Charlotte C. Gard

Breast Density and Benign Breast Disease: Risk Assessment to Identify Women at High Risk of Breast Cancer

PURPOSE Women with proliferative breast lesions are candidates for primary prevention, but few risk models incorporate benign findings to assess breast cancer risk. We incorporated benign breast disease (BBD) diagnoses into the Breast Cancer Surveillance Consortium (BCSC) risk model, the only breast cancer risk assessment tool that uses breast density. METHODS We developed and validated a competing-risk model using 2000 to 2010 SEER data for breast cancer incidence and 2010 vital statistics to adjust for the competing risk of death. We used Cox proportional hazards regression to estimate the relative hazards for age, race/ethnicity, family history of breast cancer, history of breast biopsy, BBD diagnoses, and breast density in the BCSC. RESULTS We included 1,135,977 women age 35 to 74 years undergoing mammography with no history of breast cancer; 17% of the women had a prior breast biopsy. During a mean follow-up of 6.9 years, 17,908 women were diagnosed with invasive breast cancer. The BCSC BBD model slightly overpredicted risk (expected-to-observed ratio, 1.04; 95% CI, 1.03 to 1.06) and had modest discriminatory accuracy (area under the receiver operator characteristic curve, 0.665). Among women with proliferative findings, adding BBD to the model increased the proportion of women with an estimated 5-year risk of 3% or higher from 9.3% to 27.8% (P<.001). CONCLUSION The BCSC BBD model accurately estimates womens risk for breast cancer using breast density and BBD diagnoses. Greater numbers of high-risk women eligible for primary prevention after BBD diagnosis are identified using the BCSC BBD model.

Reproducibility of BI-RADS breast density measures among community radiologists: a prospective cohort study.

Abstract:  Using data from the Vermont Breast Cancer Surveillance System (VBCSS), we studied the reproducibility of Breast Imaging Reporting and Data System (BI‐RADS) breast density among community radiologists interpreting mammograms in a cohort of 11,755 postmenopausal women. Radiologists interpreting two or more film‐screen screening or bilateral diagnostic mammograms for the same woman within a 3‐ to 24‐month period during 1996–2006 were eligible. We observed moderate‐to‐substantial overall intra‐rater agreement for use of BI‐RADS breast density in clinical practice, with an overall intra‐radiologist percent agreement of 77.2% (95% confidence interval (CI), 74.5–79.5%), an overall simple kappa of 0.58 (95% CI, 0.55–0.61), and an overall weighted kappa of 0.70 (95% CI, 0.68–0.73). Agreement exhibited by individual radiologists varied widely, with intra‐radiologist percent agreement ranging from 62.1% to 87.4% and simple kappa ranging from 0.19 to 0.69 across individual radiologists. Our findings underscore the need for additional evaluation of the BI‐RADS breast density categorization system in clinical practice.

Reported Mammographic Density: Film-Screen versus Digital Acquisition

PURPOSE To test the hypothesis that American College of Radiology Breast Imaging Reporting and Data System (BI-RADS) categories for breast density reported by radiologists are lower when digital mammography is used than those reported when film-screen (FS) mammography is used. MATERIALS AND METHODS This study was institutional review board approved and HIPAA compliant. Demographic data, risk factors, and BI-RADS breast density categories were collected from five mammography registries that were part of the Breast Cancer Surveillance Consortium. Active, passive, or waiver of consent was obtained for all participants. Women aged 40 years and older who underwent at least two screening mammographic examinations less than 36 months apart between January 1, 2000, and December 31, 2009, were included. Women with prior breast cancer, augmentation, or use of agents known to affect density were excluded. The main sample included 89 639 women with both FS and digital mammograms. The comparison group included 259 046 women with two FS mammograms and 87 066 women with two digital mammograms. BI-RADS density was cross-tabulated according to the order in which the two types of mammogram were acquired and by the first versus second interpretation. RESULTS Regardless of acquisition method, the percentage of women with a change in density from one reading to the next was similar. Breast density was lower in 19.8% of the women who underwent FS before digital mammography and 17.1% of those who underwent digital before FS mammography. Similarly, lower density classifications were reported on the basis of the second mammographic examination regardless of acquisition method (15.8%-19.8%). The percentage of agreement between density readings was similar regardless of mammographic types paired (67.3%-71.0%). CONCLUSION The study results showed no difference in reported BI-RADS breast density categories according to acquisition method. Reported BI-RADS density categories may be useful in the development of breast cancer risk models in which FS, digital, or both acquisition methods are used.

