FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3190363403> ?p ?o ?g. }

• W3190363403 endingPage "1764" @default.
• W3190363403 startingPage "1752" @default.
• W3190363403 abstract "An individual’s genetics can dramatically influence breast cancer (BC) risk. Although clinical measures for prevention do exist, non-invasive personalized measures for reducing BC risk are limited. Commonly used medications are a promising set of modifiable factors, but no previous study has explored whether a range of widely taken approved drugs modulate BC genetics. In this study, we describe a quantitative framework for exploring the interaction between the genetic susceptibility of BC and medication usage among UK Biobank women. We computed BC polygenic scores (PGSs) that summarize BC genetic risk and find that the PGS explains nearly three-times greater variation in disease risk within corticosteroid users compared to non-users. We map 35 genes significantly interacting with corticosteroid use (FDR < 0.1), highlighting the transcription factor NRF2 as a common regulator of gene-corticosteroid interactions in BC. Finally, we discover a regulatory variant strongly stratifying BC risk according to corticosteroid use. Within risk allele carriers, 18.2% of women taking corticosteroids developed BC, compared to 5.1% of the non-users (with an HR = 3.41 per-allele within corticosteroid users). In comparison, there are no differences in BC risk within the reference allele homozygotes. Overall, this work highlights the clinical relevance of gene-drug interactions in disease risk and provides a roadmap for repurposing biobanks in drug repositioning and precision medicine. An individual’s genetics can dramatically influence breast cancer (BC) risk. Although clinical measures for prevention do exist, non-invasive personalized measures for reducing BC risk are limited. Commonly used medications are a promising set of modifiable factors, but no previous study has explored whether a range of widely taken approved drugs modulate BC genetics. In this study, we describe a quantitative framework for exploring the interaction between the genetic susceptibility of BC and medication usage among UK Biobank women. We computed BC polygenic scores (PGSs) that summarize BC genetic risk and find that the PGS explains nearly three-times greater variation in disease risk within corticosteroid users compared to non-users. We map 35 genes significantly interacting with corticosteroid use (FDR < 0.1), highlighting the transcription factor NRF2 as a common regulator of gene-corticosteroid interactions in BC. Finally, we discover a regulatory variant strongly stratifying BC risk according to corticosteroid use. Within risk allele carriers, 18.2% of women taking corticosteroids developed BC, compared to 5.1% of the non-users (with an HR = 3.41 per-allele within corticosteroid users). In comparison, there are no differences in BC risk within the reference allele homozygotes. Overall, this work highlights the clinical relevance of gene-drug interactions in disease risk and provides a roadmap for repurposing biobanks in drug repositioning and precision medicine. Breast cancer (BC) is the most commonly diagnosed cancer in women. Over 2 million new cases were diagnosed and 600,000 deaths occurred in 2018 worldwide,1Bray F. Ferlay J. Soerjomataram I. Siegel R.L. Torre L.A. Jemal A. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries.CA Cancer J. Clin. 2018; 68: 394-424Crossref PubMed Scopus (39163) Google Scholar highlighting a clear need for implementation of primary prevention strategies. However, preventive measures are limited, and many involve invasive surgical procedures (such as mastectomy) or drugs (such as tamoxifen) with moderate to severe side effects. Personalized measures involving existing medications are a unique opportunity for prevention. This may include repurposing existing treatments that treat other indications on the basis of individuals’ genetics for breast cancer risk reduction. Alternatively, this may include altered prescribing of medications that may be harmful and increase risk within genetically susceptible individuals. For example, vitamin B12 intake is not associated with BC risk among the Women’s Health Study participants2Lin J. Lee I.M. Cook N.R. Selhub J. Manson J.E. Buring J.E. Zhang S.M. Plasma folate, vitamin