{
  "abstract": "Introduction Previous research has suggested that colorectal neoplasia is associated with alterations to faecal microbiome signatures, however findings are inconsistent. We aimed to explore the diagnostic utility of microbial markers in colorectal neoplasia risk prediction models.Methods This was exploratory research assessing whether the addition of faecal microbiome signatures improved the performance of existing colorectal neoplasia risk prediction models when applied to the COLO-COHORT dataset. COLO-COHORT is a multi-site, observational study aiming to develop a risk prediction tool for colorectal neoplasia.For this analysis, two existing logistic regression models were selected (the FIT-based, FAST score1 and the non-FIT-based, Law score2). Firstly, the AUROC of the baseline models was evaluated for the outcomes of: any neoplasia, advanced neoplasia and cancer. Subsequently, two faecal microbiome signals (bacterial species with well-described associations with CRC) were added to the refitted logistic regressions and the AUROC was recalculated.Results COLO-COHORT data were available for 1959 participants. Colorectal neoplasia was present in 858 participants (43.8%). Advanced neoplasia and cancer were detected in 305 (15.6%) and 49 (2.5%) participants respectively.Predictive performance of the baseline models and following the addition of the microbiome markers are shown below:Conclusions This analysis demonstrated that the addition of microbiome-based markers did not significantly improve the performance of either model for ‘any’ or ‘advanced’ neoplasia. Our findings could suggest these species improve detection of CRC in non-FIT based models, but incremental value beyond FIT may be constrained. Due to small CRC sample size, this finding should be considered hypothesis-generating only.References Cubiella J, Digby J, Rodríguez-Alonso L, et al. The fecal hemoglobin concentration, age and sex test score: Development and external validation of a simple prediction tool for colorectal cancer detection in symptomatic patients. Int J Cancer 2017;140(10):2201–11.Law C, Rampal S, Roslani AC, Mahadeva S. Development of a risk score to stratify symptomatic adults referred for colonoscopy. J Gastroenterol Hepatol. 2014;29(11):1890–6.Abstract O22 Table 1OutcomeModelAUROC (95% CI)Any neoplasiaFAST0.682 (0.659-0.706) + F.nucleatum0.684 (0.660-0.707) + P.micra0.682 (0.659-0.706)Law0.627 (0.602-0.651) + F.nucleatum0.601 (0.576-0.626) + P.micra0.603 (0.578-0.628)Advanced neoplasiaFAST 0.729 (0.695-0.762) + F.nucleatum0.729 (0.696-0.762) + P.micra0.729 (0.696-0.762)Law 0.600 (0.567-0.633) + F.nucleatum0.601 (0.568-0.634) + P.micra0.607 (0.574-0.640)CancerFAST 0.868 (0.831-0.904) + F.nucleatum0.870 (0.833-0.907) + P.micra0.848 (0.805-0.891)Law 0.654 (0.581-0.728) + F.nucleatum0.685 (0.610-0.760) + P.micra0.734 (0.660-0.807)",
  "authors": [
    {
      "affiliations": [
        "Newcastle University, Newcastle upon Tyne, United Kingdom"
      ],
      "name": "Sarah Manning"
    },
    {
      "affiliations": [
        "Newcastle University, Newcastle upon Tyne, United Kingdom"
      ],
      "name": "Yashvee Dunneram"
    },
    {
      "affiliations": [
        "Newcastle University, Newcastle upon Tyne, United Kingdom"
      ],
      "name": "Colin Rees"
    },
    {
      "affiliations": [
        "Newcastle University, Newcastle upon Tyne, United Kingdom"
      ],
      "name": "Linda Sharp"
    }
  ],
  "title": "O22 Exploring the utility of faecal microbiome markers in risk prediction models for colorectal neoplasia",
  "uid": "2c7c0d93-99d2-583e-a7c7-83e1b7a2be6a"
}