Association between ultrasound morphologic features and histopathological findings of lobular carcinoma

A role for breast ultrasound Artificial Intelligence decision support in the evaluation of small invasive lobular carcinomas

OBJECTIVE

To evaluate the diagnostic performance of an Artificial Intelligence (AI) decision support (DS) system in the ultrasound (US) assessment of invasive lobular carcinoma (ILC) of the breast, a cancer that can demonstrate variable appearance and present insidiously.

METHODS

Retrospective review was performed of 75 patients with 83 ILC diagnosed by core biopsy or surgery between November 2017 and November 2019. ILC characteristics (size, shape, echogenicity) were recorded. AI DS output (lesion characteristics, likelihood of malignancy) was compared to radiologist assessment.

RESULTS

The AI DS system interpreted 100% of ILCs as suspicious or probably malignant (100% sensitivity, and 0% false negative rate). 99% (82/83) of detected ILCs were initially recommended for biopsy by the interpreting breast radiologist, and 100% (83/83) were recommended for biopsy after one additional ILC was identified on same-day repeat diagnostic ultrasound. For lesions in which the AI DS output was probably malignant, but assigned a BI-RADS 4 assessment by the radiologist, the median lesion size was 1 cm, compared with a median lesion size of 1.4 cm for those given a BI-RADS 5 assessment (p = 0.006). These results suggest that AI may offer more useful DS in smaller sub-centimeter lesions in which shape, margin status, or vascularity is more difficult to discern. Only 20% of patients with ILC were assigned a BI-RADS 5 assessment by the radiologist.

CONCLUSION

The AI DS accurately characterized 100% of detected ILC lesions as suspicious or probably malignant. AI DS may be helpful in increasing radiologist confidence when assessing ILC on ultrasound.

Predictive Value of Ultrasound Characteristics for Disease-Free Survival in Breast Cancer

The main objective of this study was to determine the predictive value of US characteristics for disease-free survival (DFS) in BC patients. We retrospectively analyzed the ultrasonic images and clinical data of BC patients who had previously undergone breast surgery at least 10 years before study enrollment and divided them into a case group and a control group according to the cutoff value of 120 months for DFS. Correlation analysis was performed to identify US characteristics as independent predictors for DFS by multivariable logistic regression and Kaplan−Meier survival analysis. A total of 374 patients were collected, including 174 patients in the case group with short-DFS and 200 patients in the control group with long-DFS. Three US characteristics (size on US, mass shape, mass growth orientation) and two clinical factors (axillary lymph node (ALN), molecular subtypes) were identified as independent predictors for DFS (p < 0.05). The ROC curve showed good performance of the multivariate linear regression model with the area under the curve being 0.777. The US characteristics of large size, irregular shape, and nonparallel orientation were significantly associated with short-DFS, which is a promising supplementary for clinicians to optimize clinical decisions and improve prognosis in BC patients.

Development and External Validation of a Simple-To-Use Dynamic Nomogram for Predicting Breast Malignancy Based on Ultrasound Morphometric Features: A Retrospective Multicenter Study

Background

With advances in high-throughput computational mining techniques, various quantitative predictive models that are based on ultrasound have been developed. However, the lack of reproducibility and interpretability have hampered clinical use. In this study, we aimed at developing and validating an interpretable and simple-to-use US nomogram that is based on quantitative morphometric features for the prediction of breast malignancy.

Methods

Successive 917 patients with histologically confirmed breast lesions were included in this retrospective multicentric study and assigned to one training cohort and two external validation cohorts. Morphometric features were extracted from grayscale US images. After feature selection and validation of regression assumptions, a dynamic nomogram with a web-based calculator was developed. The performance of the nomogram was assessed with respect to calibration, discrimination, and clinical usefulness.

