Palakal, Maya

Purpose: Women whose mammographic breast density declines within 12–18 months of initiating tamoxifen for chemoprevention or adjuvant treatment show improved therapeutic responses compared with those whose density is unchanged. We tested whether measuring changes in sound speed (a surrogate of breast density) using ultrasound tomography (UST) could enable rapid identification of favorable responses to tamoxifen. Methods: We evaluated serial density measures at baseline and at 1 to 3, 4 to 6, and 12+ months among 74 women (aged 30–70 years) following initiation of tamoxifen for clinical indications, including an elevated risk of breast cancer (20%) and diagnoses of in situ (39%) or invasive (40%) breast carcinoma, enrolled at Karmanos Cancer Institute and Henry Ford Health System (Detroit, MI, USA). For comparison, we evaluated an untreated group with screen negative mammography and frequency-matched on age, race, and menopausal status (n = 150), at baseline and 12 months. Paired t-tests were used to assess differences in UST sound speed over time and between tamoxifen-treated and untreated patients. Results: Sound speed declined steadily over the 12 month period among patients receiving tamoxifen (mean (SD): −3.0 (8.2) m/s; p = 0.001), whereas density remained unchanged in the untreated group (mean (SD): 0.4 (7.1) m/s; p = 0.75 (relative change between groups: p = 0.0009)). In the tamoxifen group, we observed significant sound speed reductions as early as 4–6 months after tamoxifen initiation (mean (SD): −2.1 (6.8) m/s; p = 0.008). Sound speed reductions were greatest among premenopausal patients (P-interaction = 0.0002) and those in the middle and upper tertiles of baseline sound speed (P-interaction = 0.002). Conclusions: UST can image rapid declines in sound speed following initiation of tamoxifen. Given that sound speed and mammographic density are correlated, we propose that UST breast imaging may capture early responses to tamoxifen, which in turn may have utility in predicting therapeutic efficacy.

Abstract Background: High percent mammographic density (MD), which reflects the relative fibroglandular tissue content of the breast, is one of the strongest breast cancer risk factors; however, the pathologic mediators of this risk are unknown. We hypothesize that analysis of breast tissue sections using deep learning approaches may characterize histologic features that underpin risk associated with high MD. Methods: Non-targeted H&E stained breast tissue sections of diagnostic image-guided breast biopsies were evaluated among 588 women enrolled following an abnormal mammogram in the Breast Radiology Evaluation and Study of Tissues (BREAST) Stamp Project (2007-2010). Overall volumetric percent MD for the biopsied breast and localized volumetric percent MD surrounding the biopsy site were determined for each participant. A deep convolutional neural network (CNN) model was trained to identify and quantitatively assess breast epithelial, stroma and fat tissue and their organizational and spatial arrangements. Least absolute shrinkage and selection operator (Lasso) regression was used to determine relationships between MD measures and pathological features. To ensure reliability of the model, a cross-validation strategy was employed to build and assess the performance of the fitted model. Finally, Spearman correlation coefficients were estimated to test the association between the predicted density values by each model (predicting overall or localized MD) and the actual MD measurements. We report the average and standard deviation (SD) of the correlation coefficients. Results: In an independent validation set, the CNN model was 95.5% accurate in classifying epithelial, stromal and fat tissue. The mean (SD) correlations between the predicted model and the actual measurements for overall and localized MD were 0.70 (0.06) and 0.65 (0.06) respectively. The amount of stroma identified (normalized to tissue area) had the highest selection probability (P-value) by the Lasso model and thus the strongest positive relationship with MD (P-value >0.9 for each MD measurement). In contrast, the amount of normalized epithelial tissue was not related to MD (P-value=0.01 for each MD measurement). No association was observed for the total normalized fat area with MD (P-value <0.31 for each MD measurement). In addition, the number of distributed epithelial regions was positively associated, whereas the distance between epithelial regions was inversely associated with overall MD (P-value <0.87 and 0.62, respectively). Conclusions: These results show that greater stromal tissue amount and spatial distribution patterns of epithelial regions, rather than total epithelial amounts, had the strongest relationships with elevated MD. Future work will determine the relationship of these MD features with biopsy diagnosis. Citation Format: Maeve Mullooly, Babak Ehteshami Bejnordi, Maya Palakal, Pamela M. Vacek, Donald L. Weaver, John A. Shepherd, Bo Fan, Amir Pasha Mahmoudzadeh, Jeff Wang, Jason M. Johnson, Sally D. Herschorn, Brian L. Sprague, Ruth M. Pfeiffer, Louise A. Brinton, Mark E. Sherman, Andrew Beck, Gretchen L. Gierach. Application of convolutional neural networks to breast biopsies to uncover tissue correlates of mammographic breast density [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 4235. doi:10.1158/1538-7445.AM2017-4235

