Diagnostic accuracy of sonographic bi-rads score in the diagnosis of breast lesions based on histopathology at Moi Teaching and Referral Hospital, Eldoret-Kenya

Abstract

Background: In Kenya, breast cancer leads in incidence with 5,985 new cases annually (12.5% of all new cancer cases) and it is third leading cause of all cancer deaths. Breast screening and diagnosis through imaging is important in the treatment and control of the disease. Breast Ultrasonography (US) should be the primary imaging tool for women who are pregnant, lactating, or younger than 30 years and for women > 40 years old it is complementary to both mammography and magnetic resonance imaging of the breast. Breast Imaging Reporting Data System (BI-RADS) is the reference for the presentation of clinical and medical results and communications. Its main objectives are to standardize the reports, facilitate the comparison of follow-up examinations and to allow a collection of data so as to follow treatment on an individual scale. However, the utilization of this standardized system has not been explored and no previous studies on diagnostic accuracy of sonographic BIRADS score exists in our setup. Therefore, the study aimed to establish the diagnostic accuracy of sonographic BI-RADS score based on histopathology in the diagnosis of breast lesions. Objectives: To determine the sonographic and histopathological findings and to establish the diagnostic accuracy of sonographic BIRADS score in the diagnosis of breast lesions based on histopathology at Moi Teaching and Referral Hospital(MTRH). Methods: This was a cross-sectional study done in MTRH from April 2021 to March 2022. A consecutive sampling technique was used to enroll 214 patients aged between 19-70years. Breast US was done using 7.5MHz linear transducer of the MINDRAY M7 ultrasound machine. Sonographic findings were described and classified into categories 0 to 6 according to the BI-RADS for breast US. A total of 295 breast US were done, however only 214 patients underwent core needle biopsy and their results were analyzed. Data collection form was used to record the demographics, clinical presentation, radiological findings (BIRADS category) and histopathological findings. BIRADS categorization and biopsy results were compared. Continuous variables were summarized using mean and categorical variables were summarized in frequencies and percentages. Cohens kappa statistics was used to determine the level of agreement between the sonographic findings and histopathology. A P-value of <0.05 was considered statistically significant. Results: Majority of the study participants were female 204 (95.3%) with the mean age 44.14years. Clinically, 72% of patients presented with breast lump only, (8.4%) breast lump with pain, (8.9%) had ulceration. Sonographically, BIRADS-4(suspicious) was commonest at 33.6% followed by BIRADS-5(highly suspicious) at 24.8% and BIRADS-3 (probably benign) at 22.4%. Histopathologically, malignant breast lesions were the commonest at (59.3%) while the rest were benign at 40.7%. Commonest malignant and benign lesions was invasive ductal carcinoma (IDC) at (59.3%)n and fibrocystic change (41.4%) respectively. Sonographic BIRADS score had sensitivity (66.67%), specificity (51.14%), NPV (51. 72%), PPV (66.14%), NLR (0.65) and PLR (1.36). Correlating signs for malignancy were hypo-echogenicity (96.2%), uncircumscribed margins (95.7%) and posterior shadowing (77.4%) and for benignity were parallel orientation (96.5%), echogenic (95.1%), well circumscribed margins (91.8%) and oval/round shape (68%). There was slight agreement between the sonographic BIRAD-score and histopathology with two examinations having κ = 0.178 (95% CI 0.090 to 0.333). Conclusion: Descriptors from the sonographic BI-RADS lexicon can be useful in differentiating benign from malignant masses. The low sensitivity and specificity for the BIRADS was due to high false positive and false negative numbers. Recommendation: use of BIRADS as a standardized reporting tool should be mandatory. Use of high resolution transducers is recommended for high quality images and in detection of smaller abnormalities.