7569 Background: We sought to characterize clinical outcomes, EGFR/RAS mutations, EGFR amplification, pharmacokinetics (PK) and pharmacogenetics (PG) in AA patients receiving erlotinib for previously treated NSCLC. Methods: Pts received erlotinib in a phase II 2-stage minimax randomized study, at the standard 150 mg daily (Arm A), or escalated after 1 cycle to achieve skin rash (Arm B). Primary endpoint was disease control rate (DCR, RECIST CR+PR+SD) at 3 months. PBMCs and tumor tissue were collected from all pts to identify SNPs in metabolizing enzymes, and tumor mutations/amplification. PKs from parent drug and active metabolite (OSI-420) were evaluated at days 1 and 28. Results: Of 42 pts randomized, 39 (37 current/former smokers) started erlotinib, 21 on Arm A, 18 on Arm B. DCR at 3 months in 37 pts to date is 32% (30 and 35%, Arms A and B, respectively). Grade ≥3 adverse events included rash (2 pts), and muscle weakness, dyspnea, dehydration, thrombosis, anemia, 1 pt each. No skin rash on the first two months was associated with early PD (10/11). PK analysis in 26 pts showed higher parent drug and metabolite Cmax and AUCs in pts with grades 0-1 vs. grades 2-3 rashes (cycle 2 OSI-420 mean Cmax of 107 ± 83 ng/mL vs. 243 ± 165 ng/mL, respectively; ANOVA p-value = 0.01). SNP presence in CYP3A4 and CYP3A5 also trends with PK, but not statistically significant due to high variability in PK parameter estimates. One EGFR exon 19 short in frame deletion mutation and a novel exon 20 missense R776L (c.2373 G>T) mutation were identified among 28 samples already analyzed. EGFR was amplified in 9/20 samples. 5 pts had Kras mutations: 3 G12C (c.32 G>T), 1 G12D (c.33 G>A), and 1 G12V (c.33 G>T). Silent mutations indicating SNPs in EGFR exon 20 Q787Q (c.2470 G>A) and Kras (-9, C>T), 9 base pairs upstream of the transcription site, were noted. Conclusions: Erlotinib has similar efficacy and toxicity in AA as reported for caucasians. Absence of rash predicted for worst outcome, but the dose-to-rash strategy did not show meaningful differences. PKs, although highly variable, correlate with rash occurrence and may relate to CYP genotype. A larger sample size will be required to evaluate the role of the PBMC/tumor SNPs identified. Author Disclosure Employment or Leadership Position Consultant or Advisory Role Stock Ownership Honoraria Research Funding Expert Testimony Other Remuneration Genentech Genentech