(PO-0121) Real-world Safety Of Palbociclib In Breast Cancer Patients In The United States

Daniel C Beachler, Cynthia de Luise, Aziza Jamal-Allial, Ruihua Yin, Devon H Taylor and Stephan Lanes

Background: Palbociclib (palb) received FDA approval in 2015 for advanced stage HR+/HER2- breast cancer and is now a standard of care.

Objectives: Provide real-world safety information on palb.

Methods: We conducted a new user cohort study of breast cancer patients initiating palb and/or fulvestrant (fulv) from 02/01/2015 to 09/31/2017 using data from the HealthCore Integrated Research Database (HIRD), a longitudinal claims database from commercial health plan members in the US. The fulv monotherapy cohort was historical, comprising patients initiating fulv from 01/01/2011 to 01/31/2015. Propensity score matching and Cox proportional hazard regression were used to compare incidence rates between groups during therapy for 42 safety codes of interest (including acute liver injury (ALI)) which were defined via ICD-9/10 code algorithms. For ALI, additional analyses, medical record validation and preparation of case narratives for cases confirmed by hepatology experts were conducted, and a contemporaneous comparator group of new users of fulv monotherapy examined from 02/01/2015 to 09/31/2017 was added.

Results: There were 2,445 patients initiating palb including 566 new users of palb + fulv. ICD-9/10 codes of interest were identified for the palb + fulv cohort, including neutropenia, anemia, interstitial lung disease/pneumonitis, and serious infections (incidence rates >20 per 100 person-years). Compared to historical fulv new users, new users of palb + fulv had more claims for neutropenia, leukopenia, anemia, stomatitis and mucositis, and ALI (Hazard Ratios (HRs)>1.6), whereas other claims, e.g., serious infections, and QT prolongation were similar between groups (HRs<1.4). Using several algorithms to identify ALI, an elevated hazard ratio of ALI in new users of palb + fulv was observed compared to the historical fulv new users after control for additional ALI risk factors (e.g., primary ALI algorithm: HR=3.0, 95% CI=1.1-8.4). However when new users of palb + fulv were compared to the contemporaneous fulv comparator group, no increased hazard ratio was observed, although imprecise (HR=0.5 (95% CI=0.1-2.2)). In addition, available ALI cases reviewed by a blinded hepatology expert, did not directly attribute any of these cases to palb or fulv, instead attributing the majority of them to be a consequence of liver metastases progression. Residual confounding effects cannot be excluded.

Conclusions: A large real-world study of new users of palb + fulv, suggests that the observed safety profile is generally consistent with the known safety profile of palb + fulv.