Vachon, Celine M.

Abstract Background: Mammographic density (MD) is a strong breast cancer risk factor. We previously reported associations of percent mammographic density (PMD) with larger and node-positive tumors across all ages, and estrogen receptor (ER)–negative status among women ages <55 years. To provide insight into these associations, we examined the components of PMD [dense area (DA) and nondense area (NDA)] with breast cancer subtypes. Methods: Data were pooled from six studies including 4,095 breast cancers and 8,558 controls. DA and NDA were assessed from digitized film-screen mammograms and standardized across studies. Breast cancer odds by density phenotypes and age according to histopathologic characteristics and receptor status were calculated using polytomous logistic regression. Results: DA was associated with increased breast cancer risk [OR for quartiles: 0.65, 1.00 (Ref), 1.22, 1.55; Ptrend <0.001] and NDA was associated with decreased risk [ORs for quartiles: 1.39, 1.00 (Ref), 0.88, 0.72; Ptrend <0.001] across all ages and invasive tumor characteristics. There were significant trends in the magnitude of associations of both DA and NDA with breast cancer by increasing tumor size (Ptrend < 0.001) but no differences by nodal status. Among women <55 years, DA was more strongly associated with increased risk of ER+ versus ER− tumors (Phet = 0.02), while NDA was more strongly associated with decreased risk of ER− versus ER+ tumors (Phet = 0.03). Conclusions: DA and NDA have differential associations with ER+ versus ER− tumors that vary by age. Impact: DA and NDA are important to consider when developing age- and subtype-specific risk models. Cancer Epidemiol Biomarkers Prev; 24(5); 798–809. ©2015 AACR.

Purpose Optimal early detection and prevention for breast cancer depend on accurate identification of women at increased risk. We present a risk prediction model that incorporates histologic features of biopsy tissues from women with benign breast disease (BBD) and compare its performance to the Breast Cancer Risk Assessment Tool (BCRAT). Methods We estimated the age-specific incidence of breast cancer and death from the Mayo BBD cohort and then combined these estimates with a relative risk model derived from 377 patient cases with breast cancer and 734 matched controls sampled from the Mayo BBD cohort to develop the BBD–to–breast cancer (BBD-BC) risk assessment tool. We validated the model using an independent set of 378 patient cases with breast cancer and 728 matched controls from the Mayo BBD cohort and compared the risk predictions from our model with those from the BCRAT. Results The BBD-BC model predicts the probability of breast cancer in women with BBD using tissue-based and other risk factors. The concordance statistic from the BBD-BC model was 0.665 in the model development series and 0.629 in the validation series; these values were higher than those from the BCRAT (0.567 and 0.472, respectively). The BCRAT significantly underpredicted breast cancer risk after benign biopsy (P = .004), whereas the BBD-BC predictions were appropriately calibrated to observed cancers (P = .247). Conclusion We developed a model using both demographic and histologic features to predict breast cancer risk in women with BBD. Our model more accurately classifies a woman's breast cancer risk after a benign biopsy than the BCRAT.

Purpose Recent advances in DNA sequencing have led to the development of breast cancer susceptibility gene panels for germline genetic testing of patients. We assessed the frequency of mutations in 17 predisposition genes, including BRCA1 and BRCA2, in a large cohort of patients with triple-negative breast cancer (TNBC) unselected for family history of breast or ovarian cancer to determine the utility of germline genetic testing for those with TNBC. Patients and Methods Patients with TNBC (N = 1,824) unselected for family history of breast or ovarian cancer were recruited through 12 studies, and germline DNA was sequenced to identify mutations. Results Deleterious mutations were identified in 14.6% of all patients. Of these, 11.2% had mutations in the BRCA1 (8.5%) and BRCA2 (2.7%) genes. Deleterious mutations in 15 other predisposition genes were detected in 3.7% of patients, with the majority observed in genes involved in homologous recombination, including PALB2 (1.2%) and BARD1, RAD51D, RAD51C, and BRIP1 (0.3% to 0.5%). Patients with TNBC with mutations were diagnosed at an earlier age (P < .001) and had higher-grade tumors (P = .01) than those without mutations. Conclusion Deleterious mutations in predisposition genes are present at high frequency in patients with TNBC unselected for family history of cancer. Mutation prevalence estimates suggest that patients with TNBC, regardless of age at diagnosis or family history of cancer, should be considered for germline genetic testing of BRCA1 and BRCA2. Although mutations in other predisposition genes are observed among patients with TNBC, better cancer risk estimates are needed before these mutations are used for clinical risk assessment in relatives.

