9538 Background: In BRAF V600 mutant melanoma, continuous treatment with BRAF/MEK inhibitors invariably begets acquired resistance. Preclinical success of intermittent dosing to overcome resistance did not translate to clinical benefit in the randomized SWOG S1320 trial, possibly due to fixed schedule of intermittent dosing (5 wk on/3 wk off cycles). Leveraging intra-tumoral competition dynamics to slow expansion of resistant tumor subpopulations via patient-specific adaptive MAPK therapy may improve outcomes. We hypothesized that adaptive drug scheduling of BRAF/MEKi combined with anti-PD-1 would lead to non-inferior or improved outcomes with less toxicity compared to a continuous triplet regimen. Methods: In a pilot prospective trial, pts with advanced BRAF V600E mutant melanoma, naïve to systemic tx in this setting, with detectable plasma circulating tumor DNA (ctDNA) were eligible (using BRAFV600E ctDNA ddPCR). Ten pts were randomized 1:1 to continuous (Arm A) or patient-specific adaptively dosed (Arm B) ENCO plus BINI, with continuous biweekly NIVO in both arms. A mathematical model was calibrated for Arm B using pts’ ctDNA values (baseline and biweekly sampling) and standard CT response imaging every 12 weeks. This model was a coupled system of differential equations to guide treatment decisions biweekly by simulating the future response on or off drugs (ENCO+BINI) based on that pt’s known ctDNA levels and response data. Therefore, pts in Arm B all had differing lengths of time that they received ENCO+BINI. Primary endpoint was feasibility of the adaptive approach, by adherence to protocol from baseline to week 12. Results: Ten pts, median age 59, 70% male, 80% with stage IV melanoma, were randomized. In Arm B, NIVO alone was often effective enough to not trigger resuming ENCO+BINI; in Arm A, pts were on E+B for a median of 61 weeks (89% of time on tx, as drugs could be held for toxicity), while in adaptive Arm B, patients were on E+B for median 11 weeks (24% of time on tx). ORR was 100% in Arm A and 60% in Arm B, all partial responses, with 2 stable disease in Arm B. At median follow-up of 45.5 months, median PFS was 26.9 months (95% CI 5.5-NA) with 3 progressors in Arm A. Median PFS was not reached (17.8-NA) in Arm B and 2 progressed (log-rank p= 0.5). Median OS was 32 months (8.8-NA) in Arm A (2-year OS 60%, 3 pts died). Median OS was not reached (17.8-NA) in Arm B (2-year OS 75%,1 pt died), log-rank p=0.18. In Arm A, there were 16 treatment-related grade 3 adverse events, with only 6 grade 3 AEs in Arm B. Conclusions: In this innovative study, we demonstrated the feasibility of patient specific mathematical modeling incorporating ctDNA and imaging, to inform adaptive dosing of ENCO+BINI with concurrent NIVO. The adaptive arm did not compromise efficacy and had lower toxicity. Successful confirmation of this approach in larger trials may favorably impact patient QOL and reduce cost of therapy. Clinical trial information: NCT03543969 .
Eroglu et al. (Thu,) studied this question.