Grading Evidence

Analysis of the colonoscopic findings in patients with rectal bleeding according to the pattern of their presenting symptoms Journal Diseases of the Colon & Publisher Springer New York ISSN 0012-3706 (Print) 1530-0358 (Online) Issue Volume 34, Number 5 / May, 1991

Abstract Patients presenting with rectal bleeding were prospectively categorized according to the pattern of their presentation into those with outlet bleeding (n=115), suspicious bleeding (n=59), hemorrhage (n=27), and occult bleeding (n=68). All patients underwent and this was complete in 94 percent. There were 34 patients with carcinoma and 69 with adenomas >1 cm diameter. The percentage of neoplasms proximal to the splenic flexure was 1 percent in outlet bleeding, 24 percent with suspicious bleeding, 75 percent with hemorrhage, and 73 percent with occult bleeding. Barium enema was available in 78 patients and was falsely positive for neoplasms in 21 percent and falsely negative in 45 percent. Colonoscopy is the investigation of choice in patients with suspicious, occult, or severe rectal bleeding. Bleeding of a typical outlet pattern may be investigated by flexible .

J Surg Res. 1993 Feb;54(2):136-9.

Colonoscopy for intermittent rectal bleeding: impact on patient management. Graham DJ, Pritchard TJ, Bloom AD.

Department of Surgery, Case Western Reserve University School of Medicine, Cleveland, Ohio 44106.

Abstract Rectal bleeding is a frequent presenting symptom of a number of benign anorectal disorders. However, it may also be a warning sign of more significant gastrointestinal pathology. For this reason, full colonic evaluation has been recommended in patients with intermittent bright red rectal bleeding. The purpose of this study is to evaluate the utility of colonoscopy in this setting. Data were prospectively collected on 125 performed on the surgical service at the Cleveland Wade Park Veterans Administration Medical Center during a two year period. During this period 33 patients underwent colonoscopy for the evaluation of intermittent bright red rectal bleeding. Fourteen patients had abnormal rectal exams, including in 9, mass lesions in 3, prolapse in 1, and fistula in ano in 1. Colonoscopy was normal in only 7 (21%) of the 33 patients examined. Findings in the remaining 26 included 31 polyps in 14 patients, cancer in 3, AVM in 1, diverticula in 9, hemorrhoids in 4, and other benign lesions in 5. Positive findings on had no relationship to findings at , with abnormal findings in 52% of patients with normal rectal exams and in 27% of patients with abnormal rectal exams (P = 0.187, NS). Findings at colonoscopy resulted in a change in management in 16 (48%) of patients examined. In patients with intermittent rectal bleeding, the entire colon should be evaluated regardless of findings on rectal examination, as a significant number of patients will have concomitant findings. Colonoscopy is an excellent method for colonic evaluation in this setting. (ABSTRACT TRUNCATED AT 250 WORDS)

Education improves colonoscopy appropriateness Gastrointestinal Endoscopy - Volume 67, Issue 1 (January 2008) -

Table 2 -- Nine hundred twenty-one indications in 866 appropriate colonoscopies ASGE/SIED guidelines No % Hematochezia 218 23.7 Occult fecal blood 154 16.7 Surveillance after endoscopic polypectomy (3- to 5-y intervals after adequate clearance 136 14.7 of neoplastic polyps) Persistent change in bowel habits 105 11.4 Surveillance after resection of cancer (colonoscopy to remove synchronous neoplastic 100 10.8 lesion at or around time of curative resection of cancer followed by colonoscopy at 3 y and 3 to 5 y thereafter to detect metachronous cancer) Chronic 57 6.2 Unexplained iron-deficiency anemia 54 5.9 Family history of sporadic before age 60 y: colonoscopy every 5 y 42 4.6 beginning 10 y earlier than affected relative or every 3 y if adenoma is found Abnormality on imaging 28 3.0 Unexplained weight loss 18 1.9 Chronic inflammatory bowel disease of colon, if more precise diagnosis or 9 1.0 determination of textent of activity of disease will influence immediate management In patients with ulcerative or Crohn's pancolitis for 8 y or more or left-sided colitis for 15 y or more every 1-2 y with systematic biopsies to detect dysplasia

Table 3 -- Indications in 151 inappropriate colonoscopies Indication No. % Surveillance of colonic polyps out of recommended intervals (3- to 5-y intervals after 49 32.4 adequate clearance of neoplastic polyps) Transitory or already endoscopically investigated unmodified chronic abdominal pain 30 19.9 Transitory change in bowel habit 21 13.9 Colorectal carcinoma surveillance out of guidelines (colonoscopy to remove 20 13.2 synchronous neoplastic lesion at or around time of curative resection of cancer followed by colonoscopy at 3 y and 3 to 5 y thereafter to detect metachronous cancer) Melena with upper GI source already identified 9 6.0 Screening in patients with family history of sporadic colorectal cancer before age 60 y 6 4.0 out of guidelines (colonoscopy every 5 y beginning 10 y earlier than affected relative or every 3 y if adenoma is found) Hematochezia in patients <40 y without previous rectal evaluation 4 2.6 Follow-up for inflammatory bowel diseases out of recommended intervals 3 2.0 Chronic inflammatory bowel disease of colon, if more precise diagnosis or determination of extent of activity of disease will influence immediate management In patients with ulcerative or Crohn's pancolitis for 8 y or more or left-sided colitis for 15 y or more every 1-2 y with systematic biopsies to detect dysplasia Anal symptoms 3 2.0 Rectal incontinence 2 1.3 Abnormal serologic markers (carcinoembryonic antigen, cancer antigen 19-9) 2 1.3 Metastatic adenocarcinoma of unknown origin without colonic symptoms when it will not 1 0.7 influence management Inguinal 1 0.7

Table 4 -- Pathologic findings Appropriate procedures Inappropriate procedures (n = 866) (n = 151) No. % No. % Low-grade dysplasia adenoma 92 41.8 9 6.0 Colorectal cancer 71 32.3 — — Undetermined polyps (unretrieved 25 11.4 5 3.3 polyps <5 mm) High-grade dysplasia adenoma 25 11.4 2 1.3 Inflammatory bowel disease 17 7.7 — — In situ adenocarcinoma 3 1.4 — — Total 233 26.9 16 10.6

Overuse and underuse of colonoscopy in a European primary care setting Gastrointestinal Endoscopy - Volume 52, Issue 5 (November 2000) Table 1. Major categories of clinical conditions for which colonoscopy might be considered Unexplained hemoccult positive stools or iron deficiency anemia Hematochezia Uncomplicated lower abdominal pain or change in bowel habits of 3 weeks' or more duration Evaluation of known inflammatory bowel disease Surveillance following polypectomy Surveillance or screening for colorectal cancer Miscellaneous conditions

Use of this content is subject to the Terms and Conditions Colonoscopy: A Review of Its Yield for Cancers and Adenomas by Indication American Journal of - Volume 90, Issue 3 (March 1995) - Copyright © 1995 Elsevier

353

Clinical reviews

Colonoscopy: A Review of Its Yield for Cancers and Adenomas by Indication

Douglas K. Rex M.D., F.A.C.G.

Division of Gastroenterology/Hepatology, Department of Medicine, Indiana University School of Medicine, Indianapolis, Indiana

Colonoscopy for bleeding indications (positive test, emergent or nonemergent rectal bleeding, melena with a negative upper endoscopy and iron deficiency anemia) has a substantial yield for cancers (1 per 9 to 13 colonoscopies), although slide rehydration of fecal occult blood tests decreases the yield (1 per 45 colonoscopies). Prospective studies indicate that nonbleeding colonic symptoms have a substantially lower yield for cancer than bleeding symptoms (1 per 109 colonoscopies).

Patients with indications for screening colonoscopy with a relatively high yield of cancer are those with Lynch syndrome (1 per 39 colonoscopies) and males more than 60 yr old (1 per 64 colonoscopies).

Perioperative colonoscopy in persons undergoing colorectal cancer resection has a high yield for synchronous cancer (2-3%). An initial examination in persons with long-standing has a high yield for cancer (12%). Surveillance colonoscopy after cancer resection has an intermediate yield for anastomotic cancer (1 per 74 procedures) and metachronous cancers (1 per 82 colonoscopies), although this number may overestimate the yield of metachronous cancer.

Postpolypectomy surveillance and ulcerative colitis surveillance colonoscopy have relatively low yields for cancer (1 per 317 and 360 colonoscopies, respectively). However, postpolypectomy surveillance colonoscopy, in combination with initial clearing colonoscopy, has been proven to be almost entirely effective in preventing colorectal cancer death.

Further, cancer yields for postpolypectomy surveillance should improve with implementation of new surveillance guidelines, with little or no impact on mortality. The effectiveness of ulcerative colitis surveillance is less certain. Referral of patients with low-grade dysplasia for would improve the value and effectiveness of surveillance colonoscopy in ulcerative colitis.

Adenoma yields at colonoscopy are relatively independent of indication, as evidenced by the high yield of adenomas in screening colonoscopy studies. Demographic factors, including increasing age and male gender, are important predictors of adenomas at initial colonoscopy. Age, male gender, and multiple and large adenomas at initial examination are predictors of adenomas at subsequent postpolypectomy surveillance. Persons with ureterosigmoidostomies can be screened by interval flexible sigmoidoscopy and do not require screening colonoscopy. A history of breast cancer does not predict an improved yield of colonoscopic screening compared to average-risk persons.

Reprint requests and correspondence: Douglas K. Rex, M.D., Gastroenterology/Hepatology, 550 University Blvd., Room 2300, Indianapolis, Indiana 46202-5244.

Received July 6, 1994; accepted November 7, 1994.

