Bone Marrow Transplantation (2001) 28, 289–294  2001 Nature Publishing Group All rights reserved 0268–3369/01 $15.00 www.nature.com/bmt Post-transplant complications Does consultation change management of patients receiving hematopoietic stem cell transplantation?

G Fallows1, M Rubinger1,2 and CN Bernstein1,3

1Department of Internal Medicine, University of Manitoba, 2Cancer Care Manitoba, and 3University of Manitoba Inflammatory Bowel Disease Clinical and Research Centre, Winnipeg, Manitoba, Canada

Summary: Gastrointestinal toxicity is common in the course of hema- topoietic stem cell transplantation (HSCT) following high- Gastrointestinal complications following hematopoietic dose chemotherapy alone or combined with radiotherapy. stem cell transplantations (HSCTs) are common, but it Pre-transplantation radiation and chemotherapy with cyto- is unknown how often gastroenterology consultation toxic drugs commonly results in intestinal mucosal damage. (GEC) early post BMT leads to specific changes in Bone marrow ablation leaves patients with lowered patient management. We aimed to determine the immunologic defences and susceptible to bacterial, viral reason(s) for GEC, the diagnoses found through GEC, and fungal infections. Allogeneic transplantation, which is whether the advice or intervention led to change(s) in associated with a significant risk of acute graft-versus-host management and if intervention led to any adverse out- disease (aGVHD), increases the possibility of gastrointes- come within the first 100 days post HSCT. We under- tinal complications. took a retrospective review of all patients at least 18 The impact of the gastroenterologist in the evaluation who underwent HSCT between Nov- and management of these patients is poorly defined. We (197 ؍ years old (n ember 1990 and April 1998. Of these, 79 patients had aimed to determine the nature of requests for consultation 92 consultations for a total of 163 separate GE problems with a gastroenterologist (GEC), what diagnoses were within the first 100 days post HSCT. Data were obtained found through these consultations, and whether the advice through chart review. It was determined whether the or intervention undertaken by the gastroenterologist led to intervention or advice given by the consultant led to any changes in patient management or adverse outcomes. actual changes in patient management or outcome. We found that the characteristics more likely to be associa- Methods ؍ ted with GEC included female patient vs male (P 0.03), allogeneic vs autologous transplants (P Ͻ 0.001), and leu- The charts of 197 consecutive patients over 18 years old ,(0.006 ؍ hematologic vs solid malignancies (P Overall, a definitive undergoing HSCT at the University of Manitoba between .(0.013 ؍ kemias vs lymphomas (P diagnosis for an identified complaint was made in 71% November 1990 and April 1998 were retrospectively (range 25–87%). A change in management was effected reviewed. Data collected included the type of malignancy, in 54% of cases (range 0–59%). Endoscopy led to per- preparatory chemotherapy regimens, radiation therapy foration and subsequent death in two patients (1.8%). fields and dosages, medications, type of hematopoietic stem Gastrointestinal disease was a direct cause of death in cell graft (autologous bone marrow, autologous peripheral 2.5% of all patients. In conclusion, a definite diagnosis blood progenitor cells, and allogeneic bone marrow), was reached in 71% of gastrointestinal problems and prophylactic or active anti-GVHD medications, laboratory management was effected in 54% of cases. Since endos- data, reason for consultation, timing of complication in copy was associated with a mortality of 1.8%, minimiz- relation to HSCT, endoscopic and histologic findings, ing its use for the cases in which no impact is made, patient outcomes and adverse events. In those patients should be considered. Bone Marrow Transplantation requiring GEC, the reason(s) for GEC was recorded, as was (2001) 28, 289–294. the final diagnosis determined for each problem. After Keywords: acute graft-versus-host disease; ; dys- review of each case, it was determined whether the inter- phagia; gastroenterology consultation; gastrointestinal vention or advice given by the consultant led to actual endoscopy; hematopoietic stem cell transplantation changes in patient outcome or management. A change in management was defined as initiation of new therapy, dis- continuation or dose adjustment of medications, or thera- peutic endoscopic intervention. Upper gastrointestinal Correspondence: Dr CN Bernstein, University of Manitoba, Section of Gastroenterology, 804F-715 McDermot Avenue, John Buhler Research bleeding (UGIB) refers to of old or fresh blood. Centre, Winnipeg, Manitoba, Canada, R3E-3P4 Lower gastrointestinal bleeding (LGIB) refers to passage Received 2 November 2000; accepted 5 April 2001 of melena or fresh blood per rectum. Gastroenterology in patients receiving HSCT G Fallows et al 290 Table 1 GI problems addressed in GEC Data were entered into an Excel spreadsheet (Microsoft, Excel 97) and statistical significance was determined by a Reason for No. of No. scoped No. biopsied chi-squared analysis or Student’s t-test where applicable. consultation consultations (%) (%) (% of those scoped)

