Search

Background: Genome-wide association studies identified novel breast cancer susceptibility variants that could be used to predict breast cancer in asymptomatic women. This review and modeling study aimed to investigate the current and potential predictive performance of genetic risk models.Methods: Genotypes and disease status were simulated for a population of 10,000 women. Genetic risk models were constructed from polymorphisms from meta-analysis including, in separate scenarios, all polymorphisms or statistically significant polymorphisms only. We additionally investigated the magnitude of the odds ratios (OR) for 1 to 100 hypothetical polymorphisms that would be needed to achieve similar discriminative accuracy as available prediction models [modeled range of area under the receiver operating characteristic curve (AUC) 0.70–0.80].Results: Of the 96 polymorphisms that had been investigated in meta-analyses, 41 showed significant associations. AUC was 0.68 for the genetic risk model based on all 96 polymorphisms and 0.67 for the 41 significant polymorphisms. Addition of 50 additional variants, each with risk allele frequencies of 0.30, requires per-allele ORs of 1.2 to increase this AUC to 0.70, 1.3 to increase AUC to 0.75, and 1.5 to increase AUC to 0.80. To achieve AUC of 0.80, even 100 additional variants would need per-allele ORs of 1.3 to 1.7, depending on risk allele frequencies.Conclusion: The predictive ability of genetic risk models in breast cancer has the potential to become comparable to that of current breast cancer risk models.Impact: Risk prediction based on low susceptibility variants becomes a realistic tool in prevention of nonfamilial breast cancer. Cancer Epidemiol Biomarkers Prev; 20(1); 9–22. ©2011 AACR.

Hypothesis/Commentary for This Article from Genome-Based Prediction of Breast Cancer Risk in the General Population: A Modeling Study Based on Meta-Analyses of Genetic Associations

Here, we argue that polygenic risk scores (PRSs) are different epistemic objects as compared to other biomarkers such as blood pressure or sodium level. While the latter two may be subject to variation, measured inaccurately or interpreted in various ways, blood flow has a pressure and sodium is available in a concentration that can be quantified and visualised. In stark contrast, PRSs are calculated, compiled or constructed through the statistical assemblage of genetic variants. How researchers frame and name PRSs has consequences for how we interpret and value their results. We distinguish between the tangible and inferential understanding of PRS and the corresponding languages of measurement and computation, respectively. The conflation of these frames obscures important questions we need to ask: what PRS seeks to represent, whether current ways of ‘doing PRS’ are optimal and responsible, and upon what we base the credibility of PRS-based knowledge claims.

Background We assessed the clinical utility of a first-degree breast cancer family history and polygenic risk score (PRS) to inform screening decisions among women aged 30-50 years. Methods Two established breast cancer models evaluated digital mammography screening strategies in the 1985 US birth cohort by risk groups defined by family history and PRS based on 313 single nucleotide polymorphisms. Strategies varied in initiation age (30, 35, 40, 45, and 50 years) and interval (annual, hybrid, biennial, triennial). The benefits (breast cancer deaths averted, life-years gained) and harms (false-positive mammograms, overdiagnoses) were compared with those seen with 3 established screening guidelines. Results Women with a breast cancer family history who initiated biennial screening at age 40 years (vs 50 years) had a 36% (model range = 29%-40%) increase in life-years gained and 20% (model range = 16%-24%) more breast cancer deaths averted, but 21% (model range = 17%-23%) more overdiagnoses and 63% (model range = 62%-64%) more false positives. Screening tailored to PRS vs biennial screening from 50 to 74 years had smaller positive effects on life-years gained (20%) and breast cancer deaths averted (11%) but also smaller increases in overdiagnoses (10%) and false positives (26%). Combined use of family history and PRS vs biennial screening from 50 to 74 years had the greatest increase in life-years gained (29%) and breast cancer deaths averted (18%). Conclusions Our results suggest that breast cancer family history and PRS could guide screening decisions before age 50 years among women at increased risk for breast cancer but expected increases in overdiagnoses and false positives should be expected.

Polygenic risk scores (PRSs) have become the standard for quantifying genetic liability in the prediction of disease risks. PRSs are generally constructed as weighted sum scores of risk alleles using effect sizes from genome-wide association studies as their weights. The construction of PRSs is being improved with more appropriate selection of independent single-nucleotide polymorphisms (SNPs) and optimized estimation of their weights but is rarely reflected upon from a theoretical perspective, focusing on the validity of the risk score. Borrowing from psychometrics, this paper discusses the validity of PRSs and introduces the three main types of validity that are considered in the evaluation of tests and measurements: construct, content, and criterion validity. This introduction is followed by a discussion of three topics that challenge the validity of PRS, namely, their claimed independence of clinical risk factors, the consequences of relaxing SNP inclusion thresholds and the selection of SNP weights. This discussion of the validity of PRS reminds us that we need to keep questioning if weighted sums of risk alleles are measuring what we think they are in the various scenarios in which PRSs are used and that we need to keep exploring alternative modeling strategies that might better reflect the underlying biological pathways.

Purpose: The area under the receiver operating characteristic curve (AUC) is commonly used for evaluating the improvement of polygenic risk models and increasingly assessed together with the net reclassification improvement (NRI) and integrated discrimination improvement (IDI). We evaluated how researchers described and interpreted AUC, NRI, and IDI when simultaneously assessed. Methods: We reviewed how researchers described definitions of AUC, NRI, and IDI and how they computed each metric. Next, we reviewed how the increment in AUC, NRI, and IDI were interpreted, and how the overall conclusion about the improvement of the risk model was reached. Results: AUC, NRI, and IDI were correctly defined in 63, 70, and 0% of the articles. All statistically significant values and almost half of the nonsignificant were interpreted as indicative of improvement, irrespective of the values of the metrics. Also, small, nonsignificant changes in the AUC were interpreted as indication of improvement when NRI and IDI were statistically significant. Conclusion: Researchers have insufficient knowledge about how to interpret the various metrics for the assessment of the predictive performance of polygenic risk models and rely on the statistical significance for their interpretation. A better understanding is needed to achieve more meaningful interpretation of polygenic prediction studies.