B-6, vitamin B-12, and risk of breast cancer in women.Am. J. Clin. Nutr. 2008; 87: 734-743Crossref PubMed Scopus (103) Google Scholar (who are primarily white), but it is significantly associated with reduced BC risk among Mexican women3Lajous M. Lazcano-Ponce E. Hernandez-Avila M. Willett W. Romieu I. Folate, vitamin B(6), and vitamin B(12) intake and the risk of breast cancer among Mexican women.Cancer Epidemiol. Biomarkers Prev. 2006; 15: 443-448Crossref PubMed Scopus (99) Google Scholar and Canadian BRCA1 mutation carriers.4Kim S.J. Zhang C.X.W. Demsky R. Armel S. Kim Y.I. Narod S.A. Kotsopoulos J. Folic acid supplement use and breast cancer risk in BRCA1 and BRCA2 mutation carriers: a case-control study.Breast Cancer Res. Treat. 2019; 174: 741-748Crossref PubMed Scopus (9) Google Scholar Although these differences may be due to a number of study design and methodologic differences, it may also suggest differences due to genetic background. Given the prevalence of gene-environment interactions,5Marderstein A.R. Davenport E.R. Kulm S. Van Hout C.V. Elemento O. Clark A.G. Leveraging phenotypic variability to identify genetic interactions in human phenotypes.Am. J. Hum. Genet. 2021; 108: 49-67Abstract Full Text Full Text PDF PubMed Scopus (4) Google Scholar, 6Huang W. Campbell T. Carbone M.A. Jones W.E. Unselt D. Anholt R.R.H. Mackay T.F.C. Context-dependent genetic architecture of Drosophila life span.PLoS Biol. 2020; 18: e3000645Crossref PubMed Scopus (13) Google Scholar, 7Kraft P. Aschard H. Finding the missing gene-environment interactions.Eur. J. Epidemiol. 2015; 30: 353-355Crossref PubMed Scopus (24) Google Scholar medication-associated risk reduction or increase may strongly depend on genetic factors and thus be different between individuals. Germline genetic variation is one potential reason for variable drug efficacy and adverse outcomes. To date, most discovered gene-drug interactions involve variation at individual genes8Hewett M. Oliver D.E. Rubin D.L. Easton K.L. Stuart J.M. Altman R.B. Klein T.E. PharmGKB: the pharmacogenetics knowledge base.Nucleic Acids Res. 2002; 30: 163-165Crossref PubMed Scopus (284) Google Scholar (such as CYP2C9- or VKORC1-warfarin interactions for anticoagulation,9Schwarz U.I. Ritchie M.D. Bradford Y. Li C. Dudek S.M. Frye-Anderson A. Kim R.B. Roden D.M. Stein C.M. Genetic determinants of response to warfarin during initial anticoagulation.N. Engl. J. Med. 2008; 358: 999-1008Crossref PubMed Scopus (497) Google Scholar HMGCR-statin interactions for cholesterol,10Chasman D.I. Posada D. Subrahmanyan L. Cook N.R. Stanton Jr., V.P. Ridker P.M. Pharmacogenetic study of statin therapy and cholesterol reduction.JAMA. 2004; 291: 2821-2827Crossref PubMed Scopus (387) Google Scholar or CYP2D6- or SULT1A1-tamoxifen interactions for breast cancer11Serrano D. Lazzeroni M. Zambon C.F. Macis D. Maisonneuve P. Johansson H. Guerrieri-Gonzaga A. Plebani M. Basso D. Gjerde J. et al.Efficacy of tamoxifen based on cytochrome P450 2D6, CYP2C19 and SULT1A1 genotype in the Italian Tamoxifen Prevention Trial.Pharmacogenomics J. 2011; 11: 100-107Crossref PubMed Scopus (51) Google Scholar, 12Jung J.-A. Lim H.-S. Association between CYP2D6 genotypes and the clinical outcomes of adjuvant tamoxifen for breast cancer: a meta-analysis.Pharmacogenomics. 2014; 15: 49-60Crossref PubMed Scopus (31) Google Scholar, 13Goetz M.P. Rae J.M. Suman V.J. Safgren S.L. Ames M.M. Visscher D.W. Reynolds C. Couch F.J. Lingle W.L. Flockhart D.A. et al.Pharmacogenetics of tamoxifen biotransformation is associated with clinical outcomes of efficacy and hot flashes.J. Clin. Oncol. 2005; 23: 9312-9318Crossref PubMed Scopus (698) Google Scholar), and the contribution from genome-wide variation is largely unexplored. Genome-wide association studies (GWASs) have revealed that the genetic variance for most phenotypes is spread genome wide at thousands to tens of thousands of loci with small allelic effects rather than the most significant single nucleotide polymorphisms (SNPs).14Sinnott-Armstrong N. Naqvi S. Rivas M. Pritchard J.K. GWAS of three molecular traits highlights core genes and pathways alongside a highly polygenic background.eLife. 