Challenges in Designing a Randomized Trial of Surveillance Colonoscopy in IBD

Ulcerative colitis is associated with an increased risk of developing colorectal cancer. The risk of colorectal cancer is estimated to be ˜7-14% at 25 years of disease (1,–3). As compared with the general population, the standardized incidence ratio among a Swedish population-based cohort of patients with ulcerative colitis was 5.7 (95% CI, 4.6-7.0) (4). Because much effort and money has been expended in search of the optimal approach to screen for colorectal cancer in the general population (5), it stands to reason that a colorectal cancer screening approach would be of critical importance in ulcerative colitis. Furthermore, as the colon is an organ that is easily accessible for examination, there has been great hope that dysplasia-surveillance colonoscopy would be an ideal screening tool ever since a panel of experts readdressed the histologic approach to dysplasia diagnosis (6). Many inflammatory bowel disease referral centers have been carrying out dysplasia-surveillance colonoscopy programs and recording their experience prospectively. Remarkably, for a procedure that entails some discomfort, considerable direct and indirect costs, and even risks (albeit uncommon), a randomized controlled study proving the benefit of dysplasia-surveillance colonoscopy has never been undertaken. Experts often sadly and resignedly report that this type of study is unlikely ever to be performed because of general concerns about the high risks of colorectal cancer in this group of patients. “How can we not do surveillance?” Eddy eloquently crystallized the issue regarding the pursuit of cancer screening: “…when we encounter a technology, such as mammography, that has both biological and social appeal, should we begin with a presumption that it is effective and appropriate and then proceed to promote it until evidence proves otherwise? Or should we make no initial presumption of benefit and resist promoting the technology until the evidence shows it to be effective?” (7). The issues regarding cancer screening in ulcerative colitis can be further expanded. Is dysplasia-surveillance colonoscopy so obviously successful that it necessitates that we forego the optimal form of evidence-based medicine (the randomized controlled trial)? And is a randomized controlled study of dysplasia surveillance so difficult to perform that it is simply not doable? Alternatively, how best might a randomized trial of such a screening intervention be performed? What do we want a colorectal cancer screening program to achieve in ulcerative colitis? First, the diagnosis of neoplasia or preneoplasia should be found before a state of frank cancer (or at least invasive cancer) is present. It would be expected that earlier diagnosis would limit cancer-associated morbidity and mortality. Second, the screening tool should be minimally invasive and have minimal risk, be widely available and easy to standardize, and be relatively inexpensive. The mortality rate from cervical cancer has decreased ˜70% over the past 40 years (8). Although this is widely accepted as being a result of widespread populationbased cytologic screening (Pap testing), this has never been proven in a randomized controlled trial. In the face of this view of cervical cancer screening, one might consider it arrogant to expect such proof for a colorectal cancer screening technique. However, it is critical to recognize that we do not yet have data establishing that we have in any meaningful way affected colorectal cancer incidence or outcome in patients with ulcerative colitis. Based on mathematical models and large databases of cervical cancer rates in widely screened and generally unscreened populations, screening intervals and age at screening for cervical cancer can be reliably established even in the absence of controlled studies (9,10). Furthermore, the Pap smear entails much less invasiveness, preparation, and cost. Nonetheless, it has been estimated that 80-90% of treatments have not been adequately evaluated with controlled trials (11). This should be kept in perspective before we disregard dysplasia-surveillance colonoscopy in ulcerative colitis, despite the absence of solid prospective controlled data (12,13). In contrast to other screening strategies such as mammography, which improves survival by detecting malignant disease at an early stage, the Pap is a central part of a strategy designed to prevent the development of invasive cancer. Dysplasia surveillance is designed to perform like Pap screening but in reality, and disappointingly, performs much like mammography in many instances in which malignant disease is already present at a positive screen. The other advantage to cervical cancer screening techniques over dysplasia surveillance in ulcerative colitis is that there is a graded approach to the search for and finding of cervical intraepithelial neoplasia. The process begins with the Pap screen and, based on results, progresses