|
|
| |||||||||||||||||||||||||||||||||||||||||||||||
Background After patients have undergone colonoscopic polypectomy, it is uncertain whether colonoscopic examination or a barium enema is the better method of surveillance.
Methods As part of the National Polyp Study, we offered colonoscopic examination and double-contrast barium enema for surveillance to patients with newly diagnosed adenomatous polyps. Although barium enema was performed first, the endoscopist did not know the results.
Results A total of 973 patients underwent one or more colonoscopic examinations for surveillance. In the case of 580 of these patients, we performed 862 paired colonoscopic examinations and barium-enema examinations that met the requirements of the protocol. The findings on barium enema were positive in 222 (26 percent) of the paired examinations, including 94 of the 242 colonoscopic examinations in which one or more adenomas were detected (rate of detection of adenomas, 39 percent; 95 percent confidence interval, 33 to 45 percent). The proportion of examinations in which adenomatous polyps were detected by barium enema was significantly related to the size of the adenomas (P=0.009); the rate was 32 percent for colonoscopic examinations in which the largest adenomas detected were 0.5 cm or less, 53 percent for those in which the largest adenomas detected were 0.6 to 1.0 cm, and 48 percent for those in which the largest adenomas detected exceeded 1.0 cm. Among the 139 paired examinations with positive results on barium enema and negative results on colonoscopic examination in the same location, 19 additional polyps, 12 of which were adenomas, were detected on colonoscopic reexamination.
Conclusions In patients who have undergone colonoscopic polypectomy, colonoscopic examination is a more effective method of surveillance than double-contrast barium enema.
Methods
Patients
The National Polyp Study was a randomized, controlled trial designed to assess surveillance strategies in patients after colonoscopic removal of newly diagnosed adenomas. All patients referred for initial colonoscopy or polypectomy from November 1980 to February 1990 at one of seven participating clinical centers were identified. Those with newly diagnosed adenomas who met the eligibility criteria and consented to participate were randomly assigned to undergo a colonoscopic examination three years after the initial colonoscopic examination or at both one and three years. Patients in both groups were offered a surveillance examination at six years. At each follow-up examination, the patients were offered both double-contrast barium enema and colonoscopic examination. The design of the study and follow-up results based on colonoscopic findings have been described previously.5,6 All patients provided written informed consent for their participation in the study, and the study design was approved by the institutional review board at each participating center.
Design of the Study
The study design included a prospective, blinded comparison of colonoscopy and double-contrast barium enema in detecting polyps at follow-up.5,7 Only study endoscopists and radiologists performed the examinations and interpreted the results. Each center designated colon-cleansing regimens that were acceptable to the endoscopic and radiologic investigators. The regimens consisted of cathartic agents and hydration and, beginning in 1985, polyethylene glycol for colonoscopy. The use of enemas as part of the cleansing regimen was avoided before the barium enema because the products used interfered with the mucosal coating. Barium preparations were either HD-85 (Lafayette, Lafayette, Ind.) or Liquid Polibar (EZ-EM, Westbury, N.Y.). The standard barium-enema technique involved the acquisition of multiple (120 kVp) films, including horizontal-beam films. Radiologists had access to all prior films. The colonoscopic examination was performed approximately two weeks after the barium enema.
Findings on barium enema and colonoscopic examination were reported according to the location: rectum, sigmoid colon, descending colon, splenic flexure, transverse colon, hepatic flexure, ascending colon, or cecum. The size, location, and shape of each polyp were recorded. For each anatomical segment, findings of spasm, redundancy, stool, poor mucosal coating, or diverticulosis, and the level of confidence in each such finding, were reported. For barium enema, a high level of confidence indicated that the radiologist was unlikely to miss lesions that were larger than 1.0 cm, and a low level indicated that the radiologist could easily overlook a lesion that was larger than 1.0 cm. For colonoscopic examination, a high level of confidence indicated that the endoscopist had reached the cecum and could visualize the entire colon. The adequacy of the colonoscopic preparation and the completeness of the colonoscopic examination were also recorded.
