Abstract P6-09-04: Volumetric breast density improves breast cancer risk prediction

Abstract Introduction: There is increasing interest in implementing personalized breast cancer screening strategies rather than relying on population based guidelines. Most risk models do not include breast density and two models that do rely on subjective BI-RADS categories; all have limited discriminatory ability (C-statistics ranging from 0.60-0.74). Our aim was to develop a model that includes an automated objective and numeric volumetric measurement of breast density with other known risk factors to improve risk prediction. Methods: A case-control study design was used to evaluate the association between risk factors and breast cancer diagnosis. All women diagnosed with breast cancer during 2003-13 with a digital contralateral mammogram at the University of Virginia at the time of diagnosis were eligible as cases. All women without a breast cancer diagnosis but with a digital mammogram at UVA during 2003-2008 were eligible as controls. Risk factor information was collected using a self-reported electronic questionnaire. Mean automated volumetric breast density (Volpara, NZ) was calculated for each patient as a percentage. Controls were matched to cases in a 2:1 ratio based on age group, race, and education, using the GREEDY algorithm. Case-control selections were made using the weighted sum of the absolute differences between the case and control matching factors. Conditional logistic regression using the partial likelihood function from Cox proportional hazard’s regression was used to fit risk prediction equations to the matched case-control study dataset, with stratification for each case matched set. A full model was estimated including all available covariates for use as a model performance reference standard. A reduced model was then estimated including covariates in the full model that had a Wald Chi-Square/degrees of freedom ratio > 1.0. The performance of the full and reduced models was measured using the C index and the maximum R-Square statistic. Results: The study enrolled 3,445 women; 839 cases and 2,606 controls. Multivariable analysis was conducted using 825 cases and 1,628 controls with 1 or more breast studies reported for the surveyed population. The matching process yielded balanced matching factor values between cases and controls, with no significant differences in age group (p = 0.95), race (p = 0.13), or education (p = 0.86). The full prediction model (with 97 df) yielded a C index of 0.88, and an R-Square of 0.53. The reduced model (with 15 df) had a C index of 0.83 and an R-Square of 0.54. Variables in the reduced model included: mean breast density; biopsy showing ADH, ALH/LCIS; BMI; use of HRT, contraceptives, NSAIDS; smoking; exercise; parity; diabetes; family history of breast cancer, HBOC, Li-Fraumeni, or Cowden Syndromes and/or BRCA mutation. Mean volumetric breast density was a leading independent predictor of case status in both the full (p<0.0001) and reduced models (p=0.0212). Conclusion: Volumetric breast density with other known risk factors may provide more accurate individual risk assessment, enabling clinicians to develop patient centered risk based screening protocols that better inform decision making while including patient preferences. The next steps require independent validation of the risk model in other populations. Citation Format: Jennifer A Harvey, George J Stukenborg, Wendy F Cohn, Kathy L Repich, Olivier Alonzo, Wendy M Novicoff, Martin D Yaffe, William A Knaus. Volumetric breast density improves breast cancer risk prediction [abstract]. In: Proceedings of the Thirty-Seventh Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2014 Dec 9-13; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2015;75(9 Suppl):Abstract nr P6-09-04.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.