Postcolonoscopy colorectal cancer: how low can we go?

Postcolonoscopy colorectal cancers (PCCRCs) are CRCs occurring between 6 months and 3 years after an index colonoscopy.¹ Significant variation in PCCRC rates exist between endoscopy units (3.2%–16.4%).² British Society of Gastroenterology standards advise that all units should develop a system for capturing data on PCCRCs and for investigating these cases by root cause analysis (RCA) and should aim for a PCCRC rate of less than 5% at 3 years.³

A number of factors are known to be associated with PCCRC including patients at increased risk which include those with inflammatory bowel disease (IBD), diverticular disease, those who are female, older and those with comorbidity.¹ PCCRCs are also associated with endoscopists with low adenoma detection rate (ADR), where resections of lesions are incomplete and in certain types of lesion morphology such as subtle, flat, depressed and serrated lesions, particularly in the proximal colon.

The study by Ahmad et al undertook RCA of all potential cases of PCCRC within one large, expert centre in the UK.⁴ Having sifted out incorrect inclusion of cases, 58 were true PCCRCs. The main factors associated with PCCRC were possible missed lesions, inadequate procedures and not resected or incompletely resected lesions. New or rapidly growing lesions have been proposed elsewhere to be a potential explanation for PCCRCs and guidance proposes undertaking Microsatellite Instability testing on all PCCRCs.¹ It is very interesting to note, however, that none of the lesions in this study were felt to be new or rapidly growing cancers. Therefore, all cases identified in this study had the potential to be diagnosed at initial colonoscopy and as such could all potentially have been prevented.

PCCRCs could be classified as being influenced by patient or procedural (including service) factors. The patient factors contributing to PCCRCs are listed above and well described with IBD being the standout risk factor. In some series PCCRC rates in patients with IBD are as high as 38%. It is clear from this study, however, that some of the PCCRCs were due to procedural factors, that is limitations in the quality of the colonoscopy. Good documentation via structured endoscopy reporting and photographic evidence are important markers of quality. Where these items are missing broader issues of quality may be present. These items were not always complete in this study indicating that some elements of quality may have been suboptimal. Variation in colonoscopy quality is well documented and known to be a major factor in PCCRC.^{5 6} Despite key performance indicators (KPIs) being in place in the UK, they are not universally implemented. A recent study from our research team demonstrated that across eight endoscopy units only 61% of endoscopists achieved the minimum recommended adenoma detection rate of 15%.⁷

Undertaking RCA of all PCCRC cases is essential to allow units to review and improve their practice and to allow individual endoscopists to improve the quality of their procedures. It is equally important to provide learning points more widely and this important study provides a range of helpful learning points. Ahmad et al should be commended for the rigour of their study and for having the courage to ‘air their laundry in public’. Transparency should always be commended. It should also be noted that this study is from a world leading colonoscopy centre so if this unit is finding preventable PCCRCs then almost certainly all units will, if they look for them.

RCA allows detailed study of individual cases with potential learning points but how important is benchmarking of PCCR rates? Use of KPIs and quality assurance standards has been an important driver for improving colonoscopy quality internationally, but should we be enforcing standards for PCCRC rates? One problem here is the relative rarity of PCCRCs and the potential tyranny of small numbers. Assuming a mean PCCRC rate of 8.6% and looking to detect a 50% higher PCCRC rate (12.9%) would require around 10 000 colonoscopies to achieve 80% statistical power.¹ It is also extremely difficult to get robust PCCRC data outside of organised screening programmes. This means that benchmarking at an individual level is almost impossible, however unit level benchmarking may have some value.

The last decade has seen a major emphasis on colonoscopy training and quality improvement. Much of the initial emphasis was on caecal intubation rates and more latterly on ADR as a marker of thorough colonic inspection. The link between thorough mucosal visualisation (as measured by ADR) and PCCRC is well established, and we should continue to inform individuals how they might improve their colonoscopy technique through lifelong learning.⁶ All colonoscopists should be aware of advances in the field and implement evidence-based measures to improve practice. A thorough review of this evidence is beyond the scope of this paper but some of the measures know to improve colonoscopic detection include minimal withdrawal time; position changes on extubation; good quality bowel preparation; use of antispasmodics; rectal retroflexion; use of high definition colonoscopes, chromo or virtual chromoendoscopy, use of devices such as Endocuff Vision to improve detection. One area with significant potential is the use of Artificial Intelligence to assist colonic detection with emerging evidence and several large trials underway. It is also important to note that patient experience, as measured by a validated tool such as the Newcastle ENDOPREM, is also an important factor and has been shown to correlate with other quality markers.^{8 9} All endoscopists should seek to continually improve their practice. When, as one of the world leaders in the procedure, Peter Cotton was asked ‘where did you learn ERCP’ Peter answered, ‘I am still learning’. This mantra applies to all areas of medicine but particularly to a dynamic and evolving technique such as colonoscopy.

Patient factors and clinician factors are often viewed in isolation however given the known risk factors for PCCRC, consideration should be given to combining patient factors with the clinician factors—for example, units should develop stratification systems for procedures and ensure that IBD surveillance or other higher risk (for missing PCCRCs) procedures are done by endoscopists with higher KPIs.

So how low can we go? The study by Ahmad et al identifies a significant number of cases in which PCCRC was preventable.⁴ This is in keeping with other work. All colonoscopists should deliver high-quality colonoscopy and where current practice is suboptimal then technique should be improved. Providing low-quality colonoscopy is simply not acceptable particularly when there are so many ways that technique can be improved. PCCRC rates should be reported at unit not individual level with each unit undertaking RCA to identify factors and adjusting services to allow local improvements. An audit or planning, doing, checking and acting cycle should be adopted to continuously improve practice.

Ethics statements

Patient consent for publication

Not applicable.