? aGVHD 42 (46) 41 (98) 40 (98) Results Diarrhea 33 (36) 31 (94) 29 (94) Vomiting 27 (29) 24 (89) 24 (100) UGI bleeding 23 (25) 21 (91) 12 (57) In total, 197 HSCTs were performed in the study period LGI bleeding 4 (4) 3 (75) 2 (67) on patients over 18 years old. A total of 92 GEC were 18 (20) 18 (100) 17 (94) requested on 79 of the 197 patients (40%). In total, 163 12 (13) 9 (75) 8 (89) Anorexia 4 (4) 4 (100) 4 (100) gastrointestinal problems were specified as needing to be addressed in consultation (see Table 1). The characteristics of these patients are shown in Tables 2 and 3. There were no diagnoses of CMV enterocolitis made. Significant risk factors for undergoing GEC were female gender (51% of females had GEC vs 34% of males, P ϭ Table 2 Demographics of patients undergoing HSCT 0.03) and allogeneic HSCT (52% of allogeneic transplants had GEC vs 25% of autologous transplants, P Ͻ 0.001). All HSCTs No GEC GEC The disease for which the transplant was being performed Ͻ Total patients 197 118 79 also influenced the likelihood of GEC (P 0.0001). Hema- Total consults 92 tologic malignancies were at greater risk than solid malig- Males/Females 124/73 66%/49% 34%/51% nancies (P ϭ 0.006), and patients with leukemia needed Average age 40.1 40.3 39.9 more GEC than did lymphoma patients (P ϭ 0.013). Type of HSCT Allogeneic bone marrow 110 48% 52% Autologous bone marrow and/or 87 75% 25% Acute GVHD consultations stem cells Forty-two consultations (46%) were sought on 39 patients specifically to rule out suspected aGVHD due to specific gastrointestinal symptoms (ie vomiting, diarrhea; these Table 3 Diseases for which HSCT was performed symptoms if specified in the consultation request are also included in separate analysis). These consultations occurred All HSCTs No GI consult GI consult within a mean (Ϯ standard deviation) of 31 Ϯ 12 days post (n) (%) (%) HSCT. Of these consultations, 16/42 (38%) were on patients who had been diagnosed with aGVHD previously Hematopoietic in their clinical course, 11 (26%) of whom were already Non-Hodgkin’s lymphoma 41 73 27 receiving active therapy for either presumed or documented Acute myeloid leukemia 39 38 62 Chronic myeloid leukemia 30 53 47 aGVHD. A definitive diagnosis was made in 23 (55%) of Hodgkin’s lymphoma 25 71 29 the 42 GEC, 18 of whom had aGVHD (43% of Acute lymphoblastic leukemia 17 71 29 consultations) (Table 5). Of the 18 GEC in whom aGVHD Myelodysplasia 12 58 42 was the diagnosis, six (33%) had previously been diagnosed Multiple myeloma 9 44 56 Aplastic anemia 9 33 67 with aGVHD and five (27.8%) were already being treated Myelofibrosis 2 50 50 for GVHD. When a gastroenterologist was asked to rule Solid tumors out aGVHD, management was changed in 19 of 42 patients Embryonal or germ cell cancer 8 100 0 (45%) – see Table 6. Breast cancer 5 100 0

Gastrointestinal bleeding (GIB) Upper: Twenty-three GEC (25%) were sought on 23 patients for a complaint of upper GIB. Twenty-nine upper Table 4 endoscopies were performed in this patient group in total. Bleeding occurred within a mean of 30 Ϯ 19 days of Pathologic diagnosis n HSCT. At the time of endoscopy, mean platelet counts were aGVHD 18 ϫ 9 Ϯ ϫ 9 aGVHDϩ Clostridium difficile colitis 1 25.0 10 /l 22.7 10 /l. Candida 12 A cause of the GIB was determined in 20 of 23 patients Ischemic colitis 1 (87%), six of 23 (26%) were secondary to diffuse gastric Radiation esophagitis 1 bleeding. Peptic ulcers or reflux esophagitis each accounted Reflux esophagitis 3 for five (22%) of the upper GIB (Table 5). Therapeutic ‘Non-specific’ 3 Normala 31 hemostasis was performed nine times on seven patients (30%). Two patients required a second therapeutic inter- aThe diagnosis of virus ileitis was made after normal vention and therapeutic hemostasis was successful in three biopsies were taken, and the patient went to surgery. of seven patients (43%). Two patients required surgery, one