INTRODUCTION

Colonoscopy is a powerful diagnostic and therapeutic procedure. Although colonoscopy can diagnose and treat many conditions, a principal focus of its use is the diagnosis of colorectal cancer and the diagnosis and removal of adenomatous colon polyps. Removal of adenomas is one of the greatest advantages of colonoscopy and is now known to prevent the development of colorectal cancer [1] [2] [3] . In many locations, colonoscopy has been proven superior to other methods, such as double contrast barium enema, for detection of early colorectal cancers as well as adenomas [4] [5] [6] [7] [8] . Because colorectal cancer is a major cause of mortality in most western countries, and because the potential to prevent mortality by removal of benign precursor lesions is so unusual in cancer, colonoscopy is one of the most powerful tools currently in use in clinical medicine.

Detection and treatment of colorectal cancers and adenomas is overall the most important outcome of colonoscopy. The factors that best predict the presence of colorectal cancers and adenomas at colonoscopy are different. Adenomas are largely asymptomatic lesions that are best predicted by demographic factors such as increasing age and male gender [9] [10] [11] , and to a much lesser extent, by the indication for the procedure [12] . Cancer prevalence at colonoscopy varies considerably according to the indication for the procedure. The focus of this paper is to review the relationship of procedure indication to the prevalence of colorectal neoplasia, particularly cancer, detected at colonoscopy.

METHODS

Papers were selected for review through Medline searches and through examination of reference lists in

354

TABLE 1 -- Yield of Colonoscopy in Patients with Positive Fecal Occult Blood Tests N Cancers Adenoma Slides Author Patients a a Comment Nonrehydrated Ahlquist [13] 1016 NS [4] NS [17] 70% had CS, others positive test (701 by evaluated by other TABLE 1 -- Yield of Colonoscopy in Patients with Positive Fecal Occult Blood Tests N Cancers Adenoma Slides Author Patients a a Comment CS) means Fruhmorgen [8] 127 13 [10] 51 [40] Frommer [14] 50 11 [22] 13 [26] Hardcastle [15] 618 63 [10] 266 [43] Not clear how many were repeated before work-up Kronberg [16] 209 14 [7] 86 [41] Sontag [6] 120 12 [11] 44 [37] Winawer [4] 467 50 [11] 92 [19] Increased PPV with age; PPV decreases with subsequent screens Total 1591 163 [10] 552 [35] (excluding Ahlquist) Rehydrated positive Frommer [14] 91 15 [16] 21 [23] test Mandel [17] >12,246 NS (2.2) NS Negative on repeat Kewenter b [20] 108 1 [1] NS [6] Used FS and DCBE; decreased PPV on 2nd screen Positive on repeat Elliot [18] 32 12 [38] 11 [34] Not rehydrated Hobbs [19] 16 6 [38] 5 [31] Not rehydrated Kewenter [20] 67 10 [15] 15 [22] Both rehydrated CS, Colonoscopy; PPV, positive predictive value; NS, not stated; PPV, positive predictive value; FS and DCBE, flexible sigmoidoscopy and air contrast barium enema. a Number of patients (percent). b This is the only study reviewed in this manuscript in which patients were evaluated primarily by flexible sigmoidoscopy and double contrast barium enema rather than by colonoscopy.

published articles. With rare exception, preference was given to articles using colonoscopy or primarily colonoscopy to evaluate patients. In some cases, papers were excluded because they lumped categories of indications together ( e.g., occult and overt rectal bleeding lumped together as rectal bleeding). In some situations, endoscopy programs reported their results in multiple papers over the course of time, as the number of patients studied and the length of follow-up progressively increased. This was particularly true of colonoscopy surveillance programs for ulcerative colitis. In these cases, efforts were made to only include the latest report to avoid counting individual patients more than once. Calculation of cancers per procedure for surveillance indications (postcancer resection, postpolypectomy, ulcerative colitis) reflect the total number of examinations in the reviewed studies. This total number is affected by the interval between examinations, which was often shorter than is currently recommended. Cancers per procedure for all other indications, including screening, reflect the results of a single colonoscopy per patient.

POSITIVE FECAL OCCULT BLOOD TESTS

Colonoscopy may reach its greatest utility for detection of colonic neoplasia in persons with positive fecal occult blood tests. Not only is the yield for cancers and adenomas high in persons with positive fecal occult blood tests, but for cancers there is a considerably higher percentage of earlier Dukes classes (Dukes A and B) than is found in symptomatic persons [4] . Factors influencing the yield for cancer include the individual's age, whether the test was positive or negative if repeated, and whether rehydration was used. The method of testing has a recognizable but lesser impact on the yield of adenomas (Table 1) .

In clinical trials, fecal occult blood tests have generally been studied without rehydration [4] [6] [8] [13] [14] [15] [16] , because rehydration is known to decrease the specificity and positive predictive value of occult blood testing [17] . In general, full colonic imaging has been used to evaluate a single positive test (one or more positive slides out of six). Under these circumstances, colonoscopy will detect cancer in 10% (range 7-22%) and adenomas in 35% (range 19-43%) of positive persons (Table 1) . Increasing age among persons with positive tests is highly correlated with a higher yield of colonoscopy for cancers and adenomas [4] .

The positive predictive value of fecal occult blood testing for cancer (and thus the yield of colonoscopy) can be improved by repeating the test in persons who are initially positive and by reserving colonic imaging for those who remain positive on the second test [18] [19] . Individuals in whom two consecutive tests are positive with nonrehydrated slides have the highest yield of any indication for colonoscopy, 38% for cancer and 33% for adenomas (Table 1) .

Although rehydration of fecal occult blood tests has not been advocated generally, enthusiasm for rehydration may increase because a recent prospective randomized trial of fecal occult blood testing, using primarily rehydration, demonstrated that the combination of annual fecal occult blood testing with colonoscopy performed for positive tests resulted in a 33% reduction in colorectal cancer mortality

355

TABLE 2 -- Yield of Colonoscopy in Persons with a Positive Fecal Occult Blood Test by Digital Rectal Examination Adenomas Author N Patients Cancers a a Comments TABLE 2 -- Yield of Colonoscopy in Persons with a Positive Fecal Occult Blood Test by Digital Rectal Examination Adenomas Author N Patients Cancers a a Comments Gomez [21] 9 1 [11] 2 [22] Asymptomatic Kim [22] 72 3 [4] 14 [19] Not asymptomatic Reilly [23] 356 NS [7] NS [36] Not asymptomatic Rex [24] 164 13 [8] 63 [38] Not asymptomatic NS: not stated. a Number patients (percent).

[17] . The yield of colonoscopy in persons with positive fecal occult blood tests after rehydration is 2.2% for cancer [17] , although it was higher (16%) in one smaller study [14] . Thus, from the perspective of yield of colonoscopy, the major impact of rehydration is to reduce the yield of colonoscopy for cancer, with probably less impact on the yield for adenomas. As with nonrehydrated slides, age of the person with the positive test and the results of any repeat test influence the expected yield of colonoscopy. In one study in which both initial and repeat tests were performed with rehydration, with colonic imaging performed only in those cases positive on both testings, there was significant improvement in specificity with minimal loss of sensitivity [20] .

In clinical practice, patients are often referred for colonic imaging after feces obtained by digital rectal examination test positive for fecal occult blood. Several studies have examined the yield of colonoscopy in patients with

356

TABLE 3 -- Yield of Colonoscopy in Patients with Nonemergent Rectal Bleeding Adenomas Author Comment N Patients Cancer a a Bat [33] All patients bleeding >2 months duration 101 29 [29] 34 [34] all patients over 80 yr Berkowitz Retrospective only; 69% cecal intubation 123 11 [9] 15 [12] [34] Brand [35] All patients with previous negative 221 16 [7] 51 [23] barium enema Church [32] Suspicious type 59 13 [22] 16 [27] Church [32] Outlet type 115 5 [4] 35 [30] TABLE 3 -- Yield of Colonoscopy in Patients with Nonemergent Rectal Bleeding Adenomas Author Comment N Patients Cancer a a Goulston [31] All age > 49 and bleeding < 6 months 145 15 [10] 25 [17] duration Graham [36] Prospective 33 3 [9] 14 [42] Guillem [37] 11/106 with negative barium enema had 224 33 [15] 61 [27] cancer Irvine [38] 71 5 [7] 26 [37] Swarbrick All had negative barium enema 239 23 [10] 39 [16] [39] Teague [40] All had negative barium enema, mostly 109 6 [6] 18 [17] DCBE Thoeni [41] Retrospective; barium enema and 215 27 [13] 29 [13] colonoscopy Total 1655 186 [11] 363 [22] DCBE, double contrast barium enema. a Number of patients with lesion (percent).

positive fecal occult blood tests by digital examination and have found a substantial prevalence of cancer and adenomas (21-24; Table 2) . However, patients in these studies were generally symptomatic [22] [23] [24] . There is almost no data on the yield of colonoscopy in asymptomatic persons with positive fecal occult blood tests obtained by digital examination. The specificity of fecal occult blood testing by digital rectal examination in the office could be impaired by factors such as anal trauma, inadequate control of diet, and medications associated with false positive tests, so it appears preferable to not test feces obtained by digital examination for occult blood. However, when patients with positive fecal occult blood tests by this method are referred for evaluation, there is insufficient data available at this time to justify ignoring the positive test. Either the test should be repeated properly, or the patient should undergo colonic imaging.

In clinical practice, positive fecal occult blood tests are still evaluated in some cases by barium enema. In general, air contrast barium enema is preferred over single contrast barium enema and should be performed in combination with flexible sigmoidoscopy [25] [26] [27] . Unfortunately, in clinical practice, barium enema is still usually performed as a single test [28] . Some have argued that fecal occult blood tests should not be evaluated by air contrast barium enema, because the test has poor sensitivity. Indeed, five prospective evaluations (four using air contrast studies) in which patients with positive fecal occult blood tests were evaluated by both barium enema and colonoscopy have found that barium enema misses 25- 33% of cancers and up to 57% of polyps [4] [5] [6] [7] [8] . The explanation for the low sensitivity of barium enema for cancer in patients with fecal occult blood tests may be their early (A and B) distribution of Dukes' classes, which likely makes cancers in this population more difficult to detect by barium enema than those in symptomatic patients. These alarming findings support the use of colonoscopy in all asymptomatic patients with positive fecal occult blood tests. Exceptions may be young persons with positive fecal occult blood tests, in whom the chance of finding colorectal cancer by colonoscopy is much lower, and thus the chance of missing cancer by air contrast barium enema is much less.

