braintrauma.org). 59 ICUs across Canada.60 Intensive care units that were The recent comprehensive Cochrane Database Sys- more consistent with the Canadian clinical practice tematic Review reviewed 20 trials with 12,303 random- guidelines were more likely to successfully feed patients ized participants in the study of corticosteroids for acute by enteral nutrition. This study documented that adop- TBI. This group concluded that steroids should no tion of the clinical practice guidelines should lead to Vol. 200, No. 5, May 2005 Napolitano What’s New in Trauma and Critical Care 763 improved nutrition support practice in ICUs and may tion increased during the postorder form period with- translate into better outcomes for critically ill patients out evidence of significantly hastening death. receiving nutrition support. Another prospective cohort study by the Canadian A common practice in ICU patients is to provide a Critical Care TrialsGroup examined physician estimates combination of enteral and parenteral nutrition when of ICU survival and patient outcomes. Daily assessment full caloric goal requirements are unable to be achieved of intensivist and bedside ICU nurse estimates of prob- by enteral nutrition alone. A systematic review of five ability of ICU survival were measured in 851 consecu- randomized prospective trials that compared combina- tive mechanically ventilated adult patients. ICU mortal- tion nutrition to enteral nutrition alone confirmed no ity rate was 35.7%, and 341 patients (40.1%) were difference between the groups in rates of mortality, in- assessed by a physician at least once to have a Ͻ 10% fectious complications, length of hospital stay, or venti- ICU survival probability. Independent predictors of lator days.61 So in critically ill patients who are not mal- ICU mortality were baseline APACHE II score and daily nourished and have an intact gastrointestinal tract, the factors such as multiple organ dysfunction syndrome, use of parenteral nutrition in combination with enteral use of inotropes or vasopressors, dialysis, patient prefer- nutrition provides no benefit in clinical outcomes over ence to limit life support, and physician but not nurse Ͻ enteral nutrition alone. prediction of 10% survival. This study documented that physician estimates of ICU survival Ͻ 10% were associated with subsequent life support limitation and End-of-life issues more powerfully predicted ICU mortality than illness There is increasing concern over the appropriateness and severity, evolving or resolving organ dysfunction, and quality of care provided in the ICU at the end of life.The use of inotropes or vasopressors.64 first review of the number of Americans who receive ICU care at the end of life was published in 2004, using hospital discharge data from 1999 for 6 states and the REFERENCES National Death Index.There were 552,157 deaths in the 1. ARDS Clinical Network. Available at: http://www.ardsnet.org/. 6 states in 1999, of which 38.3% occurred in the hospi- Accessed February 17, 2005. tal and 22.4% occurred after ICU admission. Using 2. Resuscitation Outcomes Consortium. Available at: https:// roc.uwctc.org. Accessed February 17, 2005. these data to project nationwide estimates, 540,000 peo- 3. Truog RD. Will ethical requirements bring critical care research ple die after ICU admission each year. Average length of to a halt? Intensive Care Med 2004 Nov 4; [Epub ahead of stay and costs were 12.9 days and $24,541, respectively, print]. 4. Miller FG, Silverman HJ. The ethical relevance of the standard for terminal ICU hospitalizations and 8.9 days and of care in the design of clinical trials. Am J Respir Crit Care Med $8,548, respectively, for non-ICU terminal hospitaliza- 2004;169:562–564. tions.This study documented that one in five Americans 5. Luce JM, Cook DJ, Martin TR, et al, American Thoracic Soci- ety.The ethical conduct of clinical research involving critically ill dies using ICU services. The authors concluded that the patients in the United States and Canada: principles and recom- doubling of persons older than 65 by 2030 will require a mendations. Am J Respir Crit Care Med 2004;170:1375–1384. system-wide expansion in ICU care for dying patients 6. Dorman T, Angood PB, Angus DC, et al, American College of unless the health-care system pursues rationing, more Critical Care Medicine. Guidelines for critical care medicine training and continuing medical education. Crit Care Med effective advanced care planning, and augmented capac- 2004;32:263–272. ity to care for dying patients in other settings.62 7. Angus DC, Kelley MA, Schmitz RJ, et al, Committee on A large proportion of deaths in the ICU are preceded Manpower for Pulmonary and Critical Care Societies (COMPACCS). Caring for the critically ill patient. Current and by withdrawal of life support. One important publica- projected workforce requirements for care of the critically ill and tion in this area documented the usefulness of a stan- patients with pulmonary disease: can we meet the requirements dardized order form for the withdrawal of life support in of an aging population? JAMA 2000;284:2762–2770. 63 8. Irwin RS, Marcus L, Lever A. The critical care professional so- the ICU. This “withdrawal of life support order form” cieties address the critical care crisis in the United States. Chest was not documented to improve quality of end-of-life 2004;125:1512–1513. care in the ICU, which may be difficult to achieve. 9. Kelley MA, Angus D, Chalfin DB, et al. The critical care crisis in the United States: a report from the profession. Chest 2004;125: The standardized order form was noted to be helpful 1514–1517. by both nurses and physicians. Medications for seda- 10. Ewart GW, Marcus L, Gaba MM, et al. The critical care medi- 764 Napolitano What’s New in Trauma and Critical Care J Am Coll Surg

cine crisis: a call for federal action: a white paper from the critical Advisory Committee. MMWR 2004 March;53 (No. RR-3): care professional societies. Chest 2004;125:1518–1521. [1–36]. Available at: http://www.cdc.gov/ncidod/hip/pneumonia/ 11. Halpern NA, Pastores SM, Greenstein RJ. Critical care medi- default.htm. Accessed February 17, 2005. cine in the United States 1985–2000: an analysis of bed num- 28. Niederman MS, Craven DE, and the ATS and IDSA Committee bers, use, and costs. Crit Care Med 2004;32:1254–1259. members. Guidelines for the management of adults with hospital- 12. ICU Physician Staffing. Available at: http://www.leapfroggroup. acquired, ventilator-associated, and healthcare-associated pneumo- org/for_hospitals/leapfrog_safety_practices/icu_physician_staffing. nia. Am J Respir Crit Care Med 2005;171:388–416. Accessed February 17, 2005. 29. Chastre J, Wolff M, Fagon JY, et al, PneumA TrialGroup. Com- 13. Pronovost PJ, Needham DM, Waters H, et al. Intensive care parison of 8 vs 15 days of antibiotic therapy for ventilator- unit physician staffing: financial modeling of the Leapfrog stan- associated pneumonia in adults: a randomized trial. JAMA dard. Crit Care Med 2004;32:1247–1253. 2003;290:2588–2598. 14. Dellinger RP,Carlet JM, Masur H, et al, Surviving Sepsis Cam- 30. Berenholtz SM, Pronovost PJ, Lipsett PA, et al. Eliminating paign Management Guidelines Committee. Surviving Sepsis catheter-related bloodstream infections in the intensive care Campaign guidelines for management of severe sepsis and septic unit. Crit Care Med 2004;32:2014–2020. shock. Crit Care Med 2004;32:858–873. 31. Jeremitsky E, Omert LA, Dunham CM, et al. The impact of 15. Levy MM, Pronovost PJ, Dellinger RP,et al. Sepsis change bun- hyperglycemia on patients with severe brain injury. J Trauma dles: converting guidelines into meaningful change in behavior 2005;58:47–50. and clinical outcome. Crit Care Med 2004;32[Suppl 11]:S595– 32. Van den Berghe G. How does blood glucose control with insulin 597. save lives in intensive care? J Clin Invest 2004;114:1187–1195. 16. Nathens AB, Curtis JR, Beale RJ, et al. Management of the 33. Drew BJ, Califf RM, Funk M, et al, American Heart Associa- critically ill patient with severe acute pancreatitis. Crit Care Med tion; Councils on Cardiovascular Nursing, Clinical Cardiology, 2004;32:2524–2536. and Cardiovascular Disease in the Young; International Society 17. Finfer S, Bellomo R, Boyce N, et al, SAFE Study Investigators. A of Computerized Electrocardiology; American Association of comparison of albumin and saline for fluid resuscitation in the Critical-Care Nurses. Practice standards for electrocardio- intensive care unit. N Engl J Med 2004;350:2247–2256. graphic monitoring in hospital settings: an American Heart As- 18. Cooper DJ, Myles PS, McDermott FT, et al, HTS Study Inves- sociation scientific statement from the Councils on Cardiovas- tigators. Prehospital hypertonic saline resuscitation of patients cular Nursing, Clinical Cardiology, and Cardiovascular Disease with hypotension and severe traumatic brain injury: a random- in the Young: endorsed by the International Society of Comput- ized controlled trial. JAMA 2004;29:1350–1357. erized Electrocardiology and the American Association of 19. Attuwaybi B, Kozar RA, Gates KS, et al. Hypertonic saline Critical-Care Nurses. Circulation 2004;110:2721–2746. prevents inflammation, injury, and impaired intestinal transit 34. Sandham JD, Hull RD, Brant RF, et al, Canadian Critical Care after gut ischemia/reperfusion by inducing heme oxygenase 1 Clinical Trials Group. A randomized, controlled trial of the use enzyme. J Trauma 2004;56:749–758; discussion 758–759. of pulmonary-artery catheters in high-risk surgical patients. 20. Moore FA, McKinley BA, Moore EE. The next generation in N Engl J Med 2003;348:5–14. shock resuscitation. Lancet 2004;363:1988–1996. 35. Schlienger RG, Kraenzlin ME, Jick SS, Meier CR. Use of beta- 21. Keenan SP,Sinuff T, Cook DJ, Hill NS. Does noninvasive pos- blockers and risk of fractures. JAMA 2004;292:1326–1332. itive pressure ventilation improve outcome in acute hypoxemic 36. Napolitano LM, Corwin HL. Efficacy of red blood cell transfu- respiratory failure? A systematic review. Crit Care Med 2004;32: sion in the critically ill. Crit Care Clin 2004;20:255–268. 2516–2523. 37. Hebert PC, McDonald BJ, Tinmouth A. Clinical consequences 22. Rumbak MJ, Newton M, Truncale T, et al. A prospective, ran- of anemia and red cell transfusion in the critically ill. Crit Care domized, study comparing early percutaneous dilational trache- Clin 2004;20:225–235. otomy to prolonged translaryngeal intubation (delayed trache- 38. McIntyre L, Hebert PC, Wells G, et al, Canadian Critical Care otomy) in critically ill medical patients. Crit Care Med 2004; Trials Group. Is a restrictive transfusion strategy safe for resusci- 32:1689–1694. tated and critically ill trauma patients? J Trauma 2004;57:563– 23. Ely EW, Shintani A, Truman B, et al. Delirium as a predictor of 568; discussion 568. mortality in mechanically ventilated patients in the intensive 39. Zimmerman JL. Use of blood products in sepsis: an evidence- care unit. JAMA 2004;291:1753–1762. based review. Crit Care Med 2004;32[Suppl 11]:S542–547. 24. The Acute Respiratory Distress Syndrome Network. Ventilation 40. Napolitano LM. Current status of blood component therapy with lower tidal volumes as compared with traditional tidal vol- in surgical critical care. Curr Opin Crit Care 2004;10:311– umes for acute lung injury and the acute respiratory distress 317. syndrome. N Engl J Med 2000;342:1301–1308. 41. Kim HW, Greenburg AG. Artificial oxygen carriers as red blood 25. Brower RG, Lanken PN, MacIntyre N, et al, National Heart, cell substitutes: a selected review and current status. Artif Organs Lung, and Blood Institute ARDS Clinical Trials Network. 2004;28:813–828. Higher versus lower positive end-expiratory pressures in patients 42. Goodnough LT, Lublin DM, Zhang L, et al. Transfusion med- with the acute respiratory distress syndrome. N Engl J Med icine service policies for recombinant factor VIIa administra- 2004;351:327–336. tion. Transfusion 2004;44:1325–1331. 26. Spragg RG, Lewis JF, Walmrath HD, et al. Effect of recombi- 43. Friederich PW, Henny CP,Messelink EJ, et al. Effect of recom- nant surfactant protein C-based surfactant on the acute respira- binant activated factor VII on perioperative blood loss in pa- tory distress syndrome. N Engl J Med 2004;351:884–892. tients undergoing retropubic prostatectomy: a double-blind 27. Centers for Disease Control and Prevention. Guidelines for pre- placebo-controlled randomised trial. Lancet 2003;361:201– venting health–care–associated pneumonia, 2003: recommen- 205. dations of CDC and the Healthcare Infection Control Practices 44. Boffard KD, Warren B, Iau P.Decreased transfusion utilization Vol. 200, No. 5, May 2005 Napolitano What’s New in Trauma and Critical Care 765

and improved outcome associated with the use of recombinant rebral perfusion pressure, and survival. J Trauma 2004;56: factor VIIa as an adjunct in trauma [abstract]. J Trauma 482–489; discussion 489–491. 2004;57:451. 55. Deogaonkar A, Gupta R, DeGeorgia M, et al. Bispectral index 45. Pusateri AE, Delgado AV, Dick EJ Jr, et al. Application of a monitoring correlates with sedation scales in brain-injured pa- granular mineral-based hemostatic agent (QuikClot) to reduce tients. Crit Care Med 2004;32:2403–2406. blood loss after grade V liver injury in swine. J Trauma 2004;57: 56. Stiefel MF, Heuer GG, Smith MJ, et al. Cerebral oxygenation 555–562; discussion 562. following decompressive hemicraniectomy for the treatment of 46. Alam HB, Chen Z, Jaskille A, et al. Application of a zeolite refractory intracranial hypertension. J Neurosurg 2004;101: hemostatic agent achieves 100% survival in a lethal model of 241–247. complex groin injury in swine. J Trauma 2004;56:974–983. 57. Millham FH, LaMorte WW. Factors associated with mortality 47. Schwaitzberg SD, Chan MW, Cole DJ, et al. Comparison of in trauma: re-evaluation of the TRISS method using the Na- poly-N-acetyl glucosamine with commercially available topical tional Trauma Data Bank. J Trauma 2004;56:1090–1096. hemostats for achieving hemostasis in coagulopathic models of 58. Santaniello JM, Luchette FA, Esposito TJ, et al. Ten year expe- splenic hemorrhage. J Trauma 2004;57[Suppl 1]:S29–32. rience of burn, trauma, and combined burn/trauma injuries 48. Najjar SF, Healey NA, Healey CM, et al. Evaluation of poly-N- comparing outcomes. J Trauma 2004;57:696–701. acetyl glucosamine as a hemostatic agent in patients undergoing 59. Hundley JC, Kilgo PD, Miller PR, et al. Non-helmeted motor- cardiac catheterization: a double-blind, randomized study. cyclists: a burden to society? A study using the national trauma J Trauma 2004;57[Suppl 1]:S38–41. data bank. J Trauma 2004;57:944–949. 49. Hesdorffer DC, Ghajar J, Iacono L. Predictors of compliance 60. Heyland DK, Dhaliwal R, Day A, et al. Validation of the Cana- with the evidence-based guidelines for traumatic brain injury dian clinical practice guidelines for nutrition support in me- care: A survey of United States trauma centers. J Trauma 2002; chanically ventilated, critically ill adult patients: results of a pro- 52:1202–1209. 50. Fakhry SM, Trask AL, Waller MA, et al, IRTC Neurotrauma spective observational study. Crit Care Med 2004;32:2260– Task Force. Management of brain-injured patients by an 2266. evidence-based medicine protocol improves outcomes and de- 61. Dhaliwal R, Jurewitsch B, Harrietha D, Heyland DK. Combi- creases hospital charges. J Trauma2004;56:492–499; discussion nation enteral and parenteral nutrition in critically ill patients: 499–500. harmful or beneficial? A systematic review of the evidence. In- 51. Alderson P,Roberts I. Corticosteroids for acute traumatic brain tensive Care Med 2004;30:1666–1671. injury. Cochrane Database Syst Rev 2005;1:CD000196. 62. Angus DC, Barnato AE, Linde-Zwirble WT,et al, Robert Wood 52. Gracias VH, Guillamondegui OD, Stiefel MF, et al. Cerebral Johnson Foundation ICU End-Of-Life Peer Group. Use of in- cortical oxygenation: a pilot study. J Trauma 2004;56:469–472; tensive care at the end of life in the United States: an epidemi- discussion 472–474. ologic study. Crit Care Med 2004;32:638–643. 53. Edouard AR, Vanhille E, Le Moigno S, et al. Non-invasive as- 63. Treece PD, Engelberg RA, Crowley L, et al. Evaluation of a sessment of cerebral perfusion pressure in brain injured patients standardized order form for the withdrawal of life support in the with moderate intracranial hypertension. Br J Anaesth 2005;94: intensive care unit. Crit Care Med 2004;32:1141–1148. 216–221. 64. Rocker G, Cook D, Sjokvist P,et al, Level of Care Study Inves- 54. Dunham CM, Ransom KJ, Flowers LL, et al. Cerebral hypoxia tigators; Canadian Critical Care Trials Group. Clinician predic- in severely brain-injured patients is associated with admission tions of intensive care unit mortality. Crit Care Med 2004;32: Glasgow Coma Scale score, computed tomographic severity, ce- 1149–1154. SYMPOSIUM

Introduction: Symposium on Surgery in the Elderly Patient, Part 2

James C Thompson, MD, FACS Galveston, TX

The symposium, “Surgery in the Elderly Patient” was spon- cluded in this issue provide critical reviews of common sored by The University of Texas Medical, the Sealy Center perioperative problems in elderly surgical patients. The on Aging, and the Department of Surgery, Galveston TX, article by Amador and Goodwin7, “Postoperative Delir- January 2004. ium in the Older Patient,” calls attention to the difficul- Our society is aging, and with luck and gifts from sci- ties in diagnosis, treatment, and prevention of this seri- ence, the trend will continue. Evidence of the impressive ous dilemma, with special emphasis on risk factors, the demographic shifts that result from this amazing increase in role of drugs, and proper supportive treatment. lifespan is everywhere around us1-4 inter alia; one clear result Grecula and Caban,8 in their contribution, “Com- is that more and more elderly patients will undergo surgical mon Orthopaedic Problems in the Elderly Patient,” call operations. In recognition of these changes, the American attention to, among other items, important aging College of Surgeons and the American Geriatric Society changes in nonosseous tissues. They summarize an ex- appointed a joint ad hoc committee to consider problems tensive list of previous studies and give clear examples of of special import in the care of elderly surgical patients, the the management of specific skeletal injuries. solution of which would greatly improve chances of suc- Effects of age-related physiologic changes on anes- cess. These problems include: thetic risk are critically evaluated by Prough9 in “Anes- 1. Preoperative evaluation of medical contraindications to thetic Pitfalls in the Elderly Patient.” Of special note is operation his discussion of the role of anesthetic outcomes to age- 2. Preoperative evaluation of risks and benefits of proposed related changes in the brain and postoperative cognitive operation dysfunction. 3. Preoperative documentation of patient’s desire for inten- Another meeting on geriatric surgery is planned for sity of postoperative care April 2005. 4. Appropriate anesthesia 5. Postoperative pain management 6. Postoperative delirium REFERENCES 7. Postoperative atelectasis (and pneumonia) 8. Postoperative deconditioning (loss of function) 1. US Department of Commerce and HHS Administration on Ag- ing, census data. Profile of Older Americans. 2002; Retrieved Additional results of this liaison between the Ameri- 2003 from http://www.aoa.gov. can College of Surgeons and the American Geriatric 2. Loran DB, Zwischenberger JB. Thoracic surgery in the elderly. J Am Coll Surg 2004;199:773–784. Society are that the College: 3. Etzioni DA, Liu JH, Maggard MA, Ko CY. The aging population 1. Will be represented on the American Geriatric Society and its impact on the surgery workforce. Ann Surg 2003;238: 170–177. council on geriatric surgery; 4. Thompson JC. Introduction. J Am Coll Surg 2004;199:760–761. 2. Will have formal programs on surgery in elderly patients at 5. Lawrence VA, Hazuda HP, Cornell JE, et al. Recovery of func- the Annual Clinical Congress; and tional independence following major abdominal surgery in el- 3. Has appointed a Task Force on Geriatric Surgery that will ders. J Am Coll Surg 2004;199:763–772. provide a venue for educational material designed to bring 6. Rosenthal RA. Nutritional concerns in the older surgical patient. J Am Coll Surg 2004;199:785–791. current knowledge and the results of new research to the 7. Amador LF, Goodwin JS. Postoperative delirium in the older practicing surgeon and to the surgical resident. patient. J Am Coll Surg 2005;200:767–773. A symposium on surgery in the elderly patient was 8. Grecula MJ, Caban ME. Common orthopaedic problems in the elderly patient. J Am Coll Surg 2005;200:774–783. held in January 2004, the first three papers of which 9. Prough DS. Anesthetic pitfalls in the elderly patient. J Am Coll have been published.2,5,6 The three contributions in- Surg 2005;200:784–794.

Luis F Amador, MD, James S Goodwin, MD

Postoperative delirium, an acute disorder of cognition Incidence/prevalence and attention after an operation, is among the most Approximately one-quarter of adults aged 65 and older common and potentially devastating psychiatric com- experience delirium during hospitalization.10 There is a plications of hospitalized postoperative elderly patients. wide range of estimates of postoperative delirium, de- Although possible at any time in the perioperative pe- pending on type of surgical procedure.11-14 For example, riod, delirium most commonly occurs during the post- delirium is estimated to occur in 35% of patients after surgical period and complicates hospital stays for more vascular operation11 and in 40% to 60% of older pa- than 2 million older people every year.1 tients after operation for hip fracture.12,13 In contrast, Studies show a pattern of underdiagnosis of this delirium after cataract operation occurs in Ͻ5% of older condition.2-4 A survey by Ely and colleagues of more adults.14 than 912 healthcare professionals, including 753 physi- Rate of delirium is likely to increase in the foreseeable cians, found that delirium was considered an underdiag- future. One reason for this is that older adults have dispro- nosed condition by 78%.5 Despite this finding, only portionably more operations than younger adults. An esti- 40% reported that they routinely screen for delirium mated 12% of adults between ages 45 and 60 have opera- and less than half of those who routinely screen indi- tions each year, compared with 21% of those aged 65 and cated using a specific tool for assessment. This survey older.13 Another reason is that the elderly population is found incongruence between the perceived significance growing; by 2030 it is projected that 1 of 5 Americans will of delirium as a serious medical problem and current be 65 years of age or older, bringing the estimated number practices of assessing for this condition. close to 62 million.15,16 This growth will be more concen- In the postoperative setting, delirium has been as- trated in those older than 85 years of age because projec- sociated with poor cognitive and functional recovery, tions indicate that this group will climb to 9.6 million by longer hospital stays, and greater hospital costs.6,7 In the year 2030, up from 2.8 million in 1990 and 4.2 million addition, delirium in the elderly is a risk factor for in 2000.16,17 This increase in the older population will be institutionalization and morbidity.8,9 Fortunately, partially a result of the number of baby boomers becoming risks of postoperative delirium can be reduced and older adults and the increase in life expectancy in old recovery from delirium can be enhanced by early age.18,19 identification, assessment, and treatment. This article will review postoperative delirium with emphasis on Risk Factors management. Although evidence is scarce on treat- The likelihood of developing delirium increases as ment of delirium in postoperative older adults, several the number of existing risk factors increases.20,21 Table 1 clinical trials about prevention and treatment of this provides risk factors associated with postoperative condition will be included. delirium.22-28 One of the greatest predisposing risk factors for developing postoperative delirium is the presence of Competing interests declared: None. neurodegenerative diseases such as Alzheimer’s or Parkin- Dr Amador’s work is supported by the Bureau of Health Professions’ Geriat- son’s disease. Precipitating risk factors are those associated ric Academic Career Award 1 K01 HP 00056-01. with the perioperative period, such as low postoperative Received May 12, 2004; Revised August 19, 2004; Accepted August 19, 2004. oxygen saturation or pain. Postoperative pain in the elderly From the Department of Internal Medicine (Amador, Goodwin), Sealy Cen- and many analgesic medications appear to increase inci- ter on Aging (Amador, Goodwin), and Department of Preventive Medicine 28 and Community Health (Goodwin), University of Texas Medical Branch, dence of delirium. Lynch and colleagues, in a prospective Galveston, TX. observational study, found that higher pain scores at rest Correspondence address: Luis F Amador, MD, Department of Internal Med- icine, 3.236 Jennie Sealy Hospital, 301 University Blvd, Galveston, TX during the first 3 postoperative days were associated with an 77555-0460 increased risk of delirium. Interestingly, the route of anes-

© 2005 by the American College of Surgeons ISSN 1072-7515/05/$30.00 Published by Elsevier Inc. 767 doi:10.1016/j.jamcollsurg.2004.08.031 768 Amador and Goodwin Postoperative Delirium in the Older Patient J Am Coll Surg

Table 1. Risk Factors Associated with Postoperative Delirium stances in cerebrospinal fluid.38 It should not be unex- Risk factors References pected that most measures of CNS function are abnor- Older age 22–25 mal in delirium because delirium represents global brain Cognitive impairment 23–25 dysfunction. None of these abnormalities provides con- Functional impairment 23–25 vincing evidence for an “underlying mechanism” for de- Decreased postoperative hemoglobin 23, 26, 27 lirium, because they all could be secondary changes not Markedly abnormal sodium, potassium, or glucose 23–25 directly related to its pathogenesis. In addition, it is not Alcohol abuse 24, 25 at all clear that the syndrome of delirium, with its diverse Noncardiac thoracic operation 25 precipitants and manifestations, will be shown to have a History of delirium 23 common mechanistic pathway. Preoperative use of narcotic 23 Preoperative use of benzodiazepine 23 Drug Toxicity Low postoperative oxygen saturation 22 Drug toxicity accounts for approximately 30% of all History of cerebrovascular disease 26 cases of delirium and plays a major role in postoperative Untreated pain 28 delirium.39 Table 2 illustrates common drugs associated with delirium.36,39,40 Drugs that affect neurotransmitter thesia (general versus regional) does not appear to affect risk functions can cause delirium, especially drugs that have of postoperative delirium.29 anticholinergic activity.39,41 Pain management in the elderly should avoid meperidine, which has been associ- Pathogenesis ated with postoperative delirium.42 Medications should Why are older adults more susceptible to delirium? As the be started at a lower dose and advanced slowly and still body ages, multiple physiologic changes occur that affect provide adequate relief. the body’s ability to respond to stress.30 Central nervous system changes with age include loss of nerve cells, de- Diagnosis creases in cerebral blood flow, and changes in neurotrans- Diagnosis of postoperative delirium can be challenging. mitter systems, such as decreased acetylcholinesterase activ- First, because clinical presentations vary, other condi- ity, carbonic anhydrase activity, muscarinic receptors, and tions such as dementia, schizophrenia, and depression, serotonin receptors.31 It should be noted that the elderly which often have similar presentations, must be ex- population is quite heterogeneous; individuals of the same cluded. This distinction is complicated by the fact that chronological age may have very different comorbid conditions and abilities to respond to stress. Table 2. Drugs Associated with Delirium Although important advances in understanding delir- Drugs with anticholinergic activity ium have been made in the last few decades, the biologic Tricyclic antidepressants basis is not clearly understood. In delirium, the electroen- Cimetidine cephalogram typically shows slowing of the dominant pos- Corticosteroids terior alpha rhythm and the appearance of abnormal slow- Digoxin wave activity.32 An exception to this finding is when Diphenhydramine Belladonna delirium accompanies alcohol withdrawal, in which fast- Dipyridamole wave activity predominates on the EEG. Theophylline In delirium, multiple brain systems are disrupted si- Promethazine multaneously. Several studies have shown a decrease in Amantadine 33,34 acetylcholine function. This decrease is also seen in Oxybutynin individuals with Alzheimer’s dementia.35 In addition to Warfarin the cholinergic system, various investigators have found Analgesics changes in other neurotransmitters, hormones, and cy- Narcotics (especially meperidine) tokines in patients with delirium.36,37 For example, a Nonsteroidal antiinflammatory agents study of patients with delirium revealed abnormal levels Benzodiazepines of endorphins, serotonin, neuropeptides, and other sub- Antiparkinsonian agents Vol. 200, No. 5, May 2005 Amador and Goodwin Postoperative Delirium in the Older Patient 769