Bone Marrow Transplantation Gastroenterology in patients receiving HSCT G Fallows et al 291 Table 5 Diagnoses found through consultation based on GI problem identified

Reason for consultation Diagnosis No. %

GVHD (n ϭ 42) CMV enteritis 1 2.4 aGVHD 18 42.9 Infective esophagitis 2 4.8 Drug-induced enteritis 1 2.4 Pseudomembranous colitis 2 4.8 Ulcer 2 4.8 No diagnosis 16 38.1 Upper GIB (n ϭ 23) Reflux esophagitis 5 21.7 Infective esophagitis 2 8.7 Chemo-esophagitis 1 4.3 Diffuse bleeding 6 26.1 Ulcer 5 21.7 Epistaxis 1 4.3 No diagnosis 3 13.0 Dysphagia/Odynophagia No endoscopy 1 5.6 (n ϭ 18) Chemo-esophagitis 1 5.6 Candida esophagitis 9 50.0 Radiation esophagitis 2 11.1 Reflux esophagitis 2 11.1 No diagnosis 3 16.7 Vomiting (n ϭ 27) aGVHD 10 37.0 Chemo-esophagitis 1 3.7 Infective esophagitis 3 11.1 Reflux esophagitis 1 3.7 Ulcer 3 11.1 Drug induced 1 3.7 No diagnosis 8 29.6 Diarrhea (n ϭ 33) Adenovirus enteritis 1 3.0 aGVHD 14 42.4 Antibiotic-associated diarrhea 1 3.0 Drug-induced diarrhea 1 3.0 Pseudomembranous colitis 2 6.1 No diagnosis 14 42.4 Abdominal pain (n ϭ 12) Adenovirus colitis 1 8.3 aGVHD 1 8.3 1 8.3 Ileus 1 8.3 Infective esophagitis 2 16.7 Candida gastritis 1 8.3 Ulcer 1 8.3 No diagnosis 4 33.3 Anorexia (n ϭ 4) aGVHD 2 50.0 Reflux esophagitis 1 25.0 No diagnosis 1 25.0 Lower GIB (n ϭ 4) Hemorrhoids 1 25.0 HSV ileitis 1 25.0 No diagnosis 2 50.0

of whom subsequently died. GEC led to a change in man- Dysphagia/Odynophagia agement, either through therapeutic intervention or change in drug therapy in 13 (57%) of consults. Upper GIB was Eighteen (20%) GEC were for a complaint of dysphagia or identified as the cause of death in three of the 79 patients odynophagia. Dysphagia or odynophagia occurred at a Ϯ consulted (3.8%). mean of 16 13 days after HSCT when the mean WBC count was 1.2 ϫ 109/l Ϯ 2.2 ϫ 109/l. Of the 18 patients consulted for this problem, 11 were on anti-fungal therapy Lower: Lower GIB was an uncommon complication and and 14 were on anti-viral therapy at the time of GEC. A was a reason for GEC in only four (4%) patients. A defini- GEC resulted in a definitive diagnosis in 14 (78%) of these tive diagnosis for the lower GIB was made in only one case. cases. The most common cause of dysphagia and odyno- GEC did not change the management in any of these cases. phagia in this patient population was Candida esophagitis,

Bone Marrow Transplantation Gastroenterology in patients receiving HSCT G Fallows et al 292 Table 6 Impact of gastroenterology (GE) consultation in HSCT Abdominal pain