RECTAL BLEEDING

Nonemergent rectal bleeding is a common symptom and is noted intermittently by about 15% of apparently well persons [29] . Some reports of the yield of colonoscopy in the evaluation of nonemergent rectal bleeding do not carefully distinguish the type or nature of the bleeding, although this certainly bears importantly on the yield of colonic imaging in these patients [30] . Patients reporting small amounts of blood on the toilet paper, blood dripping from the anus after bowel movements or being passed without bowel movements have been considered to nearly always have an anal source of bleeding [30] . Although this rule regarding anal bleeding is certainly largely accurate, a prospective evaluation of general practitioners and gastroenterologists found that they were not able to distinguish reliably anal from colonic bleeding by history alone in a significant number of patients aged more than 40 yr with rectal bleeding of less than 6 months duration [31] . Specifically, 25% of the patients with cancer were believed by history to have an anal source of bleeding [31] . In another prospective evaluation of patients with rectal bleeding, Church [32] characterized bleeding as outlet type (bright red blood seen during or after defecation, on the toilet paper, or in the toilet bowel) or as suspicious (dark red blood, and/or blood mixed with or streaked on stool, or blood associated with a change in bowel habit or mucus passage). Patients with suspicious bleeding were much more likely to have cancer than patients with outlet type bleeding, but the yield of cancer and adenomas in those with outlet type bleeding was also important (Table 3) . A variety of other reports [33] [34] [35] [36] [37] [38] [39] [40] [41] of colonoscopy in patients with nonemergent rectal bleeding, which generally did not carefully characterize the nature of bleeding, also found a significant yield of cancers and adenomas (Table 3) . Thus, nonemergent rectal bleeding portends a substantial yield of neoplasia at colonoscopy, including cancers. In the case of blood mixed with stool, this yield of cancers equals or exceeds the yield seen in patients with positive fecal occult blood tests (Tables 1 , 3) . The yield of adenomas in patients with rectal bleeding is somewhat more variable in reported series (Table 3) than that in persons with positive fecal occult blood tests (Table 1) and may reflect variable reporting of persons with small adenomas and age variation between series. As is the case with positive fecal occult blood tests, the age of the patient with rectal bleeding influences the findings at colonoscopy. Thus, in a group of octogenarians with nonemergent rectal bleeding, 29% had cancer at colonoscopy [33] (Table 3) . Likewise clinical experience and common sense suggest that some young patients with only anal-type bleeding can be safely evaluated by flexible sigmoidoscopy alone.

Colonoscopists have repeatedly reported that patients with persistent rectal bleeding after a negative barium enema have a high prevalence of cancers and adenomas at colonoscopy [35] [37] [39] [40] . Thus, colonoscopy is indicated in any patient with persistent rectal bleeding after a negative barium enema. Colonoscopy has been used in several studies to evaluate patients with acute lower gastrointestinal hemorrhage [32] [37] [42] [43] [44] . Colonoscopy has substantial diagnostic and therapeutic value in this setting beyond the detection and treatment of neoplasia [42] [43] [44] . Nevertheless, 8% of patients will have cancer and 14% will have adenomas (Table 4) . This yield of cancers is lower but comparable to that in patients with nonemergent bleeding (Table 3 , 4) .

Two small studies have reported the yield of colonoscopy in patients presenting with melena and having a normal upper endoscopy (35, 45; Table 5) . In both studies, significant numbers of colorectal cancers and polyps were detected by colonoscopy, even though in one study patients had also had negative barium enemas before undergoing colonoscopy [45] .

Colonoscopy also has a substantial yield (Table 6) for cancer in patients with iron deficiency anemia [34] [46] [47] , particularly in patients with positive fecal occult blood tests [46] . For uncertain reasons, the reported yield of colonoscopy for adenomas

TABLE 4 -- Yield of Colonoscopy for Neoplasia in Patients with Acute Hemorrhage Adenomas Author N Patients Comment Cancer a a Caos [42] 31 1 [3] 5 [16] Church [32] 27 3 [11] 1 [4] Deyhle [43] 18 1 [6] 3 [17] Guillem [37] 93 Recent major 7 [8] 15 [16] 55 Active acute 7 [13] 8 [15] Jensen [44] 80 9 [11] b Total c 224 19 [8] 32 [14]

a Number of patients with lesion (percent). b Cancers and adenomas were not separated in this paper. c Does not include Jensen paper.

TABLE 5 -- Yield of Colonoscopy for Neoplasia in Persons with Melena and Negative Upper Endoscopy Author N Patients Cancer b Adenoma b Brand [35] 26 3 [12] 6 [23] Tedesco a [45] 53 5 [9] 4 [8] b Number patients (percent). a All had negative barium enema before colonoscopy.

has been lower in patients with iron deficiency anemia than for other bleeding indications (Tables 1 - 6) .

COLONIC SYMPTOMS WITHOUT BLEEDING

Colonoscopy has been used several times to examine the prevalence of neoplasia in persons with abdominal symptoms who have no evidence of gastrointestinal bleeding (34, 48-50; Table 7) . Patients in these studies have generally had abdominal pain, altered bowel habit, and/or weight loss. In one study [49] , patients also had at least one negative fecal occult blood test. The prevalence of cancer in these studies has been generally low (Table 7) . In each study, the yield of cancer has been substantially lower than a group of patients

357

TABLE 6 -- Yield of Colonoscopy for Neoplasia in Patients with Iron Deficiency Anemia Author Comment N Patients Cancer a Adenoma a Alemayehue [46] With + FOBT 90 8 [9] NC Without + FOBT 38 1 [3] NC Berkowitz [34] Only 69% cecal 46 1 [2] 4 [9] intubation Rockey [47] 100 11 [11] 5 [5] FOBT, fecal occult blood test; NC, not clear. a Number of patients with lesions (percent).

TABLE 7 -- Yield of Colonoscopy in Patients with Abdominal Pain, Altered Bowel Habit Adenomas Comment N Patients Cancer a a Berkowitz [34] Abdominal pain 55 1 [2] 6 [11] Berkowitz [34] Altered bowel habit 79 1 [1] 8 [10] Brenna [48] 117 1 [1] 7 [6] Neugut [50] Persistent symptoms 315 15 [5] 60 [19] Rex [49] All with 1 negative 75 0 (0) 23 [31] TABLE 7 -- Yield of Colonoscopy in Patients with Abdominal Pain, Altered Bowel Habit Adenomas Comment N Patients Cancer a a FOBT Total 641 18 (2.8) 104 [16] FOBT, fecal occult blood test. a Number of patients with lesions (percent). with rectal bleeding examined at the same institution. Only one study found a significant yield of cancer [50] . Patients in this study had "persistent" nonbleeding symptoms, suggesting the possibility of significant selection bias. The prevalence of adenomas in this population largely reflects the background prevalence in an asymptomatic population [50] . Thus, abdominal symptoms with no evidence of bleeding are a relatively poor indication for colonoscopy. However, in older patients, the background prevalence of adenomas is sufficiently high that colonoscopy is likely more cost-effective than initial flexible sigmoidoscopy plus air contrast barium enema [51] , if colonic imaging is considered necessary. Likewise, colonoscopy can be considered in patients with recent onset but "persistent" nonbleeding symptoms, even if barium enema is negative [50] .

SURVEILLANCE AFTER COLORECTAL CANCER RESECTION

Clearing colonoscopy in the setting of cancer resection refers to colonoscopy performed before resection (or shortly after resection in obstructed patients) to identify and/or remove synchronous colonic neoplasia. Clearing colonoscopy has a high yield for synchronous cancer (2.2%) and adenomas (27%) [52] [53] .

Surveillance colonoscopy refers to interval colonoscopy after the clearing procedure. Most authors reporting the yield of surveillance colonoscopy after colorectal cancer resection have performed the procedure at 6-month to 2-yr intervals [48] [52] [53] [54] [55] [56] [57] [58] [59] [60] [61] [62] . Anastomotic and metachronous cancers occur during surveillance with approximately equal frequency (Table 8) . Unfortunately, most anastomotic recurrences are accompanied by intra- abdominal recurrence that is unresectable for cure [63] [64] . Therefore, detection of an anastomotic recurrence is only rarely of long-term therapeutic benefit. Detection of metachronous cancers occurs in approximately 1 in 37 patients and 1 in 82 colonoscopies (Table 8) . However, more than two-thirds of reported "metachronous" cancers are from series in which it was not fully apparent that a clearing colonoscopy was performed in the perioperative period [48] [54] [56] [57] [60] . Thus, some of these tumors were certainly synchronous lesions that were not recognized at the time of surgery. Nevertheless, even excluding these series, the detection rate of metachronous cancers appears higher than that reported in postpolypectomy surveillance studies (Tables 8 , 9) , whereas adenoma detection rates are comparable in postadenoma and postcancer resection surveillance (Tables 8 , 9) . Therefore, postcancer resection surveillance colonoscopy has a substantial yield for neoplasia. However, because detection of anastomotic recurrences seldom benefits patients, and because metachronous cancers are likely slow-growing tumors with biological characteristics similar to initial cancers, it has been recommended [65] that surveillance intervals could be safely increased relative to those usually reported [48] [52] [53] [54] [55] [56] [57] [58] [59] [60] [61] [62] . This would likely improve the cancer per procedure yield without negatively impacting survival, although this remains unproven. Surveillance at 1 yr after the clearing colonoscopy, followed by colonoscopy at 3-5 yr intervals, appears reasonable and safe for most patients.

