A Retrospective Analysis of a Serum Multi-Biomarker (CA153, TSGF, and CYFRA 21-1) Logistic Regression Model for...

Accurately predicting pathologic complete response (pCR) to neoadjuvant chemotherapy (NAC) in breast cancer remains challenging when relying on single biomarkers. We aimed to establish a composite serum model integrating CA153, tumor-supplied group of factors (TSGF), and CYFRA 21-1 to enhance predictive performance.

This retrospective study included 258 breast cancer patients who received NAC at The First Hospital of Putian between January 2021 and December 2022. Eligible patients had histologically confirmed invasive breast cancer and complete clinical, pathological, and biomarker data. Patients with distant metastasis or incomplete serum data were excluded. Serum biomarkers (CA153, TSGF, and CYFRA 21-1) were measured before NAC. Univariate and multivariate logistic regression analyses were conducted to identify independent predictors of pCR, and receiver operating characteristic (ROC) curves were used to evaluate model performance.

Serum CA153, TSGF, and CYFRA 21-1 levels were significantly lower in patients achieving pCR than in non-pCR patients (CA153: 72.10 vs. 103.82 U/mL; TSGF: 129.77 vs. 188.12 U/mL; CYFRA 21-1: 10.16 vs. 17.05 ng/mL; all P < 0.001). Moderate positive correlations were observed among the three markers. Multivariate logistic regression confirmed CA153 (OR = 1.185), TSGF (OR = 1.062), and CYFRA 21-1 (OR = 1.395) as independent predictors of non-pCR. The composite serum model demonstrated excellent discrimination (AUC = 0.967, 95% CI: 0.949-0.985), with high sensitivity (0.980) and negative predictive value (0.968), outperforming each biomarker alone.

The CA153-TSGF-CYFRA 21-1 serum composite model provides a simple, accurate, and non-invasive approach for predicting NAC response in breast cancer, with potential to support individualized treatment strategies.