Reason for consultation GE made GE changed Twelve (13%) of the consultations sought on 11 patients diagnosis management included a complaint of abdominal pain. GEC was able to No. (%) (No. (%) make a diagnosis to explain the abdominal pain in eight (67%) cases, although no single diagnosis was identified as ?aGVHD (n ϭ 42) 23 (55) 19 (45) being significantly more frequent. GEC changed manage- Upper GIB (n ϭ 23) 20 (87) 13 (56) ment in 67% of consultations. Lower GIB (n ϭ 4) 1 (25) 0 (0) Dysphagia/Odynophagia (n ϭ 18) 14 (78) 10 (56) Vomiting (n ϭ 27) 19 (70) 16 (59) Anorexia Diarrhea (n ϭ 33) 19 (58) 15 (45) Abdominal pain (n ϭ 12) 8 (67) 8 (67) Four (4%) of the GEC sought on four patients included a Anorexia (n ϭ 4) 3 (75) 3 (75) Overall (n ϭ 163) 107 (66) 84 (52) complaint of anorexia. Gastroenterology was able to make a diagnosis to explain the anorexia in three (75%) of the patients, with aGVHD being responsible for half of these complaints (n ϭ 2). Management was changed in 75% of cases. with nine (50%) patients showing either typical changes of Candida esophagitis on endoscopy, and/or biopsy findings Overall consistent with this diagnosis. Of those diagnosed with Overall, a definite diagnosis was reached in 66% of gastro- Candida esophagitis, 4/9 were on no anti-fungal therapy intestinal problems and management was effected in 52% and were subsequently started on fluconazole. GEC led to of cases. a subsequent change in management in 10 (56%) of these cases. The patients were sub-analyzed according to whether Complications their predominant symptom was odynophagia or dysphagia. ϭ A total of 113 endoscopies were performed – 76 gastro- Of those with dysphagia (n 12), five had Candida eso- scopies, 21 flexible sigmoidoscopies, and 16 colonoscopies. phagitis, two had reflux esophagitis and chemotherapy- Complications from endoscopic interventions were ident- induced and radiation-induced esophagitis each accounted ified in two patients (1.8% of procedures). Duodenal perfor- for one patient. Of interest, only one of these five patients ation complicated therapeutic endoscopy in one patient with Candida esophagitis was on anti-fungal therapy at the after epinepherine was injected into a 2 cm, actively bleed- time of consultation. In contrast, all patients with odyno- ϭ ing duodenal ulcer. The patient died due to intraperitoneal phagia (n 6) were receiving anti-fungal therapy prior to hemorrhage from the necrotic duodenal bulb and aspiration consultation. Four of these patients (67%) had Candida eso- of stomach contents after undergoing duodenectomy and phagitis and one had radiation esophagitis. Although gastrojejunostomy. A second patient died from gram- already on anti-fungal therapy, two patients with odyno- negative sepsis 1 month after a flexible sigmoidoscopy to phagia had their anti-fungal therapy intensified after consul- investigate for a complaint of diarrhea and GVHD. At auto- tation (nystatin and ketoconazole therapy was changed to psy, the cause of death was determined to be an intraabdo- fluconazole). minal abscess secondary to a perforation of the sigmoid colon. Vomiting Deaths Vomiting was a problem in 27 (29%) of the GEC and occurred within 26 Ϯ 15 days of HSCT. A definitive diag- Deaths were attributed to a gastrointestinal complaint in nosis was determined for the vomiting in 19 (70%) of five patients (2.5%, 5/197). In addition to the two adverse patients, with the most common etiology being aGVHD outcomes, two patients died of diffuse GI bleeding, and (nine patients, 37%). Consultation led to a change in patient one from extensive small bowel infarction in a patient with management in 16 (59%) of patients. cardiac tamponade.

Diarrhea Discussion Thirty-three (36%) of the GEC sought on 31 patients Gastroenterology complications early after HSCT are com- included a complaint of diarrhea. Two patients had GEC mon, but the impact of a gastroenterologist in the evalu- for diarrhea twice. Diarrhea consultations occurred 29 Ϯ ation and management of these patients is not well defined. 11 days after HSCT. GEC made a definitive diagnosis to Elsewhere, HSCTs are complicated by GIB in 7–18% of explain the diarrhea in 19 (58%) of cases, with aGVHD HSCT cases.1–4 In studies addressing causes of GIB, the accounting for the majority of these diagnoses (Table 5). most common causes are GVHD (up to 60%),5 diffuse GEC led to a change in management in 15 (45%) consul- mucosal bleeding (65–69%),1,4 and esophagitis (up to tations. 92%).2 Duodenal and gastric ulcers are uncommon early

