Editorial: Less is more for colorectal cancer diagnosis – FIT leads the way

Abstract

Symptoms have traditionally been the mainstay of clinical diagnosis and were the basis of the original NICE NG12 GP referral guidelines for patients at risk of colorectal cancer (CRC).¹ Attempts to increase earlier diagnosis of CRC by lowering referral thresholds and broadening referral criteria created further demand for colonoscopy. Despite the expansion in eligibility for urgent testing, many cancers were still identified in routinely referred patients.²

The use of Faecal immunochemical testing (FIT), introduced in the UK in 2017, was transformed by the COVID-19 pandemic. Colonoscopy came to a halt, and the need to identify patients at highest risk of CRC was facilitated by diagnostic accuracy studies of FIT, with preliminary data showing that CRC was rarely found in patients with FIT values below 10 μg/g of faeces. The emergency pandemic guidance from NHS-England provided alternative management pathways for most patients and has subsequently been supported by a large systematic review,³ and most recently by draft NICE guidance.⁴

However, these studies were too small to consider other potentially relevant factors such as age, gender, and the presence of anaemia (iron deficient or otherwise) despite clear epidemiological evidence linking CRC to these risk factors.⁵

Crooks et al tackled this issue and posited a more sophisticated pathway, including age (in 15-year bands) and haemoglobin values.⁶ The study, from a cohort of over 33,000 patients with FIT and haemoglobin results, introduced these two variables by identifying adjusted FIT thresholds above which the risk of cancer was 3%. In those with a normal haemoglobin and a FIT level between 20 and 40 μg/g faeces, only those aged over 85 years had a risk of cancer exceeding 3%. Not surprisingly, anaemia increased the risk; anaemic patients with a FIT over 20 μg/g faeces had >3% chance of CRC except in the small group aged under 40 years.

This study has raised the possibility of using evidenced differential FIT thresholds for selecting those for definitive testing, which could be sophisticated - colonoscopy for those at highest risk, CT colonography (or, perhaps, colon capsule) for intermediate risk, and safety-netting in primary care, perhaps with repeat FIT for those at lowest risk.⁷

Will this be implemented? First, since this is only one study, replication is essential, but may not be simple.⁸ An older Spanish equation failed replication, although it was a small study.⁹ Other factors may be relevant, such as platelet count and gender, although it is unlikely that these will add much predictive power. Adoption will also have to resolve the tension between the current simple guidance, with a single threshold of 10 μg/g, which is easy to implement, and accurate guidance, which uses resources more effectively. This is particularly so in cancer diagnostics where the consequences of a delayed (or entirely missed) diagnosis may be severe. However, CRC diagnostic costs have exceeded treatment costs for many years.¹⁰ Applying a personalised risk should therefore be the long-term aim, allowing those who need investigation to receive it in a timely fashion. This really is a case of ‘less is more’.

Robert P. H. Logan: Writing – original draft (equal); writing – review and editing (equal). William Hamilton: Writing – original draft (equal); writing – review and editing (equal).

This article is linked to Crooks et al paper. To view this article, visit https://doi.org/10.1111/apt.17632