Making Novel Genetic Associations With Carotid Intima-Media Thickness Using the UK Biobank

Carotid artery intima-media thickness (cIMT) can be noninvasively assessed by ultrasound measurement and has been widely used as a surrogate for subclinical atherosclerosis and predictor of cardiovascular events¹. So far only a few large genome-wide association studies (GWAS) using cIMT phenotyping have been conducted^{2, 3}. In the current issue of ATVB, Strawbridge et al. present the largest single study to date originating from the UK Biobank, which recruited 500,000 people aged 40–69 years between 2006 and 2010 across the United Kingdom. With data from consistent cIMT measurements in 22,179 patients available, the authors identified 4 novel loci associated with mean cIMT and validated 7 of 11 previously reported loci with two of them previously associated with ischemic heart disease⁴. A summary of the new and known loci associated with mean cIMT and carotid plaque is given in Table 1.

Table 1.

Newly Identified and Known Genome-Wide Associated Loci for cIMT and Carotid Plaque.

Trait	Chr:Position	Gene	SNP	Variant position	EAF	P-Value	Beta	Lipid/BP/CHD/Stroke	REF
Newly identified loci for cIMT
cIMT mean	8:123594919	FBXO23 - KLHL38	rs34557926	Intergenic	0.41	2.9×10−12	−0.011		4
cIMT mean and max	7:35427416	LOC401324 – HERPUD2	rs342988	Intergenic	0.28	3.0×10−11	0.011		4
cIMT mean	19:40827247	CYP2F2P - CYP2A6	rs111689747	Intergenic	0.01	1.2×10−10	−0.044		4
cIMT mean in women	5:83600825	VCAN - HAPLN1	rs309563	Intergenic	0.34	8.9×10−10	−0.01	Yes	4
cIMT mean	11:103798234	MIR4693 - PDGFD	rs2019090	Intergenic	0.29	8.2×10−9	0.009	Yes†	4
cIMT max	11:20533244	PRMT3 - SLC6A5	rs11025608	Intergenic	0.44	3.2×10−8	−0.01		4
Known Loci for cIMT and carotid plaque
cIMT	19:44908822	APOE – APOC1	rs7412	Missense	0.08	1.0×10−14	0.0119	Yes†	2, 20
cIMT	8:10748714	PINX1	rs200482500	Intron	0.52	7.0×10−12	0.0056	Yes	2
Carotid plaque and calcification	9:22090937	CDKN2B-AS1	rs9644862	Intron	0.39	8.7×10−12	0.131	Yes†	21, 22
cIMT	8:122389299	ZHX2	rs148147734	Intergenic	0.54	3.0×10−11	0.005		2
Carotid Plaque	4:148395284	EDNRA	rs11413744	Intergenic	0.15	4.0×10−10	−0.1586	Yes	2
cIMT (smoking interaction)	13:49549513	RCBTB1	rs3751383	Synonymous	0.21	2.9×10−09	0.029		23
cIMT and carotid plaque	7:106776021	PIK3CG	rs13225723	Splice region	0.22	3.0×10−09	0.0052		2
cIMT	1:208779832	LINC01717 - LINC01774	rs201648240	Intergenic	0.83	4.0×10−09	0.0062		3
cIMT and carotid plaque	7:106770331	CCDC71L – PRKAR2B	rs12705390	Intergenic	0.17	5.0×10−09	0.0049	Yes	3
cIMT	16:88900259	CBFA2T3	rs844396	Intron	0.30	6.0×10−09	0.0051		3
cIMT	8:6628512	MCPH1	rs2912063	Intron	0.71	9.0×10−09	0.0045	Yes	3
cIMT	8:8347494	PRAG1	rs11785239	Intron	0.65	9.0×10−09	0.0043		3
Carotid Plaque	19:11078623	LDLR	rs200495339	Intergenic	0.11	1.0×10−08	−0.1023	Yes†	3
Carotid Plaque	10:102727625	SFXN2	rs2902548	Intron	0.21	2.0×10−08	−0.141		21
cIMT in rheumatoid arthritis	3:25596864	RARB	rs116199914	3 prime UTR	0.01	2.0×10−08	0.142		24
cIMT in HIV infection	15:33613209	RYR3	rs2229116	Missense	0.18	3.0×10−08	-	Yes	25
cIMT	10:112651239	VTI1A	rs11196033	Intron	0.48	4.0×10−08	0.0042		3
cIMT	5:82342097	ATP6AP1L	rs224904	Intron	0.95	5.0×10−08	0.0088		3
cIMT	6:143287831	AIG1	rs6907215	Intron	0.27	5.0×10−08	0.004		3
cIMT	2:21063185	APOB - TDRD15	rs515135	Intergenic	0.18	8.0×10−08	0.0487	Yes†	3
cIMT	2:241654811	ATG4B	rs139302128	Intron	0.03	8.0×10−08	0.0487		3
cIMT in HIV infection	5:11047569	CTNND2	rs2907092	Intron	0.21	2.0×10−07	0.02498	Yes	26
cIMT	1:190842358	LOC440704 - RGS18	rs7529733	Intergenic	0.02	3.0×10−07	0.0431		26
cIMT in rheumatoid arthritis	12:62943756	RF00019 - RPL14P1	rs1695024	Regulatory region	0.23	3.0×10−07	0.031	Yes	24
cIMT in HIV infection	8:117750197	MED30 - EXT1	rs2280828	Intergenic	0.05	3.0×10−07	0.02708		26
cIMT	2:203407367	RAPH1 - ABI2	rs1376877	Intron	0.42	4.0×10−07	-		27
cIMT in rheumatoid arthritis	11:129982285	PRDM10	rs111703287	Intron	0.01	4.0×10−07	0.119		24
cIMT (sex interaction)	5:154227372	GALNT10	rs2081015	Intron	0.41	5.0×10−07	0.033		28
cIMT and carotid plaque	16:75298143	CFDP1	rs4888378	Intron	0.48	6.8×10−07	-	Yes†	29
cIMT in HIV infection	7:126684537	GRM8	rs17691394	Intron	0.20	9.0×10−07	-		25