2021; 10: e58615Crossref PubMed Scopus (14) Google Scholar Thus, although individual genes have been shown to strongly influence the response to treatments, the genome-wide polygenic contribution to drug response deserves further exploration. In one analysis of three statin clinical trials, the relative risk reduction of coronary artery disease for those at high polygenic risk was 46%, compared to 26% in all other individuals.15Natarajan P. Young R. Stitziel N.O. Padmanabhan S. Baber U. Mehran R. Sartori S. Fuster V. Reilly D.F. Butterworth A. et al.Polygenic risk score identifies subgroup with higher burden of atherosclerosis and greater relative benefit from statin therapy in the primary prevention setting.Circulation. 2017; 135: 2091-2101Crossref PubMed Scopus (207) Google Scholar Patients with a particular genetic risk profile receiving a drug may experience greater risk protection against (or harm toward) disease than other drug users with a different set of genetic alterations. Importantly, genetic activation of a disease pathway may be nullified (or exacerbated) through a drug’s mechanism of action. Generally, pharmacogenomic studies—which have successfully identified genes and pathways influencing the body’s response to hundreds of drugs8Hewett M. Oliver D.E. Rubin D.L. Easton K.L. Stuart J.M. Altman R.B. Klein T.E. PharmGKB: the pharmacogenetics knowledge base.Nucleic Acids Res. 2002; 30: 163-165Crossref PubMed Scopus (284) Google Scholar—have been relatively narrow in scope and have specific hypotheses linked to particular mutations or drugs. A systematic search for interactions between commonly used drugs and genome-wide variants remains unexplored. This could lead to potential drug repurposing or personalized medication prescribing for the reduction of disease risk within specific subsets of the population (e.g., those who have extremely high polygenic risk15Natarajan P. Young R. Stitziel N.O. Padmanabhan S. Baber U. Mehran R. Sartori S. Fuster V. Reilly D.F. Butterworth A. et al.Polygenic risk score identifies subgroup with higher burden of atherosclerosis and greater relative benefit from statin therapy in the primary prevention setting.Circulation. 2017; 135: 2091-2101Crossref PubMed Scopus (207) Google Scholar, 16Khera A.V. Chaffin M. Aragam K.G. Haas M.E. Roselli C. Choi S.H. Natarajan P. Lander E.S. Lubitz S.A. Ellinor P.T. Kathiresan S. Genome-wide polygenic scores for common diseases identify individuals with risk equivalent to monogenic mutations.Nat. Genet. 2018; 50: 1219-1224Crossref PubMed Scopus (891) Google Scholar, 17Kulm S. Marderstein A. Mezey J. Elemento O. A systematic framework for assessing the clinical impact of polygenic risk scores.SSRN. 2021; https://doi.org/10.2139/ssrn.3808292Crossref Scopus (0) Google Scholar or carry particular risk variants4Kim S.J. Zhang C.X.W. Demsky R. Armel S. Kim Y.I. Narod S.A. Kotsopoulos J. Folic acid supplement use and breast cancer risk in BRCA1 and BRCA2 mutation carriers: a case-control study.Breast Cancer Res. Treat. 2019; 174: 741-748Crossref PubMed Scopus (9) Google Scholar,9Schwarz U.I. Ritchie M.D. Bradford Y. Li C. Dudek S.M. Frye-Anderson A. Kim R.B. Roden D.M. Stein C.M. Genetic determinants of response to warfarin during initial anticoagulation.N. Engl. J. Med. 2008; 358: 999-1008Crossref PubMed Scopus (497) Google Scholar, 10Chasman D.I. Posada D. Subrahmanyan L. Cook N.R. Stanton Jr., V.P. Ridker P.M. Pharmacogenetic study of statin therapy and cholesterol reduction.JAMA. 2004; 291: 2821-2827Crossref PubMed Scopus (387) Google Scholar, 11Serrano D. Lazzeroni M. Zambon C.F. Macis D. Maisonneuve P. Johansson H. Guerrieri-Gonzaga A. Plebani M. Basso D. Gjerde J. et al.Efficacy of tamoxifen based on cytochrome P450 2D6, CYP2C19 and SULT1A1 genotype in the Italian Tamoxifen Prevention Trial.Pharmacogenomics J. 2011; 11: 100-107Crossref PubMed Scopus (51) Google Scholar). Extensive analysis of the pharmacogenomic interactions between a variety of medications and genome-wide data can reveal why some medication users experience adverse clinical outcomes or point to potential therapeutic repurposing opportunities for women with genetic predispositions to breast cancer risk (Figure 1). With the emergence of UK Biobank (UKB), a large publicly available dataset that combines electronic health record data with genomic, prescription, and survey questionnaire information from 500,000 individuals, studying such potential pharmacogenomic interactions at scale is possible. This cohort has already revealed the genetic influences on medication use19Wu Y. Byrne E.M. Zheng Z. Kemper K.E. Yengo L. Mallett A.J. Yang J. Visscher P.M. Wray N.R. Genome-wide association study of medication-use and associated disease in the UK Biobank.Nat. Commun. 