The endoscopist inserted the colonoscope to the cecum and then withdrew the colonoscope to the next most distal segment of the colon. Polyps detected by colonoscopic examination were recorded for each anatomical segment by endoscopists who had no knowledge of the findings on barium enema (blinded colonoscopic examination). After the endoscopist reported observations for a given segment, the study coordinator informed the endoscopist of the results of barium enema for that segment. If there was a finding on barium enema that was not seen on colonoscopic examination, the endoscopist reexamined the segment several times with the patient in different positions until satisfied that the area was clear or a lesion had been found (unblinded colonoscopic examination). With the use of this approach, the colonoscope did not need to be reinserted after the completion of the examination and the rate of polyps missed by colonoscopic examination could be determined.
All polyps detected by colonoscopic examination were removed, measured, and then classified histologically by the pathology review team according to National Polyp Study criteria as tubular adenomas or villous adenomas class A (proportion of villous component, 1 to 25 percent), B (26 to 75 percent), C (76 to 99 percent), or D (100 percent) or as nonadenomas (hyperplastic or other types, mainly normal mucosal tags).13 After the histologic evaluation, the findings of the colonoscopic examination were classified on the basis of the presence or absence of adenomas and the size of the largest adenoma as showing no polyps, nonadenomatous polyps alone, adenomas that were no larger than 0.5 cm, adenomas that were 0.6 to 1.0 cm, or adenomas that were larger than 1.0 cm. The size of an adenoma was estimated by the endoscopist with use of open forceps during colonoscopic examination.
Statistical Analysis
The primary comparison of colonoscopic examination and double-contrast barium enema was based on the classification of the findings on barium enema as positive (any polyps) or negative (no polyps) and the classification of the colonoscopic findings according to the presence and type (size and histologic characteristics) of polyps. The colonoscopic findings reported by endoscopists before the barium-enema findings were revealed were used as the reference measure for the barium-enema findings.
A secondary analysis assessed whether the same polyps detected by colonoscopic examination were also detected by barium enema. Polyps detected by both techniques within the same or an adjoining segment and whose sizes matched within 5 mm were considered a match. The matched polyps were true positives for barium enema relative to colonoscopic examination; polyps detected by colonoscopic examination but not by barium enema were considered to have been missed by barium enema or to represent a false negative result.
SAS statistical software (version 6.12, SAS Institute, Cary, N.C.) was used for all calculations. The chi-square statistic was used to compare differences in percentages; the Mantel–Haenszel chi-square test was used to compare differences in percentages with adjustment for covariates. The regression coefficient in a logistic model was used to assess whether there was a linear trend in the rate of detection with barium enema according to the size of the adenoma. All tests were two-sided, and a P value of 0.05 or less was considered to indicate statistical significance.14
Results
Patients
Among the 1418 patients who underwent randomization,6 973 underwent one or more surveillance colonoscopic examinations, representing 80 percent of those alive and eligible for surveillance colonoscopic examination. These patients underwent 1762 surveillance colonoscopic examinations and 949 barium-enema examinations, of which 881 were paired with colonoscopic examinations. Nineteen paired examinations were excluded: the cecum was not reached in 12, the time between examinations exceeded six months in 6, and both were true in 1 case. The remaining 862 paired examinations meeting the protocol requirements were performed in 580 patients. The characteristics of these 580 patients at enrollment were similar to those of the 393 patients who did not have paired examinations (Table 1).
|
The average interval between the paired examinations was 16 days (median, 10; range, 2 to 164; interquartile range, 2 to 20). There were 125 pairs (15 percent) in which the interval between examinations exceeded 30 days. Among the 862 paired examinations, 337 were performed as part of the one-year surveillance (39 percent), 361 as part of the three-year surveillance (42 percent), and 119 as part of the six-year surveillance (14 percent); 45 were performed at other times (5 percent). All were performed between November 1981 and November 1990. There were no major complications after any of the paired examinations. Findings that were reported with a high level of confidence were more frequent for colonoscopic examination than for barium enema. The confidence level for barium enema was lowest with respect to findings in the sigmoid colon because of the presence of redundancy and diverticulosis and in the ascending colon and cecum because of the presence of fecal residue and poor mucosal coating (Table 2).