to colposcopy for abnormal Pap tests, then cryotherapy, laser vaporization, or loop electric excision procedure or a combination of these for abnormal colposcopy and biopsy results. Ultimately, the cone biopsy, which is a more aggressive approach, could certainly be argued to be more appealing than a colectomy (particularly if the colectomy is accompanied by an ileostomy) with a far lesser impact on an individual's daily activities. Finally, the playing field is so much larger and more confusing [e.g., finding a dysplasia-associated lesion or mass (DALM) in a pseudopolyp “minefield”] in ulcerative colitis than that in an analogous chronic inflammatory and premalignant condition, Barrett's esophagus. The gastroenterology community has yet to undertake a randomized controlled study of dysplasia surveillance for this condition. This, despite the much greater access to these patients due to the much greater prevalence of Barrett's esophagus in the general population. This does not justify the lack of controlled data in ulcerative colitis but may underscore the inertia among us at pursuing these type of studies. A synthesis analysis of the literature, including studies in which surveillance colonoscopy was performed on more than one occasion for the majority of patients, was published in early 1994 (14), and two later studies corroborated the findings of this literature review (15,16). At initial endoscopy for dysplasia/cancer surveillance, ˜10% of subjects will have dysplasia or cancer found (14,15). This removes a large burden of the ultimate load of patients in any one practice who may ultimately develop neoplasia. If the initial endoscopy is negative for dysplasia, then only ˜3% of subjects will ultimately develop neoplasia when followed up over time (14,15). Unfortunately, the literature reports that if low-grade dysplasia is the indication for colectomy, a cancer may already be present in 19% of cases (14). Subjects with low-grade dysplasia who undergo further surveillance endoscopy have a high likelihood of ultimately progressing to a highly neoplastic lesion (high-grade dysplasia, DALM, or cancer) in 16% (14), 35% (16), or ≤54% (15) of the time. If high-grade dysplasia or DALM is the indication for colectomy, a cancer may already be present in 40-45% of cases, and of these cancers, 50% are Dukes' stages C or D (14). Others have estimated an even higher rate of cancer progression from these lesions (15). These data underscore that finding dysplasia may be too late. The high association of concurrent cancer in colons containing high-grade dysplasia diagnosed at endoscopic surveillance is not widely known. Only 54% of gastroenterologists responding to a questionnaire regarding dysplasia surveillance suspected that >20% of cases of high-grade dysplasia would be associated with concurrent cancer (17). For indefinite dysplasia, 20-30% of subjects will ultimately have an ominous lesion found (high-grade dysplasia, DALM, or cancer; 14), and thus subjects with this diagnosis must be rigorously followed up in search of the truth regarding their indefinite status. Although it can be reassuring that if the initial dysplasia-surveillance endoscopy is negative, the risk of progression to dysplasia or cancer when followed up over time may be as low as 3%, special circumstances must be regarded. First, there is a paucity of hard data in the literature specific to patients in the higher risk groups (disease duration >25 years). Because ˜10% of patients have dysplasia or cancer at their first surveillance colonoscopy (14,15), this implies that the neoplasm has been evolving before the typical surveillance initiation date of ≥8 years. Furthermore, it has been reported that some patients were seen with cancer within 1-2 years of their last negative dysplasia-surveillance colonoscopy (15). Thus for some patients, their neoplasm progression does not follow as slow a progression as we typically expound. In ˜25% of colorectal cancers in ulcerative colitis, dysplasia was not found in the colcctomy-resection specimens (15,18,19). Although the true sensitivity of the process is unknown, these caveats underscore that it is simplistic to view the ulcerative colitis to cancer progression as one that only begins after 8-10 years of disease; that it is widespread enough in flat mucosa that random biopsies will likely find it; and finally, that dysplasia evolving to cancer is always slow enough that annual or biennial are These caveats the of a controlled study to the of dysplasia if such a study were it should the risk cases and not to the for Unfortunately, there is no approach to the process as for other screening techniques in mammography, for is associated with relatively In a questionnaire study to gastroenterologists in the and as as to gastroenterology from the we to their techniques (17). found that of two to 54% perform biopsies at to biopsies from in the colon (17). However, some one biopsy and some perform biopsies at only one to (17). estimated