Application of metagenomic sequencing to improve sessile serrated lesion detection using the faecal immunochemical test

Abstract

Most population colorectal cancer (CRC) screening programs use the faecal immunochemical test (FIT), which quantifies occult haemoglobin in faeces and has proven to be effective in detecting and reducing CRC incidence and mortality. Sessile serrated lesions (SSLs) account for up to 30% of sporadic CRC, yet FIT has low sensitivity to detect them. Faecal microbial biomarkers may improve their detection, but FIT samples have not previously been used to study the SSL microbiome. We asked whether FIT samples preserve the faecal microbiome as accurately as fresh collection, and whether they can reveal a microbial signature of SSL. We first reanalysed a published dataset of paired FIT and fresh faecal samples from the same participants. Donor identity explained 63.5% of compositional variance compared with 16.7% for collection method, indicating that FIT preserves the donor-specific signature on which biomarker discovery depends. We then generated shotgun metagenomes from FIT samples of 24 adults scheduled for diagnostic colonoscopy, twelve with at least one clinically significant SSL and twelve with no neoplasia (NN) and compared each group with a published fresh faeces cohort matched on colonoscopy outcome. Collection method explained 6.3% to 7.6% of variance, and no individual species was significantly differentially abundant between methods. Between SSL and NN FIT samples, community structure did not significantly differ, and conservative differential abundance analysis returned no results. More permissive count-based models suggested a depletion of fermentative commensals in SSL. Functionally, SSL samples encoded significantly fewer KEGG orthologs and less polyamine biosynthesis capacity, while orthologs carried by several Bifidobacterium species were more abundant. Many depleted functions reflected Enterobacteriaceae gene content concentrated in a few control samples. No single feature was diagnostic on its own. A multi-feature signature, confirmed in larger cohorts and applied to samples that screening programs already collect, may increase the detection rate of SSLs.