e15197 Background: Bone is a common site of metastatic involvement in patients (pts) with PC. Bony metastases (mets) are often associated with SREs (spinal cord compression [SCC], pathologic fracture [PF], surgery to bone [SB], radiotherapy to bone [RT]). Skeletal complications cause significant morbidity and mortality. Current estimates of SRE risk come principally from randomized clinical trials. Information from routine clinical practice is limited. Methods: Using the tumor registry and electronic data stores at a large U.S. Midwest healthcare system that serves approximately 800,000 persons, we retrospectively identified all pts aged ≥18 yrs with primary PC and newly diagnosed bone mets between 1/1/95 and 12/31/09. Electronic medical records were reviewed by trained abstractors for evidence of SREs between date of bone mets diagnosis and death, loss to follow-up, or end of study for evidence of first SRE. Cumulative incidence of SREs was estimated in the presence of competing risk of death. Results: We identified a total of 420 men with primary PC and newly diagnosed bone mets; 42 pts had evidence of SREs at initial diagnosis of bone mets and were excluded from the analyses. Among the remaining 378 pts, mean (SD) age was 72.7 yrs (9.8 yrs); 38% were Caucasian and 58% were African-American. Median duration of follow-up after diagnosis of bone mets was 17.1 months (mos). At 12 mos, cumulative incidence of SREs was 31.6% (SCC, 6.1%; PF, 15.0%; SCC and/or PF, 19.1%; SB, 3.9%; RT, 24.4%) (Table). Corresponding figures at 24 mos were 45.3% (SCC, 12.5%; PF, 22.2%; SCC and/or PF, 30.2%; SB, 6.2%; RT, 34.9%). Relatively few pts (14.6%) received intravenous bisphosphonates prior to SRE. Conclusions: Pts with PC in routine clinical practice are at high risk of SREs following initial diagnosis of bone mets. [Table: see text]
Eighteen to twenty percent of breast cancer tumors show abnormal amplification of the Human Epidermal growth factor Receptor 2 (HER2) gene and increased expression of the associated protein. HER2 amplification is associated with rapid tumor proliferation and shorter disease-free and overall survival. Because women with HER2 amplification are more likely to benefit from treatment with the drug trastuzumab, testing for HER2 is recommended to guide therapy. However, little is known about use of HER2 testing in real-world settings. This study examined uptake, use, appropriateness of HER2 testing, and the relationship between HER2 test results and treatment decisions.We assessed electronic data from 3,634 patients with invasive breast cancer diagnosed from 1998 to 2007 in a large integrated health system. We collected data on patient and tumor characteristics, HER2 testing status, test results, and trastuzumab treatment.From 1998 to 2000, the percent of patients who underwent HER2 evaluation increased from 12 to 94%; <3% of women with ductal carcinoma in situ, for whom HER2 testing is not recommended, were tested. Trastuzumab use increased 5-fold after 2004, when guidelines expanded to include recommending adjuvant treatment for early-stage breast cancer in addition to metastatic treatment. Ninety-five percent of women receiving trastuzumab had a positive HER2 result. After 2004, 55% of women with invasive breast cancer and overexpression of HER2 received trastuzumab treatment; this ranged from 44% of women with localized breast cancer to 80% of women with distant metastatic disease.These findings illustrate appropriate and effective implementation of a HER2 testing strategy in a managed care setting.