that biopsies are to have a of finding the of neoplasia in a colon containing dysplasia or cancer. However, biopsies are to have a of finding cancer or dysplasia in a who has a cancer present This of biopsies would be if the was to screen only until an cancer was present or the was to do a colectomy for any of has a to dysplasia A of studies have intervals For some patients with ulcerative colitis, this may only approach by have from to to to to simply reported biopsies two biopsies two to biopsies up to biopsies or even to biopsies It should be that biopsies from the are on the of the increased prevalence of colorectal cancer in the in ulcerative colitis may be for as to how many biopsies to and surveillance biopsies is and on the part of the Thus if endoscopic surveillance is to be will have to recognize that more is it has been estimated that a of surveillance biopsies only of the of the colon biopsies for dysplasia typically to their The literature us that among for dysplasia in ulcerative colitis can and The is at high-grade dysplasia low-grade or dysplasia but is far from even at high The issue then is not that we to be in or but that we may simply be too much from general that two on a diagnosis of dysplasia is only of for a (17). and cancer screening in patients with ulcerative colitis on a that more about their risk of of and their benefit of mammography screening than do data are not available about patients with ulcerative this is generally a disease of If this cohort of patients their cancer this may for pursuing dysplasia surveillance for This would of on the as there can be the of the screen the or true dysplasia that to the but is found to have no cancer at The mammography literature reports that can the or daily of for after being do not have cancer in cancer surveillance, of may the with some of over an These generally patients with colitis may to have some of a of the of dysplasia surveillance (14), the patients may some of and from a dysplasia-surveillance For many of these patients who are the surveillance colonoscopy an to be of their disease status. if this is a report of or this can further The dysplasia-surveillance colonoscopy may have analogous as does the The of may be a for the The of may with the of the of in the a certainly promoting in the of a chronic such as ulcerative colitis The regarding the of dysplasia surveillance in ulcerative colitis have been The patients to be with these so that can in the and who about cancer screening and no if cancer was found and not to be screened at compared with a group of who not view the Others have the that patients are the best of how would the risks and of available Thus patients with about program can their and This will ultimately limit and costs, should and if will It is easy to with because we have yet to the The of the procedure should be in the and might and thus to be more and less For in a dysplasia-surveillance colonoscopy would the in In the rates on the but even by the the would for this the not of dysplasia surveillance colectomy no intervention by a This approach on the best available are at a in such a analysis in ulcerative colitis as compared for of screening mammography, in which controlled studies are available In we are not for a of on some of the data regarding in inflammatory bowel neoplasia were reported in this and may of inflammatory bowel colorectal cancer For the of in colorectal cancer may ultimately and but based on the data to this will not dysplasia-surveillance colonoscopy The on may an in The data in ulcerative neoplasia that this may the of progression in some patients with ulcerative colitis but will not patients at risk may to dysplasia of these may of a that is present in and may be an in cases that are for dysplasia is a that is in colon mucosa but in of colon cancers The finding of has been to be highly in a surveillance program of ulcerative colitis in cases who ultimately colon cancer Unfortunately, was positive in of cases, the of the In a was more in surveillance biopsies from cases who ultimately dysplasia compared with who never dysplasia of this study that was positive in of dysplasia cases and in of the This may some the of and the ultimate development of cancer and may as a in cases of dysplasia in the of patients at risk for despite an absence of of these could for of patients with ulcerative colitis with to no risk and with high is to be a of and groups have evaluated the of on random colon biopsies in patients dysplasia-surveillance studies not find However, found a and dysplasia A prospective study of 25 patients dysplasia with of to dysplasia in not to or dysplasia within years It was that biopsies would be for that an population of would be was in of patients with dysplasia or cancer so it too is not a even in This group in cases with dysplasia or and one that was for dysplasia and found that was not so specific a for dysplasia or cancer as the of of for In a later with of two colectomy specimens for and this group