Axioms
Many accepted practices related to cancer surveillance in ulcerative colitis have not been studied, but are assumed to be valid. Indeed, there are certain axiomatic statements that must be true for surveillance to be at all accepted by patients and physicians.
(1) The cancer risk is elevated in ulcerative colitis patients and is too high to ignore. There have been many epidemiological studies investigating the cancer risk in ulcerative colitis [2–12, 29–33], mostly from northern Europe or North America, where the incidence of ulcerative colitis is high and accurate and complete databases exist. From these studies, it is reasonable to assume that the lifetime incidence of colorectal cancer in a patient with pan-ulcerative colitis is approximately 6%, since a risk of this magnitude has been established for the background risk in the American population [34], and the risk of cancer-related mortality is approximately 3%. These figures are too high to ignore, assuming that there is either effective surveillance available or acceptable prophylactic treatment. Furthermore, in some countries like the USA, either prophylactic colectomy or cancer surveillance colonoscopy have become the standard of care for ulcerative colitis patients, especially those diagnosed at a young age. In older patients, especially those with severe comorbidity or disability, the case for surveillance is much less clear, since they may be more likely to die from other diseases and not be fit for proctocolectomy.
Newer studies report a lower risk of cancer than previously described. In a recent meta-analysis, Jess and colleagues calculate a lower risk of cancer than previously reported (standardised incidence ratio, 1.9; 95% CI: 1.4–5.2) [35]. More recent cohort studies report a risk of cancer less than this [29–31, 33, 36]. This apparent reduction in cancer risk may be due to the protective effect of adherence to a colonoscopic surveillance program in which colectomy is offered to those in whom dysplasia is detected. It may also reflect the use of mesalamine for which there is evidence of a protective effect against colorectal cancer in ulcerative colitis [37].
(2) Most patients would rather not have prophylactic colectomy. Colectomy prior to the development of dysplasia or cancer is sure to dramatically reduce, if not eliminate, the mortality from colorectal cancer [38]. The existence of cancer surveillance programs, whether or not they are effective, has convinced some patients that the excess cancer mortality risk with ulcerative colitis can be minimized, and that the minimized risk is preferred to the morbidity following proctocolectomy.
(3) Patients would agree to proctocolectomy if the cancer risk is very high, as it is with a positive test from surveillance. There is no point in performing surveillance colonoscopy if a patient will refuse to have a proctocolectomy for a positive test. Clinicians need to counsel patients carefully so they understand that surveillance is meant to identify the patients at very high cancer risk for proctocolectomy and allow the remaining patients to continue in the cancer surveillance program. From that approach, a majority of patients, those without dysplasia, will not have a colec-tomy recommended.
In an optimally performing program, all cancer deaths will be averted through colectomy on high risk patients, and no cancer deaths will occur among patients not having colectomy. There has been no perfectly performing surveillance program reported. Program performance is likely to improve following the development of a diagnostic test with better sensitivity and specificity than the presence or absence of dysplasia and/or with more frequent testing than is currently done.
Questions
Existing evidence can only partially answer some of the questions related to cancer surveillance in ulcerative colitis. Understanding the limits of this evidence and identifying priorities for future investigation could improve technical aspects of surveillance and, ultimately, decrease cancer-related mortality. Questions regarding expected outcomes, the method of surveillance, testing intervals and the criterion for a positive test will be addressed.
(1) How effective will a surveillance program be for reducing cancer-related mortality? Although there is no direct evidence for a survival benefit in patients undergoing surveillance, there is evidence that cancers tend to be detected at an earlier stage, and these patients have a correspondingly better prognosis [39]. However lead-time bias may contribute substantially to this apparent benefit.
The number of patients needed to be enrolled in a surveillance program who comply with all of its parameters (i.e. repeated testing with colectomy for a positive test) in order to avert one cancer death can be calculated using expected risk reductions. The number-needed-to-treat (NNT) is the inverse of the absolute risk reduction [21]. Assuming the cancer-related mortality in high risk patients is 3%, the NNT in a perfectly performing program in which colectomy for dysplasia is highly effective for prevention of death from cancer and results in the complete elimination of cancer-related mortality is 1/0.03 or 33. For an absolute risk reduction from surveillance of 1% (i.e. 3% to 2%), the NNT is 100, and for an absolute risk reduction of 2% (i.e. 3% to 1%), the NNT is 50. It is reasonable to assume that surveillance will have some benefit and the NNT most likely will fall between 33 and 100.Therefore, for every 100 patients with pan-ulcerative colitis who are entered into and faithfully comply with the parameters of a surveillance program, between one and three cancer deaths will be averted.