Table 3. Modiﬁed from Diagnostic and Statistical Manual of Mental Disorders, 4th ed. (Text Revision) Criteria for Diagnosis of Delirium A. Disturbance of consciousness with reduced ability to focus, sustain, or shift attention. B. A change in cognition or the development of a perceptual disturbance that is not better accounted for by a preexisting, established, or evolving dementia. C. The disturbance develops over a short period of time (usually hours to days) and tends to fluctuate during the course of the day. D. There is evidence from the history, physical examination, or laboratory findings that the disturbance is caused by a general medical condition, substance intoxication, substance withdrawal, or multiple etiologies. E. Delirium not otherwise specified (NOS) should be used to diagnose delirium that meets criteria for A, B, or C but not criterion D. Adapted from the American Psychiatric Association’s Diagnostic and Statistical Manual of Mental Disorders, 4th ed., text revision criteria for delirium.43 underlying dementia is a major risk factor for delirium, types: hypoactive delirium with low psychomotor behav- and often present in patients with postoperative delir- ioral activity, hyperactive delirium with high psychomotor ium. Dementia is probably the most common diagnosis activity, mixed delirium that has features of hypoactive and confused with postoperative delirium in the older adult. hyperactive psychomotor activity, and delirium without Dementia generally progresses slowly over many years psychomotor behavioral changes. In hyperactive delirium, and does not alter consciousness, as does postoperative patients may present with psychosis, agitation, and com- delirium. Careful attention to key features of postoper- bativeness, pulling out lines or Foley catheters. Hypoactive ative delirium—acute onset, fluctuating course, altered delirium is just as common as hyperactive delirium, but is consciousness, and disorganized thought—will help dis- recognized less frequently because of a decreased level of tinguish it from dementia. These conditions can be dis- arousal.38,44 tinguished by referring to the American Psychiatric As- sociation’s Diagnostic and Statistical Manual of Mental Physical examination Disorders,4th ed., text revision criteria for delir- Physical examination is often difficult in the confused ium.43 Table 3 illustrates the criteria for diagnosis of and uncooperative patient. Physicians should check vital delirium. signs, oxygen saturation, signs of trauma or infection, state of hydration, and new neurologic findings. The History Confusion Assessment Method (CAM) is a tool to help A good history will often distinguish postoperative delir- clinicians identify when delirium is probable. Table 4 ium from other conditions. If behavioral or cognitive dete- illustrates the CAM algorithm, which in medical and rioration occurs in the postoperative period, delirium surgical settings has a sensitivity of 94% to 100% and a should be suspected. History should focus on careful de- specificity of 90% to 95%.45 scriptions of the patient’s behavior, such as agitation, appar- Postoperative delirium is a frequent problem in inten- ent hallucinations, and disorientation. Nurses and family sive care units. The CAM-ICU has been specifically de- members are usually good sources for descriptions of be- veloped for identification of delirium in the intensive havior. Patients themselves can often give clear descriptions care unit and can be used on individuals who cannot of delirium, especially if they are warned preoperatively communicate verbally or use a mechanical ventilator.46 about the potential of experiencing delirium and educated In addition to history, physical examination, and diag- about its manifestations. Additional information on history of any earlier episodes of delirium, evidence of cogni- Table 4. Confusion Assessment Method tive impairment before hospitalization, and information to 1. Acute onset and fluctuating course, mental status changes from rule out alcohol or drug withdrawal as a cause is also hours to days 2. Difficulty in focusing (easily distracted or unable to follow in an helpful. interview) 3. Disorganized thinking (rambling or irrelevant conversation) Clinical presentation 4. Altered level of consciousness (from hyperalert to unable to Clinical presentation includes changes in alertness, diffi- arouse) culty staying focused, disorientation, disorganized think- A positive Confusion Assessment Method test for delirium requires the presence of acute onset and inattention (items 1 and 2), and either disorganized ing, and changes in sleep cycle. Delirium can be classified thinking or altered level of consciousness (items 3 or 4). Adapted from Inouye on the basis of psychomotor activity into four different and colleagues.45 770 Amador and Goodwin Postoperative Delirium in the Older Patient J Am Coll Surg nostic tests, instruments such as the CAM and the defined areas of risk for postoperative delirium. Consul- CAM-ICU are particularly helpful when incorporated tation reduced the incidence of delirium by one-third into routine assessments of patients with cognitive or (32 versus 50%, p ϭ 0.04), suggesting that proactive behavioral changes. These tests can be administered by a interventions may play an important role in preventing nurse practitioner, registered nurse, or physician assis- postoperative delirium. tant who frequently monitors postoperative patients and These studies show that identifying and reducing risk communicates with the physician. Diagnosis of postop- factors can decrease but by no means eliminate postoperative delirium is based on clinical observation, rather erative delirium in the elderly. A simple but effective than laboratory or radiology tests. History and physical method to reduce morbidity from postoperative delir- examination will indicate need for further diagnostic ium that is used by experienced surgeons and anesthesi- tests. ologists is to preoperatively educate the patient about its manifestations and to explain that the condition is self- Diagnostic tests limited. Much of the behavioral agitation associated Diagnostic tests are used not to establish the diagnosis of with delirium may result from the extreme anxiety felt delirium, but to identify potentially correctable factors by individuals experiencing acute global brain dysfunc- that might be contributing to delirium. Complete blood tion. Earlier warning and education, and continuous counts, serum electrolytes, creatinine, glucose, and uri- assurances, can help reduce that anxiety. nalysis are reasonable tests for most patients.36 Retro- spective studies suggest that neuroimaging may be used Management selectively for patients with delirium.47-49 Neuroimaging Management of postoperative delirium is multidimen- is probably not necessary if the initial clinical evaluation sional and spans three areas simultaneously: treating the reveals an obvious cause and there are no signs of trauma contributing illness(es), providing supportive measures, or new focal neurologic deficits. and, if needed, introducing symptom control. It is important to consider postoperative delirium as a medical Prevention emergency that merits prompt care,40 because it can be Evidence has shown that implementation of protocols precipitated by a serious or life-threatening condition targeting modifiable risk factors prevent some cases of (eg, pulmonary embolus). The practitioner needs to delirium in medical and surgical patients.50-53 A study by closely evaluate, treat, and monitor individuals with Inouye and colleagues51 used standardized protocols of postoperative delirium. six known risk factors for delirium in 852 hospitalized First, delirium should be identified and the cause(s) older adults admitted to general medical services. The treated. Frequently, more than one disorder contributes to study used interventions for sleep deprivation; immobil- delirium. An example is a 75-year-old man with chronic ity; dehydration; and visual, hearing, and cognitive im- kidney disease and postoperative delirium after an uncom- pairments. These strategies resulted in a marked reduc- plicated elective knee operation, recently started on leva- tion in the number of delirium episodes, from 9.9% for floxacin 500 mg daily for a urinary tract infection. In this the intervention group compared with 15% for the patient, the urinary tract infection and inappropriate dos- usual-care group, although there was no effect on delir- ing of levofloxacin for the renal function are two factors ium severity or rate of recurrence. precipitating the delirium. In addition, there might be mild A study conducted perioperatively on patients with preexisting cognitive impairment. femoral neck fracture showed that postoperative delir- The second area of management is supportive care. ium decreased from 61% to 48% after a program con- Supportive care provides a safe environment for the in- sisting of pre- and postoperative geriatric assessments, dividual and staff, and also includes orientation, appro- oxygen therapy, early operation, prevention and treat- priate stimulation, and nutrition.36,54 Providing an ap- ment of perioperative blood pressure falls, and treatment propriate environment is vital and can be achieved by of postoperative complications.52 Marcantonio and col- removing unnecessary objects, maintaining appropriate leagues53 used a model involving geriatric consultation room temperature and lighting, and avoiding excessive in surgical patients with hip fracture. During daily visits, noise.55 Arranging for family members to be near the the geriatrician targeted recommendations based on pre- bedside often lessens patients’ agitation. Table 5 illus- Vol. 200, No. 5, May 2005 Amador and Goodwin Postoperative Delirium in the Older Patient 771

Table 5. Supportive Measures include extrapyramidal symptoms, hypotension, akathisia, Provide adequate space and remove unnecessary objects and neuroleptic malignant syndrome. Maintain adequate lighting and room temperature Benzodiazepines have a rapid onset of action but can Avoid excessive noise and unnecessary interruption of sleep, worsen sedation and cause paradoxical agitation. Loraz- especially during night-time hours Ensure use of glasses, dentures, and hearing aids if needed epam, 0.5 to 2 mg, is preferred over diazepam and cloraz- Maintain activity level and encourage participation of self-care epate because it has a simpler metabolism and shorter half- 40 Promote orientation by using clocks, calendars, and familiar life. Benzodiazepines are the drug of choice to treat objects withdrawal from alcohol or sedative drugs and are useful Maintain appropriate nutrition and hydration adjuncts to antipsychotic medications.54 When antipsy- Minimize immobility by avoiding use of physical or chemical chotic medications and benzodiazepines are used together, restraints or unnecessary urinary catheter doses should be decreased because the drugs work synergistically. trates supportive care interventions.36,54 Multiple disci- plines such as nursing, social work, and physical therapy Outcomes of delirium can contribute to supportive care by promoting orienta- Sequels of delirium can persist for 6 months or longer in tion, safety, comfort, and reassurance.56 many individuals.60-65 In a study of patients who had The third area of management is symptom control. It operations for hip fracture, delirium persisted at the time should be reserved for agitated or disruptive individuals of hospital discharge in 39% of those who developed to ensure safety and facilitate diagnosis and treatment. postoperative delirium.60 In addition, 29% and 6% had Symptom control should never replace or delay treat- symptoms at 1 month and 6 months after hospitaliza- ment of the precipitating cause(s) of delirium. Although tion, respectively. managing disruptive behavior is the most challenging Individuals with delirium are at risk for future cognitive aspect of delirium therapy, less than one-third of older decline. Rockwood and colleagues61 examined the relation- individuals with delirium exhibit aggressive or agitated ship between an episode of delirium and subsequent de- behavior.57 mentia over 3 years with 203 patients aged 65 years and older admitted to medical services. They found the inci- Medication for symptom control dence of dementia was 5.6% per year over 3 years for those Antipsychotic medications and benzodiazepines are the without delirium and 18.1% per year for those with delir- two main classes of medications that have been studied ium. In another study of 78 elderly patients with femoral for symptom control. Haloperidol has advantages over neck fractures, dementia developed within 5 years in 69% other antipsychotic medications because it has less anti- of patients with postoperative delirium versus only 20% of cholinergic activity as compared with older antipsy- patients without postoperative delirium.62 One possible ex- chotic medications and has multiple routes of adminis- planation is that patients who develop postoperative delir- tration (oral, IM, or IV).54 Newer antipsychotic agents ium might have preoperative undiagnosed cognitive im- such as risperidone and olanzapine appear to have simi- pairment or “subclinical dementia.” Another explanation is lar efficacy and have fewer extrapyramidal side effects that delirium causes pathologic changes that promote than haloperidol.58,59 These newer agents can be used dementia. effectively in many patients with postoperative delirium, Delirium is also associated with increased mortality. although there are not any clinical trials to compare ef- Hospital mortality in patients with delirium has been ficacy on the management of symptoms of postoperative estimated between 10% and 65%.63 This is significantly delirium in the older adult. higher than patients without delirium and can be com- Lower doses, 0.5 mg, of haloperidol can be effective in pared with patients with perioperative myocardial in- the older adult. Further, older adults rarely require Ͼ5mg farction. Mortality after hospitalization of individuals per day. The goal is to control disruptive symptoms and who had delirium is higher than for those who did not avoid obtundation, which is associated with its own set of have delirium during hospitalization. A metaanalysis by complications. After symptoms are controlled, a mainte- Cole and Primeau found that mortality rates were 14% nance dose should be started and divided during the day. at 1 month and 22% at 6 months after hospital discharge Haloperidol can be tapered in 3 to 5 days. Side effects for individuals with delirium.9 Older adults with delir- 772 Amador and Goodwin Postoperative Delirium in the Older Patient J Am Coll Surg ium experience longer hospitalization stays and higher 11. Schneider F, Bohner H, Habel U, et al. Risk factors for postop- nursing home placement than those without delirium. erative delirium in vascular surgery. Gen Hosp Psychiatry 2002; 24:28–34. In a longitudinal study of patients admitted to general 12. Marcantonio E, TaT, Duthie E, Resnick NM. Delirium severity medical wards it was found that after 2 years only one- and psychomotor types: their relationship with outcomes after third of individuals who had experienced delirium still hip fracture repair. J Am Geriatr Soc 2002;50:850–857. 7 13. Ergina PL, Gold SL, Meakins JL. Perioperative care of the el- lived independently in the community. The practi- derly patient. World J Surg 1993;17:192–198. tioner should be aware of these negative outcomes when 14. Milstein A, Pollack A, Kleinman G, Barak Y. Confusion/ managing older adults with this condition. delirium following cataract surgery: an incidence study of 1-year duration. Int Psychogeriatr 2002;14:301–306. In conclusion, understanding delirium can improve 15. Hobbs FB, Damon BL. Special studies, 65ϩ in the United its prevention, diagnosis, and management. Risk of States. Current population reports. Series P23. No. 190. postoperative delirium can be reduced with careful at- Washington, DC: Bureau of the Census; 1996. tention to risk factors such as avoiding medications with 16. US Census Bureau, 2004. US interim projections by age, sex, race, and Hispanic origin. Available at: http://www.census.gov/ anticholinergic activity and careful pain management. ipc/www/usinterimproj/. Accessed March 18, 2004. Also, interventions to target problems like sleep depri- 17. US Census Bureau, 2001. Resident population, according to age, sex, race, and Hispanic origin: United States, selected years vation; dehydration; immobility; and impairments in Ϫ 50 1950 2001. Available at: http://www.census.gov/pubinfo/ cognition, vision, and hearing are useful. Systematic www/newhispmk1.html. Accessed March 18, 2004. approaches to the diagnostic workup, including history, 18. Trends and projections. Washington, DC: US Senate Special examination, screening tests, and laboratory and radiol- Committee on Aging in conjunction with the American Asso- ciation of Retired Persons, the Federal Council on Aging, and ogy tests, facilitate recognition of this condition. Early the US Administration on Aging with Aging America; 1988. identification and assessment will expedite multidimen- 19. Perls TT. The oldest old. Sci Am 2004;14:6–11. sional management of postoperative delirium and facil- 20. Inouye SK, Charpentier PA. Precipitating factors for delirium in hospitalized elderly persons: predictive model and interrelation- itate implementation of appropriate care plans and dis- ship with baseline vulnerability. JAMA 1996;275:852–857. charge planning. 21. Inouye SK, Viscoli CM, Horwitz RI, et al. A predictive model for delirium in hospitalized elderly medical patients based on admission characteristics. Ann Intern Med 1993;119:474–481. REFERENCES 22. Wang SG, Lee UJ, Goh EK, Chon KM. Factors associated with postoperative delirium after major head and neck surgery. Ann 1. Rizzo JA, Bogardus ST Jr, Leo-Summers L, et al. Multicompo- Otol Rhinol Laryngol 2004;113:48–51. nent targeted intervention to prevent delirium in hospitalized 23. Litaker D, Locala J, Franco K, et al. Preoperative risk factors for older patients: what is the economic value? Med Care 2001;39: postoperative delirium. Gen Hosp Psychiatry 2001;23:84–89. 740–752. 24. Weed HG, Lutman CV, Young DC, Schuller DE. Preoperative 2. Milisen K, Foreman MD, Wouters B, et al. Documentation of identification of patients at risk for delirium after major head delirium in elderly patients with hip fracture. J Gerontol Nurs and neck cancer surgery. Laryngoscope 1995;105:1066–1068. 2002;28:23–29. 25. Marcantonio ER, Goldman L, Mangione CM, et al. A clinical 3. Lewis LM, Miller DK, Morley JE, et al. Unrecognized delirium prediction rule for delirium after elective noncardiac surgery. in ED geriatric patients. Am J Emerg Med 1995;13:142–145. JAMA 1994;271:134–139. 4. Laurila JV,Pitkala KH, Strandberg TE, Tilvis RS. Detection and 26. Bucerius J, Gummert JF, Borger MA, et al. Predictors of delir- documentation of dementia and delirium in acute geriatric ium after cardiac surgery delirium: effect of beating-heart (off- wards. Gen Hosp Psychiatry 2004;26:31–35. pump) surgery. J Thorac Cardiovasc Surg 2004;127:57–64. 5. Ely EW, Stephens RK, Jackson JC, et al. Current opinions re- 27. Marcantonio ER, Goldman L, Orav EJ, et al. The association of garding the importance, diagnosis, and management of delirium intraoperative factors with the development of postoperative in the intensive care unit: a survey of 912 healthcare profession- delirium. Am J Med 1998;105:380–384. als. Crit Care Med 2004;32:106–112. 28. Lynch EP,Lazor MA, Gellis JE, et al. The impact of postopera- 6. Thomas R, Cameron D, Fahs M. A prospective study of delir- tive pain on the development of postoperative delirium. Anesth ium and prolonged hospital stay. Arch Gen Psychol 1988;45: Analg 1998;86:781–785. 937–940. 29. Williams-Russo P, Sharrock NE, Mattis S, et al. Cognitive ef- 7. Francis J, Kapoor W. Prognosis after hospital discharge of older fects after epidural vs general anesthesia in older adults: a ran- medical patients with delirium. J Am Geriatr Soc 1992;40:601– domized trial. JAMA 1995;274:44–50. 606. 30. Resnick NM, Marcantonio ER. How should clinical care of the 8. Inouye SK, Rushing JT, Foreman MD, et al. Does delirium aged differ? Lancet 1997;350:1157–118. contribute to poor hospital outcomes? A three-site epidemio- 31. Aging and the nervous system. In: Boers MH, Berkow R, eds. logic study. J Gen Intern Med 1998;13:234–242. The Merck Manual of Geriatrics, 3rd ed. New Jersey: Merck 9. Cole MG, Primeau FJ. Prognosis of delirium in elderly hospital Research Laboratories, 2000:380Ϫ383. patients. CMAJ 1993;149:41–46. 32. Romano J, Engel GL. Delirium. I. EEG data. Arch Neurol 10. Brown TM, Boyle MF. Delirium. BMJ 2002;325:644–647. Psychiatr 1944;51:356–377. Vol. 200, No. 5, May 2005 Amador and Goodwin Postoperative Delirium in the Older Patient 773

33. Tune L, Carr S, Hoag E, Cooper T. Anticholinergic effects of 49. Benbadis SR, Sila CA, Cristea RL. Mental status changes and drugs commonly prescribed for the elderly: potential means for stroke. J Gen Intern Med 1994;9:485–487. assessing risk of delirium. Am J Psychiatry 1993;149:1393– 50. Cole MG, Primeau F, McCusker J. Effectiveness of interven- 1394. tions to prevent delirium in hospitalized patients: a systematic 34. Mach JR, Dysken MW, Kuskowski M, et al. Serum anticholin- review. CMAJ 1996;155:1263–1268. ergic activity in hospitalized older persons with delirium: a pre- 51. Inouye SK, Bogardus ST, Charpentier PA, et al. A multicompo- liminary study. J Am Geriatr Soc 1995;43:491–495. nent intervention to prevent delirium in hospitalized older pa- 35. Eikelenboom P,Hoogendijk WJ, Jonker C, van Tilburg W. Im- tients. N Engl J Med 1999;340:669–676. munological mechanisms and the spectrum of psychiatric syn- 52. Gustafson Y, Brannstrom B, Berggren D, et al. A geriatric- dromes in Alzheimer’s disease. J Psychiatr Res 2002;36:269– anesthesiologic program to reduce acute confusional states in 280. elderly patients treated for femoral neck fractures. J Am Geriatr 36. Rummans TA, Evans JM, Krahn LE, Fleming KC. Delirium in Soc 1991;39:655–62. elderly patients: evaluation and management. Mayo Clin Proc 53. Marcantonio ER, Flacker JM, Wright RJ, et al. Reducing delir- 1995;70:989–998. ium after hip fracture: a randomized trial. J Am Geriatr Soc 37. Marshall MC, Soucy MD. Delirium in the intensive care unit. 2001;49:516–522. Crit Care Nurs Q 2003;26:172–178. 54. Meagher DJ. Delirium: Optimising management. BMJ 2001; 38. Trzepacz PT.The neuropathogenesis of delirium: a need to focus 322:144–149. our research. Psychosomatics 1994;35:374–391. 55. American Psychiatric Association. Practical guidelines for the treat- 39. Francis J. Drug-induced deliriums: diagnosis and treatment. ment of patients with delirium. Washington, DC: American Psy- CNS Drugs 1996;5:103–114. chiatric Association; 1999. 40. Burns A, Gallagley A, Byrne J. Delirium. J Neurol Neurosurg 56. Inouye SK. Delirium in hospitalized older patients: recognition Psychiatry 2004;75:362–367. and risk factors. J Geriatr Psychiatry Neurol 1999;11:118–125. 41. Feinberg M. The problems of anticholinergic adverse effects in 57. Francis J, Martin D, Kapoor WN. A prospective study of delir- older patients. Drugs Aging 1993;3:335–348. ium in hospitalized elderly. JAMA 1990;263:1097–1101. 42. Adunsky A, Levy R, Heim M, et al. Meperidine analgesia and 58. Sipahimalani A, Masand PS. Olanzapine in the treatment of delirium in aged hip fracture patients. Arch Gerontol Geriatr delirium. Psychosomatics 1998;39:422–430. 2002;35:253–259. 59. Sipahimalani A, Masand PS. Use of risperidone in delirium: case 43. American Psychiatric Association. Diagnostic and statistical reports. Ann Clin Psychiatry 1997;9:105–107. manual of mental disorders. 4th ed. Washington, DC: American 60. Marcantonio ER, Flacker JM, Michaels M, et al. Delirium is Psychiatric Association; 2000. independently associated with poor functional recovery after hip 44. Roth M. Clinical perspectives. Int Psychogeriatr 1991;3:309– fracture. J Am Geriatr Soc 2000;48:618–624. 317. 61. Rockwood K, Cosway S, Carver D, et al. The risk of dementia 45. Inouye SK, van Dyck CH, Alessi CA, et al. Clarifying confusion: and death after delirium. Age Ageing 1999;28:551–556. the Confusion Assessment Method, a new method for detection 62. Lundstrom M, Edlund A, Bucht G, et al. Dementia after delir- of delirium. Ann Intern Med 1990;113:941–948. ium in patients with femoral neck fractures. J Am Geriatr Soc 46. Ely EW, Inouye SK, Bernard GR, et al. Delirium in mechani- 2003;51:1002–1006. cally ventilated patients: validity and reliability of the confusion 63. Inouye SK. Delirium in hospitalized elderly patients: recogni- assessment method for the intensive care unit (CAM-ICU). tion, evaluation, and management. CT Med 1993;57:309–315. JAMA 2001;286:2703–2710. 64. Levkoff SE, Evans DA, Liptzin B, et al. Delirium: the occurrence 47. Koponen H, Hurri L, Stenback U, et al. Computed tomography and persistence of symptoms among elderly hospitalized pa- findings in delirium. J Nerv Ment Dis 1989;177:226–231. tients. Arch Intern Med 1992;152:334–40. 48. Naughton BJ, Moran M, Ghaly Y, Michalakes C. Computed 65. McCusker J, Cole M, Dendukuri N, et al. Delirium in older tomography scanning and delirium in elderly patients. Acad medical inpatients and subsequent cognitive and functional sta- Emerg Med 1997;4:1107–1110. tus: a prospective study. CMAJ 2001;165:575–583. Common Orthopaedic Problems in the Elderly Patient

Michael J Grecula, MD, Mabel E Caban, MD

The aging process causes significant changes in the neu- fixation of fractures, prosthetic replacement of in- romuscular and skeletal systems. The neurologic system jured or arthritic joints, repair of torn tendons and is affected by cerebral cortical atrophy, decline in neuro- ligaments, and decompression or fusion of the spine, transmitter levels, decreased cerebral blood flow, and de- have provided treatment options to achieve these creased nerve conduction velocities. Visual, hearing, and goals. Newer approaches to minimize the surgical im- vestibular functions also diminish with age. Aging skel- pact on the patient may also prove to be of benefit in etal muscle experiences a decrease in lean muscle mass, decreasing the rehabilitation time after surgical degeneration of the neuromuscular junction, and 30% interven- to 60% decrease in strength. The overall result of these tions. changes is a 20% increase in reaction time and a 35% to Thirty-five million people in the United States are 1 40% increase in falls in adults over 60 years of age. Bone currently restricted by a musculoskeletal disorder. As mass and strength decrease with age, especially in the baby-boom generation ages, it is estimated that the women, resulting in an increase in fractures after falls. number of people over 65 years of age will double over Tendons, ligaments, and joint capsules also lose signifi- the next 50 years, and this will significantly increase the 2 cant tensile strength and are more likely to sustain injury number of patients in need of musculoskeletal care. It is or degeneration. Articular cartilage changes include de- obvious that diagnosis and treatment of musculoskeletal creased chondrocyte function, and changes in matrix conditions in the older patient have and will continue to have a major impact on our health-care system. A good composition, water content, and physical properties, understanding of these conditions is important for any but the relationship between these changes and develop- health-care provider who has contact with patients in ment of arthritis are unclear. The intervertebral disc also this age category. Much effort and research should be experiences changes in cell function, matrix composi- directed into diagnosis, treatment, and prevention of tion, water content, and ability to distribute forces re- these disorders. The goal of this article is to review the sulting in degenerative spine disease. age-related changes in neuromuscular and skeletal sys- Common orthopaedic problems in the aging pa- tems, the disorders that result from these changes, and tient include acute fractures, muscle and tendon tears, the methods to treat and prevent common orthopedic joint arthritis, and spinal stenosis. Each of these dis- disorders. orders can result in significant dysfunction for the patient and loss of mobility and independence. Treat- Neurologic system ment goals are to return patients to their maximal The central nervous system suffers significant changes level of function and independence in the shortest during the aging process. It has been shown that the period of time and subject them to the least amount of cerebral cortex atrophies and experiences a 20% weight risks. Use of surgical interventions, including internal loss between the ages of 45 and 85. There is a decrease in neurotransmitter levels, decrease in cerebral blood flow, Competing interests declared: None. and decrease in nerve conduction velocity of 10% to Supported in part by grant R24HS111618 (JL Freeman) of the Agency for 15%. The visual, auditory, vestibular, and propriocep- Health Care Research and Quality. tive functions also become impaired.3 Received July 7, 2004; Revised November 24, 2004; Accepted December 8, 2004. From the Department of Orthopaedics and Rehabilitation, University of Texas Muscular system Medical Branch, Galveston, TX. Neural control of the muscles occurs through neuro- Correspondence address: Michael J Grecula, MD, Department of Orthopae- dics and Rehabilitation, University of Texas Medical Branch, 301 University muscular junctions. Motor units are formed as a single Blvd, Rt 0476, Galveston, TX 77555. motor neuron innervates a group of skeletal muscle fi-

© 2005 by the American College of Surgeons ISSN 1072-7515/05/$30.00 Published by Elsevier Inc. 774 doi:10.1016/j.jamcollsurg.2004.12.003 Vol. 200, No. 5, May 2005 Grecula and Caban Common Orthopaedic Problems in the Elderly Patient 775 bers. The aging process is associated with degeneration ficial fibrillation and thinning. On a cellular level, there of the neuromuscular junction and a decrease in total is a decrease in chondrocyte density, activity, and re- number of motor units. Motor unit loss is more specific sponse to growth factors. The extracellular matrix to the fast motor units. Some of the fast motor units are changes by showing a decrease in the water content and felt to undergo reinnervation by slow motor units and a decrease in the proteoglycan size. Molecular size be- increase the total number of slow motor units. Both type comes more variable; chondroitin sulfate content de- I and type II muscle fibers decrease in size and number, creases and keratin sulfate content increases. The num- especially the type II, fast twitch fibers. These changes ber of degraded molecules increases in concentration result in an overall decrease in muscle strength by 30% to and there is evidence of increased collagen cross-linking. 3,4 60% between the ages of 30 and 80. Mechanically, this results in a decrease in tensile strength, decrease in fatigue resistance, and increase in Skeletal system rigidity of the cartilage.2,3 Humans reach their maximum bone mass by age 25 to 30. Bone mass declines over the next several decades and this decline is faster and usually more pronounced in Intervertebral disc women. By the age of 80, women, on average, have lost The intervertebral disc undergoes some of the most dra- 50% of their cancellous mass and 30% of their cortical matic age-related changes in the musculoskeletal sys- mass. Bone loss is accelerated during the first 10 years tem.2,8 The central region of the disc (inner annulus after menopause. Men lose, on average, 30% of their fibrosis and nucleus pulposus) experiences the most ex- cortical and cancellous bone mass by the age of 80.5 tensive changes. By early adult life, the clear gelatinous Bone has the remarkable ability to regenerate. As a per- nucleus pulposus seen in children changes to a firm fi- son ages, there is a decline in this ability. Animal studies brous plate with fissures and cracks extending from the have demonstrated a three times increase in healing time periphery to the central regions. The notochordal cells in older animals when compared with younger ones.6 seen in infants disappear and chondrocyte-like cells take The underlying cause is multifactorial, including a de- their place. The number of chondrocyte-like cells then crease in osteoprogenitor cell number, a decrease in the declines sharply after skeletal maturity. inflammatory response to injury, and a decrease in the Ninety to 95% of the intervertebral disc is composed expression of osteoinductive cytokines and growth fac- of water, collagen, and proteoglycans, and each of these tors. Growth hormone is shown to decrease with age and are also affected by the aging process.9 Water content this affects the inductive potential of bone matrix, which decreases, resulting in a decreased ability to distribute 3 also decreases with age. stress. The mean collagen fibril diameter increases and variability of the collagen fibril diameters also increases. Connective tissues There is a decrease in proteoglycan concentration and a Connective tissue vascular perfusion and nutrition de- decrease in the agrecan size in the central portion of the crease with age. Cell function, matrix composition, and disc. There also appears to be an accumulation of de- organization change during the aging process, although graded matrix molecules and noncollagenous proteins. these changes have not been studied extensively.2 As fibroblasts age, they appear to flatten and elongate. Phys- Nutrition to the disc and cells occurs through diffu- iologic studies on older tendon tissues also show a sion of nutrients from blood vessels on the periphery of decrease in aerobic glycolysis and biosynthetic activity.3 the annulus fibrosus and within the vertebral bodies. Mechanical studies of the ligament-bone complex from Age-related changes, including a decline in number of individuals over 60 years of age showed the structure arteries, accumulation of degraded matrix macromole- failed at one-third the load needed for failure in a cules, and decrease in water content, may all interfere younger ligament.7 with this process. Mechanically, the intervertebral disc becomes less resilient with decreased ability to dissipate Articular cartilage energy.10 Disc shape and volume also change, which can As articular cartilage ages, it changes from a white, glossy affect spine mobility, alignment, and loads applied to the appearance to a yellow-brown tinge with localized super- facet joints, spinal ligaments, and paraspinous muscles.9 776 Grecula and Caban Common Orthopaedic Problems in the Elderly Patient J Am Coll Surg

