Mechanisms of acquired resistance to TRK inhibitors.

3104 Background: First-generation TRK tyrosine kinase inhibitors (TKIs) are approved in a tumor-agnostic fashion in more than 40 countries for patients with NTRK fusion-positive adult and pediatric cancers. While resistance to these agents has previously been described, the exact frequency with which major mechanisms of resistance emerges is not clearly understood. Methods: Patients with an NTRK-fusion-positive tumor who received a first-generation TRK TKI were eligible. We retrospectively identified those patients that had post-progression tumor tissue analyzed by next-generation sequencing (NGS). The pattern of serial resistance to a second-generation TKI was analyzed when available. Results: Eighteen patients were identified. The median age was 46 years (range 2-67). Nine unique fusions were detected in ten different tumor types. NTRK1, NTRK2, and NTRK3 fusions were found in eight (44%), one (6%), and nine (50%) patients, respectively. Thirteen patients (72%) were treated with larotrectinib and five patients (28%) received entrectinib. NGS (MSK-IMPACT n = 17, Foundation One n = 1) carried out on post-progression tissue revealed the following profile of acquired resistance: on-target resistance (83%, n = 15/18), off-target resistance (11%, n = 2/18), and no identifiable mechanism (6%, n = 1/18). Among patients with on-target resistance, the most common mutation involved the solvent front (87%, n = 13/15: n = 7 NTRK3 G623R, n = 4 NTRK1 G595R, n = 1 NTRK2 G639L, n = 1 NTRK3 G623E) followed by the gatekeeper region (13%, n = 2/15: n = 1 NTRK1 F589L, n = 1 NTRK3 F617I). Two patients developed off-target alterations. One acquired BRAF V600E mutation and the other MET amplification. Interestingly, solvent front mutation loss was observed in two patients who transitioned to and progressed on a second-generation TRK TKI. One patient with a baseline NTRK1 G595R mutation developed polyclonal resistance with acquisition of KRAS G12A and NTRK1 G667A alterations as well as NTRK1 G595R loss. The other patient with NTRK3 G623R developed an NTRK3 F617I gatekeeper mutation with NTRK3 G623R loss. Conclusions: In NTRK fusion-positive cancers, on-target resistance preferentially involving the solvent front is more frequent than off-target resistance to first-generation TKI therapy. Furthermore, the sequential use of second-generation therapy appears to alter the evolutionary kinetics of mutation retention and acquisition.