- Email: [email protected]
Increased risk of colorectal neoplasia in patients with primary sclerosing cholangitis and ulcerative colitis: A meta-analysis
Increased risk of colorectal neoplasia in patients with primary sclerosing cholangitis and ulcerative colitis: a meta-analysis Roy M. Soetikno, MD, Otto S. Lin, MD, Paul A. Heidenreich, MD, Harvey S. Young, MD, Michael O. Blackstone, MD Palo Alto, California, and Chicago, Illinois
Background: Published data on the risk of colorectal neoplasia in patients with ulcerative colitis with and without primary sclerosing cholangitis are conflicting. A meta-analysis was performed to synthesize available publications and to compare the risk of colorectal neoplasia in patients with ulcerative colitis with and without primary sclerosing cholangitis. Methods: By using MEDLINE and manual search methods, studies were identified that compared the risk of colorectal neoplasia (dysplasia and carcinoma) in patients with ulcerative colitis with and without primary sclerosing cholangitis. In addition, citations were reviewed in relevant articles and proceedings from gastroenterology meetings, and investigators were contacted when data were incomplete. The summary odds ratio (OR) was then calculated for the risk for patients with ulcerative colitis and primary sclerosing cholangitis of having colorectal neoplasia develop compared with that of patients with ulcerative colitis without primary sclerosing cholangitis. Results: Eleven studies met all eligibility criteria for the meta-analysis. Patients with ulcerative colitis and primary sclerosing cholangitis are at increased risk of colorectal dysplasia and carcinoma compared with patients with ulcerative colitis alone; OR 4.79: 95% CI [3.58, 6.41] with the Mantel-Haenszel method, and OR 5.11: 95% CI [3.15, 8.29] with the Der Simonian and Laird method. This increased risk is present even when the risk of colorectal carcinoma alone is considered; OR 4.09: 95% CI [2.89, 5.76] and OR 4.26: 95% CI [2.80, 6.48] by using, respectively, the MantelHaenszel and the Der Simonian and Laird methods. Conclusions: Patients with ulcerative colitis and primary sclerosing cholangitis have a significantly higher risk for the development of colorectal neoplasia than patients with ulcerative colitis but not primary sclerosing cholangitis. More intensive colonoscopic surveillance should be considered for patients with ulcerative colitis and primary sclerosing cholangitis. (Gastrointest Endosc 2002;56:48-54.)
Ulcerative colitis (UC), especially if extensive and of long duration, is strongly associated with primary sclerosing cholangitis (PSC), a progressive, fibrotic, cholestatic hepatobiliary inflammatory disease.1 The risk that a patient with UC will eventually have PSC develop is between 2.5% and 5%. Conversely, approximately 75% of patients with PSC have inflammatory bowel disease (IBD); the majority of Received March 21, 2001. For revision May 21, 2001. Accepted February 29, 2002. Current affiliations: VA Palo Alto, Stanford University School of Medicine, Department of Medicine, Division of Gastroenterology, and Department of Health Research and Policy, Stanford, California, and University of Chicago-Pritzker School of Medicine, Department of Medicine, Section of Gastroenterology, Chicago, Illinois. Supported in part by an American Digestive Health Foundation Outcomes Research Training Award to Dr. Soetikno. Reprint requests: Roy M. Soetikno, MD, Gastroenterology Section (111-GI), Palo Alto Veterans Affairs Health Care System, 1801 Miranda Ave., Palo Alto, CA 94304. 37/1/125367 doi:10.1067/mge.2002.125367 48
GASTROINTESTINAL ENDOSCOPY
these patients have UC.2 Patients with UC are also at increased risk for colorectal (CR) neoplasia (cancer and/or dysplasia). The risk for developing CR neoplasia becomes appreciable after a decade of colitis, especially with extensive disease; this risk increases with time.3 Broomé et al.1 published in 1992 the first report suggesting that patients who have both UC and PSC are at a higher risk for developing CR neoplasia compared with patients with UC alone. Since then, the results of many studies have supported these findings,4-11 although some have suggested the clinical significance of this risk may be low.12,13 At present, the question of whether PSC is associated with a higher risk of CR neoplasia in patients with UC remains unresolved. This issue is of clinical importance because of the significant risk of CR neoplasia in patients with UC and the increased life expectancy of patients with UC and PSC in the present era of orthotopic liver transplantation (OLT). To further investigate this issue, a meta-analysis of available studies was undertaken. VOLUME 56, NO. 1, 2002
Colonic neoplasia: patients with primary sclerosing cholangitis, ulcerative colitis
R Soetikno, O Lin, P Heidenreich, et al.
Establishing that patients with both UC and PSC are at higher risk of CR cancer would define a subset of patients in need of more intensive surveillance, regardless of the estimated duration of UC, which is notoriously difficult to determine accurately.14 It has been suggested that all such patients should undergo annual colonoscopic surveillance, irrespective of duration of UC.15 Moreover, the finding of definite dysplasia, even low grade, in these patients would become an absolute indication for colectomy. Surveillance is equally important for patients with PSC and UC who are awaiting OLT, as well as those who have undergone OLT, because transplantation does not reduce the risk of dysplasia or cancer.16
of publication (full article, abstract, letter to the editor), study design (prospective or retrospective), and reliability of the diagnosis of UC and PSC (with or without histopathologic confirmation). Studies published in full, those performed prospectively, and those that used strict criteria for the diagnosis of UC and PSC received the highest score of 4. Ambiguity or absence of information in any category resulted in a score of zero for that category. Quantitative data were abstracted on the number of patients with UC with and without PSC and the number with CR neoplasia and CR carcinoma. Quality assessment and quantitative abstraction were performed by 3 investigators (R.M.S., O.S.L., P.A.N.); they also resolved disagreements. These investigators also contacted authors of studies to clarify ambiguous information and/or to obtain missing quantitative data.
PATIENTS AND METHODS Data identification and selection
Data analysis
Three investigators (R.M.S., O.S.L., P.A.N.) independently searched the MEDLINE database for citations for January 1985 to December 2001 by using the keywords inflammatory bowel disease, ulcerative colitis, and sclerosing cholangitis. In addition, a manual search was performed with the citations in relevant articles and the proceedings from meetings of the American Gastroenterological Association, the American College of Gastroenterology, and the American Association for the Study of Liver Diseases (between 1992 and 2001). The titles and abstracts of all pertinent studies that compared the incidence of CR neoplasia in patients with UC with and without PSC were reviewed. There was no language restriction. Inclusion and exclusion criteria To be included in the meta-analysis, each study had to contain information on the size of the population at risk, that is, the number of patients with UC and PSC and the observed number of patients with CR neoplasia. In addition, the study had to contain information concerning the prevalence of CR neoplasia in a control population, patients with UC who did not have PSC. If such information was unavailable in the published article, additional information needed to reconstruct a two-by-two table was requested from the authors. Any study comparing patients with UC and PSC to patients with UC alone with an end point of CR neoplasia was included. Both populations had to be followed for the detection of CR neoplasia by using similar methods. Because it is well known that the occurrence of CR neoplasia varies according to many factors,17 including geographic region, it was important that the control group consisted of comparable patients; that is, the population at risk and the control population had to come from similar geographic regions. Case control studies were excluded because these mandate by design nonrandom selection of case and control patients. Quality assessment and quantitative abstraction of data The methodological quality of the relevant studies was assessed. Each study was given a score based on the type VOLUME 56, NO. 1, 2002
For the outcomes considered, the effect of the presence or absence of PSC was calculated by using odds ratios (ORs). The summary OR and its 95% confidence interval (CI) were calculated and homogeneity was tested with the Mantel-Haenszel method.18,19 To avoid problems if any cell in the two-by two tables contained zero, a modification of the Mantel-Haenszel technique was used (addition of a correction factor of 0.5 to each cell). In addition, the Der Simonian and Laird method was used to summarize data in a random-effects model.19 Subgroup analyses were performed (complete articles vs. abstracts and letters; high methodologic score studies vs. low score studies), all planned a priori, to test the strength of the common OR. The corrected Mantel-Haenszel chi-square test (one degree of freedom, two-tailed) was used to test whether ORs were different from 1. Finally, an estimate was made of the number of studies with null results needed to overturn the conclusion of the summary data with the Rosenthal method (Manual of OR2X2X2ka, Department of Clinical Epidemiology and Biostatistics, McMaster University, Hamilton, Ontario, Canada). All calculations were performed with Excel 97 software (Microsoft Corp., Redmond, Wash.) and OR2X2Xka software (McMaster University).