2019; 10: 1891Crossref PubMed Scopus (42) Google Scholar and dosage,20Lavertu A. McInnes G. Tanigawa Y. Altman R.B. Rivas M.A. LPA and APOE are associated with statin selection in the UK Biobank.bioRxiv. 2020; https://doi.org/10.1101/2020.08.28.272765Crossref Scopus (0) Google Scholar,21McInnes G. Altman R.B. Drug Response Pharmacogenetics for 200,000 UK Biobank Participants.Pac. Symp. Biocomput. 2021; 26: 184-195PubMed Google Scholar the population prevalence of known pharmacogenomic variation,22McInnes G. Lavertu A. Sangkuhl K. Klein T.E. Whirl-Carrillo M. Altman R.B. Pharmacogenetics at scale: An analysis of the UK Biobank.Clin. Pharmacol. Ther. 2021; 109: 1528-1537Crossref PubMed Scopus (17) Google Scholar the incidence of drug side effects,21McInnes G. Altman R.B. Drug Response Pharmacogenetics for 200,000 UK Biobank Participants.Pac. Symp. Biocomput. 2021; 26: 184-195PubMed Google Scholar and the genetic and non-genetic characteristics of treatment-resistant depression.23Fabbri C. Hagenaars S.P. John C. Williams A.T. Shrine N. Moles L. Hanscombe K.B. Serretti A. Shepherd D.J. Free R.S. et al.Genetic and clinical characteristics of treatment-resistant depression using primary care records in two UK cohorts.medRxiv. 2020; https://doi.org/10.1101/2020.08.24.20178715Crossref Scopus (0) Google Scholar We use UKB to identify interactions between genetic and medication data by first testing for interaction effects between drug exposures and a polygenic score (PGS) for BC. Although PGS-by-exposure interaction tests have been widely performed across a range of phenotypes and exposures,24Rudolph A. Song M. Brook M.N. Milne R.L. Mavaddat N. Michailidou K. Bolla M.K. Wang Q. Dennis J. Wilcox A.N. et al.Joint associations of a polygenic risk score and environmental risk factors for breast cancer in the Breast Cancer Association Consortium.Int. J. Epidemiol. 2018; 47: 526-536Crossref PubMed Google Scholar, 25Shi M. O’Brien K.M. Weinberg C.R. Interactions between a Polygenic Risk Score and Non-genetic Risk Factors in Young-Onset Breast Cancer.Sci. Rep. 2020; 10: 3242Crossref PubMed Scopus (1) Google Scholar, 26Aschard H. Zaitlen N. Lindström S. Kraft P. Variation in predictive ability of common genetic variants by established strata: the example of breast cancer and age.Epidemiology. 2015; 26: 51-58Crossref PubMed Scopus (7) Google Scholar, 27Kramer I. Hooning M.J. Mavaddat N. Hauptmann M. Keeman R. Steyerberg E.W. Giardiello D. Antoniou A.C. Pharoah P.D.P. Canisius S. et al.Breast Cancer Polygenic Risk Score and Contralateral Breast Cancer Risk.Am. J. Hum. Genet. 2020; 107: 837-848Abstract Full Text Full Text PDF PubMed Scopus (10) Google Scholar, 28Ahmad S. Rukh G. Varga T.V. Ali A. Kurbasic A. Shungin D. Ericson U. Koivula R.W. Chu A.Y. Rose L.M. et al.Gene × physical activity interactions in obesity: combined analysis of 111,421 individuals of European ancestry.PLoS Genet. 2013; 9: e1003607Crossref PubMed Scopus (137) Google Scholar, 29Aschard H. Tobin M.D. Hancock D.B. Skurnik D. Sood A. James A. Vernon Smith A. Manichaikul A.W. Campbell A. Prins B.P. et al.Evidence for large-scale gene-by-smoking interaction effects on pulmonary function.Int. J. Epidemiol. 2017; 46: 894-904PubMed Google Scholar, 30Ye Y. Chen X. Han J. Jiang W. Natarajan P. Zhao H. Interactions Between Enhanced Polygenic Risk Scores and Lifestyle for Cardiovascular Disease, Diabetes, and Lipid Levels.Circ Genom Precis Med. 2021; 14: e003128Crossref PubMed Scopus (8) Google Scholar, 31Meyers J.L. Cerdá M. Galea S. Keyes K.M. Aiello A.E. Uddin M. Wildman D.E. Koenen K.C. Interaction between polygenic risk for cigarette use and environmental exposures in the Detroit Neighborhood Health Study.Transl. Psychiatry. 