Abstract Double-edged “Soil”: Stromal Microenvironment in Breast Cancer Development Mustapha Abubakar, MD, PhD1; Shaoqi Fan, MPH1; Máire A. Duggan, MD, FRCPC2; Ruth M. Pfeiffer, PhD1; Scott Lawrence, M.S.3; Kathryn Richert-Boe, MD4; Andrew G. Glass, MD4; Teresa M. Kimes, MS4; Jonine D. Figueroa, PhD, MPH5; Thomas E. Rohan, MBBS, PhD6; Gretchen L. Gierach, PhD, MPH1. Affiliations 1Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institute of Health (NIH), USA2Department of Pathology and Laboratory Medicine, University of Calgary, Calgary, T2N2Y9, Alberta, Canada3Molecular and Digital Pathology Laboratory, Cancer Genomics Research Laboratory, Leidos Biomedical Research, Inc., Frederick, MD 217024Kaiser Permanente Center for Health Research, Portland, Oregon 5Usher Institute of Population Health Sciences and Informatics, The University of Edinburgh, Scotland, UK6Department of Epidemiology and Population Health, Albert Einstein College of Medicine, Bronx, New York, 10461 Deceased Background Over a century after Paget’s postulation of the “seed” and “soil” hypothesis of breast cancer metastasis, understanding of the role of the “soil” (i.e., local or distant stromal microenvironment) in supporting and promoting the growth and dissemination of “seed” tumor cells has increased considerably. However, the role of the stromal microenvironment in early stages of breast cancer development, including tissue origins of etiologic heterogeneity, remains poorly understood. To date, the prevailing model of breast cancer development involves a linear sequence of epithelial changes that begin from normal epithelium, progress to flat epithelial atypia (FEA), develop into atypical ductal hyperplasia, transform into ductal carcinoma in situ (DCIS), and culminate in the emergence of invasive breast cancer. An alternative model proposes the evolution of FEA from usual ductal hyperplasia, but this is not supported by a clear genetic link between the two. Notably, established models of breast cancer development are predicated almost exclusively on sequences of epithelial changes. Although recent efforts have shed light on the significance of stromal microenvironment in DCIS to invasive breast cancer progression, its role in breast cancer development following benign breast disease (BBD) is less well-studied. MethodsIn this case-control study, nested within a cohort of 15,395 BBD patients from Kaiser Permanente Northwest (1970-2012) who were followed for subsequent development of invasive breast cancer, we evaluated archival diagnostic, formalin-fixed and paraffin-embedded tissue blocks using high-accuracy machine learning algorithms for the detailed characterization of stromal microenvironment on digitized H&E-stained breast biopsy sections. Stromal phenotypes, including total stromal-to-epithelial ratio (TSER), dense stromal-to-epithelial ratio (DSER), loose stromal-to-epithelial ratio (LSER) and stromal cellular density (SCD) were defined based on the distributions of total stromal area, dense (mostly fibrous/collagenized, inter-lobular) stroma, loose (mostly pale, intra-lobular or remodeled) stroma, and stromal cellularity, respectively. Relationships between stromal features and invasive breast cancer incidence through 2012 were assessed in multivariable conditional logistic regression models adjusted for BBD histological classification, body mass index, menopausal status/menopausal hormone therapy use, parity and age at first live birth (AFLB), family history of breast cancer (FHBC), oophorectomy, and mammographic density. Analyses were performed overall and by BBD histologic classification. We also evaluated associations between stromal features and breast cancer risk according to tumor characteristics that define divergent etiologic pathways, namely estrogen receptor (ER) expression and histologic grade. Results The current analysis is comprised of 486 cases and 487 controls, representing 95% of the case-control study population, for whom digitized H&E-stained sections were suitable for image analysis. The median age at diagnosis was 51.4 (range=18-86) years. ~55% of the participants were either overweight (30.1%) or obese (24.5%) at BBD diagnosis, and 69% had non-proliferative disease (NPD), 28% proliferative disease (PD) without atypia, and 3% atypical hyperplasia (AH). 