Polygenic risk scores (PRSs), which often aggregate results from genome-wide association studies, can bridge the gap between initial discovery efforts and clinical applications for the estimation of disease risk using genetics. However, there is notable heterogeneity in the application and reporting of these risk scores, which hinders the translation of PRSs into clinical care. Here, in a collaboration between the Clinical Genome Resource (ClinGen) Complex Disease Working Group and the Polygenic Score (PGS) Catalog, we present the Polygenic Risk Score Reporting Standards (PRS-RS), in which we update the Genetic Risk Prediction Studies (GRIPS) Statement to reflect the present state of the field. Drawing on the input of experts in epidemiology, statistics, disease-specific applications, implementation and policy, this comprehensive reporting framework defines the minimal information that is needed to interpret and evaluate PRSs, especially with respect to downstream clinical applications. Items span detailed descriptions of study populations, statistical methods for the development and validation of PRSs and considerations for the potential limitations of these scores. In addition, we emphasize the need for data availability and transparency, and we encourage researchers to deposit and share PRSs through the PGS Catalog to facilitate reproducibility and comparative benchmarking. By providing these criteria in a structured format that builds on existing standards and ontologies, the use of this framework in publishing PRSs will facilitate translation into clinical care and progress towards defining best practice.

The area under the receiver operating characteristic (ROC) curve (AUC) is commonly used for assessing the discriminative ability of prediction models even though the measure is criticized for being clinically irrelevant and lacking an intuitive interpretation. Every tutorial explains how the coordinates of the ROC curve are obtained from the risk distributions of diseased and non-diseased individuals, but it has not become common sense that therewith the ROC plot is just another way of presenting these risk distributions. We show how the ROC curve is an alternative way to present risk distributions of diseased and non-diseased individuals and how the shape of the ROC curve informs about the overlap of the risk distributions. For example, ROC curves are rounded when the prediction model included variables with similar effect on disease risk and have an angle when, for example, one binary risk factor has a stronger effect; and ROC curves are stepped rather than smooth when the sample size or incidence is low, when the prediction model is based on a relatively small set of categorical predictors. This alternative perspective on the ROC plot invalidates most purported limitations of the AUC and attributes others to the underlying risk distributions. AUC is a measure of the discriminative ability of prediction models. The assessment of prediction models should be supplemented with other metrics to assess their clinical utility.

Polygenic risk scores (PRS), often aggregating the results from genome-wide association studies, can bridge the gap between the initial variant discovery efforts and disease risk estimation for clinical applications. However, there is remarkable heterogeneity in the reporting of these risk scores due to a lack of adherence to reporting standards and no accepted standards suited for the current state of PRS development and application. This lack of adherence and best practices hinders the translation of PRS into clinical care. The ClinGen Complex Disease Working Group, in a collaboration with the Polygenic Score (PGS) Catalog, have developed a novel PRS Reporting Statement (PRS-RS), updating previous standards to the current state of the field and to enable downstream utility. Drawing upon experts in epidemiology, statistics, disease-specific applications, implementation, and policy, this 23-item reporting framework defines the minimal information needed to interpret and evaluate a PRS, especially with respect to any downstream clinical applications. Items span detailed descriptions of the study population (recruitment method, key demographics, inclusion/exclusion criteria, and phenotype definition), statistical methods for both PRS development and validation, and considerations for potential limitations of the published risk score and downstream clinical utility. Additionally, emphasis has been placed on data availability and transparency to facilitate reproducibility and benchmarking against other PRS, such as deposition in the publicly available PGS Catalog (www.PGScatalog.org). By providing these criteria in a structured format that builds upon existing standards and ontologies, the use of this framework in publishing PRS will facilitate translation of PRS into clinical care and progress towards defining best practices.SummaryIn recent years, polygenic risk scores (PRS) have become an increasingly studied tool to capture the genome-wide liability underlying many human traits and diseases, hoping to better inform an individual’s genetic risk. However, a lack of tailored reporting standards has hindered the translation of this important tool into clinical and public health practice with the heterogeneous underreporting of details necessary for benchmarking and reproducibility. To address this gap, the ClinGen Complex Disease Working Group and Polygenic Score (PGS) Catalog have collaborated to develop the 23-item Polygenic Risk Score Reporting Statement (PRS-RS). This framework provides the minimal information expected of authors to promote the validity, transparency, and reproducibility of PRS by requiring authors to detail the study population, statistical methods, and potential clinical utility of a published score. The widespread adoption of this framework will encourage rigorous methodological consideration and facilitate benchmarking to ensure high quality scores are translated into the clinic.

Background: Although uniform colonoscopy screening reduces colorectal cancer (CRC) mortality, risk-based screening may be more efficient. We investigated whether CRC screening based on polygenic risk is a cost-effective alternative to current uniform screening, and if not, under what conditions it would be. Methods: The MISCAN-Colon model was used to simulate a hypothetical cohort of US 40-year-olds. Uniform screening was modeled as colonoscopy screening at ages 50, 60, and 70 years. For risk-stratified screening, individuals underwent polygenic testing with current and potential future discriminatory performance (area under the receiver-operating curve [AUC] of 0.60 and 0.65-0.80, respectively). Polygenic testing results were used to create risk groups, for which colonoscopy screening was optimized by varying the start age (40-60 years), end age (70-85 years), and interval (1-20 years). Results: With current discriminatory performance, optimal screening ranged from once-only colonoscopy at age 60 years for the lowest-risk group to six colonoscopies at ages 40-80 years for the highest-risk group. While maintaining the same health benefits, risk-stratified screening increased costs by $59 per person. Risk-stratified screening could become cost-effective if the AUC value would increase beyond 0.65, the price per polygenic test would drop to less than $141, or risk-stratified screening would lead to a 5% increase in screening participation. Conclusions: Currently, CRC screening based on polygenic risk is unlikely to be cost-effective compared with uniform screening. This is expected to change with a greater than 0.05 increase in AUC value, a greater than 30% reduction in polygenic testing costs, or a greater than 5% increase in adherence with screening. National Cancer Institute as part of the Cancer Intervention and Surveillance Modeling Network (CISNET) [U01CA152959] This work was supported by the National Cancer Institute (grant number U01CA152959) as part of the Cancer Intervention and Surveillance Modeling Network (CISNET), with a supplement from the Evaluation of Genomic Applications in Practice and Prevention (EGAPP). Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the National Cancer Institute or Centers for Disease Control and Prevention.