|
Polyps were detected in 392 of the 862 colonoscopic examinations (45 percent); adenomas were detected in 242 colonoscopic examinations (28 percent), and adenomas that were more than 1.0 cm were identified in 23 colonoscopic examinations (3 percent) (Table 3). Findings on barium enema were positive in 222 of the 862 paired examinations (26 percent) and in 139 of the 392 colonoscopic examinations in which one or more polyps were detected (rate of detection of polyps, 35 percent; 95 percent confidence interval, 31 to 40 percent). In the 470 pairs of examinations in which colonoscopic examination did not detect polyps, the findings on barium enema were negative in 387 cases (rate of agreement for negative findings, 82 percent; 95 percent confidence interval, 79 to 86 percent). The results of barium enema and colonoscopic examination were concordant in 526 (61 percent) of the examinations: 139 were concordant for polyps, and 387 were concordant for the absence of polyps.
|
0.5 cm) were detected, the findings on paired barium enema were positive in 49 examinations (rate of detection, 32 percent; 95 percent confidence interval, 25 to 39 percent). Of the 64 colonoscopic examinations in which the largest adenoma was 0.6 to 1.0 cm, the findings on paired barium enema were positive in 34 examinations (rate of detection, 53 percent; 95 percent confidence interval, 40 to 66 percent). Adenomas that were greater than 1.0 cm were detected in 23 colonoscopic examinations (3 percent), and the results of the corresponding barium enema were positive in 11 examinations (rate of detection, 48 percent; 95 percent confidence interval, 24 to 67 percent). The rate of detection for barium enema was significantly related to the size of the largest adenoma detected by colonoscopic examination (P=0.009). In the paired examinations, two malignant polyps were detected by surveillance colonoscopic examination and were included in the group of adenomas that were larger than 1.0 cm.1 One of these cancers was detected by barium enema, and one was not. Both these cancers were 1.5-cm cecal cancers. Three additional malignant polyps were detected by surveillance colonoscopic examination, but in these cases paired examinations were not performed.1
Characteristics of the Polyps Identified
The polyps detected by colonoscopic examination were matched according to size and location with polyps detected by barium enema and were the basis for the secondary analysis of the rate of detection associated with barium enema (Table 4). The 862 blinded colonoscopic examinations that were matched with barium-enema examinations detected 791 polyps, of which 375 were adenomas (47 percent) and 416 were nonadenomas (53 percent). Barium enema detected 160 of the 791 polyps detected by colonoscopic examination (20 percent). The rate of detection of adenomas by barium enema was significantly related to the size of the adenomas (P<0.001) and was lower for the 270 adenomas that were 0.5 cm or less (21 percent; 95 percent confidence interval, 17 to 26 percent) than for those that were 0.6 to 1.0 cm (42 percent; 95 percent confidence interval, 31 to 54 percent) and those that were more than 1.0 cm (46 percent; 95 percent confidence interval, 26 to 67 percent). The location of the adenomas also significantly affected the rate of detection by barium enema: the rate was higher for polyps on the left side of the colon, even after adjustment for differences in the size of the adenomas between the right and left sides of the colon (P=0.01). The rate of detection of any hyperplastic polyps by barium enema was similar to the rate of detection of adenomas that were 0.5 cm or smaller (P=0.25).
|
An unblinded colonoscopic examination was done during 139 examinations in which the results of blinded colonoscopic examination of a specific location differed from those for barium enema. Nineteen additional polyps were detected by unblinded colonoscopic examination, of which 12 were adenomas (Table 5). Therefore, the rate of missed adenomas with blinded colonoscopic examination was 20 percent, since 12 were detected by unblinded colonoscopic examination in addition to the 47 adenomas already detected by blinded colonoscopic examination. The rate of missed small adenomas was 26 percent (11 of 43) and of adenomas of 0.6 to 1.0 cm, 6 percent (1 of 16). No adenomas that were greater than 1.0 cm were missed.
|
Since its introduction in the early 1970s, colonoscopy has become an important diagnostic and therapeutic tool for the examination of the colon.15 Before colonoscopy became available, barium enema was the primary means of detection of polyps, and their removal required surgical colotomy.15 The results of studies have supported the transition from the use of barium enema to the use of colonoscopy, suggesting a greater rate of detection of colonic neoplasia with colonoscopic examination and highlighting its usefulness for biopsy and polypectomy.12,16 Our study design permitted a direct blinded comparison of colonoscopic examination with barium enema without interfering with complete colonoscopy in each patient.7 We assessed whether colonoscopic examination alone, barium enema alone, or both were needed for surveillance.