Outcomes of age-related changes Table 1. Ten Most Common Fractures, Shown in Percent- ages, of Medicare Beneﬁciaries above 65 Years Old, Be- Physiologic and structural changes that occur during the tween July 1991 and June 1992 aging process result in an increase in certain diseases and Fracture site % injuries. Neurologic changes result in a 20% increase in Hip 38 reaction time, decreased sensory function, visual acuity, Wrist 19 vestibular function, proprioception, and balance. These Proximal humerus 11 changes contribute to the reported 35% to 40% increase Ankle 9 in falls in people older than 60 years of age. Loss of Pelvis 6 muscle strength and reflexes limits the ability to reduce Radius/ulna 4 the force generated from a fall, and incidence of injury Tibia/fibula 4 Distal humerus 3 from a fall increases. In fact, falls are the leading cause of Femur 3 fatal injuries in the United States population over 70 Patella 3 years of age.3 Excluded were incomplete information, bone cancer, previous fractures, and The decline in bone mass and strength combined vertebral fractures.11 with increased incidence of falls in the elderly results in an increased number of fractures. About 850,000 frac- rupture to be 6.4% in patients aged 60 to 70 and 20.4% 20 tures occurred annually in the United States in people 65 in patients aged 80 years or older. years or older,11 resulting in $14 billion in health-care Although a direct relationship has not been found expenditures in 1995.12 More than half of all women will between the aging changes in articular cartilage and de- sustain an osteoporotic fracture in their lifetime.12-14 velopment of osteoarthritis, incidence of arthritis has Women present with hip fractures more often than men, also been shown to increase with age. The incidence of osteoarthritis of the hip is greater in women than men, and the increase in incidence of fractures occurs earlier in but the gender ratio is nearly even in osteoarthritis of the women; about 10 years after menopause. Increased in- knee. One study projected that as many as 500,000 new cidence in fractures in men occurs after age 70.15,16 Data symptomatic cases of idiopathic osteoarthritis of the hip obtained from the Medicare Beneficiary database from and knee may arise annually in the Caucasian popula- 1991 to 1992 reported the 10 most common areas of tion of the United States.21 fracture, as illustrated in Table 1.11 Hip fractures repre- Structural and mechanical changes in the interverte- sented the most common area of fracture in the elderly bral disc result in degeneration of the disc, as evidenced population and account for nearly 60% of all hospital 17 by a loss of disc height, bulging of the annulus, disc admissions for fracture care in the United States. herniation, and osteophytes. As the disc degenerates, it Degeneration and decrease in strength of the tendon affects spine mobility, alignment, and load distribution and ligaments results in tears either spontaneously or to the surrounding structures. Additional changes in- after minor trauma. These injuries are seen most com- clude facet joint arthritis, capsular and ligament thick- monly in the rotator cuff, biceps tendon, quadriceps ening, or instability with or without displacement of the tendon insertion to the patella, Achilles tendon, and adjacent vertebral bodies.9 Lumbar disc degeneration posterior tibial tendon. The incidence of these injuries is can be observed as early as the second decade of life in difficult to ascertain because their occurrence is often men and the third decade in women. By age 49, 97% of insidious or the diagnosis is delayed or not reported at lumbar discs examined in autopsy specimens have evi- all. Using the rotator cuff as an example, a study of 200 dence of macroscopic changes in the nucleus pulposus.22 unselected elderly patients with an average age of 78 As degeneration progresses, spinal stenosis may result as years revealed clinical findings consistent with a rotator the bulging annulus, hypertrophied ligaments and facet cuff tear in 7% of subjects.18 Anatomic dissections of joints, osteophytes, or displacement of the vertebral cadavers over 60 years of age at death, found the inci- bodies can narrow the spinal canal and neural foramen. dence of full-thickness tears to be 30%.19 Increased in- This may result in cauda equina or nerve root impinge- cidence of tendon injuries with aging is demonstrated by ment and symptoms of spinal stenosis.9 a large population-based study in the United Kingdom, Back pain and stiffness are among the most common which identified the odds ratio of an Achilles tendon complaints of the elderly patient23 and rank as the third Vol. 200, No. 5, May 2005 Grecula and Caban Common Orthopaedic Problems in the Elderly Patient 777 leading cause of chronic illness in women over 65 years of age and the fourth leading cause of chronic illness in men over 65 years of age.24 Radiographic incidence of lumbar spinal stenosis has been reported to be 1.7% to 10% when large series of myelograms have been reviewed.25,26 These findings have been shown to increase with age,27,28 and most studies indicate a male prepon- derance and the average age of onset in the fifth or sixth decade.29-31 The correlation between structural changes seen on x-ray or MRI does not correlate well with the symptoms. The absolute incidence of symptomatic spinal stenosis in the general population remains to be quantified. Data provided by the National Ambulatory Medical Care Survey estimate incidence of spinal stenosis as 3.9% of all visits made for mechanical low back pain.32 Together, the increase results in falls, fractures, tendon and ligament injuries, arthritis, and spinal stenosis result in significant loss of mobility and independence for the older person. Because of slower healing time, duration of disability is further increased after the injury or surgical intervention. Patients then become less involved in society, less productive, and often need home assistance or placement in longterm care facilities. As the population ages, this will become an increasing burden on the health-care system and society in general. Focus should Figure 1. Anterior/posterior radiograph of a subcapital, intracapsu- be directed on improving the treatment, outcomes, and lar femoral neck fracture displaced into varus. prevention of these problems. rate to 15%.33 It becomes obvious that surgical treat- Treatment ment of these fractures has become the standard of care, Advances in surgical care for these disorders have signif- except for the rare patient whose medical comorbidities icantly improved treatment and outcomes of many of preclude a surgical procedure. these orthopaedic problems. A historical review of the Current treatment options for fractures of the proxi- treatment of hip fractures, one of the most common mal femur are tailored to fracture type. Fracture patterns problems in the elderly patient, eloquently demonstrates of the proximal femur can be divided into two distinctly both the significant advances that have been made and different types. The more proximal fractures occur at the the continued obstacles to achieving full recovery in all base of the femoral head and are intracapsular (Fig. 1). patients. Before availability of surgical intervention, This fracture pattern is prone to nonunion and disrup- fractures of the proximal femur were treated either with tion of the femoral head blood supply. The more distal bed rest and traction or placement of a hip spica cast. A fractures occur through the intertrochanteric region, review of treatment methods for femoral neck fractures which is extracapsular (Fig. 2). The bone in this region is at one institution from 1930 to 1952 illustrates the sig- more vascular but also tends to be more osteoporotic. nificant advance of surgical treatment. When treated These fractures have a higher union rate but are more with closed reduction and spica cast application, the predisposed to shortening and residual deformity. nonunion rate for these fractures was 76% and avascular Surgical treatment options for intracapsular femoral necrosis of the femoral head occurred in 78%. When neck fractures include closed reduction and internal fix- treated with closed reduction and internal fixation, the ation, replacement of the femoral head with a hemiar- nonunion rate dropped to 20% and avascular necrosis throplasty, or replacement of both the femoral head and 778 Grecula and Caban Common Orthopaedic Problems in the Elderly Patient J Am Coll Surg

For the extracapsular intertrochanteric fractures, internal fixation is the most common method of treatment (Fig. 5) and advances in the treatment have been in the design of implants to minimize the occurrence of hardware failure, fixation failure, shortening, and deformity. Initial fixation devices were rigid and did not allow for compression of the soft cancellous bone during the healing process. This resulted in either hardware breakage, failure of fixation in the bone, or penetration of the hardware through the femoral head. A major advance was seen with the development of a sliding device that allowed controlled compression across the fracture site. Failure rate was reduced from 54% to 9% with the use of this sliding hip screw.35 In addition to hardware development, advances have also been made in defin- ing proper surgical technique. The ideal position of the screw within the head has been identified to be within 5 mm of the subchondral bone in the center of the head on both the anterior/posterior and lateral x-ray projections. When placed in this position authors have found the screw cutout rate decreased to 0% compared to a 25% to 58% incidence of cutout when not placed within this position.36,37 Shortening through the fracture site still remains a problem and averages 15.7 mm in unstable fracture pat- Figure 2. Anterior/posterior radiograph of an intertrochanteric fracture displaced into varus. acetabulum with a total hip replacement. Each treatment has its potential benefits and risks. Use of closed reduction and internal fixation (Fig. 3) requires a less- invasive surgical approach and if union occurs without the occurrence of avascular necrosis, the result can be excellent and enduring. Use of a hemiarthroplasty (Fig. 4) or total hip replacement removes the risks of nonunion and avascular necrosis but adds the problems of prosthetic replacements, including dislocation, loosening, or need for removal secondary to deep sepsis. In a prospective randomized clinical trial of these three methods, the 13-year followup results showed overall similar mortality rates with all three treatment methods. Pain relief was best in the total hip replacement group. The need for a revision operation was only 7% in the total hip replacement group, compared with 33% and 24% in the internal fixation group and hemiarthroplasty group, respectively. Early complication rate was higher in the total hip replacement group which experienced a 20% dislocation rate.34 Other considerations also include increased sur- Figure 3. Anterior/posterior radiograph of a subcapital, intracapsu- gical time and increased costs of the implants associated lar femoral neck fracture treated with internal ﬁxation with cannu- with the total hip replacement treatment. lated cancellous screws placed percutaneously. Vol. 200, No. 5, May 2005 Grecula and Caban Common Orthopaedic Problems in the Elderly Patient 779

patients were walking at their preinjury ability and only 17% of patients reached their preinjury level of function for activities of daily living.45 Review of surgical treatment advances for other common problems in the elderly patient has similarly improved outcomes for the patient but there also can be a significant loss of function and prolonged rehabilitation. Fractures of the distal radius and proximal humerus are the most common upper extremity fractures in the elderly.11 Nonoperative care of these fractures results in less disability than in hip fractures, as patients usually remain ambulatory. Loss of joint mobility, decreased muscle strength, residual deformity, and overall functional loss of the injured extremity are common. Surgical options of external fixation or internal fixation of the distal radius fractures and internal fixation or prosthetic replacement of the proximal humerus fractures have become more common. Complete tears of the rotator cuff of the shoulder or Achilles tendon at the ankle can be treated nonoperatively but usually require prolonged re-

Figure 4. Anterior/posterior radiograph of a cemented unipolar hemiarthroplasty used to treat an intracapsular femoral neck fracture. terns.38 This results in increased pain and decreased mobility for the patient. Other devices have been developed in an attempt to reduce this problem but superior results have yet to be demonstrated.39-44 Despite these advances, proximal femoral fractures are still associated with significant morbidity and mortality. Care of these patients is often associated with preexisting malnutrition, postoperative delirium, decubitus ulcers, cardiopulmonary disease, urinary tract infections, postoperative infections, and thromboembolic disease. One study showed an average of three complications per patient, of which 26% were severe.45 Mortality rates at 1 year postoperatively have been reported to be 13% to 24%.46,47 Asig- nificant proportion of surviving patients experience pro- Figure 5. Anterior/posterior radiograph of a sliding hip screw and longed disability. At 4 months after injury, only 43% of side plate used to treat an intertrochanteric hip fracture. 780 Grecula and Caban Common Orthopaedic Problems in the Elderly Patient J Am Coll Surg habilitation and the patient often will not regain full tial of perioperative morbidity or mortality, which is strength and mobility. Surgical repair is aimed at im- reported to increase with age in some series.61,62 proving functional recovery from these injuries.48,49 One of the most successful and cost effective of all Acute fractures usually result in an abrupt change in surgical advances has been the ability to replace arthritic the individual’s functional status, but degenerative joints, especially the hip and knee.63-67 Benefits of total changes of the spine and joints usually result in a slow joint replacement include relief of pain and improved progressive functional loss. Spinal stenosis typically pre- mobility; continued independence of the individual and sents as an insidious onset of low back pain and stiffness longterm success rates have been reported to be as high with an aching, cramping type of pain extending into as 90% to 95%.68-70 Review of Medicare claims data the buttocks and posterior thigh. Classic symptoms of show the highest rate of joint replacement to be in the neurogenic claudication also include pain, burning, 75-to-79-year-old age group.71 The age group with the numbness, or paresthesias radiating into the posterior or highest increase in joint replacements between 1994 and posterolateral aspect of the lower limb. Pain is exacer- 2000 was the over-85-year-old group.72 Although joint bated by activity, especially standing with the trunk ex- replacement has been shown to be successful and cost tended. Relief may be obtained with rest or simply bend- effective in this age group also, risk of major complica- ing forward, such as leaning on a shopping cart. tions has been reported to be 3.5 times higher.73-75 Frequently, neurologic examination is normal.9 Because Limitations still exist in surgical treatment of ortho- osteoarthritis of the hip is also fairly common in the paedic problems in the elderly patient. Osteoporotic elderly and may coexist in as many as 33% of patients fractures can be difficult to treat because of the limited with spinal stenosis,50,51 the cause of the patient’s pain area of good-quality bone for fixation and slower healing may be confusing. Osteoarthritic hip pain usually pre- time of the fractures. Work continues on development of sents as an aching or sharp mechanical type of pain in the less invasive, minimally invasive, or arthroscopic ap- groin or lateral hip with extension into the anterior thigh proaches to minimize soft tissue injury from surgical or knee. Radiographic evaluation can be helpful in es- intervention and on ways to improve fixation in osteo- tablishing the diagnosis but severity of the changes often porotic bone using new devices or bone substitutes. De- does not correlate with clinical symptoms. generative tendon or ligament tears can be difficult to Decisions about the appropriate interventions for de- repair because of large defects or the degenerative nature generative diseases are usually based on amount of pain, of the surrounding tissues, and advances have been made functional limitations, and quality-of-life issues. Severe in new materials for tissue repair and tissue substitutes. neurologic deterioration from spinal stenosis is uncom- Surgical treatment of degenerative disorders of the spine mon and symptoms may stabilize despite progressive or joints exposes the patient to the risks of the operation changes on imaging studies. Johnsson and colleagues52 and postoperative complications. Work continues on studied the natural history of spinal stenosis and found improving techniques and materials to minimize the im- at mean followup of 49 months of observation, 15% of pact of the operation and reduce complication rates. In the patients showed improvement, 70% showed no the future, the ability to genetically replicate tissue will change, and 15% worsened. Nonoperative treatment provide additional opportunities for surgical care of the options include NSAIDS, analgesics, exercise programs, injured or degenerative musculoskeletal system. bracing, traction, or epidural steroid injections, and reported rates of improvement with these treatments vary Prevention from 15% to 43%.52-54 The most common surgical in- It becomes obvious that treatment of orthopaedic prob- tervention for symptomatic spinal stenosis is decom- lems in the elderly patient continues to be a significant pressive laminectomy with reported success rates of challenge, with life-threatening or major functional loss 70% to 90% in the elderly patient.55-58 Fusion, with or consequences. The best method of treatment is to pre- without instrumentation, may be added if there is sig- vent problems from occurring and major emphasis nificant deformity or instability. A few studies have com- should be placed on preventive measures. A routine ex- pared nonoperative and operative treatment and sup- ercise program has been shown to reduce the effects of port operative treatment as providing better results.59,60 aging on the skeletal muscle, soft tissues, and bone. This Surgical treatment does expose the patient to the poten- maintains strength, protective reflexes, coordination, Vol. 200, No. 5, May 2005 Grecula and Caban Common Orthopaedic Problems in the Elderly Patient 781 and balance, and reduces incidence of falls and 7. Woo SL, Hollis JM, Adams DJ, et al. Tensile properties of the fractures.4,76-79 Osteoporosis prevention also includes human femur-anterior cruciate ligament-tibia complex. The effects of specimen age and orientation. Am J Sports Med 1991; proper diet with sufficient calcium (1,500 mg/day) and 19:217–225. vitamin D (400 IU/day).80 Recent recommendations on 8. Buckwalter JA. Aging and degeneration of the human interver- estrogen and estrogen with progesterone for postmeno- tebral disc. Spine 1995;20:1307–1314. 9. Garfin SR, Herkowitz HN, Mirkovic S. Spinal stenosis. Instr pausal women have impacted the prevention and treat- Course Lect 2000;49:361–374. ment of osteoporosis.81 Controversy aroused from data 10. Iatridis JC, Setton LA, Weidenbaum M, Mow VC. Alter- published by the Women’s Health Initiative Steering ations in the mechanical behavior of the human lumbar nucleus pulposus with degeneration and aging. J Orthop Res Committee demonstrating the increased risk of stroke in 1997;15:318–322. women taking estrogen. There is a protective effect 11. Centers for Disease Control and Prevention. Incidence and costs against suffering hip fractures, vertebral fractures, and to Medicare of fractures among Medicare beneficiaries aged (greater or equal) 65 years—United States, July 1991–June overall fractures while using hormonal therapy that 1992. Morb Mortal Wkly Rep 1996;45:877–883. 81-83 ceases quickly after discontinuation of its use. The 12. Feldstein AC, Nichols GA, Elmer PJ, et al. Older women with US FDA endorses lower doses of estrogen, for the short- fractures: patients falling through the cracks of guideline- est duration and, particularly for the treatment of osteo- recommended osteoporosis screening and treatment. J Bone Joint Surg 2003;85A:2294–2302. porosis, to consider taking other nonestrogen therapies. 13. Barrett JA, Baron JA, Karagas MR, Beach ML. Fracture risk in the Biphosphonates for women with low bone mineral den- U.S. Medicare population. J Clin Epidemiol 1999;52:243–249. sity at the hip and older females with osteoporotic frac- 14. Ross PD. Osteoporosis. Frequency, consequences, and risk factors. Arch Intern Med 1996;156:1399–1411. tures and osteopenia have been effective by reducing 15. Davenport MG, Adelakun A, McCormick C. Osteoporosis and bone resorption.12,84-86 Patients should also be counseled hip fractures in the Medicare population, 1992–1996. Health on the detrimental effects of excessive alcohol intake, Care Financ Rev 1999;21:123–128. 80 16. Kenny AM, Joseph C, Taxel P,Prestwood KM. Osteoporosis in smoking, and caffeine use on bone metabolism. older men and women. Conn Med 2003;67:481–486. In summary, aging has been shown to significantly 17. Hobar CE, Carbone L,TylavskyF.Medicare fracture data for the affect the musculoskeletal system. These changes lead to state of Tennessee, 1994–1998. Tenn Med 2002;95:231–235. an increase in number of falls, fractures, and degenera- 18. Fuchs S, Chylarecki C, Langenbrinck A. Incidence and symptoms of clinically manifest rotator cuff lesions. Int J Sports Med tive changes of the joints, tendons, ligaments, and spine. 1999;20:201–205. Resultant fractures, ligament or tendon tears, pain, or 19. Lehman C, Cuomo F, Kummer FJ, Zuckerman JD. The inci- nerve compression can significantly affect mobility and dence of full thickness rotator cuff tears in a large cadaveric population. Bull Hosp Jt Dis 1995;54:30–31. independence of the individual. Surgical advances have 20. van der Linden PD, Sturkenboom MC, Herings RM, et al. significantly improved treatment outcomes of these Increased risk of achilles tendon rupture with quinolone anti- problems but still hold significant risks. The primary bacterial use, especially in elderly patients taking oral corticosteroids. Arch Intern Med 2003;163:1801–1807. goal in treating the elderly patient should be directed to 21. Wilson MG, Michet CJ, Ilstrup DM, Melton LJ. Idiopathic the education and prevention of these disorders. symptomatic osteoarthritis of the hip and knee: a population- based incidence study. Mayo Clin Proc 1990;65:1214–1221. 22. Miller JA, Schmatz C, Schultz AB. Lumbar disc degeneration: REFERENCES correlation with age, sex, and spine level in 600 autopsy specimens. Spine 1988;13:173–178. 1. United States Bone and Joint Decade Web site. Available at: 23. Bressler HB, Keyes WJ, Rochon PA, Badley E.The prevalence of http://www.usbjd.org. low back pain in the elderly. A systematic review of the literature. 2. Buckwalter JA, Heckman JD, Petrie DP.An AOA critical issue: Spine 1999;24:1813–1819. aging of the North American population: new challenges for 24. Goel V, Iron K, Williams JI. Indicators of health determinants orthopaedics. J Bone Joint Surg 2003;85A:748–758. and health status. In: Goel V, Williams JI, Anderson GM, et al, 3. Bostrom MP,Buckwalter JA.The physiology of aging. In: Koval KJ, eds. Patterns of health care in Ontario. The ICES Practice Atlas. ed. Orthopaedic knowledge update 7. Rosemont, IL: American 2nd ed. Ottawa: Canadian Medical Association; 1996:5–26. Academy of Orthopaedic Surgeons; 2002:85–93. 25. Roberson GH, Llewellyn HJ, Taveras JM. The narrow lumbar 4. Kirkendall DT,Garrett WE.The effects of aging and training on spinal canal syndrome. Radiology 1973;107:89–97. skeletal muscle. Am J Sports Med 1998;26:598–602. 26. De Villiers PD, Booysen EL. Fibrous spinal stenosis. A report on 5. Buckwalter JA, Glimcher MJ, Cooper RR, Recker R. Bone bi- 850 myelograms with a water-soluble contrast medium. Clin ology. II: formation, form, modeling, remodeling, and regula- Orthop 1976;115:140–144. tion of cell function. J Bone Joint Surg 1995;77A:1276–1289. 27. Boden SD, Davis DO, Dina TS, et al. Abnormal magnetic- 6. Bak B, Andreassen TT.The effect of aging on fracture healing in resonance scans of the lumbar spine in asymptomatic subjects. A the rat. Calcif Tissue Int 1989;45:292–297. prospective investigation. J Bone Joint Surg 1990;72A:403–408. 782 Grecula and Caban Common Orthopaedic Problems in the Elderly Patient J Am Coll Surg

28. Sasaki K. Magnetic resonance imaging findings of the lumbar massive rotator cuff tears in patients older than 70 years. J root pathway in patients over 50 years old. Eur Spine J 1995;4: Shoulder Elbow Surg 1999;8:26–30. 71–76. 49. Nestorson J, Movin T, Moller M, Karlsson J. Function after 29. MartinelliTA, Wiesel SW.Epidemiology of spinal stenosis. Instr Achilles tendon rupture in the elderly: 25 patients older than 65 Course Lect 1992;41:179–181. years followed for 3 years. Acta Orthop Scand 2000;71:64–68. 30. Verbiest H. Pathomorphologic aspects of developmental lumbar 50. Surin V, Hedelin E, Smith L. Degenerative lumbar spinal stenosis: stenosis. Orthop Clin North Am 1975;6:177–196. results of operative treatment. Acta Orthop Scand 1982;53:79–85. 31. Paine KW. Clinical features of lumbar spinal stenosis. Clin Or- 51. Swezey RL. Overdiagnosed sciatica and stenosis, underdiag- thop 1976;115:77–82. nosed hip arthritis. Orthopedics 2003;26:173–174. 32. Hart LG, Deyo RA, Cherkin DC. Physician office visits for low 52. Johnsson KE, Rosen I, Uden A. The natural course of lumbar back pain. Frequency, clinical evaluation, and treatment pat- spinal stenosis. Clin Orthop 1992;279:82–86. terns from a U.S. national survey. Spine 1995;20:11–19. 53. Simotas AC. Nonoperative treatment for lumbar spinal stenosis. 33. Cleveland M, Fielding JW. A continuing end-result study of Clin Orthop 2001;384:153–161. intracapsular fracture of the neck of the femur. J Bone Joint Surg 54. Atlas SJ, Deyo RA, Keller RB, et al. The Maine Lumbar Spine 1954;36A:1020–1030. Study, Part III. 1-year outcomes of surgical and nonsurgical 34. Ravikumar KJ, Marsh G. Internal fixation versus hemiarthro- management of lumbar spinal stenosis. Spine 1996;21:1787– plasty versus total hip arthroplasty for displaced subcapital frac- 1794. tures of femur—13 year results of a prospective randomised 55. Benz RJ, Ibrahim ZG, Afshar P, Garfin SR. Predicting compli- study. Injury 2000;31:793–797. cations in elderly patients undergoing lumbar decompression. 35. Heyse-Moore GH, MacEachern AG, Evans DC. Treatment of Clin Orthop 2001;384:116–121. intertrochanteric fractures of the femur. A comparison of the 56. Jonsson B, Stromqvist B. Lumbar spine surgery in the elderly. Richards screw-plate with the Jewett nail-plate. J Bone Joint Complications and surgical results. Spine 1994;19:1431–1435. Surg 1983;65B:262–267. 57. Katz JN, Lipson SJ, Chang LC, et al. Seven- to 10-year outcome 36. Kyle RF, Cabanela ME, Russell TA, et al. Fractures of the prox- of decompressive surgery for degenerative lumbar spinal steno- imal part of the femur. Instr Course Lect 1995;44:227–253. sis. Spine 1996;21:92–98. 37. Baumgaertner MR, Curtin SL, Lindskog DM, Keggi JM. The 58. Sanderson PL, Wood PL. Surgery for lumbar spinal stenosis in value of the tip-apex distance in predicting failure of fixation of old people. J Bone Joint Surg 1993;75B:393–397. peritrochanteric fractures of the hip. J Bone Joint Surg 1995; 59. Johnsson KE, Uden A, Rosen I. The effect of decompression on 77A:1058–1064. the natural course of spinal stenosis. A comparison of surgically 38. Jacobs RR, McClain O, Armstrong HJ. Internal fixation of in- treated and untreated patients. Spine 1991;16:615–619. tertrochanteric hip fractures: a clinical and biomechanical study. 60. Amundsen T, Weber H, Nordal HJ, et al. Lumbar spinal steno- Clin Orthop 1980;146:62–70. sis: conservative or surgical management? A prospective 10-year 39. Bridle SH, Patel AD, Bircher M, Calvert PT. Fixation of inter- study. Spine 2000;25:1424–1435. trochanteric fractures of the femur. A randomised prospective 61. Deyo RA, Cherkin DC, Loeser JD, et al. Morbidity and mor- comparison of the gamma nail and the dynamic hip screw. tality in association with operations on the lumbar spine. The J Bone Joint Surg 1991;73B:330–334. influence of age, diagnosis, and procedure. J Bone Joint Surg 40. Saudan M, Lubbeke A, Sadowski C, et al. Pertrochanteric frac- 1992;74A:536–543. tures: is there an advantage to an intramedullary nail? A random- 62. Ciol MA, Deyo RA, Howell E, Kreif S. An assessment of surgery ized, prospective study of 206 patients comparing the dynamic for spinal stenosis: time trends, geographic variations, compli- hip screw and proximal femoral nail. J Orthop Trauma 2002; cations, and reoperations. J Am Geriatr Soc 1996;44:285–290. 16:386–393. 63. Laupacis A, Bourne R, Rorabeck C, et al. The effect of elective 41. Ahrengart L, Tornkvist H, Fornander P, et al. A randomized total hip replacement on health-related quality of life. J Bone study of the compression hip screw and Gamma nail in 426 Joint Surg 1993;75A:1619–1626. fractures. Clin Orthop 2002;401:209–222. 64. Chang RW,Pellisier JM, Hazen GB. A cost-effectiveness analysis 42. Watson JT, Moed BR, Cramer KE, Karges DE. Comparison of of total hip arthroplasty for osteoarthritis of the hip. JAMA the compression hip screw with the Medoff sliding plate for 1996;275:858–865. intertrochanteric fractures. Clin Orthop 1998;348:79–86. 65. Liang MH, Cullen KE, Larson MG, et al. Cost-effectiveness of 43. Butt MS, Krikler SJ, Nafie S, Ali MS. Comparison of dynamic total joint arthroplasty in osteoarthritis. Arthritis Rheum 1986; hip screw and gamma nail: a prospective, randomized, con- 29:937–943. trolled trial. Injury 1995;26:615–618. 66. Lavernia CJ, Guzman JF,Gachupin-Garcia A. Cost effectiveness 44. Parker MJ, Pryor GA. Gamma versus DHS nailing for extracap- and quality of life in knee arthroplasty. Clin Orthop 1997;345: sular femoral fractures. Meta-analysis of ten randomised trials. 134–139. Int Orthop 1996;20:163–168. 67. Hawker G, Wright J, Coyte P,et al. Health-related quality of life 45. van Balen R, Steyerberg EW, Polder JJ, et al. Hip fracture in after knee replacement. J Bone Joint Surg 1998;80A:163–173. elderly patients: outcomes for function, quality of life, and type 68. Garellick G, Herberts P, Stromberg C, Malchau H. Long-term of residence. Clin Orthop 2001;390:232–243. results of Charnley arthroplasty. A 12–16-year follow-up study. 46. Aharonoff GB, Koval KJ, Skovron ML, Zuckerman JD. Hip J Arthroplasty 1994;9:333–340. fractures in the elderly: predictors of one year mortality. J Or- 69. Kavanagh BF, Wallrichs S, Dewitz M, et al. Charnley low- thop Trauma 1997;11:162–165. friction arthroplasty of the hip.Twenty-year results with cement. 47. Huddleston JM, Whitford KJ. Medical care of elderly patients J Arthroplasty 1994;9:229–234. with hip fractures. Mayo Clin Proc 2001;76:295–298. 70. Onsten I, Besjakov J, Carlsson AS. Improved radiographic sur- 48. Worland RL, Arredondo J, Angles F,Lopez-Jimenez F.Repair of vival of the Charnley prosthesis in rheumatoid arthritis and Vol. 200, No. 5, May 2005 Grecula and Caban Common Orthopaedic Problems in the Elderly Patient 783

osteoarthritis. Results of new versus old operative techniques in 79. Province MA, Hadley EC, Hornbrook MC, et al. The effects 402 hips. J Arthroplasty 1994;9:3–8. of exercise on falls in elderly patients. A preplanned meta- 71. Mahomed NN, Barrett JA, Katz JN, et al. Rates and outcomes of analysis of the FICSIT Trials. Frailty and Injuries: Coopera- primary and revision total hip replacement in the United States tive Studies of Intervention Techniques. JAMA 1995;273: Medicare population. J Bone Joint Surg 2003;85A:27–32. 1341–1347. 72. Bourne RB, Maloney WJ, Wright JG. An AOA critical issue. 80. NIH Consensus Development Panel on Osteoporosis Pre- The outcome of the outcomes movement. J Bone Joint Surg vention, Diagnosis, and Therapy. Osteoporosis prevention, 2004;86A:633–640. diagnosis, and therapy. JAMA 2001;285:785–795. 73. Ekelund A, Rydell N, Nilsson OS. Total hip arthroplasty in 81. Anderson GL, Limacher M, Assaf AR, et al. Effects of conju- patients 80 years of age and older. Clin Orthop 1992; gated equine estrogen in postmenopausal women with hysterec- 281:101–106. tomy: The Women’s Health Initiative randomized controlled 74. Boettcher WG. Total hip arthroplasties in the elderly. Mor- trial. JAMA 2004;291:1701–1712. bidity, mortality, and cost effectiveness. Clin Orthop 1992; 82. Banks E, Beral V, Reeves G, et al. Fracture incidence in relation 274:30–34. to the pattern of use of hormone therapy in postmenopausal 75. Pettine KA, Aamlid BC, Cabanela ME. Elective total hip arthro- women. JAMA 2004;291:2212–2220. plasty in patients older than 80 years of age. Clin Orthop 1991; 83. Hulley SB, Grady D. The WHI Estrogen-Alone Trial—do 266:127–132. things look any better? JAMA 2004;291:1769–1771. 76. Daley MJ, Spinks WL. Exercise, mobility and aging. Sports Med 84. Ettinger MP. Aging bone and osteoporosis: strategies for pre- 2000;29:1–12. venting fractures in the elderly. Arch Intern Med 2003;163: 77. Schoutens A, Laurent E, Poortmans JR. Effects of inactivity and 2237–2246. exercise on bone. Sports Med 1989;7:71–81. 85. Miller PD. Efficacy and safety of long-term bisphosphonates in 78. Wolf SL, Barnhart HX, Ellison GL, Coogler CE. The effect of postmenopausal osteoporosis. Exp Opin Pharmacother 2003;4: Tai Chi Quan and computerized balance training on postural 2253–2258. stability in older subjects. Atlanta FICSIT Group. Frailty and 86. Bone HG, Hosking D, Devogelaer JP, et al. Ten years’ experi- Injuries: Cooperative Studies on Intervention Techniques. Phys ence with alendronate for osteoporosis in postmenopausal Ther 1997;77:371–381. women. N Engl J Med 2004;350:1189–1199. Anesthetic Pitfalls in the Elderly Patient

