Predicting hormone receptor status in invasive breast cancer through radiomics analysis of long-axis and short-axis ultrasound planes

Abstract

The hormone receptor (HR) status plays a significant role in breast cancer, serving as the primary guide for treatment decisions and closely correlating with prognosis. This study aims to investigate the predictive value of radiomics analysis in long-axis and short-axis ultrasound planes for distinguishing between HR-positive and HR-negative breast cancers. A cohort of 505 patients from two hospitals was stratified into discovery (Institute 1, 416 patients) and validation (Institute 2, 89 patients) cohorts. A comprehensive set of 788 ultrasound radiomics features was extracted from both long-axis and short-axis ultrasound planes, respectively. Utilizing least absolute shrinkage and selection operator (LASSO) regression analysis, distinct models were constructed for the long-axis and short-axis data. Subsequently, radiomics scores (Rad-scores) were computed for each patient. Additionally, a combined model was formulated by integrating data from long-axis and short-axis Rad-scores along with clinical factors. The diagnostic efficacy of all models was evaluated by the area under the receiver operating characteristic (ROC) curve (AUC). The long-axis and short-axis models, consisting of 11 features and 15 features, respectively, were established, yielding AUCs of 0.743 and 0.751 in the discovery cohort, and 0.795 and 0.744 in the validation cohort. The calculated long-axis and short-axis Rad-scores exhibited significant differences between HR-positive and HR-negative groups across all cohorts (all p < 0.001). Univariate analysis identified ultrasound-reported tumor size as an independent predictor. The combined model, incorporating long-axis and short-axis Rad-scores along with tumor size, achieved superior AUCs of 0.788 and 0.822 in the discovery and validation cohorts, respectively. The combined model effectively distinguishes between HR-positive and HR-negative breast cancers based on ultrasound radiomics features and tumor size, which may offer a valuable tool to facilitate treatment decision making and prognostic assessment.

Keywords: Breast cancer; Hormone receptor; Radiomics; Ultrasound.

Figure 1

Flowchart illustrating the inclusion and exclusion criteria for the study population. HR hormone receptor.

Figure 2

The flowchart depicting the radiomics analysis conducted in this study. LASSO least absolute shrinkage and selection operator, Rad-score radiomics score, DCA decision curve analysis, ROC receiver operating characteristic.

Figure 3

Radiomic feature selection using the least absolute shrinkage and selection operator (LASSO) logistic regression. The selection of tuning parameter (Lambda) in the LASSO model. (A,C) The area under the curve (AUC) was plotted versus log (Lambda). The dashed lines indicate the selected optimal log(Lambda) value and the location of one standard error. The optimal Lambda values of 0.015879 and 0.012864 with log (Lambda) of −4.142732 and −4.353298 were chosen according to tenfold cross-validation. (B,D) Graphs of variation of the radiomics characteristic coefficients with log(Lambda) for long-axis and short-axis ultrasound planes, respectively. AUC area under the curve.

Figure 4

The distribution of radiomics scores (Rad-scores). Long-axis Rad-scores for each patient across both hormone receptor (HR) positive and HR negative breast cancers, as well as the distribution of short-axis Rad-scores, are assessed in both the discovery cohort (A,C) and the validation cohort (B,D), respectively. Rad-score radiomics score.

Figure 5

Performances for hormone receptor (HR) positive prediction. Receiver operating characteristic curves were generated for the long-axis model, short-axis model, and combined model in both the discovery (A) and validation (B) cohorts. AUC area under the curve.

Figure 6

Calibration curves of the three models. Calibration curves constructed for the long-axis model, short-axis model, and combined model within both the discovery (A) and validation (B) cohorts. These curves described a good fitness between predicted and observed outcomes for hormone receptor (HR) positive and HR negative breast cancers across all three models. The ideal prediction was depicted by the gray line. A closer fitness to the gray line represented a well-calibrated model.

Figure 7

Decision curve analyses for the three models. Decision curves were generated for the long-axis model, short-axis model, and combined model. The combined model exhibited a superior net benefit compared to the other two models across the majority of threshold probabilities.

Figure 8

Combined model nomograms. Nomograms were constructed by integrating the long-axis radiomics score (Rad-score), short-axis Rad-score, and tumor size to differentiate between hormone receptor (HR) positive and HR negative breast cancers. Patient 1 (A) was a 72-year-old female diagnosed with a 3.0-cm-sized breast lesion. Her long-axis and short-axis Rad-scores were 0.907 and 0.943, respectively. The total score calculated was 257 points, corresponding to a high probability of HR positivity (0.940). Pathological examination confirmed HR positive breast cancer in this patient. Patient 2 (B) was a 47-year-old female diagnosed with a 5.0-cm-sized breast lesion. Her long-axis and short-axis Rad-scores were 0.499 and 0.537, respectively. The total score obtained was 178 points, indicating a lower probability of HR positivity (0.357). Pathological analysis confirmed HR negative breast cancer in this case. L-Rad-score long-axis radiomics score, S-Rad-score short-axis radiomics score.