Results

Through feature selection, three morphometric features were identified as being the most optimal for predicting malignancy, and all regression assumptions of the prediction model were met. Combining all these predictors, the nomogram demonstrated a good discriminative performance in the training cohort and in the two external validation cohorts with AUCs of 0.885, 0.907, and 0.927, respectively. In addition, calibration and decision curves analyses showed good calibration and clinical usefulness.

Conclusions

By incorporating US morphometric features, we constructed an interpretable and easy-to-use dynamic nomogram for quantifying the probability of breast malignancy. The developed nomogram has good generalization abilities, which may fit into clinical practice and serve as a potential tool to guide personalized treatment. Our findings show that quantitative morphometric features from different ultrasound machines and systems can be used as imaging surrogate biomarkers for the development of robust and reproducible quantitative ultrasound dynamic models in breast cancer research.

Radiological Underestimation of Tumor Size Influences the Success Rate of Re-Excision after Breast-conserving Surgery

Objective

Failure to achieve adequate margins after breast-conserving surgery often leads to re-excision, either by repeat breast-conserving surgery (BCS) or by mastectomy. Despite the high frequency of this problem, the success rate of achieving adequate margins by repeat BCS is not well documented. The objective of this study was to determine the success rate of repeat BCS and identify the factors influencing that rate.

Materials and Methods

A retrospective review was performed of all women undergoing repeat BCS for inadequate margins after initial BCS in our breast unit between 2013 and 2019. Univariate and multivariate analyses were carried out to identify the factors influencing how often adequate margins were achieved after repeat BCS.

Results

One hundred fifty-four patients underwent repeat BCS after initially inadequate margins, of which adequate margins were achieved in 82%. Patients with successful repeat BCS had smaller tumors, had less underestimation of tumor size on imaging, and were less likely to have had cavity shaves taken at their initial BCS. A tumor size more than 50% larger than predicted by imaging was independently associated with failure of repeat BCS in multivariate analysis (odds ratio: 3.6, 95% CI: 1.41-9.20, p = 0.007). Underestimation of tumor size by imaging was commoner and more extensive in patients with larger tumors and those with ductal carcinoma in situ.

Conclusion

Re-excision by cavity shaves has a high success rate and should be offered to all patients who are deemed suitable for the procedure. Patients whose tumors are more than 50% larger than predicted by imaging should be counseled about the higher risk of failure.

Association between Imaging Features using the BI-RADS and Tumor Subtype in Patients with Invasive Breast Cancer

BACKGROUND

Different molecular breast cancer subtypes present different biologic features, treatment options, and clinical prognoses. The breast cancer imaging phenotype may help precisely classify breast cancer in a non-invasive manner.

OBJECTIVE

To identify the association between the imaging and clinicopathologic features of invasive breast cancer according to the molecular subtype.

METHODS

We retrospectively reviewed the electronic medical records of 313 consecutive women with breast cancer who underwent surgery between March 2018 and February 2019. Preoperative imaging studies were also reviewed and the association between the clinicopathologic and imaging features was evaluated according to the molecular subtype.

RESULTS

On mammography, the presence of microcalcifications was correlated with the human epidermal factor receptor 2-positive subtype (67%, 14/21). Luminal A and B tumors were more likely to have a spiculated margin (57% [63/110] and 41% [34/81]), while human epidermal factor receptor 2-positive and triple-negative breast cancers were more likely to have an indistinct margin (56% [10/18] and 35% [17/48]). On ultrasonography, luminal A tumors were likely to be depicted as masses with an irregular shape (85%, 115/136) and spiculated margin (49%, 66/136). On magnetic resonance imaging, triple-negative breast cancer appeared as a mass (n=13) that frequently had an irregular shape (62%, 8/13) but was more likely to be oval or round (39%, 5/13) than other subtypes.

CONCLUSION

Some imaging features on mammography, ultrasonography, and magnetic resonance imaging could be useful predictors of the molecular subtype of breast cancer and may aid precision medicine development for patients with breast cancer according to the subtype.