Association of Cardiovascular and Metabolic Disease Genes with Psoriasis

To the Editor:

Psoriasis is a chronic immune-mediated and hyperproliferative disorder of the skin that affects 2–3% of the population. Psoriasis is associated with an increased incidence of several cardiovascular and metabolic co-morbidities, including coronary artery disease (CAD), hypertension, obesity, hyperlipidemia, and type 2 diabetes (T2D) (Davidovici et al., 2010). The association of these cardiovascular and metabolic diseases with psoriasis could be due to shared genetic risk variants, shared environmental triggers, activation of common inflammatory pathways, or a combination of these factors. Here, we evaluated the hypothesis that some of the increased risk of cardiovascular and metabolic diseases in psoriasis is derived from shared genetic risk factors.

Using the genome-wide association studies (GWAS) Catalog (available at www.genome.gov/gwastudies and accessed in December 2011), we selected 363 SNPs that showed significant association with CAD, hypertension, body mass index (BMI), hyperlipidemia (total, LDL and HDL cholesterol levels and triglyceride levels), or T2D. Selected SNPs met genome-wide significance (p < 5×10⁻⁸) in at least two GWAS or were significant in the latest meta-analysis of GWAS. Additional SNPs or loci were also included based on the latest expert opinions (McCarthy, 2010; O'Donnell and Nabel, 2011; Peden and Farrall, 2011). Detailed information about the selected SNPs is provided in Table S1.

We evaluated the selected cardiovascular and metabolic SNPs for association with psoriasis in four psoriasis GWAS cohorts: the GAIN cohort including 1368 psoriasis cases and 1348 controls (Nair et al., 2009), an unpublished psoriasis cohort from Sweden including 725 cases and 438 controls, a Washington University/University of California San Francisco cohort including 211 psoriasis cases and 502 controls (Liu et al., 2008), and the Wellcome Trust Case-Control Consortium (WTCCC) cohort including 2178 psoriasis cases and 5175 controls (Strange et al., 2010). Further details of the Swedish cohort are described in the Supplemental Methods. IMPUTE2 was used to impute the ungenotyped SNPs by using phase 3 HapMap and 1000 Genomes pilot project CEU haplotypes as a reference. SNPTEST was used to associate the imputed dosage for each SNP with psoriasis status separately in each study population with the adjustment of the first three principal components from an MDS analysis of population stratification. The association test results for these SNPs with relatively high confidence (PROPER_Info > 0.5) were then combined by meta-analysis with the META using inverse-variance method based on a fixed-effect model. The false discovery rate method was used to correct for multiple testing (FDR_q < 0.05).

We first examined the associations between all selected SNPs and psoriasis status (Table 1 and Table S2). After adjusting for multiple testing with the false discovery rate method, seven SNPs were associated with psoriasis status (FDR_q < 0.05, Table 1). The alleles associated with increased risk of dyslipidemia (rs2247056, rs3177928, rs492602, and rs181362), increased blood pressure levels (rs805303, rs653178, and rs3184504), and increased CAD risk (rs3184504) were associated with increased risk of psoriasis (Table 1). The top three SNPs (rs2247056, rs3177928, and rs805303) were located in the HLA gene region which is a known psoriasis susceptibility locus. After further adjustment for the top psoriasis risk allele HLA-C*06:02 (determined by imputation as described in (Chen et al., 2012), the associations for rs2247056 and rs3177928 were mitigated (p = 0.06 and 0.07, respectively), while the association for rs805303 persisted (p = 0.005). Since multiple psoriasis risk alleles are identified in HLA loci (Chen et al., 2012), we cannot rule out the significant association for rs805303 is derived from linkage disequilibrium (LD) with other psoriasis HLA risk alleles. Interestingly, we identified 4 non-HLA SNPs with evidence of shared genetic risk between psoriasis and cardiovascular and metabolic diseases: rs492603 in FUT2, rs181362 in UBE2L3, and rs653178 and rs3184504 (in complete LD with each other) near or in SH2B3. FUT2 encodes an alpha-(1,2)fucosyltransferase that determines secretor status of blood group antigens on epithelial cells and in bodily secretions and has been recently associated with susceptibility to psoriasis and Crohns’s disease (Ellinghaus et al., 2012), type 1 diabetes (Smyth et al., 2011), primary sclerosing cholangitis (Folseraas et al., 2012), and norovirus infection (Carlsson et al., 2009). UBE2L3 encodes an ubiquitin-conjugating enzyme involved in cell proliferation and immune function and is associated with susceptibility to celiac disease and rheumatoid arthritis (Zhernakova et al., 2011), Crohn’s disease (Fransen et al., 2010), and systemic lupus erythematosus (Wang et al., 2012). The adaptor protein encoded by SH2B3 plays pleiotropic signaling roles in regulating lymphocyte differentiation, induction of VCAM-1 and E-selectin on endothelial cells by TNF-α, and thrombus formation (Devalliere and Charreau, 2011), and thus might explain its dual role in susceptibility to multiple autoimmune diseases and endothelium-related cardiovascular diseases (Jin et al., 2012).

