Abstract Background The 21-gene recurrence score (RS) can predict chemotherapy benefit in estrogen receptor-positive, human epidermal growth factor receptor-2-negative (ER+/HER2-) early breast cancer patients. Age would influence the interaction between RS and chemotherapy effect. The current study aimed to determine RS thresholds which were predictive of chemotherapy benefit in young and old women, respectively. Methods Patients diagnosed with pN0–1, ER+/HER2- breast cancer between 2009 and 2016 were retrospectively reviewed. Propensity score matching was performed according to chemotherapy usage. After stratifying patients with different cutoffs of age, the RS threshold indicating chemotherapy benefit in each age strata were determined by cox proportional hazard models. Results A total of 1227 patients were included. The median age was 58 years and the median RS was 24. After matching, the RS thresholds suggesting chemotherapy benefit varied with age. For patients ≤55 years, chemotherapy benefit was observed in those having RS > 25 (P = 0.03), with 4-year invasive disease-free survival (IDFS) of 97.0 and 89.3% in patients receiving chemotherapy or not. While patients derived no benefit from chemotherapy if they had RS ≤25 (P = 0.66, 4-year IDFS: 95.3% vs. 94.6%). For patients > 55 years, adjuvant chemotherapy was associated with better prognosis in those with RS > 36 (P = 0.014, 4-year IDFS: 94.7% vs. 76.2%), but not in those having RS ≤36 (P = 0.13, 4-year IDFS: 92.3% vs. 95.8%). Conclusions Old patients need higher RS thresholds to demonstrate the chemotherapy benefit. Further efforts are warranted to investigate the association between age and predictive RS thresholds.