RESULTS Study selection Computer-based and manual bibliographic searches identified 22 pertinent articles.1,4-13,20-30 Eleven articles that contained 12 studies met the inclusion criteria.1,4-11,13,23 Eight studies were published as full articles in peer-reviewed journals,1,5-10,23 2 were published in abstract form,4,11 and 2 studies were published in a single letter to the editor13 in response to the original report of Broomé et al.1 However, one study23 had to be excluded from the analysis because, despite correspondence with the authors, it was not possible to construct the two-bytwo table required for the analysis. Of the 11 studies excluded from the analysis, two used indirect comparison groups. In one retrospecGASTROINTESTINAL ENDOSCOPY
49
R Soetikno, O Lin, P Heidenreich, et al.
Colonic neoplasia: patients with primary sclerosing cholangitis, ulcerative colitis
Table 1. Studies of the risk of CR neoplasia in patients with UC with and without PSC included in the meta-analysis Patients (UC with PSC/UC without PSC) Study; publication type Broomé, 19921; full article
Study setting Surveillance program (SP) at S. Hospital in Stockholm SP at Lahey Clinic
Choi, 1992,13 part I; letter to the editor Choi, 1992,13 part II PSC patients at Lahey Clinic Ahnen, 19954; SP (Denver Dysplasia in abstract UC Study)
No. included
No. with neoplasia
No. with CR carcinoma
5/67
4/5
1/1
5/199
2/43
1/21
62/2,149
2/41
2/41
11/68
4/4
1/0
Broomé,19955; full article
Hospitals in Stockholm
40/80
16/10
6/1
Brentnall, 19966; full article
SP at University of Washington
20/25
9/4
0/0
Bansal, 19967; full article
Veterans Affairs computerized database
201/11,245
17/354
17/354
Shetty, 19998; full article
Computerized records at Cleveland Clinic
132/196
33/11
17/5
Marchesa, 19979; full article
Proctocolectomy patients at Cleveland Clinic
27/1185
18/145
4/49
Leidenius, 199710; full article
Helsinki University Central Hospital
45/45
13/4
4/1
Timmermanns, 199611 abstract
University of Heidelberg Hospital
16/457
—
7/27
564/16280
118/621
60/500
Total
tive study, the incidence of CR neoplasia in 35 patients was compared with 4 historical control groups reported in 1966, 1971, 1977, and 1978.22 In another study, the relative risk of CR carcinoma in patients with PSC and UC seen at the Mayo Clinic was compared with that in patients with UC from central Sweden.12 Three other studies were excluded because a comparison group was not included.25,28,29 One study was excluded because its case-control design mandated nonrandom selection of case and control patients.27 Finally, 5 studies of the incidence of CR neoplasia in patients after OLT for PSC were excluded because they lacked control groups or the control group was from a cohort of patients from a markedly different period.20,21,24,26,30 Quality assessment and quantitative abstraction of data A scale of 0 to 4 was used for the methodologic quality scores. Four primary authors (Drs. Broomé, Sonnenberg, Lashner, and Sandborn) were contact50
GASTROINTESTINAL ENDOSCOPY
Conclusions of study “UC patients with PSC seem to run an increased risk of developing colonic dysplasia …” “… risk of CR neoplasia is not significantly influenced by the presence of PSC.” Ibid “… aneuploidy occurred more commonly in patients with PSC than in those without PSC.” “… PSC and UC have a significantly higher risk of developing colorectal neoplasia…” “…PSC and UC represent a subset of patients with UC who are at markedly increased risk …” “Sclerosing cholangitis is associated with a strong risk of developing colon cancer…” “UC patients with PSC are at increased risk of developing CR cancer or dysplasia …” “UC patients with PSC are at a significantly increased risk for dysplasia or cancer…” “… risk of CR dysplasia and cancer in patients with UC is increased by concomitant PSC.” “Patients with UC-associated PSC exhibit an increased risk to develop malignant tumor …”
ed. Two were able to supply corrective or missing data.1,5,7 The primary author of 2 articles from the same institution confirmed that there was no overlap of patients in the 2 articles.1,5 Of 16,844 patients that were included in this analysis, 560 had CR carcinoma. Of the 564 patients with UC and PSC included, 60 had CR carcinoma. There was 100% agreement in quality assessment and 91% agreement in quantitative abstraction. The latter was resolved by consensus review of the publications in question. Risks for colorectal neoplasia The outcomes considered in the 11 studies varied. Two included DNA aneuploidy and low-grade dysplasia.1,4 Because the objective of the analysis was to determine whether the presence of PSC in patients with UC increases the risk of neoplasia, patients with findings limited to DNA aneuploidy were excluded. The distribution of patients with dysplasia and/or carcinoma reported for each study VOLUME 56, NO. 1, 2002
Colonic neoplasia: patients with primary sclerosing cholangitis, ulcerative colitis
R Soetikno, O Lin, P Heidenreich, et al.
Figure 1. Effect of primary sclerosing cholangitis on risk of developing colorectal neoplasia in patients with ulcerative colitis.
Figure 2. Effect of primary sclerosing cholangitis on risk of developing colorectal carcinoma in patients with ulcerative colitis.
is shown in Table 1. The common OR of developing CR neoplasia (dysplasia or carcinoma) was calculated first, using data from 10 studies1,4-10,13 because one study reported only data on colorectal carcinoma.11 The presence of PSC was associated with a higher risk for the development of CR neoplasia in patients with UC (OR 4.79: 95% CI [3.58, 6.41]) (Fig. 1). The common OR based on the random-effects model was calculated as OR 5.11: 95% CI [3.15, 8.29]. Heterogeneity was not detected (Q statistic –18.1, p = 0.03). Because the histopathologic diagnosis of dysplasia is subject to interobserver variation, an analogous analysis was performed to evaluate only outcomes limited to the presence or absence of adenocarcinoma, with data from all 11 studies. This analysis found that the presence of PSC is associated with an increased risk for the development of CR carcinoma (OR 4.09: 95% CI [2.89, 5.76]) (Fig. 2). The OR for the development of CR carcinoma calculated by the random-effect model was 4.26: 95% CI [2.80, 6.48]. Heterogeneity was present (Q statistic –11.2, p = 0.35).
OR recalculated in each case (results not shown). A dominant study was not identified.
