Incorporating European GWAS findings improve polygenic risk prediction accuracy of breast cancer among East Asians

Abstract

Previous genome‐wide association studies (GWASs) have been largely focused on European (EUR) populations. However, polygenic risk scores (PRSs) derived from EUR have been shown to perform worse in non‐EURs compared with EURs. In this study, we aim to improve PRS prediction in East Asians (EASs). We introduce a rescaled meta‐analysis framework to combine both EUR (N = 122,175) and EAS (N = 30,801) GWAS summary statistics. To improve PRS prediction in EASs, we use a scaling factor to up‐weight the EAS data, such that the resulting effect size estimates are more relevant to EASs. We then derive PRSs for EAS from the rescaled meta‐analysis results of EAS and EUR data. Evaluated in an independent EAS validation data set, this approach increases the prediction liability‐adjusted Nagelkerke's pseudo R2 by 40%, 41%, and 5%, respectively, compared with PRSs derived from an EAS GWAS only, EUR GWAS only, and conventional fixed‐effects meta‐analysis of EAS and EUR data. The PRS derived from the rescaled meta‐analysis approach achieved an area under the receiver operating characteristic curve (AUC) of 0.6059, higher than AUC = 0.5782, 0.5809, 0.6008 for EAS, EUR, and conventional meta‐analysis of EAS and EUR. We further compare PRSs constructed by single‐nucleotide polymorphisms that have different linkage disequilibrium (LD) scores and minor allele frequencies (MAFs) between EUR and EAS, and observe that lower LD scores or MAF in EAS correspond to poorer PRS performance (AUC = 0.5677, 0.5530, respectively) than higher LD scores or MAF (AUC = 0.589, 0.5993, respectively). We finally build a PRS stratified by LD score differences in EUR and EAS using rescaled meta‐analysis, and obtain an AUC of 0.6096, with improvement over other strategies investigated.

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

“…We previously demonstrated that linearly combining PRS from European and non-European training samples improves cross-population prediction accuracy 7 . However, these previous results did not incorporate causal effects and used P+T, which is highly inaccurate despite its widespread use 11,[13][14][15][16][17][18]23,31,[52][53][54][55][56] , as PolyPred obtained up to 164% greater accuracy than P+T. In conclusion, PolyPred and its summary statistic-based analogues substantially improve cross-population polygenic prediction accuracy, ameliorating health disparities 13 .…”

“…Among the published methods, BOLT-LMM attained the highest prediction accuracy in all target populations (differences between BOLT-LMM and SBayesR were small and not statistically significant, but the difference between BOLT-LMM and PRS-CS was statistically significant in non-British Europeans); as noted above, the higher accuracy of BOLT-LMM vs. SBayesR and PRS-CS does not imply that BOLT-LMM is a superior method, as BOLT-LMM analyzes individual-level training data whereas SBayesR and PRS-CS analyze summary statistics. P+T was much less accurate than the other methods (despite its widespread recent use 11,[13][14][15][16][17][18]23,31,[52][53][54][55][56] ), suffering relative losses of 37-50% vs. BOLT-LMM. We thus used BOLT-LMM as a benchmark, conservatively assessing the statistical significance of improvements vs. BOLT-LMM via genomic block-jackknife across 200 genomic regions (Methods).…”

“…PolyPred-S and PolyPred-P performed slightly worse than PolyPred for both trait architectures, but were substantially and significantly more accurate than the corresponding constituent methods (SBayesR for PolyPred-S, PRS-CS for PolyPred-P). Among the remaining methods, BOLT-LMM was consistently the most accurate and P+T was consistently the least accurate method, far underperforming the other methods (despite its widespread recent use 11,[13][14][15][16][17][18]23,31,[52][53][54][55][56] ). We note that the higher accuracy of BOLT-LMM vs. SBayesR and PRS-CS does not imply that BOLT-LMM is a superior method, as BOLT-LMM analyzes individual-level training data whereas SBayesR and PRS-CS analyze summary statistics (there also exist other methods that analyze individual-level training data, e.g.…”

“…Polygenic risk scores (PRS) can identify individuals at elevated risk of complex diseases, providing opportunities for preventative action [1][2][3][4][5][6] . However, many studies have shown that PRS based on European training data attain lower accuracy when applied to populations of non-European ancestry [7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][25][26] . This loss of accuracy is primarily driven by LD differences [12][13][14][15] , allele frequency differences (including populationspecific SNPs) 13,14,27 , and causal effect size differences [12][13][14][28][29][30][31] , though differences in heritability also play a minor role 13,14,32 .…”

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

“…Polygenic risk scores (PRS) can identify individuals at elevated risk of complex diseases, providing opportunities for preventative action [1][2][3][4][5][6] . However, many studies have shown that PRS based on European training data attain lower accuracy when applied to populations of non-European ancestry [7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][25][26] . This loss of accuracy is primarily driven by LD differences [12][13][14][15] , allele frequency differences (including populationspecific SNPs) 13,14,27 , and causal effect size differences [12][13][14][28][29][30][31] , though differences in heritability also play a minor role 13,14,32 .…”