found more of than and a greater of with and dysplasia Because in that were indefinite or negative for dysplasia, these that may be an earlier in the cancer than These data have been both corroborated and These that is likely not available in other centers would to be standardized for of these on a random biopsy but by cancer risk that earlier than dysplasia, may limit the likelihood of the cancers that are seen are seen after negative surveillance or in cases of not obviously associated with This would be true if the finding of these were associated with in surveillance or even ultimately greater for and may ultimately be to greater of cancer risk, and could ultimately be established in such that not will the In no are widely available that could dysplasia surveillance at this time. at the reported that a group of patients for dysplasia surveillance cancer compared with a group who not have This is an because the dysplasia-surveillance group two that follow up and the dysplasia compared with for the cohort of patients who were first seen with their patients at can on and in the make and has proven to be effective The dysplasia-surveillance from the In this patients the surveillance program and were seen at a later time with colorectal of the patients in surveillance, cancers, of which the of the and of two were These data are of an increased colorectal cancer rate among patients who However, this direct has the lack of appropriate group as in the Thus at the literature is only that dysplasia surveillance is than no generally have a high of for DALM and when these lesions are the majority colectomy (17). It effort to perform a biopsy of a mass lesion in flat but a effort to perform a biopsy of mass lesions in the of In mass lesions when are present is often simply not Thus the colon with a if dysplasia surveillance is an endoscopic of a with ulcerative colitis who has of a a colon and As there is no evidence that we can reliably the pseudopolyp from we have to consider that is in any of (or of findings in a with ulcerative colitis. of the of the in the of in the is in to the finding of an A in an of colitis is a The advantage of the DALM is that it among that lesions in ulcerative colitis may invasive cancers and not simply issue to consider is any no how should a a (or in a with ulcerative colitis be in the way as if it in a we expect that of the general population have A of cases reported in the literature that if a has an and mucosa that is of dysplasia, then it should be In the dysplasia-surveillance found patients with (15). In patients, the flat mucosa dysplasia, and In patients dysplasia in the no cancers of years. In the patients not of these of invasive and after two or no neoplastic lesions were found at patients not associated with dysplasia and no neoplasia was found at endoscopic at and years or at colectomy for disease in the other patients. Finally, one with an dysplasia in the colon and cases reported are not so regarding the finding of The experience two patients initial neoplastic lesions were not At some dysplasia was found in one of these patients, and was found in the A low-grade dysplasia at the initial surveillance followed by negative within and then at was found to have an Ultimately, this colectomy after this and two of invasive cancer were present. The experience patients who with of cancer However, one of these patients a not to to and it is not from the of the how many of were within of the cases only one and in two of Thus the data from these two can not further for the of in ulcerative colitis. What then should be the approach to colon in ulcerative colitis? If a has a in the from the of colitis, it is more likely that this is a to the and to and not If a has a in any of the including in the colitis it is to and not colectomy, that the is These might be if the is than 40 when the of any is At the the of two lesions for and ulcerative colitis, for that may the at high risk of colon cancer should any with the regarding surveillance Finally, for the in the colitis there is for but to date it would be to this lesion as a DALM and Alternatively, in a than years may have of the colitis as an one can consider with This approach is only if at least one surveillance endoscopy no other evidence of In the population-based in the literature, there is a risk of for colorectal cancer in disease (95% CI, and for with colitis the risk was (95% CI, other databases reported risks of colorectal cancer among patients with disease from to It is critical in colitis for to their data based on patients who have or for colitis will have or their risk of colorectal cancer. the data for patients with disease and colons to a reported risk of colorectal cancer of (95% CI, evidence the cancer from ulcerative cancers compared with cancers the increased risk by In a large on colorectal cancer in both ulcerative colitis and the age at developing the the duration of the and of the cancers, the of dysplasia, and the were A group reported