Subgroup analyses Subgroup analyses for publication status were performed. The common OR (for CR neoplasia) among studies published in full was 5.02 (95% CI [3.68, 6.86]). Studies reported as abstracts or letters had a common OR of 3.30 (95% CI [1.19, 9.28]). Studies with high methodological scores (i.e., 3 or 4) turned out to be those published as complete articles, whereas studies published as abstracts or letters to the editor all had scores less than 3. To test whether any study had a dominant effect on the common OR, each study was excluded one by one and the common VOLUME 56, NO. 1, 2002
Estimation of the number of unpublished studies needed to alter conclusions The calculated number of unpublished studies with null results required to change the conclusion of the present meta-analysis was 163. This value represents the number of studies needed to render the findings null. It has been suggested that if this number is at least 5 times the number of retrieved studies plus 10, then the findings of the meta-analysis are not influenced by publication bias.31 DISCUSSION The present meta-analysis indicates an approximate 4-fold increased risk for CR neoplasia in patients with UC and PSC, compared with those with UC alone. This increased risk does not change appreciably when the presence or absence of carcinoma is the only outcome considered. The inclusion of only high-quality studies is one of the important determinants of the validity of a meta-analysis. In the present synthesis, studies reported in abstract form and as letters to the editor were included because the number of peer-reviewed articles reporting the risk of CR neoplasia in patients with UC with and without PSC is limited. The effects of these partially published studies was analyzed critically, and the conclusion of the meta-analysis does not change even if they are excluded. It is well known that studies with null results are less likely to be published. Thus, a calculation was made of the number of unpublished studies with null results required to change our conclusion: the result was 163 studies. Although the applicability of GASTROINTESTINAL ENDOSCOPY
51
R Soetikno, O Lin, P Heidenreich, et al.
Colonic neoplasia: patients with primary sclerosing cholangitis, ulcerative colitis
Rosenthal’s method is heuristic, this large number indicates that publication bias is unlikely to be the sole cause of the high summary OR. Our conclusion from the 11 studies examined is that patients with UC and PSC have a significantly higher risk for CR neoplasia than patients with UC alone. The apparent association of PSC with an increased risk for the development of CR neoplasia in patients with UC is intriguing, although no satisfying explanation has been forthcoming. Alterations in bile acids have been suspected to occur in PSC and to account for the association with cancer and dysplasia; however, direct evidence for this is currently lacking. In fact, biliary secretion of bile acids in PSC is either normal or reduced.32 The lowered frequency of dysplasia in patients treated with ursodiol observed in a recently reported study was attributed to a reduced concentration of deoxycholic acid.33 However, the concentration of this bile acid is already reduced in the bile of patients with PSC, so that whatever chemoprotection effect ursodiol has would be due to an action34 other than an effect on a preexisting high fecal level of bile acids. For these reasons, the decrease in risk for CR neoplasia is probably not related to alteration of colonic bile acids.35,36 A more likely explanation is found in the subset of patients who have both UC and PSC. As with patients with UC at increased risk for both dysplasia and cancer,3,7 patients with PSC and UC are more likely to have extensive colitis of long duration.37 Therefore, the increased risk in patients with both PSC and UC may be the consequence of association with pancolitis and long duration of disease. In this subset, the risk of dysplasia of about 27% and that of cancer of 14% reflects the risks associated with extensive disease of duration in excess of 25 years.12 The estimation of duration of disease based on history is often inexact, and the typically long subclinical phase of PSC before cholestatic symptoms (e.g., 10-15 years),14 coupled with the duration of UC at the onset of PSC (e.g., 10 years),37 may be sufficient to ensure that a large proportion of this subset actually have UC that has been present for over 25 years. Although logistic regression analysis in one study1 found that association of PSC with dysplasia was independent of duration, this finding may have been influenced by underestimation of duration based on clinical history. At present, it is not possible to exclude some causal contribution of PSC itself to the progression from long-standing extensive UC to neoplasia. The carcinogenic events that might be promoted by the presence of PSC remain poorly understood. One possibility would be an immunogenetic association of PSC with abnormal p53 expression.38 Alteration of 52
GASTROINTESTINAL ENDOSCOPY
this DNA repair protein might favor the development of neoplastic changes.39,40 PSC may also promote the expression of sialosyl-Tn antigen,41 an antigen that appears to precede the presence of dysplasia in UC. Finally, antibodies directed against nuclear and cytoplasmic constituents of neutrophils (ANCA) have been of particular interest because of their high prevalence in patients with PSC and UC.42 Whether these antibodies are indicative of a specific underlying immunologic disturbance that allows for defective antigen recognition is unknown and requires further study. Tung et al.33 recently reported lower rates of colorectal dysplasia in 59 patients with UC and PSC who were treated with ursodiol. Although the potential of ursodiol as a chemopreventive agent appears promising, this study involved a small cohort of patients followed for an unspecified period at a single institution. Whether the increased rate of CR neoplasia in patients with UC and PSC shown in the present meta-analysis can be reduced by treatment with ursodiol is unclear at present. Although the current meta-analysis found that patients with UC and PSC have a significantly higher risk for CR neoplasia, the major cause of morbidity and mortality in these patients is usually endstage liver disease and its complications, which include cholangiocarcinoma but not colon cancer.12,14,43,44 Thus, whether aggressive screening colonoscopy or early proctocolectomy would result in a significant life-expectancy advantage is unclear. PSC is now one of the leading indications for OLT. Patients with PSC who undergo OLT are now living longer than previously, with 1- and 3-year survival rates of, respectively, 85% and 75%.45 The development of CR neoplasia in these patients could be a major cause of mortality.24 The present analysis indicates the need for a careful management strategy to prevent or detect the development of CR neoplasia in this group of high-risk patients. For the present this would include yearly surveillance colonoscopy with 4 biopsy specimens taken at 10-cm intervals, and colectomy would be indicated for definite dysplasia of any grade, provided this diagnosis is confirmed by a pathologist with expertise in the interpretation of dysplasia. Upon entering a surveillance program, patients must be educated as to the increased risk associated with the combination of UC and PSC. They must be motivated to undergo surveillance on a yearly basis, as well as to submit to colectomy if dysplasia is detected and confirmed by a pathologist with the requisite expertise. In these patients, colonoscopy surveillance should be performed annually regardless of the duration of UC. Future studies of VOLUME 56, NO. 1, 2002
Colonic neoplasia: patients with primary sclerosing cholangitis, ulcerative colitis
patients who participate in programs of annual surveillance, including colectomy for definite dysplasia, will determine whether patients with PSC and UC actually have an accelerated development of cancer. REFERENCES 1. Broomé U, Lindberg G, Lofberg R. Primary sclerosing cholangitis in ulcerative colitis—a risk factor for the development of dysplasia and DNA aneuploidy? Gastroenterology 1992;102: 1877-80. 2. Lee YM, Kaplan MM. Primary sclerosing cholangitis. N Engl J Med 1995;332:924-33. 3. Ekbom A, Helmick C, Zack M, Adanii HO. Ulcerative colitis and colorectal cancer: a population-based study. N Engl J Med 1990;323:1228-33. 4. Ahnen DJ, McHugh JBD, Arsenault LL, Warren G. Does primary sclerosing cholangitis (PSC) increase the risk of colon cancer in patients with chronic ulcerative colitis (CUC) [abstract]? Gastroenterology 1993;104:A658. 5. Broomi U, Lofberg R, Veress B, Eriksson LS. Primary sclerosing cholangitis and ulcerative colitis: evidence for increased neoplastic potential. Hepatology 1995;22:1404-8. 6. Brentnall TA, Haggitt RC, Rabinovitch PS, Kimmey MB, Bronner MP, Levine DS, et al. Risk and natural history of colonic neoplasia in patients with primary sclerosing cholangitis and ulcerative colitis. Gastroenterology 1996;110:331-8. 7. Bansal P, Sonnenberg A. Risk factors of colorectal cancer in inflammatory bowel disease. Am J Gastroenterol 1996;91:44-8. 8. Shetty K, Rybicki L, Brzezinski A, Carey WD, Lashner BA. The risk for cancer or dysplasia in ulcerative colitis patients with primary sclerosing cholangitis. Am J Gastroenterol 1999;94:1643-9. 9. Marchesa P, Lashner BA, Lavery IC, Milsom J, Hull TL, Strong SA, et al. The risk of cancer and dysplasia among ulcerative colitis patients with primary sclerosing cholangitis. Am J Gastroenterol 1997;92:1285-8. 10. Leidenius MH, Farkkila MA, Karkkainen P, Taskinen EI, Kellokumpu IH, Hockerstedt KA. Colorectal dysplasia and carcinoma in patients with ulcerative colitis and primary sclerosing cholangitis. Scan J Gastroenterol 1997;32:706-11. 11. Timmermanns G, Heuschen UA, Stern J, Herfath CH. Enhanced colorectal cancer risk in patients with ulcerative colitis associated with PSC [abstract]. Gastroenterology 1996; 110:A603. 12. Loftus EV, Sandborn WJ, Tremaine WJ, Mahoney DW, Zinsmeister AR, Offord KP, et al. Risk of colorectal neoplasia in patients with primary sclerosing cholangitis. Gastroenterology 1996;110:432-40. 13. Choi PM, Nugent FW, Rossi RL. Relationship between colorectal neoplasia and primary sclerosing cholangitis in ulcerative colitis. Gastroenterology 1992;103:1707-9. 14. Aadland E, Schrunpf E, Fausa O, Elgjo K, Heilo A, Aakhus T, Gjone E. Primary sclerosing cholangitis: a long-term followup study. Scand J Gastroenterol 1987;22:655-64. 15. Eaden JA, Mayberry JF. Colorectal cancer complicating ulcerative colitis: a review. Am J Gastroenterol 2000;95:2710-9. 16. Lashner BA, Provencher KS, Seidner DL, Knesebeck A, Brzezinski A. The effect of folio acid supplementation on the risk for cancer or dysplasia in ulcerative colitis. Gastroenterology 1997;112:29-32. 17. Ransohoff DF. Colorectal cancer. In: Everhart YE, editor. Digestive diseases in the United States: epidemiology and impact. US Department of Health and Human Services, Public Service, National Institutes of Health, National VOLUME 56, NO. 1, 2002
18. 19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31. 32.