(2) What is the best testing method for cancer surveillance? Using colonoscopy with multiple biopsies of the colon to detect dysplasia is the best and most accepted strategy employed for cancer surveillance in ulcerative colitis. Since dysplasia can be present focally, and not necessarily diffusely, biopsies must be taken throughout the colon. Histological interpretation can be problematic in the presence of inflammation and surveillance endoscopy is best performed when the disease is in remission. It is necessary to take at least 33 biopsies from around the colon to have a 90% sensitivity for the detection of dysplasia [40], but even the most intensive sampling protocols sample less than 0.05% of the colon. While it has not been studied, it seems to be a reasonable trade-off between sensitivity and cost/morbidity to sample the colon with four biopsies taken from each 10 cm colonic segment and of any raised or strictured lesions. C5
Intensive biopsy protocols, in which 33 or more biopsies are taken at each surveillance colonoscopy, have an impact on pathology services. Techniques to target biopsies to those sites likely to be harboring dysplasia would reduce this burden.
Dysplasia can be visualised at white light colonoscopy [41], which is able to detect prominent polypoid lesions. Careful mucosal inspection, even without chromoendos-copy will improve detection of dysplasia. The finding that there is a significant correlation between the mean duration of colonoscopy and detection rate of flat dysplasia seems intuitive [42].
Chromoendoscopy using indigo carmine or methylene blue highlights architectural abnormalities allowing flat or depressed dysplastic lesions to be seen with a related increase in diagnostic yield [43–47]. This was highlighted in the study by Rutter and colleagues in which one hundred patients with ulcerative colitis attending for surveillance colonoscopy underwent back-to-back colonoscopies with indigo carmine dye spraying at the second examination. No dysplasia was found in the 2904 random biopsies taken, whereas dye spraying and targeted biopsies revealed seven dysplastic lesions [48]. Combining chromoendoscopy with high magnification endoscopy allows the surface staining pattern to be analyzed. Kiesslich and colleagues have used this technique to differentiate between neoplastic and non-neoplastic tissue with a sensitivity and specificity of 93% [49].
Chromoendoscopy with targeted biopsies reduces the burden on the histopathologist while increasing the yield of surveillance colonoscopy. There is a trade-off in terms of increased procedural time, with examinations taking approximately 10 minutes longer in expert hands [49, 50]. Colonoscopists will require adequate training in this technique and guidelines exist to aid optimal performance [51].
The addition of endomicroscopy allows in vivo histology to analyse lesions identified by chromoendoscopy, which further increases the yield of intra-epithelial neoplasia as compared to chromoendoscopy alone [52]. The diagnostic yield for intraepithelial neoplasia was improved 4.75-fold using this technique [43].
Other emerging endoscopic technologies may aid in the visualisation of architectural abnormalities and distinguish neoplastic from non-neoplastic lesions (narrow band imaging, fluoroendoscopy, optical coherence tomography and confocal laser microscopy) [43, 53–56], but it remains to be seen to what extent these technologies will be utilized in daily practice.
