Personalized N-of-1 combinations based on molecular profiles in advanced malignancies: Immunotherapy group analysis of the I-PREDICT N-of-1 precision oncology study.

Abstract

2628 Background: While current precision oncology typically targets a single biomarker, most cancers have multiple pathogenic alterations that often differ from patient to patient. We evaluated whether individualized, N-of-1 biomarker-based combinations that included immune checkpoint inhibitors (ICIs) would improve outcomes in patients (pts) with advanced cancers. Methods: In the I-PREDICT prospective trial (NCT02534675), 70 pts with advanced poor-prognosis cancers received ICI-based regimens. Personalized combinations were based on tissue/ctDNA next-generation sequencing (NGS) and Molecular Tumor Board review. An "i-Matching Score" was developed using TMB ≥16 mut/Mb—selected because analysis showed it was a more robust outcome predictor than TMB ≥10 or continuous TMB (though both correlated with some outcome parameters)—alongside scores reflecting the matching of other targeted therapies (TT) to tumor alterations. Regimens used individualized dose reductions and intra-patient dose titration for safety. Results: Tumors harbored a median of five pathogenic alterations (range, 1–18); 97% of pts had unique molecular profiles. To optimize biomarker-matching, pts received 48 distinct therapy regimens, many previously unstudied; for the first-in-human (FIH) combinations, initial doses were lowered, and intra-patient dose modifications were made for tolerance. Regimens included ICI monotherapy (20%), ICI + TT (61%), and others. Mean initial dose was 79.5% of the FDA-approved dose; 26% of pts had dose adjustments (6% increase; 20% decrease). Initial doses were lower for multi-drug regimens (P=0.002), ICI+TT (P<0.001), and FIH regimens (P=0.005) vs. established regimens. Grade ≥3 treatment-related serious adverse events (SAEs) occurred in 14% of 70 pts; notably, the SAE rate for ICI monotherapy (established safety profile) was 15%, and no combination subgroup significantly exceeded this rate. FIH regimens and high i-Matching groups (>50% vs. ≤50%) trended towards fewer SAEs (P=0.06/0.08). TMB ≥16 independently predicted improved disease control rate (DCR; SD ≥6mo/PR/CR), PFS, and OS (P=0.006/0.001/0.016); i-Matching Score >50% vs. ≤50% significantly correlated with higher DCR (74% vs 27%, P<0.001), longer PFS (10.4 vs 3.8 mo, P<0.001) and OS (19.1 vs 9.1 mo, P=0.016). In pts receiving ICI+TT (n=51), the i-Matching Score outperformed TMB as an outcome predictor. Conclusions: Individualized, N-of-1 ICI-based combinations, molecularly matched to complex genomic profiles, can be safely given with initial dose reduction and intra-patient dose modification. Higher TMB and higher degrees of biomarker matching (i-Matching Score) correlated with improved outcomes, with the latter performing best in pts receiving ICI+TT combinations. This molecularly guided multi-agent therapy paradigm warrants further study. Clinical trial information: NCT02534675 .