British Society of Breast Radiology Annual Scientific Meeting 2017

Abstract

Introduction Comparison of contrast enhanced spectral mammography (CESM) and breast MRI in monitoring response to neoadjuvant chemother-apy in patients with breast cancer according to different patterns of enhancement and histological type of cancer. Method 29 women diagnosed with breast cancer at Guy's Hospital between October 2015 and May 2016 receiving neoadjuvant chemotherapy were consented to the study. Patients have been offered both CESM and breast MRI before, during and at the end of their treatment. Quantification of the response was done using RECIST criteria. Patterns of enhancement, histological tumour type and final pathological response to treatment were assessed. Results 26 patients were considered to have a response to treatment on both CESM and MRI compared to 23 cases on final pathology specimen. Negative predictive value (NPV) was much higher on breast MRI (98%) compared to CESM (46%) while positive predictive value (PPV) was higher on CESM (70%) compared to MRI (37 %) Enhancement pattern was clumped or mass like in 19, diffuse or punctate in 7. Patients with higher DCIS component were likely to have some residual enhancement which has been underestimated on CESM and slightly overestimated on MRI, especially when compared to those cases with low and intermediate grade DCIS. Conclusion CESM has the potential of monitoring response to neoadjuvant chemotherapy especially in patients with pure invasive breast cancers , however in lesions with associated DCIS component CESM tends to underestimate disease the extent of the disease. O.2 Utility of using mammographic density and clinical risk factors to identify higher risk women in an average-risk screening cohort. What is necessary? What is sufficient? Introduction In the average risk population, mammographic density better predicts breast cancer than risk models using clinical risk factors. This study evaluated the consistency of several risk models to identify patients at higher risk within this population. Methods This 2:1 age-and screen-matched case control study sampled all unilateral screen-detected breast cancer cases from a Canad-ian breast screening program, diagnosed among digitally screened women aged 40-75 (2009-2013). Clinical risk factor data and fully-automated area-based mammographic density assessments were obtained for 392 cases and 817 controls, and used to derive patient-specific risk estimates from models that included density and clinical risk factors alone and in combination. Agreement was assessed using Intraclass Correlation Coefficient (ICC) and Kappa. Results Agreement between model risk estimates was highly variable (ICC=0-0.98, Kappa=0-0.88) and was very poor between models including density alone and models with combinations of clinical risk factors (ICC=0.039, Kappa=0.043). Agreement was almost perfect between a model including density, family history and age and a model including density and all clinical risk factors (ICC=0.98, Kappa=0.88). Conclusion Risk models with varying sets of predictors generate different risk estimates for the same woman and could significantly alter follow-up recommendations , especially for higher risk women. A risk model that includes mammographic density, family history and age provides a practicable and pragmatic solution for identifying higher risk women in the population-based average risk screening setting. Introduction The increased use of neoadjuvant therapy means there is a need for pre-operative prediction of prognosis. We aimed to assess the prog-nostic value of pre-operative factors and combine them to form a predictive model. Methods Consecutive patients with invasive breast cancer undergoing breast ultrasound (US) had the lesion diameter, mean stiffness (kPa) at shearwave elastography (SWE), presentation (screening or symptomatic) core grade and pre-operative nodal status recorded prospectively. Subsequent breast cancer specific survival (BCSS) was ascertained for 3 equal sized groups based on US size and stiffness using Kaplan-Meier survival curves. BCSS according to core grade, presentation and pre-operative nodal status were also produced. Multivariate analysis used cox proportional hazards and a prognostic model was assessed using ROC curves. Results Among 520 patients, 40 breast cancer deaths were recorded at mean follow-up of 5.0 years. BCSS for three equal groups based on SWE were 98%, 92% and 86% (p=0.0001) and on US size 99%, 96% and 82% (p<0.0001). At multivariate analysis all factors except pre-operative nodal status retained significance. A model based on these 4 factors gave BCSS for three equal sized groups of 100%, 95% and 82% (p<0.0001). The model gave identical prognostic information to the Nottingham Prognostic Index (NPI) (AUC 0.86 for both). Conclusion We propose a pre-operative model based on US size, stiffness, core grade and presentation. If validated the model could be used to assess the appropriateness of neoadjuvant therapy. O.5 A fresh look at CT staging in breast cancer: can we do better?