Terminal duct lobular units (TDLUs) are the predominant source of future breast cancers, and lack of TDLU involution (higher TDLU counts, higher acini count per TDLU and the product of the two) is a breast cancer risk factor. Numerous breast cancer susceptibility single nucleotide polymorphisms (SNPs) have been identified, but whether they are associated with TDLU involution is unknown. In a pooled analysis of 872 women from two studies, we investigated 62 established breast cancer SNPs and relationships with TDLU involution. Poisson regression models with robust variance were used to calculate adjusted per‐allele relative risks (with the non‐breast cancer risk allele as the referent) and 95% confidence intervals between TDLU measures and each SNP. All statistical tests were two‐sided; P < 0.05 was considered statistically significant. Overall, 36 SNPs (58.1%) were related to higher TDLU counts although this was not statistically significant (p = 0.25). Six of the 62 SNPs (9.7%) were nominally associated with at least one TDLU measure: rs616488 (PEX14), rs11242675 (FOXQ1) and rs6001930 (MKL1) were associated with higher TDLU count (p = 0.047, 0.045 and 0.031, respectively); rs1353747 (PDE4D) and rs6472903 (8q21.11) were associated with higher acini count per TDLU (p = 0.007 and 0.027, respectively); and rs1353747 (PDE4D) and rs204247 (RANBP9) were associated with the product of TDLU and acini counts (p = 0.024 and 0.017, respectively). Our findings suggest breast cancer SNPs may not strongly influence TDLU involution. Agnostic genome‐wide association studies of TDLU involution may provide new insights on its biologic underpinnings and breast cancer susceptibility.

Abstract Purpose: Expression of the tumor suppressor gene E-cadherin is diminished in lobular breast cancers and has been implicated in epithelial mesenchymal transition. We assessed risk factor associations for breast cancer stratified by low vs. high E-cadherin protein expression in a pooled analysis within the Breast Cancer Association Consortium (BCAC) studies. Methods: E-cadherin tumor tissue staining was performed centrally at the NCI on formalin-fixed paraffin-embedded tissue microarray (TMA) sections representing 6,010 breast cancer patients from 12 US and European BCAC studies. TMAs were digitally scanned and scored using the SlidePath Digital Image Hub (Leica Biosystems, Wetzlar, Germany). For 5,896 cancers with evaluable tumors, E-cadherin was visually scored as estimated percent of positive cells times stain intensity (0, 1+, 2+, 3+) (score range 0-300). E-cadherin low was defined as tumors with a score &lt; 100. Risk factor associations for low vs. high E-cadherin expressing tumors were evaluated by logistic regression, adjusted for age and study site, and stratified by ER status and histologic subtype. To assess the consistency in results by study, meta-analyses were performed using the random-effects model. All statistical tests were two-sided. Results: E-cadherin low cancers comprised 20% of tumors and were associated with lobular histology, well/moderately differentiated cancers, &gt; 2cm in size, and HER2-negative status (χ2, P &lt; 0.003 for all factors). E-cadherin low status was associated with family history of breast cancer (FH) (OR = 1.32, 95% CI = 1.09-1.60, P-het = 0.005) and ever use of menopausal hormones (OR = 1.26, 95% CI = 1.02-1.56, P-het = 0.03). Study specific meta-plots showed consistent effects across studies for menopausal hormone therapy (I2 = 0.0%, p = 0.64); however, E-cadherin loss by FH did show evidence of heterogeneity by study (I2 = 54%, P = 0.03). Differences in E-cadherin expression remained significant for FH, and menopausal hormone use when further adjusted for ER (P &lt; 0.05). We also found relationships with E-cadherin loss to vary by BMI, number of live births, age at first birth, and oral contraceptive use in stratified analysis by ER status and histologic subtype. Conclusion: This large pooled analysis shows that breast cancer risk factor associations may differ by E-cadherin expression independent of ER status, suggesting that it may represent a marker of etiologic heterogeneity. Citation Format: Hisani N. Horne, Hannah Oh, Mark E. Sherman, Maya Palakal, Stephen H. Hewitt, Marjanka Schmidt, Javier Benitez, Roger Milne, Hermann Brenner, Heli Nevanlinna, Arto Mannermaa, Georgia Chenevix-Trench, Fergus Couch, Peter Devilee, Diana Eccles, Maartje Hooning, Anthony J. Swerdlow, Nick Orr, Melissa A. Troester, Renata Cora, Paul D. Pharoah, Montserrat Garcia-Closas, Jonine D. Figueroa. Breast cancer risk factor associations by loss of E-cadherin tumor tissue expression: A pooled analysis of 5,896 cases in 12 studies from the Breast Cancer Association Consortium (BCAC). [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 3451.