Objective Adding risk factors to a prediction model often increases the area under the receiver operating characteristic curve (AUC) only slightly, particularly when the AUC of the model was already high. We investigated whether a risk factor that minimally improves the AUC may nevertheless improve the predictive ability of the model, assessed by integrated discrimination improvement (IDI). Study Design and Setting We simulated data sets with risk factors and event status for 100,000 hypothetical individuals and created prediction models with AUCs between 0.50 and 0.95. We added a single risk factor for which the effect was modeled as a certain odds ratio (OR 2, 4, 8) or AUC increment (ΔAUC 0.01, 0.02, 0.03). Results Across all AUC values of the baseline model, for a risk factor with the same OR, both ΔAUC and IDI were lower when the AUC of the baseline model was higher. When the increment in AUC was small (ΔAUC 0.01), the IDI was also small, except when the AUC of the baseline model was >0.90. Conclusion When the addition of a risk factor shows minimal improvement in AUC, predicted risks generally show minimal changes too. Updating risk models with strong risk factors may be informative for a subgroup of individuals, but not at the population level. The AUC may not be as insensitive as is frequently argued.

Direct-to-consumer genetic testing companies aim to predict the risks of complex diseases using proprietary algorithms. Companies keep algorithms as trade secrets for competitive advantage, but a market that thrives on the premise that customers can make their own decisions about genetic testing should respect customer autonomy and informed decision making and maximize opportunities for transparency. The algorithm itself is only one piece of the information that is deemed essential for understanding how prediction algorithms are developed and evaluated. Companies should be encouraged to disclose everything else, including the expected risk distribution of the algorithm when applied in the population, using a benchmark DNA dataset. A standardized presentation of information and risk distributions allows customers to compare test offers and scientists to verify whether the undisclosed algorithms could be valid. A new model of oversight in which stakeholders collaboratively keep a check on the commercial market is needed.

Background We recently developed CoCites, a citation-based search method that is designed to be more efficient than traditional keyword-based methods. The method begins with identification of one or more highly relevant publications (query articles) and consists of two searches: the co-citation search, which ranks publications on their co-citation frequency with the query articles, and the citation search, which ranks publications on frequency of all citations that cite or are cited by the query articles. Methods We aimed to reproduce the literature searches of published systematic reviews and meta-analyses and assess whether CoCites retrieves all eligible articles while screening fewer titles. Results A total of 250 reviews were included. CoCites retrieved a median of 75% of the articles that were included in the original reviews. The percentage of retrieved articles was higher (88%) when the query articles were cited more frequently and when they had more overlap in their citations. Applying CoCites to only the highest-cited article yielded similar results. The co-citation and citation searches combined were more efficient when the review authors had screened more than 500 titles, but not when they had screened less. Conclusions CoCites is an efficient and accurate method for finding relevant related articles. The method uses the expert knowledge of authors to rank related articles, does not depend on keyword selection and requires no special expertise to build search queries. The method is transparent and reproducible.

To explain how the intended use of polygenic risk scores (PRSs) in healthcare guides the design and evaluation of prediction studies. The advances in gene discovery in common complex diseases have fueled the interest in the potential of PRSs to predict risks and improve the prevention and early detection of disease. As the predictive ability of a PRS differs between populations and settings, it is important that prediction studies are designed and evaluated with the intended use of the risk scores in mind, but this is rarely done. The intended use indicates in whom and how the PRS will be used in healthcare and for what purpose. This intended use dictates what outcome needs to be predicted in which population using which predictors. It also tells which other variables or clinical risk models might be available to improve the prediction. The intended use also provides the necessary context to evaluate whether the predictive ability of the PRS or the risk model that includes PRS is high enough for the score to be potentially useful in healthcare. The intended use should be leading risk prediction research.

At the turn of the millennium, when the Human Genome Project was about to publish the first full sequence of human DNA, about 100 websites were offering direct-to-consumer genetic testing. Most sites offered DNA testing to determine paternity, but about 15% offered health-related genetic tests that determined carrier status of DNA mutations for hereditary disorders or recommended nutrients for a healthier lifestyle. Their test offer was premature and businesses were closed. The commercial interest renewed when genome-wide association studies published their first findings, faded a few years later, and revived again recently as whole genome sequencing becoming affordable. This chapter describes a short history, the status, and challenges of direct-to-consumer genetic testing.

Irreplicability is framed as crisis, blamed on sloppy science motivated by perverse stimuli in research. Structural changes to the organization of science, targeting sloppy science (e.g., open data, pre-registration), are proposed to prevent irreplicability. While there is an unquestionable link between sloppy science and failures to replicate/reproduce scientific studies, they are currently conflated. This position can be understood as a result of the erosion of the role of theory in science. The history, sociology, and philosophy of science reveal alternative explanations for irreplicability to show it is part of proper, informative and valuable science. Irreplicability need not equate research waste. Sloppy science is the problem, also when results do replicate. Hence, the solution should focus on opposing sloppy research.