Donald S Prough, MD

Although anesthesiology has many subspecialties, only but rather those factors that seem to relate to aging per recently have anesthesiologists begun to organize the se, including physiologic changes and pharmacologic subspecialty of geriatric anesthesiology. The need for changes that accompany aging and the emerging prob- increased emphasis on geriatric anesthesia is best illus- lem of postoperative cognitive dysfunction (POCD), trated by considering the rapid growth in the aging popu- which is a recently recognized syndrome that is more lation and the expected growth of workload in specific sur- common and more severe in elderly patients. gical subspecialties over the next 2 decades (Fig. 1).1 Perhaps the reason for the delayed emergence of geriatric PHYSIOLOGIC CHANGES anesthesiology is that most anesthesiologists routinely Up through the age of 60 years, both basal organ func- provide anesthesia for operations in elderly, frail pation and physiologic reserve (the difference between tients. In discussing anesthesia for elderly patients, I will basal and maximal organ function) are well maintained use the conventional cut-off age of 65, but acknowledge (Fig. 3).2 Subsequently, physiologic reserve diminishes. that the physiology of vigorous older patients more This description of average changes fails to distinguish be- closely resembles that of younger patients. tween aging individuals who experience slower, milder de- Both aging and the five-category Physical Status (PS) clines in functional reserve and those in whom declines are classification developed by the American Society of An- more severe, ie, the frail elderly. esthesiologists (ASA) (Fig. 2)2 are associated with in- Despite large differences in the rate of age-related decreased incidence of postoperative complications. In the cline, aging has defied a universal definition other than healthiest population of patients, those who are ASA PS chronological age. The most obvious evidence of diffuse 1 (free of systemic disease) and 2 (mild systemic disease), cellular deterioration includes the physical manifesta- the curve that relates age to major complications in- tions of aging, such as facial wrinkling, hair loss, and creases gradually until about age 70. For patients in ASA muscle atrophy. Just as these processes vary markedly PS 3, which includes less healthy patients, the incidence among individuals, so do aging processes in other organs of major complications begins to ascend at an earlier age that may influence the ability to withstand stress posed and increases more rapidly. For those in ASA PS 4, by an operation and anesthesia. At present, the most which includes patients with diseases that are a constant popular theory of aging, oxidative stress theory, states threat to life, the curve rises even more steeply. The that cumulative oxidative damage, attributable in part to strong relationship between PS, age, and outcomes dem- superoxide and hydroxide radicals, limits lifespan.3 One onstrates that increasing age is associated with both an of the most important targets of oxidative damage is increasing incidence of intercurrent diseases and with DNA, which gradually accumulates strand breaks and declining physiologic reserve. Geriatric anesthesia neces- base modifications.4 In all organ systems, progressive sitates distinction between problems that relate to age- DNA damage inevitably leads to diminished production dependent decreases in functional reserves from problems of proteins or production of abnormal proteins and pro- that result from intercurrent diseases, such as cardiovascular gressive loss of function. In the brain, for example, DNA disease. damage can impair learning, memory, and neuronal sur- The focus of this article is not intercurrent disease vival.5 Stem cells, particularly hematopoietic stem cells, gradually lose their capacity for self-regeneration, which Competing interests declared: None. decreases the ability of tissues to adapt to stress.6 On a Received June 29, 2004; Revised December 9, 2004; Accepted December 9, subcellular level, the mitochondria gradually lose func- 2004. From the Department of Anesthesiology, University of Texas Medical tion as mitochondrial DNA is damaged and critical en- Branch, Galveston, TX. zymes become altered.7 Another factor accompanying Correspondence address: Donald S Prough, MD, Department of Anesthesi- ology, University of Texas Medical Branch, 301 University Blvd, Galveston, aging is chronic low-grade inflammation, which is evi- TX 77555-0591. denced by increased levels of the cytokines, tumor ne-

© 2005 by the American College of Surgeons ISSN 1072-7515/05/$30.00 Published by Elsevier Inc. 784 doi:10.1016/j.jamcollsurg.2004.12.010 Vol. 200, No. 5, May 2005 Prough Anesthetic Pitfalls in the Elderly Patient 785

creases in heart rate; in healthy, older individuals, in- Abbreviations and Acronyms creases in cardiac output are primarily accomplished by ASA ϭ American Society of Anesthesiologists increasing end-diastolic volume and by increasing stroke ϭ BIS Bi-Spectral Index volume at a similar ejection fraction (Fig. 4).11,12 Myo- MAC ϭ minimum alveolar concentration POCD ϭ postoperative cognitive dysfunction cardial work, measured per heart beat, increases. This PS ϭ Physical Status change is accompanied by increased myocardial stiffness and increased arterial stiffness;13 even healthy older patients tolerate fluid loading less than younger patients crosis factor, and interleukin-6.8 The aging immune and are less tolerant of changes in rhythm, particularly system demonstrates diminished function,9 reportedly loss of atrial contribution to cardiac filling. Older adults more in elderly patients who have experienced severe have decreased exercise capacity because of a multifacto- emotional stress.10 Regardless of whether elderly surgical rial interaction between changes in cardiac function, vas- patients have overt evidence of organ system dysfunc- cular function, and skeletal muscle mass; some, but not tion, these processes will inevitably have reduced func- all of these changes can be delayed by systematic exercise tion and reserve. What follows are examples of changes (Fig. 5).14 in specific organ systems. Autonomic changes Cardiovascular changes With aging, the intrinsic function of both the sympa- Although management considerations are similar in thetic and parasympathetic nervous systems decreases. older patients and younger patients with cardiovascular Sympathetic nervous system activity may be compensa- disease, older patients without overt cardiovascular dis- torily increased by diminution of baroreceptor activity.15 ease still suffer an inevitable loss of functional reserve as Consequently, plasma norepinephrine levels increase aging progresses. In healthy, young individuals, increases with age and plasma norepinephrine levels correlate with in cardiac output are accomplished primarily by in- mean arterial blood pressure.15

Figure 1. Over the next 15 years, the percentage of persons 65 years of age or older will increase more rapidly than any other age group. (Reprinted from: Etzioni DA, Liu JH, Maggard MA, Ko CY. The aging population and its impact on the surgery workforce. Ann Surg 2003;238:170–177, with permission.1) 786 Prough Anesthetic Pitfalls in the Elderly Patient J Am Coll Surg

hypercapnia are markedly diminished.19,20 As the pulmonary system ages, there is a small, clinically unimpor- tant increase in true shunt and a substantial increase in ventilation-perfusion mismatching. Both changes are highly individually variable. From an anesthetic standpoint, one important effect of aging on pulmonary function is the increase in closing capacity, which is the lung volume at which small air- ways begin to close. Closing capacity increases more rapidly than functional residual capacity, which is the Figure 2. Preoperative Physical Status (PS), as defined in the American amount of air left in the lungs at the end of a normal tidal Society of Anesthesiologists (ASA) classification system, strongly influ- exhalation, so that closing capacity exceeds functional ences perioperative morbidity. In any category, elderly patients suffer more residual capacity in older patients in the supine posi- complications, but elderly patients also are more likely to have diseases 19 that result in classification as ASA PS 3 or 4. (Reprinted from: Muravchick tion. Under anesthesia, virtually every patient older S. Anesthesia for the elderly. In: Cucchiara RF, Miller ED Jr, Reves JG, et al, than age 45 will have some gas-exchange abnormalities eds. Anesthesia. 5th ed. Philadelphia: Churchill Livingstone; 2000:2140– during a surgical procedure under anesthesia. These ab- 2156, with permission.2) normalities are easily managed by increasing FiO2 or by minor changes in ventilatory management, although re- One important consequence of impaired autonomic duced pulmonary reserves also may contribute to post- activity is impaired thermoregulation in response to operative pulmonary complications. stress. Under nitrous oxide/isoflurane anesthesia, patients aged 60 to 80 years failed to vasoconstrict to con- Renal and hepatic changes serve heat until core temperature was Ͼ1°C lower than Renal mass diminishes with aging, with even more sub- in patients aged 30 to 50 years.16 Older patients cool to stantial drops in renal blood flow and glomerular filtra- a greater extent during anesthesia and require a longer tion rate.21 In the absence of renal disease, serum creat- period of time for rewarming (Fig. 6).17 inine remains relatively constant during aging because Another consequence related to impaired auto- production of creatinine, a product of muscle catabo- nomic activity is response to subarachnoid block (spi- lism, decreases at a rate similar to the rate of decrease in nal anesthesia). In a prospective series of 952 patients the glomerular filtration rate.21 Healthy older patients in receiving spinal anesthesia, risk of developing hypotension (defined for this study as systolic blood pressure Ͻ90 mmHg) was 2.5 times higher in patients older than 40 years.18 Considering the somewhat higher blood pressure expected in older individuals, the definition of hypotension in that population could be set at a level Ͼ90 mmHg, in which case the risk in older patients would be proportionately greater. Had the study separated groups into patients of more advanced age, such as a group exceeding age 65, the risk of hypotension with spinal anesthesia would also likely have been greater.

Pulmonary changes Figure 3. For any organ system, “functional reserve” represents the difference between basal (minimal) and maximal organ system func- Aging-related deterioration of pulmonary function does tion. The age-related decline in functional reserve may not be clinically not produce symptoms in unstressed individuals. Never- apparent until demands made on the organ system are increased by theless, vital capacity, maximum voluntary ventilation, stress, disease, polypharmacy, or surgical intervention. (Reprinted from: Muravchick S. Anesthesia for the elderly. In: Cucchiara RF, Miller and respiratory compliance are lower in aging ED Jr, Reves JG, et al, eds. Anesthesia. 5th ed. Philadelphia: Churchill individuals and the respiratory responses to hypoxia and Livingstone; 2000:2140–2156, with permission.2) Vol. 200, No. 5, May 2005 Prough Anesthetic Pitfalls in the Elderly Patient 787

Figure 4. Relationship between heart rate (A), end-diastolic volume (EDV) (B), and stroke volume (C), versus cardiac output. In older individuals, exercise-induced increases in cardiac output are achieved with a lower heart rate, higher EDV, and higher stroke volume. (Reprinted from: Rodeheffer RJ, Gerstenblith G, Becker LC, et al. Exercise cardiac output is maintained with advancing age in healthy human subjects: cardiac dilatation and increased stroke volume compensate for a diminished heart rate. Circulation 1984;69:203–213, with permission.12) whom serum creatinine is normal will predictably have a advanced age. Both hepatocyte mass and liver blood substantially lower glomerular filtration rate than flow are reduced by 12% to 40% in elderly patients,24 younger patients. The relationship between aging pro- but liver blood flow in relation to hepatocyte volume cesses and chronic renal failure remains unclear; but the does not change with age.25 Much of the decrease in majority of patients with renal failure are older.22 hepatic blood flow may be related to diminished portal Other changes in renal function associated with aging venous flow.26 Pulse echo Doppler demonstrated a 14% may also be important in the perioperative period. Older decrement in portal blood flow in patients aged 56 to 70 experimental subjects do not conserve sodium as well as years in comparison to those younger than 56 years of younger subjects, in association with lower plasma renin age.27 Oxidative drug metabolism in the liver is slowed in activity, lower plasma aldosterone, and higher plasma aging. As a consequence, drugs that are metabolized by levels of atrial natriuretic peptide.21 Antidiuretic hor- hepatic oxidation (eg, diazepam) are cleared less rapidly; mone levels are also increased.23 Older patients are less drugs that are metabolized primarily by glucuronidation able to excrete either a water or sodium load. The com- (eg, lorazepam) are metabolized similarly in young and bination of a decreased glomerular filtration rate and old subjects. Rapidly cleared drugs, such as propranolol decreased aldosterone levels may explain the greater ten- and verapamil, the clearance of which is more dependent dency of older individuals to develop hyperkalemia on hepatic blood flow, are eliminated less rapidly in when given supplemental potassium. older individuals, and elimination of slowly cleared Hepatic function generally is well maintained with drugs, such as warfarin and lidocaine, is minimally in- 788 Prough Anesthetic Pitfalls in the Elderly Patient J Am Coll Surg

Ϫ ϭ Figure 5. Multifactorial contributions to reduced exercise capacity in older adults. A VO2 difference oxygen ϭ extraction. VO2 max maximal oxygen consumption per minute. (Reprinted from: Oxenham H, Sharpe N. Cardiovascular aging and heart failure. Eur J Heart Fail 2003;5:427–434, with permission.14) fluenced by aging.24 Individual variability in liver func- lower concentrations because of increased uptake by ad- tion is sufficiently great among elderly patients that ac- ipose tissue.29 Water-soluble drugs, such as aspirin, lith- curate prediction of drug metabolism in specific patients ium, and ethanol, will attain higher plasma concentra- is difficult. tions as will those, such as neuromuscular blocking

PHARMACOKINETIC AND PHARMACODYNAMIC CHANGES In addition to pharmacokinetic changes in drug metabolism associated with aging, it is important to re- member that pharmacodynamic changes may also alter drug responses. Because of gradual age-related deterioration in homeostatic responses, elderly patients may have exaggerated responses to antihyper- tensive agents, anticoagulants, and sulfonylureas.28 Even such ubiquitous agents as nonsteroidal antiinflammatory drugs are associated with a greater incidence of gastrointestinal and renal complications in elderly patients.28 Changes in body composition that accompany aging Figure 6. (A) Mean age and intraoperative decrease in body temper- may influence the distribution of drugs. In both men ature (°C) are plotted for 97 patients using these age groups: 35 to 52 and women, adipose tissue increases slightly with aging, (n ϭ 15), 53 to 59 (n ϭ 16), 60 to 65 (n ϭ 17), 66 to 69 (n ϭ 17), 70 to 76 (n ϭ 17), and 77 to 92 (n ϭ 15) years. (B) Mean age and accompanied by decreases of lean body mass and total postoperative hours required to rewarm to 36°C are plotted for 97 body water. Depending on the volumes of distribution patients using these age groups: 35 to 52 (n ϭ 15), 53 to 59 (n ϭ 16), of various drugs given during anesthesia, there may be 60 to 65 (n ϭ 17), 66 to 69 (n ϭ 17), 70 to 76 (n ϭ 17), and 77 to 92 ϭ substantial differences in the pharmacological effects of (n 15) years. (Reprinted from: Frank SM, Beattie C, Christopherson R, et al. Epidural versus general anesthesia, ambient operating room anesthetic and adjuvant agents. Lipid-soluble drugs, temperature, and patient age as predictors of inadvertent hypothermia. such as diazepam and verapamil, will attain somewhat Anesthesiology 1992;77:252–257, with permission.17) Vol. 200, No. 5, May 2005 Prough Anesthetic Pitfalls in the Elderly Patient 789 agents, that bind to muscle. In general, older patients are more subject to adverse drug reactions, usually as a consequence of exaggeration of the expected effects of drugs.28,29 Because of the reduction of renal function that accompanies aging and the reduction of hepatocyte mass, metabolism and elimination of many drugs will be impaired. Responses to narcotic analgesics represent an area of particular concern, because of anecdotal reports that older individuals are more sensitive to the effects of morphine when used for postoperative analgesia. Risk of respiratory depression after morphine administration is greater in older patients.30 Although older patients require similar doses of morphine in the postanesthesia care unit to achieve adequate analgesia,31 smaller doses are required for sustained postoperative analgesia when patient-controlled analgesia is used.32 Because of multiple changes in drug distribution, drug clearance, and drug responses that occur in elderly patients and that vary markedly among patients, it is essential to carefully titrate all drugs to appropriate end points and to observe carefully for adverse responses. Figure 7. Changes in minimum alveolar anesthetic (MAC) values for four potent inhalational anesthetics, expressed as a fraction of MAC Anesthetic agents at age 40, as a function of age. Overall, MAC decreased by 6.7% per increasing decade of life. (Reprinted from: Eger EI. Age, minimum In general, older patients require less anesthetic agent. alveolar anesthetic concentration, and minimum alveolar anes- The potency of inhalational agents is characteristically thetic concentration-awake. Anesth Analg 2001;93:947–953, expressed in terms of minimum alveolar concentration with permission.33) (MAC), which is the alveolar concentration at which 50% of patients will not move in response to skin inci- or older, although the effect-site concentration for the sion. Alveolar concentration is roughly equivalent to narcotic remifentanil was not different in older individ- blood concentration but is more easily measured as ex- uals. Although propofol is a short-acting hypnotic agent, haled end-tidal agent concentration. Increasing age is psychomotor function recovers more slowly in elderly associated with decreasing MAC, ie, less inhalational than in young patients (Fig. 8).37 Ninety minutes after anesthetic is required to anesthetize older than younger discontinuing a propofol infusion, patients aged 65 to individuals and older individuals awaken at significantly 85 years had still not returned to baseline psychomotor lower end-tidal concentrations (Fig. 7).33 In patients function, and patients aged 20 to 50 years old returned aged 18 to 30 years, the MAC of desflurane is 7.25%; in to baseline within 30 minutes after discontinuation of patients older than 65 years of age, the MAC of desflu- propofol.37 rane is 5.17%.34 Concentration of sevoflurane at which In clinical practice, increased sensitivity of older pa- patients lose their ability to respond to verbal commands tients to anesthetic agents is likely to be demonstrated as (a concentration that is lower than MAC) is 0.72% in excessive decreases in blood pressure at higher concen- patients aged 18 to 39 years, in contrast to 0.59 in pa- trations. Typically, anesthesiologists gauge depth of an- tients aged 65 to 85 years.35 esthesia based on stress-induced autonomic responses. There are also important differences in responses of Anesthesiologists usually assume that a patient whose older patients to IV agents. Ouattara and colleagues36 blood pressure is high under anesthesia probably re- reported that effect-site concentration for IV hypnotic quires more anesthetic agent and a patient whose blood propofol when administered for tracheal intubation was pressure is low requires less. Although assessment of au- significantly lower by about one-third in patients age 65 tonomic responses is a relatively inaccurate way to assess 790 Prough Anesthetic Pitfalls in the Elderly Patient J Am Coll Surg

