Introduction: The NMPOA cohort started in 2004 with the purpose of testing a breast screening model for underserved women in limited resource countries. In the city of Porto Alegre (South Brazil with 1.5 million population), breast cancer incidence rate is 127/100.000 (Brazilian NCI,2010) with increasing mortality due to 45% of Stage III and IV and no public health care early detection program, it was mandatory to test the feasibility of an organized screening program for women with very low income and level of education (64% under 6 years of schooling). In this scenario, there was always a concern of over diagnosis that will increase costs (more biopsies and surgeries) and under diagnosis by the radiologists involved in the program. We present here the results of all breast biopsies performed in this period, and its correlation with the attributed BI-RADS classification in order to optimize care and costs. Also, this study has a long term proposal and other results have been published concerned to the baseline profile of the population (Cancer Epidemiol Biomarkers Prev. 2010 Oct;19(10):2673–9. Epub 2010 Aug 17).

Methods:Since April 2004, 5592 women from 40 to 69 years were enrolled in the program and underwent to clinical breast exam and annual mammography screening at NMPOA. Initially, lesions were classified as B1, B2 or B0. After complementary views or exams, B0 was classified as B3, B4 (A, B or C) or B5. B4 and B5 were considered positive tests and immediately submitted to biopsy (open surgical or ultrasound guided core biopsy). B3 was followed according to recommended in BI-RADS fourth edition, and subsequently classified as benign (annual follow up) or suspicious findings (biopsy performed). The evaluation included histological analysis of breast biopsy specimens by two independent pathologists. Predictive positive value (PPV), predictive negative value (PNV) and accuracy (Acc) were calculated.

Results: A total of 259 breast lesions were initially classified as BI-RADS 0 (recall rate of 2.8%), and further classification of these was: 166 B3 (64.1%), 60 B4 (23.2%) and 33 B5 (12.7%). Thirty seven B3 (14.3%) were considered suspicious after 6 months and submitted to biopsy. Correlation results of the 130 biopsies performed are summarized on [Table1]. PPV of 63.4% (52.8%-73.1%), PNV of 97.3% (85.8%-99.9%) and an Acc of 73.1% were observed. In this sample, 71.7% of BC were ductal invasive carcinoma, 11.7% ductal carcinoma in situ, 5% lobular invasive carcinoma and 11.6% were other types of neoplasia.

Conclusion: Recall rate is in accordance with the medical audit benchmarks recommended of less than 10% published in BI-RADS fourth edition. NMPOA is model for limited resource countries in breast cancer imaging diagnosis. The radiologists involved in the program are classifying breast lesions in accordance with the established parameters, with a high accuracy in diagnosis. Even though we are dealing with a poorly educated cohort not used to any form of screening, our adherence is 57% in 12 months and 71% in 24 months.

Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P5-09-01.