33.
34.
35.
R Soetikno, O Lin, P Heidenreich, et al.
Institute of Diabetes and Digestive and Kidney Diseases. Washington, DC: US Government Printing Office; 1994. NIH publication no. 94-1447. Hedges LV, Olkin I. Statistical methods for meta-analysis. San Diego: Academic Press; 1985. Petitti DB. Meta-analysis decision analysis and cost-effectiveness analysis. Methods for quantitative synthesis in medicine. New York: Oxford University Press; 1994. Higashi H, Yanaga K, Marsh JW, Tzakis A, Kakizoe S, Starzl TE. Development of colon cancer after liver transplantation for primary sclerosing cholangitis associated with ulcerative colitis. Hepatology 1990;11:477-80. Bleday R, Lee E, Jessurun J, Heine J, Wong WD. Increased risk of early colorectal neoplasms after hepatic transplant in patients with inflammatory bowel disease. Dis Colon Rectum 1993;36:908-12. Gurbuz AK, Giardiello FM, Bayless TM. Colorectal neoplasia in patients with ulcerative colitis and primary sclerosing cholangitis. Dis Colon Rectum 1995;38:37-41. D’Haens GR, Lashner BA, Hanauer SB. Pericholangitis and sclerosing cholangitis are risk factors for dysplasia and cancer in ulcerative colitis. Am J Gastroenterol 1993;88:1174-8. Narumi S, Roberts JP, Emond JC, Lake J, Asoher NL. Liver transplantation for sclerosing cholangitis. Hepatology 1995; 22:451-7. Pinczowski D, Ekjbom A. Is there an increased risk of colorectal cancer among ulcerative colitis patients with primary sclerosing cholangitis [abstract]? Gastroenterology 1995;108: A29. Shaked A, Colonna JO, Goldstein L, Busuttil RW. The interrelation between sclerosing cholangitis and ulcerative colitis in patients undergoing liver transplantation. Ann Surg 1992; 215:598-603. Nuako KW, Ahlquist DA, Sandborn WJ, Mahoney DW, Siems DM, Zinsmeister AR. Primary sclerosing cholangitis and colorectal carcinoma in patients with chronic ulcerative colitis: a case-control study. Cancer 1998;82:822-6. Kornfeld D, Ekbom A, Ihre T. Is there an excessive risk for colorectal cancer in patients with ulcerative colitis and concomitant primary sclerosing cholangitis? A population based study. Gut 1997;41:522-5. Kingham JG. Risk of colon cancer in patients with ulcerative colitis is potentiated by the presence of primary sclerosing cholangitis. Hepatology 1997;25:254. Loftus EV Jr, Aguilar HI, Sandborn WJ, Tremaine WJ, Krom RA, Zinsmeister AR, et al. Risk of colorectal neoplasia in patients with primary sclerosing cholangitis and ulcerative colitis following orthotopic liver transplantation. Hepatology 1998;27:685-90. Rosenthal R. The “file drawer problem” and tolerance for null results. Psychol Bull 1979;86:638-41. Stiehl A, Rudolph G, Sauer P, Theilmann L. Biliary secretion of bile acids and lipids in primary sclerosing cholangitis: influence of cholestasis and effect of ursodeoxycholic acid treatment. J Hepatol 1995;23:283-9. Tung BY, Emond MH, Haggitt RC, Bronner MP, Kimmey MB, Kowdley KV, et al. Ursodiol use is associated with lower prevalence of colonic neoplasia in patients with ulcerative colitis and primary sclerosing cholangitis. Ann Intern Med 2001; 134:89-95. Hawk ET, Viner IL. Chemoprevention in ulcerative colitis: narrowing the gap between clinical practice and research. Ann Intern Med 2001;134:158-60. Blackstone MO. An association of fecal bile acids and colon cancer in ulcerative colitis? Gastroenterology 1988;95:575-6. GASTROINTESTINAL ENDOSCOPY
53
R Soetikno, O Lin, P Heidenreich, et al.
Colonic neoplasia: patients with primary sclerosing cholangitis, ulcerative colitis
36. Siegel JH, Barnes S, Morris JS. Bile acids in liver disease associated with inflammatory bowel disease. Digestion 1977; 15:469-81. 37. Fausa O, Schrumpf E, Elgjo K. Relationship of inflammatory bowel disease and primary sclerosing cholangitis. Semin Liver Dis 1991;11:31-9. 38. van Milligen de Wit AW, van Deventer SJ, Tytgat GN. Immunogenetic aspects of primary sclerosing cholangitis: implications for therapeutic strategies. Am J Gastroenterol 1995;90:893-900. 39. Ilyas M, Talbot IC. p53 Expression in ulcerative colitis: a longitudinal study. Gut 1995;37:802-4. 40. Burmer GC, Rabinovitch PS, Haggitt RC, Crispin DA, Brentnall TA, Kolli VR, et al. Neoplastic progression in ulcerative colitis: histology, DNA content, and loss of p53 allele. Gastroenterology 1992;103:1602-10.
54
GASTROINTESTINAL ENDOSCOPY
41. Itzkowitz SH, Young E, DuBois D, Harpaz N, Bodian C, Chen A, et al. Sialosyl-Tn antigen is prevalent and precedes dysplasia in ulcerative colitis: a retrospective case-control study. Gastroenterology 1996;110:694-704. 42. Seibold F, Slametschka D, Gregor M, Weber P. Neutrophil autoantibodies: a genetic marker in primary sclerosing cholangitis and ulcerative colitis. Gastroenterology 1994;107: 532-6. 43. Broomé U, Glaumann H, Hellers G, Nilsson B, Sorstad J, Hultcrantz R. Liver disease in ulcerative colitis: an epidemiological and follow up study in the county of Stockholm. Gut 1994;35:84-9. 44. Lebovics E, Palmer M, Woo J, Schaffner F. Outcome of primary sclerosing cholangitis. Arch Intern Med 1987;147:729-31. 45. Wiesner RH. Current concepts in primary sclerosing cholangitis. Mayo Clin Proc 1994;69:969-82.