Table 1.

Top 10 metabolic SNPs associated with psoriasis status in the meta-analysis of four studies.

Chr.	SNP	Nearby Gene	Minor/major allele	Effect on metabolic phenotypes (allele effect)1	Effect on psoriasis (allele effect/p value)				β/p value (meta-analysis)	OR (95% CI)	FDR_q value3
					GAIN1,2	Sweden1,2	WashU/UCSF1,2	WTCCC1,2
6	rs2247056	HLA-C	T/C	C↑TG	C↑/3.32E-04	C↑/0.02	C↑/0.45	C↑/4.21E-38	0.43/3.25E-35	1.53(1.43–1.64)	< 0.0001
6	rs3177928	HLA-DRA	A/G	A↑TC, LDLC	A↑/5.59E-07	A↑/0.07	A↑/0.21	A↑/2.66E-29	0.45/1.20E-32	1.57(1.46–1.69)	< 0.0001
6	rs805303	BAT2, BAT5	A/G	A↓SBP, DBP, and HTN	A↓/6.17E-02	A↓/3.42E-01	A↓/7.69E-01	A↓/3.13E-07	−0.15/8.30E-07	0.86(0.81–0.91)	< 0.0001
19	rs492602	FUT2	A/G	G↑TC	G↑/3.30E-03	G↑/6.64E-01	G↑/7.16E-01	G↑/6.59E-05	0.13/8.23E-06	1.14(1.07–1.20)	0.0007
22	rs181362	UBE2L3	T/C	T↓/HDLC	T↑/1.40E-01	T↑/8.18E-01	T↑/2.62E-01	T↑/2.77E-04	0.14/1.43E-04	1.15(1.07–1.23)	0.0102
12	rs653178	ATXN2, SH2B3	C/T	T↓DBP	T↓/2.56E-02	T↑/6.65E-01	T↓/1.07E-02	T↓/5.77E-03	−0.11/2.08E-04	0.90(0.85–0.95)	0.0124
12	rs3184504	SH2B3	T/C	T↑CAD, DBP, SBP	T↑5.60E-02	T↓/4.76E-01	T↑/1.00E-02	T↑/6.51E-03	0.10/6.22E-04	1.11(1.04–1.17)	0.0318
12	rs11065987	BRAP	G/A	G↓TC, LDLC	G↑/6.61E-02	G↓/8.57E-01	G↑/2.59E-02	G↑/2.39E-02	0.09/1.81E-03	1.09(1.03–1.16)	0.0811
2	rs7593730	RBMS1	T/C	C↑T2D	C↑/3.01E-01	C↑/8.17E-03	C↓/5.81E-01	C↑/1.73E-02	0.11/2.34E-03	1.11(1.04–1.19)	0.093
10	rs7901695	TCF7L2	C/T	C↑T2D	C↓/1.87E-01	C↓/3.81E-01	C↓/1.98E-01	C↓/4.49E-02	−0.09/5.26E-03	0.92(0.86–0.97)	0.1881

Abbreviations: TC, total cholesterol levels; LDLC, low-density-lipoprotein cholesterol levels; HDLC, high-density-lipoprotein cholesterol levels; TG, triglyceride levels; SBP, systolic blood pressure; DBP, diastolic blood pressure; HTN, hypertension; CAD, coronary artery disease; T2D, type 2 diabetes; OR (odds ratios); 95% CI (95% confidence interval).

¹↑/↓, increasing/decreasing effects on relevant metabolic traits and psoriasis risk.

²All p values were adjusted by the first three principal components from an MDS analysis of population stratification.

³FDR_q, false discovery rate adjusted p value based on 358 SNPs in supplementary table 3.

With a combined 4482 psoriasis cases and 7463 controls, our meta-analysis had 80% power to detect genetic variants with OR = 1.2 at a significance level alpha of 0.05, assuming 5% of population allele frequency. Given that many cardiovascular and metabolic trait SNPs have ORs less than 1.2, we sought to determine whether cardiovascular and metabolic risk SNPs could be having a real, but more subtle effect on psoriasis susceptibility. We therefore constructed a weighted gene risk score (wGRS) to investigate the aggregate effects of the risk alleles associated with cardiovascular and metabolic traits between psoriasis patients and controls. Prior studies have shown that such genetic risk scores, which estimate an individual’s overall genetic burden, have increased ability to discriminate between cases and controls (Chen et al., 2011). The GRS were weighted according to the effect size of the risk alleles and a common set of SNPs were examined across all cohorts (see Supplemental Methods). A small but significant difference for the wGRS of CAD, total cholesterol levels, and TG levels was seen (p < 0.006 for adjusting for multiple testing of 8 traits, Table 2). The wGRS of total cholesterol levels did not reach significance after removing the top associated SNPs (rs3177928 and rs492602, p = 0.42, Table 1 and 2). No difference between psoriasis cases and controls was observed regarding the wGRS of hypertension, T2D, LDL cholesterol levels, HDL cholesterol levels, and BMI (Table 2).

