Carcinogenesis in ulcerative colitis : Is surveillance worthwhile ?

Observations that dysplastac epithelium preJicts neoplastic change in the coli tic bowel have led to the adoption of surveillance policies in ulcerative colitis parients, usually involving colonoscopy irrespective of symptomatology. Few satisfactory prospective studies of surveillance have been carried out. The present data suggest that patients should be screened at 10 years an<l investigated when symptomatic. Annual clinical review with five-yearly colonoscopic examination may be more reasonable, lead to better compliance, detect the occasional early cancer, and reduce the colonoscopic load. Can J Gastroenterol 1990;4(7):397-399

T l IE ASSOCIATION BETWEEN Ul- cerative colitis and colonic cancer has been recognized for many years (I).
In 1967 Morson anJ Pang (2) showed that the presence of <lysplastic epithelium in the rectum was a predictor of neoplastic change elsewhere in the colitic bowel.These observations have led to the aJoption of surveillance policies in pattents with ulcerative colitis.Those at risk are patients who have had an extensive attack of ulcerative colitis, the period of risk beginning IO years after the initial attack.Many gastro-enterolog1sts now perform regu lar colonoscop1es in this group (3).The presence of either high grade dysplasia or a <lysplasia-associated mass or lesion ( 4) are taken as indications for surgery.

CANCER RISK AND SURVEILLANCE STUDIES
The cumulative cancer risk in patients with extensive colitis is approximately 7% at 20 years anJ 17% at 30 (5).A number of surveillance  studies have confirmed an increased incidence of cancer, and have concluded that surveillance may be an effective means of reducing mortality in ulcerative colitis-associated cancer (4,6,11).This review of the literature does not support this view, and if the cost of surveillance, both financial and human, are taken in to consideration, their benefit is subject t0 doubt.
Most studies have been retrospective with incomplete follow-up.
They have included patients referred tO tertiary centres with new symptoms and individuals with cancers diagnosed using techniques other than the surveillance method (including those in whom the diagnosis was made at operation for continuous symptoms).In these programs a number of patients had advanced malignant disease at the time of diagnosis.

STUDY DESIGN
In order to answer the question 'does surveillance work?', the term 'survei llance' must first be defined.Patients newly referred to gastroenterologists with symptomatic ulcerative colitis are often colonoscoped to assess disease activity and extent; others are colonoscoped Juring their period of illness because of new symptoms or because they have been referred from other centres; and some are colonoscoped who surface after many years of not being followed up.These patients are often assessed or screened by colonoscopy, but this does not constitute 'surveillance'.Surveillance is an ongoing commitment to a program in which patients are fo llowed up irrespective of symptomatology with a view to early detection of a malignant or premalignant lesion.Studies should be prospective and the survei llance technique to be tested agreed to at the outset.
There must be a policy agreed to on the management of dysplasia.In a number of studies, patients appear to have been operated on at random, sometimes with dysplasia being taken into account; a t other times it is ignored.
There must be an agreed to endpoint which should include some definition of success or failu re.For example the discovery of a Duke's stage C lesion probably represent!:,failure of the pro-gram, anJ an A or B success only if it is identified anJ treated by following the surveillance protocol.
Finally, the study should be carried out on an 'intention to survey' has,s analogous to the 'intention to treat' protocols used in Jrug triab.This means that patient:.who have defau lted from the trial and then Jevelop cancer must be counted as failures.This latter point is of some importance because authors have drawn aLtcntion to an apparently increased incidence of cancer in defaulters; thb could be construed as evidence in favour of !:,Urveillance on the grounds that had they been properly followed up their neoplastic dise<1se wou ld have heen diagnosed earlier.ln fact it represents fai lure of the surveillance program.

PUBLISHED SURVEILLANCE SERIES
Published series are I istcd in Table I.It can he seen thaL the avernge numher of colonoscopies per patient ranges from 0.92 to 3.3.If it is accepted that surveillance implies regular annual colonoscopy, and Lhat the initial colono, scopy is screening not :.urvcillance, then few of these centres have in practice instituted a surveillance program.
Table 2 shows the amalgamated data separating colonic cancer into Duke's stages and the circumstances by which cancer was diagnosed: 39% had Duke's stage C or worse, while 35% and 26% had stages B and A, respectively.These figures underline t he inadequacy of current techniques to diagnose cancer in patients known to be at risk, who were seen in first class gastroenterology units where modern investigational techniques were available.
Only 24% of carcinomas were de, tected as a result of the surveillance program itself, assuming colonoscopy to be the accepted means of surveil, lance, and only six of 49 cancers ( 12%) were both early (A or B) and dis, covered by survei llance.

POLICY ON MANAGEMENT OF DYSPLASIA
The management of dysplasia remains controversial and is made more difficult by the interpretive problems reason in itself for surgery because the incidence of cancer in this group is low.
In the majo rity of stud ies colectomy has been advised only on the basis of high grade dysplasia, a foca l les ion assoc iated with dysplasia or cancer itself.In some cases patients we re operated for low grade dysplas ia.In some it may have been because of the length of history , in oth ers symptoms, in ochers, possibly, clinical acumen.If, however, low grade dysplasia is nor regarded as a reason for universal total colcctomy patients found co have cancer as a result of operations for low grade dysplasia sho uld no t be regarded as successes of a surveillance program.If t h is is don e it can be seen that only 20% of cancers were found as a direct result of surve illance.Even chis 1s probably an overestimate because 1t h as been assumed that in all 10 patients coleccomy was performed as a result of the surveillance procedure alone rather tha n debility; this was not specifically stated in all patients.
'COST' OF SURVEILLANCE Cancer surveillance In ulcerative colitis regular surveilla nce program may be reassuring to some, bu t to others it 1s a constant reminder that they harbo ur a premalignant lesion.These com1derations lead to a high level of defaulting.Finally, the financial costs of surveillance studies must he taken into cons1deratton .Accordi ng ro the available data it appea rs that the cost effecti veness of surveillance programs in ulcerative colitis is low.

CONCLUSIONS
T here is at present no answer to the question "Is surveilla nce in ulcera tt vc colitis worthwhile?"In ~pite of evide nce that extensive ulcerative colitis carncs an increased risk of carcinogenesis, there have been few satisfaccory prospective studies.Wha t data there are suggest chat patients should be screened at IO years and investigated whe n symptomatic.Effective regular annual sc reening by colonoscopy 1s not being undertaken in many centres, and where n is, few early cancers arc detected.It may be mo re reasonable to consider annual clmical review with perhaps five-yearly colonoscopic examination.This policy might lead to better compli,mcc, detect occasional earl y cancers, and reduce the colonoscopic load.noma m ulceranvc colm~.Gastroenterolobry 1985;89: 1342-6.9. Brostri>m 0, Llilberg R, 0,1

TABLE 1
Ulcerative colitis surveillance studies

TABLE 3
Outcome o f 66 patients with dysploslo in ulcerative colitis