Abstract Background Chemotherapy is integral to many curative-intent regimens for HER2-positive breast cancer, but it contributes substantial toxicity and treatment burden, motivating interest in de-escalation strategies paired with HER2-targeted therapy. We conducted a systematic review and meta-analysis to quantify comparative effectiveness and safety of chemotherapy de-escalation approaches. Methods PubMed, EMBASE, and Scopus were searched from inception to December 16, 2025 for English, peer-reviewed adult human randomized controlled trials (RCTs), cohort, or case–control studies evaluating chemotherapy omission, reduction, substitution, or shortening versus a contemporaneous comparator. Study selection, data extraction, and risk-of-bias assessment (Cochrane Risk of Bias 2 for RCTs; ROBINS-I for non-randomized studies) were performed by all authors with discrepancies resolved by discussion. Random-effects meta-analyses were performed in RevMan 5.4 using risk ratios (RR) for pCR, odds ratios (OR) for serious adverse events, and hazard ratios (HR) for time-to-event outcomes. Results Six studies (total N = 1,770) met inclusion criteria, largely in operable early to locally advanced settings with varied de-escalation backbones. Pooled estimates showed no clear differences in disease-free survival (HR 0.97, 95% CI 0.45–2.12; p = 0.95; I²=64%), recurrence-free survival (HR 0.85, 95% CI 0.30–2.39; p = 0.76; I²=75%), or pCR (RR 0.53, 95% CI 0.13–2.21; p = 0.19; I²=90%), while serious adverse events were reduced (OR 0.37, 95% CI 0.15–0.88; p = 0.03; I²=62%). The pooled overall survival estimate (HR 0.98, 95% CI 0.96–0.99; p = 0.004) was driven predominantly by a single large-sample study and should be interpreted with caution given sparse event counts and the exploratory nature of survival endpoints in most included trials. Conclusion Chemotherapy de-escalation may reduce severe toxicity in selected patients with HER2-positive early breast cancer, but heterogeneity in de-escalation strategies, study designs, and outcome definitions and imprecision limit definitive inference. Most included trials were not powered for long-term survival endpoints, and pooled survival estimates should be regarded as hypothesis-generating rather than practice-defining. Adequately powered, randomized trials with prespecified non-inferiority margins and standardized endpoints are needed to establish the safety and efficacy of specific de-escalation approaches.
Setiawan et al. (Thu,) studied this question.