dure. In elderly patients undergoing orthopaedic operations, use of BIS was associated with less isoflurane usage and faster emergence.38 If comparable depths of propofol anesthesia, as measured by BIS, are maintained, blood pressure responses to anesthesia are similar in younger and older patients.36 Does the BIS index have any specific relevance to longterm surgical outcomes in elderly patients? Al- though data so far have been published only in abstract form, Weldon and colleagues39 examined the relationship between the BIS index (in effect, the depth of an- Figure 8. Number of dots missed in Trieger’s dot test in younger esthesia) and longterm outcomes in a prospective, obser- (20 to 50 years old) versus older (60 to 70 years old) before and vational study in which anesthesiologists were blinded to after propofol sedation during subarachnoid block anesthesia. *p Ͻ 0.05 versus younger group; **p Ͻ 0.01 versus younger group; †p Ͻ BIS monitoring. Conventionally, the influence of anes- 0.05 versus baseline; p Ͻ 0.01 versus baseline. (Reprinted from: thetics is assumed to barely extend beyond the end of the Shinozaki M, Usui Y, Yamaguchi S, et al. Recovery of psychomotor operation, with little if any longterm effect. Weldon and function after propofol sedation is prolonged in the elderly. Can J 39 Anaesth 2002;49:927–931, with permission.37) associates correlated longterm outcomes with the BIS index in 907 adult patients. Anesthetic management was not altered based on the BIS index; but they recorded the anesthetic depth, it is currently the most common index and later correlated it with other variables. For the method. Correlation with exhaled end-tidal gas concen- purposes of this study, young patients were defined as trations are helpful but fail to provide absolute assess- younger than 40, middle-aged patients as 40 to 59 years, ment of depth of anesthesia. and elderly patients as 60 years or older. At all ages, a low Are there alternative ways to adjust the dose of anes- Ͻ thetic agents? Because the target organ for general anes- BIS index ( 40) correlated with higher 1-year mortal- thesia is the brain, there are rough correlations between ity; and the correlation was particularly striking in el- the raw EEG and depth of anesthesia. In individual pa- derly patients. Of 305 elderly patients, 16.7% who had Ͻ tients, there are striking differences between the EEG in BIS index levels 40 were dead within 1 year (Table 1). the awake state versus moderate sedation (such as might Presumably this was not mediated by intraoperative he- be seen with moderate doses of midazolam) versus light modynamic effects because blood pressure and heart rate general anesthesia or deep general anesthesia. Recogni- were managed similarly in all patients. Perhaps the tion of these patterns is variable and not quantifiable. mechanism is related to alterations of the inflammatory One recent development in anesthesia is a monitor response or to neuroendocrine responses that are influ- that processes the EEG pattern to estimate anesthetic enced by deep anesthesia. Clearly, these data must be depth. The Bi-Spectral Index (BIS) (Aspect Medical Sys- carefully examined when they are published in peer- tems) calculates several parameters from the raw EEG and then converts them into a single number that re- Table 1. Incidence of Death Within 1 Year Versus Predomi- flects anesthetic depth. One argument for the use of BIS nant Bi-Spectral Index Level is that it detects patients who are inadequately anesthe- Age group <40 (%) 40–60 (%) >60 (%) tized but do not yet demonstrate autonomic evidence All ages (n ϭ 907) 7.8 6.8 3.1 (ie, increased blood pressure and heart rate) of arousal. Young (Ͻ40 y; n ϭ 291) 3.1 2.1 4.1 Some clinicians use BIS to ensure adequate depth of Middle-aged anesthesia and also to avoid excessive depth. In quanti- (40–59 y; n ϭ 311) 9.3* 4 0 tative terms, the goal is to maintain the BIS index be- Elderly tween 40 and 60 during general anesthesia. A potential (Ն60 y; n ϭ 305) 16.7* 12.6 4.2 additional benefit of BIS monitoring is that more precise *p ϭ 0.05. (From Weldon BC, Mahla ME, van der Aa MT, Monk TG. Advancing age adjustment of anesthetic depth may result in somewhat and deeper intraoperative anesthetic levels are associated with higher first year more prompt awakening at the end of a surgical proce- death rates. Anesthesiology 2002;96:A1097, with permission.39) Vol. 200, No. 5, May 2005 Prough Anesthetic Pitfalls in the Elderly Patient 791 reviewed form and other investigators must confirm these provocative findings. Differences in drug elimination in older patients are most important in elimination of neuromuscular blocking agents. Marked differences, regardless of age, in individual responses to muscle relaxants obscure the fact that older patients, on average, eliminate muscle relaxants more slowly. In elderly patients, as in younger surgical patients, neuromuscular blockers are frequently used to avoid movement and provide light anesthesia. One way to describe recovery from neuromuscular blocking agents is the recovery index, which is defined as the time between 25% recovery and 75% recovery of neuromuscular function after administration of neuromuscular blockers (Fig. 9).40 Agents that have particu- Figure 9. The time needed for clinical recovery from neuromuscular blockade. Recovery index (RI), time required for spontaneous recovery larly prolonged durations of action in older patients in- from 25% to 75% of the control-evoked neuromuscular response, is clude pancuronium and vecuronium. signiﬁcantly increased in older adults (shaded bars) for nondepolarizing relaxants that require organ-based clearance from plasma but is little different for atracurium, cisatracurium, or mivacurium because they are Neurologic changes hydrolyzed in plasma. (Reprinted from: Muravchick S. Anesthesia for A wealth of animal and human research demonstrates the geriatric patient. In: Barash PG, Cullen BF, Stoelting RK, eds. Clinical anesthesia. 4th ed. Philadelphia: Lippincott Williams & Wilkins; substantial changes in the brain with aging, including 2001:1205–1216, with permission.40) diminished cortical volume,41 accumulated markers of chronic stress,42,43 DNA damage,44 and diminished hip- POSTOPERATIVE pocampal (memory) function.45 Terry and colleagues41 examined 51 brains at autopsy in patients who had been COGNITIVE DYSFUNCTION clinically and neuropathologically normal; although to- Postoperative cognitive dysfunction (POCD) is most com- tal numbers of neurons and overall neuronal density monly described after cardiac operations, after which a sub- were unchanged, aging brains showed statistically signif- stantial percentage of patients experience deterioration icant decrements in brain weight, cortical thickness in in cognitive function that persists for years after the op- 47 the midfrontal and superior temporal areas, and the ratio eration. Among 261 patients who underwent coronary of neurons to glia in the midfrontal and inferior parietal artery bypass operation under cardiopulmonary bypass, 53% demonstrated cognitive decline at discharge, which areas. Based on their analysis, the brain would lose about decreased to 36% at 6 weeks and 24% at 6 months, 250 g in weight (about one-sixth of normal brain weight before increasing to 42% at 5 years. More recently, nu- at age 30) from age 30 through age 80. In comparison to merous investigators have demonstrated that POCD a series that examined patients with degenerative neuro- commonly occurs after noncardiac operations in elderly logic diseases, loss of brain weight was considerably re- 48,49 46 and middle-aged patients. Minor operations per- duced. Rogers and colleagues have suggested that formed on an outpatient basis appear to be associated marked variation among individuals in rate of decline in with less risk of POCD.50 In addition to anesthetic intellectual function associated with aging could be ex- agents, a variety of perioperative factors, eg, pain, de- plained by large individual differences in hypothalamic- pression, and fear of loss of function, could produce pituitary-adrenal activity. Specifically, chronic increases POCD.51 Influence of anesthesia must be critically in corticosterone levels could explain specific aging- evaluated. related processes such as hippocampal dysfunction and The first study to call attention to POCD in patients neuronal loss.45 The relationship between anesthetic other than cardiac surgical patients was published in outcomes and aging-related changes in the brain is just Lancet in 1998.48 The authors examined multiple factors beginning to be investigated. that were associated with deterioration in discrete objec- 792 Prough Anesthetic Pitfalls in the Elderly Patient J Am Coll Surg tive testing of various intellectual functions at 1 week Table 2. Analysis of Predictors of Postoperative Cognitive and 3 months after a surgical procedure. At 1 week, Dysfunction after Major Operation in Adults Univariate Multivariate patients had a relatively high incidence of POCD; even Risk factors for POCD p value p value at 3 months, about 10% of the entire population had Age Ͻ0.001* Ͻ0.001* POCD, but the most striking difference was that POCD at 1 week POCD occurred in 14% of patients older than 70 years postoperative Ͻ0.01* NS of age and 7% of those aged 60 to 69. At 1 week after Operation type 0.04* NS surgical procedure, cognitive testing could easily be dis- Baseline MMSE NS NS rupted by residual pain, analgesics, stress, altered nutri- Depression at 3 months postoperative Ͻ0.01* NS tion, and numerous other possible factors. At 3 months Anxiety at 3 months after the operation, most acute effects should be re- postoperative NS NS solved. As controls, nonsurgical patients underwent cog- Pain at 3 months nitive testing; in contrast to surgical patients, about 3% postoperative NS NS of nonsurgical patients showed some deterioration. *Signifies statistical significance in either the first-stage univariate analysis of factors or the subsequent multivariate model. Surprisingly, reoperation and duration of anesthesia, MMSE, Mini-Mental State Examination; POCD, postoperative cognitive which are factors that seemed likely to correlate with dysfunction. (From Monk TG, Garvan CW, Dede DE, et al. Predictors of postoperative POCD, were not statistically significantly related to cognitive dysfunction following major surgery. Anesthesiology 2001;95:A50, POCD at the 3-month interval.48 Monk and col- withpermission.52) leagues52 compared the incidence of POCD in surgical patients younger than 40 years, 40 to 59 years, and older longer hospital stays. These data suggest that POCD is than 60 years at 1 week and 3 months postoperatively. related to brain oxygenation during anesthesia, although By multivariate analysis, the only predictor of POCD more quantitative measurements of brain oxygenation that remained significant at 3 months after the operation are necessary to make that statement. was age (Table 2). Other factors, including POCD at 1 Other mechanisms of POCD have been explored. week, type of operation, baseline mental status, and de- Patients who developed new cognitive deficits after lapa- pression, did not contribute to the multivariate model. roscopic cholecystectomy were more likely to have There is a prominent influence of age on POCD that higher concentrations of stable metabolites of nitric ox- may well influence quality of life, although other authors ide.55 Other investigators have addressed the question of report that POCD does not correlate well with patients’ whether random variability in cognitive testing could subjective perception of how well they function.49,53 result in apparent POCD, but concluded that the effect Monk and colleagues54 also correlated the occurrence of random variability was insufficient to explain the of POCD with cerebral near-infrared spectroscopy. Ce- magnitude and frequency of effects.56 Surprisingly, pa- rebral near-infrared spectroscopy is based on the ability tients older than 60 years of age who were randomized to of near-infrared light to penetrate the skull, where the postoperative inpatients or outpatients after minor op- light encounters deoxygenated and oxygenated hemo- erations were more likely to have POCD if they were globin, thereby changing the returning optical signal. older than 70 and if they were managed as inpatients.50 Because cerebral near-infrared spectroscopy cannot This suggests that hospitalization per sé is associated measure absolute value of brain oxyhemoglobin satura- with POCD. tion, the investigators defined a clinically important de- One potentially important question is whether re- saturation as a drop in brain saturation to Ͻ 80% of the gional anesthesia reduces incidence of POCD. In gen- baseline and then calculated desaturation area as the eral, data suggest that overall surgical mortality is not number of minutes that the brain was below that 80% influenced by choice of regional or general anesthe- threshold, multiplied by the difference between 80% of sia.57,58 Similarly, incidence of POCD at 3 months post- baseline and the actual percent of baseline. Elderly pa- operation is not influenced by choice of regional or gen- tients who did not develop POCD had few desatura- eral anesthesia.59 This suggests that factors other than the tions and a small total area of desaturation. In contrast, specific influences of anesthesia on the brain are respon- 62% of patients who had POCD had desaturations, had sible for POCD. Recently, Selwood and Orrell60 cau- a substantial desaturation area, and also had slightly tioned that research to date on POCD has sufficiently Vol. 200, No. 5, May 2005 Prough Anesthetic Pitfalls in the Elderly Patient 793 important methodologic limitations that no firm con- 14. Oxenham H, Sharpe N. Cardiovascular aging and heart failure. clusions are possible. Further studies are necessary. Eur J Heart Fail 2003;5:427–434. 15. Pfeifer MA, Weinberg CR, Cook D, et al. Differential changes In summary, elderly surgical patients pose two sets of of autonomic nervous system function with age in man. Am J problems. As in younger patients, intercurrent disease, par- Med 1983;75:249–258. ticularly cardiovascular disease, renal disease, and lung dis- 16. Kurz A, Plattner O, Sessler DI, et al. The threshold for thermo- regulatory vasoconstriction during nitrous oxide/isoflurane an- ease, may influence anesthetic management and postoper- esthesia is lower in elderly than in young patients. Anesthesiol- ative care. Even apparently robust elderly patients may have ogy 1993;79:465–469. important reductions of reserve capacity in major organ 17. Frank SM, Beattie C, Christopherson R, et al. Epidural versus general anesthesia, ambient operating room temperature, and systems. Important pharmacologic changes include inpatient age as predictors of inadvertent hypothermia. Anesthe- creased sensitivity to sedative and anesthetic drugs and pro- siology 1992;77:252–257. longation of the duration of action of other common anes- 18. Carpenter RL, Caplan RA, Brown DL, et al. Incidence and risk thetic drugs and adjuvants, particularly certain muscle factors for side effects of spinal anesthesia. Anesthesiology 1992; 76:906–916. relaxants. Finally, POCD is an area of increasing interest. 19. Allen SJ. Respiratory considerations in the elderly surgical pa- Ultimately, investigators may uncover evidence that certain tient. Clinics in Anesthesiology 1986;4:899–930. anesthetics are more associated with POCD. Perhaps spe- 20. Peterson DD, Pack AI, Silage DA, Fishman AP.Effects of aging on ventilatory and occlusion pressure responses to hypoxia and cific interventions can be identified to reduce the incidence hypercapnia. Am Rev Respir Dis 1981;124:387–391. and severity of POCD. 21. Lindeman RD. Renal physiology and pathophysiology of aging. Contrib Nephrol 1993;105:1–12. 22. Lamb EJ, O’Riordan SE, Delaney MP.Kidney function in older people: pathology, assessment and management. Clin Chim REFERENCES Acta 2003;334:25–40. 1. Etzioni DA, Liu JH, Maggard MA, Ko CY. The aging popula- 23. Luckey AE, Parsa CJ. Fluid and electrolytes in the aged. Arch tion and its impact on the surgery workforce. Ann Surg 2003; Surg 2003;138:1055–1060. 238:170–177. 24. Jansen PL. Liver disease in the elderly. Best Pract Res Clin Gas- 2. Muravchick S. Anesthesia for the elderly. In: Cucchiara RF, Miller troenterol 2002;16:149–158. ED Jr, Reves JG, et al, eds. Anesthesia. 5th ed. Philadelphia: 25. Wakabayashi H, Nishiyama Y, Ushiyama T, et al. Evaluation of Churchill Livingstone; 2000:2140–2156. the effect of age on functioning hepatocyte mass and liver blood 3. Golden TR, Hinerfeld DA, Melov S. Oxidative stress and aging: flow using liver scintigraphy in preoperative estimations for sur- beyond correlation. Aging Cell 2002;1:117–123. gical patients: comparison with CT volumetry. J Surg Res 2002; 4. Van Remmen H, Hamilton ML, Richardson A. Oxidative dam- 106:246–253. age to DNA and aging. Exerc Sport Sci Rev 2003;31:149–153. 26. Zoli M, Iervese T, Abbati S, et al. Portal blood velocity and flow 5. Lu T, Pan Y, Kao SY, et al. Gene regulation and DNA damage in in aging man. Gerontology 1989;35:61–65. the ageing human brain. Nature 2004;429:883–891. 27. Gillardon F, Kiprianova I, Sandkuhler J, et al. Inhibition of 6. Van Zant G, Liang Y. The role of stem cells in aging. Exp He- caspases prevents cell death of hippocampal CA1 neurons, but matol 2003;31:659–672. not impairment of hippocampal long-term potentiation follow- 7. Ozawa T. Genetic and functional changes in mitochondria as- ing global ischemia. Neuroscience 1999;93:1219–1222. sociated with aging. Physiol Rev 1997;77:425–464. 28. Turnheim K. When drug therapy gets old: pharmacokinetics 8. Pedersen BK, Bruunsgaard H. Possible beneficial role of exercise and pharmacodynamics in the elderly. Exp Gerontol 2003;38: in modulating low-grade inflammation in the elderly. Scand 843–853. J Med Sci Sports 2003;13:56–62. 29. Bressler R, Bahl JJ. Principles of drug therapy for the elderly 9. Miller RA. The aging immune system: primer and prospectus. patient. Mayo Clin Proc 2003;78:1564–1577. Science 1996;273:70–74. 30. Cepeda MS, Farrar JT, Baumgarten M, et al. Side effects of 10. Esterling BA, Kiecolt-Glaser JK, Glaser R. Psychosocial modu- opioids during short-term administration: effect of age, gender, lation of cytokine-induced natural killer cell activity in older and race. Clin Pharmacol Ther 2003;74:102–112. adults. Psychosom Med 1996;58:264–272. 31. Aubrun F, Bunge D, Langeron O, et al. Postoperative morphine 11. Aronow WS, Stein PD, Sabbah HN, Koenigsberg M. Resting consumption in the elderly patient. Anesthesiology 2003;99: left ventricular ejection fraction in elderly patients without evi- 160–165. dence of heart disease. Am J Cardiol 1989;63:368–369. 32. Woodhouse A, Mather LE. The influence of age upon opioid 12. Rodeheffer RJ, Gerstenblith G, Becker LC, et al. Exercise car- analgesic use in the patient-controlled analgesia (PCA) environ- diac output is maintained with advancing age in healthy human ment. Anaesthesia 1997;52:949–955. subjects: cardiac dilatation and increased stroke volume com- 33. Eger EI. Age, minimum alveolar anesthetic concentration, and pensate for a diminished heart rate. Circulation 1984;69:203– minimum alveolar anesthetic concentration-awake. Anesth Analg 213. 2001;93:947–953. 13. Kawaguchi M, Hay I, Fetics B, Kass DA. Combined ventricular 34. Gold MI, Abello D, Herrington C. Minimum alveolar concen- systolic and arterial stiffening in patients with heart failure and tration of desflurane in patients older than 65 yr. Anesthesiology preserved ejection fraction: implications for systolic and dia- 1993;79:710–714. stolic reserve limitations. Circulation 2003;107:714–720. 35. Katoh T, Bito H, Sato S. Influence of age on hypnotic require- 794 Prough Anesthetic Pitfalls in the Elderly Patient J Am Coll Surg

ment, bispectral index, and 95% spectral edge frequency ISPOCD investigators. International Study of Post-Operative associated with sedation induced by sevoflurane. Anesthesiology Cognitive Dysfunction. Lancet 1998;351:857–861. 2000;92:55–61. 49. Johnson T, Monk T, Rasmussen LS, et al. Postoperative cogni- 36. Ouattara A, Boccara G, Lemaire S, et al. Target-controlled infu- tive dysfunction in middle-aged patients. Anesthesiology 2002; sion of propofol and remifentanil in cardiac anaesthesia: influ- 96:1351–1357. ence of age on predicted effect-site concentrations. Br J Anaesth 50. Canet J, Raeder J, Rasmussen LS, et al. Cognitive dysfunction 2003;90:617–622. after minor surgery in the elderly. Acta Anaesthesiol Scand 2003; 37. Shinozaki M, Usui Y, Yamaguchi S, et al. Recovery of psychomo- 47:1204–1210. tor function after propofol sedation is prolonged in the elderly. 51. Bekker AY, Weeks EJ. Cognitive function after anaesthesia in the Can J Anaesth 2002;49:927–931. elderly. Best Pract Res Clin Anaesthesiol 2003;17:259–272. 38. Wong J, Song D, Blanshard H, et al. Titration of isoflurane 52. Monk TG, Garvan CW, Dede DE, et al. Predictors of postop- using BIS index improves early recovery of elderly patients un- erative cognitive dysfunction following major surgery. Anesthe- dergoing orthopedic surgeries. Can J Anaesth 2002;49:13–18. siology 2001;95:A50. 39. Weldon BC, Mahla ME, van der Aa MT, Monk TG. Advancing 53. Dijkstra JB, Jolles J. Postoperative cognitive dysfunction versus age and deeper intraoperative anesthetic levels are associated complaints: a discrepancy in long-term findings. Neuropsychol with higher first year death rates. Anesthesiology 2002;96: Rev 2002;12:1–14. A1097. 54. Monk TG, Weldon BC, Weldon JE, van der Aa MT. Cerebral 40. Muravchick S. Anesthesia for the geriatric patient. In: Barash oxygen desaturations are associated with postoperative cognitive PG, Cullen BF, Stoelting RK, eds. Clinical anesthesia. 4th ed. dysfunction in elderly patients. Anesthesiology 2002;96:A40. Philadelphia: Lippincott Williams & Wilkins; 2001:1205– 55. Iohom G, Szarvas S, Larney V, et al. Perioperative plasma con- 1216. centrations of stable nitric oxide products are predictive of cog- 41. Terry RD, DeTeresa R, Hansen LA. Neocortical cell counts in nitive dysfunction after laparoscopic cholecystectomy. Anesth normal human adult aging. Ann Neurol 1987;21:530–539. Analg 2004;99:1245–1252. 42. Heuser I, Lammers CH. Stress and the brain. Neurobiol Aging 56. Rasmussen LS, Siersma VD. Postoperative cognitive dysfunc- 2003;24[Suppl 1]:S69–S76. tion: true deterioration versus random variation. Acta Anaesthe- 43. O’Brien JT.The ‘glucocorticoid cascade’ hypothesis in man. Br J siol Scand 2004;48:1137–1143. Psychiatry 1997;170:199–201. 57. Wu CL, Anderson GF, Herbert R, et al. Effect of postoperative 44. Mattson MP. Gene-diet interactions in brain aging and neuro- epidural analgesia on morbidity and mortality after total hip degenerative disorders. Ann Intern Med 2003;139:441–444. replacement surgery in medicare patients. Reg Anesth Pain Med 45. Meaney MJ, O’Donnell D, Rowe W,et al. Individual differences 2003;28:271–278. in hypothalamic-pituitary-adrenal activity in later life and hip- 58. Christopherson R, Beattie C, Frank SM, et al. Perioperative pocampal aging. Exp Gerontol 1995;30:229–251. morbidity in patients randomized to epidural or general anes- 46. Rogers RL, Meyer JS, Mortel KF. After reaching retirement age thesia for lower extremity vascular surgery. Anesthesiology physical activity sustains cerebral perfusion and cognition. J Am 1993;79:422–434. Geriatr Soc 1990;38:123–128. 59. Rasmussen LS, Johnson T, Kuipers HM, et al. Does anaesthesia 47. Newman MF, Kirschner JL, Phillips-Bute B, et al. Longitudinal cause postoperative cognitive dysfunction? A randomised study assessment of neurocognitive function after coronary-artery by- of regional versus general anaesthesia in 438 elderly patients. pass surgery. N Engl J Med 2001;344:395–402. Acta Anaesthesiol Scand 2003;47:260–266. 48. Moller JT, Cluitmans P, Rasmussen LS, et al. Long-term post- 60. Selwood A, Orrell M. Long term cognitive dysfunction in older operative cognitive dysfunction in the elderly. ISPOCD1 study. people after non-cardiac surgery. BMJ 2004;328:120. COLLECTIVE REVIEW

Schistosomal Portal Hyptertension

Adeyemi O Laosebikan, MBBS, FCS(SA), Sandie R Thomson, ChM, FRCS(Ed & Eng), Namasha M Naidoo, MBChB, FCS(SA)

Schistosomiasis is a multisystemic disease caused by spe- presence of IL-12, before egg challenge, verified reduced cies of the Schistosoma protozoa (S mansoni, S japonica, fibrosis and granuloma size.17 As disease progresses, S hematobium, S intercalatum, and S mekogni).1-3 First Th-1 response is downregulated, diminishing reaction documented in ancient Egypt, it has become an en- to new incoming eggs, but the Th-2 response that leads demic disease in 74 tropical and subtropical countries, to antibody and eosinophil release lingers on.11,17 But with 200 million people afflicted and 600 million at only 2% to 10% of humans with schistosomiasis de- risk.4-6 velop hepatosplenic disease.15,18 Immunogenetic studies Schistosomiasis is the most common cause of portal suggest that polymorphism within the IFN gamma receptor 7 hypertension, typically caused by S mansoni and char- (R1) gene influences hepatic disease severity, implying a ge- acterized by portal vein obstruction from intrahepatic netic predisposition.13,15 8 periportal fibrosis originally described by Symmers. Varices develop because of defective fibrogenesis, hy- Schistosomal portal hypertension usually presents in the droxyproline deficiency within the vascular tissue, and gas- young with variceal bleeding, and association with other troesophageal lymphatic congestion.19 Chronic venous causes of portal hypertension is common and influences 2,9,10 congestion, mucosal hypoxia, and atrophy predispose to the disease pattern and outcomes. variceal rupture during periods of negative intraesophageal pressure.20 This is also influenced by neurohumoral re- Pathogenesis of schistosomiasis sponses that cause splanchnic hyperemia.21 The ova induce a cell-mediated, delayed hypersensitivity reaction that initiates a vascular granulomatous Clinicopathologic features disease characterized by presinusoidal pyelophlebitis The disease is prevalent in the young (10 to 30 years), and thromboembolism subsequently replaced by pro- with a long latent period ranging from 2 to 25 years.13,22 gressive fibrosis.9,11,12 This correlates with worm load, 13,14 Initial intestinal symptoms are nonspecific, unreliable, and may later affect the postsinusoidal vessels. The 23 egg granulomata activate antigen-specific CD4 Th-1 and depend on other host factors. Variceal formation is 14,24 and Th-2 cells, inducing release of specific immuno- more frequent than it is in cirrhosis. modulating antifibrogenic (interleukin [IL]-12, inter- Hepatosplenic manifestation permits differentiation feron [IFN] gamma) and fibrogenic (tumor necrosis fac- from the genuinely cirrhotic process because organ size is 25,26 tor [TNF] ␣, IL-4, IL-10, transforming growth factor related to egg output. The liver is firm with a larger left [TGF] ␤1) cytokines, respectively.11,15 In a study using lobe because of its preferential blood flow, but it shrinks as 26,27 severe combined immunodeficient (SCID) mice, with the disease advances. Evaluation of the cut liver surface normal macrophages but no functional B or T lympho- gives the appearance of clay pipestems because of fibroob- cytes, TNF ␣ produced by the Th-2 cells led to an instructive bands along the portal tracts.8,25 Microscopy crease in egg production and granuloma formation.16 In reveals focal matrix degradation characterized by fragmen- another study, animals sensitized to egg antigens in the tation and dispersion of collagen fibers, hyperplasia of elas- tic tissue, destruction of portal vein radicles, and disarray of 14,28 Competing interests declared: None. smooth muscle fibers. Congestive splenomegaly may

Received June 29, 2004; Revised October 26, 2004; Accepted November 17, result from portal hypertension, lymphoid hyperplasia, or 2004. secondary hypersplenism.29 Visceral nodularity occurs in From the Departments of Surgery, Greys Hospital, Pietermaritzburg (Laosebikan), and the Nelson R Mandela School of Medicine (Thomson, about 10% of patients and may signify a more acute type Naidoo), University of KwaZulu Natal, Durban, South Africa. infection.26 Signs of advanced disease are anemia, ascites, Correspondence address: SR Thomson, Department of Surgery, Nelson R Mandela School of Medicine, University of KwaZulu Natal, Private Bag X 7, anasarca from low albumin, and gallbladder thickening Congella 4013, Durban, South Africa. from adjacent liver fibrosis.4,13,30

© 2005 by the American College of Surgeons ISSN 1072-7515/05/$30.00 Published by Elsevier Inc. 795 doi:10.1016/j.jamcollsurg.2004.11.017 796 Laosebikan et al Schistosomal Portal Hypertension J Am Coll Surg

extrahepatic shunts.5 Hemodynamic studies in schisto- Abbreviations and Acronyms somal disease reveal a unique hyperdynamic circulatory DSRS ϭ distal splenorenal shunt syndrome characterized by marked increased cardiac in- EBL ϭ band ligation EIS ϭ injection of sclerosants dex and decreased peripheral vascular resistance, splenic GEDS ϭ gastroesophageal decongestion and splenectomy hyperflow, and reduced mesenteric blood flow, with IFN ϭ interferon ϭ mild increase in portal flow compared with the total IL interleukin 18,27,36-39 US ϭ ultrasonography hepatic blood flow. There is also elevated intra- splenic and portal venous pressure and a normal or slightly elevated sinusoidal pressure.36 The evidence for 36,37 Diagnosis tachycardia is conflicting. The increase in azygous blood flow is much less than that observed for other Early diagnosis is vital for therapy and prognosis espe- 39 cially in endemic areas.23 Direct confirmation is based causes of portal hypertension. This syndrome may be on finding ova in stools, and in 60% of patients, rectal caused by nitric oxide and other vasoactive media- 5,27,37,38 biopsies will be positive.31 The traditional hatching test tors, but is sustained by splenomegaly and splenic 23,27,37 and Kato-Katz’s thick smear stool examination tech- hyperflow. nique will increase the yield.13 Serology, including In early stages, hepatic perfusion is usually normal, or indirect hemagglutination, immunofluorescence, and increased paradoxically, because of the focal distribution of ELISA, is indispensable,13,25 but biochemical and hema- the inflammation, periportal neovascularization, and com- tologic markers are not.18,31 pensatory hepatic arterial hypertrophy and flow.36,40 The Ultrasound (US) is widely used to demonstrate hepa- hepatic venous pressure gradient is normal and does not reflect tosplenic lesions. Its simplicity, safety, specificity (95%), true splanchnic hemodynamics and disease severity.36 With and sensitivity (97%) have enabled it to replace wedge disease progression, compensatory changes become in- biopsy as the gold standard.1 In cases of diagnostic doubt adequate.23 A high hepatic venous pressure gradient or in regions of low endemicity, wedge biopsy is pre- might suggest sinusoidal involvement or associated viral ferred to needle biopsy because the focal nature of or alcoholic hepatitis.36,37 perivascular changes makes the latter inaccurate.4,13,31 Ultrasound assesses organ size, portal hemodynam- Schistosomal treatment ics, and ascites, which are important in staging and Failure to treat schistosomiasis leads to ongoing liver followup.1,32 Liver US grading for mansoniasis is damage with its sequelae.41 Praziquantel, the drug of based on the thickness of the portal vein radicle.33,34 choice, is contraindicated in pregnancy and severe he- US systems, including the Homeida, Arafa, and patic, renal, and cardiac failure. Single or repeated doses Doehering-Schwerdtfeger systems, have largely been 34 replaced by the World Health Organization system, range from 20 to 40 mg/kg. The alternative, oxam- 42,43 which has four grades. Grade 0 represents the absence niquine, has significant neuropsychiatric side effects. of echogenic thickening of any portal vein radicles (Ͻ3 mm); Pathologic resolution is evidenced by the reduction in I, mild (3 to 5 mm); II, moderate (6 to 7 mm); and III, liver volume, the degree of periportal thickening, and 10 marked (Ͼ7 mm). These features are not seen in cirrho- biochemical markers. It occurs more quickly in the sis.23,34,35 Intrahepatic capsular and septal calcification young and those with early disease, but takes 6 months give a “turtle neck” appearance unique to japonica infec- in the majority of humans.34,44 tion.25 Doppler US demonstration of reduction in portal In advanced disease, fibrosis increases despite a nega- flow and the size of collaterals are good prognostic indi- tive egg output.44,45 Parasite eradication is confirmed if cators of reduced bleeding and survival in S japonica stool or rectal biopsy remains egg free after 6 months. disease.23 Large-scale treatment and public health maneuvers have paid dividends in the hyperendemic areas of Brazil and Pathophysiology China.13,25 Vaccine development may further prevent Portal hypertension long precedes any change in liver the immunopathologic changes associated with the function, with development of both intrahepatic and disease.11,25,46 Vol. 200, No. 5, May 2005 Laosebikan et al Schistosomal Portal Hypertension 797

Treatment of schistosomal portal hypertension efficacy of central shunts,7,22,41,50 but with less risk of Schistosomal portal hypertension is unique, with a distinct encephalopathy.8,55 pathophysiology, different from cirrhotic and prehepatic GEDS, developed in 1964 by Hassab,56 is widely causes.36 Liver function is generally well preserved and the practiced in Egypt20,38 and Brazil.27,57 The critical step is main source of morbidity and mortality is bleeding from the perihiatal devascularization of the abdominally varices or portal hypertensive gastropathy.13,25,36,41,47 Trans- reached lower 3 to 4 inches of the esophagus, with liga- plantation is rarely required, as opposed to cirrhosis, in tion of the left and short gastric vessels, diverting the which longterm outcomes are largely determined by the venous drainage from the precarious gastroesophageal liver reserve.47-49 junction caudally to the nonvulnerable retroperitoneal Treatments fall into primary and secondary prophylaxis collaterals.20,39 Intraoperative and postoperative hemo- categories. Endotherapy, interventional radiology, and a dynamic studies show reversal of the hyperdynamic cir- variety of surgical options have been applied as they have culation toward normalization from the time the splenic for all causes of portal hypertension. Their use varies de- artery is ligated through to complete azygoportal discon- pending on clinical presentation, available modalities, and nection, a pattern not seen in cirrhosis.27,39 institutional expertise and bias.47 The many variations of Controversy still surrounds the complete but exten- operations have evolved in attempts to improve efficacy or sive devascularization and esophageal transection of reduce morbidity, or as specific modifications, because of Sugiura and Futagawa58 and Hassab’s limited transab- the pathophysiology of the disease.13,38,48,50 Interventions dominal approach without esophageal transection.59 aimed at preventing bleeding are presented in the context of The Sugiura and Futagawa operation has never been treatment of portal hypertension in general and schistoso- extensively applied to schistosomal portal hypertension, mal portal hypertension in particular. and Hassab believes that the extensive gastric devascularization worsens congestion and accentuates rebleed- 38,39 Primary prophylaxis ing. GEDS also avoids the potential complications of esophageal transection.39 Encephalopathy is negligible Primary prophylaxis aims to reduce the very high mor- because portal perfusion is maintained and splenectomy tality of the initial bleed from high-risk varices.47 The eliminates the hemolysis and hypersplenism that may be ideal option must be safe and cost effective, with mini- observed after DSRS.22,57,60 mal side effects. Predictive indices, derived from studies DSRS, developed by Warren and colleagues50 in 1967 in cirrhotic patients, do not provide sufficient evidence and modified by the addition of the splenopancreatic for widespread prophylaxis, and uniform application disconnection in 1984, selectively drains the dangerous would mean unnecessary treatment and exposure to po- esophagogastric venous complex into the systemic circu- tential complications for the majority of patients.47,51-53 lation while maintaining portal perfusion. Splenic vein Toour knowledge, no studies have been done in schis- mobilization is the most challenging step, especially in tosomiasis with regard to nonselective beta-blockers and the presence of thin dilated veins and marked spleno- endotherapy. Several reviews concluded that prophylac- megaly. There is a rapid decrease in variceal pressure and tic sclerotherapy in portal hypertension in general, immediate decongestion of the spleen once the shunt is should be adopted only in the context of a controlled 47,51,54 opened, which culminates in the reduction of spleno- trial. megaly.59 It requires a learning curve before consistent Operations have been used as primary prophylaxis results are obtainable, and is feasible in children without and definitive therapy. For better understanding, the op- great difficulty.61,62 erations and their rationales are described. Modifications of this procedure in unfavorable anatomy do not preclude shunt selectivity.24,41,63-65 Thrombosis of Operations and rationale for their use the mesenteric end of the splenic vein may occur with po- Two major procedures, gastroesophageal decongestion tential portal vein thrombosis and its debilitating sequelae, and splenectomy (GEDS), specifically evolved for schis- rebleeding, ascites, or, rarely, mesenteric infarction.59,66 tosomiasis, and distal splenorenal shunt (DSRS), origi- The risk of rebleeding is highest within the first postop- nally devised for cirrhosis, have been widely applied to erative month (10% to 15%) and is attributed to shunt schistosomal patients. It is believed that they equal the occlusion or a large splenic vein and splenic hyperflow, 798 Laosebikan et al Schistosomal Portal Hypertension J Am Coll Surg secondary to renal venous hypertension that occurs during earlier resuscitation and stabilization. Of the 10 patients the process of gastroesophageal decompression.7,48 subjected to emergency surgery in the series by Bessa and One of the advantages of DSRS is that it does not associates,73 9 had staple transection of the esophagus. accelerate the rate of progression of underlying liver dis- Bleeding was effectively controlled in 90%, but varices ease beyond that of the disease’s natural history.48 The recurred in 30%. high venous pressure within the intestinal wall after DSRS Five of the ten studies61,64,69,70,74 summarized in Table 2 appears beneficial, inhibiting the reabsorption of nitroge- have data on emergency DSRS. The numbers are small nous breakdown products.67 Longterm hepatopetal flow is except for Pitanga’s experience,70 and the full article was preserved,7,48 and postshunt encephalopathy is low in the not available. Preoperative management was not clearly schistosomal cohorts.61,68 Hypersplenism may persist in up stated in most of these studies, many variations of DSRS to 69% of patients, though in most it is asymptomatic.64,69 were performed, and results were not stratified accord- Because of the frequent occurrence of hypersplenism ing to liver reserve. The operative mortality ranged from in schistosomal portal hypertension, the extensive na- 13.3% to 20.0%, and postoperative encephalopathy re- ture of GEDS, and its simplicity, splenectomy with or bleeding rates were not always stated. without limited devascularization is a considered op- Although survivors of extensive operations share ben- tion. It deals with the crucial hemodynamic factor that efits with those operated electively, emphasis is now on sustains portal hypertension, splenic hyperflow, and re- the less demanding and simpler staple esophageal tran- moves a large area of portal bed with collateral veins in section for patients in whom endotherapy failed.47,63 the gastrolienal ligament.57,59 Longterm treatment Prophylactic surgery The goal is to prevent longterm rebleeding and its provoc- Prophylactic surgery was common practice until the ative mortality, with minimal morbidity. The optimal ap- 1980s.51 Data are shown in Tables 1 and 2. The results of proach remains controversial, but access to appropriate prophylactic surgery are difficult to analyze as a separate level of care and compliance influence treatment choice.61 entity because of the heterogeneity of the series and the absence of detail to determine critical end points for Audit of noncomparative reports comparison.38,64,70-72 Data summarized in Tables 1 and 2 suggest that the In view of the significant surgical mortality and potential results of GEDS are excellent. In 351 schistosomal por- morbidity, the many procedural variations, and lack of ad- tal hypertensive patients, Hassab38 reported an operative equate predictive indices for bleeding to apply selective pro- mortality of 3%. This increased to 10% when the expe- phylaxis, prophylactic surgery is unjustifiable.47 rience of trainees was included. Variceal rebleeding in this and other series occurred in 11% to 12%, with a rate The acute bleed of encephalopathy at less than 1%.38,71,75 Therapeutic strategy is constantly changing with techno- Regarding DSRS,7,24,61,63,64,69,70,74,76,77 (Table 2), the larg- logic advances, especially in minimally invasive techniques. est series were from Egypt7,76,77 and South America.63,69,70,74 As with other causes of portal hypertension, somatostatin Five studies were prospective. Three studies documented or its analogues in combination with endotherapy, at the variations of the standard procedure. Operative mortality time of diagnostic endoscopy in a resuscitated patient, re- averaged 3.7% (0% to 10%), with the lower figures from main the gold standard, reserving transjugular intrahepatic the experienced centers. The variceal rebleeding and en- portosystemic shunts (TIPS) or surgery for failures.47 The cephalopathy rates averaged 3.6% (range 0% to 8%) and benefit of ab initio surgery is shrouded by concerns such as 2.1% (range 0% to 6%), respectively. Encephalopathy was availability of expertise, the risk of major blood transfusion, not disabling, and was easily controlled by dietary manip- and the significant morbidity and mortality.41,47 ulation.7 The 5-year survival rate was between 72% and 100%. Morbidity frequency must be interpreted in the Emergency surgery context of followup rates that ranged from less than 2 years The only available data on emergency GEDS (Table 1) to 16 years, and variable liver reserve. are from Hassab,38 who performed 39 operations. The Despite the deficiencies in these studies, results tend initial 38% mortality rate was reduced to 12% with to support the use of DSRS in pure schistosomal portal o.20 o ,My2005 May 5, No. 200, Vol.