Background & aimsRelative risk of colorectal cancer (CRC) decreases with age among individuals with a family history of CRC. However, no screening recommendations specify less frequent screening with increasing age. We aimed to determine whether such a refinement would be cost effective.MethodsWe determined the relative risk for CRC for individuals based on age and number of affected first-degree relatives (FDRs) using data from publications. For each number of affected FDRs, we used the Microsimulation Screening Analysis model to estimate costs and effects of colonoscopy screening strategies with different age ranges and intervals. Screening was then optimized sequentially, starting withthe youngest age group, and allowing the interval of screening to change at certain ages. Strategies with an incremental cost effectiveness ratio below $100,000 per quality-adjusted lifeyear were considered cost effective.ResultsFor people with 1 affected FDR (92% of those with a family history), screening every 3years beginning at an age of 40 years is most cost effective. If no adenomas are found, the screening interval can gradually be extended to 5 and 7 years, at ages 45 and 55 years, respectively. From a cost-effectiveness perspective, individuals with more affected FDRs should start screening earlier and at shorter intervals. However, frequency can be reduced if no abnormalities are found.ConclusionsUsing a microsimulation model, we found that for individuals with afamily history of CRC, it is cost effective to gradually increase the screening interval if several subsequent screening colonoscopies have negative results and no new cases of CRC are found in family members.

The area under the receiver operating characteristic (ROC) curve (AUC) is commonly used for assessing the discriminative ability of prediction models even though the measure is criticized for being clinically irrelevant and lacking an intuitive interpretation. Most of the criticism is traced back to the fact that the ROC curve was introduced as the discriminative ability of a binary classifier across all its possible thresholds. Yet, this is not the curve’s only interpretation. Every tutorial explains how the coordinates of the ROC curve are obtained from the risk distributions of diseased and non-diseased individuals, but it has not become common sense that the ROC plot is another way of presenting these risk distributions. This alternative perspective on the ROC plot invalidates most limitations of the AUC and attributes others to the underlying risk distributions. The separation of these distributions, represented by the area between the curves (AUC), is a more straightforward and intuitive measure of the inference of the discriminative ability of prediction models.

We systematically investigated how 11 overlapping meta-analyses on the association between CYP2C19 loss-of-function alleles and clinical efficacy of clopidogrel could yield contradictory outcomes. The results of the meta-analyses differed because more recent meta-analyses included more primary studies and some had not included conference abstracts. Conclusions differed because between-study heterogeneity and publication bias were handled differently across meta-analyses. All meta-analyses on the clinical end point observed significant heterogeneity and several reported evidence for publication bias, but only one out of eight statistically significant meta-analyses concluded that therefore the association was unproven and one other refrained from quantifying a pooled estimate because of heterogeneity. For the end point stent thrombosis, all meta-analyses reported statistically significant associations with CYP2C19 loss-of-function alleles with no statistically significant evidence for heterogeneity, but only three had investigated publication bias and also found evidence for it. One study therefore concluded that there was no evidence for an association, and one other doubted the association because of a high level of heterogeneity. In summary, meta-analyses on the association between CYP2C19 loss-of-function alleles and clinical efficacy of clopidogrel differed widely with regard to assessment and interpretation of heterogeneity and publication bias. The substantial heterogeneity and publication bias implies that personalized antiplatelet management based on genotyping is not supported by the currently available evidence.Genet Med advance online publication 19 June 2014.

Purpose: Family history of colorectal cancer (CRC) is a known risk factor for CRC and encompasses both genetic and shared environmental risks. Methods: We conducted a systematic review to estimate the impact of family history on the natural history of CRC and adherence to screening. Results: We found high heterogeneity in family-history definitions, the most common definition being one or more first-degree relatives. The prevalence of family history may be lower than the commonly cited 10%, and confirms evidence for increasing levels of risk associated with increasing family-history burden. There is evidence for higher prevalence of adenomas and of multiple adenomas in people with family history of CRC but no evidence for differential adenoma location or adenoma progression by family history. Limited data regarding the natural history of CRC by family history suggest a differential age or stage at cancer diagnosis and mixed evidence with respect to tumor location. Adherence to recommended colonoscopy screening was higher in people with a family history of CRC. Conclusion: Stratification based on polygenic and/or multifactorial risk assessment may mature to the point of displacing family historybased approaches, but for the foreseeable future, family history may remain a valuable clinical tool for identifying individuals at increased risk for CRC.

The US National Institutes of Health (NIH) recently announced that they would limit the number of grants per scientist and redistribute their funds across a larger group of researchers. The policy was withdrawn a month later after criticism from the scientific community. Even so, the basis of this defunct policy was flawed and it merits further examination. The amount of grant support would have been quantified using a new metric, the Grant Support Index (GSI), and limited to a maximum of 21 points, the equivalent of three R01 grants. This threshold was decided based upon analysis of a new metric of scientific output, the annual weighted Relative Citation Ratio, which showed a pattern of diminishing returns at higher values of the GSI. In this commentary, we discuss several concerns about the validity of the two metrics and the quality of the data that the NIH had used to set the grant threshold. These concerns would have warranted a re-analysis of new data to confirm the legitimacy of the GSI threshold. Data-driven policies that affect the careers of scientists should be justified by nothing less than a rigorous analysis of high-quality data.

Except in rare cases, obesity tends to be a consequence of both an unhealthy lifestyle and a genetic susceptibility to gain weight. With more than 200 common genetic variants identified, there is a growing interest in developing personalized preventive and treatment strategies to predict an individual's obesity risk. We review the literature on the prediction of obesity and show that models based on the established genetic variants have poorer predictive ability than traditional predictors, such as family history of obesity and childhood obesity. Current findings suggest that opportunities for precision medicine in common obesity may be limited.

