Background: D+T, I+N, N and P are approved for the treatment of BRAF+ MM. Without head to head trials of D+T v. I+N or N/P comparative effectiveness of real world outcomes are needed to better inform treatment decisions. Methods: Physicians from the Cardinal Health Oncology Provider Extended Network identified BRAF+ MM patients initiating 1L with D+T, I+N, or N/P after 01/01/2014 through 6/31/2017. Treatment and clinical data (including target lesion measurements) were entered into electronic case-report forms which were validated by clinical research staff. 1L ORR using RECIST v1.1 and TTF was compared between D+T and I+N, N/P. Odds ratio (OR) for objective response (complete or partial) using multivariate logistic regression and hazard ratio (HR) for TTF using a Cox proportional hazards models adjusted for patient factors were calculated. Results: 53 providers contributed 345 patients. Patient characteristics, ORR, median TTF, and model estimates are shown in the table. ORR was significantly lower in N/P than in D+T (42.3% v 60.6%, p=0.01); adjusted OR was not significantly different. Median TTF was significantly longer in D+T versus I+N (11.4 v 4.6 mo, p<0.001) but not N/P (12.0 mo, p=0.26). Risk of treatment failure was 3.2 times greater (HR=3.2; 95% CI: 2.3-4.6) among I+N v D+T. Conclusions: In the largest community-based study of 1L outcomes for BRAF+ MM the ORR was greatest for 1L D+T. TTF for 1L D+T was longer than I+N but similar to N/P. Adjusting for LDH/liver metastases, a potential community provider selection bias, results in a non-significant difference in ORR but lower risk of TTF for D+T v I+N remained.

* p<0.05 v D+T