2013; 3: e290Crossref PubMed Scopus (39) Google Scholar they have not been extensively applied for studying drug-related risk across a range of medications. Furthermore, the biological drivers (mechanisms, pathways, and processes) of observed PGS-by-exposure interactions are unclear. In the present study, we apply a PGS-drug interaction test to identify corticosteroids as a modulator of polygenic risk in BC and use a SNP-based gene-set enrichment analysis to highlight potential pathways and mechanisms of this interaction. Finally, we assessed stratification of BC risk in UKB by corticosteroid use and genetic variation. Overall, our results demonstrate the potential of prospective cohorts such as UKB in drug repurposing, risk prediction, and pharmacogenomics by using statistical genetic and epidemiologic approaches (Figure 2). The UK Biobank (UKB) team previously processed the UKB data32Bycroft C. Freeman C. Petkova D. Band G. Elliott L.T. Sharp K. Motyer A. Vukcevic D. Delaneau O. O’Connell J. et al.The UK Biobank resource with deep phenotyping and genomic data.Nature. 2018; 562: 203-209Crossref PubMed Scopus (1419) Google Scholar and deposited the data for research in the scientific community. We accessed the UKB data under application ID 47137. We extracted the set of women with self-reported British European ancestry and excluded all individuals with sex chromosome aneuploidy, excess heterozygosity, or outlier genotype missing rates. Breast cancer (BC) was diagnosed with procedural classifications (OPCS) and medical classifications (ICD9 and ICD10) from longitudinal health record information and self-reports during the baseline assessment. We used 174X in ICD9 information and C50X in ICD10 information. In OPCS data, we used B27, B28, or B29. We used self-report code 1002. We removed any individuals with a BC diagnosis prior to or reported at the baseline assessment. This left 212,335 women with no prior BC diagnosis at the time of baseline assessment: 11,730 incident BC cases diagnosed after the baseline assessment (determined through the longitudinal health records) and 200,605 population controls. Individuals were censored at death, hospital records end date, or BC diagnosis. Imputed SNP data were provided by the UKB team, calculated as described previously,32Bycroft C. Freeman C. Petkova D. Band G. Elliott L.T. Sharp K. Motyer A. Vukcevic D. Delaneau O. O’Connell J. et al.The UK Biobank resource with deep phenotyping and genomic data.Nature. 2018; 562: 203-209Crossref PubMed Scopus (1419) Google Scholar and used for genetic analyses. We use a polygenic score (PGS) to summarize breast cancer risk from genetic loci across the genome. The PGS can be computed with the standard calculation of a weighted sum of trait-associated SNPs (PGS = ∑iβiXi). We refer to βi as the weight assigned to SNP i and Xi as the number of minor alleles. However, determining the appropriate weights for each SNP (βi) and which SNPs to include in the PGS leads to user and parameter bias and variable accuracies as a result. As such, we use two breast cancer GWASs to calculate scores. Michailidou et al.33Michailidou K. Lindström S. Dennis J. Beesley J. Hui S. Kar S. Lemaçon A. Soucy P. Glubb D. Rostamianfar A. et al.Association analysis identifies 65 new breast cancer risk loci.Nature. 2017; 551: 92-94Crossref PubMed Scopus (524) Google Scholar released summary statistics from a GWAS meta-analysis of 61,282 individuals with BC diagnoses across 68 cohorts. These were used for the creation of original PGSs. Mavaddat et al.34Mavaddat N. Michailidou K. Dennis J. Lush M. Fachal L. Lee A. Tyrer J.P. Chen T.H. Wang Q. Bolla M.K. et al.Polygenic risk scores for prediction of breast cancer and breast cancer subtypes.Am. J. Hum. Genet. 2019; 104: 21-34Abstract Full Text Full Text PDF PubMed Scopus (287) Google Scholar released PGS weights for two pre-computed scores in the PGS catalog35Lambert S.A. Gil L. Jupp S. Ritchie S.C. Xu Y. Buniello A. McMahon A. Abraham G. Chapman M. Parkinson H. et al.The Polygenic Score Catalog as an open database for reproducibility and systematic evaluation.Nat. Genet. 2021; 53: 420-425Crossref PubMed Scopus (25) Google Scholar based on a GWAS meta-analysis of 94,075 individuals with BC diagnoses across 69 cohorts (an updated GWAS from Michailidou et al.33Michailidou K. Lindström S. Dennis J. Beesley J. Hui S. Kar S. Lemaçon A. Soucy P. Glubb D. Rostamianfar A. et al.Association analysis identifies 65 new breast cancer risk loci.Nature. 2017; 551: 92-94Crossref PubMed Scopus (524) Google Scholar). One score is based on hard-thresholding stepwise forward regression, while the other score is based on LASSO penalized regression. Neither score used UKB to estimate or train SNP weights. For both Michailidou et al.33Michailidou K. Lindström S. Dennis J. Beesley J. Hui S. Kar S. Lemaçon A. Soucy P. Glubb D. Rostamianfar A. et al.Association analysis identifies 65 new breast cancer risk loci.Nature. 