Association Between the Preoperative Systemic Immune- Inflammation Index and Pathologic Complete Response AfterNeoadjuvant Therapy for Breast Cancer A Retrospective Cohort Study With Internal Validation

Abstract Background Pathologic complete response (pCR) after neoadjuvant systemic therapy is an important treatment endpoint in breast cancer. Peripheral blood inflammatory indices may provide low-cost pretreatment predictive information, but their incremental value beyond clinicopathologic characteristics remains uncertain. Methods We retrospectively analyzed 429 registry records corresponding to 427 medical record identifiers. The primary outcome was defined using a strict operational definition of pCR restricted to records coded as exactly 0 or 1 in the original field. The systemic immune-inflammation index (SII) was calculated as platelet count x neutrophil count/lymphocyte count. Multivariable logistic regression was used to evaluate the association between SII and pCR after adjustment for age, clinical T stage, clinical N stage, estrogen receptor (ER) status, progesterone receptor (PR) status, and HER2 immunohistochemistry (IHC). A clinical model was compared with a model that additionally included SII. We report the area under the receiver operating characteristic curve (AUC), Brier score, calibration, and internal validation based on 500 bootstrap resamples. Results Strictly defined pCR status was available for 389 patients, of whom 80 achieved pCR (20.6%). The complete-case dataset for the primary model included 343 patients and 74 pCR events. Median SII was lower in the pCR group than in the non-pCR group (448.35 vs 544.23; P = 0.010). Higher log-transformed SII was independently associated with a lower probability of pCR (odds ratio [OR], 0.56; 95% confidence interval [CI], 0.37–0.85; P = 0.006). In the same 343-patient complete-case dataset, adding SII increased the apparent AUC from 0.716 to 0.731. Optimism-corrected AUCs were 0.686 for the clinical model and 0.700 for the clinical-plus-SII model. Across ascending SII quartiles, pCR rates were 24.7%, 26.8%, 19.6%, and 10.8%, respectively. In a sensitivity analysis using the expanded pCR definition, the corresponding OR was 0.70 (95% CI, 0.48–1.03; P = 0.074). Conclusions Preoperative SII was independently associated with pCR and provided a small but consistent improvement in internal predictive performance beyond a clinicopathologic model. SII may serve as a low-cost adjunct to routinely available clinicopathologic information, although its transportability requires validation in independent cohorts.