13% of BBD biopsies contained simple fibroadenoma, ~2% complex fibroadenoma, 8% sclerosing adenosis, 5% radial scar, and 14% columnar cell lesions. The median (range) values (%) of total, dense, and loose connective tissue stroma were 39.3% (0.6-89.9%), 25.1% (0.1-84.5%), and 8.6% (0.2-59.0%), respectively. On average, BBD lesions contained ~6 times more stroma than epithelium. Average TSER, DSER, LSER, and SCD were 6.3, 4.4, 1.8, and 7.5%, respectively. Overall, increasing TSER was associated with decreasing risk of breast cancer [OR(95% CI)Q4 vs Q1=0.51(0.32, 0.82); p-trend=0.009]. The protective effect of TSER was, however, stronger in relation to DSER [OR(95% CI)Q4 vs Q1=0.48(0.29, 0.79); p-trend=0.007] than LSER [OR(95% CI)Q4 vs Q1=0.84(0.52, 1.36); p-trend=0.70]. Conversely, increasing SCD was statistically significantly associated with increasing breast cancer risk [OR(95% CI)Q4 vs Q1=2.21 (1.38, 3.56); p-trend=0.001]. Although findings were stronger among patients with NPD than PD, there was no heterogeneity in the association by BBD histology. Of the stromal features, DSER and SCD were most predictive of breast cancer risk but these were not independent of one another. To test their joint association with risk, we combined categories [low (<25th percentile), intermediate (25th–75th percentile), and high (>75th percentile) for each variable] in a composite, stromal disruption (SD), variable as follows: 1) no SD (high DSER and low SCD); 2) minimal disruption (high DSER and intermediate SCD, or vice versa); 3) moderate SD (intermediate DSER and high SCD, or vice versa); and 4) substantial SD (low DSER and high SCD). BBD patients with moderate [OR(95% CI)=1.74(1.01, 2.99)] or substantial [OR(95% CI)=2.70(1.51, 4.84)] SD were at statistically significantly elevated risk of breast cancer than those with no SD. Younger women, those with proliferative BBD, parous and AFLB <30 years, positive FHBC, absent involution, and being postmenopausal were statistically significantly more likely to develop BBDs with substantial SD than those with no SD. Substantial SD was associated with elevated risk of both ER+ [OR(95% CI)=1.97(1.16, 3.36)] and ER- [OR(95% CI)=2.09(0.77, 5.69)] breast cancer. In terms of grade, substantial SD was more strongly associated with risks of high [OR(95% CI)=3.17(1.28, 7.85)] and intermediate [OR(95% CI)=2.30(1.10, 4.83)] than low [OR(95% CI)=1.61(0.74, 3.50)] grade tumors overall (p-heterogeneity=0.44). This association was stronger among patients with NPD [OR(95% CI) substantial SD vs no SD=5.75(2.04, 16.09); 2.83(1.15, 6.97); and 1.36(0.53, 3.46) for high, intermediate, and low grade tumors, respectively (p-heterogeneity=0.03)]. Because AH has been implicated in the development of ER+/low grade but not ER+/high grade tumors, we further evaluated the role of AH and SD in ER+ breast cancer risk: contrasting patterns were observed in associations between AH, substantial SD, and risk of ER+ tumors defined by levels of histologic grade. While AH more strongly predisposed to risk of low [OR(95% CI)=6.32(1.09, 20.08)] than high [OR(95% CI)=1.04(0.10, 11.24)] grade ER+ tumors, substantial SD more strongly predisposed to risk of high [OR(95% CI)=5.28(1.54, 18.10)] than low [OR(95% CI)=1.52(0.76, 3.06)] grade ER+ tumors.Conclusion Components of the stromal microenvironment in BBD showed disparate associations with breast cancer risk factors and risk of subsequent invasive breast cancer. In particular, increasing ratio of dense (mostly fibrous/collagenized and inter-lobular), but not loose (mostly pale/myxoid, intra-lobular, remodeled), connective tissue stroma to epithelium was strongly associated with reduced risk of breast cancer. Conversely, increasing stromal cellularity was associated with increasing risk of breast cancer. In combination, decreasing amounts of dense stroma and concomitant increase in loose stroma, epithelial volume, and stromal cellularity resulted in a stromal disruption phenotype that was strongly associated with increased breast cancer risk overall, but particularly of aggressive high grade tumors. These results were independent of BBD histologic diagnosis. Many of the observed risk factor associations with stromal microenvironment features were consistent with their breast cancer risk relationships, suggesting that stromal changes may reflect cumulative exposure to breast cancer risk factors. These findings provide new etiologic insights into stromal role in breast cancer risk, including tissue origins of breast cancer etiologic heterogeneity, with the potential to aid risk stratification and clinical decision-making for BBD patients. Citation Format: Mustapha Abubakar, Shaoqi Fan, Maire A. Duggan, Ruth M. Pfeiffer, Scott Lawrence, Kathryn Richert-Boe, Andrew Glass, Teresa M. Kimes, Jonine D. Figueroa, Thomas E. Rohan, Gretchen L. Gierach. Double-edged “soil”: Stromal microenvironment in breast cancer development [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr SY25-02.