The Evaluation of Genomic Applications in Practice and Prevention (EGAPP) Working Group found insufficient evidence to recommend prostate cancer antigen 3 (PCA3) testing to inform decisions for when to rebiopsy previously biopsy-negative patients for prostate cancer or to inform decisions to conduct initial biopsies for prostate cancer in at-risk men (e.g., previous elevated prostate-specific antigen test or suspicious digital rectal examination).The EGAPP Working Group found insufficient evidence to recommend PCA3 testing in men with cancer-positive biopsies to determine if the disease is indolent or aggressive in order to develop an optimal treatment plan.Based on the available evidence, the overall certainty of clinical validity to predict the diagnosis of prostate cancer using PCA3 is deemed "low." The EGAPP Working Group discourages clinical use for diagnosis unless further evidence supports improved clinical validity.Based on the available evidence, the overall certainty of net health benefit is deemed "low." The EGAPP Working Group discourages clinical use unless further evidence supports improved clinical outcomes.It has been suggested that PCA3 testing in the general male population might lead to earlier diagnosis and management changes (e.g., earlier detection and earlier initiation or higher rates of medical interventions) that improve outcomes. EGAPP Working Group found no direct evidence to support this possibility, so we sought indirect evidence aimed at documenting the extent to which PCA3 testing alters prostate cancer diagnosis or management, alone and in combination with traditional clinical management factors, and the extent to which this testing improves health outcomes.Assay-related evidence was deemed adequate for the PROGENSA PCA3 assay approved by the US Food and Drug Administration, available from Gen-Probe. Very few studies were available that investigated preanalytical effects, analytical performance, and diagnostic accuracy of other quantitative assays for PCA3.Evidence on clinical validity was rated inadequate to derive any conclusions about performance of PCA3 testing to inform decisions for when to rebiopsy previously biopsy-negative patients for prostate cancer, or to inform decisions to conduct initial biopsies for prostate cancer in at-risk men (e.g., previous elevated prostate-specific antigen test or suspicious digital rectal examination). Furthermore, there was little evidence to derive any conclusions about performance of PCA3 testing in men with cancer-positive biopsies to determine if the disease is indolent or aggressive in order to develop an optimal treatment plan.No studies were available to provide direct evidence on the balance of benefits and harms related to PCA3 testing for diagnosis and management in the general male population. Evidence for other populations (e.g., high risk) was not evaluated in the review.Early diagnosis of prostate cancer is central to minimizing morbidity and mortality. Prevention of prostate cancer mortality is a public health priority. Improvements in outcomes associated with PCA3 testing could have important impacts.

Aims B-type natriuretic peptide (BNP) and C-reactive protein (CRP) predict atrial fibrillation (AF) risk. However, their risk stratification abilities in the broad community remain uncertain. We sought to improve risk stratification for AF using biomarker information. Methods and results We ascertained AF incidence in 18 556 Whites and African Americans from the Atherosclerosis Risk in Communities Study (ARIC, n =10 675), Cardiovascular Health Study (CHS, n = 5043), and Framingham Heart Study (FHS, n = 2838), followed for 5 years (prediction horizon). We added BNP (ARIC/CHS: N-terminal pro-B-type natriuretic peptide; FHS: BNP), CRP, or both to a previously reported AF risk score, and assessed model calibration and predictive ability [C-statistic, integrated discrimination improvement (IDI), and net reclassification improvement (NRI)]. We replicated models in two independent European cohorts: Age, Gene/Environment Susceptibility Reykjavik Study (AGES), n = 4467; Rotterdam Study (RS), n = 3203. B-type natriuretic peptide and CRP were significantly associated with AF incidence ( n = 1186): hazard ratio per 1-SD ln-transformed biomarker 1.66 [95% confidence interval (CI), 1.56–1.76], P < 0.0001 and 1.18 (95% CI, 1.11–1.25), P < 0.0001, respectively. Model calibration was sufficient (BNP, χ 2 = 17.0; CRP, χ 2 = 10.5; BNP and CRP, χ 2 = 13.1). B-type natriuretic peptide improved the C-statistic from 0.765 to 0.790, yielded an IDI of 0.027 (95% CI, 0.022–0.032), a relative IDI of 41.5%, and a continuous NRI of 0.389 (95% CI, 0.322–0.455). The predictive ability of CRP was limited (C-statistic increment 0.003). B-type natriuretic peptide consistently improved prediction in AGES and RS. Conclusion B-type natriuretic peptide, not CRP, substantially improved AF risk prediction beyond clinical factors in an independently replicated, heterogeneous population. B-type natriuretic peptide may serve as a benchmark to evaluate novel putative AF risk biomarkers.

Purpose: The promise of personalized genomics for common complex diseases depends, in part, on the ability to predict genetic risks on the basis of single nucleotide polymorphisms. We examined and compared the methods of three companies (23andMe, deCODEme, and Navigenics) that have offered direct-to-consumer personal genome testing. Methods: We simulated genotype data for 100,000 individuals on the basis of published genotype frequencies and predicted disease risks using the methods of the companies. Predictive ability for six diseases was assessed by the AUC. Results: AUC values differed among the diseases and among the companies. The highest values of the AUC were observed for age-related macular degeneration, celiac disease, and Crohn disease. The largest difference among the companies was found for celiac disease: the AUC was 0.73 for 23andMe and 0.82 for deCODEme. Predicted risks differed substantially among the companies as a result of differences in the sets of single nucleotide polymorphisms selected and the average population risks selected by the companies, and in the formulas used for the calculation of risks. Conclusion: Future efforts to design predictive models for the genomics of common complex diseases may benefit from understanding the strengths and limitations of the predictive algorithms designed by these early companies.

Purpose Prediction models for age-related macular degeneration (AMD) based on case-control studies have a tendency to overestimate risks. The aim of this study is to develop a prediction model for late AMD based on data from population-based studies. Design Three population-based studies: the Rotterdam Study (RS), the Beaver Dam Eye Study (BDES), and the Blue Mountains Eye Study (BMES) from the Three Continent AMD Consortium (3CC). Participants People (n = 10 106) with gradable fundus photographs, genotype data, and follow-up data without late AMD at baseline. Methods Features of AMD were graded on fundus photographs using the 3CC AMD severity scale. Associations with known genetic and environmental AMD risk factors were tested using Cox proportional hazard analysis. In the RS, the prediction of AMD was estimated for multivariate models by area under receiver operating characteristic curves (AUCs). The best model was validated in the BDES and BMES, and associations of variables were re-estimated in the pooled data set. Beta coefficients were used to construct a risk score, and risk of incident late AMD was calculated using Cox proportional hazard analysis. Cumulative incident risks were estimated using Kaplan–Meier product-limit analysis. Main Outcome Measures Incident late AMD determined per visit during a median follow-up period of 11.1 years with a total of 4 to 5 visits. Results Overall, 363 participants developed incident late AMD, 3378 participants developed early AMD, and 6365 participants remained free of any AMD. The highest AUC was achieved with a model including age, sex, 26 single nucleotide polymorphisms in AMD risk genes, smoking, body mass index, and baseline AMD phenotype. The AUC of this model was 0.88 in the RS, 0.85 in the BDES and BMES at validation, and 0.87 in the pooled analysis. Individuals with low-risk scores had a hazard ratio (HR) of 0.02 (95% confidence interval [CI], 0.01–0.04) to develop late AMD, and individuals with high-risk scores had an HR of 22.0 (95% CI, 15.2–31.8). Cumulative risk of incident late AMD ranged from virtually 0 to more than 65% for those with the highest risk scores. Conclusions Our prediction model is robust and distinguishes well between those who will develop late AMD and those who will not. Estimated risks were lower in these population-based studies than in previous case-control studies. Financial Disclosure(s) The author(s) have no proprietary or commercial interest in any materials discussed in this article.

