Integrated Genetic Risk Stratification Identifies High Adenoma Detection Rates at Colonoscopy in a Real-World US Health System.

Genetic probability model combining polygenic risk score, family history and Lynch syndrome pathogenic variants, applied to colonoscopy prioritisation.

The authors conducted a retrospective chart review of 800 adults from a US health-system biobank who underwent colonoscopy between 2010 and 2025, with investigators blinded to genetic risk status. Participants were randomly sampled across four predefined genetic risk categories (Low, Intermediate, High, Very High; 200 in each) using an externally developed and validated genetic probability model incorporating polygenic risk scores, family history and Lynch syndrome pathogenic variants; 608 patients were retained after exclusion criteria. Overall adenoma detection rate was 43.59% and rose from 26.27% (Low) to 41.38%, 54.08% and 68.35% (Very High), with P for trend below 0.001 and adjusted odds ratios of 1.77 (1.04-3.01), 2.40 (1.48-3.90) and 4.18 (2.21-7.91) versus Low risk. The absolute difference between Very High and Low risk groups reached 42.08 percentage points. The model improved discrimination beyond baseline predictors (area under the curve 0.739 versus 0.715; ΔAUC = 0.024; P = .03).

Synthesis written by Geno'X. For the full original abstract, please refer to the source publication.

A 42 percentage-point absolute gap in adenoma detection between the extreme strata is striking, yet the model's discrimination gain is only 0.024 in area under the curve: most of the separation comes from age, sex and indication, not the genetic score. This is the classic limitation of polygenic scores applied to triage — they reorganise a stratum-sampled population well, an individual patient far less so. Useful for prioritising resources at health-system scale, insufficient for deciding on a colonoscopy in clinic.

Analysis by Dr Thibaut Benquey