A nomogram integrating clinical and multimodal ultrasound features for pretreatment prediction of HER2 status in breast cancer: a retrospective study.

BACKGROUND: The effective management of heterogeneous breast cancer relies on accurate human epidermal growth factor receptor 2 (HER2) status assessment to guide targeted therapy. Currently, core needle biopsy remains the gold standard for accurately determining HER2 status. However, its high cost and invasive nature render it unsuitable for large-scale screening or for dynamic monitoring of HER2 status during treatment. Furthermore, due to intratumoral heterogeneity, a single invasive biopsy sample may not fully represent the entire lesion. Additionally, HER2 expression in breast cancer can change during chemotherapy. These limitations highlight the need for a non-invasive assessment tool. To address this gap, this study aims to develop a predictive model for HER2 status by integrating clinical and multimodal ultrasound features. METHODS: We retrospectively enrolled 142 consecutive breast cancer patients from Yunnan Cancer Hospital's Breast Center (March 2022-August 2024). The least absolute shrinkage and selection operator (LASSO) regression method was applied to filter variables and select predictors, and multivariate logistic regression was used to construct a nomogram. Model performance was assessed through: (I) calibration curves with Hosmer-Lemeshow test, (II) discrimination via area under the ROC curve (AUC), and (III) clinical utility by decision curve analysis. Internal validation was performed using bootstrap resampling with 1,000 replicates and optimism correction. RESULTS: Six variables of breast cancer patients were selected through the LASSO regression method: maximum lesion diameter, hyperechoic halo presence, Adler grade, shape, lymph node status, and calcifications. Multivariate analysis confirmed three independent predictors of HER2 status (P<0.05): hyperechoic halo presence [odds ratio (OR) =2.655; 95% confidence interval (CI): 1.204-5.990], lymph node status (OR =0.325; 95% CI: 0.133-0.745), and Adler grade (OR =3.093; 95% CI: 1.672-6.110). The resulting nomogram demonstrated good discrimination (AUC =0.728; 95% CI: 0.639-0.817), with bootstrap-corrected AUC of 0.708 (95% CI: 0.620-0.796). The model also exhibited good calibration (calibration slope =0.928; Hosmer-Lemeshow test P=0.86). Furthermore, decision curve analysis confirmed the clinical utility of the nomogram, showing a superior net benefit across a wide range of risk thresholds. CONCLUSIONS: The validated nomogram offers a clinically valuable tool for the pretreatment assessment of HER2 status. While it cannot replace pathological biopsy, it may serve as a complementary approach to identify patients with a high likelihood of HER2 positivity. This could potentially guide biopsy strategies or aid treatment decisions, particularly in cases where biopsy is inconclusive or not feasible.