2017; 551: 92-94Crossref PubMed Scopus (524) Google Scholar and Mavaddat et al.34Mavaddat N. Michailidou K. Dennis J. Lush M. Fachal L. Lee A. Tyrer J.P. Chen T.H. Wang Q. Bolla M.K. et al.Polygenic risk scores for prediction of breast cancer and breast cancer subtypes.Am. J. Hum. Genet. 2019; 104: 21-34Abstract Full Text Full Text PDF PubMed Scopus (287) Google Scholar scores, we applied a quality-control pipeline to handle ambiguous SNPs, account for potential sequencing errors, and focus on high-quality variants. We removed SNPs with missing chromosome, position, or effect size information. We removed multi-allelic SNPs, indels, SNPs with ambiguous strand flips (A/T, G/C), SNPs not present in UKB, and SNPs with poor imputation (information score: INFO < 0.9). We flipped alleles if different strands were used in the summary statistics file compared to UKB (A/C in the summary statistics file but T/G in UKB) and reversed alleles if necessary (A/C in file but C/A in UKB). For Michailidou et al. scores, odds ratio estimates provided by Michailidou et al. were transformed into b estimates via log transformation. We created multiple Michailidou et al.33Michailidou K. Lindström S. Dennis J. Beesley J. Hui S. Kar S. Lemaçon A. Soucy P. Glubb D. Rostamianfar A. et al.Association analysis identifies 65 new breast cancer risk loci.Nature. 2017; 551: 92-94Crossref PubMed Scopus (524) Google Scholar scores with different sets of SNPs and selected the optimal one for downstream analysis. To select SNPs for scoring five different Michailidou et al. scores, we used clumped summary statistics at five p value thresholds of BC significance: 0.05, 0.005, 0.0005, 10−5, and 5 × 10−8. We set the r2 value = 0.1 and a window size equal to 250 kb. We performed clumping by using a linkage disequilibrium panel of 503 European individuals from the 1000 Genomes project. In all, we created seven scores: five Michailidou et al. scores and two Mavaddat et al.34Mavaddat N. Michailidou K. Dennis J. Lush M. Fachal L. Lee A. Tyrer J.P. Chen T.H. Wang Q. Bolla M.K. et al.Polygenic risk scores for prediction of breast cancer and breast cancer subtypes.Am. J. Hum. Genet. 2019; 104: 21-34Abstract Full Text Full Text PDF PubMed Scopus (287) Google Scholar scores. All scores were standardized such that the mean was equal to zero and the standard deviation was equal to one. To select the optimal Michailidou et al.33Michailidou K. Lindström S. Dennis J. Beesley J. Hui S. Kar S. Lemaçon A. Soucy P. Glubb D. Rostamianfar A. et al.Association analysis identifies 65 new breast cancer risk loci.Nature. 2017; 551: 92-94Crossref PubMed Scopus (524) Google Scholar and Mavaddat et al.34Mavaddat N. Michailidou K. Dennis J. Lush M. Fachal L. Lee A. Tyrer J.P. Chen T.H. Wang Q. Bolla M.K. et al.Polygenic risk scores for prediction of breast cancer and breast cancer subtypes.Am. J. Hum. Genet. 2019; 104: 21-34Abstract Full Text Full Text PDF PubMed Scopus (287) Google Scholar scores, we assessed both the model fit to all UKB participants and the model prediction within a held-out individual set. To assess model prediction, we trained logistic regression models containing a single PGS within 90% of UKB and predicted disease liability within the held-out remaining 10% of UKB. We measured the correlation between model predictions and future disease status. This was repeated across ten iterations, and each iteration contained a distinct random prediction set of individuals. We statistically compared the ten correlation values from one score to the correlation values from another score by using paired t tests. For assessing model fit, we used logistic regression models of breast cancer risk. We calculated the Nagelkerke R2 of each PGS (compared to a null model with no PGS covariate) and selected the optimal Michailidou et al. score and optimal Mavaddat et al. score. We included body mass index, age, menopause status, number of births, at least one birth indicator, and the first ten genotype principal components as covariates or features in all models—we refer to these as the baseline covariates. We assessed the relationship between scores through pairwise Pearson correlation. Because of modest correlation