Accuracy of breast ultrasound BI-RADS classification and final pathological assessment of breast lesions submitted to core biopsy or fine needle aspiration of a breast diagnostic referral center in South Brazil.

Abstract

Abstract Abstract #2002 Background: Breast cancer (BC) is the leading cancer related mortality in women from 40-69 years old in Brazil. In the current year, about 49,400 women are expected to be diagnosed with BC, 4,880 of these in the state of Rio Grande do Sul. Until preventive strategies can be implemented, or better therapies developed, BC mortality reduction can only come from improvement in detection. Radiologists involved in breast imaging should ensure that they have the necessary skills to carry out core biopsy (CB) and/or fine-needle aspiration (FNA) under all forms of image guidance. This quality control is crucial in countries of limited resources, where the costs generated by benign breast biopsies affect the whole public health system. This paper aims to provide a quality parameter on best practice for diagnostic interventional breast procedures and standards.
 Methods: 1,071 consecutive women with breast imaging (BI) abnormalities referred to the BI Center of the Moinhos de Vento Hospital for ultrasound (Us) guided breast CB or FNA between Oct/2003 and Oct/2007 were eligible. All patients in the study population underwent clinical and radiological examination. CB was performed by Us-guided automated gun method. The pathology reports were performed by 2 independent pathologists. All breast findings were classified according to the Us BIRADS system and were correlated to pathology results. An Us BIRADS 4 or 5 was considered a positive test. Sensitivity (Sen), specificity (Spe), predictive positive value (PPV), predictive negative value (PNV) and accuracy (Acc) were calculated.
 Results: Patient medium age was 45.5y (13-92y). Correlation results are summarized on table 1. In this sample, 88.7% of BC were ductal invasive carcinoma, 4.6% lobular invasive carcinoma, 1.2% ductal carcinoma in situ, 1.2% undifferentiated neoplasia, and 4.3% were other types of neoplasia. Sen of 93.3% (CI 87.7%-96.5%), Spe of 95% (93.3%-96.2%), PPV of 75.2% (68.3%Â81.1%), PNV of 98.8% (97.8%-99.4%) and an Acc of 94.7% were observed. 
 Discussion: Us BIRADS classification allows an adjusted standardization of radiologists images interpretation, reproducing the correlation between this and the pathology. Periodic performance audit is the most meaningful way to demonstrate the success or failure in detecting otherwise occult BC, the ultimate indicator of BI performance. Radiologists involved in BI in our center have the requirements to carry out CB and/or FNA under Us image guidance, once the observed results are in accordance with the ACR Practice Guideline, published in 2005.
 
 Citation Information: Cancer Res 2009;69(2 Suppl):Abstract nr 2002.