VOLUME 56, NO. 1, 2002
Background: Published data on the risk of colorectal neoplasia in patients with ulcerative colitis with and without primary sclerosing cholangitis are conflicting. A meta-analysis was performed to synthesize available publications and to compare the risk of colorectal neoplasia in patients with ulcerative colitis with and without primary sclerosing cholangitis. Methods: By using MEDLINE and manual search methods, studies were identified that compared the risk of colorectal neoplasia (dysplasia and carcinoma) in patients with ulcerative colitis with and without primary sclerosing cholangitis. In addition, citations were reviewed in relevant articles and proceedings from gastroenterology meetings, and investigators were contacted when data were incomplete. The summary odds ratio (OR) was then calculated for the risk for patients with ulcerative colitis and primary sclerosing cholangitis of having colorectal neoplasia develop compared with that of patients with ulcerative colitis without primary sclerosing cholangitis. Results: Eleven studies met all eligibility criteria for the meta-analysis. Patients with ulcerative colitis and primary sclerosing cholangitis are at increased risk of colorectal dysplasia and carcinoma compared with patients with ulcerative colitis alone; OR 4.79: 95% CI [3.58, 6.41] with the Mantel-Haenszel method, and OR 5.11: 95% CI [3.15, 8.29] with the Der Simonian and Laird method. This increased risk is present even when the risk of colorectal carcinoma alone is considered; OR 4.09: 95% CI [2.89, 5.76] and OR 4.26: 95% CI [2.80, 6.48] by using, respectively, the MantelHaenszel and the Der Simonian and Laird methods. Conclusions: Patients with ulcerative colitis and primary sclerosing cholangitis have a significantly higher risk for the development of colorectal neoplasia than patients with ulcerative colitis but not primary sclerosing cholangitis. More intensive colonoscopic surveillance should be considered for patients with ulcerative colitis and primary sclerosing cholangitis. (Gastrointest Endosc 2002;56:48-54.)
Ulcerative colitis (UC), especially if extensive and of long duration, is strongly associated with primary sclerosing cholangitis (PSC), a progressive, fibrotic, cholestatic hepatobiliary inflammatory disease.1 The risk that a patient with UC will eventually have PSC develop is between 2.5% and 5%. Conversely, approximately 75% of patients with PSC have inflammatory bowel disease (IBD); the majority of Received March 21, 2001. For revision May 21, 2001. Accepted February 29, 2002. Current affiliations: VA Palo Alto, Stanford University School of Medicine, Department of Medicine, Division of Gastroenterology, and Department of Health Research and Policy, Stanford, California, and University of Chicago-Pritzker School of Medicine, Department of Medicine, Section of Gastroenterology, Chicago, Illinois. Supported in part by an American Digestive Health Foundation Outcomes Research Training Award to Dr. Soetikno. Reprint requests: Roy M. Soetikno, MD, Gastroenterology Section (111-GI), Palo Alto Veterans Affairs Health Care System, 1801 Miranda Ave., Palo Alto, CA 94304. 37/1/125367 doi:10.1067/mge.2002.125367 48
GASTROINTESTINAL ENDOSCOPY
these patients have UC.2 Patients with UC are also at increased risk for colorectal (CR) neoplasia (cancer and/or dysplasia). The risk for developing CR neoplasia becomes appreciable after a decade of colitis, especially with extensive disease; this risk increases with time.3 Broomé et al.1 published in 1992 the first report suggesting that patients who have both UC and PSC are at a higher risk for developing CR neoplasia compared with patients with UC alone. Since then, the results of many studies have supported these findings,4-11 although some have suggested the clinical significance of this risk may be low.12,13 At present, the question of whether PSC is associated with a higher risk of CR neoplasia in patients with UC remains unresolved. This issue is of clinical importance because of the significant risk of CR neoplasia in patients with UC and the increased life expectancy of patients with UC and PSC in the present era of orthotopic liver transplantation (OLT). To further investigate this issue, a meta-analysis of available studies was undertaken. VOLUME 56, NO. 1, 2002
Colonic neoplasia: patients with primary sclerosing cholangitis, ulcerative colitis
R Soetikno, O Lin, P Heidenreich, et al.
Establishing that patients with both UC and PSC are at higher risk of CR cancer would define a subset of patients in need of more intensive surveillance, regardless of the estimated duration of UC, which is notoriously difficult to determine accurately.14 It has been suggested that all such patients should undergo annual colonoscopic surveillance, irrespective of duration of UC.15 Moreover, the finding of definite dysplasia, even low grade, in these patients would become an absolute indication for colectomy. Surveillance is equally important for patients with PSC and UC who are awaiting OLT, as well as those who have undergone OLT, because transplantation does not reduce the risk of dysplasia or cancer.16
of publication (full article, abstract, letter to the editor), study design (prospective or retrospective), and reliability of the diagnosis of UC and PSC (with or without histopathologic confirmation). Studies published in full, those performed prospectively, and those that used strict criteria for the diagnosis of UC and PSC received the highest score of 4. Ambiguity or absence of information in any category resulted in a score of zero for that category. Quantitative data were abstracted on the number of patients with UC with and without PSC and the number with CR neoplasia and CR carcinoma. Quality assessment and quantitative abstraction were performed by 3 investigators (R.M.S., O.S.L., P.A.N.); they also resolved disagreements. These investigators also contacted authors of studies to clarify ambiguous information and/or to obtain missing quantitative data.
PATIENTS AND METHODS Data identification and selection
Data analysis
Three investigators (R.M.S., O.S.L., P.A.N.) independently searched the MEDLINE database for citations for January 1985 to December 2001 by using the keywords inflammatory bowel disease, ulcerative colitis, and sclerosing cholangitis. In addition, a manual search was performed with the citations in relevant articles and the proceedings from meetings of the American Gastroenterological Association, the American College of Gastroenterology, and the American Association for the Study of Liver Diseases (between 1992 and 2001). The titles and abstracts of all pertinent studies that compared the incidence of CR neoplasia in patients with UC with and without PSC were reviewed. There was no language restriction. Inclusion and exclusion criteria To be included in the meta-analysis, each study had to contain information on the size of the population at risk, that is, the number of patients with UC and PSC and the observed number of patients with CR neoplasia. In addition, the study had to contain information concerning the prevalence of CR neoplasia in a control population, patients with UC who did not have PSC. If such information was unavailable in the published article, additional information needed to reconstruct a two-by-two table was requested from the authors. Any study comparing patients with UC and PSC to patients with UC alone with an end point of CR neoplasia was included. Both populations had to be followed for the detection of CR neoplasia by using similar methods. Because it is well known that the occurrence of CR neoplasia varies according to many factors,17 including geographic region, it was important that the control group consisted of comparable patients; that is, the population at risk and the control population had to come from similar geographic regions. Case control studies were excluded because these mandate by design nonrandom selection of case and control patients. Quality assessment and quantitative abstraction of data The methodological quality of the relevant studies was assessed. Each study was given a score based on the type VOLUME 56, NO. 1, 2002
For the outcomes considered, the effect of the presence or absence of PSC was calculated by using odds ratios (ORs). The summary OR and its 95% confidence interval (CI) were calculated and homogeneity was tested with the Mantel-Haenszel method.18,19 To avoid problems if any cell in the two-by two tables contained zero, a modification of the Mantel-Haenszel technique was used (addition of a correction factor of 0.5 to each cell). In addition, the Der Simonian and Laird method was used to summarize data in a random-effects model.19 Subgroup analyses were performed (complete articles vs. abstracts and letters; high methodologic score studies vs. low score studies), all planned a priori, to test the strength of the common OR. The corrected Mantel-Haenszel chi-square test (one degree of freedom, two-tailed) was used to test whether ORs were different from 1. Finally, an estimate was made of the number of studies with null results needed to overturn the conclusion of the summary data with the Rosenthal method (Manual of OR2X2X2ka, Department of Clinical Epidemiology and Biostatistics, McMaster University, Hamilton, Ontario, Canada). All calculations were performed with Excel 97 software (Microsoft Corp., Redmond, Wash.) and OR2X2Xka software (McMaster University).
