The pathway alteration load as a pan-cancer determinant of outcome of targeted therapies: Results from the Drug Rediscovery Protocol (DRUP).

Background: Although targeted treatment (TT) has improved outcomes for many patients with cancer, the emergence of primary or secondary resistance remains a common challenge. A frequently suggested contributing mechanism to this resistance is the existence of parallel activated oncogenic signaling pathways that remain untargeted and support tumor survival. Therefore, we hypothesized that an increased number of altered oncogenic signaling pathways (Pathway Alteration Load, PAL) predicts reduced benefit of TT which targets only one altered pathway. Methods: The Drug Rediscovery Protocol (DRUP) is a prospective, pan-cancer, non-randomized clinical trial (NCT02925234) in which patients with treatment-refractory metastatic cancer with an actionable molecular profile are treated with matched approved targeted- and immunotherapies outside their registered indication. All patients treated with TT for whom clinical outcome and whole genome sequencing results of a pre-treatment biopsy were available were included. The PAL was determined by counting the number of proliferation pathways impacted by somatic driver alterations, using an established pathway classification scheme (Sanchez-Vega et al., Cell, 2018). Influence of PAL on clinical benefit rate (CBR, defined as objective response or stable disease ≥ 16 weeks according to RECIST v1.1 or RANO), progression-free survival (PFS) and overall survival (OS) was evaluated. Cox proportional-hazards models, stratified by tumor type, were used to adjust outcomes for drug, performance status, previous treatment lines, tumor mutational burden and cancer cell ploidy. Results were validated using the independent Hartwig database of metastatic cancers. Results: In all 154 patients treated with single pathway-directed TT, PAL was 0 in 3, 1 in 20, 2 in 37, 3 in 56, 4 in 31, 5 in 6 and 6 in 1 patient(s). Patients with a PAL below the median of 3 (N= 60) demonstrated a higher CBR (41.7% vs 25.5%, odds ratio [OR] 0.48, P = 0.051), longer PFS (median 4.7 vs 2.9 months, adjusted hazard ratio [aHR] 1.70, P = 0.020) and OS (median 13.7 vs 5.6 months, aHR 3.80, P< 0.001), compared to those with PAL ≥3. Results were consistent across different PAL cut-offs and when applying PAL continuously (aHR for PFS 1.41, P = <0.001; aHR for OS 1.77, P = <0.001). In the independent Hartwig cohort, 258 patients treated with single pathway-directed TT were identified and similar results were found for CBR (54.2% vs 36.7%, OR 2.04, P = 0.009) and PFS (7.0 vs 4.2 months, aHR 1.55, P = 0.009), but not for OS (14.7 vs 8.2 months, aHR 1.37, P = 0.111). Conclusions: In our population, PAL emerged as a pan-cancer predictor of resistance to TT. Our findings provide support for more refined patient selection for TT and highlight the rationale for combinatorial treatment strategies, especially in patients with multiple affected pathways.