Bone Marrow Transplantation Gastroenterology in patients receiving HSCT G Fallows et al 293 post HSCT and account for only 6–10% of cases of UGI HSCT patients.10 Fungal esophagitis occurred in only 1% bleeds.4,5 Although endoscopy may identify the cause of of patients undergoing HSCT in this study, despite the fact upper GIB in up to 91%4 (a study looking only at ‘gross that we do not uniformly provide fluconazole prophylaxis. GIB’), the majority of causes of bleeding are not amenable Of interest was the fact that Candida caused the majority to endoscopic treatment. Several studies have therefore of cases of dysphagia, and that in contrast to patients with concluded that endoscopy is not useful as a therapeutic odynophagia, the majority were not receiving anti-fungal endeavour.1,2,4 therapy. The majority of patients with odynophagia and In a study evaluating the risk of generalized bleeding in dysphagia in our study had a fungal cause and it is possible this patient population, Nevo et al6 found moderate to sev- that institution of anti-fungal therapy prior to endoscopy ere GIB complicated 14% of their cases. They found that would be cost effective for management of this compli- bleeding in general was more prevalent in patients with cation by reducing the need for endoscopy. Endoscopy per- allogeneic and unrelated stem cell transplants compared to haps should be reserved for those patients who do not autologous transplants. Bleeding was determined to be an respond to fluconazole therapy. indicator of poor prognosis, but was not a direct cause of The majority of cases of persistent and vomiting death in the majority of cases. GIB is an indicator of poor post HSCT are secondary to aGVHD. Wu et al11 found that prognosis in patients requiring ICU admission after HSCT,7 81% cases of persistent nausea and vomiting were due to but death due to GIB occurs in only about 0.2–3% of aGVHD, although elsewhere GVHD accounted for as few cases.1,5 as 26% of nausea and vomiting.12 In our study, as in pre- In our study, GIB occurred in 12% (23/197) of patients vious studies, this complication occurred at a mean of 26 undergoing HSCT. As in previous studies, the most com- Ϯ 15 days, and aGVHD accounted for 37% of cases. Eso- mon reason for GI bleeding was secondary to diffuse phagitis was responsible for the majority of the remaining mucosal hemorrhage (26%) with reflux esophagitis and cases (19%). peptic ulcer disease each being the cause of upper GI bleed- Acute diarrhea after HSCT is common and is a compli- ing in 22% of cases. Therapeutic intervention was cation in up to 43% of patients post HSCT.13 A study attempted in 30% of cases of upper GI bleeding and was addressing causes of acute diarrhea in this patient popu- successful in controlling bleeding in 13% of these cases. In lation showed that acute GVHD accounted for 48% of addition to being a successful intervention in some patients, cases, while infectious diarrhea (C. difficile, astrovirus and endoscopy resulted in a substantial impact on patient man- adenovirus) accounted for 13%. Despite extensive investi- agement in GI bleeding, with 87% of cases having an ident- gation, no clear etiology could be found for 39% of their ifiable lesion and 56% having their management affected cases of self-limiting diarrhea. Diarrhea has been a reason by endoscopic intervention. We feel this is a sufficient for GEC elsewhere in 9% of cases.4 Previous studies have success rate to warrant pursuing upper endoscopy. shown that the most likely etiology for diarrhea post HSCT The incidence of odynophagia and dysphagia is about is aGVHD (48–52%).4,13 In our study, we found 58% of 5%4 in patients undergoing HSCT. Historically, common diarrhea cases had an identifiable etiology. The majority of causes of these complaints were either infectious esophag- identified cases were acute aGVHD (42%). itis or chemo/radiotherapy-induced esophagitis. In an early Most of the gastroenterology consultations included study, McDonald et al8 found 54% of patients with symp- endoscopy (91%) and most endoscopies were associated toms had esophageal infection, the majority of which were with biopsies (83%). The complication rate after upper virus induced. Wheeler et al9 showed that 47% of patients endoscopy of 1.8% must be viewed in the context of the with dysphagia had fungal infections while 13% had viral nature of the patient population (typically immunosup- infections. A more recent study by Vishny et al2 showed a pressed with severe thrombocytopenia). This rate is much reduced incidence of viral esophagitis, presumably due to higher than the complication rate in immunocompetent prophylactic acyclovir, with all cases of odynophagia being patients (0.01–1%).14,15 One of the complications occurred attributed to at least grade III to IV esophagitis. Seventy- after therapeutic endoscopy for a bleeding lesion and one six percent of those patients had chest radiotherapy pre- was a perforation after a flexible sigmoidoscopy. This com- viously, only 27% had documented fungal infections, and plication rate needs to be assessed in the context of no cases of viral esophagitis were identified. Another recent the patient’s ability to withstand emergency surgical study by Johnson et al4 showed that infective esophagitis intervention. (Candida or viral) accounted for 26% of patients with Further exploration is required in order to determine odynophagia in a study able to determine a cause of whether empiric therapy would be useful in minimizing odynophagia in 79% of cases. endoscopic therapy in these patients. For instance, the Our study was able to determine the cause of dysphagia majority of patients with dysphagia have Candida esophag- and odynophagia in 78% of cases. Fifty percent of the cases itis. Endoscopy could potentially be reduced by empirically of odynophagia and dysphagia in this study were secondary treating odynophagia or dysphagia with fluconazole and to fungal infection and no cases of viral esophagitis were reserving endoscopy for those patients who fail to improve found. Again, this may reflect the fact that these patients after 48 h. In summary, gastrointestinal complications received prophylactic acyclovir therapy (78% were still on requiring GEC occur in approximately 40% of patients with anti-viral therapy at time of consultation), but only 5/9 were HSCT within the first 100 days. Although GEC affects still receiving anti-fungal therapy. management in at least half of the cases, further study is Prophylactic fluconazole has been shown to decrease the required to look at whether endoscopy could be minimized risk of both systemic and superficial fungal infections in in this patient population.

