Decoding molecular, clinico-pathological, and outcome differences between early-onset (EO, <50 years) and late-onset (LO, ≥50 years) mismatch repair deficient (dMMR)/microsatellite instability-high (MSI-H) colorectal cancer (CRC): Insights from real-world data.

19 Background: The incidence of EO CRC is rising globally. While clinico-molecular features of EO mismatch repair proficient/microsatellite stable CRC have been described, less is known about EO dMMR/MSI-H tumors, which are highly immunogenic and may represent a biologically distinct subset. This study provides a comprehensive comparison of EO versus LO dMMR/MSI-H CRC. Methods: We retrospectively analyzed CRC samples profiled by next-generation sequencing (NGS) of DNA/RNA and immunohistochemistry (IHC) at a CLIA-certified lab (Caris Life Sciences; Phoenix, AZ). Patients (pts) with confirmed dMMR/MSI-H CRC were stratified into EO and LO cohorts. Consensus Molecular Subtypes (CMS) were determined from RNA expression profiles using a 600-gene classifier. Statistical comparisons used chi-square, Fisher’s exact or Mann-Whitney U tests with multiple-testing correction (q<0.05). Overall survival (OS) in months (mo) was estimated from insurance claims data using Cox proportional hazards models and log-rank tests. Results: Among 3,846 dMMR/MSI-H CRC cases (EO=496; LO=3,350), EO pts were more often male (65.3% vs 39.9%, p<0.001) and Hispanic (12.7% vs 7.5%, p<0.001), while LO pts were enriched for White race (71.8% vs 38.1%, p<0.001). EO tumors were more frequently left-sided or rectal compared to LO (27% vs 13% and 15% vs 5%, respectively, p<0.001). EO tumors more often showed loss of MSH2/MSH6 expression by IHC, while LO tumors more frequently demonstrated loss of MLH1/PMS2 (all p<0.001). EO tumors had higher mutation rates of KRAS (52.4% vs 19.7%, p<0.001), APC (67.8% vs 35.2%, p<0.001), PIK3CA (44.3% vs 31.1%, p<0.001), PTEN (28.6% vs 21.1%, p=0.0002), MLH1 (35.6% vs 14.9%, p<0.001) and MSH2 (25.1% vs 9.1%, p<0.001%). LO tumors more often harbored BRAF (54.7% vs 5.6%, p<0.001) and KMT2D mutations (55.6% vs 46.6%, p=0.0002). TMB-H rates were comparable (97.0% vs 97.6%), while PD-L1 expression by IHC (SP142) was higher in LO (20.2% vs 6.7%, p<0.001). EO tumors were enriched for CMS3 and CMS4 subtypes (11% vs 8% and 13% vs 7%, respectively, p<0.05). EO pts demonstrated superior OS compared to LO (HR=0.66, 95% CI: 0.56-0.77, 61.7 vs 41.4 mo, p <0.00001), including among those treated with immune checkpoint inhibitors (HR=0.53, 95% CI: 0.40-0.69, 71.9 vs 39.3 mo, p <0.00001). Conclusions: EO dMMR/MSI-H CRC exhibits distinct demographic, clinical and molecular characteristics compared to LO disease. The markedly lower prevalence of BRAF mutations may suggest that a greater proportion of EO cases arise from hereditary predisposition rather than sporadic. Enrichment of divergent oncogenic drivers, such as KRAS and APC , further underscores unique tumorigenic pathways in EO disease, supporting opportunities for tailored therapeutic strategies.