SURVEILLANCE AFTER REMOVAL OF ADENOMAS

Several studies [1] [48] [59] [66] [67] [68] [69] [70] [71] [72] have reported the yield of surveillance colonoscopy after endoscopic removal of adenomatous colon polyps (Table 9) . One of the largest and most recent of these was the National Polyp Study [71] . Most studies report examinations performed 1-3 yr after a colonoscopy that was believed to have cleared the colon of neoplasia. The combined results of these remarkably consistent studies indicate a substantial incidence of adenomas and a very low incidence of cancers in the 3 yr after a clearing colonoscopy (Table 9) . By comparison to reference populations, two surveillance studies have calculated that colonoscopy and polypectomy reduces the incidence of colorectal cancers and mortality from colorectal cancer [66] [71] during the surveillance interval. This is certainly supported by the finding of only one death from colorectal cancer [66] among 3153 patients entered into surveillance. Low mortality is related to very early Dukes stages among the 19 observed cancers during surveillance. Combining the studies, the number of colonoscopies performed per cancer detected is 317. However, colonoscopy intervals for most patients were less than the 3 yr recently found to be adequate

358

TABLE 8 -- Yield of Surveillance Colonoscopy Postcancer Resection N Anastomotic Metachronous Author N Patients Colonoscopies Recurrence a Cancer a Adenoma a Brady [54] 207 445 6 (2.9) 9 (4.3) NC Brenna [48] 51 51 0 (0) 2 (3.9) 20 [40] Carlsson b [52] 129 358 NS 1 (0.7) 43 [33] Hall [55] 54 54 0 (0) 1 [2] 3 [6] Jahn [56] 529 1244 9 (1.7) 13 (2.4) NC Juhl [57] 133 316 9 (6.8) 4 [3] 51 [38] Kronberg [53] 239 710 5 (2.1) 4 (1.7) 32 [13] Larson [58] 68 135 2 (2.9) 3 (4.4) 24 [35] McFarland [59] 74 237 2 (2.7) 0 (0) 25 [34] Nava b [60] 240 304 17 (7.1) 11 (4.6) NC Unger [61] 56 56 1 (1.8) 0 (0) 12 [21] Weber [62] 75 197 0 (0) 2 (2.6) 13 [17] TABLE 8 -- Yield of Surveillance Colonoscopy Postcancer Resection N Anastomotic Metachronous Author N Patients Colonoscopies Recurrence a Cancer a Adenoma a Total c 1855 4107 51 (3.0) 50 (2.7) 223 [25] NC, Not clear; NS, not stated. a Number of patients with given finding at one or more examinations (percent of patients entering surveillance). b Patients may overlap between these two studies. c Percentages are based only on studies giving data for given result.

TABLE 9 -- Yield of Surveillance Colonoscopy Postadenoma Removal N Colonos- Author N Patients copies Cancers a Adenoma a Brenna [48] 44 44 0 (0) 14 [32] Jorgenson [66] 1056 2721 10 (0.9) 347 [33] Matek [67] 592 1281 2 (0.3) 94 [16] McFarland [59] 84 252 2 (2.3) 29 [12] Nava [68] 44 105 0 (0) 26 [60] Neugut [69] 118 165 0 (0) 35 [30] Waye [70] 133 133 0 (0) 75 [56] Winawer [71] 973 1311 5 (0.5) 280 [29] Woolfson [72] 109 109 0 (0) 54 [50] Total 3153 6121 19 (0.6) 954 [30] a Number of patients with given lesion at one or more surveillance examinations (percent of patients entering surveillance). by the National Polyp Study [71] . Use of a 3-yr interval before the first surveillance examination in most adenoma patients would reduce the number of surveillance colonoscopies to detect one cancer to approximately 200, with almost certainly very little impact on survival. In fact, survival might improve, considering that deaths related to colonoscopy complications currently exceed deaths from colorectal cancer in surveillance studies [66] . A low yield for cancers in a surveillance program is expected after effective clearing colonoscopy. Thus, although the yield of surveillance colonoscopy for cancers is low, the value of surveillance colonoscopy clearly exceeds that implied by the cancer yield. That is, the combination of initial clearing colonoscopy and subsequent surveillance examinations dramatically prevents colorectal cancer incidence and mortality in persons with adenomas [1] . The overall incidence of adenomas in surveillance examinations is remarkable (Table 9) , considering that these patients have already had a clearing colonoscopy, and approaches or exceeds the prevalence of adenomas found in initial examinations (Table 1 - 7) . However, detected adenomas during surveillance are generally smaller than those at the initial examination [66] [67] [68] [69] [70] [71] [72] . Detection of patients with adenomas greater than or equal to 1 cm in size varies from 2.5% in the National Polyp Study [71] to 9.3% in the Funen Adenoma Follow-Up Study [66] . The explanation for this variation is not clear but is partly accounted for by exclusion of patients with index polyps greater than or equal to 3 cm in size in the National Polyp Study. Older patients, males, and those with multiple or large (>1 cm in size) adenomas at the initial colonoscopy are more likely to have adenomas at the surveillance examination [66] [71] . Multiple adenomas on initial examination predicts advanced (>1 cm in size, high grade dysplasia, or invasive cancer focus) adenomas at follow-up. Currently, most patients with adenomas can undergo follow-up colonoscopy at 3 yr, with subsequent examinations at 3 yr if adenomas are detected and at 5 yr if no neoplasia is found [73] . Future work will likely focus on whether low-risk groups such as those with only tubular adenomas less than 1 cm in size can safely undergo surveillance at 5 or more yr.

POSITIVE FAMILY HISTORY

A history of colorectal cancer in one or more first-degree relatives is associated with an increased risk of colorectal cancer [74] [75] [76] [77] . The risk is increased with multiple affected relatives and decreasing age of affected relatives [74] . Several groups have reported the yield of screening colonoscopy in asymptomatic persons with a positive family history of colorectal cancer (78-87; Table 10) . In general, these studies have not prescreened subjects with fecal occult blood tests (Table 10) , although there have been exceptions [83] [86] . In addition, several of the studies have been retrospective

359

TABLE 10 -- Yield of Screening Colonoscopy in Persons with a Positive Family History Author Design N a Cancersb Adenomasb Comments Baker [78] Retrospective 201 4 [2] 50 [25] No FOBT Mean age 52.4 One perforation occurred Grossman Prospective; ? 154 0 (0) 28 [18] 23% had adenoma(s) [79] systematic FOBT in those age 50 34% under age 50 Gryska [80] Retrospective 49 2 [4] 31 [63] No FOBT Mean age 54.5 TABLE 10 -- Yield of Screening Colonoscopy in Persons with a Positive Family History Author Design N a Cancersb Adenomasb Comments Guillem [81] Controlled 181 0 (0) 26 [18] Controls (8.8%) Prospective adenomas; male No FOBT gender and increased age also predictors Luchtefeld Retrospective 160 0 (0) 17 [11] 14% had adenoma(s) [82] No FOBT in those age 50 46% 50 years McConnell Prospective 125 0 (0) 15 [12] 15% had adenoma(s) [83] Systematic FOBT in those age 50 31% under age 50 Meagher Prospective 136 1 [1] 49 [36] [84] No FOBT Orrom [85] 90% first-degree 114 2 [2] 21 [18] 25% had adenoma(s) relatives Prospective in those age 50 No FOBT Mean age 51 Rex [86] Controlled 109 0 (0) 33 [30] Controls (26%) Prospective; adenomas; male systematic FOBT gender and increased All age 50 age also predictors Rozen [87] Prospective 39 0 (0) 14 [36] Two or more relatives Total 1268 9 (0.7) 284 [22] N, number of subjects screened; FOBT, fecal occult blood test. a Number of subjects with lesion (percent). surveys (Table 10) , and thus the extent to which subjects were truly asymptomatic can be questioned. None of the studies described inclusion of subjects who may have been members of familial colon cancer kindreds. However, certain studies did include subjects with more than one affected relative, and one studied only patients with two or more relatives [87] . In eight uncontrolled studies and two controlled studies (Table 10) , a total of 1268 asymptomatic persons with a positive family history underwent colonoscopy. Nine cancers were detected (0.7%), and there was a 22% prevalence of adenomas. This represents one cancer detected per 141 screening colonoscopies. However, six of the nine cancers were detected in studies that were retrospective, and all nine cancers were in studies that did not prescreen subjects with fecal occult blood tests (Table 10) . This is of some importance in comparing the results to studies of asymptomatic average-risk persons (Table 11) , all of which were prospective and generally excluded individuals with positive fecal occult blood tests. Of the seven prospective (Table 10) screening colonoscopy studies of persons with a positive family history of colorectal cancer, two of which systematically tested for occult blood and excluded positives, one cancer was detected per 286 screening examinations, and the prevalence of adenomas was unchanged at 22%. In a controlled trial of screening colonoscopy involving 109 persons with a positive family history of colorectal cancer and 496 persons with only average risk, positive family history of a single first-degree relative with colorectal cancer diagnosed at an age greater than 60 yr was not shown to predict a higher prevalence of either cancer or adenoma [86] . There was a trend toward a predictive effect if the affected relative was less than 60 yr old at diagnosis [86] . The other controlled trial [81] had a case control design and did find that family history was predictive of a higher risk of adenomas, but the prevalence of adenomas in controls was unexpectedly low (Table 10) . The prevalence of adenomas in uncontrolled trials, when adjusted for age [79] [82] [83] [85] , was not higher than that seen in screening colonoscopy trials in average risk persons (Table 11) . It is uncertain why the increased lifetime risk of colorectal cancer in persons with a positive family history has thus far not been associated with a more impressive yield of cancers and adenomas during screening colonoscopy in these individuals. It should be noted that one

360

TABLE 11 -- Yield of Screening Colonoscopy in Asymptomatic Average-Risk Persons Author Comment N Cancers a Adenomas a Foutch [93] Prospective; age 50 114 2 (1.7) 47 [41] Veterans; all had negative FOBT Guillem [81] No FOBT Included subjects 52 over age 50 0 (0) 7 [13] 50 Johnson [94] Prospective; age 50 89 0 (0) 21 [24] All had negative FOBT Lieberman [95] Prospective; age 50 105 1 [1] 43 [41] Veterans all had negative FOBT Rex [86] Prospective; age 50 496 3 (0.6) 128 [26] All had negative FOBT Total 856 6 (0.7) 246 [29] N, number of subjects screened; FOBT, fetal occult blood test. a Number of subjects with lesion (percent). controlled trial using flexible sigmoidoscopy did find that positive family history was a significant predictor of adenomas [88] .

