William Hsu1, Xinkai Zhou2, Antonia Petruse3, Ngan Chau3, Stephanie Lee-Felker4, Anne Hoyt4, Neil Wenger5, David Elashoff2, Arash Naeim3. 1. Department of Radiological Sciences, University of California, Los Angeles, CA. Electronic address: whsu@mednet.ucla.edu. 2. Department of Medicine, Statistics Core, David Geffen School of Medicine, University of California, Los Angeles, CA. 3. Jonsson Comprehensive Cancer Center, David Geffen School of Medicine, University of California, Los Angeles, CA. 4. Department of Radiological Sciences, University of California, Los Angeles, CA. 5. Division of General Internal Medicine, Department of Medicine, David Geffen School of Medicine, University of California, Los Angeles, CA.
Abstract
PURPOSE: To analyze women with suspicious findings (assessed as Breast Imaging Reporting and Data System [BI-RADS] 4), examining the value of clinical and imaging predictors in predicting cancer diagnosis. PATIENTS AND METHODS: A set of 2138 examinations (1978 women) given a BI-RADS 4 with matching pathology results were analyzed. Predictors such as patient demographics, clinical risk factors, and imaging-derived features such as BI-RADS assessment and qualitative breast density were considered. Independent predictors of breast cancer were determined by univariate analysis and multivariate logistic regression. RESULTS: In univariate analysis, age, race, body mass index, age at first live birth, BI-RADS assessment, qualitative breast density, and risk triggers were found to be independent predictors. In multivariate analysis, age, BI-RADS score, breast density, race, presence of a lump, and number of risk triggers were the most predictive. An integrative logistic regression model achieved a performance of 0.84 cross-validated area under the curve. No variable was a constant independent predictor when stratifying the population on the basis of the BI-RADS score. CONCLUSION: While BI-RADS assessment remains the strongest predictor of breast cancer, the inclusion of clinical risk factors such as age, breast density, presence of a lump, and number of risk triggers derived from guidelines improves the specificity of identifying individuals with imaging descriptors associated with BI-RADS 4A and 4B that are more likely to be diagnosed with breast cancer.
PURPOSE: To analyze women with suspicious findings (assessed as Breast Imaging Reporting and Data System [BI-RADS] 4), examining the value of clinical and imaging predictors in predicting cancer diagnosis. PATIENTS AND METHODS: A set of 2138 examinations (1978 women) given a BI-RADS 4 with matching pathology results were analyzed. Predictors such as patient demographics, clinical risk factors, and imaging-derived features such as BI-RADS assessment and qualitative breast density were considered. Independent predictors of breast cancer were determined by univariate analysis and multivariate logistic regression. RESULTS: In univariate analysis, age, race, body mass index, age at first live birth, BI-RADS assessment, qualitative breast density, and risk triggers were found to be independent predictors. In multivariate analysis, age, BI-RADS score, breast density, race, presence of a lump, and number of risk triggers were the most predictive. An integrative logistic regression model achieved a performance of 0.84 cross-validated area under the curve. No variable was a constant independent predictor when stratifying the population on the basis of the BI-RADS score. CONCLUSION: While BI-RADS assessment remains the strongest predictor of breast cancer, the inclusion of clinical risk factors such as age, breast density, presence of a lump, and number of risk triggers derived from guidelines improves the specificity of identifying individuals with imaging descriptors associated with BI-RADS 4A and 4B that are more likely to be diagnosed with breast cancer.
Authors: Karthikeyan Marthay; Maya Mazuwin Yahya; Tengku Ahmad Damitri Al-Astani Tengku Din; Wan Zainira Wan Zain; Juhara Haron; Michael Pak-Kai Wong; Rosenelifaizur Ramely; Wan Muhammad Mokhzani Wan Mokhter; Siti Rahmah Hashim Isa Merican; Mohd Nizam Mohd Hashim Journal: Cureus Date: 2022-03-01