Abstract Purpose: To compare deep learning (DL) approaches applied to hematoxylin and eosin (H&E)-stained whole slide images (WSIs) from women with benign breast disease (BBD) to predict risk of developing invasive breast cancer (BC). Method: Two deep convolutional neural networks (CNNs) based on a customized 16-layer CNN (known as VGG-16 by Visual Geometry Group, University of Oxford) and an automated CNN (Google’s AutoML) were trained using H&E-stained WSIs to identify distinct histological features on diagnostic BBD biopsies that characterize BBD patients who were (cases, n=347) and were not (controls, n=347) subsequently diagnosed with invasive BC. The CNNs consisted of multiple convolutions, max pooling, fully connected, etc., layers. To incorporate our data into the VGG network, we customized the network architecture and hyperparameters to enhance the classification performances. For AutoML, we used the system's default network with standard hyperparameters. The trained model was then tested on a held-out set of 140 patients (70 cases and 70 controls). The quantitative performance was evaluated using accuracy (ACC), sensitivity (SE), precision (PR), area under the receiver operating characteristic curve (AUROC), etc. For qualitative results, we generated heatmaps using weights and feature maps from the final convolution layer of our customized CNN. Heatmaps were superimposed onto original H&E images to highlight different unique features (such as pattern, texture, color, and morphology). Results: We found both deep learning methods to demonstrate remarkable ability in predicting case-control status in the held-out set (AUROC= 90% and 89% for customized CNN and AutoML, respectively). However, our customized CNN outperformed AutoML in terms of ACC (83.57% (95% confidence interval (CI): 76-89%) vs 77.86% (95%CI: 70-84%), respectively); SE (82.85% (95%CI: 72-91%) vs 77.86% (95%CI: 70-84%), respectively); PR (84.05% (95%CI: 73-92%) vs 81.97% (95%CI: 70-91%), respectively); F1 score (83.45% (95%CI: 76-89%) vs 76.34% (95%CI: 68-83%), respectively); as well as error rates (0.16% (95%CI: 0.11-0.24%) vs 0.22% (95%CI: 0.16-0.30%), respectively). Heatmaps revealed specific stromal and epithelial features that were distinct between case and control images. Conclusion: By using routinely available H&E-stained WSIs, we developed a customized CNN that outperformed AutoML in distinguishing future BC cases from controls in a BBD population. The qualitative results identified stromal and epithelial regions in the BBD biopsies that were highly predictive of being a case versus control and vice versa thereby providing etiologic clues into breast cancer development following BBD. Future research will focus on leveraging DL to better understand the histologic basis of BBD progression to invasive BC. Citation Format: Monjoy Saha, Mustapha Abubakar, Thomas E. Rohan, Ruth M. Pfeiffer, Máire A. Duggan, Kathryn Richert-Boe, Jonine D. Figueroa, Jonas S. Almeida, Gretchen L. Gierach. Comparison of deep learning approaches applied to hematoxylin and eosin-stained whole slide images from women with benign breast disease to predict risk of developing invasive breast cancer. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 5393.
e19630 Background: Bony metastases (mets) are a common source of morbidity in patients (pts) with cancer, cause spinal cord compression (SCC), pathological fracture (PF), and bone pain, and often require radiotherapy (RT) and/or surgery to bone (SB). Methods: Using the tumor registry and electronic data stores at a large U.S. Midwest healthcare system that serves approximately 800,000 persons, we retrospectively identified all pts aged ≥18 years with primary breast, lung, or prostate cancer diagnosed between 1995 and 2009. Registry and electronic medical records were then used to identify pts with diagnosis of bone mets at initial cancer diagnosis or at recurrence. Trained technicians reviewed medical records for occurrence of SCC, PF, RT and SB—outcomes that have been collectively referred to as SREs. Cumulative incidence of these events was calculated in the presence of competing risk of death. Results: We identified 378 pts with breast cancer and bone mets, 272 with lung cancer and bone mets, and 420 with prostate cancer and bone mets. SREs were present at initial diagnosis of bone mets in 23% of breast cancer pts, 24% of lung cancer pts, and 10% of prostate cancer pts (Table). At 12 months, cumulative incidence of SREs was 57.3% for breast cancer (SCC, 5.1%; PF, 37.9%; SCC and/or PF, 40.2%; SB, 5.9%; RT, 27.4%), 62.7% for lung cancer (SCC, 8.9%; PF, 37.8%; SCC and/or PF, 43.3%; SB, 3.5%; RT, 32.4%), and 38.4% for prostate cancer (SCC, 7.9%; PF, 21.3%; SCC and/or PF, 26.7%; SB, 4.0%; RT, 22.9%). Use of bisphosphonates was largely confined to pts with breast cancer. Conclusions: Though breast, lung, and prostate cancers differ considerably in presentation, clinical course, and treatment, SREs are a common and serious problem in all three cancers among patients with bone mets. [Table: see text]