Because screening colonoscopy trials do not clearly define family history groups (other than Lynch syndrome; see below) with increased risk, recommendations for screening colonoscopy must be based on lifetime colorectal cancer risk. For example, interval screening colonoscopy appears clearly appropriate in persons with multiple affected first-degree relatives or first-degree relatives diagnosed at less than or equal to 55 yr of age [74] . Screening should probably begin 10-15 yr before the age of the youngest affected relative. Even persons with a single affected first-degree relative aged more than 55 yr appear to be at some increased risk [74] .

Two small studies [89] [90] have examined the yield of screening colonoscopy in asymptomatic relatives of affected probands with Lynch syndrome colorectal cancers (Table 12) . The Lynch syndromes are characterized by autosomal dominant inheritance of colorectal cancer predisposition, with a predominance of right-sided cancers. Recommendations have generally been for potentially affected relatives to undergo interval colonoscopy beginning at ages 20-25 yr. Application of this recommendation in the two screening studies has yielded an unexpectedly high prevalence of adenomas in a relatively young population, as well as a 2% prevalence of cancer (Table 12) . In a third study including both initial and interval examinations, cancers were detected in six of 160 asymptomatic family members [91] . These results support the screening recommendation. The yield of continued interval colonoscopy in individuals who are initially negative is unknown, but almost certainly lower because of preselection of persons less likely to carry the responsible gene. Reports of cancers developing within 3-5 yr of negative colonoscopy [92] suggests that surveillance intervals should not be expanded beyond 2-3 yr at this time. Development of a commercially available genetic screen is expected to identify at-risk individuals in some Lynch syndrome kindreds, which would in turn significantly affect the yield of screening and surveillance colonoscopy.

ASYMPTOMATIC AVERAGE-RISK PERSONS

Five studies have evaluated screening colonoscopy in asymptomatic average-risk persons (81, 86, 93-95; Table 11) . Four studies included only subjects aged 50 yr or older. Each of the five studies was prospective, and four required that all subjects have negative fecal occult blood tests. A total of four asymptomatic persons with colorectal cancer were detected by fecal occult blood testing and were excluded from the studies before colonoscopy [86] [94] . This feature of the studies is important in comparing them to uncontrolled screening trials of colonoscopy in persons with a positive family history (see above), which did not in all cases prescreen subjects for fecal occult blood. Two of the studies were performed in veterans [93] [95] . The largest study was performed in a group of volunteer health care professionals [86] . Combining results from the five studies, the yield of cancers is low (0.7%), and the prevalence of adenomatous polyps is 29%. One cancer was detected per 143 colonoscopies. Increasing age and male gender were both strong predictors of the presence of adenomas [86] [94] . All six of the cancers and nearly all of the adenomas greater than or equal to 1 cm were in males more than 60 yr of age. If only males aged greater than 60 yr are considered, then 1.6% had cancer and 38% had adenomas [86] [93] [94] [95] . In this group, one cancer was detected per 64 screening colonoscopies performed. Approximately 10% of males more than 60 yr of age will have an adenoma greater than or equal to 1 cm in size [86] [88] [89] [90] [93] [94] [95] . The high yield of screening colonoscopy in males is consistent with recent reports that colorectal cancer incidence and mortality are now each approximately 1.6-fold greater in males than females in the United State [96] . Screening colonoscopy in average-risk males more than 60 yr old appears justified because it detects and treats significantly more cancers and advanced adenomas than some other accepted indications for colonoscopy. ULCERATIVE COLITIS SURVEILLANCE

The degree of increased risk of colorectal cancer in patients with long-standing ulcerative colitis remains controversial

361

TABLE 12 -- Yield of Screening Colonoscopy in Family Members of Lynch Syndrome Patients Author Comment N Cancer a Adenoma a Lanspa [89] Mean age 44 0 (0) 13 [30] 42.2 yr Love [90] Mean age 34 (5%) 1 cancer in 5 [15] 34.5 yr a 20 year old N, number of subjects screened. a Number of subjects with lesion (percent).

[97] . The increased risk is related to the extent of colitis and the duration of disease. Variable results have been obtained with regard to the risk associated with early age at onset. Of the areas in which colonoscopy interfaces with colorectal neoplasia, ulcerative colitis surveillance is the one area in which the actual efficacy of colonoscopy as a diagnostic tool can be challenged [97] . Several lines of evidence argue that biopsy surveillance for dysplasia is not a reliable concept. First, extensive biopsies taken with endoscopic forceps from surgical colectomy specimens containing cancer fail to detect any dysplasia in flat mucosa in up to 25% of cases [98] [99] . To have a 95% probability of detecting the highest grade of dysplasia present, 64 biopsies during a surveillance colonoscopy would be needed [100] . Yet published series indicate consistently that less than 20 biopsies typically are taken [48] [101] [102] [103] [104] [105] [106] [107] [108] [109] . Indeed, surveillance series document cases in which cancers developed with no previous documentation of dysplasia by surveillance colonoscopy [109] . Compliance with surveillance may be difficult to maintain in well patients [97] , and a number of patients lost to follow-up have presented later with cancer [101] [108] . Dysplasia itself is subject to variability in its pathological interpretation [97] , and the extent of this variability in clinical practice is uncertain. The most common abnormality detected by surveillance biopsies is low-grade dysplasia, and proper management of low-grade dysplasia remains uncertain. About 80% of patients with low-grade dysplasia do not develop cancer or high-grade dysplasia during 5 yr of follow-up [101] . Some do, however, progress to high-grade dysplasia [101] and may progress from low-grade dysplasia directly to cancer despite interval surveillance colonoscopies showing no evidence of either dysplasia or cancer [109] .

The yield of colonoscopy is higher when patients with long-standing ulcerative colitis present for an initial colonoscopic screening [103] [109] and has been calculated to be as high as 12% for cancer [110] . Such an initial examination has detected high-grade dysplasia in up to 8% of patients [103] . High-grade dysplasia has a definite association with cancer or progression to cancer [103] , and colectomy in these patients clearly appears to be beneficial. The yield of colonoscopy is lower when it is performed as a surveillance study in patients who are initially negative for dysplasia or who are entered into a surveillance program as

TABLE 13 -- Yield of Colonoscopy for Neoplasia in Miscellaneous Conditions Author Population N Patients Cancer a Adenoma a Berg [115] Uretero- 19 0 (0) 6 [32] sigmoidostomy Stewart [116] Uretero- 34 0 (0) 5 [15] sigmoidostomy Uehling [117] Uretero- 14 0 (0) 5 [29] sigmoidostomy Longo [118] Benign anorectal 102 1 [1] 9 [9] conditions Rex [119] Breast cancer age 50 116 0 (0) 21 [18]

a Number of patients with lesion (percent).

their disease duration reaches 7-10 yr [108] . Under these circumstances, surveillance colonoscopy typically is performed every 1-2 yr. In these studies [48] [101] [103] [104] [105] [107] [108] , detection of cancer occurred at a rate of one cancer per 360 surveillance colonoscopies. Another author recently calculated the yield of cancers in prospective surveillance studies as 1 per 476 examinations [108] . High-grade dysplasia and dysplasia-associated lesion or mass may also be detected and are not in all instances associated with cancer at colectomy [48] [101] [102] [103] [104] [105] [106] [107] [108] [109] . However, both lesions are clear indications for colectomy [110] [111] , and inclusion of these lesions in a category of severe neoplasia improves the apparent value of surveillance colonoscopy. Likewise, low-grade dysplasia is also predictive of cancer and has been recently suggested to be an indication for prompt colectomy [110] . Low-grade dysplasia is by far the most common neoplastic finding in ulcerative colitis [95] [107] [108] [109] . Although controversial, performance of colectomy for low-grade dysplasia would almost certainly increase the effectiveness of preventing colorectal cancer deaths by surveillance colonoscopy in ulcerative colitis. Further, treatment of low-grade dysplasia as an indication for colectomy would improve the apparent value and yield of surveillance colonoscopy for detection of significant neoplasia.

Other than the generally accepted delay of 15 yr after onset of symptoms before onset of increased cancer risk, patients with left-sided colitis are similar to pancolitis patients with regards to development of dysplasia and may have an incidence of dysplasia equal to that of patients with pancolitis [102] .

Certainly, interval surveillance colonoscopy is an appropriate option for management of long- standing ulcerative colitis. However, concerns over effectiveness [97] [110] suggest that patients with very long-standing panulcerative colitis (>20 yr) should be offered the option of prophylactic colectomy. MISCELLANEOUS CONDITIONS

Table 13 shows the yield of colonoscopy in several miscellaneous conditions. Ureterosigmoidostomy is clearly associated with an increased risk of colorectal cancer [112] [113] [114] .

362

The cancers arise at or in the vicinity of the ureterosigmoidostomy sites. Screening colonoscopy has not yet detected cancers in asymptomatic persons with ureterosigmoidostomies [115] [116] [117] (Table 13) . However, considering the age of the patients, a significant number of adenomas are detected. The adenomas have been found either at or distal to the ureterosigmoidostomy sites. The sites are easily within reach of a 60-cm flexible sigmoidoscope, so there is little rationale for colonoscopy in these patients, and interval flexible sigmoidoscopy should be adequate.

