Development of genome-wide polygenic risk scores for lipid traits and clinical applications for dyslipidemia, subclinical atherosclerosis, and diabetes cardiovascular complications among East Asians

Tam, Claudia H. T.; Lim, Cadmon K. P.; Luk, Andrea O. Y.; Ng, Alex C. W.; Lee, Heung-man; Jiang, Guozhi; Lau, Eric S. H.; Fan, Baoqi; Wan, Raymond; Kong, Alice P. S.; Tam, Wing-hung; Ozaki, Risa; Chow, Elaine Y. K.; Lee, Ka-fai; Siu, Shing-chung; Hui, Grace; Tsang, Chiu-chi; Lau, Kam-piu; Leung, Jenny Y. Y.; Tsang, Man-wo; Kam, Grace; Lau, Ip-tim; Li, June K. Y.; Yeung, Vincent T. F.; Lau, Emmy; Lo, Stanley; Fung, Samuel; Cheng, Yuk-lun; Chow, Chun-chung; Hu, Miao; Yu, Weichuan; Tsui, Stephen K. W.; Huang, Yu; Lan, Huiyao; Szeto, Cheuk-chun; Tang, Nelson L. S.; Ng, Maggie C. Y.; So, Wing-yee; Tomlinson, Brian; Chan, Juliana C. N.; Ma, Ronald C. W.

doi:10.1186/s13073-021-00831-z

Table 3 Association between quintiles of polygenic risk scores and carotid intima-media thickness in adult women (n = 781)

From: Development of genome-wide polygenic risk scores for lipid traits and clinical applications for dyslipidemia, subclinical atherosclerosis, and diabetes cardiovascular complications among East Asians

Polygenic risk score	Cohorts of adult women	Quintile of polygenic score						Model 1		Model 2
Polygenic risk score	Cohorts of adult women	Q1	Q2	Q3	Q4	Q5	Test	P_{model 1}	P_Q	P_{model 2}	P_Q
PRS_TC	Cohort 1	0.0481 (0.0466–0.0497)	0.0496 (0.0482–0.0511)	0.0503 (0.0487–0.0520)	0.0501 (0.0486–0.0516)	0.0497 (0.0482–0.0514)	Linear trend	0.0534		0.1003
	Cohort 2	0.0481 (0.0450–0.0515)	0.0504 (0.0453–0.0561)	0.0529 (0.0479–0.0584)	0.0550 (0.0496–0.0610)	0.0512 (0.0479–0.0547)	Linear trend	0.1032		0.0758
							Meta-analysis	0.0182	0.3547	0.0315	0.2560
	Cohort 1	0.0495 (0.0488–0.0503)				0.0497 (0.0482–0.0514)	Top 20% vs others	0.6893		0.7445
	Cohort 2	0.0518 (0.0493–0.0544)				0.0512 (0.0479–0.0547)	Top 20% vs others	0.8236		0.9732
							Meta-analysis	0.7591	0.7336	0.7667	0.8891
PRS_TG	Cohort 1	0.0499 (0.0485–0.0514)	0.0498 (0.0481–0.0515)	0.0491 (0.0475–0.0507)	0.0494 (0.0481–0.0507)	0.0498 (0.0480–0.0516)	Linear trend	0.7416		0.7173
	Cohort 2	0.0515 (0.0471–0.0563)	0.0514 (0.0481–0.0548)	0.0512 (0.0477–0.0549)	0.0542 (0.0457–0.0643)	0.0507 (0.0469–0.0548)	Linear trend	0.6197		0.9116
							Meta-analysis	0.6430	0.7068	0.7048	0.9963
	Cohort 1	0.0495 (0.0488–0.0503)				0.0498 (0.0480–0.0516)	Top 20% vs others	0.9536		0.9879
	Cohort 2	0.0518 (0.0495–0.0542)				0.0507 (0.0469–0.0548)	Top 20% vs others	0.4509		0.9879
							Meta-analysis	0.8786	0.4574	0.9917	0.9849
PRS_HDL	Cohort 1	0.0512 (0.0497–0.0527)	0.0490 (0.0475–0.0505)	0.0496 (0.0479–0.0513)	0.0483 (0.0469–0.0498)	0.0498 (0.0482–0.0514)	Linear trend	0.3786		0.3242
	Cohort 2	0.0511 (0.0470–0.0556)	0.0542 (0.0466–0.0630)	0.0498 (0.0459–0.0539)	0.0532 (0.0491–0.0576)	0.0508 (0.0475–0.0542)	Linear trend	0.9307		0.8528
							Meta-analysis	0.4194	0.7165	0.3205	0.8838
	Cohort 1	0.0512 (0.0497–0.0527)	0.0492 (0.0484–0.0500)				Bottom 20% vs others	0.0212		0.0160
	Cohort 2	0.0511 (0.0470–0.0556)	0.0518 (0.0495–0.0542)				Bottom 20% vs others	0.4012		0.4524
							Meta-analysis	0.0533	0.1286	0.0405	0.1373
PRS_LDL	Cohort 1	0.0481 (0.0467–0.0495)	0.0498 (0.0484–0.0513)	0.0489 (0.0475–0.0503)	0.0514 (0.0494–0.0536)	0.0495 (0.0482–0.0509)	Linear trend	0.0485		0.0954
	Cohort 2	0.0509 (0.0452–0.0574)	0.0516 (0.0458–0.0581)	0.0533 (0.0485–0.0585)	0.0545 (0.0504–0.0590)	0.0490 (0.0465–0.0517)	Linear trend	0.9030		0.7916
							Meta-analysis	0.0570	0.5840	0.0972	0.7430
	Cohort 1	0.0496 (0.0488–0.0504)				0.0495 (0.0482–0.0509)	Top 20% vs others	0.7463		0.7438
	Cohort 2	0.0526 (0.0500–0.0553)				0.0490 (0.0465–0.0517)	Top 20% vs others	0.2775		0.3636
							Meta-analysis	0.9340	0.2564	0.9800	0.3328

Data are presented as geometric mean (95% CI) stratified by quintile categories of polygenic risk score. Intima-media thickness was natural log (ln) transformed. P_{model 1} and P_{model 2} values were obtained from linear regression with the adjustment for covariates included in models 1 and 2, respectively. Model 1: principal components and age. Model 2: principal components, age, body mass index, and systolic blood pressure. Results from individual cohorts were meta-analyzed using a fixed effects model. P_Q refers to the p value of Cochran’s Q-statistics in heterogeneity test. Associations between PRSs and carotid intima-media thickness were examined in two different ways: (1) we examined a linear trend across the quintile categories. (2) We tested a hypothesis that a high PRS for TC, TG, and LDL-C (a low PRS for HDL-C) was associated with intima media thickness by comparing the top (bottom) 20% with the remaining 80% of the PRS distribution

Back to article page

ISSN: 1756-994X

Contact us

Submission enquiries: editorial@genomemedicine.com
General enquiries: info@biomedcentral.com

Genome Medicine

Contact us