RESULTS Study selection Computer-based and manual bibliographic searches identified 22 pertinent articles.1,4-13,20-30 Eleven articles that contained 12 studies met the inclusion criteria.1,4-11,13,23 Eight studies were published as full articles in peer-reviewed journals,1,5-10,23 2 were published in abstract form,4,11 and 2 studies were published in a single letter to the editor13 in response to the original report of Broomé et al.1 However, one study23 had to be excluded from the analysis because, despite correspondence with the authors, it was not possible to construct the two-bytwo table required for the analysis. Of the 11 studies excluded from the analysis, two used indirect comparison groups. In one retrospecGASTROINTESTINAL ENDOSCOPY
49
R Soetikno, O Lin, P Heidenreich, et al.
Colonic neoplasia: patients with primary sclerosing cholangitis, ulcerative colitis
Table 1. Studies of the risk of CR neoplasia in patients with UC with and without PSC included in the meta-analysis Patients (UC with PSC/UC without PSC) Study; publication type Broomé, 19921; full article
Study setting Surveillance program (SP) at S. Hospital in Stockholm SP at Lahey Clinic
Choi, 1992,13 part I; letter to the editor Choi, 1992,13 part II PSC patients at Lahey Clinic Ahnen, 19954; SP (Denver Dysplasia in abstract UC Study)
No. included
No. with neoplasia
No. with CR carcinoma
5/67
4/5
1/1
5/199
2/43
1/21
62/2,149
2/41
2/41
11/68
4/4
1/0
Broomé,19955; full article
Hospitals in Stockholm
40/80
16/10
6/1
Brentnall, 19966; full article
SP at University of Washington
20/25
9/4
0/0
Bansal, 19967; full article
Veterans Affairs computerized database
201/11,245
17/354
17/354
Shetty, 19998; full article
Computerized records at Cleveland Clinic
132/196
33/11
17/5
Marchesa, 19979; full article
Proctocolectomy patients at Cleveland Clinic
27/1185
18/145
4/49
Leidenius, 199710; full article
Helsinki University Central Hospital
45/45
13/4
4/1
Timmermanns, 199611 abstract
University of Heidelberg Hospital
16/457
—
7/27
564/16280
118/621
60/500
Total
tive study, the incidence of CR neoplasia in 35 patients was compared with 4 historical control groups reported in 1966, 1971, 1977, and 1978.22 In another study, the relative risk of CR carcinoma in patients with PSC and UC seen at the Mayo Clinic was compared with that in patients with UC from central Sweden.12 Three other studies were excluded because a comparison group was not included.25,28,29 One study was excluded because its case-control design mandated nonrandom selection of case and control patients.27 Finally, 5 studies of the incidence of CR neoplasia in patients after OLT for PSC were excluded because they lacked control groups or the control group was from a cohort of patients from a markedly different period.20,21,24,26,30 Quality assessment and quantitative abstraction of data A scale of 0 to 4 was used for the methodologic quality scores. Four primary authors (Drs. Broomé, Sonnenberg, Lashner, and Sandborn) were contact50
GASTROINTESTINAL ENDOSCOPY
Conclusions of study “UC patients with PSC seem to run an increased risk of developing colonic dysplasia …” “… risk of CR neoplasia is not significantly influenced by the presence of PSC.” Ibid “… aneuploidy occurred more commonly in patients with PSC than in those without PSC.” “… PSC and UC have a significantly higher risk of developing colorectal neoplasia…” “…PSC and UC represent a subset of patients with UC who are at markedly increased risk …” “Sclerosing cholangitis is associated with a strong risk of developing colon cancer…” “UC patients with PSC are at increased risk of developing CR cancer or dysplasia …” “UC patients with PSC are at a significantly increased risk for dysplasia or cancer…” “… risk of CR dysplasia and cancer in patients with UC is increased by concomitant PSC.” “Patients with UC-associated PSC exhibit an increased risk to develop malignant tumor …”
ed. Two were able to supply corrective or missing data.1,5,7 The primary author of 2 articles from the same institution confirmed that there was no overlap of patients in the 2 articles.1,5 Of 16,844 patients that were included in this analysis, 560 had CR carcinoma. Of the 564 patients with UC and PSC included, 60 had CR carcinoma. There was 100% agreement in quality assessment and 91% agreement in quantitative abstraction. The latter was resolved by consensus review of the publications in question. Risks for colorectal neoplasia The outcomes considered in the 11 studies varied. Two included DNA aneuploidy and low-grade dysplasia.1,4 Because the objective of the analysis was to determine whether the presence of PSC in patients with UC increases the risk of neoplasia, patients with findings limited to DNA aneuploidy were excluded. The distribution of patients with dysplasia and/or carcinoma reported for each study VOLUME 56, NO. 1, 2002
Colonic neoplasia: patients with primary sclerosing cholangitis, ulcerative colitis
R Soetikno, O Lin, P Heidenreich, et al.
Figure 1. Effect of primary sclerosing cholangitis on risk of developing colorectal neoplasia in patients with ulcerative colitis.
Figure 2. Effect of primary sclerosing cholangitis on risk of developing colorectal carcinoma in patients with ulcerative colitis.
is shown in Table 1. The common OR of developing CR neoplasia (dysplasia or carcinoma) was calculated first, using data from 10 studies1,4-10,13 because one study reported only data on colorectal carcinoma.11 The presence of PSC was associated with a higher risk for the development of CR neoplasia in patients with UC (OR 4.79: 95% CI [3.58, 6.41]) (Fig. 1). The common OR based on the random-effects model was calculated as OR 5.11: 95% CI [3.15, 8.29]. Heterogeneity was not detected (Q statistic –18.1, p = 0.03). Because the histopathologic diagnosis of dysplasia is subject to interobserver variation, an analogous analysis was performed to evaluate only outcomes limited to the presence or absence of adenocarcinoma, with data from all 11 studies. This analysis found that the presence of PSC is associated with an increased risk for the development of CR carcinoma (OR 4.09: 95% CI [2.89, 5.76]) (Fig. 2). The OR for the development of CR carcinoma calculated by the random-effect model was 4.26: 95% CI [2.80, 6.48]. Heterogeneity was present (Q statistic –11.2, p = 0.35).
OR recalculated in each case (results not shown). A dominant study was not identified.
Subgroup analyses Subgroup analyses for publication status were performed. The common OR (for CR neoplasia) among studies published in full was 5.02 (95% CI [3.68, 6.86]). Studies reported as abstracts or letters had a common OR of 3.30 (95% CI [1.19, 9.28]). Studies with high methodological scores (i.e., 3 or 4) turned out to be those published as complete articles, whereas studies published as abstracts or letters to the editor all had scores less than 3. To test whether any study had a dominant effect on the common OR, each study was excluded one by one and the common VOLUME 56, NO. 1, 2002
Estimation of the number of unpublished studies needed to alter conclusions The calculated number of unpublished studies with null results required to change the conclusion of the present meta-analysis was 163. This value represents the number of studies needed to render the findings null. It has been suggested that if this number is at least 5 times the number of retrieved studies plus 10, then the findings of the meta-analysis are not influenced by publication bias.31 DISCUSSION The present meta-analysis indicates an approximate 4-fold increased risk for CR neoplasia in patients with UC and PSC, compared with those with UC alone. This increased risk does not change appreciably when the presence or absence of carcinoma is the only outcome considered. The inclusion of only high-quality studies is one of the important determinants of the validity of a meta-analysis. In the present synthesis, studies reported in abstract form and as letters to the editor were included because the number of peer-reviewed articles reporting the risk of CR neoplasia in patients with UC with and without PSC is limited. The effects of these partially published studies was analyzed critically, and the conclusion of the meta-analysis does not change even if they are excluded. It is well known that studies with null results are less likely to be published. Thus, a calculation was made of the number of unpublished studies with null results required to change our conclusion: the result was 163 studies. Although the applicability of GASTROINTESTINAL ENDOSCOPY
51
R Soetikno, O Lin, P Heidenreich, et al.