Colonoscopy in a group of patients with benign anorectal conditions but no other symptoms had a very low yield (118; Table 13) . Simple anal symptoms appear comparable to nonbleeding colonic symptoms as a poor predictor of colorectal cancer.

Screening colonoscopy in women with a personal history of breast cancer has a low yield that is identical to that obtained in average-risk women [119] . This negative result supports the conclusions of two recent meta-analyses [120] [121] that found that women with breast cancer do not have a clinically significant increased risk of colorectal cancer.

SUMMARY OF CANCER YIELD OF COLONOSCOPY BY INDICATION

Table 14 provides the number of colonoscopies needed to detect one cancer for a variety of indications, based on the papers reviewed in this report. These calculations provide a general sense of how various indications compare with regard to their predictive value for cancer.

With regard to cancers, a positive fecal occult blood test is perhaps the single best indication for colonoscopy, because the positive predictive value for cancer is high, and the cancers have relatively early Dukes stages. More data is needed to know the positive predictive value of fecal occult blood tests performed after digital rectal examinations. Colonoscopy is superior to barium enema for cancer detection in patients with positive fecal occult blood tests and is the procedure of choice. All other bleeding indications for colonoscopy (rectal bleeding, severe hemorrhage, melena with a negative upper endoscopy, iron deficiency anemia) also have a high positive predictive value for cancer. Colonoscopy is indicated in patients with persistent rectal bleeding despite a negative barium enema.

Nonbleeding indications (abdominal pain, altered bowel habit) portend a lower yield of cancers when colonoscopy is used. This is particularly true if absence of bleeding is documented by several criteria [history, normal hemoglobin, negative fecal occult blood test [49] ]. Excluding the results of one study [50] , about 10 times as many colonoscopies must be performed in this group to detect one cancer compared to bleeding indications (Table 14) . However, "persistent" nonbleeding symptoms may be associated with a higher yield of cancers [50] .

Screening colonoscopies have had intermediate yields for

TABLE 14 -- Yield of Colonoscopy by Indication Procedures to Detect One Indication Cancer Two consecutive positive FOBT; neither rehydrated 2.7 Rectal bleeding--nonemergent 8.9 Positive FOBT--nonrehydrated 9.8 Melena with negative EGD 9.9 Acute lower GI hemorrhage 11.8 Iron deficiency anemia 13 Colonic symptoms without bleeding 36 Screening Lynch syndromes 39 Positive FOBT--rehydrated 45 Screening average-risk males 60 years 64 Surveillance after cancer resection: anastomotic recurrence 74 Surveillance after cancer resection 82 metachronous cancer Colonic symptoms without bleeding 109 exclusion of reference 50 Screening positive family history; non-Lynch kindred 141 Screening average-risk persons age 50 143 Screening positive family history prospective studies only 286 Postpolypectomy surveillance 317 a Prospective U.C. surveillance 360 FOBT, fecal occult blood test; EGD, esophagogastroduodenoscopy.

a Observation of National Polyp Study recommendations could reduce this number to approximately 200.

cancers (Table 14) . Proven predictors of higher yield are Lynch syndrome, increasing age, and male gender. Although positive family history clearly portends an increased lifetime incidence of colorectal cancer, cross-sectional screening colonoscopy studies have yet to define non-Lynch kindred groups with a positive family history in whom the yield of screening colonoscopy is increased, and screening recommendations must be based on lifetime colorectal cancer risk.

Perioperative clearing colonoscopy in patients with colorectal cancer has a substantial yield for synchronous cancers. An initial colonoscopy in persons with long-standing ulcerative colitis has a high yield for cancers and high-grade dysplasia.

Postcancer resection surveillance has an intermediate yield for cancer, whereas the yield of cancer by postpolypectomy and ulcerative colitis surveillance is relatively low (Table 14) . Postpolypectomy surveillance cancer yields will probably improve substantially with widespread observation of new postpolypectomy surveillance guidelines [73] , with little expected impact on mortality. A low yield of postpolypectomy surveillance for cancer is expected after effective clearing colonoscopy. In combination with clearing colonoscopy, postpolypectomy surveillance has been proven to prevent colorectal cancer. Future research will

363 likely define subgroups of adenoma patients in whom surveillance intervals can be further increased. The yield and effectiveness of ulcerative colitis surveillance could be improved by referring low-grade dysplasia patients for colectomy.

SUMMARY OF YIELD OF COLONOSCOPY FOR ADENOMAS BY INDICATION

Relative to cancers, the yield of adenomas is fairly independent of indication (Tables 1 - 3) , as evidenced by the high yield of adenomas in screening colonoscopy studies. Increasing age and male gender clearly predict adenomas and advanced adenomas, as does Lynch syndrome history. Previous cancer resection or polypectomy is associated with a high rate of incident adenomas, despite an earlier colonoscopy that "cleared' the colon of neoplasia. Increasing age, male gender, large adenomas, and multiple adenomas at the initial examination are predictors of adenomas at surveillance, and multiple adenomas predict advanced adenomas at surveillance.

Screening flexible sigmoidoscopy appears adequate for patients with ureterosigmoidostomies. Breast cancer alone is not an indication for screening colonoscopy.

ACKNOWLEDGEMENT

The author thanks Gregory Foutch and David Lieberman for supplying data and Chris Lambert for typing the manuscript. REFERENCES

1. Winawer SJ, Zauber AG, Ho MN, et al. Prevention of colorectal cancer by colonoscopic polypectomy. N Engl J Med 1993;329:1977-81.

2. Selby JV, Friedman GD, Quesenberry CP, et al. Effect of occult blood testing on mortality from colorectal cancer. A case control study. Ann Intern Med 1993;118:1-6.

3. Newcomb PA, Norfleet RG, Storer BE, et al. Screening sigmoidoscopy and colorectal cancer mortality. J Natl Cancer Inst 1992;84:1572-75.

4. Winawer SJ, Flehinger BJ, Schottenfeld D, et al. Screening for colorectal cancer with fecal occult blood testing and sigmoidoscopy. J Natl Cancer Inst 1991;85:1311-8.

5. Gilbertsen VA, McHugh R, Schuman L, et al. The earlier detection of colorectal cancers. A preliminary report of the results of the occult blood study. Cancer 1980;45:2899-901.

6. Sontag SJ, Durczak C, Aranha GV, et al. Fecal occult blood testing for colorectal cancer in Veteran's Administration hospital. Am J Surg 1983;145:89-93.

7. Elliot MS, Levenstein JH, Wright JP. Faecal occult blood testing in the detection of colorectal cancer. Br J Surg 1984;71:785-6.

8. Fruhmorgen P, Demling L. Early detection of colorectal carcinoma with a modified guaiac test: A screening examination in 6000 humans. Acta Gastroenterol Belg 1978;41:682-7.

9. Arminski TC, McLean DW. Incidence and distribution of adenomatous polyps of the colon and rectum based on 1000 autopsy examinations. Dis Colon Rectum 1964;7:249-61.

10. Vatn MH, Stalsberg H. The prevalence of polyps of the in Oslo: An autopsy study. Cancer 1982;49:819-25.

11. Rickert RR, Auerbach O., Garfinkel L, et al. Adenomatous lesions of the large bowel; an autopsy survey. Cancer 1979;43:1847-57.

12. Ransohoff DF, Lang CA. Small adenomas detected during fecal occult blood screening for colorectal cancer: The impact of serendipity. JAMA 1990;264:76-8.

13. Ahlquist DA, Wieand HS, Moertel CH, et al. Accuracy of fecal occult blood screening for colorectal neoplasia. JAMA 1993;269:1262-67.

14. Frommer DJ, Kapparis A, Brown MK. Improved screening for colorectal cancer by immunological detection of occult blood. Br Med J 1988;296:1092-4.

15. Hardcastle JD, Chamberlain J, Sheffield J, et al. Randomized, controlled trial of faecal occult blood screening for colorectal cancer. Lancet 1989;1160-4.

16. Kronborg O, Fenger C Sondergaard O, et al. Initial mass screening for colorectal cancer with fecal occult blood test. Scand J Gastroenterol 1987;22:677-86.

17. Mandel JS, Bond JH, Church TR, et al. Reducing mortality from colorectal cancer by screening for fecal occult blood. N Engl J Med 1993;328:1365-71.

18. Elliot MS, Levenstein JH, Wright JP. Faecal occult flood testing in the detection of colorectal cancer. Br J Surg 1984;71:785-6.

19. Hobbs FDR, Cherry RC, Fielding JWL, et al. Acceptability of opportunistic screening for occult gastrointestinal blood loss. BMJ 1992;304:483-6.

20. Kewenter J, Engaras B, Haglind E, et al. Value of retesting subjects with a positive Hemoccult in screening for colorectal cancer. Br J Surg 1990;77:1349-51.

21. Gomez JA, Diehl AK. Admission stool guaiac test: Use and impact on patient management. Am J Med 1992;92:603-6.

22. Kim WR, Banerjee B. Outcome of endoscopic evaluation in patients with positive fecal occult blood test (FOBT) on digital rectal exam (DRE) without frank blood loss. Gastrointest Endosc 1993;39:293.

23. Reilly JM, Ballantyne GH, Fleming FX, et al. Evaluation of the occult blood test in screening for colorectal neoplasms. Am Surg 1990;56:119-23.

24. Rex DK, Weddle RA, Lehman GA, et al. Flexible sigmoidoscopy plus air contrast barium enema versus colonoscopy for suspected lower gastrointestinal bleeding. Gastroenterology 1990;98:855-61.

25. Baker SR, Alterman DD. False negative barium enema in patients with sigmoid cancer and coexistent diverticula. Gastrointest Radiol 1985;10:171-3.

26. Williams CB, Macrae FA, Bartram CI. A prospective study of diagnostic methods in adenoma follow-up. Endoscopy 1982;14:74-5.

27. Vellacott KD, Amar SS, Hardcastle JD. Comparison of rigid and flexible fiberoptic sigmoidoscopy with double contrast barium enema. Br J Surg 1982;69:399-400.