A large genotyping project within the Breast Cancer Association Consortium (BCAC) recently identified 41 associations between single nucleotide polymorphisms (SNPs) and overall breast cancer (BC) risk. We investigated whether the effects of these 41 SNPs, as well as six SNPs associated with estrogen receptor (ER) negative BC risk are modified by 13 environmental risk factors for BC. Data from 22 studies participating in BCAC were pooled, comprising up to 26,633 cases and 30,119 controls. Interactions between SNPs and environmental factors were evaluated using an empirical Bayes‐type shrinkage estimator. Six SNPs showed interactions with associated p ‐values ( p int ) <1.1 × 10 −3 . None of the observed interactions was significant after accounting for multiple testing. The Bayesian False Discovery Probability was used to rank the findings, which indicated three interactions as being noteworthy at 1% prior probability of interaction. SNP rs6828523 was associated with increased ER‐negative BC risk in women ≥170 cm (OR = 1.22, p = 0.017), but inversely associated with ER‐negative BC risk in women <160 cm (OR = 0.83, p = 0.039, p int = 1.9 × 10 −4 ). The inverse association between rs4808801 and overall BC risk was stronger for women who had had four or more pregnancies (OR = 0.85, p = 2.0 × 10 −4 ), and absent in women who had had just one (OR = 0.96, p = 0.19, p int = 6.1 × 10 −4 ). SNP rs11242675 was inversely associated with overall BC risk in never/former smokers (OR = 0.93, p = 2.8 × 10 −5 ), but no association was observed in current smokers (OR = 1.07, p = 0.14, p int = 3.4 × 10 −4 ). In conclusion, recently identified BC susceptibility loci are not strongly modified by established risk factors and the observed potential interactions require confirmation in independent studies.

Abstract Background: Chronic lymphocytic leukemia/small lymphocytic lymphoma (CLL) is the most common adult leukemia in the United States. Prior studies in Olmsted County, Minnesota, showed a CLL age and sex adjusted incidence rate of 6.6 per 100,000 during 1985-89. Subsequently, the definition of CLL is now based on the 2008 International Workshop on CLL (IWCLL) Guidelines. Individuals having a CLL immunophenotype by flow cytometry are classified as CLL if the absolute B-cell count (ABC) is &gt;5 x 109/L or classified as monoclonal B-cell lymphocytosis (MBL) if ABC is &lt;5 x 109/L. To assess the impact of these changes on the incidence of CLL and clinically discovered MBL, we conducted an incidence study in Olmsted County for the years 2000-2010. Methods Using diagnostic indexes available through the Rochester Epidemiology Project and the Mayo CLL database, we identified all CLL and clinically identified MBL cases from 2000-2010. Incidence rates and demographics were compared between the older National Cancer Institute-Working Groups 1996 guidelines (NCI-WG 96) defined as absolute lymphocyte count (ALC)&gt; 5 x 109/L and the IWCLL 2008 guidelines defined as ABC &gt; 5 x 109/L. Incidence rates were age and sex adjusted using the 2010 US white population. Time to treatment (TTT) survival analyses was from time of diagnosis to date of first treatment or last follow-up and Cox regression analysis was used. Results Using the NCI-WG 96 classification, there were 8 MBL cases and 115 CLLs in whom 61% were Rai stage 0, 34% were Rai stage 1-2, and 5% were Rai stage 3-4. Using the IWCLL 2008 classification, there were 44 MBL cases and 79 CLLs in whom 43% Rai stage 0, 49% were Rai stage 1-2 and 8% Rai stage 3-4. Median ALC at diagnosis was 8.1 x 109/L for CLL and 5.9 x 109/L for MBL. Median ABC count at diagnosis was 7.1 x 109/L for CLL and 2.8 x 109/L for MBL. The age and sex adjusted incidence rate (per 100,000) for CLL for 2000-2010 was 10.0 using NCI-WG 96 and 6.8 using IWCLL 2008. Rates for MBL were 0.66 using NCI-WG 96 and 3.5 using IWCLL 2008. The median TTT for CLL was 9.2 years using NCI-WG 96 and 6.5 years by IWCLL 2008. Discussion This study demonstrates that reclassification of CLL using the IWCLL 2008 guidelines decreases CLL incidence relative to the NCI-WG 96 criteria. For the first time, this population based study also establishes an incidence rate for clinically identified MBL. Approximately 50% of previously labeled Rai 0 CLL are now reclassified as MBL. Re-classification increases the percent of Rai stage 1-2 and decreases the percent of Rai 0 cases at diagnosis. Reclassification also substantially shortens the median TTT for newly diagnosed CLL cases classified by the 2008 criteria relative to the NCI-WG 96 criteria. In summary, the IWCLL 2008 reclassification has significant implications for incidence studies, staging, and prognosis of CLL and MBL. Citation Format: Aaron D. Norman, Timothy G. Call, Curtis A. Hanson, Neil E. Kay, Clive S. Zent, Wei Ding, James R. Cerhan, Sara J. Achenbach, Kari G. Rabe, Celine M. Vachon, Emily J. Hallberg, Tait D. Shanafelt, Susan L. Slager. Incidence of chronic lymphocytic leukemia and monoclonal B-cell lymphocytosis in Olmsted county, 2000-2010: impact of the 2008 International Workshop on CLL Guidelines. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 2287. doi:10.1158/1538-7445.AM2013-2287

