Chemotherapy-induced neutropenia and treatment efficacy in advanced non-small-cell lung cancer: a pooled analysis of 6 randomized trials

Background Chemotherapy-induced neutropenia (CIN) has been demonstrated to be a prognostic factor in several cancer conditions. We previously found a significant prognostic value of CIN on overall survival (OS), in a pooled dataset of patients with advanced non-small-cell lung cancer (NSCLC) receiving first line chemotherapy from 1996 to 2001. However, the prognostic role of CIN in NSCLC is still debated. Methods We performed a post hoc analysis pooling data prospectively collected in six randomized phase 3 trials in NSCLC conducted from 2002 to 2016. Patients who never started chemotherapy and those for whom toxicity data were missing were excluded. Neutropenia was categorized on the basis of worst grade during chemotherapy: absent (grade 0), mild (grade 1-2), or severe (grade 3-4). The primary endpoint was OS. Multivariable Cox model was applied for statistical analyses. In the primary analysis, a minimum time (landmark) at 180 days from randomization was applied in order to minimize the time-dependent bias. Results Overall, 1529 patients, who received chemotherapy, were eligible; 572 of them (who received 6 cycles of treatment) represented the landmark population. Severe CIN was reported in 143 (25.0%) patients and mild CIN in 135 (23.6%). At multivariable OS analysis, CIN was significantly predictive of prognosis although its prognostic value was entirely driven by severe CIN (hazard ratio [HR] of death 0.71; 95%CI: 0.53-0.95) while it was not evident with mild CIN (HR 1.21; 95%CI: 0.92-1.58). Consistent results were observed in the out-of-landmark group (including 957 patients), where both severe and mild CIN were significantly associated with a reduced risk of death. Conclusion The pooled analysis of six large trials of NSCLC treatment shows that CIN occurrence is significantly associated with a longer overall survival, particularly in patients developing severe CIN, confirming our previous findings.