OBJECTIVES: Three trials evaluating CDK4&6i+fulvestrant have recently reported overall survival (OS) data among women with HR+/HER2− advanced breast cancer. The objective of this research was to compare the OS of post-menopausal (including, but not limited to, natural, surgical or ovarian suppression) women treated with CDK4&6i+fulvestrant who progressed on or immediately after (<12 months) adjuvant endocrine therapy (ET) or while receiving their first ET for advanced disease. METHODS: A SLR identified RCTs based on pre-defined eligibility criteria. Data from the intention-to-treat (ITT) populations and subgroups were extracted. There was a suggestion of a departure from the proportional hazards (PH) assumption. Therefore, a time-to-event NMA based on Fractional Polynomials (FP) proposed by Jansen (2011) was conducted in a Bayesian framework allowing non-PH. A range of first and second order FP were considered. For each FP, random effect (RE) and fixed effects (FE) were examined. The best fitting model was selected using the deviance information criteria (DIC). Afterwards, the relative treatment effects were assessed, and the survival curves displayed. RESULTS: A FE model with P1 = -0.5 & P2 = 0 was the best fitting model for this network. The mean difference (95% Credible Interval [CrI]) in expected survival at month 80 for palbociclib+fulvestrant vs abemaciclib+fulvestrant was -11.44 (-21.86 to 2.85) months and for ribociclib+fulvestrant vs abemaciclib+fulvestrant was 2.16 (-6.70 to 10.65) months. The overlapping 95% CrI indicate no differences between the CDK4&6i. CONCLUSIONS: The results showed no differences in OS between the CDK4&6i+fulvestrant based on the restricted network. The FP NMA allowed us to address uncertainties associated with violation of PH assumption. Interpretation is limited by important uncertainties around the estimates of relative treatment effects, suggesting heterogeneity among trial populations generated by differences in key prognostic factors.