One versus Two Breast Density Measures to Predict 5- and 10-Year Breast Cancer Risk

Background: One measure of Breast Imaging Reporting and Data System (BI-RADS) breast density improves 5-year breast cancer risk prediction, but the value of sequential measures is unknown. We determined whether two BI-RADS density measures improve the predictive accuracy of the Breast Cancer Surveillance Consortium 5-year risk model compared with one measure. Methods: We included 722,654 women of ages 35 to 74 years with two mammograms with BI-RADS density measures on average 1.8 years apart; 13,715 developed invasive breast cancer. We used Cox regression to estimate the relative hazards of breast cancer for age, race/ethnicity, family history of breast cancer, history of breast biopsy, and one or two density measures. We developed a risk prediction model by combining these estimates with 2000–2010 Surveillance, Epidemiology, and End Results incidence and 2010 vital statistics for competing risk of death. Results: The two-measure density model had marginally greater discriminatory accuracy than the one-measure model (AUC, 0.640 vs. 0.635). Of 18.6% of women (134,404 of 722,654) who decreased density categories, 15.4% (20,741 of 134,404) of women whose density decreased from heterogeneously or extremely dense to a lower density category with one other risk factor had a clinically meaningful increase in 5-year risk from <1.67% with the one-density model to ≥1.67% with the two-density model. Conclusion: The two-density model has similar overall discrimination to the one-density model for predicting 5-year breast cancer risk and improves risk classification for women with risk factors and a decrease in density. Impact: A two-density model should be considered for women whose density decreases when calculating breast cancer risk. Cancer Epidemiol Biomarkers Prev; 24(6); 889–97. ©2015 AACR.

Performance of diagnostic mammography differs in the United States and Denmark.

Diagnostic mammography is the primary imaging modality to diagnose breast cancer. However, few studies have evaluated variability in diagnostic mammography performance in communities, and none has done so between countries. We compared diagnostic mammography performance in community‐based settings in the United States and Denmark. The performance of 93,585 diagnostic mammograms from 180 facilities contributing data to the US Breast Cancer Surveillance Consortium (BCSC) from 1999 to 2001 was compared to that of all 51,313 diagnostic mammograms performed at Danish clinics in 2000. We used the imaging workups final assessment to determine sensitivity, specificity and an estimate of accuracy: area under the receiver‐operating characteristics (ROCs) curve (AUC). Diagnostic mammography had slightly higher sensitivity in the United States (85%) than in Denmark (82%). In contrast, it had higher specificity in Denmark (99%) than in the United States (93%). The AUC was high in both countries: 0.91 in United States and 0.95 in Denmark. Denmarks higher accuracy may result from supplementary ultrasound examinations, which are provided to 74% of Danish women but only 37% to 52% of US women. In addition, Danish mammography facilities specialize in either diagnosis or screening, possibly leading to greater diagnostic mammography expertise in facilities dedicated to symptomatic patients. Performance of community‐based diagnostic mammography settings varied markedly between the 2 countries, indicating that it can be further optimized.

A coarsened multinomial regression model for perinatal mother to child transmission of HIV

BackgroundIn trials designed to estimate rates of perinatal mother to child transmission of HIV, HIV assays are scheduled at multiple points in time. Still, infection status for some infants at some time points may be unknown, particularly when interim analyses are conducted.MethodsLogistic regression models are commonly used to estimate covariate-adjusted transmission rates, but their methods for handling missing data may be inadequate. Here we propose using coarsened multinomial regression models to estimate cumulative and conditional rates of HIV transmission. Through simulation, we compare the proposed models to standard logistic models in terms of bias, mean squared error, coverage probability, and power. We consider a range of treatment effect and visit process scenarios, while including imperfect sensitivity of the assay and contamination of the endpoint due to early breastfeeding transmission. We illustrate the approach through analysis of data from a clinical trial designed to prevent perinatal transmission.ResultsThe proposed cumulative and conditional models performed well when compared to their logistic counterparts. Performance of the proposed cumulative model was particularly strong under scenarios where treatment was assumed to increase the risk of in utero transmission but decrease the risk of intrapartum and overall perinatal transmission and under scenarios designed to represent interim analyses. Power to estimate intrapartum and perinatal transmission was consistently higher for the proposed models.ConclusionCoarsened multinomial regression models are preferred to standard logistic models for estimation of perinatal mother to child transmission of HIV, particularly when assays are missing or occur off-schedule for some infants.

Joint relative risks for estrogen receptor-positive breast cancer from a clinical model, polygenic risk score, and sex hormones

