Patients with long-standing ulcerative colitis (UC) are at increased risk of developing colorectal cancer1 and thus require colonoscopy for the early detection of precursor lesions. When high-grade intraepithelial neoplasias or early cancers are detected, immediate colectomy is recommended. Other aspects of cancer prevention remain controversial.2 Conflicting observations on the natural history and on the management of low-grade intraepithelial neoplasia have been published.3, 4 Some of these discrepancies may be related to diagnostic problems; there is, as we have observed in a previous study,5 no general consensus on the histological characteristics of low-grade intraepithelial neoplasias.6, 7 Evidence is particularly poor with respect to patients with negative index endoscopy because it is difficult to detect neoplastic precursor lesions by standard colonoscopy. They are frequently missed despite target biopsies of all suspicious lesions and four-quadrant random biopsies.8, 9 When index colonoscopy is negative and a later colonoscopy shows lesions, it is not known whether these lesions have newly developed or had been missed by the index examination. According to current guidelines, colonoscopic surveillance every 1–2 years with random and targeted biopsies is the state of the art,10, 11 but it is conceivable that follow-up intervals could be prolonged with a better detection rate.
Novel endoscopic techniques, fluorescence-guided colonoscopy (FGC),5, 12 confocal laser endomicroscopy,13 chromoendoscopy,14 and narrow band imaging,15 have been claimed to improve the detection rate. However, it is unknown how many neoplastic lesions are still missed by these new techniques. The natural history of neoplastic precursor lesions detected by these novel methods and in particular the fate of patients with negative examinations has not yet been investigated.
After having shown that the detection rate of flat low-grade intraepithelial neoplasia in patients with long-standing extensive UC is improved by FGC,5 we speculated that, as a consequence of better detection, patients with normal FGC can confidently be considered true negatives. In order to test this hypothesis we conducted a prospective long-term study. In this study we also compared the findings in FGC and in the subsequently resected colon and of patients with neoplastic precursor lesions. This should allow assessing the reliability of FGC.
- Top of page
- PATIENTS AND METHODS
In this prospective long-term study, FGC turned out to be reliable for the detection of neoplasias in patients with long-standing colitis.
In the group of patients with an FGC diagnosis of neoplasia, colectomy confirmed the endoscopic diagnosis in all cases, and therefore the positive predictive value of the FGC diagnosis of “neoplasia” was 100%. This is better than previously reported with standard colonoscopy. In all our positive patients, low-grade dysplasia was found at index FGC, and this type of neoplasia is notoriously difficult to diagnose with standard colonoscopy.21, 22 However, despite its good results, FGC had shortcomings. More than one-third of the patients diagnosed with low-grade neoplasia at index FGC had synchronous more severe lesions—high-grade dysplasia or cancer—in their colectomy specimens. This may be due to several reasons: first, complete sampling of all neoplastic areas of the colon mucosa may not be feasible with FGC. Also, some neoplasias may exhibit weak fluorescence or no fluorescence at all. Other authors have hoped to improve the results by using endoscopic inspection of fluorescence instead of analyzing fluorescence curves.12, 23, 24 Their results in detection of neoplastic lesions in UC are controversial and follow-up data have not yet been published.
Theoretically, an optimal endoscopic method for diagnosing neoplasias would identify all neoplastic foci, and it would allow to endoscopically remove all neoplastic foci at minimal risk, leaving the rest of the colon tumor-free. FGC is far from being such an ideal method. However, even if an optimal method would be developed in the future, a considerable degree of diagnostic uncertainty would remain because the histological diagnosis of dysplasia, in particular of low-grade dysplasia, depends, among other things, on subjective impressions. In our study we tried to minimize this problem by blinding two—and in cases of disagreement, three—expert pathologists. In cases of disagreement, a consensus diagnosis was made, i.e., the diagnosis of the two agreeing pathologists was considered correct. Two of our pathologists had previously taken active part in the elaboration of multinational histological criteria for dysplasia but they still disagreed among themselves in some cases. Consensus diagnosis as defined in our study was not more accurate than were individual pathologists, but FGC would have more accurately predicted the findings at colectomy if we had taken the worst diagnosis in each case instead of the consensus diagnosis. This observation may have clinical consequences and should be further tested in future prospective studies.
The invasive growth pattern of the two cancers may have been an additional reason why the lesions were not detected at index FGC and histological diagnoses were conflicting. Although fluorescence was high, the depth of endoscopic biopsies may not have been sufficient to diagnose cancer. In colorectal polyps it has been shown that biopsies are not adequate for grading neoplasia. In 63% of advanced neoplasia only low-grade intraepithelial neoplasia was diagnosed by biopsies.25
The low false-negative rate of FGC is a consequence of the multifocal occurrence of neoplasias in patients with long-standing UC; when at least some out of a multitude of neoplastic areas show strong fluorescence and are thus biopsied, the presence of neoplasia is correctly diagnosed. Along the same line of reasoning, the absence of fluorescence is easily recognizable, and FGC is a good method to exclude neoplasia. Normal FGC findings most likely reflect the absence of neoplasia and are true-negative. This explains the low progression rate in our study: neoplasia developed in only 6% of our initially negative patients and only after more than 7 years of follow-up, 20 years after the onset of colitis. Correspondingly, our patients with neoplasia at index FGC had suffered from colitis for 16 years on average. The higher progression rate reported in studies with standard colonoscopy is most likely due to a high rate of neoplasias missed at index endoscopy and not to a selection of more severe cases.
The progression rate was low in our study despite the fact that surveillance was performed with conventional colonoscopy. We could of course have missed neoplastic lesions, but this seems rather unlikely because no interval cancers were detected. Therefore, one could speculate that an accurate index colonoscopy with whichever method is crucial for a low progression rate and negative findings at surveillance colonoscopy.
The overall result of our approach, index FGC and colectomy in patients with neoplasia, is excellent. All 31 initially neoplasia-free patients who entered the follow-up program and also the eight patients with neoplasia who underwent immediate colectomy survived during the long, on average, 8-year observation period.
It will be of interest to conduct further studies where an entire collective of patients with long-standing colitis, with and without neoplasia, will be randomized to different managing schedules and followed up for prolonged time periods. Long-term data, including comparisons of FGC with other detection methods, are required before changing present surveillance strategies. Finally, larger trial will be needed; the small sample size is a weak point of our study.
In conclusion, FGC is a new method for detection and exclusion of intraepithelial neoplasia in patients with inflammatory bowel disease with a high sensitivity and specificity. The overall outcome of our management approach is excellent. Patients with negative findings at FGC can feel safe, since progression is rare and occurs late. Surveillance intervals might be prolonged for up to 4 years. Patients with multiple low-grade detected neoplasia at FGC should be advised to undergo colectomy.