Table 2.

Metabolic weighted gene risk score between controls and psoriasis cases.¹

Studies	CAD	Hypertension	T2D	TC	LDLC	HDLC	TG	BMI
Gain
Control	2.17±0.30	1.78±0.20	3.59±0.43	39.44±4.69	26.52±5.18	25.94±3.23	105.74±15.44	3.62±0.52
Case	2.20±0.30	1.79±0.20	3.59±0.43	39.73±4.78	26.22±5.33	25.82±3.20	107.40±15.13	3.65±0.49
P12	0.01	0.59	0.96	0.12	0.17	0.34	0.01	0.10
Sweden
Control	2.13±0.29	1.84±0.19	3.47±0.41	39.56±4.83	25.71±5.03	25.76±3.18	104.68±14.85	3.66±0.54
Case	2.14±0.30	1.81±0.20	3.51±0.43	39.59±5.07	26.04±5.41	25.96±3.31	104.87±14.82	3.69±0.52
P12	0.52	0.01	0.28	0.60	0.39	0.28	0.77	0.34
WashU/UCSF
Control	2.18±0.31	1.78±0.19	3.57±0.44	39.78±4.61	26.23±5.49	25.84±3.27	106.01±15.27	3.59±0.53
Case	2.19±0.28	1.76±0.20	3.57±0.42	39.82±4.64	26.47±4.75	26.26±2.99	109.62±15.93	3.54±0.52
P12	0.81	0.33	0.52	0.77	0.61	0.10	0.04	0.31
WTCCC
Control	2.16±0.30	1.78±0.20	3.54±0.44	39.39±4.70	26.27±5.27	26.01±3.19	105.1±15.26	3.64±0.52
Case	2.18±0.30	1.79±0.20	3.55±0.43	39.77±4.88	26.51±5.33	25.98±3.27	106.24±15.12	3.67±0.52
P12	0.002	0.22	0.95	0.003	0.003	0.69	0.001	0.03
Meta   OR(95%CI)	1.31(1.15–1.50)	0.88(0.60–1.29)	1.01(0.92–1.10)	1.01(1.01–1.02)	1.01(0.99–1.02)	1.00(0.99–1.01)	1.01 (1.00–1.01)	1.11(1.02–1.20)
P23	<0.00015	0.50	0.91	0.0025	0.35	0.97	<0.00015	0.01
P34	0.73	0.02	0.66	0.64	0.05	0.17	0.47	0.41

Abbreviations: CAD, coronary artery disease; T2D, type 2 diabetes; TC, total cholesterol levels; LDLC, low-density-lipoprotein cholesterol levels; HDLC, high-density-lipoprotein cholesterol levels; TG, triglyceride levels; BMI, body mass index.

¹Mean ± SD was presented; in total, 1348 controls and 1368 cases, 438 controls and 725 cases, 502 controls and 211 cases, and 5175 controls and 2178 cases, were included in the Gain, Sweden, WashU/UCSF, and WTCCC studies, respectively; the SNP lists used for constructing the weighted gene risk score for each trait were listed in Supplementary Table 2; the statistical significance for testing 8 traits was 0.006.

²P1, p value for 1 unit increase of weighed gene risk score with adjustment for gender and the covariates for population stratification using the logistic regression model.

³P2, p value for the meta-analysis over four studies.

⁴P3, p value for Cochran’s Q test of the heterogeneity.

⁵P value = 0.0003 after removing rs3184504 from the weighted gene risk score of CAD; P value = 0.42 after removing rs3177928 and rs492602 from the weighted gene risk score of TC; P value = 0.0025 after removing rs2247056 from the weighted gene risk score of TG.

Multiple cardiovascular and metabolic co-morbidities have been observed in psoriasis patients. Here we examined whether the co-manifestation of these conditions is a result of shared genetic factors. The data presented here suggest that patients with psoriasis are enriched for certain common genetic variants (HLA, FUT2, UBE2L3, SH2B3) that predispose to increased risk of dyslipidemia, hypertension, and increased CAD risk itself. Further evaluation of these genes and pathways may be important to elucidate whether particular subsets of psoriasis patients carry a higher burden of risk for the development of these cardiovascular and metabolic co-morbidities.