Background: It is matter of current debate which would be the best chemotherapy backbone of neoadjuvant HER2-targeted therapy for HER2+ BC. The TRAIN 2 trial showed no significant difference in terms of pathological complete response (pCR) when anthracyclines–based (CTA) or anthracyclines–free regimens (CT) were combined with dual HER2 blockade. However, it remains unclear how anthracyclines may influence the relative benefit across different anti-HER2 treatments. Methods: A systematic review was conducted which included all phase II/III randomized clinical trials (RCTs) comparing different neoadjuvant regimens for HER2+ BC. pCR (yT0/isN0) was the outcome of interest. Indirect comparisons of all combination of anti-HER2 agents with CTA or CT were estimated with a random-effects frequentist NMA. Estimated pCR rates were inferred adopting a Bayesian NMA. Results: 17 RCTs (3933 patients) were included. Overall, 8 arms were identified, comprising all possible combinations of CTA and CT with trastuzumab (H), lapatinib (L) and dual HER2 blockade (D) but also CTA and D only. Odds ratios (OR) for pCR and 95% confidence interval (CI) of selected NMA comparisons are shown in the table. Estimated rates of pCR for each treatment and 95% credible interval (CrI) are reported in the table. Conclusions: Through indirect comparisons, no significant pCR gain was found for CTA vs CT when combined to D, H and L. In particular, considering double vs single-agent anti-HER2 regimens, D-CT remains superior to H-CTA, supporting a possible omission of anthracyclines when dual anti-HER2 block is used. On the contrary, our pooled estimate suggests a more relevant role for anthracyclines when comparing H-CT/A vs CTA. Moreover, we estimated a 4% pCR gain for D-CTA vs D-CT, and an 8% higher pCR rate for H-CTA vs H-CT.