Introduction
Colonoscopic surveillance reduces colorectal cancer (CRC) incidence and mortality in Lynch syndrome (LS) carriers. Still, its invasiveness and high costs highlight the need for a non-invasive approach to predict the presence of neoplastic lesions.
Aims & Methods
This study aimed to identify faecal bacterial biomarkers to discern between carriers with colorectal lesions and those with normal colon.
Two independent Spanish cohorts (a training and a validation cohort) were recruited through two multicenter studies (Table 1). Participants provided a faecal sample before surveillance colonoscopy to quantify the abundance of F. prausnitzii (FPRA) and its phylogroups I and II (PHGI and PHGII), P. stomatis (PTST), B. fragilis (BCTF), B. thetaiotaomicron (BCTT), B46 (best BLAST match S. variabile), G. morbillorum (GMLL), B48 (best BLAST match R. faecis), and E. coli (ECO) through qPCR. Eubacteria (EUB) was quantified as the total bacterial load. Participants were classified based on colonoscopy findings as normal colonoscopy (NC), non-advanced adenoma (NAA), and advanced neoplasia (AN, including advanced adenoma and CRC). A bacterial signature for neoplasia prediction was defined and validated in the training and validation cohorts, respectively.
Table 1. Main characteristics of the study participants. NC, normal colonoscopy; NAA, non-advanced adenoma; AA, advanced adenoma; CRC, colorectal cancer.
|
|
| Training cohort | Validation cohort |
|---|
| Age | mean (range) |
| 51 (25-83) | 47 (20-75) |
| Sex (female) | n (%) |
| 32 (64.0) | 75 (65.8) |
| Diagnosis | NC (%) |
| 40 (80.0) | 71 (64.5) |
| | NAA (%) |
| 6 (12.0) | 35 (31.8) |
| | AA (%) |
| 3 (6.0) | 3 (2.8) |
| | CRC (%) |
| 1 (2.0) | 1 (0.9) |
| Mutated gene | High penetrance
| MLH1 (%) - MSH2 (%) - EPCAM (%)
| 18 (36.0) - 9 (18.0) - 0 (0.0) | 37 (33.6) - 25 (22.9) - 1 (0.9) |
| | Low penetrance | MSH6 (%) - PMS2 (%)
| 16 (32.0) - 7 (14.0) | 37 (33.6) - 10 (9.0) |
| Total | n |
| 50 | 110 |
Results
The bacterial abundances were analysed in the training cohort according to age, sex, body mass index (BMI), germline mutation, and colonoscopy findings. Significant moderate positive correlations were identified between the abundances of BCTF and age (r=0.325, p=0.040), and PHGII and BMI (r=0.399, p=0.010). A significant negative correlation was identified between GMLL abundance and BMI (r=-0.331, p=0.036). Patients harbouring MLH1 mutations exhibited a significantly higher abundance of FPRA compared to PMS2 carriers (p=0.048). Patients with neoplasia (NAA+AN) presented significantly lower abundances of BCTF (p=0.027) and B48 (p=0.038) than patients with NC. Patients with AN presented a significantly lower abundance of PTST compared to patients with NC or NAA (p=0.032).
Although all the bacterial markers achieved a sensitivity of 100% to predict the presence of neoplasia, none of them presented a specificity higher than 25% by itself. To maintain sensitivity and increase specificity a bacterial signature was defined including ECO, BCTF, GMLL, FPRA, PHGI, PHGII, and B46 combined with the inherited mutation. The signature showed 100% sensitivity (SS), 52.50% specificity (SP), 34.48% positive predictive value (PPV), and 100% negative predictive value (NPV). This signature was subsequently validated demonstrating 100% SS, 56.45% SP, 53.44% PPV, and 100% NPV.
Conclusion
A new faecal bacterial signature to predict the presence of colorectal neoplasia in LS carriers has been identified, showcasing its potential usefulness to improve and personalise LS surveillance, optimising endoscopic resources, and providing a less invasive and cost-effective approach.
Disclosure
S. Taboada-López, M. Malagón, S. Ramió-Pujol and M. Serra-Pagès are employees of GoodGut.