Clinical Outcomes and Biomarker Profiles of Elderly Pretreated NSCLC Patients from the BATTLE Trial

BackgroundTreating elderly non–small-cell lung cancer (NSCLC) patients in the salvage setting is challenging because of concerns of intolerance to therapy. Here we report outcomes (survival and toxicity) of elderly patients on the Biomarker-Integrated Approaches of Targeted Therapy for Lung Cancer Elimination (BATTLE) trial.MethodsTwo hundred and fifty-five chemorefractory NSCLC patients received tumor molecular analysis, and were randomized to erlotinib, erlotinib-bexarotene, vandetanib, or sorafenib. Retrospective subgroup analyses were conducted comparing outcomes among age groups (< 65 versus ≥ 65 years; < 70 versus ≥ 70 years; < 75 versus ≥ 75 years), treatments, and sex.ResultsMedian age was 62 years (range, 26–84); 38% were aged 65 years or more. No significant differences among age groups were seen in rates of biopsy-related pneumothorax, treatment-related death, compliance, grade 3 to 4 hematologic toxicities, response rate, nor overall survival. However, older women aged 65 years or more had more grade 3 to 4 nonhematologic toxicities (p = 0.05). Elderly men aged 65 years or more (p = 0.008) had a higher disease-control rate at 8 weeks and a better progression-free survival (PFS) (p = 0.0068). Elderly women aged 70 years or more had a trend toward higher 8-week disease-control rate (p = 0.06). Older men aged 65 years or more treated with vandetanib had a better median PFS (p = 0.03) whereas PFS of older women aged 70 years or more was worse (p = 0.03) compared with younger patients. Elderly men aged 70 years or more treated with sorafenib had a higher overall survival compared with younger men (p = 0.04). Tumor tissue biomarkers show distinct differences by sex and age.ConclusionFit elderly NSCLC patients should be considered for salvage targeted therapy. In this subset of patients, older men seem to have significant clinical benefit from certain agents. Tumor biomarker analysis demonstrates sex and age variations, and is hypothesis-generating