Colonic neoplasia: patients with primary sclerosing cholangitis, ulcerative colitis
Rosenthal’s method is heuristic, this large number indicates that publication bias is unlikely to be the sole cause of the high summary OR. Our conclusion from the 11 studies examined is that patients with UC and PSC have a significantly higher risk for CR neoplasia than patients with UC alone. The apparent association of PSC with an increased risk for the development of CR neoplasia in patients with UC is intriguing, although no satisfying explanation has been forthcoming. Alterations in bile acids have been suspected to occur in PSC and to account for the association with cancer and dysplasia; however, direct evidence for this is currently lacking. In fact, biliary secretion of bile acids in PSC is either normal or reduced.32 The lowered frequency of dysplasia in patients treated with ursodiol observed in a recently reported study was attributed to a reduced concentration of deoxycholic acid.33 However, the concentration of this bile acid is already reduced in the bile of patients with PSC, so that whatever chemoprotection effect ursodiol has would be due to an action34 other than an effect on a preexisting high fecal level of bile acids. For these reasons, the decrease in risk for CR neoplasia is probably not related to alteration of colonic bile acids.35,36 A more likely explanation is found in the subset of patients who have both UC and PSC. As with patients with UC at increased risk for both dysplasia and cancer,3,7 patients with PSC and UC are more likely to have extensive colitis of long duration.37 Therefore, the increased risk in patients with both PSC and UC may be the consequence of association with pancolitis and long duration of disease. In this subset, the risk of dysplasia of about 27% and that of cancer of 14% reflects the risks associated with extensive disease of duration in excess of 25 years.12 The estimation of duration of disease based on history is often inexact, and the typically long subclinical phase of PSC before cholestatic symptoms (e.g., 10-15 years),14 coupled with the duration of UC at the onset of PSC (e.g., 10 years),37 may be sufficient to ensure that a large proportion of this subset actually have UC that has been present for over 25 years. Although logistic regression analysis in one study1 found that association of PSC with dysplasia was independent of duration, this finding may have been influenced by underestimation of duration based on clinical history. At present, it is not possible to exclude some causal contribution of PSC itself to the progression from long-standing extensive UC to neoplasia. The carcinogenic events that might be promoted by the presence of PSC remain poorly understood. One possibility would be an immunogenetic association of PSC with abnormal p53 expression.38 Alteration of 52
GASTROINTESTINAL ENDOSCOPY
this DNA repair protein might favor the development of neoplastic changes.39,40 PSC may also promote the expression of sialosyl-Tn antigen,41 an antigen that appears to precede the presence of dysplasia in UC. Finally, antibodies directed against nuclear and cytoplasmic constituents of neutrophils (ANCA) have been of particular interest because of their high prevalence in patients with PSC and UC.42 Whether these antibodies are indicative of a specific underlying immunologic disturbance that allows for defective antigen recognition is unknown and requires further study. Tung et al.33 recently reported lower rates of colorectal dysplasia in 59 patients with UC and PSC who were treated with ursodiol. Although the potential of ursodiol as a chemopreventive agent appears promising, this study involved a small cohort of patients followed for an unspecified period at a single institution. Whether the increased rate of CR neoplasia in patients with UC and PSC shown in the present meta-analysis can be reduced by treatment with ursodiol is unclear at present. Although the current meta-analysis found that patients with UC and PSC have a significantly higher risk for CR neoplasia, the major cause of morbidity and mortality in these patients is usually endstage liver disease and its complications, which include cholangiocarcinoma but not colon cancer.12,14,43,44 Thus, whether aggressive screening colonoscopy or early proctocolectomy would result in a significant life-expectancy advantage is unclear. PSC is now one of the leading indications for OLT. Patients with PSC who undergo OLT are now living longer than previously, with 1- and 3-year survival rates of, respectively, 85% and 75%.45 The development of CR neoplasia in these patients could be a major cause of mortality.24 The present analysis indicates the need for a careful management strategy to prevent or detect the development of CR neoplasia in this group of high-risk patients. For the present this would include yearly surveillance colonoscopy with 4 biopsy specimens taken at 10-cm intervals, and colectomy would be indicated for definite dysplasia of any grade, provided this diagnosis is confirmed by a pathologist with expertise in the interpretation of dysplasia. Upon entering a surveillance program, patients must be educated as to the increased risk associated with the combination of UC and PSC. They must be motivated to undergo surveillance on a yearly basis, as well as to submit to colectomy if dysplasia is detected and confirmed by a pathologist with the requisite expertise. In these patients, colonoscopy surveillance should be performed annually regardless of the duration of UC. Future studies of VOLUME 56, NO. 1, 2002
Colonic neoplasia: patients with primary sclerosing cholangitis, ulcerative colitis
patients who participate in programs of annual surveillance, including colectomy for definite dysplasia, will determine whether patients with PSC and UC actually have an accelerated development of cancer. REFERENCES 1. Broomé U, Lindberg G, Lofberg R. Primary sclerosing cholangitis in ulcerative colitis—a risk factor for the development of dysplasia and DNA aneuploidy? Gastroenterology 1992;102: 1877-80. 2. Lee YM, Kaplan MM. Primary sclerosing cholangitis. N Engl J Med 1995;332:924-33. 3. Ekbom A, Helmick C, Zack M, Adanii HO. Ulcerative colitis and colorectal cancer: a population-based study. N Engl J Med 1990;323:1228-33. 4. Ahnen DJ, McHugh JBD, Arsenault LL, Warren G. Does primary sclerosing cholangitis (PSC) increase the risk of colon cancer in patients with chronic ulcerative colitis (CUC) [abstract]? Gastroenterology 1993;104:A658. 5. Broomi U, Lofberg R, Veress B, Eriksson LS. Primary sclerosing cholangitis and ulcerative colitis: evidence for increased neoplastic potential. Hepatology 1995;22:1404-8. 6. Brentnall TA, Haggitt RC, Rabinovitch PS, Kimmey MB, Bronner MP, Levine DS, et al. Risk and natural history of colonic neoplasia in patients with primary sclerosing cholangitis and ulcerative colitis. Gastroenterology 1996;110:331-8. 7. Bansal P, Sonnenberg A. Risk factors of colorectal cancer in inflammatory bowel disease. Am J Gastroenterol 1996;91:44-8. 8. Shetty K, Rybicki L, Brzezinski A, Carey WD, Lashner BA. The risk for cancer or dysplasia in ulcerative colitis patients with primary sclerosing cholangitis. Am J Gastroenterol 1999;94:1643-9. 9. Marchesa P, Lashner BA, Lavery IC, Milsom J, Hull TL, Strong SA, et al. The risk of cancer and dysplasia among ulcerative colitis patients with primary sclerosing cholangitis. Am J Gastroenterol 1997;92:1285-8. 10. Leidenius MH, Farkkila MA, Karkkainen P, Taskinen EI, Kellokumpu IH, Hockerstedt KA. Colorectal dysplasia and carcinoma in patients with ulcerative colitis and primary sclerosing cholangitis. Scan J Gastroenterol 1997;32:706-11. 11. Timmermanns G, Heuschen UA, Stern J, Herfath CH. Enhanced colorectal cancer risk in patients with ulcerative colitis associated with PSC [abstract]. Gastroenterology 1996; 110:A603. 12. Loftus EV, Sandborn WJ, Tremaine WJ, Mahoney DW, Zinsmeister AR, Offord KP, et al. Risk of colorectal neoplasia in patients with primary sclerosing cholangitis. Gastroenterology 1996;110:432-40. 13. Choi PM, Nugent FW, Rossi RL. Relationship between colorectal neoplasia and primary sclerosing cholangitis in ulcerative colitis. Gastroenterology 1992;103:1707-9. 14. Aadland E, Schrunpf E, Fausa O, Elgjo K, Heilo A, Aakhus T, Gjone E. Primary sclerosing cholangitis: a long-term followup study. Scand J Gastroenterol 1987;22:655-64. 15. Eaden JA, Mayberry JF. Colorectal cancer complicating ulcerative colitis: a review. Am J Gastroenterol 2000;95:2710-9. 16. Lashner BA, Provencher KS, Seidner DL, Knesebeck A, Brzezinski A. The effect of folio acid supplementation on the risk for cancer or dysplasia in ulcerative colitis. Gastroenterology 1997;112:29-32. 17. Ransohoff DF. Colorectal cancer. In: Everhart YE, editor. Digestive diseases in the United States: epidemiology and impact. US Department of Health and Human Services, Public Service, National Institutes of Health, National VOLUME 56, NO. 1, 2002
18. 19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31. 32.
