Aims
Colonoscopy surveillance for Lynch syndrome is burdensome and post-colonoscopy colorectal cancer (CRC) still occurs. Non-invasive faecal volatile organic compounds (VOCs) might guide optimal colonoscopy intervals.

Methods
Prospective, multi-centre study in which individuals with Lynch syndrome collected a faecal sample prior to high-quality surveillance colonoscopy. Samples were analysed using field asymmetric ion mobility spectrometry (FAIMS) and our well-established machine learning pipeline including 10-fold cross validation, to assess diagnostic performance of faecal VOC patterns for relevant neoplasia: advanced neoplasia (CRC, advanced adenomas [AA] and advanced serrated lesions [ASL]) and non-advanced adenomas (NAA). On sensitivity analysis, individuals with and without neoplasia were matched 1:1 on possible confounders: gender, age, BMI, smoking and diet. Using gas chromatography time-of-flight mass spectrometry (GC-TOF-MS), individual faecal VOCs were identified from a random subset of 13 NAA and 14 controls.

Results
Of the 132 included individuals (57% female, median age 51y, 86% ≥2 previous colonoscopies), 3 had CRC, 3 AA, 3 ASL and 32 NAA as most relevant neoplasia. Faecal VOC patterns showed a 66% positivity rate and a sensitivity and negative predictive value of, respectively, 100% and 100% for advanced neoplasia (54% specificity), and 88% and 89% for relevant neoplasia (44% specificity). On sensitivity analysis (n=9 versus n=9 [advanced neoplasia], n=35 versus n=35 [relevant neoplasia]), specificity for advanced neoplasia improved to 89% at equal sensitivity (100%) whereas sensitivity for relevant neoplasia decreased to 79% at equal specificity (44%). NAA presence was associated with decreased faecal VOC abundance of butanal, dimethyldisulfide, dimethyltrisulfide, hydrazinecarboxamide and 2-hexanone.
Ranging from extremely burdensome [0] to not burdensome [10], median patient acceptability regarding faeces collection was 7 (IQR 6 – 9), with “not burdensome” being more prevalent among patients under 39y than over 60y irrespective of gender (OR 0.484, p-value 0.045).

Conclusions
Faecal VOC patterns seem to detect relevant neoplasia in Lynch syndrome with high sensitivity and moderate specificity, with the latter potentially improving upon correction for external confounders. Individual faecal VOCs provide pathophysiological insights and, following validation, may be translated into a diagnostic test. These results provide a perspective on faecal VOCs enabling personalised colonoscopy surveillance in Lynch syndrome.

Joris van Lingen

Appearances