Abstract Background: Women with benign breast disease (BBD) have increased breast cancer risk, and histologic features of the lesion stratify individual risk. We hypothesized that the concentration and type of terminal duct lobular units in BBD tissue is associated with future breast cancer risk. Methods: A nested case-control series of 770 breast cancer cases and 1273 age- and year of biopsy-matched controls was identified within a cohort of 13,647 women with BBD from 1967-2001 (median follow-up 15.3 years). One representative H&E slide from each biopsy was reviewed by a single pathologist, who recorded total # of lobules present, and # of fibrocystic lobules (total and # with epithelial proliferation). Samples were also evaluated for lobular involution and primary BBD histology. Lobule density (LD) was calculated as # of lobules/cm2 in each tissue section. Breast cancer associations with LD's (total, fibrocystic, proliferative) were examined with conditional logistic regression. Odds ratios are reported per 1 SD increase in LD. Results: Each LD measure was significantly associated with involution and primary BBD histology (p&lt;0.0001 for each). Fibrocystic (OR 1.17, 95% CI 1.06-1.28) and proliferative (OR 1.37, 95% CI 1.16-1.61) LD were each associated with case status (p≤0.001) but total LD did not reach significance (OR 1.09, 95% CI 1.0-1.20; p=0.06). LD associations were somewhat attenuated after adjusting for involution (total OR 1.04 p=0.41; fibrocystic OR 1.13 p=0.02; proliferative OR 1.28 p=0.002) and primary BBD histology (total OR 1.04 p=0.49; fibrocystic OR 1.08 p=0.15; proliferative OR 1.18 p=0.07). Conclusions: The density of proliferative lobules in benign breast biopsy tissue is related to other BBD histologic features but may provide additional breast cancer risk stratification among women with proliferative BBD. Additional research will more fully evaluate the ability of this promising novel measure to predict breast cancer risk. Lobule Density Measures (lobules per square cm) All values are Medians Total Lobule Density Fibrocystic Lobule Density Proliferative Lobule Density Overall 17.5 4.5 0.7 Cases 18.1 4.9 1.3 Controls 17.0 4.2 0.4 Primary BBD Histology Nonproliferative disease 16 3.3 0 Proliferative disease 18.6 5.8 2.7 Atypical hyperplasia 20.9 6.2 3.3 Involution Category None 28.7 5.3 1.0 Partial 19.0 5 1.2 Complete 8.5 2.1 0 Citation Format: Amy C. Degnim, Lynn C. Hartmann, Tanya L. Hoskin, Marlene H. Frost, V Shane Pankratz, Derek C. Radisky, Lorelle L. Benetti, Celine M. Vachon, Teresa M. Allers, Robert A. Vierkant, Daniel W. Visscher. Density of breast lobules in benign breast tissue and association with future breast cancer risk. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 155. doi:10.1158/1538-7445.AM2013-155