between scores, we combined the optimal Michailidou et al. score (clumped with p = 10−5, r2 = 0.1, and window size = 250 kb; a total of 908 SNPs included) and the optimal Mavaddat et al. score (LASSO penalized regression with p < 0.001; a total of 2,661 SNPs included) into a combined score by taking the mean between the two (a total of 3,342 unique SNPs included between the two scores). We use a Cox proportional-hazards model to calculate hazard ratios (HRs) of PGSs. Individuals were binned into deciles, and disease incidence was calculated within each decile and compared to the 45th–55th percentile individuals. The combined score was used in downstream analyses. Because of minimal competing risks, we use a Cox proportional-hazards model for all survival analyses as opposed to a Fine-Gray model. At the baseline assessment, individuals were asked whether they were regularly taking any medications. One limitation of this survey is that further information was not obtained; specifically, duration, dosage, and personal reasons for taking each medication are unknown. We downloaded a mapping from UKB medications to Anatomical Therapeutic Chemical (ATC) classification system36Santos R. Ursu O. Gaulton A. Bento A.P. Donadi R.S. Bologa C.G. Karlsson A. Al-Lazikani B. Hersey A. Oprea T.I. Overington J.P. A comprehensive map of molecular drug targets.Nat. Rev. Drug Discov. 2017; 16: 19-34Crossref PubMed Scopus (878) Google Scholar codes provided by Wu et al.19Wu Y. Byrne E.M. Zheng Z. Kemper K.E. Yengo L. Mallett A.J. Yang J. Visscher P.M. Wray N.R. Genome-wide association study of medication-use and associated disease in the UK Biobank.Nat. Commun. 2019; 10: 1891Crossref PubMed Scopus (42) Google Scholar We mapped medications to level 4 ATC groups. We identified 99 ATC groups with at least 1,000 users. We removed three ATC groups with a strong marginal association with BC risk (false discovery rate [FDR] < 0.01) by using 99 distinct Cox proportional-hazards models with the baseline covariates (minus PGS). Because UKB is not a randomized controlled trial, these possibly include medications preferentially taken by individuals in high-risk groups, which could be difficult to account for in analysis and interpret within results. We first tested for interactions between the breast cancer PGS and medication use on breast cancer risk. Using the 96 remaining ATC groups, we computed 96 Cox proportional-hazards models with baseline covariates, each medication group, and the combined PGS as main effects and an interaction term between medication group and the combined PGS. We used FDR for multiple hypothesis-testing corrections. We fitted a logistic regression model with the same covariates to compute Nagelkerke R2 as a measure of variation explained by the PGS within users versus non-users. We extracted 3,342 SNPs used for computation of the combined PGS and next tested for interactions between SNPs and corticosteroid use by using 3,342 Cox proportional-hazards models. We used the baseline covariates, one SNP, and corticosteroid use (defined with the S01BA coding) as main effects and tested for the significance of the interaction term between the SNP and S01BA. We used FDR for multiple testing corrections and determined statistical significance at FDR < 0.1. To examine evidence for coordinated interactions, we assessed the concordance of the sign of the marginal GWAS effect estimates to the sign of the interaction effects in UKB at SNPs with an interaction p value below 0.05. We calculated significance at risk-increasing and -decreasing SNPs by using a binomial test with probability equal to 0.5. To further analyze the strongest SNP-corticosteroid interaction (rs62119267), we contrasted carriers of at least one C allele to non-carriers within su" @default.
• W3190363403 created "2021-08-16" @default.
• W3190363403 creator A5013206062 @default.
• W3190363403 creator A5042388690 @default.
• W3190363403 creator A5049212930 @default.
• W3190363403 creator A5050661907 @default.
• W3190363403 creator A5061702711 @default.
• W3190363403 creator A5086709630 @default.
• W3190363403 date "2021-09-01" @default.
• W3190363403 modified "2023-10-12" @default.
• W3190363403 title "A polygenic-score-based approach for identification of gene-drug interactions stratifying breast cancer risk" @default.
• W3190363403 cites W1634951127 @default.
