Generalizability of “GWAS Hits” in Clinical Populations: Lessons from Childhood Cancer Survivors

Im, Cindy; Qin, Na; Wang, Zhaoming; Qiu, Weiliang; Howell, Carrie R.; Sapkota, Yadav; Moon, Wonjong; Chemaitilly, Wassim; Gibson, Todd M.; Mulrooney, Daniel A.; Ness, Kirsten K.; Wilson, Carmen L.; Morton, Lindsay M.; Armstrong, Gregory T.; Bhatia, Smita; Zhang, Jinghui; Hudson, Melissa M.; Robison, Leslie L.; Yasui, Yutaka

doi:10.1016/j.ajhg.2020.08.014

Cited by 13 publications

(7 citation statements)

References 64 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our results corroborate previous results that predictions within the UK Biobank are often more accurate than off-cohort predictions to the same target ancestry 62–64 . This raises the question of whether the higher within-UK Biobank prediction accuracy is inflated by cohort effects.…”

Section: Discussionsupporting

confidence: 91%

“…We applied PolyPred to predict 23 diseases and complex traits in Biobank Japan 61 and 7 complex traits in Uganda-APCDR, an African-ancestry cohort 33,34 (Methods, Supplementary Table 3). We performed these experiments to avoid training effect sizes and testing predictions in the same cohort, which may produce inflated prediction accuracies 52,62–64 . We again used UK Biobank British training data (average N =325K) to estimate SNP effect sizes, and used 500 individuals from the target population to estimate mixing weights.…”

Section: Resultsmentioning

confidence: 99%

See 1 more Smart Citation

Leveraging fine-mapping and non-European training data to improve cross-population polygenic risk scores

Weissbrod

Kanai

Shi

et al. 2021

Preprint

View full text Add to dashboard Cite

Polygenic risk scores (PRS) based on European training data suffer reduced accuracy in non-European target populations, exacerbating health disparities. This loss of accuracy predominantly stems from LD differences, MAF differences (including population-specific SNPs), and/or causal effect size differences. Here, we propose PolyPred, a method that improves trans-ethnic polygenic prediction by combining two complementary predictors: a new predictor that leverages functionally informed fine-mapping to estimate causal effects (instead of tagging effects), addressing LD differences; and BOLT-LMM, a published predictor. In the special case where a large training sample is available in the non-European target population (or a closely related population), we propose PolyPred+, which further incorporates the non-European training data, addressing MAF differences and causal effect size differences. We applied PolyPred to 49 diseases and complex traits in 4 UK Biobank populations using UK Biobank British training data (average N=325K), and observed statistically significant average relative improvements in prediction accuracy vs. BOLT-LMM ranging from +7% in South Asians to +32% in Africans (and vs. LD-pruning + P-value thresholding (P+T) ranging from +77% to +164%), consistent with simulations. We applied PolyPred+ to 23 diseases and complex traits in UK Biobank East Asians using both UK Biobank British (average N=325K) and Biobank Japan (average N=124K) training data, and observed statistically significant average relative improvements in prediction accuracy of +24% vs. BOLT-LMM and +12% vs. PolyPred. In conclusion, PolyPred and PolyPred+ improve trans-ethnic polygenic prediction accuracy, ameliorating health disparities.

show abstract

Section: Discussionsupporting

confidence: 91%

Section: Resultsmentioning

confidence: 99%

Leveraging fine-mapping and non-European training data to improve cross-population polygenic risk scores

Weissbrod

Kanai

Shi

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…We applied PolyPred and its summary statistic-based analogues to predict 23 diseases and complex traits in Biobank Japan 41 and 7 complex traits in Uganda-APCDR, an African-ancestry cohort 42 , 43 ( Methods , Supplementary Table 3 ). We performed these experiments to avoid training effect sizes and testing predictions in the same cohort, which may produce inflated prediction accuracies 33 , 58 – 60 . We again used UK Biobank British training data (average N =325K) to estimate SNP effect sizes, and used 500 individuals from the target population to estimate mixing weights.…”

Section: Resultsmentioning

confidence: 99%

Leveraging fine-mapping and multipopulation training data to improve cross-population polygenic risk scores