Most early studies comparing colonoscopic examination and barium enema were retrospective and did not adjust for important variables.8,9,10,11,12 These studies usually did not provide descriptions of the way in which polyps identified by both barium enema and colonoscopic examination were matched, and they often did not report the histologic appearance or the size of the polyps. Areas of concern in many studies have included the unequal expertise of the examiners, the use of disparate cleansing regimens for colonoscopic examination and barium enema, and the lack of blinding during examinations. In some studies, single-contrast and double-contrast examinations of the colon were combined. We used only double-contrast examinations, the allowable cleansing regimens were agreed on at a consensus meeting before the initiation of the study, all examinations were performed by endoscopic and radiologic investigators at each center who had similar levels of experience, and the examinations were carried out in a blinded fashion. During the study, we held regular meetings to ensure continued uniformity of practice and adherence to the protocol.
The prospective nature of the study allowed uniform documentation of the size, location, and pathological characteristics of the polyps.5 The process of matching polyps detected by both barium enema and colonoscopic examination required accurate mapping of polyps by radiologists and endoscopists at each examination. Several factors were considered in matching polyps. Endoscopic assessment of location may vary, especially in redundant portions of the colon. Telescoping of the bowel can increase the difficulty of pinpointing the tip of the endoscope. For this reason, determining whether a polyp seen on a barium enema is precisely the same as that seen on colonoscopic examination may not be possible. An inaccuracy could result in a false positive or a false negative result for barium enema or a false negative result for colonoscopic examination. Also, the results of direct measurement of a polyp on an x-ray film can be at variance with estimates of size seen endoscopically or direct measurements of a removed polyp. We used the size on endoscopy as the reference size, since more polyps were detected by endoscopy than by barium enema.
We found that colonoscopic examination detected many more polyps than barium enema; about half of these polyps were adenomas, and the remainder were primarily normal mucosal tags, with some hyperplastic polyps.13 The rate of detection with barium enema was related to the size of the adenomas.
A major question is whether the polyps that were not detected by barium enema are important in the long-term outcome of the patients. On the basis of prior observations in the National Polyp Study, it is clear that few clinically significant abnormalities are found after the initial colonoscopic examination.6 Only 3 percent of the surveillance colonoscopic examinations detected adenomas that were greater than 1.0 cm. This finding is understandable, given current knowledge of the slow rate of progression of adenomas to carcinomas. The best estimate of the average time that it takes for a new polyp to grow and transform into cancer is 10 to 20 years.17,18
Is combining the two procedures beneficial? Performing both would result in additional costs, inconvenience, and risk. With the use of both procedures we found only 19 additional polyps in 18 patients — 12 of which, in 11 patients, were adenomas. We did not, however, conduct a formal cost-effectiveness study.
Colonoscopy was used as the reference measure with the knowledge that it is not perfect and does miss polyps. In our study, the rate of missed adenomas was 20 percent for colonoscopic examination, and all missed polyps were 1.0 cm or smaller. This rate is similar to the rates reported in other studies. A study of back-to-back colonoscopic examinations by Hixson et al. found that 15 percent of adenomatous polyps were missed, and all were less than 1.0 cm.19 Rex et al. reported that back-to-back colonoscopic examinations missed 25 percent of adenomas that were less than 1.0 cm in size, 6 percent of adenomas that were at least 1.0 cm, and 24 percent of adenomas overall.20 In addition, the colonoscope may not reach the cecum in all cases; the rate of complete examination ranges from 80 to 95 percent.21,22,23 The rate of complete colonoscopic examination in our study (in which patients had already undergone a complete colonoscopic examination one or more years earlier) was 99 percent (868 of 881 examinations). Using colonoscopic examination as the reference procedure produced an unavoidable bias in favor of this procedure. We minimized this bias by using equivalent cleansing regimens for each type of examination, having investigators perform all the examinations, and using a blinded comparison agreed on before the study was initiated.7
Colonoscopic examination has become the preferred way of examining the colon for both the detection and the removal of polyps, replacing diagnostic barium enema as a means of surveillance. Our study supports this evolving clinical practice. The low rate of detection of large adenomas with barium enema is a drawback to the use of this radiologic technique as the primary surveillance tool. A double-contrast barium enema can be performed in cases in which the colonoscope does not reach the cecum.
