Vivian Viallon

Prediction of breast cancer risk by genetic risk factors, overall and by hormone receptor status.

Objective There is increasing interest in adding common genetic variants identified through genome wide association studies (GWAS) to breast cancer risk prediction models. First results from such models showed modest benefits in terms of risk discrimination. Heterogeneity of breast cancer as defined by hormone-receptor status has not been considered in this context. In this study we investigated the predictive capacity of 32 GWAS-detected common variants for breast cancer risk, alone and in combination with classical risk factors, and for tumours with different hormone receptor status. Material and methods Within the Breast and Prostate Cancer Cohort Consortium, we analysed 6009 invasive breast cancer cases and 7827 matched controls of European ancestry, with data on classical breast cancer risk factors and 32 common gene variants identified through GWAS. Discriminatory ability with respect to breast cancer of specific hormone receptor-status was assessed with the age adjusted and cohort-adjusted concordance statistic (AUROCa). Absolute risk scores were calculated with external reference data. Integrated discrimination improvement was used to measure improvements in risk prediction. Results We found a small but steady increase in discriminatory ability with increasing numbers of genetic variants included in the model (difference in AUROCa going from 2.7% to 4%). Discriminatory ability for all models varied strongly by hormone receptor status. Discussion and conclusions Adding information on common polymorphisms provides small but statistically significant improvements in the quality of breast cancer risk prediction models. We consistently observed better performance for receptor-positive cases, but the gain in discriminatory quality is not sufficient for clinical application.

Predictiveness curves in virtual screening

BackgroundIn the present work, we aim to transfer to the field of virtual screening the predictiveness curve, a metric that has been advocated in clinical epidemiology. The literature describes the use of predictiveness curves to evaluate the performances of biological markers to formulate diagnoses, prognoses and assess disease risks, assess the fit of risk models, and estimate the clinical utility of a model when applied to a population. Similarly, we use logistic regression models to calculate activity probabilities related to the scores that the compounds obtained in virtual screening experiments. The predictiveness curve can provide an intuitive and graphical tool to compare the predictive power of virtual screening methods.ResultsSimilarly to ROC curves, predictiveness curves are functions of the distribution of the scores and provide a common scale for the evaluation of virtual screening methods. Contrarily to ROC curves, the dispersion of the scores is well described by predictiveness curves. This property allows the quantification of the predictive performance of virtual screening methods on a fraction of a given molecular dataset and makes the predictiveness curve an efficient tool to address the early recognition problem. To this last end, we introduce the use of the total gain and partial total gain to quantify recognition and early recognition of active compounds attributed to the variations of the scores obtained with virtual screening methods. Additionally to its usefulness in the evaluation of virtual screening methods, predictiveness curves can be used to define optimal score thresholds for the selection of compounds to be tested experimentally in a drug discovery program. We illustrate the use of predictiveness curves as a complement to ROC on the results of a virtual screening of the Directory of Useful Decoys datasets using three different methods (Surflex-dock, ICM, Autodock Vina).ConclusionThe predictiveness curves cover different aspects of the predictive power of the scores, allowing a detailed evaluation of the performance of virtual screening methods. We believe predictiveness curves efficiently complete the set of tools available for the analysis of virtual screening results.

Discrimination measures for survival outcomes: Connection between the AUC and the predictiveness curve

Finding out biomarkers and building risk scores to predict the occurrence of survival outcomes is a major concern of clinical epidemiology, and so is the evaluation of prognostic models. In this paper, we are concerned with the estimation of the time-dependent AUC--area under the receiver-operating curve--which naturally extends standard AUC to the setting of survival outcomes and enables to evaluate the discriminative power of prognostic models. We establish a simple and useful relation between the predictiveness curve and the time-dependent AUC--AUC(t). This relation confirms that the predictiveness curve is the key concept for evaluating calibration and discrimination of prognostic models. It also highlights that accurate estimates of the conditional absolute risk function should yield accurate estimates for AUC(t). From this observation, we derive several estimators for AUC(t) relying on distinct estimators of the conditional absolute risk function. An empirical study was conducted to compare our estimators with the existing ones and assess the effect of model misspecification--when estimating the conditional absolute risk function--on the AUC(t) estimation. We further illustrate the methodology on the Mayo PBC and the VA lung cancer data sets.