no increased risk of colorectal cancer among their group of patients with However, this was a population in which the of years after diagnosis was these only patients with years of the risk of colorectal cancer was (95% CI, It from these studies that colitis is a risk in disease for developing colorectal cancer duration is less a risk some of the data that early age at diagnosis of colitis may the risk of cancer development The development of cancer in colon the of endoscopic surveillance difficult and in some instances Thus the of a prospective controlled study of cancer surveillance in disease is even less than that for ulcerative colitis. This does not that a controlled study should not be cases with colorectal would have to be and would have to for a study might best begin at the time of colitis because some have early development of colon cancer years from disease in as many as of cases In colitis should the concerns regarding cancer risk as does ulcerative colitis. Thus who dysplasia surveillance in ulcerative colitis should likely a surveillance program in disease controlled data proving the of dysplasia surveillance in ulcerative colitis may not be to the of such a study to disease because of the specific to The risks of colorectal cancer begin to at 8-10 years of and after years of disease It would be to a increased based on years of disease many patients already colorectal neoplasia at their first screening who are diagnosed with ulcerative colitis after the age of 40 years to be at increased risk the duration of their including at of years (4). In the population-based data to the risk of colorectal cancer was increased among patients with age at and duration of disease (4). For subjects diagnosed before age with who years of disease an risk of Alternatively, subjects diagnosed after age with colitis not have increased risks of developing cancer. groups reported an increased risk of colorectal cancer among patients who of It is or not patients with other increased or in there are other of that colorectal cancer risk Nonetheless, in based on a prospective screening it would appropriate to the analysis based on age at of and concurrent diagnosis of This a of to such a study and the but is because it is enough to approach to patients based on these Finally, it is that screening be performed in to colons at patients with endoscopic might more for the of optimal biopsies from the surveillance that to a 50% in colorectal cancer mortality over a patients would be in of a cohort This is based on the of an incidence of colorectal cancer of over years and a 50% mortality This analysis no to and a over years. Alternatively, it would appropriate to the group for dysplasia If patients are who have years of the expected incidence of cancer at years of disease who would have already cancer) might be as high as a survival by a to the of subjects study in a over years and a a 50% mortality from the colorectal cancer and for a 50% in mortality with the patients would be study no to and patients study a to performed an analysis by over and this increased the study to As part of population-based of we have from subjects with ulcerative colitis of the of with ulcerative these reported years of If population is to in and inflammatory bowel disease referral should be to patients over a to at this centers would to be The analysis would be performed on an to who ultimately as as based on in It is the outcome for but for highly to in such a program are the majority of patients with ulcerative to in the study would be randomized to one of two group would undergo surveillance colonoscopy The group would undergo annual at a and both groups would undergo The group would undergo colonoscopy only as Although one might that it would be difficult to patients to the risk of not being that when patients are with the more often forego screening Thus the would to be to the to dysplasia surveillance including the A would be to is no in the if the is to At least should be for biopsy with at least biopsies the and an increased of biopsies biopsies should be because these are with a higher incidence of colorectal cancer in ulcerative colitis. At at which a biopsy is an biopsy would be A of this would be so as to other It is that at a further biopsy is and for to be performed at one biopsies would be by the but would be to a central two at dysplasia diagnosis in the colon would review This would a direct and approach to the surveillance would be patients with dysplasia of any would undergo patients with indefinite dysplasia would undergo increased if and colonoscopy within this approach, it would not be to the issue regarding dysplasia surveillance in but it could the to the group at the gastroenterology and inflammatory bowel disease a study that not some of endoscopic Nonetheless, this type of study would be and and, is only approach could a with a study that is only years. This would some controlled data on surveillance intervals some of more surveillance programs in