• W3190363403 cites W1756412699 @default.
• W3190363403 cites W1989607209 @default.
• W3190363403 cites W1992204010 @default.
• W3190363403 cites W1992724001 @default.
• W3190363403 cites W2032594110 @default.
• W3190363403 cites W2034351801 @default.
• W3190363403 cites W2036298961 @default.
• W3190363403 cites W2057783610 @default.
• W3190363403 cites W2085127466 @default.
• W3190363403 cites W2096261918 @default.
• W3190363403 cites W2098023402 @default.
• W3190363403 cites W2104689123 @default.
• W3190363403 cites W2104775804 @default.
• W3190363403 cites W2105440238 @default.
• W3190363403 cites W2127069089 @default.
• W3190363403 cites W2132192895 @default.
• W3190363403 cites W2137422676 @default.
• W3190363403 cites W2139849788 @default.
• W3190363403 cites W2142879323 @default.
• W3190363403 cites W2144201925 @default.
• W3190363403 cites W2169795969 @default.
• W3190363403 cites W2171705978 @default.
• W3190363403 cites W2187787607 @default.
• W3190363403 cites W2239090104 @default.
• W3190363403 cites W2247766769 @default.
• W3190363403 cites W2345356016 @default.
• W3190363403 cites W2411366655 @default.
• W3190363403 cites W2487801709 @default.
• W3190363403 cites W2548616668 @default.
• W3190363403 cites W2558217333 @default.
• W3190363403 cites W2590902687 @default.
• W3190363403 cites W2613483874 @default.
• W3190363403 cites W2765253743 @default.
• W3190363403 cites W2781651514 @default.
• W3190363403 cites W2794844698 @default.
• W3190363403 cites W2886752110 @default.
• W3190363403 cites W2889646458 @default.
• W3190363403 cites W2895486342 @default.
• W3190363403 cites W2898585601 @default.
• W3190363403 cites W2900668971 @default.
• W3190363403 cites W2901952863 @default.
• W3190363403 cites W2907259814 @default.
• W3190363403 cites W2917254076 @default.
• W3190363403 cites W2920825919 @default.
• W3190363403 cites W2942053120 @default.
• W3190363403 cites W2958615822 @default.
• W3190363403 cites W2973641376 @default.
• W3190363403 cites W2979639908 @default.
• W3190363403 cites W2991034145 @default.
• W3190363403 cites W3000500728 @default.
• W3190363403 cites W3008564772 @default.
• W3190363403 cites W3009501570 @default.
• W3190363403 cites W3023819387 @default.
• W3190363403 cites W3038503389 @default.
• W3190363403 cites W3091099371 @default.
• W3190363403 cites W3108978936 @default.
• W3190363403 cites W3109005117 @default.
• W3190363403 cites W3111318171 @default.
• W3190363403 cites W3119797572 @default.
• W3190363403 cites W3129339917 @default.
• W3190363403 cites W3133859624 @default.
• W3190363403 cites W3134184401 @default.
• W3190363403 cites W3137375284 @default.
• W3190363403 cites W3144454794 @default.
• W3190363403 cites W3184293501 @default.
• W3190363403 doi "https://doi.org/10.1016/j.ajhg.2021.07.008" @default.
• W3190363403 hasPubMedCentralId "https://www.ncbi.nlm.nih.gov/pmc/articles/8456164" @default.
• W3190363403 hasPubMedId "https://pubmed.ncbi.nlm.nih.gov/34363748" @default.
• W3190363403 hasPublicationYear "2021" @default.
• W3190363403 type Work @default.
• W3190363403 sameAs 3190363403 @default.
• W3190363403 citedByCount "5" @default.
• W3190363403 countsByYear W31903634032022 @default.
• W3190363403 countsByYear W31903634032023 @default.
• W3190363403 crossrefType "journal-article" @default.
• W3190363403 hasAuthorship W3190363403A5013206062 @default.
• W3190363403 hasAuthorship W3190363403A5042388690 @default.
• W3190363403 hasAuthorship W3190363403A5049212930 @default.
• W3190363403 hasAuthorship W3190363403A5050661907 @default.
• W3190363403 hasAuthorship W3190363403A5061702711 @default.
• W3190363403 hasAuthorship W3190363403A5086709630 @default.
• W3190363403 hasBestOaLocation W31903634031 @default.
• W3190363403 hasConcept C104317684 @default.
• W3190363403 hasConcept C116834253 @default.
• W3190363403 hasConcept C121608353 @default.
• W3190363403 hasConcept C126322002 @default.

• next