Crossing the toxicity threshold: a repeated landmark analysis of early toxicities and treatment modification during first-line CDK4/6 inhibitor therapy.

BACKGROUND: Whether early treatment-related toxicities during cyclin-dependent kinase 4/6 inhibitor (CDK4/6i) therapy reflect treatment activity or simply longer treatment exposure remains unclear. Previous studies have largely evaluated toxicities cumulatively, potentially introducing immortal time bias into toxicity-outcome associations. We evaluated the prognostic significance of early toxicities and treatment modifications using repeated landmark analyses in patients receiving first-line CDK4/6i therapy for HR+/HER2- metastatic breast cancer. METHODS: This retrospective study included 404 patients receiving first-line CDK4/6i plus endocrine therapy. Repeated landmark analyses at 8, 12, and 24 weeks evaluated associations of early toxicities and treatment modifications with subsequent progression-free survival (PFS) and overall survival (OS). RESULTS: During treatment, any-grade neutropenia occurred in 88.1% of patients, hepatotoxicity in 22.5%, QTc prolongation in 17.6%, and dose modification in 37.4%. At both the 8- and 12-week landmarks, grade 1-2 neutropenia was consistently associated with improved PFS and OS in multivariable analyses (all p<0.05; 8-week PFS: 33.2 vs. 22.1 months; OS: 90.6 vs. 40.8 months). Grade 3-4 neutropenia showed less consistent associations with survival but remained associated with improved OS at 12 weeks. Dose modification was independently associated with shorter PFS at the 12- and 24-week landmarks (all p<0.05). Associations between neutropenia and survival were attenuated at the 24-week landmark. CONCLUSION: Grade 1-2 neutropenia was the most consistent toxicity-based predictor of favorable survival and may represent a potential clinical marker of treatment activity. Associations with grade 3-4 neutropenia were less consistent. Dose modification was associated with shorter PFS, whereas hepatotoxicity and QTc prolongation showed no consistent independent prognostic associations.