Intragenic rearrangement burden: A novel biomarker to predict immune checkpoint blockade response in TMB-low cancers

Background: Immune checkpoint blockade (ICB) efficacy is limited, and biomarkers like Tumor Mutation Burden (TMB) fail in TMB-low cancers (e.g., breast, ovarian) despite many being immunologically "hot". This paradox suggests an uncharacterized neoantigen source. We investigated intragenic rearrangements (IGRs)-cryptic structural mutations-as this source, hypothesizing IGR burden is a superior biomarker for T-cell inflammation and ICB benefit in these tumors. Methods: We performed a pan-cancer WGS analysis (ICGC, n=1,033) correlating IGR burden with T-inflamed signatures. We used multivariate regression to compare IGRs against TMB, indels, fusions, and SCNA on immune infiltration, validating in TNBC (n=513), HGSC (n=33), and ESAD cohorts. Predictive power for ICB response was assessed by Kaplan-Meier, Cox regression, and ROC analysis in ESAD (durvalumab), HGSC (NACT + pembrolizumab), and mTNBC trials. We also analyzed a breast cancer immunopeptidomic dataset (n=26) to correlate IGR burden with presented neopeptides. Results: Pan-cancer analysis showed IGR-high and TMB-high are distinct T-inflamed subsets. In TMB-low cancers, IGR burden was the most significant covariate for T-cell inflammation (p=0.0098), not TMB (p=0.224). In TNBC, IGR burden was the only significant genomic marker for the T-inflamed signature (p=0.008). Mechanistically, IGR burden strongly correlated with presented IGR-derived neopeptides (Pearson R=0.54) and showed evidence of immunoediting.In clinical trials, IGR burden predicted ICB benefit. In HGSC (NACT+Pembro), high IGR burden predicted superior OS (AUROC=0.80; 5-yr OS ~85% vs ~30%, HR=0.17, p=0.009), outperforming PD-L1 (HR=0.77, p=0.65). In ESAD (durvalumab), IGR burden (AUROC=0.74) also outperformed PD-L1 (AUROC=0.46). Critically, IGR burden was predictive, not just prognostic. High IGR burden predicted improved OS in mTNBC with chemo-IO (HR=0.18, p=0.074) but poorer OS with chemo-only (HR=4.29, p=0.032). Conclusions: IGR burden is a powerful predictive biomarker for ICB response in TMB-low malignancies where TMB and PD-L1 fail. Supported by immunopeptidomic evidence of neoantigen presentation, it outperforms existing markers, distinguishes true ICB benefit from prognosis, and resolves the "hot" TMB-low tumor paradox. IGR burden represents a critical new tool for patient selection.