Statistics from Altmetric.com
We welcome the joint work of the British Society of Gastroenterology and the Association of Coloproctology of Great Britain and Ireland in commissioning guidelines for colorectal cancer screening in high risk groups (Gut 2002;51(suppl 5):V13–14). In the absence of direct evidence from randomised trials for most of the groups, the various authors have balanced a wealth of recent genetic and epidemiological evidence estimating an individual’s levels of elevated risk against the risks associated with screening. The end result is the recognition that within the label “high risk” there is a spectrum of risks such that colonoscopic screening and surveillance must be tailored accordingly. This avoids the ineffective, costly, and potentially harmful “blanket-type” approach, which formerly prevailed.
However, within the guideline series, there is one exception—screening and surveillance in patients with acromegaly. We and other researchers1–3 have repeatedly stated that the studies undertaken by the authors of these guidelines have overestimated the risk of colorectal cancer in this patient group. They report a 13–14-fold increase in risk based on colorectal cancer detection rates among acromegalics undergoing colonoscopy at Barts compared with cancer rates from published series of colonoscopic screening in non-acromegalic subjects (see table 3 in Jenkins and Besser4). All of these “control” studies are in mixed race US populations and lack data to permit age-sex comparisons. This simply does not rank as a well designed controlled study and the recommendations should not receive grade B status.
In the same manner as relative risks for those with relatives with colorectal cancer are estimated from population based data (summarised in Dunlop5), we have argued for a similar approach for patients with acromegaly.1 Based on three population based studies, we calculated (by fixed effects meta-analysis) a relative risk of 2.0 (95% confidence intervals 1.4–2.9) for colon and rectal cancer combined in acromegalics.1,2 Considered in terms of absolute risk, with an approximate 2% cumulative risk of colorectal cancer by age 70 years in the general UK population, the estimated risk in acromegalics would be 4%. Notably, the incidence in the Barts series is 4.5% (10 of 222)—not a quarter if the estimate of a 13-fold increase was applied. Acromegalic patients thus have a modest increase in colorectal cancer risk, not a high risk. In first degrees relatives with a strong family history (a high risk group), Dunlop5 recommends early colonoscopic screening and then wait until 55 years for repeat colonoscopy if initial screening is clear (the majority). How can Jenkins and Fairclough justify early screening and colonoscopy five yearly thereafter in all of their acromegalics?
Atkin and Saunders6 have demonstrated that colonoscopic surveillance following initial screening in the non-acromegalic population is determined primarily by clinicopathological findings. These basic guidelines must also be applied to the acromegalic patient population. Jenkins and Fairclough have stated that elevated serum IGF-I levels may predict for recurrent adenomas in acromegalic patients and “should be offered screening at three year intervals”. This is based on data from only eight acromegalic patients with recurrent adenomas7 and should not replace other well recognised predictors of recurrence.
Looked at in the context of (other high risk) groups, the guidelines for colorectal cancer screening in patients with acromegaly are inconsistent. The aggressive approach to colonoscopic screening recommended by Jenkins and Fairclough should be seriously questioned.
We thank Dr Renehan and colleagues for their comments on our data, which they have also made previously. We do not claim that our data are perfect in all respects but it seems to us, on the basis of the data we have collected in our own series and that of other groups, that patients with acromegaly should be regarded as having a significantly increased risk of colorectal neoplasia. The two contrary studies referred to by Renehan et al are his own and that of one other author who relied upon retrospective data acquired more than 50 years ago. These data and those from the population based studies preferred by Renehan suffer from flaws of their own. The morbidity associated with acromegaly has changed in the last 25 years, probably related to the increased survival associated with aggressive and effective treatment of the cardiovascular and metabolic complications of the disease. Our data and those of others show that the prevalence of colonic neoplasia in acromegaly is age dependent. Thus it is only now that patients are surviving long enough to develop this complication, and valid comparative data must therefore be acquired contemporaneously, to take account of the changing pattern of morbidity associated with increased longevity.
We are aware of at least 12 other prospective studies evaluating colonoscopic screening in acromegaly, in addition to our original report from St Bartholomew’s Hospital1. These include one by Renehan et al in which they reported three asymptomatic patients in whom a cancer was detected. Among such studies the optimum comparison must be simultaneous screening of asymptomatic non-acromegalic subjects. A combined comparison of the data from all series using these control groups, none of which involved mixed race US populations, gives a relative risk of colon cancer in acromegaly of 13.4. We think it is prudent to accept the evidence of an increased risk of colon cancer, derived from these clinical observations rather than from theoretical calculations, and to screen acromegalic patients systematically until the current hypothesis is confirmed or refuted. The rarity of acromegaly means that the increase in workload for the majority of individual endoscopy units is likely to be minimal.