Advances in genomics have near-term impact on diagnosis and management of monogenic disorders. For common complex diseases, the use of genomic information from multiple loci (polygenic model) is generally not useful for diagnosis and individual prediction. In principle, the polygenic model could be used along with other risk factors in stratified population screening to target interventions. For example, compared to age-based criterion for breast, colorectal, and prostate cancer screening, adding polygenic risk and family history holds promise for more efficient screening with earlier start and/or increased frequency of screening for segments of the population at higher absolute risk than an established screening threshold; and later start and/or decreased frequency of screening for segments of the population at lower risks. This approach, while promising, faces formidable challenges for building its evidence base and for its implementation in practice. Currently, it is unclear whether or not polygenic risk can contribute enough discrimination to make stratified screening worthwhile. Empirical data are lacking on population-based age-specific absolute risks combining genetic and non-genetic factors, on impact of polygenic risk genes on disease natural history, as well as information on comparative balance of benefits and harms of stratified interventions. Implementation challenges include difficulties in integration of this information in the current health-care system in the United States, the setting of appropriate risk thresholds, and ethical, legal, and social issues. In an era of direct-to-consumer availability of personal genomic information, the public health and health-care systems need to prepare for an evidence-based integration of this information into population screening.

Aim Migraine, in particular with aura, has been associated with an increased risk for ischemic stroke and coronary heart disease. The underlying mechanism is unknown. In a cross-sectional case control study we investigated whether an enhanced risk of atherosclerosis in migraineurs explains this increased cardiovascular risk. Methods Subjects were participants from the population-based Erasmus Rucphen Family study. Atherosclerosis was assessed in 360 migraineurs (209 without aura and 151 with aura) and 617 subjects without migraine or severe headache. Atherosclerosis was quantified by intima media thickness, pulse wave velocity and ankle-brachial index. Results Migraineurs, especially with aura, were found more likely to smoke, have diabetes or a modestly decreased HDL-cholesterol. No differences were found for the atherosclerosis parameters. Conclusion In this large population-based study, migraineurs have no increased risk of atherosclerosis. Therefore, enhanced atherosclerosis is an unlikely explanation for the increased cardiovascular risk seen in migraineurs.

For several years personal genome tests have been offered directly to consumers via the internet. Based on single genome scans, these direct-toconsumer (DTC) tests predict susceptibility to common multifactorial diseases, such as Type 2 diabetes, coronary heart disease and nonfamilial cancer, inform about predisposition to drug response, report carrier status for monogenic diseases, or provide all of the above. The market is served by a few major players, such as 23andMe and deCODEme, and numerous lesser-known companies such as YouScript, GenePlanet and Theranostics [101–105]. The market for DTC personal genome testing is steadily increasing, even though the evidence on the predictive ability and clinical utility of these tests is limited. The few available studies have shown that predicted risks of multifactorial diseases differed between companies and were sometimes even contradictory for the same individual [1,106], but large-scale studies on the predictive ability are lacking. From prediction studies that investigated genetic risk models based on different but comparable selections of SNPs, we know that the predictive ability of genetic testing for multifactorial diseases is low to moderate, except when one or more SNPs had substantial impact on disease risk [2]. From this indirect evidence it is concluded that the DTC offer of genetic testing for multifactorial diseases is premature. To date, most of the discussion about DTC personal genome tests has focused on the prediction of these multifactorial diseases and less attention has been given to the predictive ability and clinical utility of pharmacogenetic testing. Yet, many companies offer pharmacogenetic testing to inform about the genetic susceptibility to drug response and side effects of treatment [3], such as the efficacy of clopidogrel and the risk

In recent years, developments in genomics technologies have led to the rise of commercial personal genome testing (PGT): broad genome-wide testing for multiple diseases simultaneously. While some commercial providers require physicians to order a personal genome test, others can be accessed directly. All providers advertise directly to consumers and offer genetic risk information about dozens of diseases in one single purchase. The quantity and the complexity of risk information pose challenges to adequate pre-test and post-test information provision and informed consent. There are currently no guidelines for what should constitute informed consent in PGT or how adequate informed consent can be achieved. In this paper, we propose a tiered-layered-staged model for informed consent. First, the proposed model is tiered as it offers choices between categories of diseases that are associated with distinct ethical, personal or societal issues. Second, the model distinguishes layers of information with a first layer offering minimal, indispensable information that is material to all consumers, and additional layers offering more detailed information made available upon request. Finally, the model stages informed consent as a process by feeding information to consumers in each subsequent stage of the process of undergoing a test, and by accommodating renewed consent for test result updates, resulting from the ongoing development of the science underlying PGT. A tiered-layered-staged model for informed consent with a focus on the consumer perspective can help overcome the ethical problems of information provision and informed consent in direct-to-consumer PGT.