28. Rex DK, Lappas JC, Maglinte DDT, et al. Barium enema utilization within a defined geographic region: A survey. Gastrointest Radiol 1990;15:265-7.

29. Dent OF, Goulston KJ, Zubrzycki J, et al. Bowel symptoms in an apparently well population. Dis Colon Rectum 1986;29:243-7.

30. Forde KA, Waye JD. Is there a need to perform full colonoscopy in a middle-age person with episodic bright red blood per rectum and internal hemorrhoids? Am J Gastroenterol 1989;84:1227- 28.

31. Goulston KJ, Cook I, Dent OF. How important is rectal bleeding in the diagnosis of bowel cancer and polyps? Lancet 1986;261-5.

32. Church JM. Analysis of the colonoscopic findings in patients with rectal bleeding according to the pattern of their presenting symptoms. Dis Colon Rectum 1991;34:391-5.

33. Bat L, Pines A, Shemesh E, et al. Colonoscopy in patients aged 80 years or older and its contribution to the evaluation of rectal bleeding. Postgrad Med J 1992;68:355-8.

34. Berkowitz I, Kaplan M. Indications for colonoscopy. An analysis based on indications and diagnostic yield. S Aft Med J 1993;83:245-8.

35. Brand EJ, Sullivan BH, Sivak MV, et al. Colonoscopy in the diagnosis of unexplained rectal bleeding. Ann Surg 1980;192:111-3.

36. Graham DJ, Pritchard TJ, Bloom AD. Colonoscopy for intermittent rectal bleeding: Impact on patient management. J Surg Res 1993;54:136-9.

37. Guillem JG, Forde KA, Treat MR, et al. The impact of colonoscopy

364 on the early detection of colonic neoplasms in patients with rectal bleeding. Ann Surg 1987;206:606-11.

38. Irvine EJ O'Connor J, Frost RA, et al. Prospective comparison of double contrast barium enema plus flexible sigmoidoscopy v colonoscopy in rectal bleeding. Gut 1988;28:1188-93.

39. Swarbrick ET, Fevre DI, Hunt RH, et al. Colonoscopy for unexplained rectal bleeding. Br Med J 1978;2:1685-87.

40. Teague RH, Manning AP, Thornton JR, et al. Colonoscopy for investigation of unexplained rectal bleeding. Lancet 1978;1350-52.

41. Thoeni RF, Venbrux AC. The value of colonoscopy and double-contrast barium-enema examinations in the evaluation of the patients with subacute and chronic lower intestinal bleeding. Radiology 1983;146:603-7.

42. Caos A, Benner KG, Manier J, et al. Colonoscopy after Golytely preparation in acute rectal bleeding. J Clin Gastroenterol 1986;1:46-9.

43. Deyhle P, Blum AL, Nuesch JS. Emergency colonoscopy in the management of the acute peranal hemorrhage. Endoscopy 1974;6:229-32.

44. Jensen DM, Machicado GA. Diagnosis and treatment of severe hematochezia. Gastroenterology 1988;95:1569-74.

45. Tedesco FJ, Pickens CA, Griffin JW, et al. Role of colonoscopy in patients with unexplained melena: Analysis of 53 patients. Gastrointest Endosc 1981;27:221-3.

46. Alemayehu G, Jarnerot G. Same-day upper and lower endoscopy in patients with occult bleeding, melena, hematochezia, and/or microcytic anemia. Scand J Gastroenterol 1993;28:667- 72.

47. Rockey DC, Cello JP. Evaluation of the in patients with iron-deficiency anemia. N Engl J Med 1993;329:1691-5.

48. Brenna E, Skreden K, Waldum HL, et al. The benefit of colonoscopy. Scand J Gastroenterol 1990;25:81-8.

49. Rex DK, Lehman GA, Mark D, et al. Colonoscopy vs. sigmoidoscopy/ACBE for colonic evaluation of symptomatic patients without evidence of bleeding. Gastrointest Endosc 1993;39:302 (abstract).

50. Neugut AI, Garbowski GC, Waye JD, et al. Diagnostic yield of colorectal neoplasia with the use of colonoscopy for abdominal pain, change in bowel habits, and rectal bleeding. Am J Gastroenterol 1993;88:1179-84.

51. Rex DK. Determining indications for primary colonoscopy: How can we predict the need for polypectomy? Am J Gastroenterol 1993;88:1154-6.

52. Carlsson G, Petrelli NJ, Nava HR, et al. The value of colonoscopic surveillance after curative resection for colorectal cancer or synchronous adenomatous polyps. Arch Surg 1987;122:1261-3.

53. Kronborg O, Hage E, Deichgraeber E. The remaining colon after radical surgery for colorectal cancer: The first three years of a prospective study. Dis Colon Rectum 1983;26:172-6.

54. Brady PG, Straker RJ, Goldschmid S. Surveillance colonoscopy after resection for colon carcinoma. South Med J 190;83:765-8.

55. Hall C, Griffin CH, Dykes PW, et al. Hemoccult does not reduce the need for colonoscopy in surveillance after curative resection for colorectal cancer. Gut 1993;34:227-9.

56. Jahn H, Joergensen OD, Kronborg O, et al. Can hemoccult-II replace colonoscopy in surveillance after radical surgery for colorectal cancer and after polypectomy? Dis Colon Rectum 1992;35:253-6.

57. Juhl G, Larson GM, Mullins R, et al. Six-year results of annual colonoscopy after resection of colorectal cancer. World J Surg 1990;14:255-61.

58. Larson GM, Bond SJ, Shallcross C, et al. Colonoscopy after curative resection of colorectal cancer. Arch Surg 1986;121:535-40.

59. McFarland RJ, Becciolini C, Lallemand RC. The value of colonoscopic surveillance following a diagnosis of colorectal cancer or adenomatous polyp. Eur J Surg Oncol 1991;17:514-8.

60. Nava HR, Pagana TJ. Postoperative surveillance of colorectal carcinoma. Cancer 1982;19:1043-7.

61. Unger SW, Wanebo HJ. Colonoscopy: An essential monitoring technique after resection of colorectal cancerAm J Surg 1983;145:71-6.

62. Weber CA, Deveney KE, Pellegrini CA, et al. Routine colonoscopy in the management of colorectal carcinoma. Am J Surg 1986;152:87-92.

63. Malcolm AW, Perencevich NP, Olson RM, et al. Analysis of recurrence patterns following curative resections for carcinoma of the colon and rectum. Surg Gynecol Obstet 1981;152:131-6.

64. Ballantyne GA, Modlin IM. Editorial: Postoperative follow-up for colorectal cancer: Who are we kidding? J Clin Gastroenterol 1988;10:359-64.

65. Schrock TR. Colonoscopy for coloectal cancer: Too much, too little, just right. Gastrointest Endosc 1993;39:848-50.

66. Jorgensen OD, Kronborg O, Fenger C. The Funen adenoma follow-up study: Incidence and death from colorectal carcinoma in an adenoma surveillance program. Scand J Gastroenterol 1993;28:869-74.

67. Matek W, Guggenmoos-Holsmann I, Demling L. Follow-up patients with colorectal adenomas. Endoscopy 1985;17:175-81.

68. Nava H, Carlsson G, Petrelli NJ, et al. Follow-up colonoscopy in patients with colorectal adenomatous polyps. Dis Colon Rectum 1987;30:465-8.

69. Neugut AI, Johnsen CM, Forde KA, et al. Recurrence rates for colorectal polyps. Cancer 1985;55:1586-9.

70. Waye JD, Braunfeld S. Surveillance intervals after colonoscopic polypectomy. Endoscopy 1982;14:79-81.

71. Winawer SJ, Zauber AG, O'Brien MJ, et al. Randomized comparison of surveillance intervals after colonoscopic removal of newly diagnosed adenomatous polyps. N Engl J Med 1993;328:901- 6.

72. Woolfson IK, Eckholdt GJ, Wetzel CR, et al. Usefulness of performing colonoscopy one year after endoscopic polypectomy. Dis Colon Rectum 1990;33:389-93.

73. Bond JH. Polyp guideline: Diagnosis, treatment and surveillance for patients with nonfamilial colorectal polyps. Ann Intern Med 1993;119:836-43.

74. St. John DJB, McDermott FT, Hopper JL, et al. Cancer risk in relatives of patients with common colorectal cancer. Ann Intern Med 1993;118:785-90.

75. Lovett E. Family studies in cancer of the colon and rectum. Br J Surg 1976;63:13-8.

76. Macklin MT. Inheritance of cancer of the and large intestine in man. J Natl Cancer Inst 1960;24:551-71.

77. Woolf CM. A genetic study of carcinoma of the large intestine. Am J Hum Genet 1958;10:42-7.

78. Baker JW, Gathright B, Timmcke AE, et al. Colonoscopic screening of asymptomatic patients with a family history of colon cancer. Dis Colon Rectum 1990;33:926-30.

79. Grossman S, Milos ML. Colonoscopic screening of persons with suspected risk factors for colon cancer. Gastroenterology 1988;94:395-400.

80. Gryska PV, Cohen AM. Screening asymptomatic patients at high risk for colon cancer with full colonoscopy. Dis Colon Rectum 1987;30:18-20.

81. Guillem JG, Forde KA, Treat MR, et al. Colonoscopic screening for neoplasms in asymptomatic first-degree relatives of colon cancer patients. Dis Colon Rectum 1992;35:523-9.

82. Luchtefeld MA, Syverson D, Solfelt M, et al. Is colonoscopic screening appropriate in asymptomatic patients with family history of colon cancer? Dis Colon Rectum 1991;34:763-8.

83. McConnell JC, Nizin JS, Slade MS. Colonoscopy in patients with a primary family history of colon cancer. Dis Colon Rectum 1990;33:105-7.

84. Meagher AP, Stuart M. Colonoscopy in patients with a family history of colorectal cancer. Dis Colon Rectum 1992;35:315-21.