Only lead SNP in selected loci with a significant genome-wide p-value < 1.0×10⁻⁰⁶ are shown here. SNP: Single Nucleotide Polymorphism; EAF: Effect Allele Frequency; BP: Blood Pressure; CHD: Coronary Heart Disease; REF: Reference; cIMT: Carotid Intima-Media Thickness. The loci that are associated with lipid levels, blood pressure, coronary heart disease, or stroke are indicated as “Yes”.

^†The loci that are associated with coronary artery disease. Newly identified loci for cIMT were the ones from Strawbridge et al., and previous published loci are listed under known loci for cIMT and carotid plaque with reference provided.

While some of the loci identified had previous associations with factors known to be associated with cIMT, other loci appear to have no obvious relevance to the known pathophysiology cIMT. For instance, the APOE-APOC1 (rs1065853) and MCPH1 (rs2912062) loci were among the strongest loci identified in this study. APOE-APOC1 locus was associated with both cIMT mean and max, and MCPH1 locus was associated with cIMT mean. Both loci have been previously described (as rs7412 and rs2912063)³. APOE-APOC1 locus is known to be critically involved in lipoprotein expression, lipid metabolism, and coronary artery disease, factors that make its relationship to cIMT very likely. Conversely, the MCPH1 gene found on chromosome 8 encodes a DNA damage response protein, named Microcephalin 1, which may play an important role in G2/M checkpoint arrest via maintenance of inhibitory phosphorylation of cyclin-dependent kinase 1 but has no known role in vessel wall biology⁵. Mutations in this gene have been associated with primary autosomal recessive microcephaly and premature chromosome condensation syndrome. Further studies are needed to understand the function of MCPH1 in vascular disease and how it is associated with cIMT.

Among the other novel loci discovered in this study, three loci, rs34557926 (mapped gene: FBXO23 - KLHL38), rs2019090 (mapped gene: MIR4693 – PDGFD), and rs111689747 (CYP2F2P - CYP2A6), were associated with mean cIMT, one locus, rs11025608 (mapped gene: PRMT3 - SLC6A5) was associated with maximum cIMT, and another locus, LOC401324 (rs342988), showed associations with both mean and maximum IMT (Table 1). FBXO23 encodes Tetraspanin 17, which regulates VE-cadherin expression and is required for lymphocyte transmigration⁶. PDGFD is known to stimulate smooth muscle cell proliferation in atherosclerosis⁷. CYP2A6 is part of cytochrome P450 enzymes that participate in the biotransformation of several xenobiotics that has been categorized as pharmaceuticals and toxic agents include nicotine⁸. PRMT3 was shown previously to have implications in cIMT and stroke⁹. KLHL38 has been associated with cardiac electrophysiology and parameters of the QRS complex¹⁰.