BACKGROUNDThe 1996 National Cancer Institute Working Group (NCI‐WG 96) guidelines classified disease in individuals who had a B‐cell clone with chronic lymphocytic leukemia (CLL) immunophenotype as CLL if their absolute lymphocyte count was ≥5 × 109/L. The 2008 International Workshop on CLL guidelines (IWCLL 2008) classified disease as CLL if the absolute B‐cell count was ≥5 × 109/L or as monoclonal B‐cell lymphocytosis (MBL) if the absolute B‐cell count was <5 × 109/L. The objective of the current study of Olmsted County, Minnesota, was to assess the effects of these changes on incidence rates and presentation from 2000 to 2010.METHODSUsing diagnostic indices available through the Rochester Epidemiology Project and the Mayo CLL database, the authors identified all patients with newly diagnosed CLL and high‐count MBL from 2000 to 2010. Age‐specific and sex‐specific incidence rates were determined.RESULTSAccording to NCI‐WG 96 criteria, there were 115 patients with CLL and 8 patients with MBL during the period studied. Using the IWCLL 2008 classification, there were 79 patients with CLL and 40 patients with MBL. Rai stage distribution (low risk, intermediate risk, and high risk) using NCI‐WG 96 criteria was 60.9%, 33.9%, and 5.2%, respectively, compared with 43%, 49.4%, and 7.6%, respectively, using IWCLL 2008 criteria. The age‐adjusted and sex‐adjusted incidence rates (per 100,000) for CLL and MBL were 10.0 and 0.66, respectively, using NCI‐WG 96 criteria versus 6.8 and 3.5, respectively, using IWCLL 2008 criteria. The median time to treatment according to NCI‐WG 96 criteria was 9.2 years versus 6.5 years with IWCLL 2008 criteria.CONCLUSIONSUse of the IWCLL 2008 guidelines reduced the incidence of CLL, altered the distribution of initial Rai stage at diagnosis, and shortened the median time to treatment. Cancer 2014;120:2000–2005. © 2014 American Cancer Society.

Abstract Part of the substantial unexplained familial aggregation of breast cancer may be due to interactions between common variants, but few studies have had adequate statistical power to detect interactions of realistic magnitude. We aimed to assess all two-way interactions in breast cancer susceptibility between 70 917 single nucleotide polymorphisms (SNPs) selected primarily based on prior evidence of a marginal effect. Thirty-eight international studies contributed data for 46 450 breast cancer cases and 42 461 controls of European origin as part of a multi-consortium project (COGS). First, SNPs were preselected based on evidence (P &lt; 0.01) of a per-allele main effect, and all two-way combinations of those were evaluated by a per-allele (1 d.f.) test for interaction using logistic regression. Second, all 2.5 billion possible two-SNP combinations were evaluated using Boolean operation-based screening and testing, and SNP pairs with the strongest evidence of interaction (P &lt; 10−4) were selected for more careful assessment by logistic regression. Under the first approach, 3277 SNPs were preselected, but an evaluation of all possible two-SNP combinations (1 d.f.) identified no interactions at P &lt; 10−8. Results from the second analytic approach were consistent with those from the first (P &gt; 10−10). In summary, we observed little evidence of two-way SNP interactions in breast cancer susceptibility, despite the large number of SNPs with potential marginal effects considered and the very large sample size. This finding may have important implications for risk prediction, simplifying the modelling required. Further comprehensive, large-scale genome-wide interaction studies may identify novel interacting loci if the inherent logistic and computational challenges can be overcome.