Table 1. Hassab (GEDS) Procedure Operative Emergency Variceal Survival Author and Type of study Total, n Elective Emergency Prophylaxis mortality mortality rebleeding Encephalopathy 5-y country (duration, y) (schistosomiasis) (n) (n) (n) (%) (%) (%) (%) (%) Hassab,38 Egypt Retrospective (12) 232 (178) ns ns 92 10.0 12.0 12.0 Ͻ1.0 ns Lu,71 China Retrospective (10) 73 (57) 32 ns 41 ns ns 11.3 0.0 85.5 GEDS, gastroesophageal decongestion and splenectomy; ns, not specified. asbkne al et Laosebikan

Table 2. Distal Splenorenal Shunt (DSRS) Elective Emergency Therapeutic operative operative Overall Variceal Survival Type of study Schistosomiasis Therapeutic emergency Prophylaxis mortality mortality mortality rebleed Encephalopathy 5-y

Lead author and country (duration, y) (n) elective (n) (n) elective (n) (%) (%) (%) (%) (%) (%) Hypertension Portal Schistosomal Pitanga70 (Brazil) Abstract (16) 340 (340) 185 88 67 ns ns 8.0 2.8 0.3 ns Abrantes72 (Brazil) Abstract (15) 308 (308) 298 10 75 ns ns 1.9 1.9 1.3 ns Salam7 (Egypt) Retrospective (15) 312 (170) 170 0 0 2.0 0.0 8.0 4.0 6.0 92.0 Raia74 (Brazil) Prospective (4) 69 (69) 54 15 0 0.0 13.3 13.3 ns ns ns Ezzat68 (Egypt) Prospective (Ͻ2) 60 (60) 60 0 0 1.7 0.0 8.3 6.7 5.1 88.0 Aboul-Enein24 (Egypt) Prospective (Ͻ2) 35 (35) 35 0 0 0.0 0.0 0.0 0.0 0.0 100.0 Modiba61 (South Africa) Prospective (7) 41 (32) 30 2 0 6.0 ns ns 6.0 3.0 ns Jin64 (China) Review (10) 302 (28) 190 10 102 6.0 20.0 ns 8.0 1.0 72.0–100.0 Bessa77 (Egypt) Prospective (Ͻ2) 20 (20) 20 0 0 10.0 0.0 15.0 0.0 0.0 85.0 Orozco83 (Mexico) Review (20) 177 (22) 167 10 0 ns ns 14.0 6.0 7.0 69.0 ns, not specified. 799 800 Laosebikan et al Schistosomal Portal Hypertension J Am Coll Surg hypertension with good short- and intermediate-term type of operation rather than disease (DSRS, 80%; outcomes. GEDS, 79%). This is in keeping with the 92% 5-year survival in schistosomal patients reported by Salam and Operation for schistosomiasis: comparative studies colleagues7 compared with 76% and 65% in cirrhotics Available comparative data of the various approaches are and mixed etiology patients, respectively. Myburgh’s se- scant. Five studies gave comparative data (Table 3).22,76,78-80 ries41 with a 14 years followup showed superior out- Followup ranged from 4 to 9 years. Operations were per- comes in the noncirrhotic group, which contained 5 formed at intervals varying from 1 to 6 weeks after stabili- schistosomal patients. Variceal rebleeding and encepha- zation. In two of the studies, fewer than half of the patients lopathy occur less often and remain less disabling in pa- had pure schistosomal disease.79,80 The comparative groups tients with schistosomiasis than in cirrhosis pa- varied and included central splenorenal shunts, injection tients.7,22,48,61,62,81 The data presented in this section tend to sclerotherapy (EIS), and the modified Sugiura procedure. favor DSRS over GEDS on the basis of reduced risk of The study by Da Silva and coworkers22 established the su- rebleeding, but at the expense of a higher encephalopathy periority of DSRS over central splenorenal shunt (CSRS) rate. with respect to rebleeding, encephalopathy, and overall mortality. For the same reason, CSRS arm of the study was Splenectomy abandoned by Raia and colleagues.78 Splenectomy serves a dual purpose: devascularizing the Overall, the operative mortality, overall mortality, and gastroesophageal junction while simultaneously correct- rebleeding rates for DSRS and GEDS were similar. The ing cytopenia. Al-Awami and colleagues82 had 6 deaths encephalopathy rate was lower with GEDS (2.3%, range in 48 hypersplenic portal hypertensive patients treated 0% to 6%) compared with DSRS (12.3%, range 10% to with splenectomy. In a study of 113 schistosomal pa- 14.8%). In the study by Raia and associates,78 overall tients, El Masri and Hassan83 reported a 20% mortality mortality was significantly lower (50%) in GEDS pa- and a 50% rebleeding rate. tients, a fact they attributed to their long 5- to 10-year In 25 children with hepatosplenic schistososmiasis, a followup. prohibitive mortality of 30% to 40% was reported be- Statistical projections by Raia and assoicates59 suggest cause of overwhelming sepsis and recurrent bleeding.32 that a series of at least 400 patients in each group would This approach, coupled with autotransplantation of possibly demonstrate a significant difference between 100 g of spleen and benzyl penicillin every 3 weeks, is DSRS and GEDS. These comparative data tend to favor currently subject to analysis.37 GEDS over DSRS. It is concluded that splenectomy alone is a poor option, especially in children. In the young, a DSRS is Surgery for schistosomiasis: comparison with feasible62 and preferable to better preserve immunologic nonschistosomal disease function. The outcomes of GEDS performed in pure schistosomiasis are comparable to those of the Sugiura procedure in Endotherapy nonschistosomal disease.20,58 The study by Da Silva and Variceal eradication with endotherapy can be achieved coauthors22 showed better survival in patients with schis- by injection of sclerosants (EIS) and band ligation tosomiasis subjected to total or selective shunts when (EBL). Endotherapy is now used for both immediate compared with patients in other trials, including those and longterm treatment of esophageal varices worldwide with alcoholic cirrhosis. DSRS showed a clear advantage because of its simplicity and the high surgical mortality over CSRS. in those with poor liver reserve or unsuitable for other Comparing DSRS and GEDS in both schistosomiasis therapeutic modalities.47,84 Repeated sessions are re- patients and those with cirrhotic/mixed disease, the sur- quired for the high rebleeding rate (between 20% and vival rate in the schistosomal group was 75% for GEDS 50%), and there is a 10% longterm failure rate.85-87 Data and 90% for DSRS.76 For the cirrhotic and mixed group, are conflicting with regard to survival advantage.84,85 there was no difference between therapies, with rates of EBL was introduced in 1986, and various prospective 73% and 72%, respectively. The same report showed randomized studies88 have established its superiority that longterm survival was similar when stratified by over EIS.89 It is standardized and highly effective in the o.20 o ,My2005 May 5, No. 200, Vol.

Table 3. Comparative Studies on Surgical Procedures Operative Overall Variceal Survival Lead author Type of study n Elective Emergency Pt mortality mortality rebleed Encephalopathy >5y and country (duration, y) (schistosomiasis) (n) (n) Procedure No. (%) (%) (%) (%) (%) Raja67 (Brazil) Prospective (4) 94 (94) 94 0 DSRS 30 ns 14.8 22.2 14.8 ns GEDS 32 ns 7.1 21.4 0.0 ns CSRS† 32 abandoned Abdel-Wahab*79 (Egypt) Retrospective (9) 225 (80) 103 36 DSRS 64 1.56 7.8 4.6 9.3 ns GEDS 75 1.35 12.0 17.3 1.3 ns EIS* 86 ns 18.0 21.0 ns ns Ezzat68 (Egypt) Prospective (5) 66 (66) 66 0 DSRS 50 3.3 ns 10.0 10.0 90.0 GEDS 16 3.1 ns 19.0 6.0 75.0 Da Silva22 (Brazil) Prospective 62 (62) 62 0 DSRS 30 ns 0.0 7.0 7.0 ns CSRS 32 ns 6.0 13.0 26.0 ns Abouna80 (Kuwait) Prospective (6) 50 (17) 13 4 DSRS 4 14.3 7.2 14.3 14.3 79.0 Sugiura 10 7.7 0.0 3.4 0.0 93.0 *Gastric varices. †CSRS abandoned because of prohibitive mortality. CSRS, central splenorenal shunt; DSRS, distal splenorenal shunt; EIS, injection of sclerosants; GEDS, gastroesophageal decongestion and splenectomy; ns, not specified. asbkne al et Laosebikan

Table 4. Endotherapy Lead author Type of study Total patients Schistosomiasis Sessions Variceal Efﬁcacy Mortality, and country (duration, mo) (schistosomiasis) Therapy (n) (n) rebleeding (%) (%) 5-y (%) Sakai90 (Brazil) Retrospective (48–132) 97 (97) EIS 22 7.56 50.0 72.7 ns EIS after surgery 75 4.93 36.0 97.3 ns Hypertension Portal Schistosomal Elsayed94 (Egypt) Prospective (17–24) 178 (91) EIS 45 4 38.6 52.0 12.0 EIS ϩ propranolol 46 4 14.3 52.0 12.0 Bessa91 (Egypt) Prospective (14–31) 60 (60) EIS (acute bleeding) 20 5 0.0 85.0 25.0 EIS (unfit for surgery) 20 6 20.0 100.0 30.0 EIS (postsurgery) 20 5 15.0 100.0 20.0 Cordeiro92 (Brazil) Prospective (60) 50 (50) EIS 32 ns 28.1 ns 3.1 CONTROL 18 ns 44.5 ns 27.7 Siqueira53 (Brazil) Prospective (18) 40 (40) EIS 20 3.72 5.0 90.0 ns EBL 20 3.05 0.0 100.0 ns EBL, band ligation; EIS, injection of sclerosants; ns, not specified. 801 802 Laosebikan et al Schistosomal Portal Hypertension J Am Coll Surg presence of acute bleeding and at controlling large and Child’s A and only 6% had previous operations. In the spurting varices. It has a higher eradication rate, with fewer study by Al Karawi and colleagues,85 68% of the patients sessions, lower rebleeding rate, and fewer procedure-related with schistosomiaisis were Child’s A compared with 30% complications and mortality. with nonschistosomal disease. Both showed that mortality A summary of studies on endotherapy is seen in Table 4. rates were higher in the nonschistosomal group, despite a Interseries comparison is hampered because of different higher rebleeding rate in the schistosomal group in Al therapy or population comparators. Kawari’s study. It can be concluded from these studies that Sakai and associates’90 analysis of EIS showed that mortality is related to liver reserve or previous surgery rather bleeding control was superior in the postsurgery group, than the EIS per se. Compliance remains the Achilles’ heel confirming the efficacy of salvage EIS reported by Ezzat of endotherapy,88 so some authors prefer the one-time op- and coworkers76 and Bessa and Helmy.91 tion of operation for longterm treatment.61 Cordeiro92 documented a survival benefit for EIS over a nontreatment control group. This advantage was of Secondary drug prophylaxis such a magnitude that future studies should use another Secondary drug prophylaxis refers to the longterm use of treatment arm as comparator rather than a no-treatment pharmacologic agents to prevent rebleeding and reduce control. mortality. Propranolol has been found effective, reduc- Excluding patients with gastric varices who are best ing rebleeding risk in cirrhotic patients by 30%.54 Con- treated with acrylate glue,93 Elsayed and colleagues94 traindications include chronic airway disease, peripheral showed that the addition of propranolol was more effec- vascular disease, diabetes, and pregnancy, and compli- tive in the longterm than EIS alone. ance remains a problem.51,54,84 The few studies done in Siqueira and coworkers’53 small series is the only one schistosomal patients who had bled are summarized in to compare EIS with EBL, and followup was no longer Table 6. than 1 year. They showed that variceal eradication with In a study with only a 1-year followup, Kiire97 re- EBL was obtained faster with fewer complications, but ported a fourfold reduction in rebleeding and a mortal- these differences failed to reach significance. ity rate cut in half because of propanolol. The latter In summary, EIS after operation or with propranolol failed to reach statistical significance. A larger study with appears to be more effective than EIS alone. EBL may be patients followed up over 2 years, with a dose-response superior to EIS, but the optimal endotherapy may be a propanolol treatment arm, showed a significant reduc- combined approach starting with EBL when varices are tion in rebleeding and mortality.98 This was attributed to large and as they are shrunk, EIS84,95 using small volumes effective beta-blockade, with the reduction being greater of sclerosant. This requires fewer sessions than EBL in those with a large liver and increased portal vein alone without increasing complications. diameter. Mies and colleagues,37 in a small study using intensive Endotherapy comparison of schistosomiasis with hemodynamic monitoring, shed some insight into the nonschistosomal disease pathophysiology. Patients were to receive propanolol. Table5 summarizes the results of the two comparative stud- Those achieving a Ͼ30% reduction in portal vein were ies of EIS. In Sakai and coworkers’96 nonrandomized study, to managed with endotherapy and continued propano- all patients with schistosomiasis were Child’s A, and 69% lol therapy. Clinically stable beta-blockade was attained had previous operations. In the cirrhotic group, 35% were with very high doses of propranolol, taking an average of 21

Table 5. Comparative Studies of Endotherapy Variceal Lead author Type of study Sessions rebleeding Efﬁcacy Mortality, and country (duration, y) Group n (n) (%) (%) 5 y (%) Sakai96 (Brazil) Prospective (2) Schistosomiasis 78 4–6 13.0 ns 1.0 Nonschistosomisis 71 4–6 48.0 ns 55.0 Al Karawi85 (Saudi Arabia) Prospective (6) Schistosomiasis 45 5.1 22.2 78.0 24.4 Nonschistosomisis 66 5.5 7.5 59.0 44.0 ns, not specified. Vol. 200, No. 5, May 2005 Laosebikan et al Schistosomal Portal Hypertension 803

Table 6. Secondary Prophylaxis Variceal Lead author and Type of study Total patients Schistosomiasis rebleeding Mortality, country (duration, y) (schistosomiasis) Therapy (n) (%) 5 y (%) El Tourabi98 (Sudan) Prospective (2) 82 (82) Propranolol 42 1.0 3.0 Placebo 40 8.0 7.0 Kiire97 (Zimbabwe) Prospective (1) 50 (37) Propranolol 17 20.0 4.0 Placebo 18 80.0 20.0 Mies37 (Brazil) Prospective (ns) 11 (11) Propranolol ϩ surgery 11 ns ns Propranolol ϩ EIS Abandoned EIS, injection of sclerosants; ns, not specified. days of treatment. The desired reduction in portal pressure access to tertiary medical facilities. There are no con- was not achieved and all patients had operations. trolled trials to support the superiority of the available Beta-blockade did produce a significant reduction of options. Splenectomy alone is plagued by a high rebleed- the mean arterial pressure, a 15% transient reduction of ing rate and its combination with endotherapy is yet to cardiac output with no change in heart rate, a decrease in be evaluated. DSRS effectively controls variceal bleed- azygous blood flow, and compensated pulmonary hy- ing, with some risk of encephalopathy, especially in the pertension. This lack of desired clinical effect was be- presence of mixed pathology. GEDS provides very low lieved to be because of increased hepatic blood flow and encephalopathy rates, though there is less effective con- the first-pass effect in the functionally preserved liver. trol of bleeding, which may require salvage endotherapy. They concluded that propranolol corrects the hyperdy- In the presence of massive splenomegaly or severe hyper- namic circulation, reducing the entire azygous and por- splenism it should be preferred to DSRS. tal venous flow, but maintains total hepatic blood flow by increasing hepatic arterial flow. It is this reduction in azygous blood flow rather than a reduction in portal REFERENCES venous pressure that offers protection from rebleeding. 1. Abdel-Wahab MF, Strickland GT. Abdominal ultrasonography These clinical and hemodynamic studies support for assessing morbidity from schistosomiasis. 2. Hospital stud- nonselective beta-blocker use in secondary prophylaxis, ies. Trans R Soc Trop Med Hyg 1993;87:135–137. but higher doses are likely to be required in schistosomal 2. Mohamed A, Elsheikh A, Ghandour Z, Al Karawi M. Impact of hepatitis C virus infection on schistosomal liver disease. Hepa- patients when compared with cirrhotic patients. togastroenterology 1998;45:1492–1496. In conclusion, determination of the ideal treatment 3. Biays S, Stich AH, Odermatt P, et al. A foci of Schistosomiasis relies on our understanding of the disease mechanism, mekongi rediscovered in Northeast Cambodia: cultural percep- which, at present, requires further clarification. With tion of the illness; description and clinical observation of 20 severe cases. Trop Med Int Health 1999;4:662–673. complete eradication of the parasitosis, liver and splenic 4. Ravera M, Reggiori A, Cocozza E, et al. Clinical and endoscopic changes appear reversible, but longterm followup is re- aspects of hepatosplenic schistosomiasis in Uganda. Eur J Gas- quired to critically evaluate the results of any treatment troenterol Hepatol 1996;8:693–697. modality because patients with portal hypertension may 5. McCormick PA. Pathophysiology and prognosis of esophageal varices. Scand J Gastroenterol 1994;207[Suppl]:1–5. go for 10 years or more between successive episodes of 6. McCormick PA, Burroughs AK. Relation between liver pathol- bleeding. Public health measures should be used in areas ogy and prognosis in patients with portal hypertension. World of high endemicity. Compliance remains the main det- J Surg 1994;18:171–175. rimental factor to the longterm success of propranolol in 7. Salam AA, Ezzat FA, Abu-Elmagd KM. Selective shunt in schistosomiasis in Egypt. Am J Surg 1990;160:90–92. primary and secondary prophylaxis, and endotherapy, 8. Symmers W. Note on a new form of liver cirrhosis due to the which has become the more favored practice, especially presence of ova of bilharzia haematobia. J Pathol Bacteriol 1903; in high-risk patients and surgical failures. EBL appears 9:237–239. equally effective as, and safer than, EIS. 9. Kamal S, Madwar M, Bianchi L, et al. Clinical, virological and Surgery provides a one-time treatment of variceal histopathological features: long-term follow-up in patients with chronic hepatitis C co-infected with S. mansoni. Liver 2000;20: bleeding, especially for the noncompliant patient, in 281–289. portal hypertensive gastropathy, and in those without 10. Bassily S, Dunn MA, Farid Z, et al. Chronic hepatitis B in 804 Laosebikan et al Schistosomal Portal Hypertension J Am Coll Surg

patients with schistosomiasis mansoni. J Trop Med Hyg pathological syndromes in male CBA/J inbred mice with 1983;86:67–71. chronic Schistosoma mansoni infections. Am J Pathol 11. Hoffmann KF, Caspar P,Cheever AW, Wynn TA. IFN-gamma, 1993;142:703–714. IL-12, and TNF-alpha are required to maintain reduced liver 30. Cerri GG, Alves VA, Magalhaes A. Hepatosplenic schistosomi- pathology in mice vaccinated with Schistosoma mansoni eggs asis mansoni: ultrasound manifestations. Radiology 1984;153: and IL-12. J Immunol 1998;161:4201–4210. 777–780. 12. Colley DG. Immune responses to a soluble schistosomal egg 31. Davidson RN, Houston S, Kiire CF. Schistosomal periportal antigen preparation during chronic primary infection with fibrosis in Zimbabwe: use of ultrasound in patients with esoph- Schistosoma mansoni. J Immunol 1975;115:150–156. ageal varices. Trans R Soc Trop Med Hyg 1991;85:380–382. 13. Da Silva LC, Carrilho FJ. Hepatosplenic schistosomiasis. Patho- 32. Brandt CT, Caneca OA, Tavares DJ, Avila JL. Surgical hepato- physiology and treatment. Gastroenterol Clin North Am 1992; splenic schistosomiasis mansoni in children: a Doppler duplex 21:163–177. study of the portal vein and the hepatic artery. Trans R Soc Trop 14. Francis TI. Hepatic fibrosis due to Schistosoma Mansoni in Med Hyg 1995;89:70–71. Nigerians. Trop Geogr Med 1971;23:239–245. 33. Abdel-Wahab MF, Esmat G, Farrag A, et al. Grading of hepatic 15. Dessein AJ, Hillaire D, Elwali NE, et al. Severe hepatic fibrosis schistosomiasis by the use of ultrasonography. Am J Trop Med in Schistosoma mansoni infection is controlled by a major locus Hyg 1992;46:403–408. that is closely linked to the interferon-gamma receptor gene. 34. Hatz C, Jenkins JM, Ali QM, et al. A review of the literature on Am J Hum Genet 1999;65:709–721. the use of ultrasonography in schistosomiasis with special refer- 16. Amiri P,Locksley RM, ParslowTG, et al.Tumournecrosis factor ence to its use in field studies. 2. Schistosoma mansoni. Acta alpha restores granulomas and induces parasite egg-laying in Trop 1992;51:15–28. schistosome-infected SCID mice. Nature 1992;356:604–607. 35. Homeida M, Abdel-Gadir AF, Cheever AW, et al. Diagnosis of 17. Wynn TA, Sher A. IL-12 as an adjuvant for vaccines designed to pathologically confirmed Symmers’ periportal fibrosis by ultra- prevent infection and immunopathology by schistosomes. Res sonography: a prospective blinded study. Am J Trop Med Hyg Immunol 1995;146:582–590. 1988;38:86–91. 18. Souza MR, Toledo CF, Borges DR. Thrombocytemia as a pre- 36. Denie C, Vachiery F, Elman A, et al. Systemic and splanchnic dictor of portal hypertension in schistosomiasis. Dig Dis Sci hemodynamic changes in patients with hepatic schistosomiasis. 2000;45:1964–1970. Liver 1996;16:309–312. 19. Aboul-Enein A, Arafa S, Sakr M. Pathogenesis of varices in 37. Mies S, Neto OB, Beer A Jr, et al. Systemic and hepatic hemo- schistosomal portal hypertension. Dig Dis Sci 1994;39:39–45. dynamics in hepatosplenic Manson’s schistosomiasis with and 20. Hassab MA, Younis MT, el Kilany MS. Gastroesophageal decon- without propranolol. Dig Dis Sci 1997;42:751–761. gestion and splenectomy in the treatment of esophageal varices 38. Hassab MA. Gastroesophageal decongestion and splenectomy secondary to bilharzial cirrhosis: anatomical and experimental stud- in the treatment of esophageal varices in bilharzial cirrhosis: ies. Surgery 1968;63:731–737. further studies with a report on 355 operations. Surgery 1967; 21. McCormick PA, Biagini MR, Dick R, et al. Octreotide inhibits 61:169–176. the meal-induced increases in the portal venous pressure of cirrhotic patients with portal hypertension: a double-blind, 39. Hassab MA. Gastro-esophageal decongestion and splenectomy placebo-controlled study. Hepatology 1992;16:1180–1186. GEDS (Hassab), in the management of bleeding varices. Review 22. Da Silva LC, Strauss E, Gayotto LC, et al. A randomized trial for of literature. Int Surg 1998;83:38–41. the study of the elective surgical treatment of portal hyperten- 40. Conn HO. A randomized comparison of three types of surgery sion in mansonic schistosomiasis. Ann Surg 1986;204:148– in schistosomal portal hypertension: many fewer answers than 153. questions. Hepatology 1994;20:526–528. 23. Kardorff R, Olveda RM, Acosta LP, et al. Hepatosplenic mor- 41. Myburgh JA. Selective shunts: the Johannesburg experience. bidity in schistosomiasis japonica: evaluation with Doppler Am J Surg 1990;160:67–74. sonography. Am J Trop Med Hyg 1999;60:954–959. 42. Colley DG, Garcia AA, Lambertucci JR, et al. Immune re- 24. Aboul-Enein A. Esophageal varices in schistosomiasis: treatment sponses during human schistosomiasis. XII. Differential respon- with a modified Warren’s distal splenorenal shunt. J Cardiovasc siveness in patients with hepatosplenic disease Am J Trop Med Surg (Torino) 1985;26:164–167. Hyg 1986;35:793–802. 25. Ross AG, Sleigh AC, Li Y, et al. Schistosomiasis in the People’s 43. Prescott NM. The economics of schistosomiasis chemotherapy. Republic of China: prospects and challenges for the 21st cen- Parasitol Today 1987;3:21–24. tury. Clin Microbiol Rev 2001;14:270–295. 44. Colley DG, Evan SW. Immunoregulation and World Health 26. Kardorff R, Gabone RM, Mugashe C, et al. Schistosoma Assembly resolution 54.19: why does treatment control morbid- mansoni-related morbidity on Ukerewe Island, Tanzania: clini- ity? Parasitol Int 2004;53:143–150. cal, ultrasonographical and biochemical parameters. Trop Med 45. Mohamed-Ali Q, Doehring-Schwerdtfeger E, Abdel-Rahim Int Health 1997;2:230–239. IM, et al. Ultrasonographical investigation of periportal fi- 27. de Cleva R, Pugliese V, Zilberstein B, et al. Systemic hemody- brosis in children with Schistosoma mansoni infection: re- namic changes in mansonic schistosomiasis with portal hyper- versibility of morbidity seven months after treatment with tension treated by azygoportal disconnection and splenectomy. praziquantel. Am J Trop Med Hyg 1991;44:444–451. Am J Gastroenterol 1999;4:1632–1637. 46. Flores-Villanueva PO, Zheng XX, Strom TB, Stadecker MJ. 28. Andrade ZA, Peixoto E. Pathology of periportal fibrosis involu- Recombinant IL-10 and IL-10/Fc treatment down-regulate egg tion in human schistosomiasis. Rev Inst Med Trop Sao Paulo antigen-specific delayed hypersensitivity reactions and egg gran- 1992;34:263–272. uloma formation in schistosomiasis. J Immunol 1996;156: 29. Henderson GS, Nix NA, Montesano MA, et al. Two distinct 3315–3320. Vol. 200, No. 5, May 2005 Laosebikan et al Schistosomal Portal Hypertension 805

