Background: Comprehensive breast cancer risk prediction models enable identifying and targeting women at high-risk, while reducing interventions in those at low-risk. Breast cancer risk prediction models used in clinical practice have low discriminatory accuracy (0.53–0.64). Machine learning (ML) offers an alternative approach to standard prediction modeling that may address current limitations and improve accuracy of those tools. The purpose of this study was to compare the discriminatory accuracy of ML-based estimates against a pair of established methods—the Breast Cancer Risk Assessment Tool (BCRAT) and Breast and Ovarian Analysis of Disease Incidence and Carrier Estimation Algorithm (BOADICEA) models. Methods: We quantified and compared the performance of eight different ML methods to the performance of BCRAT and BOADICEA using eight simulated datasets and two retrospective samples: a random population-based sample of U.S. breast cancer patients and their cancer-free female relatives (N = 1143), and a clinical sample of Swiss breast cancer patients and cancer-free women seeking genetic evaluation and/or testing (N = 2481). Results: Predictive accuracy (AU-ROC curve) reached 88.28% using ML-Adaptive Boosting and 88.89% using ML-random forest versus 62.40% with BCRAT for the U.S. population-based sample. Predictive accuracy reached 90.17% using ML-adaptive boosting and 89.32% using ML-Markov chain Monte Carlo generalized linear mixed model versus 59.31% with BOADICEA for the Swiss clinic-based sample. Conclusions: There was a striking improvement in the accuracy of classification of women with and without breast cancer achieved with ML algorithms compared to the state-of-the-art model-based approaches. High-accuracy prediction techniques are important in personalized medicine because they facilitate stratification of prevention strategies and individualized clinical management.

Background:: Elevated levels of the cell cycle protein cyclin E, and low levels of its inhibitor, p27^Kip1, have been associated with a poor prognosis following breast cancer. Some studies have found that germline mutations in the breast cancer susceptibility gene, BRCA1, are also associated with an inferior survival rate. The relationship between cyclin E/p27^Kip1 levels, BRCA1 status and outcome has not been studied in detail. Patients and methods:: We analyzed a historical cohort of 288 Ashkenazi Jewish women who were diagnosed with breast cancer between 1980 and 1995 and were previously tested for BRCA1/2 mutations. Protein levels of cyclin E and p27^Kip1 were assessed by immunohistochemistry. Breast cancer-specific survival (BCSS) was the main outcome measured. Results:: The median follow-up was 8 years. Thirty tumors carried germline BRCA1 mutations. These tumors were more likely to have high cyclin E protein levels [odds ratio (OR) 9.5; P <0.001] and low p27^Kip1 protein levels (OR 2.8; P=0.03) than tumors from patients without BRCA1/2 mutations. High cyclin E expression level was the strongest predictor of BRCA1 germline mutations (multivariate OR 4.7; P=0.004). On univariate analysis, high cyclin E protein levels [relative risk (RR) 2.6; P <0.001] and low p27^Kip1 protein levels (RR 2.3; P=0.006) were significant prognostic factors for a poorer BCSS. In Cox multivariate models, high cyclin E levels remained an independent indicator of poor outcome only in the subgroup of patients who did not receive chemotherapy (P=0.002). Conclusions:: In this ethnically restricted cohort, a high level of cyclin E is a characteristic of BRCA1-related breast cancer, and is a marker of poor prognosis following breast cancer, particularly in the absence of adjuvant chemotherapy.

The genetic basis for nonmedullary forms of thyroid cancer (NMTC) is less well established than that of medullary thyroid cancer. However, epidemiological and family studies suggest that a proportion of NMTC may be due to inherited predisposition. To estimate the familial risk of thyroid cancer, we conducted a hospital-based case-control study at the Princess Margaret Hospital in Toronto, Ontario, Canada, and at 2 university hospitals in Montréal, Québec, Canada. We obtained pedigrees from 339 unselected patients diagnosed with NMTC and from 319 unaffected ethnically matched controls. Family histories of cancer were obtained from the cases and controls for 3292 first degree relatives of cases and controls. Seventeen cases (5.0%) and 2 controls (0.6%) reported at least one first degree relative with thyroid cancer. In relatives of patients with thyroid cancer, the incidence of any type of cancer (including NMTC) was 38% higher than in relatives of controls (incidence rate ratio, 1.4; 95% confidence interval, 1.1–1.7). The relative risk for thyroid cancer was 10-fold higher in relatives of cancer patients than in controls (incidence rate ratio, 10.3; 95% confidence interval, 2.2–47.6). Our findings suggest that hereditary or other familial factors are important in a small proportion of NMTC. Molecular studies are needed to determine the genetic basis of cancer susceptibility in these families.

Breast, colorectal, ovarian, and endometrial cancers constitute approximately 30% of newly diagnosed cancer cases in Switzerland, affecting more than 12,000 individuals annually. Hundreds of these patients are likely to carry germline pathogenic variants associated with hereditary breast ovarian cancer (HBOC) or Lynch syndrome (LS). Genetic services (counseling and testing) for hereditary susceptibility to cancer can prevent many cancer diagnoses and deaths through early identification and risk management. Cascade screening is the systematic identification and testing of relatives of a known mutation carrier. It determines whether asymptomatic relatives also carry the known variant, needing management options to reduce future harmful outcomes. Specific aims of the CASCADE study are to (1) survey index cases with HBOC or LS from clinic-based genetic testing records and determine their current cancer status and surveillance practices, needs for coordination of medical care, psychosocial needs, patient-provider and patient-family communication, quality of life, and willingness to serve as advocates for cancer genetic services to blood relatives, (2) survey first- and second-degree relatives and first-cousins identified from pedigrees or family history records of HBOC and LS index cases and determine their current cancer and mutation status, cancer surveillance practices, needs for coordination of medical care, barriers and facilitators to using cancer genetic services, psychosocial needs, patient-provider and patient-family communication, quality of life, and willingness to participate in a study designed to increase use of cancer genetic services, and (3) explore the influence of patient-provider communication about genetic cancer risk on patient-family communication and the acceptability of a family-based communication, coping, and decision support intervention with focus group(s) of mutation carriers and relatives. CASCADE is a longitudinal study using surveys (online or paper/pencil) and focus groups, designed to elicit factors that enhance cascade genetic testing for HBOC and LS in Switzerland. Repeated observations are the optimal way for assessing these outcomes. Focus groups will examine barriers in patient-provider and patient-family communication, and the acceptability of a family-based communication, coping, and decision-support intervention. The survey will be developed in English, translated into three languages (German, French, and Italian), and back-translated into English, except for scales with validated versions in these languages. Descriptive analyses will include calculating means, standard deviations, frequencies, and percentages of variables and participant descriptors. Bivariate analyses (Pearson correlations, chi-square test for differences in proportions, and t test for differences in means) will assess associations between demographics and clinical characteristics. Regression analyses will incorporate generalized estimating equations for pairing index cases with their relatives and explore whether predictors are in direct, mediating, or moderating relationship to an outcome. Focus group data will be transcribed verbatim and analyzed for common themes. Robust evidence from basic science and descriptive population-based studies in Switzerland support the necessity of cascade screening for genetic predisposition to HBOC and LS. CASCADE is designed to address translation of this knowledge into public health interventions. ClinicalTrials.gov NCT03124212; https://clinicaltrials.gov/ct2/show/NCT03124212 (Archived by WebCite at http://www.webcitation.org/6tKZnNDBt).