A dysplasia-associated lesion or mass (DALM) correlates with a high risk for cancer and prophylactic colectomy can be advocated [57, 58]. On colonoscopic appearance, it can be difficult to distinguish a DALM from a sporadic adeno-matous polyp. Although it has been suggested that such lesions may be safely removed provided surrounding flat mucosa shows no dysplasia, recurrent raised dysplastic lesions occur in half these patients [58–61]. If colectomy is deferred, intensive follow-up with 3–6 monthly colonoscopy is advised. C5
Problematical issues involve pseudopolyps and strictures. Pseudopolyps have been associated with double the risk of colorectal cancer [62, 63]. Multiple pseudopolyps cannot be adequately biopsied and could easily harbor dysplastic tissue. These patients need to be informed of the poor sensitivity of surveillance, and the benefits of prophylactic colectomy. C5 Likewise, colonic strictures that do not allow passage of the colonoscope and adequate sampling could, and very often do, harbor dysplasia [64]. Once again, these patients should be considered for prophylactic colectomy since surveillance of these patients is insensitive. C5
Research is progressing in the area of alternatives to histological assessment of dysplasia as a surveillance strategy. Detection of acquired DNA changes in biopsy samples (aneuploidy, p53 and k-ras mutations) or glycosylation abnormalities (sialyl-Tn) may contribute to the sensitivity of surveillance [65]. DNA fingerprinting or fluorescent in situ hybridization has been used to demonstrate genomic instability in rectal biopsies as a marker of neoplasia in inflammatory bowel disease [66]. Analysis of DNA from stool samples allows analysis of DNA from more cells than pinch biopsies alone. Extraction of DNA from faecal samples and subsequent analysis for mutations associated with sporadic colon cancer has been performed [67, 68], and it remains to be seen whether this non-invasive testing strategy will have a place in surveillance in ulcerative colitis.
Patients would be considered to have a positive test if either dysplasia or a genetic abnormality is present. Of course, improved sensitivity will be at the cost of specificity and result in increased numbers of false positives. The penalty for lowering specificity is high – a proctocolectomy in a patient who might not have developed cancer. These alternative tests would be acceptable for use in a surveillance program if the cost were relatively low, the availability high, the gain in sensitivity great and the loss of specificity minimal.
(3) What is the best testing interval? The more tests that are carried out in a lifetime, the higher the likelihood that dysplasia will be detected and treated prior to the development of cancer and that cancer-related mortality will be reduced. Of course, the more tests that are done, the higher will be the cost, morbidity and patient intolerance to colonoscopy. A balance between benefits and costs needs to be struck [69, 70].
While lesions could progress at different rates, the mean value for the time between the development of low grade dysplasia and cancer (lead time) is believed to be three years [71, 72]. Therefore, testing at intervals longer than three years should be discouraged, as the majority of patients who develop cancer would not have had an opportunity for dysplasia to be detected at surveillance examinations.
The risk of developing cancer or dysplasia increases with increasing duration of disease. The benefits of frequent testing (short interval) also increase with increasing duration of disease. It can be concluded that uniform testing intervals over a lifetime of disease is not an efficient way to allocate the performance of costly and invasive test procedures. A decision analysis suggests that efficient testing is characterized by decreasing the testing interval with increasing duration of disease [71]. One reasonable method, which certainly can be adjusted according to patient and physician preferences, specifies testing every 3 years for the first 20 years of disease, every 2 years for the next 10 years of disease, and yearly thereafter. C5 Complications of colonoscopy are rare but can be serious or life-threatening. However, this approach would require at least 20 tests over a 40-year lifetime of disease, with most allocated in the later years when the risk is the highest and with the associated cumulative risk of multiple procedures. The risk of bleeding or perforation is less for diagnostic colonoscopy than for therapeutic colonoscopy (up to 0.35% and 2.3% respectively [73]).
(4) What is the best criterion for a positive test? The type of dysplastic lesion to be used as a criterion for a positive test is best determined by weighing the trade-off between sensitivity and specificity. Sensitivity is defined as the proportion of patients with disease who are positive for the test in question. Likewise, specificity is defined as the proportion of patients without disease who are negative for the test in question. A standard 2×2 contingency table for n patients (n = a + b + c + d) is shown in Table 17.2.
Table 17.2 Contingency table for calculating sensitivity and specificity of dysplasia for the diagnosis of cancer.
| Cancer | No Cancer | |
| Dysplasia | a | b |
| No Dysplasia | c | d |
In normally distributed populations, sensitivity (a / [a + c]) and specificity (d / [b + d]) are stable values that do not vary with prevalence of disease. Sensitivity and specificity will vary though, when the “cut-point” or the criterion for a positive test changes. For example, if the criterion for a positive test changes from high grade dysplasia to low grade dysplasia, the sensitivity will increase (more “a”s and less “c”s) and the specificity will decrease (more “b”s and less “d”s). As the criterion for a positive test changes, there is a trade-off between sensitivity and specificity – as one increases, the other decreases.
Related posts:
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