47. Terblanche J, Krige JE, Bornman PC. The current treatment of for management of variceal bleeding in patients with esophageal varices. S Afr J Surg 1997;35:59–62. schistosomal hepatic fibrosis. Ann Surg 1986;204:566–573. 48. Henderson JM. Role of distal splenorenal shunt for long-term 69. Abrantes WL, Carvalho MA, Rabelo GD, Drumond DAF. Dis- management of variceal bleeding. World J Surg 1994;18:205– tal splenic renal shunt in schistosomotics (Abstract of World 210. Congresses). Dig Dis Sci 1986;31:398S. 49. Maffei-Faccioli A, Gerunda GE, Neri D, et al. Selective variceal 70. Pitanga LC. Selective splenorenal anastomosis—technical de- decompression and its role relative to other therapies. Am J Surg tails and results of 340 patients subjected in 15 years (Abstract of 1990;160:60–66. World Congresses). Dig Dis Sci 1986;31:390S. 50. Warren WD, Zeppa R, Fomon JJ. Selective trans-splenic de- 71. Lu XS. [Long-term results of the Hassab operation in portal compression of gastroesophageal varices by distal splenorenal hypertension: 10 year’s follow-up]. Zhonghua Wai Ke Za Zhi shunt. Ann Surg 1967;166:437–455. 1990;28:143–146, 189. 51. Greig JD, Garden OJ, Carter DC. Prophylactic treatment of 72. Abrantes WL, de Carvalho MA, Rabelo GD, Drumond DA. patients with esophageal varices: is it ever indicated? World [Prophylactic selective splenorenal anastomosis in the hepato- J Surg 1994;18:176–184. splenic form of schistosomiasis]. AMB Rev Assoc Med Bras 52. Idezuki Y. General rules for recording endoscopic findings of 1983;29:160–162. esophagogastric varices (1991). Japanese Society for Portal Hy- 73. Bessa SM, Helmy I, Hamam SM, el Kayal ES. Esophageal tran- pertension. World J Surg 1995;19:420–422. section by the EEA stapler for bleeding esophageal varices in 53. Siqueira ES, Rohr MR, Libera ED, et al. Band ligation or scle- schistosomal hepatic fibrosis. Surg Gynecol Obstet 1988;166: rotherapy as endoscopic treatment for esophageal varices in 17–22. schistosomotic patients: results of a randomized study. HPB 74. Raia S, Mies S, Alfieri JF.Portal hypertension in mansonic schis- Surg 1998;11:27–32. tosomiasis. World J Surg 1991;15:176–187. 54. Lebrec D. Long-term management of variceal bleeding: the 75. Hassab MA. Nonshunt operations in portal hypertension with- place of pharmacotherapy. World J Surg 1994;18:229–232. out cirrhosis. Surg Gynecol Obstet 1970;131:648–654. 55. De Jesus Lopes F, Haddad C. Late clinical, biochemical, endo- 76. Ezzat FA, Abu-Elmagd KM, Aly MA, et al. Selective shunt ver- scopic and electroencephalographic evaluation of patients with sus nonshunt surgery for management of both schistosomal and schistosomal portal hypertension treated with distal splenorenal nonschistosomal variceal bleeders. Ann Surg 1990;212:97–108. shunt. Int Surg 1998;83:37–42. 77. Bessa SM, Helmy I, el Sheikh SO, et al. The distal splenorenal 56. Hassab MA. Gastroesophageal decongestion and splenectomy. shunt in patients with variceal bleeding due to schistosomal A method of prevention in treatment of bleeding esophageal hepatic fibrosis. Surg Gynecol Obstet 1987;165:143–147. varices associated with bilharzial hepatic fibrosis: preliminary 78. Raia S, da Silva LC, Gayotto LC, et al. Portal hypertension in report. J Int Coll Surgeons 1964;41:5. schistosomiasis: a long-term follow-up of a randomized trial 57. Raja S, Da Silva LC, Gayotto LC. Portal hypertension in schis- comparing three types of surgery. Hepatology 1994;20:398– tosomiasis: A long term follow-up of a randomized trial com- 403. paring three types of surgery. Hepatology 1994;20:398–403. 79. Abdel-Wahab M, el Ebidy G, Gad el-Hak N, et al. Fundal 58. Sugiura M, Futagawa S. A new technique for treating esophageal varices. J Thorac Cardiovasc Surg 1973;66:677–685. varices: problem and management. Hepatogastroenterology 59. Raia S, Mies S, Macedo AL. Surgical treatment of portal hyper- 1999;46:849–854. tension in schistosomiasis. World J Surg 1984;8:738–752. 80. Abouna GM, Baissony H, Al-Nakib BM, et al. The place of 60. Strauss E, Costa Gayotto LC, Antonelli R, et al. Systemic surgi- Sugiura operation for portal hypertension and bleeding esoph- cal shunts and splenomegaly as causes of haemolysis in portal ageal varices. Surgery 1987;101:91. hypertension in mansonic schistosomiasis. Evaluation through 81. Rikkers LF, Sorrell WT, Jin G. Which portosystemic shunt is serum levels of haptoglobin, hemopexin and bilirubins. J Hepa- best? Gastroenterol Clin North Am 1992;21:179–196. tol 1986;2:340–350. 82. al-Awami SM, al Salem AH, Knox-Macaulay H, et al. Splenec- 61. Modiba MC, Koto Z, Lowan TA, et al. Distal splenorenal shunt tomy for hematological disorders in the Eastern Province of for non-cirrhotic variceal bleeding in black South Africans. S Afr Saudi Arabia. Int Surg 1992;77:60–63. J Surg 1994;32:87–90. 83. El Masri SH, Hassan MA. Splenectomy and vasoligation for 62. Maksoud JG, Miles S, Pinto VC. Distal splenorenal shunt in patients with haematemesis secondary to bilharzial hepatic fi- children. J Pediatr Surg 1978;13:335–340. brosis. Br J Surg 1982;69:314–316. 63. Orozco H, Mercado MA, Takahashi T, et al. Role of the distal 84. Terblanche J, Stiegmann GV, Krige JE, Bornman PC. Long- splenorenal shunt in management of variceal bleeding in Latin term management of variceal bleeding: the place of varix injec- America. Am J Surg 1990;160:86–89. tion and ligation. World J Surg 1994;18:185–192. 64. Jin GL. Current status of the distal splenorenal shunt in China. 85. Al Karawi MA, Sheikh Mohamed AR, Ahmed AM, et al. Long- Am J Surg 1990;160:93–97. term outcome of endoscopic sclerotherapy of variceal bleeding: 65. Madiba TE, Thomson SR, Neijenhuis PA. Distal splenocaval comparative study between schistosomiasis and others. Hepato- shunt as an alternative to distal splenorenal shunt in a patient gastroenterology 1996;43:287–292. with a retro-aortic left renal vein. J R Coll Surg Edinb 1997;42: 86. Mohamed AE, al Karawi MA, al Otaibi R, Hanid MA. Results 359–360. of sclerotherapy in 100 patients—comparison of the outcome 66. Orozco H, Juarez F,Santillan P,et al.Ten years of selective shunts between schistosomiasis and hepatitis B. Hepatogastroenterol- for hemorrhagic portal hypertension. Surgery 1988;103:27–31. ogy 1989;36:333–336. 67. Rikkers LF. Portal hemodynamics, intestinal absorption, and 87. Hashizume M, Kitano S, Koyanagi N, et al. Endoscopic injec- postshunt encephalopathy. Surgery 1983;94:126–133. tion sclerotherapy for 1,000 patients with esophageal varices: a 68. Ezzat FA, Aly MA, Bahgat OD, et al. Distal splenorenal shunt nine-year prospective study. Hepatology 1992;15:69–75. 806 Laosebikan et al Schistosomal Portal Hypertension J Am Coll Surg

88. Tait IS, Krige JE, Terblanche J. Endoscopic band ligation of sclerotherapy and propranolol in the prevention of esophageal varices. Br J Surg 1999;86:437–446. rebleeding from esophageal varices: a randomised study. Gut 89. Stiegmann GV,Goff JS, Michaletz-Onody PA, et al. Endoscopic 1996;38:770–774. sclerotherapy as compared with endoscopic ligation for bleeding 95. Dhiman RK, Chawla YK. A new technique of combined endo- esophageal varices. N Engl J Med 1992;326:1527–1532. scopic sclerotherapy and ligation for variceal bleeding. World J 90. Sakai P,Boaventura S, Ishioka S, et al. Sclerotherapy of bleeding Gastroenterol 2003;9:1090–1093. esophageal varices in schistosomiasis. Comparative study in pa- 96. Sakai P,Boaventura S, Capacci ML, et al. Endoscopic sclerother- tients with and without previous surgery for portal hyperten- apy of bleeding esophageal varices. A comparative study of re- sion. Endoscopy 1990;22:5–7. sults in patients with schistosomiasis and cirrhosis. Endoscopy 91. Bessa SM, Helmy I. Injection sclerotherapy for esophageal var- 1988;20:134–136. ices caused by schistosomal hepatic fibrosis. Surgery 1985;97: 97. Kiire CF. Controlled trial of propranolol to prevent recurrent 164–168. variceal bleeding in patients with non-cirrhotic portal fibrosis. 92. Cordeiro F.Variceal sclerosis in schistosomotic patients: a 5-year BMJ 1989;20:298:1363–1365. follow-up study. Gastrointest Endosc 1990;36:475–478. 98. El Tourabi H, el Amin AA, Shaheen M, et al. Propranolol 93. Lo GH, Lai KH, Cheng JS. Bleeding gastric varices: Stick to the reduces mortality in patients with portal hypertension sec- sticky glue. Hepatology 2001;33:1060–1064. ondary to schistosomiasis. Ann Trop Med Parasitol 1994;88: 94. Elsayed SS, Shiha G, Hamid M, et al. Sclerotherapy versus 493–500. IMAGES FOR SURGEONS

Fibromuscular Dysplasia of the Carotid Artery

Wei Zhou, MD, Ruth L Bush, MD, FACS, Peter L Lin, MD, Alan B Lumsden, MD, FACS Baylor College of Medicine, Houston, TX

A 64-year-old woman had carotid duplex evaluation for bilateral carotid bruits. A Phillips HDI 5000 SonoCT system equipped with a linear 7-4 MHz transducer was used. The carotid vessel was examined first in transverse view, showing a normal proximal internal carotid artery (ICA) and carotid bulb. Flow was then sampled in the longitudinal axis at a 60-degree angle with the aid of color flow Doppler. Multiple areas of color flow disturbance associated with velocity increases were identified in the mid and tion from a stagnation of flow leading to distal emboli- distal ICA (A). Multiple areas of focal hyperechoic thicken- zation. Occlusion and embolization may occur as the ing alternating with thin dilated arterial wall, correlating result of dissection or aneurysm formation. with areas of color flow disturbance, were visualized on gray Surgical intervention is indicated for symptomatic scale imaging (B). Carotid angiography demonstrated a patients and patients with high-grade stenosis. Graduated string-of-beads appearance confirming fibromuscular dys- intraluminal dilatation under direct vision has been used plasia (FMD) (C). Because of minimal stenosis and absence successfully in patients continuing postoperatively on anti- of symptoms, the patient was treated medically with antiplatelet therapy. Percutanous angioplasty has become in- platelet agents and had no evidence of disease progression creasingly popular, with excellent short-term results. Med- on followup duplex at 1 year. ical management with antiplatelet therapy is an effective Carotid FMD is a rare condition evidenced in only approach for asymptomatic patients with minimal steno- 1% to 2% of routine carotid angiographies.1,2 Unlike sis.3,4 Followup is warranted to identify both disease pro- atherosclerotic disease of the carotid artery, vascular in- gression and confirm arterial patency after intervention. volvement in FMD tends to be in the segment of mid to distal ICA. Traditionally, the diagnosis is made by con- REFERENCES ventional selective angiography. With heightened suspi- 1. Morris GC Jr, Lechter A, DeBakey ME. Surgical treatment of fibromuscular disease of the carotid arteries. Arch Surg 1968; 96:636–643. cion, diagnosis of carotid FMD may be established reli- 2. Stanley JC, Gewertz BL, Bove EL, et al. Arterial fibrodysplasia. His- ably by duplex sonography in experienced hands. topathologic character and current etiologic concepts. Arch Surg 1975;110:561–566. Only 10% of patients with FMD have complications. 3. Kelly T Jr, Morris GC Jr. Arterial fibromuscular disease. Observations The symptoms are secondary to encroachment on the on pathogenesis and surgical management. Am J Surg 1982;143: 232–236. vessel lumen and a reduction of flow causing hypoper- 4. Stewart MT, Moritz MW, Smith R 3rd, et al. The natural history of fusion. Thrombi may form in the areas of mural dilata- carotid fibromuscular dysplasia. J Vasc Surg 1986;3:305–310.

Adnan Z Rizvi, MD, John A Kaufman, MD, Pamela Smith, MD, Mark L Silen, MD, MBA, FACS Oregon Health & Science University, Portland, OR

A 16-year-old female patient presented with a 2-month history of gastrointestinal bleeding and anemia. Upper and lower endoscopy, small bowel enteroclysis, and a Meckel’s technetium scan were normal. She presented to our institution with continued gastrointestinal bleeding and a hematocrit of 16%. A repeat Meckel’s technetium scan, colonoscopy, and upper endoscopy with enteroscopy to the mid-jejunum were normal. Angiography (A) revealed a vascular lesion in the ileum suggestive of an arteriovenous malformation. Laparotomy (B) confirmed a 2-cm nodular vascular malformation located 4.5 feet proximal to the ileocecal valve. A small bowel resection and primary anas- tamosis were performed. Gross inspection (C) revealed a 0.5-cm soft, red, raised nodule on the mucosal surface. Histology (D) confirmed an arteriovenous malformation. The patient was discharged on postoperative day 4 with a stable hematocrit. Arteriovenous malformations (AVM) of the small intestine are rare, and the true incidence is unknown. The terms arteriovenous malformation, angiodysplasia, telangiectasia, and hemangioma describe distinct entities but have been used interchangeably to describe the same lesion. The non- uniform classification of intestinal vascular lesions in the literature has led to some of the confusion and inaccurate reporting.

Although intestinal AVM and angiodysplasia have important because AVM may not be visible or palpable similar angiographic characteristics, the histology and intraoperatively. If a classic AVM is seen on angiography, pathophysiology are distinct. On angiography, both le- the injection of methylene blue or indigo carmine into sions show rapid filling, tortuous vessels, localized the feeding vessel marks the involved segment of bowel opaque terminal “berry”-like structures, and early filling for subsequent resection.1 Other techniques to aid in of dilated veins.1 Histologically, an AVM is an intercon- localization include transillumination, intraoperative nection of aberrant arteries and veins with thick hyper- endoscopy, selective mesenteric PO2 measurements, and trophic walls.2 These lesions are likely congenital, and intraoperative Doppler ultrasound examination. the majority are seen in patients under 30 years of age. In Once the diagnosis in made, surgical resection is the contrast, an intestinal angiodysplasia has dilated thin- usual treatment. The procedure can be performed either walled submucosal vessels, often lined only by endothe- open or laparoscopically if the location is confirmed pre- lium. Angiodysplasia represents a degenerative process operatively. Percutaneous angiographic embolization of from intermittent or partial obstruction of submucosal these lesions has been reported. For solitary AVM, resec- veins, leading to capillary dilation, resulting in an arte- tion is curative, but patients with multiple intestinal 3 riovenous connection. Intestinal angiodysplasia occurs AVM may pose a difficult surgical problem. in the elderly population and are predominately located in the right colon. Patients with small intestinal AVM present with in- REFERENCES testinal bleeding that is either massive or chronic in na- 1. Crawford ES, Roehm JO Jr, McGavran MH. Jejunoileal arterio- ture. The diagnosis of a small intestinal AVM is difficult. venous malformation: localization for resection by segmental Endoscopy is often the initial test performed to exclude bowel staining techniques. Ann Surg 1980;191:404–409. gastric, duodenal, and colonic sources of bleeding. To 2. Eastman J, Nazek M, Mangels D. Localized arteriovenous malformation of the jejunum. Arch Pathol Lab Med 1994;118:181– assess the jejunum and ileum, further studies may in- 183. clude small bowel enteroclysis, 99m technetium-labeled 3. Richardson JD. Vascular lesions of the intestines. Am J Surg red blood cell scintigraphy, angiography, and, more re- 1991;161:284–293. cently, wireless-capsule endoscopy.4 Selective mesenteric 4. Saurin JC, Delvaux M, Gaudin JL, et al. Diagnostic value of endoscopic capsule in patients with obscure digestive bleeding: angiography remains the most reliable method of diag- blinded comparison with video push-enteroscopy. Endoscopy nosing small intestinal AVM. Preoperative localization is 2003;35:576–584. SURGEON AT WORK

Percutaneous Cephalic Vein Cannulation (in the Deltopectoral Groove), with Ultrasound Guidance

Jack LeDonne, MD

Central venous access can be achieved by cannulating opened and sonoconductive gel was placed inside the various veins. Most commonly, access is achieved at the sterile sheath, and the 7.5-mHz probe was placed inside bedside through the subclavian (infraclavicular ap- the sheath. Sterile conductive gel was placed on the skin proach), the internal jugular, and the femoral veins. Less overlying the deltopectoral groove, and the cephalic vein commonly, the external jugular, the axillary, and the sub- was again visualized. The depth of the vein was noted clavian (supraclavicular approach) veins are accessed at and the appropriate-angled needle guide was attached to the bedside. Surgical exposure (cut-down) of the ce- the probe. The needle was then guided into the cephalic phalic, greater saphenous, and external jugular veins can vein under direct vision. All cephalic vein cannulations be done in the critical care setting, but is safer and more were videotaped (See Video available at http://www. convenient in an operating room. Other specialized ve- journalacs.org). nous access, such as translumbar puncture of the inferior vena cava is done extremely rarely, and must be per- Patient 1 formed in the interventional radiology suite. In August 2003, a 44-year-old man, who weighed Traditionally, central venous catheters are inserted 530 pounds, was admitted to the medical intensive care “blindly,” using various anatomic landmarks, such as the unit with pneumonia, and central venous access was clavicle, the sternal notch, sternocleidomastoid, and the requested. For optimal infection control, the axillary or carotid, to guide the needle into the appropriate vein. subclavian veins are preferable to the internal jugular.2 Since 1984, visualization of the target vein by ultrasound, The 7.5-mHz probe will penetrate only to a depth of and real-time needle guidance have been available.1 4 cm, which prevents visualization of the axillosubcla- This article describes the author’s experience with vian vein, in a patient of this size. The probe was placed sonoguided percutaneous cephalic venipuncture in four over the right deltopectoral groove. The right axillary morbidly obese patients (five attempts). These are the vein and artery were not visualized, but there was a first reported cases of central venous access using a per- prominent vein, easily seen in the center of the sonofield. cutaneous cephalic vein approach. This vein was approximately 2.5 cm deep to the skin, it was opening and closing with respiration, and impor- Technique tantly, was not accompanied by an artery. It was identi- Five central venous catheters were placed. Two were fied as the cephalic vein and was punctured with sono- placed in the medical intensive care unit (MICU), two guidance. There was free flow of blood into the syringe, were placed on a medical-surgical floor, and one port was but the guide wire would not pass. So a low right internal placed in the operating room. All patients were exam- jugular triple lumen catheter (Arrow) was placed. ined with a Site-Rite III (Dymax) portable ultrasound, as part of the procedure. Once a suitable cephalic vein Patient 2 was identified, the patient was prepared with chlorhexi- In February 2004, a 546-pound man was admitted with dine 2%, and maximal barrier protection was observed. pneumonia and respiratory failure. He was coagulo- An 18- or 21-gauge Needle Guide Kit (Dymax) was pathic with an international normalized ratio of 2.2 and on ventilatory support. The medical staff requested a

Competing interests declared: None. central venous catheter for access. The right cephalic vein was visualized ultrasonically, punctured with one Received May 12, 2004; Accepted June 14, 2004. From the Department of Surgery, Greater Baltimore Medical Center, attempt, and a triple-lumen catheter was placed without Baltimore, MD. difficulty. Correspondence address: Jack LeDonne, MD, Department of Surgery, Greater Baltimore Medical Center, 6701 N Charles St, Baltimore, MD Later in this admission, the patient underwent trache- 21204. ostomy for longterm management of his respiratory fail-

© 2005 by the American College of Surgeons ISSN 1072-7515/05/$30.00 Published by Elsevier Inc. 810 doi:10.1016/j.jamcollsurg.2004.06.025 Vol. 200, No. 5, May 2005 LeDonne Percutaneous Cephalic Vein Cannulation 811 ure. The pulmonologist requested placement of a port taneous cannulation would appear to be infinitely easier (Bard Access) because difficult venous access was antic- for the patient and the physician. ipated in the future. In the operating room, the left 3. The safety (ie, risk reduction) ramifications are enor- cephalic vein was visualized and punctured, and blood mous.6,7 Approaching the cephalic vein percutaneously in flow without resistance was noted. But, the guide wire a 540-pound, coagulopathic patient on a ventilator elimi- would not pass. At this point, the left subclavian vein nates pneumothorax as a potential complication.The mor- was punctured blindly, but again the guide wire would bidity from cephalic hemorrhage does not approach the not advance. A left internal jugular approach was per- potential catastrophe of subclavian or jugular hemorrhage. formed without difficulty. How do we know that the vein in question is the cephalic? Basically, the same way that we identify the Patient 3 cephalic vein in the course of surgical exposure. That is, In February 2004, a 300-pound man underwent a series by detecting a prominent venous structure in the deltopec- of emergent thrombectomies and subsequent below- toral groove, one that is not accompanied by an artery. In knee amputation for acute arterial thrombosis. Central conclusion, sonoguided, percutaneous cephalic vein venous access was requested and the right cephalic vein cannulation is a safe, effective, and convenient method was easily identified with ultrasound. The vein was of obtaining central venous access in a difficult patient punctured in one attempt and a triple-lumen catheter population. placed without difficulty.

Patient 4 REFERENCES In April 2004, a 400-pound woman was admitted with 1. Legler D, Nugent M. Doppler localization of the internal jugular a urinary tract infection to a floor bed. There was no vein facilitates central venous cannulation. Anesthesiology 1984; peripheral access, so central access was requested. A 60:481–482. small left cephalic vein was noted, and it was thought to 2. O’Grady NP, et al. Guidelines for the prevention of intravascular be marginally acceptable. A Micro-puncture kit (Cook) catheter-related infections. Mortality and Morbidity Weekly Report (Centers for Disease Control and Prevention) 2002;51:RR–10. was opened. The 21-gauge needle was inserted into the 3. Mermel LA, McCormick RD, Springman SR, Maki DG. The vein after two attempts, and the 0.018-inch guide wire pathogenesis and epidemiology of catheter-related infection with passed easily, followed by a triple-lumen catheter. pulmonary artery Swan-Ganz catheters: a prospective study uti- lizing molecular subtyping. Am J Med 1991;91(suppl):S197– Discussion S205. 4. Goetz AM, Wagener MM, Miller JM, Muder RR. Risk of infec- Sonoguided percutaneous cephalic vein cannulation has tion due to central venous catheters: effect of site placement and limited usefulness. It will not supplant the usual routes for catheter type. Infect Control Hosp Epidemiol 1998;19:842–845. central venous access, namely, the axillary-subclavian, the 5. Merrer J, De Jonghe B, Golliot F, et al. Complications of femoral internal and external jugular, and the femoral veins. But and subclavian venous catheterization in critically ill patients: a randomized controlled trial. JAMA 2001;286:700–707. in selected patients, the cephalic vein offers considerable 6. Randolph AG, Cook DJ, Gonzales CA, Pribble CG. Ultrasound benefits: guidance for placement of central venous catheters: a metaanalysis of the literature. Crit Care Med 1996;24:2053–2058. 1. The catheter exit site lies on the chest wall, not on the neck 7. Rothschild JM. Ultrasound guidance of central vein catheteriza- or the groin. This is a favored location from the infection tion. In: Markowitz AJ, ed. Making health care safer: A critical control perspective.2-5 analysis of patient safety practices. Agency for Healthcare Re- 2. In comparison with surgical exposure (cut-down), percu- search and Quality 2001:244–252. LETTERS

Complete Esophageal Diversion: sion of the primary repair was necessary; another needed A Simpliﬁed, Easily Reversible resection of the repair, using Ivor Lewis esophagogastrec- Technique tomy and esophagogastrostomy. Six other extremely sick patients were treated with simple, cervical diversion and exclusion, without any attempt at primary repair, as de- Paul D Kiernan, MD, FACS, John Rhee, MD, FACS, scribed previously and elsewhere.2 All but one patient Lucas Collazo, MD, FACS, Vivian Hetrick, RN, recovered nicely, and none required delayed primary re- Betty Vaughan, RN, Paula Graling, RN pair or reconstruction. Mean hospitalization in this Falls Church, VA group was 30 days. One patient, operated on after incur- We read with interest the article by Koniaris and col- ring esophageal perforation at coronary revascularization with left ventricular aneurysm repair, died from brain death leagues1 in the December 2004 issue: “Complete esoph- secondary to air embolism related to the aneurysm repair. ageal diversion: A simplified, easily reversible tech- There were no identifiable problems related to the patient’s nique.” The authors describe how, in addition to diversion and exclusion at the time of death. Otherwise, all primary repair and tube drainage, “complete esophageal patients recovered nicely, with simple outpatient take- diversion can be extremely helpful in controlling ongo- down and closure of the cervical esophagocutaneous fis- ing thoracic contamination and sepsis” secondary to tula, usually performed 3 to 4 months later. Only three thoracic esophageal perforation. We, too, have resorted of the six patients required any esophageal dilatation at to diversion, usually when ongoing sepsis prohibits pri- the distal cervical or gastroesophageal ligation site, and mary repair or resection and reanastomosis. those patients, only once. All perforations healed natu- Koniaris and colleagues1 perform primary repair and rally, and none of these six patients has any dysphagia at tube drainage in addition to proximal esophageal diver- mean followup of 36 months. sion; we more often make no attempt at primary repair in what should be a highly select group of patients “too sick to operate on.” We have found that with complete REFERENCES esophageal diversion and exclusion, neither primary re- 1. Koniaris LG, Spector SA, Staveley-O’Carroll KF. Complete pair nor resection and reconstruction are necessary.2 Our esophageal diversion: A simplified, easily reversible technique. method of proximal (cervical) diversion is essentially as J Am Coll Surg 2004;199:991–993. 1 2. Kiernan PD, Sheridan MJ, Elster E, et al. Thoracic esophageal described by Koniaris and colleagues; we use three perforations. South Med J 2003;96:158–163. number-1 chromic catgut ligatures as our absorbable 3. Urschel HC Jr, Razzuk MA, Wood RE, et al. Improved manage- ligatures. As described by others,3 distal esophageal di- ment of esophageal perforation: Exclusion and diversion in conversion of gastric reflux must also be accomplished for tinuity. Ann Surg 1974;179:587–591. complete esophageal exclusion. So, we similarly ligate the esophagogastric junction, placing gastrostomy for suction, jejunostomy for nutrition, and tube thoracos- Reply tomy for pleural drainage. From March 3, 2000, through December 10, 2004, Kevin F Staveley-O’Carroll, MD, PhD we have treated 9 patients as described previously out of Hershey, PA a total of 27 patients who presented with thoracic esophageal perforation during the same time period. Three Seth A Spector, MD patients were treated with primary repair and cervical Leonidas G Koniaris, MD, FACS diversion as described by Koniaris and associates,1 all Miami, FL surviving, with mean hospitalization of 36 days. One patient required subsequent esophageal dilatation, and We appreciate the insightful comments by Kiernan and two patients required reoperation. In one patient, revi- coworkers about our recent article.1 They clearly have a

© 2005 by the American College of Surgeons ISSN 1072-7515/05/$30.00 Published by Elsevier Inc. 812 doi:10.1016/j.jamcollsurg.2005.01.023 Vol. 200, No. 5, May 2005 Letters 813 long and successful experience in the treatment of esopha- 1. The authors do not indicate whether the fiberoptic exam- geal perforations.2 The points made demonstrate a large inations were performed by the same examiner, nor do they degree of agreement in both principles and technique of indicate if the examinations were recorded for comparison. operative approach. As such, we encourage surgeons to The impression is that they simply made a clinical judg- familiarize themselves with the steps described.1,2 We ment, without impartial observation by blinded reviewers, also agree with Kiernan and coworkers that primary re- that the mobility, vocal cord lengthening, and sensation were normal. pair of thoracic perforations may not be essential, par- 2. Fiberoptic examination, even with video recording, is not ticularly after delayed diagnosis or when thoracotomy is the standard to determine paresis. The usual finding in a contraindicated, but our experience consists of only two subtle vocal cord paresis is increased amplitude of mucosal such patients. In 1 patient, conservative management was wave on the affected side. Usually, the paretic side has used after a 1-month delay in diagnosis; subsequently we normal mobility. Amplitude of mucosal wave can be de- encountered a persistent extraesophageal collection that has termined only on stroboscopy, and cannot be seen with remained slow to resolve. Regardless, we agree that the op- solid light fiberoptic laryngoscopy. timal algorithm for esophageal perforation should include 3. The gold standard to determine recurrent laryngeal nerve proximal diversion, wide thoracic drainage, and G-tube injury is electromyography. Subtle findings of paresis that placement. Based on our experience, primary repair of are suspected on stroboscopy must be confirmed by laryn- the perforation should be undertaken if the patient can geal electromyography before a definite diagnosis of recur- tolerate it. We also do not advocate ligating the gastro- rent nerve injury can be ascertained. Laryngeal electromyography can also determine injury to the motor branch esophageal junction; instead, we prefer to simply place a of the superior laryngeal by testing the cricothyroid G-tube for decompression. Regardless of the details, a muscle. thoughtful approach to managing esophageal perforation is required. Esophageal diversion as described is the Overall, use of simple (unrecorded?) solid light fiber- least morbid approach and is critical for controlling up- optic laryngoscopy cannot be used to determine that an per contamination. injury to the recurrent laryngeal nerve has not occurred. The only accurate method to state that a nerve injury has not occurred is laryngeal electromyography. REFERENCES The author’s assertion that “the ILN was not dam- 1. Koniaris LG, Spector SA, Staveley-O’Carroll KF. Complete aged” is completely unsubstantiated, and their premise esophageal diversion: A simplified, easily reversible technique. that “voice changes after thyroidectomy without recur- J Am Coll Surg 2004;199:991–993. rent laryngeal nerve injury” is not defensible. 2. Kiernan PD, Sheridan MJ, Elster E, et al. Thoracic esophageal perforations. South Med J 2003;96:158–163. The importance of this distinction is not trivial. The recurrent laryngeal nerve is a delicate nerve whose response to manipulation is entirely unpredictable. The assertion that these 40 patients had voice changes that were independent of recurrent laryngeal nerve injury is Voice Changes after Thyroidectomy thoroughly misleading to thyroid surgeons.Thyroid sur- Without Recurrent Laryngeal geons must maintain complete vigilance to their surgical Nerve Injury dissections, and must understand that many patients will suffer voice changes because of manipulations of the Allen D Hillel, MD, FACS recurrent nerve, despite their best efforts. Any thought Seattle, WA that voice changes after thyroidectomy are not a result of recurrent laryngeal nerve injury is likely to decrease this In their article, Sinagra and colleagues1 propose that vigilance. 87% (40 of 46) of patients noted voice changes after thyroidectomy that were not a result of recurrent laryn- REFERENCE geal nerve injuries. The assertion that the recurrent la- 1. Sinagra DL, Montesinos MR, Tacchi VA, et al. Voice changes ryngeal nerve was not injured is unsubstantiated. There after thyroidectomy without recurrent laryngeal nerve injury. are three deficits in the authors’ methodology. J Am Coll Surg 2004;199:556–560. 814 Letters J Am Coll Surg

