Abstract Introduction T-DM1 treatment significantly improved overall survival and had a lower incidence of grade ≥3 adverse events (AEs) vs capecitabine plus lapatinib in patients (pts) with HER2-positive advanced breast cancer (BC) in the EMILIA study, including pts with treated, asymptomatic CNS metastases (mets). KAMILLA is an ongoing, single-arm, open-label, phase 3b global safety study of T-DM1 in pts with HER2-positive locally-advanced or metastatic BC (target n=2220). In this interim analysis we describe clinical characteristics, safety, and efficacy in pts with stable CNS mets at baseline (BL). Methods Eligible pts received prior HER2-directed therapy and chemotherapy and progressed on or after most recent treatment for advanced BC, or within 6 months of completing adjuvant therapy. Pts with asymptomatic CNS mets were eligible, including pts with stable CNS disease with prior radiation therapy. Pts received T-DM1 3.6 mg/kg every 3 weeks until unacceptable toxicity, withdrawal of consent, or disease progression. This exploratory analysis describes pts with BL CNS mets. Median progression-free survival (PFS) was estimated using the Kaplan-Meier method. Results As of April 4, 2016, data were available for 2017 treated pts, of whom 399 (20%) had BL CNS mets, with a median follow-up of 33 months. Table 1 presents demographic and BL characteristics. The incidence of AEs was similar between pts with and without BL CNS mets. Serious AEs (SAEs) occurred in 112 pts with BL CNS mets (28%) vs 311 of 1618 pts without BL CNS mets (19%), and the frequency of SAEs was comparable to that observed in EMILIA. However, nervous system disorder SAEs, such as seizure, epilepsy, and brain edema, occurred more frequently in pts with BL CNS mets (30 pts [8%]) vs 18 pts [1%] with no BL CNS mets. Mean T-DM1 treatment duration was 8.0 months (median 8 cycles; range 1–55) in pts with BL CNS mets and 9.4 months (median 9 cycles; range 1–57) in pts with no BL CNS mets. Median PFS was 5.5 months in pts with BL CNS mets vs 7.9 months in pts with no BL CNS mets. A decrease in the size of brain target lesions was observed during T-DM1 treatment in 84 of 126 pts with measurable CNS lesions. Table 1. BL characteristics CNS mets at BL (n=399)No BL CNS mets (n=1618)Median age, yrs (range)52 (28–83)55 (26–88)ECOG performance status, n (%) 0193 (48)929 (57)1174 (44)605 (37)232 (8)83 (5)Hormone receptor status, n (%) ER and/or PR positive246 (62)992 (61)ER and PR negative149 (37)593 (37)Median time since initial BC diagnosis, yrs (range)4.8 (0–28)5.0 (0–53)Median time since first metastasis, yrs (range)2.4 (0–25)2.6 (0–35)Median number of prior therapies for metastatic BC (range)3 (0–>10)2 (0–>10) Conclusions This subgroup analysis of KAMILLA is the largest reported cohort of pts with CNS mets treated with T-DM1. The overall safety profile of T-DM1 in pts with BL CNS mets was comparable to that of pts without CNS mets. As might be expected in pts with CNS disease, serious neurological AEs occurred more frequently in pts with BL CNS mets vs those without. Response to T-DM1 was seen in the CNS in pts with BL CNS mets, however, median PFS was lower in pts with BL CNS mets vs those without BL CNS mets. Citation Format: Montemurro F, Ellis P, Delaloge S, Wuerstlein R, Anton A, Button P, Lindegger N, Barrios C. Safety and efficacy of trastuzumab emtansine (T-DM1) in 399 patients with central nervous system metastases: Exploratory subgroup analysis from the KAMILLA study [abstract]. In: Proceedings of the 2016 San Antonio Breast Cancer Symposium; 2016 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2017;77(4 Suppl):Abstract nr P1-12-10.