BackgroundModels that predict the risk of estrogen receptor (ER)-positive breast cancers may improve our ability to target chemoprevention. We investigated the contributions of sex hormones to the discrimination of the Breast Cancer Surveillance Consortium (BCSC) risk model and a polygenic risk score comprised of 83 single nucleotide polymorphisms.MethodsWe conducted a nested case-control study of 110 women with ER-positive breast cancers and 214 matched controls within a mammography screening cohort. Participants were postmenopausal and not on hormonal therapy. The associations of estradiol, estrone, testosterone, and sex hormone binding globulin with ER-positive breast cancer were evaluated using conditional logistic regression. We assessed the individual and combined discrimination of estradiol, the BCSC risk score, and polygenic risk score using the area under the receiver operating characteristic curve (AUROC).ResultsOf the sex hormones assessed, estradiol (OR 3.64, 95% CI 1.64–8.06 for top vs bottom quartile), and to a lesser degree estrone, was most strongly associated with ER-positive breast cancer in unadjusted analysis. The BCSC risk score (OR 1.32, 95% CI 1.00–1.75 per 1% increase) and polygenic risk score (OR 1.58, 95% CI 1.06–2.36 per standard deviation) were also associated with ER-positive cancers. A model containing the BCSC risk score, polygenic risk score, and estradiol levels showed good discrimination for ER-positive cancers (AUROC 0.72, 95% CI 0.65–0.79), representing a significant improvement over the BCSC risk score (AUROC 0.58, 95% CI 0.50–0.65).ConclusionAdding estradiol and a polygenic risk score to a clinical risk model improves discrimination for postmenopausal ER-positive breast cancers.

Trends in Hispanic and non-Hispanic white cesarean delivery rates on the US-Mexico border, 2000-2015

Background Cesarean delivery occurs in one in three US births and poses risks for mothers and infants. Hispanic cesarean rates were higher than non-Hispanic white rates in the US in 2016. In 2009, cesarean rates among Hispanics on the US-Mexico border exceeded rates among US Hispanics. Since 2009, rates have declined nationwide, but border Hispanic rates have not been studied. Objective To compare cesarean delivery rates and trends in Hispanics and non-Hispanic whites in border and nonborder counties of the four US border states before and after 2009. Study Design We used data from birth certificates to calculate percentages of cesarean deliveries among all births and births to low-risk nulliparous women during 2000–2015, and among births to low-risk women with and without a previous cesarean during 2009–2015. We calculated 95% confidence intervals around rates and used regular and piecewise linear regression to estimate trends for four ethnic-geographic subpopulations defined by combinations of Hispanic ethnicity and border-nonborder status. Results Of the four subpopulations, border Hispanic rates were highest every year for all cesarean outcomes. In 2015 they were 38.3% overall, 31.4% among low-risk nulliparous women, and 21.1% and 94.6% among low-risk women without and with a previous cesarean, respectively. Nonborder Hispanic rates in 2015 were lowest for all outcomes but repeat cesarean. Rates for all four subpopulations rose steadily during 2000–2009. Unlike rates for non-Hispanic whites, border and nonborder Hispanic rates did not decline post-2009. Most of the border Hispanic excess can be attributed to higher cesarean rates in Texas. Discussion Border Hispanic cesarean rates remain higher than those among other Hispanics and non-Hispanic whites in border states and show no signs of declining. This continuing disparity warrants further analysis using individual as well as hospital, environmental and other contextual factors to help target prevention measures.

A Bayesian hierarchical model for estimating and partitioning Bernstein polynomial density functions

A Bayesian hierarchical model for simultaneously estimating and partitioning probability density functions is presented. Individual density functions are flexibly modeled using Bernstein densities, which are mixtures of beta densities whose parameters depend only on the number of mixture components. A prior distribution is placed on the number of mixture components, and the mixture weights are expressed as increments of a distribution function G . A Dirichlet process prior is placed on G and the parameters of the Dirichlet process, the baseline distribution and the precision parameter, are treated as random. A mixture of a product of beta densities is used to partition subjects into groups, with subjects in the same group sharing information via a common baseline distribution. Inference is carried out using Markov chain Monte Carlo. A computing algorithm based on the constructive definition of the Dirichlet process is offered, for both a fixed number of groups and an unknown number of groups. When the number of groups is unknown, a birth-death algorithm is used to make inference regarding the number of groups. The model is demonstrated using radiologist-specific distributions of percent mammographic density.

Reproducibility of BI-RADS Breast Density Measures Among Community Radiologists: A Prospective Cohort Study: Reproducibility of BI-RADS Breast Density

Abstract:  Using data from the Vermont Breast Cancer Surveillance System (VBCSS), we studied the reproducibility of Breast Imaging Reporting and Data System (BI‐RADS) breast density among community radiologists interpreting mammograms in a cohort of 11,755 postmenopausal women. Radiologists interpreting two or more film‐screen screening or bilateral diagnostic mammograms for the same woman within a 3‐ to 24‐month period during 1996–2006 were eligible. We observed moderate‐to‐substantial overall intra‐rater agreement for use of BI‐RADS breast density in clinical practice, with an overall intra‐radiologist percent agreement of 77.2% (95% confidence interval (CI), 74.5–79.5%), an overall simple kappa of 0.58 (95% CI, 0.55–0.61), and an overall weighted kappa of 0.70 (95% CI, 0.68–0.73). Agreement exhibited by individual radiologists varied widely, with intra‐radiologist percent agreement ranging from 62.1% to 87.4% and simple kappa ranging from 0.19 to 0.69 across individual radiologists. Our findings underscore the need for additional evaluation of the BI‐RADS breast density categorization system in clinical practice.