Centennial Perspective on In the last decade of the 20th century, the pernicious Burn Treatment effector was found to target the immune system, eliciting high levels of interleukin-1 and interleukin-2 and instigat- Martin Allgöwer, MD FACS(HON), ASA(HON), FRCS(HON) ing activation-induced cell death of lymphocytes. The pa- Basel, Switzerland tient, in this same decade, had been discovered to have a cascade of cytokines, a systemic inflammatory response, In your Centennial Perspective on burn treatment,1 Profes- and an immune system incapable of dealing with infec- th sor Burke skims over the 20 century and highlights the tion.6 It seems this effector is the reason necrotic tissue importance of federal financing of basic research and insti- causes “the life-threatening physiologic, anatomic, immu- tutional organization of centers of expertise focusing the nometabolic and bacterial events.” Interestingly, the de- burn team’s attention on its patients, as forces behind those scription of burn patients as having a systemic inflamma- improvements in burn care that developed mostly in the tory response only confirms what was revealed one century last 50 years. Indeed, we learn that in the first half of the earlier. At the end of the 19th century, Bardeen,7 after sev- century, surgeons’ efforts were limited to attempts to pre- eral autopsies of burn victims, described the condition as an serve life by relief of complications, with the hope that “internal inflammation.” natural host resistance and repair mechanisms might pre- Professor Burke leaves it to surgeons of the 21st cen- vail. Intensified research efforts to manage battle casualties tury to completely solve the problems posed by the el- during wartime led to appreciation of the role of fluids for derly, very extensive burns, and those with inhalation shock therapy.This extended the patient’s life long enough injury. Indeed, they might achieve this very soon, armed that at a later period, infection developed, which was then with the knowledge of this pernicious effector, for it has dealt with by antibiotics. These treatments “did not dra- been found that topical application of cerium nitrate matically improve overall results . . . [but] . . . delayed the fixes and denatures the effector in the necrotic tissue, time of death” and “the overall death rate after extensive preventing its migration into the circulation, and conse- burn injury...wasnotreduced very much.” quently preventing immune system suppression and even Professor Burke then reveals that the major improve- death.6 Elderly and very extensively burned patients treated ment in survival rates occurred with use of very early exci- with cerium nitrate have become survivors.8 sion that removes the “necrotic tissue (the source of the Surgeons who will use cerium nitrate will have a much physiologic abnormalities . . .).” Professor Burke then goes simpler experience than that with early excision, trans- on to say that this necrotic tissue is the “cause of the life- fusions, and anesthesia and so forth, all at a critical early threatening physiologic, anatomic, immunometabolic and bacterial events known as the ‘complications of burn in- time as the need for surgical procedure is reduced and jury’ . . . ,” which “do not occur if the scar is removed early.” can be scheduled more amiably.This inexpensive topical It would be of great value to your readers to explain treatment will also allow for greater success in third why this is so. In 1963, skin was shown to become per- world countries and in mass casualty management, as in 9 st nicious when subjected to burning.2 This finding was terrorist strikes, for example, things that the 21 cen- supported by US Public Health Service research grant tury will be more concerned about. GM 08678-02. Consequently, a pernicious agent was extracted from burned skin, extensively analyzed, and REFERENCES identified as an aggregation of lipid proteins that can kill 1. Burke JF. Burn treatment’s evolution in the 20th century (cen- 3 even isolated germ-free mice, indicating that infection tennial article). J Am Coll Surg 2005;200:152Ϫ153. is not necessary for death. The American Burn Associa- 2. Allgöwer M, Burri C, Gruber UF, Nagel G. Toxicity of burned mouse tion in 1971 awarded the EI Evans Memorial prize for skin in relation to burn temperature. Surg Forum 1963;14:37–39. 3. Allgöwer M, Cueni LB, Stadtler K, Schoenenberger GA. Burn research demonstrating this pernicious effector. Knowl- toxin in mouse skin. J Trauma 1973;13:95–111. edge of this detail begged for excision as early as possible 4. Burke JF, Quinby WC, Bondoc CC. Primary excision and to prevent the pernicious effector from migrating from prompt grafting as routine therapy for the treatment of thermal the necrotic scar into the circulation, and it was Profes- burns in children. Surg Clin N Am 1976;56:477–494. 5. Tompkins RG, Burke JF, Schoenfeld DA, et al. Prompt eschar sor Burke’s team that demonstrated the clinical success excision: a treatment system contributing to reduced burn mor- of this procedure.4,5 tality. Ann Surg 1986;204:272–281. Vol. 200, No. 5, May 2005 Letters 815

6. Allgöwer M, Schoenenberger GA, Sparkes BG. Burning the larg- I want to thank Dr Ross for his in-depth historical per- est immune organ. Burns 1995;21[Suppl 1]:S7–S47. spective on the discovery of the sympathetic paralysis we 7. Bardeen CR. A review of the pathology of superficial burns. John 1 Hopkins Hospital Reports 1898;7:137–145. know as Bernard-Horner syndrome. But in his first sen- 8. Scheidegger D, Sparkes BG, Luscher N, et al. Survival in major tence he perpetuates the misconception that enophthal- burn injuries treated by one bathing in cerium nitrate. Burns mos is part of the syndrome. What is actually seen is a 1992;18:296–300. 9. Sparkes B. Treating mass burns in warfare, disaster or terrorist perception that the globe is retracted because the sym- strikes. Burns 1997;23:238–247. pathetic retractors of the lower lid are also relaxed, narrowing the palpebral fissure and giving the observer the illusion of enophthalmos. This weakness of the lower lid Reply retractors is now called upside-down ptosis and is considered the third sign of the syndrome along with ptosis and John F Burke, MD, FACS miosis. Boston, MA Although it is true that the earliest descriptions included enophthalmos, testing with an exophthalmom- Professor Allgo¨wer makes an important statement in eter was not available as it is today. More recent au- identifying the generation of a highly toxic material after thors2,3 have addressed this issue and do not want the the burning of skin capable of producing extensive im- myth of enophthalmos to continue. munosuppression. Absorption of this from the burn eschar in a burned patient undoubtedly produces immu- nosuppression, making it important to remove the REFERENCES eschar or neutralize the toxic material as soon as possible 1. Ross IB.The role of Claude Bernard and others in the discovery of after injury. But, optimal treatment of burn injury not Horner’s Syndrome. J Am Coll Surg 2004;199:976–980. only removes the dead tissue but also replaces the lost 2. Lepore FE. Enophthalmos and Horner’s syndrome. Arch Neurol skin in all its physiologic, functional and cosmetic prop- 1983;40:460. 3. van der Wiel HL, van Gijn J. No enophthalmos in Horner’s erties in the shortest possible time after injury. So, de- syndrome. J Neurol Neurosurg Psychiatry 1987;50:498–499. stroyed skin must be removed and immediately followed by physiologic, primarily wound closure using skin autograft or a functional bioengineered skin replacement. If clinical trials of cerium nitrate prove successful in eliminating the immunosuppressive product produced Reply by burn eschar, it will not eliminate all of the pathologic events produced by burn eschar. The complications of Ian Ross, MD, FACS burn injury may be delayed in onset, as they have several Jackson, MS times in the past century by improvement in care, but optimal burn treatment will continue to require removal Although the globe is not truly sunken in cases of of burn eschar and immediate physiologic wound clo- Horner’s syndrome, and so the term enophthalmos is sure as soon as possible after injury. not really appropriate, it has been my experience that the traditional clinical triad, “ptosis, miosis, and enophthalmos,” is still widely used. Perhaps this is be- The Role of Claude Bernard and cause it so vividly conveys the initial clinical impression. I thank Dr Reader for his explanation of upside- Others in the Discovery of down ptosis and hope that my article is not going to Horner’s Syndrome perpetuate the misconception that enophthalmos is part of the condition. Most physicians, I believe, do August L Reader III, MD, FACS understand that the globe is not retracted, but I San Francisco, CA should have been more precise. EVIDENCE-BASED SURGERY

One of the several problems facing surgeons trying to Desai and colleagues report their findings of compar- enroll patients in randomized clinical trials, is the per- ative patency of saphenous veins and radial artery grafts ception that patients in one arm of the trial or the other, for coronary artery bypass, and arrive at a conclusion will be at risk for an adverse outcome. In this issue of the and recommendation for patients with severe native ar- Journal, Vist and colleagues, in a review from the Co- tery stenosis. chrane Collaboration, review the outcomes of more than Squadrone and colleagues report their findings of 30,000 patients enrolled in clinical trials and an addi- the efficacy of continuous positive airway pressure tional 20,000 with similar diseases who refused the op- (CPAP) on the incidence of endotracheal intubation portunity to enroll in a trial. All surgeons who are seri- and other variables, in patients who have hypoxemia ously interested in practice of evidence-based surgery after major elective abdominal operations. Their re- will be interested in their findings. sults were so conclusive that the trial was stopped In a Cochrane review of percutaneous coronary artery early; one arm of the trial was much better than the interventions compared with surgical coronary artery other. Their conclusions have important implications bypass, Bakhai and colleagues emphasize the need for for surgeons managing their patients in intensive care additional studies to identify the population that stands settings. to benefit from one or the other procedure. Education Editor

ABSTRACT ence lists of relevant articles and wrote to over 250 in- Outcomes of patients who participate in vestigators to try to obtain further information. randomised controlled trials compared to Selection Criteria: Randomised studies and cohort similar patients receiving similar interventions studies with data on clinical outcomes of RCT partici- who do not participate (Cochrane Review) pants and similar patients who received similar treatment outside of RCTs. Vist GE, Hagen KB, Devereaux P, Bryant D, Kristoffersen DT, Oxman AD Data Collection and Analysis: At least two reviewers independently assessed studies for inclusion, assessed A substantive amendment to this systematic review was study quality and extracted data. Study authors were last made on 24 August 2004. Cochrane reviews are contacted for additional information. regularly checked and updated if necessary. Main Results: We included five randomised studies Background: Some people believe that patients who (yielding 6 comparisons) and 50 non-randomised co- take part in randomised controlled trials (RCTs) face hort studies (85 comparisons), with 31,140 patients risks that they would not face if they opted for non-trial treated in RCTs and 20,380 patients treated outside treatment. Others think that trial participation is bene- RCTs. In the randomised studies, patients were invited ficial and the best way to ensure access to the most up to to participate in an RCT or not; these comparisons pro- date physicians and treatments. vided limited information because of small sample sizes Objectives: To assess the effects of patient participa- (a total of 412 patients) and the nature of the questions tion in RCTs (‘trial effects’) independent both of the they addressed. There was statistically significant heter- Ͻ ϭ effects of the clinical treatments being compared (‘treat- ogeneity (P 0.00001, I2 89.0%) among the 73 ment effects’) and any differences between patients who dichotomous outcome comparisons; none of the poten- participated in RCTs and those who did not. tial explanatory factors we investigated helped to explain this heterogeneity. No statistically significant differences Search Strategy: In May 2001, we searched The were found for 59 of the 73 comparisons. Ten compar- Cochrane Central Register of Controlled Trials isons reported statistically significant better outcomes (CENTRAL), MEDLINE, EMBASE, The Cochrane for patients treated within RCTs, and four comparisons Methodology Register, SciSearch and PsycINFO for reported statistically significant worse outcomes for pa- potentially relevant studies. Our search yielded over tients treated within RCTs. There were no statistically 10,000 references. In addition, we reviewed the refer- significant differences in heterogeneity (P ϭ 0.53, I2 ϭ

0%) or in outcomes (SMD 0.01, 95% CI -0.10 to 0.12) Selection Criteria: Only RCTs comparing stents used of patients treated within and outside RCTs in the 18 with PTCA with CABG were included. Participants comparisons which had used continuous outcomes. were adults with stable angina or ACS and unstable angina and had either single or multiple vessel disease. Reviewers’ Conclusions: This review indicates that Published and unpublished sources were considered. participation in RCTs is not associated with greater risks than receiving the same treatment outside RCTs. These Data Collection and Analysis: Outcomes included results challenge the assertion that the results of RCTs composite event rate (major adverse cardiac event, are not applicable to usual practice. event free survival), death, acute myocardial infarction (AMI), repeat revascularisation and binary reste- Citation: Vist GE, Hagen KB, Devereaux P,Bryant D, nosis as well as information on design and baseline Kristoffersen DT, Oxman AD. Outcomes of patients characteristics. Quality assessment was completed inde- who participate in randomised controlled trials com- pendently. Meta-analyses are presented as odds ratios, pared to similar patients receiving similar interventions 95% confidence intervals (CI) using a fixed-effect who do not participate. The Cochrane Database of Method- model. Heterogeneity between trials was assessed. ology Reviews 2004, Issue 4. Art. No.: MR000009.pub2. DOI: 10.1002/14651858.MR000009.pub2. Main Results: Nine studies (3519 patients) were included. Four RCTs included patients with multiple vessel disease, five focused on single vessel disease. Four ABSTRACT studies reported beyond 1 year. No statistical differences Percutaneous transluminal coronary were observed between CABG and stenting for meta- angioplasty with stents versus coronary artery analysis of mortality or AMI, but there was heterogene- bypass grafting for people with stable angina ity. Composite cardiac event and revascularisation rates or acute coronary syndromes (Cochrane were lower for CABG than for stents. Odds ratios result- Review) ing from meta-analysis of event rate data at 1 year were, odds ratio 0.43 (95% CI 0.35 to 0.54) and at 3 years, Bakhai A, Hill RA, Dundar Y, Dickson R, Walley T odds ratio 0.37 (95% CI 0.29 to 0.48). Odds ratios for A substantive amendment to this systematic review was revascularisation at 1 year were, odds ratio 0.18 (95% CI last made on 01 July 2004. Cochrane reviews are regu- 0.13 to 0.25) and at 3 years, odds ratio 0.09 (95% CI larly checked and updated if necessary. 0.02 to 0.34). Binary restenosis at 6 months (single vessel trials) favoured CABG, odds ratio 0.29 (95% CI 0.17 Background: Coronary artery bypass graft surgery to 0.51). (CABG) replaces obstructed vessels with ones from other parts of the body. Alternatively, obstructions are Reviewers’ Conclusions: CABG is associated with remodelled using catheter-based techniques such as per- reduced rates of major adverse cardiac events, mostly cutaneous coronary angioplasty with the use of stents. driven by reduced repeat revascularisation. However, the Though less invasive, stenting techniques are limited by RCT data are limited by follow-up, unrepresentative the re-narrowing of treated vessels (restenosis). We ex- samples and rapid development of both surgical tech- amined evidence on cardiac-related outcomes occurring niques and stenting. Research on real-world patient after CABG or stenting, with implications for resource population or patient level data meta-analyses may identify risk factors and groupings who may benefit most use, resource allocation and informing patient choice. from one strategy over the other. Objectives: To examine evidence from randomised Citation: Bakhai A, Hill RA, Dundar Y, Dickson R, controlled trials (RCTs) on benefit of stents or CABG in Walley T. Percutaneous transluminal coronary angio- reducing cardiac events in people with stable angina or plasty with stents versus coronary artery bypass grafting acute coronary syndrome (ACS). for people with stable angina or acute coronary syn- Search Strategy: CENTRAL (Issue 2 2004), EM- dromes. The Cochrane Database of Systematic Reviews BASE (1990 to 2004), MEDLINE (1990 to 2004) and 2004, Issue 4. Art. No.: CD004588.pub3. DOI: handsearching to July 2004. 10.1002/14651858.CD004588.pub3. Vol. 200, No. 5, May 2005 Evidence-Based Surgery A43

ABSTRACT ABSTRACT N Engl J Med 2004 Nov 25;351(22):2302–2309 JAMA 2005 Feb 2;293(5):589–595

Continuous positive airway pressure for A randomized comparison of radial-artery and treatment of postoperative hypoxemia: a saphenous-vein coronary bypass grafts randomized controlled trial Desai ND, Cohen EA, Naylor CD, Fremes SE; Radial Squadrone V, Coha M, Cerutti E, Schellino MM, Biolino Artery Patency Study Investigators P, Occella P, Belloni G, Vilianis G, Fiore G, Cavallo F, Division of Cardiac Surgery, Sunnybrook and Women’s Ranieri VM; Piedmont Intensive Care Units Network College Health Sciences Centre, University of Toronto, (PICUN) Toronto, Canada Dipartimento di Anestesia, Azienda Ospedaliera S. Gio- Background: In the past decade, the radial artery has vanni Battista-Molinette, Universita di Torino, Italy frequently been used for coronary bypass surgery despite Background: Hypoxemia complicates the recovery of concern regarding the possibility of graft spasm. Graft 30% to 50% of patients after abdominal surgery; endo- patency is a key predictor of long-term survival. We tracheal intubation and mechanical ventilation may be therefore sought to determine the relative patency rate of required in 8% to 10% of cases, increasing morbidity radial-artery and saphenous-vein grafts in a randomized and mortality and prolonging intensive care unit and trial in which we controlled for bias in the selection of hospital stay. The objective was to determine the effec- patients and vessels. tiveness of continuous positive airway pressure com- Methods: We enrolled 561 patients at 13 centers. The pared with standard treatment in preventing the need left internal thoracic artery was used to bypass the ante- for intubation and mechanical ventilation in patients rior circulation. The radial-artery graft was randomly who develop acute hypoxemia after elective major abdominal surgery. assigned to bypass the major vessel in either the inferior (right coronary) territory or the lateral (circumflex) ter- Methods: Randomized, controlled, unblinded study ritory, with the saphenous-vein graft used for the oppos- with concealed allocation conducted between June 2002 ing territory (control). The primary end point was graft and November 2003 at 15 intensive care units of the occlusion, determined by angiography 8 to 12 months Piedmont Intensive Care Units Network in Italy. Con- postoperatively. secutive patients who developed severe hypoxemia after major elective abdominal surgery. The trial was stopped Results: Angiography was performed at one year in for efficacy after 209 patients had been enrolled. Patients 440 patients: 8.2 percent of radial-artery grafts and 13.6 were randomly assigned to receive oxygen (n ϭ 104) or percent of saphenous-vein grafts were completely oc- oxygen plus continuous positive airway pressure (n ϭ ϭ cluded (P 0.009). Diffuse narrowing of the graft (the 105). The primary end point was incidence of endotra- angiographic “string sign”) was present in 7.0 percent of cheal intubation; secondary end points were intensive radial-artery grafts and only 0.9 percent of saphenous- care unit and hospital lengths of stay, incidence of pneu- ϭ vein grafts (P 0.001). The absence of severe native- monia, infection and sepsis, and hospital mortality. vessel stenosis was associated with an increased risk of Results: occlusion of the radial-artery graft and diffuse narrowing Patients who received oxygen plus continuous positive airway pressure had a lower intubation rate (1% of the graft. Harvesting of the radial artery was well ϭ tolerated. vs 10%; P .005; relative risk [RR], 0.099; 95% confidence interval [CI], 0.01-0.76) and had a lower occur- Conclusions: Radial-artery grafts are associated with a rence rate of pneumonia (2% vs 10%, RR, 0.19; 95% lower rate of graft occlusion at one year than are CI, 0.04-0.88; P ϭ .02), infection (3% vs 10%, RR, saphenous-vein grafts. Because the patency of radial- 0.27; 95% CI, 0.07-0.94; P ϭ .03), and sepsis (2% vs artery grafts depends on the severity of native-vessel ste- 9%; RR, 0.22; 95% CI, 0.04-0.99; P ϭ .03) than did nosis, such grafts should preferentially be used for target patients treated with oxygen alone. Patients who re- vessels with high-grade lesions. ceived oxygen plus continuous positive airway pressure A44 Evidence-Based Surgery J Am Coll Surg also spent fewer mean (SD) days in the intensive care pressure died in the hospital while 3 deaths occurred unit (1.4 [1.6] vs 2.6 [4.2], P ϭ .09) than patients among those treated with oxygen alone (P ϭ .12). treated with oxygen alone. The treatments did not affect Conclusions: Continuous positive airway pressure the mean (SD) days that patients spent in the hospital may decrease the incidence of endotracheal intubation (15 [13] vs 17 [15], respectively; P ϭ .10). None of those and other severe complications in patients who develop treated with oxygen plus continuous positive airway hypoxemia after elective major abdominal surgery. CONTINUING MEDICAL EDUCATION PROGRAM JACS CME-1 FEATURED ARTICLES, VOLUME 200, MAY 2005 Postoperative delirium in the older patient Amador LF, Goodwin JS J Am Coll Surg 2005;200:767–773 Safety of carotid endarterectomy in 2,443 elderly patients: Lessons from nonagenarians—are we pushing the limit? Teso D, Edwards RE, Frattini JC, et al J Am Coll Surg 2005;200:734–741

Objectives: After reading the featured articles pub- Online, at http://jacscme.facs.org, or you can earn one lished in this issue of the Journal of the American College CME credit if you submit this page by fax (see instruc- of Surgeons (JACS) participants in the JACS CME pro- tions in box below). gram should be able to demonstrate increased under- JACS CME Online provides four articles from each standing of the material specific to the article featured issue for two credits per month. The articles this and be able to apply relevant information to clinical month on JACS CME Online are: practice. Objectives are stated at the beginning of each Postoperative delirium in the older patient. Amador featured article; the questions follow with five response LF, Goodwin JS. choices, and a critique discussing the objective. Safety of carotid endarterectomy in 2,443 elderly pa- The American College of Surgeons is accredited by tients: Lessons from nonagenarians—are we pushing the the Accreditation Council for Continuing Medical Ed- limit? Teso D, Edwards RE, Frattini JC, et al. ucation (ACCME) to sponsor continuing medical edu- Does bone marrow metastasis correlate with sentinel cation for physicians. The JACS CME program fulfills lymph node metastasis in breast cancer patients? Troc- the ACCME essentials. ciola SM, Hoda S, Osborne MP, et al. The American College of Surgeons designates this GATA-3 expression as a predictor of hormone re- educational activity for a maximum of 1 Category 1 credits toward the AMA Physician’s Recognition Award. sponse in breast cancer. Parikh P, Palazzo JP, Rose LJ, Each physician should claim only those credits that he/ et al. she actually spent in the educational activity. You can earn two CME credits using JACS CME

Questions: Wendy Cowles Husser, MA, MPA Executive Editor, JACS 633 N Saint Clair Street, Chicago, IL 60611 312-202-5306 (ph) 312-202-5027 (fax) [email protected]

Postoperative delirium in the older patient management of symptoms should be reserved for agi- Amador LF, Goodwin JS tated or disruptive individuals to ensure safety and facil- J Am Coll Surg 2005;200:767–773 itate diagnosis and treatment. Symptom control should never replace or delay treatment of the precipitating Learning Objectives: After study of this article, the cause(s) of delirium. Although managing disruptive surgeon should be able to: identify risk factors; describe behavior is the most challenging aspect of delirium ther- the diagnostic workup, including the Confusion Assess- apy, less than one-third of older individuals with delir- ment Method (CAM); describe management options; ium exhibit aggressive or agitated behavior. Antipsy- and describe the outcomes of postoperative delirium in chotics are not indicated to promote a normal sleep the older adult. cycle, stop the progression of the delirium, promote cognitive functions, or to prevent dementia after surgery. Question 1 All of the following are true about postoperative delir- Safety of carotid-endarterectomy in 2,443 elderly ium except: patients Lessons from nonagenarians—are we pushing the a) acute onset limit? b) fluctuating course Teso D, Edwards RE, Frattini JC, et al c) disorganized thought d) unaltered consciousness J Am Coll Surg 2005;200:734–741 e) changes in sleep cycle Learning Objectives: After study of this article, sur- Critique: The presentation of postoperative delirium geons should be able to discuss the importance of age includes changes in alertness, difficulty staying focused, and symptoms in determining outcomes in patients disorientation, disorganized thinking, and changes in with carotid stenosis treated with carotid endarterec- sleep cycle. The patient with postoperative delirium has tomy. The surgeon will be able to advise patients regard- disturbances of consciousness and cognition (or percep- ing the safety of carotid endarterectomy for elderly pation) that develop over a short period of time, caused by tients and be familiar with current reports suggesting a general medical or surgical condition, substance intox- increased risk associated with carotid angioplasty and ication, substance withdrawal, or multiple concurrent stenting in geriatric patients. etiologies. This disturbance of consciousness causes a reduced ability to focus, sustain, or shift attention. This Question 1 disturbance in consciousness and the acute onset differ- A primary care physician has referred a 92-year-old male entiate delirium from dementia. Patients generally de- patient with an asymptomatic focal 99% right internal velop dementia over months or years and consciousness carotid artery stenosis. The patient has treated hyperten- is mostly intact in the beginning of dementia. sion and a remote history of asymptomatic coronary artery disease, but walks 1 to 2 miles a day without Question 2 becoming short of breath. The referring physician has Antipsychotic medications for postoperative delirium suggested the possibility of carotid angioplasty and are indicated in order to: stenting. Which of the following is TRUE: a) stop the progression of the delirium a) Nonagenarians with asymptomatic carotid stenosis do not b) control symptoms of agitation require treatment because their short life expectancy does c) regulate the sleep cycle not justify the risk of carotid endarterectomy. d) promote cognitive functions b) All geriatric patients with high-grade carotid stenosis e) prevent the development of dementia after surgery should be treated with carotid angioplasty and stenting. c) Nonagenarians treated with carotid endarterectomy, as a Critique: The management of postoperative delir- group, have much higher rates of postoperative myocardial ium is multidimensional and spans three areas simul- infarction compared with younger patients. taneously: treating the contributing illness(es), provid- d) If this patient is treated with carotid endarterectomy, the ing supportive measures; and, if needed, introducing same postoperative hospital length of stay as an average symptom control. Antipsychotic medication for the younger patient can be expected. Vol. 200, No. 5, May 2005 Continuing Medical Education Program A47 e) If this patient develops amaurosis fugax, he should have an quires careful discussion of the risks and benefits of urgent carotid endarterectomy performed. CEA, because symptomatic presentation and age over 90 years are strong predictors of mortality after CEA in Critique: With increasing life expectancy in Western multivariable analysis (Table 4 in article). society, the extreme elderly are presenting for medical care with increasing frequency. Landmark studies in the United States that provide evidence for the benefit of References carotid endarterectomy (CEA) for treatment of carotid 1. Alamowitch S, Eliasziw M, Algra A, et al. North American Symp- stenosis, compared to treatment with medical therapy, tomatic Carotid Endarterectomy Trial (NASCET) Group. Risk, causes, and prevention of ischaemic stroke in elderly patients with such as the North American Symptomatic Carotid End- symptomatic internal-carotid-artery stenosis. Lancet 2001;357: arterectomy (NASCET) trial of symptomatic patients 1154–1160. and the ACAS trial of asymptomatic patients, excluded 2. Roques XF, Baudet EM, Clerc F. Results of carotid endarterec- patients older than 80 years. Several studies, however, tomy in patients 75 years of age and older. J Cardiovasc Surg [Torino] 1991;32:726–731. provide evidence that CEA provides rates of stroke-free survival for elderly patients similar to those for younger Question 2 patients; these include two randomized controlled studies that provide evidence of the benefit of CEA in elderly Which of the following risk factors does NOT predict 1,2 adverse outcomes after carotid endarterectomy: patients. Carotid ar ter y angioplasty and stenting is a newer option for the treatment of carotid artery stenosis. a) history of preoperative stroke Application to elderly patients, however, should be un- b) age over 65 years dertaken with caution because early results of such trials c) hypertension have demonstrated significantly inferior results to those d) renal disease obtained with stenting and angioplasty in younger e) diabetes patients. Critique: A history of symptomatic presentation such Nonagenarians with asymptomatic carotid stenosis as a stroke, is a strong predictor of adverse outcomes after treated with CEA can expect excellent perioperative out- CEA, including an increased perioperative risk of mor- comes similar to those in younger patients, when treated tality or stroke. A history of transient ischemic attack by high-volume surgeons in high-volume hospitals (Ta- (TIA) is also associated with an increased risk, although ble 6 in article). Nonagenarians in the United States who to a lesser degree. A history of amaurosis fugax, however, are in excellent health have a life expectancy of several confers less risk. Many studies have demonstrated the years; surgeons with excellent outcomes when perform- safety of carotid endarterectomy in octogenarians; nona- ing CEA may reasonably offer the procedure to patients genarians are at increased risk for mortality after CEA who are otherwise good risk. The risk of perioperative compared with younger patients. It is likely, however, stroke after CEA may be lower than that after carotid that only symptomatic nonagenarians have an increased angioplasty and stenting in geriatric patients and CEA risk; asymptomatic nonagenarians may have as low a risk should be considered the current therapy of choice in as younger patients. Renal disease is a strong predictor of patients over 80 years old. The risk of perioperative car- mortality after CEA. Diabetes confers a small increased diac complications, such as myocardial infarction, is not risk of both perioperative mortality and stroke. Treated higher in elderly patients (Table 3 in article). Treatment hypertension is associated with a reduced risk of mortal- of symptomatic carotid stenosis in a nonagenarian re- ity with CEA (Tables 4 and 5 in article).