Another novel aspect of this study was the sex-specific analysis of cIMT. Given the known differences in cardiovascular disease onset for men versus women such an analysis might reveal factors exclusive to each gender. While the men-only analysis largely resembled the combined sex analysis, the authors identified a female-specific locus on chromosome 5. This VCAN - HAPLN1 locus` lead SNP (rs309563) is an eQTL for a versican (VCAN) antisense molecule, VCAN1-AS1. It has no linkage disequilibrium with current or the previously reported lead SNP for this locus, which makes it a very interesting discovery.

VCAN is a large chondroitin sulphate proteoglycan present in the adventitia and intima of normal blood vessels and plays a prominent role in vascular remodeling processes. Following vascular injury, smooth muscle cells become activated by growth factors and cytokines, produced by infiltrating inflammatory cells, platelets, and the injured endothelium, and secrete hyaluronan and proteoglycans, especially versican, stabilizing the extracellular matrix¹¹. The response to retention hypothesis of atherosclerosis involves a critical role for such extracellular matrix proteins in retention of lipoproteins within the vascular wall. While there is accumulation of versican in pathological intimal thickening lesions, there is a sharp decline in early fibroatheroma, and near absence in late fibroatheroma¹². These two unambiguous phases of the necrotic core formation relative to ‘early’ or ‘late’ necrosis are in part defined on the basis of the relative extent of extracellular matrix. Early phase necrosis, a transitional lipid pool, is associated with an appreciable decrease in the expression of extracellular matrix proteins versican, hyaluronan, and biglycan together with infiltrating macrophages, which are undergoing apoptosis or necrosis. On the contrary, the necrotic core underlying the ‘late fibroatheroma’ is noticeably deficient in extracellular matrix proteins, including versican and other proteoglycans and collagen, and exhibits greater presence of free cholesterol recognized as clefts. These necrotic cores may show calcification, intraplaque hemorrhage, surrounding neoangiogenesis, and inflammation, all of which are often seen in advanced carotid plaques^13–15. Thus versican appears to contribute to intimal expansion and lesion progression whereas its degradation may be seen only in more advanced atherosclerosis. Thus, these data suggest that genetic differences between men and women leading to differential expression of VCAN in the vascular wall may be one factor underlying differences in cIMT and atherosclerosis but requires further work for confirmation.

The results of the current analysis should also be placed within the context of important limitations. Although cIMT has been adopted as a surrogate for predicting cardiovascular events, recommendations regarding the use of cIMT for risk prediction are conflicting. Little or no additional prognostic value has been found by adding cIMT to traditional risk factor scores, such as the Framingham Risk Score^16–18. Variations in both how cIMT is assessed (i.e. ultrasound methodology) and in the populations studied, especially with regards to age and ethnicity may also influence its predictive power. Because cIMT itself may not accurately reflect the atherosclerotic disease process and per se is not a direct indicator of this disease, the meaning of genetic correlations with cIMT cannot be directly be extrapolated to atherosclerosis. Among the GWAS significant loci have been discovered to date (Table 1), only 6 loci were identified in these studies that also have been associated with coronary artery disease, i.e. PDGFD, APOE-APOC1, CDKN2B-AS1 (9p21), LDLR, APOB, and CFDP1¹⁹. Nonetheless, cIMT still represents one of the few available techniques for quantitative non-invasive assessment of atherosclerosis and response to pharmacotherapies.

In closing, the work of Strawbridge et al. using the largest single study sample to date from the UK Biobank shed new light on our understanding of the genetic associations with cIMT. The identification of four novel loci and validation of 7 of 11 previously reported loci advances our understanding of the biology of cIMT and suggests novel avenues for further exploration. The finding of a sex-specific locus in women is especially exciting given its probable link with the biology of atherosclerosis. Taking age, sex, and ethnic background into consideration, the integration of genetics and cIMT might become a powerful tool to identify new targets for primary prevention of atherosclerosis.