Bone Marrow Transplantation Gastroenterology in patients receiving HSCT G Fallows et al 294 References 8 McDonald GB, Sharma P, Hackman RC. Esophageal infec- tions in immunosuppressed patients after marrow transplan- 1 Kaur S, Cooper G, Fakult S, Lazarus H. Incidence and out- tation. Gastroenterology 1985; 88: 1111–1117. come of overt gastrointestinal bleeding in patients undergoing 9 Wheeler RR, Peacock JEJ, Cruz JM, Richter JE. Esophagitis bone marrow transplantation. Dig Dis Sci 1996; 41: 598–603. in the immunocompromised host: role of endoscopy in diag- 2 Vishny ML, Blades EW, Creger RJ, Lazarus HM. Role of nosis. Rev Infect Dis 1987; 9:88–96. upper endoscopy in evaluation of upper gastrointestinal symp- 10 Goodman JL, Winston DJ, Greenfield RA et al. A controlled toms in patients undergoing bone marrow transplantation. trial of fluconazole to prevent fungal infections in patients Cancer Invest 1994; 12: 384–389. undergoing bone marrow transplantation. New Engl J Med 3 Wolford JL, McDonald GB. A problem-oriented approach to 1992; 326: 845–851. intestinal and disease after marrow transplantation. J Clin 11 Wu D, Hockenberry DM, Brentnall TA et al.Persistent Gastroenterol 1988; 10: 419–33. nausea and anorexia after marrow transplantation: a pro- 4 Johnson MC, Pinkas H, Elfenbein GJ. Endoscopy in the bone spective study of 78 patients. Transplantation 1998; 66: marrow transplant patient presenting with GI symptoms: is it 1319–1324. useful? Gastroint Endosc 1994; 40: P42 (Abstr.) 12 Spencer GD, Hackman RC, McDonald GB et al. A prospec- 5 Wolford JL, McDonald GB. Gastrointestinal bleeding after tive study of unexplained nausea and vomiting after marrow marrow transplant: a prospective study of risk factors, etiology transplantation. Transplantation 1986; 42: 602–607. and outcome. Gastroenterology 1987; 92: 1697 (Abstr.) 13 Cox GJ, Matsui SM, Lo RS et al. Etiology and outcome of 6 Nevo S, Swan V, Enger C et al. Acute bleeding after bone diarrhea after marrow transplantation: a prospective study. marrow transplantation (BMT) – incidence and effect on sur- Gastroenterology 1994; 107: 1398–1407. vival. A quantitative analysis in 1,402 patients. Blood 1998; 14 Chung SC, Leung JW, Sung JY et al. Injection or heat probe 91: 1469–1477. for bleeding ulcer. Gastroenterology 1991; 100:33–37. 7 Price KJ, Thall PF, Kish SK et al. Prognostic indicators for 15 Selby JV, Friedman GD. US Preventive Services Task Force. blood and marrow transplant patients admitted to an intensive Sigmoidoscopy in the periodic health examination of asympto- care unit. Am J Crit Care Med 1998; 158: 876–884. matic adults. JAMA 1989; 261: 594–601.

Bone Marrow Transplantation