| Sažetak (engleski) | Purpose Based on the risk of locoregional recurrence (LRR), postmastectomy radiotherapy (PMRT) is recommended in T1-T2pN1 breast carcinoma (BC). We aimed to elucidate our institutional strategies underlying selection of these patients for PMRT. In the no-PMRT subset, we compared various lymph node (LN) staging systems’ abilities to predict 5-year overall and locoregional-free survival (OS/LRFS). Methods We retrospectively enrolled 548 women with T1-T2pN1 BC undergoing mastectomy and axillary LN dissection. Depending on PMRT delivery, the participants were divided into the PMRT and no-PMRT groups. Predictors of OS/LRFS were calculated for the no-PMRT group only. Based on Cox regression modelling, the number of positive LNs (PLN), negative LNs (NLN), LN ratio (LNR), log odds of PLN (LODDS), and modified LNR (mLNR) were modelled, each respectively, with OS model covariates (age, grade III, lymphovascular invasion [LVI], tumor size, hormone receptor [HR] status) and LRFS model covariates (age, grade III, LVI). The C-statistic, Akaike information criterion, and likelihood ratio χ2 of the models were compared. Results Median follow-up was 60.5 (18–82), 61 (28–82), and 60 (18–80) months for the entire cohort, PMRT, and no-PMRT group, respectively. The PMRT and no-PMRT groups had comparable OS (p = 0.235). LRFS was better (p = 0.030) in the PMRT group comprising 105 subjects (19.16%) who were younger, more likely to have a highergrade, HR–, HER2+ tumors, more PLNs, fewer NLNs, Ki-67 ≥ 20%, LVI, and extranodal extension (p ≤ 0.001). In the no-PMRT group, LNR-based OS/LRFS models exhibited superior prognostic performance. Conclusion In early-stage BC patients undergoing mastectomies, LN dissections and no PMRT, we propose LNR-based multivariable models to predict OS/LRFS with superior accuracy. |