85. Orrom WJ, Brzezinski WS, Wiens EW. Heredity and colorectal cancer: A prospective, community-based, endoscopic study. Dis Colon Rectum 1990;33:490-3.

86. Rex DK, Lehman GA, Ulbright TM, et al. Colonic neoplasia in asymptomatic persons with negative fecal occult flood tests: Influence of age, gender, and family history. Am J Gastroenterol 1993;88:825-31.

87. Rozen P, Ron E. A cost analysis of screening methodology for family members of colorectal cancer patients. Am J Gastroenterol 1989;84:1548-51.

88. Cannon-Albright LA, Skolnick MH, Bishop DT, Lee RG, Burt RH. Common inheritance of susceptibility to colonic adenomatous polyps and associated colorectal cancers. N Engl J Med 1988;319:533-7.

89. Lanspa SJ, Lynch HT, Smyrk TC, et al. Colorectal adenomas in the Lynch syndromes. Gastroenterology 1990;98:1117-22.

90. Love RR, Morrissey JF. Colonoscopy in asymptomatic individuals with a family history of colorectal cancer. Arch Intern Med 1984;144:2209-11.

91. Vasen HFA, Jafer FCA, Menko FH. Screening for hereditary non-polyposis

365 colorectal cancer: A study of 22 kindreds in the Netherlands. Am J Med 1989;86:278-81.

92. Lanspa SJ, Jenkins JX, Cavalieri J, et al. Surveillance in Lynch Syndrome: How aggressive? Am J Gastroenterol 1994;89:1978-80.

93. Foutch PG, Mai H, Pardy K, et al. Flexible sigmoidoscopy may be ineffective for secondary prevention of colorectal cancer in asymptomatic, average-risk men. Dig Dis Sci 1991;36:924-8.

94. Johnson DA, Gurney MS, Volpe RJ, et al. A prospective study of the prevalence of colonic neoplasms in asymptomatic patients with an age-related risk. Am J Gastroenterol 1990;85:969-74.

95. Lieberman DA, Smith FW. Screening for colon malignancy with colonoscopy. Am J Gastroenterol 1991;86:946-51.

96. Chu KC, Tarone RE, Chow W-H, Hankey BF, Ries LAG. Temproal patterns in colorectal cancer: Incidence, survival and mortality from 1950 through 1990. J Natl Cancer Inst 1994;86:997- 1006.

97. Collins RH, Feldman M, Fordtran JS. Sounding board: Colon cancer, dysplasia, and surveillance in patients with ulcerative colitis. N Engl J Med 1987;316:1654-8.

98. Ransohoff DF, Riddell RH, Levin B. Ulcerative colitis and colonic cancer: Problems in assessing the diagnostic usefulness of mucosal dysplasia. Dis Colon Rectum 1985;28:383-8.

99. Taylor BA, Pemberton JH, Carpenter HA, et al. Dysplasia in chronic ulcerative colitis: Implications for colonoscopic surveillance. Dis Colon Rectum 1992;35:950-6.

100. Rubin CE, Haggitt RC, Burmer GC, et al. DNA aneuploidy in colonic biopsies predicts future development of dysplasia in ulcerative colitis. Gastroenterology 1992;103:1611-20.

101. Lennard-Jones JE, Melville DM, Morson BC, et al. Precancer and cancer in extensive ulcerative colitis: Findings among 401 patients over 22 years. Gut 1990;31:800-6.

102. Nugent FW, Haggitt RC, Gilpin PA. Cancer surveillance in ulcerative colitis. Gastroenterology 1991;100:1241-8.

103. Rosenstock E, Farmer RG, Petras R, et al. Surveillance for colonic carcinoma in ulcerative colitis. Gastroenterology 1985;89:1342-6.

104. Rutegard J, hsgren L, Stenling R, et al. Ulcerative colitis: Cancer surveillance in an unselected population. Scand J Gastroenterol 1988;23:139-45.

105. Lofberg R, Brostrom O, Karlen P, et al. Colonoscopic surveillance in long-standing total ulcerative colitis: A 15-year follow-up study. Gastroenterology 1990;99:1021-31.

106. Lashner BA, Kane SV, Hanauer SB. Colon cancer surveillance in chronic ulcerative colitis: Historical cohort study. Am J Gastroenterol 1990;85:1083-7.

107. Leidenius M, Kellokumpu I, Husa A, Riihela M, Sipponen P. Dysplasia and carcinoma in longstanding ulcerative colitis: An endoscopic and histological surveillance programme. Gut 1991;32:1521-5.

108. Lynch DAF, Lobo AJ, Sobala GM, et al. Failure of colonoscopic surveillance in ulcerative colitis. Gut 1993;34:1075-80.

109. Woolrich AJ, Dasilva MD, Korelitz BI. Surveillance in the routine management of ulcerative colitis: The predictive value of low-grade dysplasia. Gastroenterology 1992;103:431-8.

110. Bernstein CN, Shanahan F, Weinstein WM. Are we telling the truth about surveillance colonoscopy in ulcerative colitis? Lancet 1994;343:71-4.

111. Blackstone MO, Riddell R, Rogers BHG, et al. Dysplasia-associated lesion or mass (DALM) detected by colonoscopy in long-standing ulcerative colitis: An indication for colectomy. Gastroenterology 1981;80:366-74.

112. Urdaneta LF, Duffell D, Creevy CD, et al. Late development of primary carcinoma of the colon following ureterosigmoidostomy: A report of three cases and literature review. Ann Surg 1966;164:503-13.

113. Parsons CD, Thomas MD, Garrett RA. Colonic adenocarcinoma: A delayed complication of ureterosigmoidostomy. Urol 1977;118:31-4.

114. Eraklis AJ, Folkman MJ. Adenocarcinoma at the site of the ureterosigmoidostomies for exstrophy of the bladder. J Pediatr Surg 1978;13:73-4.

115. Berg NO, Fredlund P, Mansson W, et al. Surveillance colonoscopy and biopsy in patients with ureterosigmoidostomy. Endoscopy 1987;19:60-3.

116. Stewart M, Macrae FA, Williams CB. Neoplasia and ureterosigmoidostomy: A colonoscopy survey. Br J Surg 1982;69:414-6.

117. Uehling DT, Starling JR, Gilchrist KW. Surveillance colonoscopy after ureterosigmoidostomy. J Urol 1982;127:34-6.

118. Longo WE, Dean PA, Virgo KS, et al. Colonoscopy in patients with benign anorectal disease. Dis Colon Rectum 1993;36:368-71.

119. Rex DK, Sledge GS, Harper PA, et al. Colonic adenomas in asymptomatic women with a history of breast cancer. Am J Gastroenterol 1993;88:2009-14.

120. Eisen GM, Sandler RS. Are women with breast cancer more likely to develop colorectal cancer? Critical review and meta-analysis. J Clin Gastroenterol 1994;19:57-63.

121. Schoen RE, Weissfeld JL, Kuller LH. Are women with breast, endometrial or ovarian cancer at increased risk for colorectal cancer? Am J Gastroenterol 1994;89:835-42. Copyright © 2010 Elsevier Inc. All rights reserved. - www.mdconsult.com )

Bookmark URL: /das/journal/view/0/N/303645?ja=1103&PAGE=1.html&issn=0002- 9270&source=MI

The role of endoscopy in dyspepsia Gastrointestinal Endoscopy - Volume 66, Issue 6 (December 2007)

Table 2 -- Alarm features for patients with dyspepsia Age >50 y, with new onset symptoms Family history of upper-GI malignancy Unintended weight loss GI bleeding or iron deficiency anemia Progressive Odynophagia Persistent Palpable mass or lymphadenopathy

High-risk features confer an increased risk of significant GI pathology in patients aged less than 50 years with dyspepsia Gastrointestinal Endoscopy - Volume 67, Issue 6 (May 2008) - Copyright © 2008 American Society for Gastrointestinal Endoscopy - About This Journal DOI: 10.1016/j.gie.2007.12.038

DOI 10.1016/j.gie.2007.12.038

To the Editor:

We welcomed the American Society for Gastrointestinal Endoscopy (ASGE) guidelines concerning the role of endoscopy in the management of dyspepsia.[1] The guidelines indicate that EGD should be performed in all patients aged ≥50 years with new-onset dyspepsia, which most clinicians would agree would be associated with a reasonable diagnostic yield. However, the guidelines also advocate EGD as an initial investigation in patients aged <50 years if “high-risk” features coexist. The merit of EGD in this setting is more contentious. We sought to determine whether 4 ASGE-defined high-risk features (weight loss, dysphagia, vomiting, and anemia) conferred increased risk of significant pathology in patients aged <50 years.

Records from all EGDs performed over 23 years at our unit were reviewed. We identified 1119 patients aged <50 years with dyspepsia and high-risk features, 9522 with dyspepsia without high-risk features, and 1915 patients aged ≥50 with dyspepsia and high-risk features. In these 3 groups we calculated the prevalence of cancer and significant benign conditions (severe , esophageal strictures, esophageal , and peptic ulcers >5 mm).

The prevalence of cancer was significantly higher in patients aged <50 years in the presence of high-risk features (1.4%) compared with those without (0.4%), odds ratio (OR) 3.54, 95% CI, 1.38-9.09, P < .0001. Significant benign lesions were also significantly more common in patients aged <50 years with high-risk features (10.4%) compared to those without (5.6%), OR 1.85, 95% CI, 1.43-2.39, P < .0001. Duodenal ulcer was the most common benign lesion. In patients aged ≥50 years with dyspepsia and high-risk features, the prevalence of cancer (6.9%) was significantly higher than in patients aged <50 years, with (OR 3.46, 95% CI, 2.34- 5.10, P < .0001) or without (OR 18.45, 95% CI, 9.04-37.69, P < .0001) high-risk features.

Our findings support current ASGE guidelines regarding the use of EGD as an initial investigation of dyspepsia in patients aged <50 years in the presence of high-risk features.