RESUMO
Background: In clinical trials of patients with anaplastic lymphoma kinase (ALK)-positive non-small-cell lung cancer (NSCLC) treated with crizotinib, evaluation of the relationship between the percentage of ALK-positive cells by fluorescence in situ hybridization (FISH)-particularly near the cut-off defining positive status-and clinical outcomes have been limited by small sample sizes. Patients and methods: Data were pooled from three large prospective trials (one single-arm and two randomized versus chemotherapy) of crizotinib in patients with ALK-positive NSCLC determined by Vysis ALK Break Apart FISH using a cut-off of ≥15% ALK-positive cells. Logistic regression and proportional hazards regression analyses were used to explore the association of percent ALK-positive cells with objective response and progression-free survival (PFS), respectively. Results: Of 11 081 screened patients, 1958 (18%) were ALK positive, 7512 (68%) were ALK negative, and 1540 (14%) were uninformative. Median percentage of ALK-positive cells was 58% in ALK-positive patients and 2% in ALK-negative patients. Of ALK-positive patients, 5% had 15%-19% ALK-positive cells; of ALK-negative patients, 2% had 10%-14% ALK-positive cells. Objective response rate for ALK-positive, crizotinib-treated patients with ≥20% ALK-positive cells was 56% (n = 700/1246), 55% (n = 725/1312) for those with ≥15% ALK-positive cells, and 38% for those with 15%-19% ALK-positive cells (n = 25/66). As a continuous variable, higher percentages of ALK-positive cells were estimated to be associated with larger differences in objective response and PFS between crizotinib and chemotherapy; however, tests for interaction between treatment and percentage of ALK-positive cells were not significant (objective response, P = 0.054; PFS, P = 0.17). Conclusions: Patients with ALK-positive NSCLC benefit from treatment with crizotinib across the full range of percentage of ALK-positive cells, supporting the clinical utility of the 15% cut-off. The small number of patients with scores near the cut-off warrant additional study given the potential for misclassification of ALK status due to technical or biologic reasons.