Population Attributable and Preventable Fractions: Cancer Risk Factor Surveillance, and Cancer Policy Projection

The proportions of new cancer cases and deaths that are caused by exposure to risk factors and that could be prevented are key statistics for public health policy and planning. This paper summarizes the methodologies for estimating, challenges in the analysis of, and utility of, population attributable and preventable fractions for cancers caused by major risk factors such as tobacco smoking, dietary factors, high body fat, physical inactivity, alcohol consumption, infectious agents, occupational exposure, air pollution, sun exposure, and insufficient breastfeeding. For population attributable and preventable fractions, evidence of a causal relationship between a risk factor and cancer, outcome (such as incidence and mortality), exposure distribution, relative risk, theoretical-minimum-risk, and counterfactual scenarios need to be clearly defined and congruent. Despite limitations of the methodology and the data used for estimations, the population attributable and preventable fractions are a useful tool for public health policy and planning.

A treelet transform analysis to relate nutrient patterns to the risk of hormonal receptor-defined breast cancer in the European Prospective Investigation into Cancer and Nutrition (EPIC)

OBJECTIVE Pattern analysis has emerged as a tool to depict the role of multiple nutrients/foods in relation to health outcomes. The present study aimed at extracting nutrient patterns with respect to breast cancer (BC) aetiology. DESIGN Nutrient patterns were derived with treelet transform (TT) and related to BC risk. TT was applied to twenty-three log-transformed nutrient densities from dietary questionnaires. Hazard ratios (HR) and 95 % confidence intervals computed using Cox proportional hazards models quantified the association between quintiles of nutrient pattern scores and risk of overall BC, and by hormonal receptor and menopausal status. Principal component analysis was applied for comparison. SETTING The European Prospective Investigation into Cancer and Nutrition (EPIC). SUBJECTS Women (n 334 850) from the EPIC study. RESULTS The first TT component (TC1) highlighted a pattern rich in nutrients found in animal foods loading on cholesterol, protein, retinol, vitamins B12 and D, while the second TT component (TC2) reflected a diet rich in β-carotene, riboflavin, thiamin, vitamins C and B6, fibre, Fe, Ca, K, Mg, P and folate. While TC1 was not associated with BC risk, TC2 was inversely associated with BC risk overall (HRQ5 v. Q1=0·89, 95 % CI 0·83, 0·95, P trend<0·01) and showed a significantly lower risk in oestrogen receptor-positive (HRQ5 v. Q1=0·89, 95 % CI 0·81, 0·98, P trend=0·02) and progesterone receptor-positive tumours (HRQ5 v. Q1=0·87, 95 % CI 0·77, 0·98, P trend<0·01). CONCLUSIONS TT produces readily interpretable sparse components explaining similar amounts of variation as principal component analysis. Our results suggest that participants with a nutrient pattern high in micronutrients found in vegetables, fruits and cereals had a lower risk of BC.

Fractions of fatal crashes attributable to speeding: Evolution for the period 2001–2010 in France

Road safety is a major concern in the West, especially in France. Among all the established risk factors for fatal crashes, speed is specific in two ways: every road-user is exposed to it, and it increases not only crash rates but also the severity of crashes. Thus, speed regulation is of primary importance in road-safety policy and has also generated much public debate. To contribute to this debate, we constructed a power-model which relates the number of fatal crashes to speed raised to the power four. Despite its simplicity, this model fitted the data well. Notably, it enabled the fractions of fatal crashes attributable to various levels of speeding to be estimated. Data for secondary roads over the period 2001-2010 showed that the fraction of fatal crashes attributable to high-level speeding (>20kph over the speed limit) decreased from 25% to 6% and that attributable to medium-level speeding (10-20kph over the speed limit) decreased from 13% to 9%, whereas that attributable to low-level speeding progressively increased from 7% to 13%. Similar trends were observed on main roads. These results highlight the effectiveness of the speed regulation policies introduced during the study period with respect to high-level speeding. They also suggest that future policy should focus on low and medium-level speeding in order further to reduce road deaths significantly, since these levels now correspond to the major fraction of fatal crashes.

On the robustness of the generalized fused lasso to prior specifications

Using networks as prior knowledge to guide model selection is a way to reach structured sparsity. In particular, the fused lasso that was originally designed to penalize differences of coefficients corresponding to successive features has been generalized to handle features whose effects are structured according to a given network. As any prior information, the network provided in the penalty may contain misleading edges that connect coefficients whose difference is not zero, and the extent to which the performance of the method depend on the suitability of the graph has never been clearly assessed. In this work we investigate the theoretical and empirical properties of the adaptive generalized fused lasso in the context of generalized linear models. In the fixed

Empirical comparison study of approximate methods for structure selection in binary graphical models

p