Obesity is predominantly caused by an unhealthy lifestyle, but genetic factors also contribute to people’s susceptibility to gain weight. Rare high-risk mutations have been identified that cause extreme and early-onset obesity in a fraction of the population, while numerous common low-risk loci have been identified through genome-wide association studies that contribute to obesity in the general population. As insights into the contribution of genetic variation to obesity increase, the interest in using genetic variants to predict who is at risk to gain weight has also increased. Before constructing risk models, however, one needs to have a clear view on what form of obesity needs to be predicted, why, in whom, and for what purpose. Obesity is multifactorial, in that it results from an interplay between genes and environmental risk factors, such that models solely based on genetic variants will be unlikely to reach high predictive ability, as we illustrate using the literature on currently identified BMI-associated loci. Furthermore, it seems that irrespective of the poor predictive ability, communicating genetic information does not seem effective in making people adopt a healthy lifestyle. While using genetic information in the prediction of obesity is a legitimate aim, we believe that the most valuable contribution of gene discovery studies lies in their contribution to elucidate new physiological pathways that underlie obesity susceptibility, which in turn could lead to the identification of therapeutic targets and make its way into mainstream health care.

Objective We propose a new measure of assessing the performance of risk models, the area under the prediction impact curve (auPIC), which quantifies the performance of risk models in terms of their average health impact in the population. Study Design and Setting Using simulated data, we explain how the prediction impact curve (PIC) estimates the percentage of events prevented when a risk model is used to assign high-risk individuals to an intervention. We apply the PIC to the Atherosclerosis Risk in Communities (ARIC) Study to illustrate its application toward prevention of coronary heart disease. Results We estimated that if the ARIC cohort received statins at baseline, 5% of events would be prevented when the risk model was evaluated at a cutoff threshold of 20% predicted risk compared to 1% when individuals were assigned to the intervention without the use of a model. By calculating the auPIC, we estimated that an average of 15% of events would be prevented when considering performance across the entire interval. Conclusion We conclude that the PIC is a clinically meaningful measure for quantifying the expected health impact of risk models that supplements existing measures of model performance.

The discrimination of a risk prediction model measures that model's ability to distinguish between subjects with and without events. The area under the receiver operating characteristic curve (AUC) is a popular measure of discrimination. However, the AUC has recently been criticized for its insensitivity in model comparisons in which the baseline model has performed well. Thus, 2 other measures have been proposed to capture improvement in discrimination for nested models: the integrated discrimination improvement and the continuous net reclassification improvement. In the present study, the authors use mathematical relations and numerical simulations to quantify the improvement in discrimination offered by candidate markers of different strengths as measured by their effect sizes. They demonstrate that the increase in the AUC depends on the strength of the baseline model, which is true to a lesser degree for the integrated discrimination improvement. On the other hand, the continuous net reclassification improvement depends only on the effect size of the candidate variable and its correlation with other predictors. These measures are illustrated using the Framingham model for incident atrial fibrillation. The authors conclude that the increase in the AUC, integrated discrimination improvement, and net reclassification improvement offer complementary information and thus recommend reporting all 3 alongside measures characterizing the performance of the final model.

Extracellular deposition of amyloid beta peptide (A beta) has been implicated as a critical step in the pathogenesis of Alzheimer's disease (AD). In Down syndrome (DS), Alzheimer's disease is assumed to be caused by the triplication and overexpression of the gene for amyloid precursor protein (APP), located on chromosome 21. Plasma concentrations of A beta 1-40 and A beta 1-42 were determined in a population based study of 506 persons with DS, who were screened annually for dementia. We used Cox proportional hazards models to determine the risk of dementia. Demented persons with DS have a significantly higher plasma A beta 1-40 concentration than the nondemented (p = 0.05). Those with the highest concentrations of A beta 1-40 and A beta 1-42 have a higher risk to develop dementia. The risk to develop dementia during follow-up (mean 4.7 years) increased to 2.56 (95% confidence interval, 1.39-4.71) for A beta 1-42 and 2.16 (95% confidence interval, 1.14-4.10) for A beta 1-40. High plasma concentration of plasma A beta 1-40 and A beta 1-42 are determinants of the risk of dementia in persons with DS. (C) 2012 Elsevier Inc. All rights reserved.

Various modeling methods have been proposed to estimate the potential predictive ability of polygenic risk variants that predispose to various common diseases. However, it is unknown whether differences between them affect their conclusions on predictive ability. We reviewed input parameters, assumptions and output of the five most common methods and compared their estimates of the area under the receiver operating characteristic (ROC) curve (AUC) using hypothetical data representing effect sizes and frequencies of genetic variants, population disease risk and number of variants. To assess the accuracy of the estimated AUCs, we aimed to reproduce the AUCs of published empirical studies. All methods assumed that the combined effect of genetic variants on disease risk followed a multiplicative risk model of independent genetic effects, but they either assumed per allele, per genotype or dominant/recessive effects for the genetic variants. Modeling strategy and input parameters differed. Methods used simulation analysis or analytical formulas with effect sizes quantified by odds ratios (ORs) or relative risks. Estimated AUC values were similar for lower ORs (0.7) due to variants with strong effects, differences in estimated AUCs between methods increased. The simulation methods accurately reproduced the AUC values of empirical studies, but the analytical methods did not. We conclude that despite differences in input parameters, the modeling methods estimate similar AUC for realistic values of the ORs. When one or more variants have stronger effects and AUC values are higher, the simulation methods tend to be more accurate. European Journal of Human Genetics (2012) 20, 1270-1274; doi:10.1038/ejhg.2012.89; published online 30 May 2012

Finding eligible studies for meta-analysis and systematic reviews relies on keyword-based searching as the gold standard, despite its inefficiency. Searching based on direct citations is not sufficiently comprehensive. We propose a novel strategy that ranks articles on their degree of co-citation with one or more "known" articles before reviewing their eligibility.In two independent studies, we aimed to reproduce the results of literature searches for sets of published meta-analyses (n = 10 and n = 42). For each meta-analysis, we extracted co-citations for the randomly selected 'known' articles from the Web of Science database, counted their frequencies and screened all articles with a score above a selection threshold. In the second study, we extended the method by retrieving direct citations for all selected articles.In the first study, we retrieved 82% of the studies included in the meta-analyses while screening only 11% as many articles as were screened for the original publications. Articles that we missed were published in non-English languages, published before 1975, published very recently, or available only as conference abstracts. In the second study, we retrieved 79% of included studies while screening half the original number of articles.Citation searching appears to be an efficient and reasonably accurate method for finding articles similar to one or more articles of interest for meta-analysis and reviews.