Abstract Purpose: Breast biopsies with benign findings (benign breast disease [BBD]) outnumber biopsies revealing in-situ or invasive cancer 4:1, providing an opportunity to identify histopathological features that predict breast cancer risk. Greater degrees of involution of terminal duct lobular units (TDLUs), the structures within the breast that produce milk and the primary source of breast cancer precursors, have been inversely associated with breast cancer risk among women with BBD. We recently developed measures to quantitate levels of TDLU involution, which demonstrate high inter/intra pathologist reproducibility. Here we assessed whether TDLU counts/100mm2 and median TDLU span (microns), two measures inversely related to degree of TDLU involution, are associated with subsequent breast cancer risk among women with BBD. Methods: From the Mayo BBD cohort (n = 9,376), we evaluated benign biopsies from 99 women who later developed breast cancer (cases) and 145 age-matched controls who did not develop breast cancer. Digitized images of biopsy sections were reviewed to enumerate TDLUs/mm2 and measure median TDLU span (microns) for up to ten normal TDLUs. Breast cancer risk factors were available from questionnaires or medical records. To assess associations with breast cancer risk, subjects were categorized into quartile levels of TDLU counts and median TDLU spans, based on data from controls. Odds ratios (ORs) and 95% confidence intervals (CI) were estimated using multivariable conditional logistic regression models, adjusted for histologic type of BBD, family history of breast cancer and menopausal hormone use. Results: Compared to controls, women who developed breast cancer had higher median number of TDLU counts/100mm2 (28 vs 20, p = 0.03) and larger TDLU spans (300 vs 267 microns, p = 0.14). In multivariable models higher TDLU counts (quartile4 vs. quartile1, OR = 2.44, 95%CI = 0.96-6.19, ptrend = 0.02) and larger TDLU span measures (quartile4 vs. quartile1, OR = 2.83, 95%CI = 1.13-7.06, ptrend = 0.03) were associated with subsequent diagnosis of breast cancer. Combinatorial metrics of TDLU counts with median TDLU span measures identified women at higher risk; specifically, women above the median for both TDLU span and TDLU counts had an OR = 3.75 (95%CI = 1.40-10.00, ptrend = 0.008), compared with women below the median for TDLU span and TDLU counts. Conclusion: These data show that lack of TDLU involution, as measured by increased persistence of TDLU counts, larger median TDLU spans and cross-classification using these measures was associated with increased breast cancer risk among women with BBD, extending prior work in this cohort based on other TDLU involution metrics. Future studies to identify determinants of TDLU involution, its association with breast cancer risk and its potential as an intermediate endpoint in prevention studies warrant consideration. Citation Format: Jonine D. Figueroa, Ruth Pfeiffer, Maya Palakal, Amy C. Degnim, Derek Radisky, Lynn C. Hartmann, Marlene Frost, Melody L. Stallings Mann, Louise A. Brinton, Daphne Papathomas, Daniel Visscher, Mark E. Sherman. Standardized measures of lobular involution and subsequent breast cancer risk among women with benign breast disease. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 4682. doi:10.1158/1538-7445.AM2015-4682