et al. 2022

View full text Add to dashboard Cite

Polygenic risk scores (PRS) based on European training data suffer reduced accuracy in non-European target populations, exacerbating health disparities. This loss of accuracy predominantly stems from LD differences, MAF differences (including population-specific SNPs), and/or causal effect size differences. PRS based on training data from the non-European target population do not suffer from these limitations, but are currently limited by much smaller training sample sizes. Here, we propose PolyPred, a method that improves cross-population polygenic prediction by combining two complementary predictors: a new predictor that leverages functionally informed fine-mapping to estimate causal effects (instead of tagging effects), addressing LD differences; and BOLT-LMM, a published predictor. In the special case where a large training sample is available in the non-European target population (or a closely related population), we propose PolyPred+, which further incorporates the non-European training data, addressing MAF differences and causal effect size differences. PolyPred and PolyPred+ require individual-level training data (for their BOLT-LMM component), but we also propose analogous methods that replace the BOLT-LMM component with summary statistic-based components if only summary statistics are available. We applied PolyPred to 49 diseases and complex traits in 4 UK Biobank populations using UK Biobank British training data (average N=325K), and observed statistically significant average relative improvements in prediction accuracy vs. BOLT-LMM ranging from +7% in South Asians to +32% in Africans (and vs. LDpruning + P-value thresholding (P+T) ranging from +77% to +164%), consistent with simulations. We applied PolyPred+ to 23 diseases and complex traits in UK Biobank East Asians using both UK Biobank British (average N=325K) and Biobank Japan (average N=124K) training data, and observed statistically significant average relative improvements in prediction accuracy of +24% vs. BOLT-LMM and +12% vs. PolyPred. The summary statistic-based analogues of PolyPred and PolyPred+ attained similar improvements. In conclusion, PolyPred and PolyPred+ improve cross-population polygenic prediction accuracy, ameliorating health disparities.

show abstract

“…rs3093956 is a known hit for EAA-Hannum [ 15 ]; however, it has low LD ( r 2 =0.075) with rs28366133. Multiple striking DMRs were observed between survivors and controls in the HLA region, which might be due to the fact that genotoxic cancer treatments modified the epigenome among other physiological alterations in survivors and hence altered functional genomic links (e.g., eQTL, mQTL, and eQTM) [ 23 , 58 ], which may lead to either disruption or introduction of genetic associations with EAA. For the same reason, a substantial proportion (e.g., 20 out of 39 IEAA-associated SNPs) of previously reported genetic associations, including the most notable rs2736099 ( TERT ), were not replicated in our study.…”

Section: Discussionmentioning

confidence: 99%

Genome-wide association studies identify novel genetic loci for epigenetic age acceleration among survivors of childhood cancer

et al. 2022

Self Cite

View full text Add to dashboard Cite

Background Increased epigenetic age acceleration (EAA) in survivors of childhood cancer is associated with specific treatment exposures, unfavorable health behaviors, and presence of certain chronic health conditions. To better understand inter-individual variability, we investigated the genetic basis underlying EAA. Methods Genome-wide association studies of EAA based on multiple epigenetic clocks (Hannum, Horvath, PhenoAge, and GrimAge) were performed. MethylationEPIC BeadChip array and whole-genome sequencing data were generated with blood-derived DNA from participants in the St. Jude Lifetime Cohort Study (discovery: 2138 pre-existing and 502 newly generated data, all survivors; exploratory: 282 community controls). Linear regression models were fit for each epigenetic age against the allelic dose of each genetic variant, adjusting for age at sampling, sex, and cancer treatment exposures. Fixed-effects meta-analysis was used to combine summary statistics from two discovery data sets. LD (Linkage disequilibrium) score regression was used to estimate single-nucleotide polymorphism (SNP)-based heritability. Results For EAA-Horvath, a genome-wide significant association was mapped to the SELP gene with the strongest SNP rs732314 (meta-GWAS: β=0.57, P=3.30×10-11). Moreover, the stratified analysis of the association between rs732314 and EAA-Horvath showed a substantial heterogeneity between children and adults (meta-GWAS: β=0.97 vs. 0.51, I2=73.1%) as well as between survivors with and without chest/abdominal/pelvic-RT exposure (β=0.64 vs. 0.31, I2=66.3%). For EAA-Hannum, an association was mapped to the HLA locus with the strongest SNP rs28366133 (meta-GWAS: β=0.78, P=3.78×10-11). There was no genome-wide significant hit for EAA-PhenoAge or EAA-GrimAge. Interestingly, among community controls, rs732314 was associated with EAA-Horvath (β=1.09, P=5.43×10-5), whereas rs28366133 was not associated with EAA-Hannum (β=0.21, P=0.49). The estimated heritability was 0.33 (SE=0.20) for EAA-Horvath and 0.17 (SE=0.23) for EAA-Hannum, but close to zero for EAA-PhenoAge and EAA-GrimAge. Conclusions We identified novel genetic variants in the SELP gene and HLA region associated with EAA-Horvath and EAA-Hannum, respectively, among survivors of childhood cancer. The new genetic variants in combination with other replicated known variants can facilitate the identification of survivors at higher risk in developing accelerated aging and potentially inform drug targets for future intervention strategies among vulnerable survivors.

show abstract

Generalizability of “GWAS Hits” in Clinical Populations: Lessons from Childhood Cancer Survivors

Cited by 13 publications

References 64 publications

Leveraging fine-mapping and non-European training data to improve cross-population polygenic risk scores

Leveraging fine-mapping and non-European training data to improve cross-population polygenic risk scores

Leveraging fine-mapping and multipopulation training data to improve cross-population polygenic risk scores

Genome-wide association studies identify novel genetic loci for epigenetic age acceleration among survivors of childhood cancer

Contact Info

Product

Resources

About