33.
34.
35.
R Soetikno, O Lin, P Heidenreich, et al.
Institute of Diabetes and Digestive and Kidney Diseases. Washington, DC: US Government Printing Office; 1994. NIH publication no. 94-1447. Hedges LV, Olkin I. Statistical methods for meta-analysis. San Diego: Academic Press; 1985. Petitti DB. Meta-analysis decision analysis and cost-effectiveness analysis. Methods for quantitative synthesis in medicine. New York: Oxford University Press; 1994. Higashi H, Yanaga K, Marsh JW, Tzakis A, Kakizoe S, Starzl TE. Development of colon cancer after liver transplantation for primary sclerosing cholangitis associated with ulcerative colitis. Hepatology 1990;11:477-80. Bleday R, Lee E, Jessurun J, Heine J, Wong WD. Increased risk of early colorectal neoplasms after hepatic transplant in patients with inflammatory bowel disease. Dis Colon Rectum 1993;36:908-12. Gurbuz AK, Giardiello FM, Bayless TM. Colorectal neoplasia in patients with ulcerative colitis and primary sclerosing cholangitis. Dis Colon Rectum 1995;38:37-41. D’Haens GR, Lashner BA, Hanauer SB. Pericholangitis and sclerosing cholangitis are risk factors for dysplasia and cancer in ulcerative colitis. Am J Gastroenterol 1993;88:1174-8. Narumi S, Roberts JP, Emond JC, Lake J, Asoher NL. Liver transplantation for sclerosing cholangitis. Hepatology 1995; 22:451-7. Pinczowski D, Ekjbom A. Is there an increased risk of colorectal cancer among ulcerative colitis patients with primary sclerosing cholangitis [abstract]? Gastroenterology 1995;108: A29. Shaked A, Colonna JO, Goldstein L, Busuttil RW. The interrelation between sclerosing cholangitis and ulcerative colitis in patients undergoing liver transplantation. Ann Surg 1992; 215:598-603. Nuako KW, Ahlquist DA, Sandborn WJ, Mahoney DW, Siems DM, Zinsmeister AR. Primary sclerosing cholangitis and colorectal carcinoma in patients with chronic ulcerative colitis: a case-control study. Cancer 1998;82:822-6. Kornfeld D, Ekbom A, Ihre T. Is there an excessive risk for colorectal cancer in patients with ulcerative colitis and concomitant primary sclerosing cholangitis? A population based study. Gut 1997;41:522-5. Kingham JG. Risk of colon cancer in patients with ulcerative colitis is potentiated by the presence of primary sclerosing cholangitis. Hepatology 1997;25:254. Loftus EV Jr, Aguilar HI, Sandborn WJ, Tremaine WJ, Krom RA, Zinsmeister AR, et al. Risk of colorectal neoplasia in patients with primary sclerosing cholangitis and ulcerative colitis following orthotopic liver transplantation. Hepatology 1998;27:685-90. Rosenthal R. The “file drawer problem” and tolerance for null results. Psychol Bull 1979;86:638-41. Stiehl A, Rudolph G, Sauer P, Theilmann L. Biliary secretion of bile acids and lipids in primary sclerosing cholangitis: influence of cholestasis and effect of ursodeoxycholic acid treatment. J Hepatol 1995;23:283-9. Tung BY, Emond MH, Haggitt RC, Bronner MP, Kimmey MB, Kowdley KV, et al. Ursodiol use is associated with lower prevalence of colonic neoplasia in patients with ulcerative colitis and primary sclerosing cholangitis. Ann Intern Med 2001; 134:89-95. Hawk ET, Viner IL. Chemoprevention in ulcerative colitis: narrowing the gap between clinical practice and research. Ann Intern Med 2001;134:158-60. Blackstone MO. An association of fecal bile acids and colon cancer in ulcerative colitis? Gastroenterology 1988;95:575-6. GASTROINTESTINAL ENDOSCOPY
53
R Soetikno, O Lin, P Heidenreich, et al.
Colonic neoplasia: patients with primary sclerosing cholangitis, ulcerative colitis
36. Siegel JH, Barnes S, Morris JS. Bile acids in liver disease associated with inflammatory bowel disease. Digestion 1977; 15:469-81. 37. Fausa O, Schrumpf E, Elgjo K. Relationship of inflammatory bowel disease and primary sclerosing cholangitis. Semin Liver Dis 1991;11:31-9. 38. van Milligen de Wit AW, van Deventer SJ, Tytgat GN. Immunogenetic aspects of primary sclerosing cholangitis: implications for therapeutic strategies. Am J Gastroenterol 1995;90:893-900. 39. Ilyas M, Talbot IC. p53 Expression in ulcerative colitis: a longitudinal study. Gut 1995;37:802-4. 40. Burmer GC, Rabinovitch PS, Haggitt RC, Crispin DA, Brentnall TA, Kolli VR, et al. Neoplastic progression in ulcerative colitis: histology, DNA content, and loss of p53 allele. Gastroenterology 1992;103:1602-10.
54
GASTROINTESTINAL ENDOSCOPY
41. Itzkowitz SH, Young E, DuBois D, Harpaz N, Bodian C, Chen A, et al. Sialosyl-Tn antigen is prevalent and precedes dysplasia in ulcerative colitis: a retrospective case-control study. Gastroenterology 1996;110:694-704. 42. Seibold F, Slametschka D, Gregor M, Weber P. Neutrophil autoantibodies: a genetic marker in primary sclerosing cholangitis and ulcerative colitis. Gastroenterology 1994;107: 532-6. 43. Broomé U, Glaumann H, Hellers G, Nilsson B, Sorstad J, Hultcrantz R. Liver disease in ulcerative colitis: an epidemiological and follow up study in the county of Stockholm. Gut 1994;35:84-9. 44. Lebovics E, Palmer M, Woo J, Schaffner F. Outcome of primary sclerosing cholangitis. Arch Intern Med 1987;147:729-31. 45. Wiesner RH. Current concepts in primary sclerosing cholangitis. Mayo Clin Proc 1994;69:969-82.
VOLUME 56, NO. 1, 2002