Lung and colorectal cancer are the first and second lead. ing causes of death from cancer among Americans. It is only natural that there should be considerable interest in trying to prevent deaths from these diseases through early detection and treatment. However, widespread screening is expensive and logistically difficult, therefore the ability of screening tests to reduce cancer mortality must be well established before they are put into common practice. The basic principles of screening, including the characteristics of diseases and tests suitable for screening, biases that can affect uncontrolled studies of screening tests, and costs of screening are discussed in part 2 of this article.1 In this article, we critically analyze the data regarding the effectiveness of screening tests in reducing mortality from lung and colorectal cancer and discuss the controversies surrounding their use. As in our previous article, the quality of the data for each test will
Prostate cancer is the second most common cause of death from cancer in men in the United States. Digital rectal examination is the oldest and most commonly used screening test for prostate cancer, but as yet there are no studies which demonstrate its effectiveness.A case-control study was conducted among members of a large health maintenance organisation to estimate the effect of screening digital rectal examination on mortality from prostate cancer. 150 men, aged 40-84 when cancer was diagnosed, who developed fatal prostate cancer, and 299 male controls matched for age who did not die from prostate cancer were studied. A history of screening digital rectal examination during the 10 years before the date on which cancer was-diagnosed was determined from medical records.A similar proportion of men who died from prostate cancer and controls had undergone at least one screening digital rectal examination during the 10 year interval (odds ratio = 0.84, 95% confidence interval 0.48 to 1.46). Similar results were obtained when a shorter interval (such as five years before diagnosis) during which screening histories were evaluated was considered, or in analyses in which men with a history of benign prostatic hypertrophy were excluded.The data suggest that screening digital rectal examination does not reduce mortality from prostate cancer to any appreciable degree.
Abstract Purpose Benign breast disease (BBD) is a strong breast cancer risk factor but identifying patients that might develop invasive breast cancer remains a challenge. Methods By applying machine-learning to digitized H&E-stained biopsies and computer-assisted thresholding to mammograms obtained circa BBD diagnosis, we generated quantitative tissue composition metrics and determined their association with future invasive breast cancer diagnosis. Archival breast biopsies and mammograms were obtained for women (18-86 years) in a case-control study, nested within a cohort of 15,395 BBD patients from Kaiser Permanente Northwest (1970-2012), followed through mid-2015. Cases (n=514) who developed incident invasive breast cancer and controls (n=514) were matched on BBD diagnosis age and plan membership duration. Results Increasing epithelial area on the BBD biopsy was associated with increasing breast cancer risk [Odds ratio(OR) 95% confidence interval(CI) Q4 vs Q1 =1.85(1.13-3.04);P trend =0.02]. Conversely, increasing stroma was associated with decreased risk in non-proliferative, but not proliferative, BBD (P heterogeneity =0.002). Increasing epithelium-to-stroma proportion [OR(95%CI) Q4 vs Q1 =2.06(1.28-3.33);P trend =0.002] and percent mammographic density (MBD) [OR(95%CI) Q4 vs Q1 =2.20(1.20-4.03);P trend =0.01] were independently and strongly predictive of increased breast cancer risk. In combination, women with high epithelium-to-stroma proportion/high MBD had substantially higher risk than those with low epithelium-to-stroma proportion/low MBD [OR(95%CI)=2.27(1.27-4.06);P trend =0.005], particularly among women with non-proliferative [P trend =0.01] versus proliferative [P trend =0.33] BBD. Conclusion Among BBD patients, increasing epithelium-to-stroma proportion on BBD biopsies and percent MBD at BBD diagnosis were independently and jointly associated with increasing breast cancer risk. These findings were particularly striking for women with non-proliferative disease (comprising ∼70% of all BBD patients), for whom relevant predictive biomarkers are lacking.