Genetic testing for personalizing diet and wellness programs is performed without extensive counseling that informs about the potential implications of knowing one's genotype status. Genetic counseling seems redundant for genes that impact the effect of diet on biomarkers such as cholesterol and blood pressure, but the same genes may have pleiotropic effects that cannot be ignored. A well-known example is the APOE gene, which is implicated in cholesterol regulation and is a major risk factor for Alzheimer's disease. Not fully informing participants about the major pleiotropic effects of genes has ethical implications and invalidates informed consent.

Human longevity and personality traits are both heritable and are consistently linked at the phenotypic level. We test the hypothesis that candidate genes influencing longevity in lower organisms are associated with variance in the five major dimensions of human personality (measured by the NEO-FFI and IPIP inventories) plus related mood states of anxiety and depression. Seventy single nucleotide polymorphisms (SNPs) in six brain expressed, longevity candidate genes (AFG3L2, FRAP1, MAT1A, MAT2A, SYNJ1, and SYNJ2) were typed in over 1,000 70-year old participants from the Lothian Birth Cohort of 1936 (LBC1936). No SNPs were associated with the personality and psychological distress traits at a Bonferroni corrected level of significance (P60 years). Because we selected a specific set of longevity genes based on functional genomics findings, further research on other longevity gene candidates is warranted to discover whether they are relevant candidates for personality and psychological distress traits. (C) 2011 Wiley Periodicals, Inc.

Fueled by the successes of genome-wide association studies, numerous studies have investigated the predictive ability of genetic risk models in type 2 diabetes. In this paper, we review these studies from a methodological perspective, focusing on the variables included in the risk models as well as the study designs and populations investigated. We argue and show that differences in study design and characteristics of the study population have an impact on the observed predictive ability of risk models. This observation emphasizes that genetic risk prediction studies should be conducted in those populations in which the prediction models will ultimately be applied, if proven useful. Of all genetic risk prediction studies to date, only a few were conducted in populations that might be relevant for targeting preventive interventions.

Serum concentrations of low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), triglycerides (TGs) and total cholesterol (TC) are important heritable risk factors for cardiovascular disease. Although genome-wide association studies (GWASs) of circulating lipid levels have identified numerous loci, a substantial portion of the heritability of these traits remains unexplained. Evidence of unexplained genetic variance can be detected by combining multiple independent markers into additive genetic risk scores. Such polygenic scores, constructed using results from the ENGAGE Consortium GWAS on serum lipids, were applied to predict lipid levels in an independent population-based study, the Rotterdam Study-II (RS-II). We additionally tested for evidence of a shared genetic basis for different lipid phenotypes. Finally, the polygenic score approach was used to identify an alternative genome-wide significance threshold before pathway analysis and those results were compared with those based on the classical genome-wide significance threshold. Our study provides evidence suggesting that many loci influencing circulating lipid levels remain undiscovered. Cross-prediction models suggested a small overlap between the polygenic backgrounds involved in determining LDL-C, HDL-C and TG levels. Pathway analysis utilizing the best polygenic score for TC uncovered extra information compared with using only genome-wide significant loci. These results suggest that the genetic architecture of circulating lipids involves a number of undiscovered variants with very small effects, and that increasing GWAS sample sizes will enable the identification of novel variants that regulate lipid levels. European Journal of Human Genetics (2011) 19, 813-819; doi:10.1038/ejhg.2011.21; published online 30 March 2011

The rapid identification of genetic markers for multifactorial diseases from genome-wide association studies is fuelling interest in investigating the predictive ability and health care utility of genetic risk models. Various measures are available for the assessment of risk prediction models, each addressing a different aspect of performance and utility. We developed PredictABEL, a package in R that covers descriptive tables, measures and figures that are used in the analysis of risk prediction studies such as measures of model fit, predictive ability and clinical utility, and risk distributions, calibration plot and the receiver operating characteristic plot. Tables and figures are saved as separate files in a user-specified format, which include publication-quality EPS and TIFF formats. All figures are available in a ready-made layout, but they can be customized to the preferences of the user. The package has been developed for the analysis of genetic risk prediction studies, but can also be used for studies that only include non-genetic risk factors. PredictABEL is freely available at the websites of GenABEL ( http://www.genabel.org ) and CRAN ( http://cran.r-project.org/).

Eur J Clin Invest 2011; 41 (9): 1004–1009 Summary Points • The rapid and continuing progress in gene discovery for complex diseases is fuelling interest in the potential application of genetic risk models for clinical and public health practice. • The number of studies assessing the predictive ability is steadily increasing, but the quality and completeness of reporting vary. • A multidisciplinary workshop sponsored by the Human Genome Epidemiology Network developed a checklist of 25 items recommended for strengthening the reporting of Genetic RIsk Prediction Studies (GRIPS), building on the principles established by prior reporting guidelines. • These recommendations aim to enhance the transparency of study reporting and thereby to improve the synthesis and application of information from multiple studies that might differ in design, conduct or analysis. • A detailed Explanation and Elaboration document is published as an accompanying article [1].

Eur J Clin Invest 2011; 41 (9): 1010–1035 Summary Points • The rapid and continuing progress in gene discovery for complex diseases is fuelling interest in the potential application of genetic risk models for clinical and public health practice. • The number of studies assessing the predictive ability is steadily increasing, but they vary widely in completeness of reporting and apparent quality. • Transparent reporting of the strengths and weaknesses of these studies is important to facilitate the accumulation of evidence on genetic risk prediction. • A multidisciplinary workshop sponsored by the Human Genome Epidemiology Network developed a checklist of 25 items recommended for strengthening the reporting of Genetic RIsk Prediction Studies (GRIPS), building on the principles established by prior reporting guidelines. • These recommendations aim to enhance the transparency, quality and completeness of study reporting and thereby to improve the synthesis and application of information from multiple studies that might differ in design, conduct or analysis.