Abstract
Screening for colorectal cancer (CRC) holds unusual position in the current public health policy of many countries. Reasons for that are numerous. Firstly, CRC is a serious epidemiological problem in aging societies. Cancer occurs in both genders unlike in other cancers for which screening programmes are accepted. Secondly, it has not been yet decided universally which of the screening methods is optimal. There are several methods available, which have been studied and introduced into practice in different parts of the world; but it seems everyone is doing different thing. The plethora of methodological opportunities is striking. On one side of the spectrum there is a faecal occult blood testing using old guaiac method repeated as often as every year, with colonoscopy for positive stool test. On the other side of the spectrum, a once-in-a-lifetime primary screening colonoscopy is suggested as a possible option. 1 Good news is that despite diverse methodology of CRC screening, it is not only cost-effective but also cost-saving – an exceptional feature compared with other screenings. 2
It appeared over the last few years, that public health policy makers have currently the choice between the two main options. One is the faecal testing using different methods requiring usually frequent repetitions. The other is infrequent endoscopic screening using sigmoidoscopy (followed by colonoscopy for polyp bearers in the distal colon) or colonoscopy (every 10 years or once-in-a-lifetime).
The recent high-quality randomized controlled trials performed in the UK, Italy, and the USA3–5 have shown surprising and unforeseen effect: a SINGLE sigmoidoscopy examination provides a very long-lasting effect, up to 11–13 years, with regard to CRC mortality and incidence reduction. The curves showing this effect comparing CRC mortality for intervention group and control group are diverging still at the moment of 11–13 years, suggesting that the effect may be even much longer than only that period of time.
In this issue, the study by Geir Hoff et al. 6 started in Norway 30 years ago provides data with as long as 26 years of follow up. This is the longest follow-up period available in the literature concerning CRC-screening randomized controlled trials and is an important piece of information providing insight into the effectiveness of endoscopic screening for CRC. Unfortunately, the number of patients recruited was not sufficient to provide statistical significance in all analyses and also probably for ethical reasons it was not possible to avoid colonoscopy in patients of control group 13 years after initial randomization. But even despite that, the 60% reduction of CRC incidence in the intervention group after 26 years of follow up is impressive. The mortality reduction (84%) sounds even better, but the lack of statistical significance slightly disappoints. The intervention in this study over the period of 26 years sounds a bit complex, but in fact it was nearly in full accordance with current knowledge and recommendations. It consisted of: (a) initial sigmoidoscopy with clearing colonoscopy for polyp bearers in the distal colon; (b) two surveillance colonoscopies for polyp bearers only, 2 and 6 years after initial examination; and (c) colonoscopy 13 years after initial examination, with subsequent surveillance colonoscopies only for patients with advanced findings either 18 years or 23 years after initial sigmoidoscopy.
The study provides other notable observations that may shed light on future design of screening programmes. In the screening group, the authors observed a sustained reduction in CRC incidence for the first 20 years, with a starting raise in the incidence in the last 6 years of follow-up. It has been courageously interpreted as an effect of a less stringent surveillance protocol following polypectomy in the second half of the follow-up period (time period II). Although this is one of a plausible explanations, it is rather poorly documented in this study. First, only two out of the seven cancers in the screening group were diagnosed in screening attendees during follow-up period (one cancer was diagnosed at first screening examination and other four cancers in non-attendees). Second, before time period II, both screening and control groups were offered screening colonoscopy, with comparable attendance rates. Therefore, we would expect that the relative reduction in CRC incidence was attenuated in time period II, but in fact it was maintained (hazard ratios: period I: 0.30, 95% CI 0.06–1.44; period II: 0.43, 95% CI 0.15–1.22). So, it not only suggests that the effect on incidence in the screening group was maintained with less stringent surveillance protocol, but also that the effect of screening colonoscopy at a mean age of 66.9 years could have been less potent that the effect of flexible sigmoidoscopy with subsequent surveillance at a mean age of 54.5 years. Despite offering screening colonoscopy in both groups at the mean age of 66.9 years, the cumulative hazard ratios for CRC continue to diverge as if there was no colonoscopy in the control group. Although a microsimulation modelling study has shown that beginning screening at age 50 was consistently better that at age 60, 7 the relatively poor effect of screening colonoscopy at age 66.9 is really unexpected and striking. We have no good explanation for this phenomenon and think that it requires immediate further studies.
All previous screening sigmoidoscopy randomized controlled trials, together with Hoff’s study here, speak for the unique value of endoscopic screening for CRC. We only need formal and final confirmation of CRC mortality and incidence reduction of primary screening colonoscopy from randomized controlled trials which are underway.8,9 The exceptional feature of endoscopic screening, confirmed in the current study, is that its effect is very long lasting and it may be probably performed only once in a life time, providing that a quality of examination is high. 10 The emerging issue that needs to be addressed thoroughly is the choice of optimal age to begin endoscopy screening.