Supported in part by a grant from the National Cancer Institute (CA-26852) and by the Tavel-Reznik Fund.
Source Information
From the Departments of Medicine (S.J.W.), Epidemiology and Biostatistics (A.G.Z.), and Pathology (S.S.S.), Memorial Sloan-Kettering Cancer Center, New York; the Department of Radiology, Medical College of Wisconsin, and Froedtert Memorial Lutheran Hospital, Milwaukee (E.T.S.); the Departments of Medicine (J.H.B.) and Radiology (H.A.), Minneapolis Veterans Affairs Medical Center, Minneapolis; the Department of Medicine, Mount Sinai Hospital, New York (J.D.W.); the Department of Radiology, Massachusetts General Hospital, Boston (D.H.); the Department of Radiology, Cedars–Sinai Medical Center, Los Angeles (J.A.H.); the Department of Medicine, Valley Presbyterian Hospital, Van Nuys, Calif. (M.S.); and the Mallory Institute of Pathology, Boston Medical Center, Boston (M.J.O., L.S.G.). Other authors were Walter J. Hogan, M.D. (Department of Medicine, Medical College of Wisconsin, and Froedtert Memorial Lutheran Hospital, Milwaukee); Mansho Khilnani, M.D. (Department of Radiology, Mount Sinai Hospital, New York; deceased); Frederick W. Ackroyd, M.D. (Department of Surgery, Massachusetts General Hospital, Boston); Joel F. Panish, M.D. (Department of Medicine, Cedars–Sinai Medical Center, Los Angeles); Larry Kussin, M.D. (Department of Radiology, Valley Presbyterian Hospital, Van Nuys, Calif.); and Martin Edelman, M.D. (Department of Radiology, Memorial Sloan-Kettering Cancer Center, New York).
Address reprint requests to Dr. Winawer at the Gastroenterology and Nutrition Service, Department of Medicine, Memorial Sloan-Kettering Cancer Center, 1275 York Ave., New York, NY 10021, or at winawers{at}mskcc.org.
References
In addition to the authors, the following persons were members of the National Polyp Study Work Group: Memorial Sloan-Kettering Cancer Center and Mount Sinai Hospital, New York — R. Kurtz, M. Shike, H. Gerdes, C.J. Lightdale, C. Miller, L. Hornsby-Lewis, A. Szporn, N. Harpaz, M. Fleisher, M.N. Ho, B. Diaz, J. Lapidus, R.A. Paden, M. Mandelman, H. Nazario, H. Colon, P. Kadvan; Minneapolis Veterans Affairs Medical Center, Minneapolis — S. Ewing, T. Dobson; Froedtert Memorial Lutheran Hospital, Milwaukee — J. Helm, R. Komorowski, F. Loo, P. Nemeth, E. McLaughlin, J. Geenan, R.P. Venu, G.K. Johnson, P. Miller (deceased), N. DeBoer; Massachusetts General Hospital, Boston — S. Hedberg (deceased), P. Shellito, G.R. Dickersin, N. Horton; Cedars–Sinai Medical Center, Los Angeles — J. Sherman, S. Geller, M. Kojimoto; Valley Presbyterian Hospital, Van Nuys, Calif. — M. Auslander, D. Kasimian, C. Scoggins; and Pathology Review Center, Boston — C. Magrath.
| |||||||||||||||||||||||||||||||||||||||||||||||
Related Letters:
Comparison of Colonoscopy and Double-Contrast Barium Enema
Glick S. N., Fibus T., Fister M. R., Balfe D. M., Anderson J. C., Birk J. W., Shaw R. D., Zauber A. G., Winawer S. J., Stewart E. T.
Extract |
Full Text
N Engl J Med 2000;
343:1728-1730, Dec 7, 2000.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
