Interaction of Resistin and Systolic Blood Pressure in Psoriasis Severity

To the Editor, in a prospective cohort study, Snekvik et al. (2017) observed that obesity doubles the incidence risk for psoriasis. This finding may be attributable to the inflammatory role of adipokines: adipose-tissue associated hormones including resistin, leptin, adiponectin, and others. Resistin induces inflammatory markers, including TNF-alpha and IL-6 (Johnston et al. 2008). Independent associations have been reported between psoriasis, elevated serum resistin levels, and cardiovascular disease (CVD) (Muse et al. 2015; Huang et al. 2015; Takahashi et al. 2013; Pina et al. 2015). Furthermore, resistin levels are elevated in hypertensive patients (Zhang et al.2017; Papadopoulos Dimitrios P. et al. 2007) and are positively correlated with systolic blood pressure (SBP) (Norata et al. 2007; Makni et al. 2013), suggesting resistin may be a risk factor for hypertension, or vice versa. Hypertension, adipocyte activation, resistin recruitment of TNF/IL-6 and subsequent skin inflammation may lead to psoriasis intensification in susceptible cohorts. Resistin, therefore, is a possible link between psoriasis and increased cardiovascular risk.

We initially examined the correlation between resistin levels and psoriasis severity (expressed as Psoriasis Area and Severity Index (PASI) score) in a primary patient cohort identified through the Murdough Family Center for Psoriasis. Resistin, leptin, HDL, and LDL levels in the serum of these patients were measured. We included only those participants with quantified resistin levels (n=100). In Figure 1, correlation between PASI and resistin levels are shown, with outliers winsorized to the 5^th and 95^th percentiles (Ghosh and Vogt 2012). Although outliers were found, assessment of all other measures of these patients were within standard levels for our cohort. Therefore we have presented all resistin and PASI data and probed for whether or not this relationship may be modified by covariates including: sex, race, smoking, atherosclerosis status, age, BMI, LDL/HDL ratio, and SBP (transformed from continuous to a binary variable based on median SBP (normal: <131 mmHg, high: >131mmHg)). We have depicted this data and correlation analysis between PASI and resistin stratified by systolic blood pressure category. Supplementary Figure 1 (S1) depict resistin-PASI relationship for those with 1) Low SBP <131mmHg (n=50), 2) High SBP > 131mmHg (n=50), and combined, with scatter representing observed data and solid colored lines representing unadjusted trend lines.

Figure 1:

PASI-resistin relationship stratified by systolic blood pressure category. Graphs depict resistin-PASI relationship for those with 1) Low SBP <120mmHg (n=34), 2) High SBP > 120mmHg (n=66), and 3) all patients (n=100), with scatter representing observed data and solid lines representing unadjusted trend lines. Outliers were winsorized to the 5^th and 95^th percentiles (Ghosh 2012).

Table 1 (column 1) summarizes patient characteristics. Median PASI score was 7.95 [3.6–15.5]. Median resistin was 6,362.6 [2,883–12,079] pg/L. Mean SBP was 128.1 + 16.8 mmHg/ml. The median SBP was 131 mmHg/ml. The median resistin in the non-hypertensive group was 6,012.8 pg/L compared to a median resistin of 6,750.7 pg/L in the hypertensive group (p=0.39, Wilcoxon test).

Table 1.

Multivariable analyses of PASI-resistin association.

Characteristic (n=100)	Descriptive Median (IQR) or N%	Adjusted Model 12		Adjusted Model 23
		Coefficient (95% CI)	P value	Coefficient (95% CI)	P value
PASI	7.95 (3.6 – 15.5)
Resistin (pg/L)	6362.62 (2883 – 12079)	0.0004 (0.0002 – 0.0006)	<0.0001	0.00009 (0.0002 – 0.0004)	0.562
Age (Years)	46.2 (33.4 – 58.1)	−0.12 (−0.27 – 0.03)	0.117	−0.12 (−0.26 – 0.03)	0.114
Female	48.0%	4.12 (−0.41 – 8.67)	0.057	4.19 (−0.24 – 8.62)	0.064
Caucasian	87.0%	--	--	--	--
Smoking	38.0%	--	--	--	--
Current use of ACE inhibitors	14.0%	--	--	--	--
BMI (kg/m2)	28.8 (24.3 – 34.4)	--	--	--	--
Atherosclerosis1	60.7%	--	--	--	--
Psoriatic Arthritis	14%	--	--	--	--
SBP (mmHg)	131 (115.5–139)	--	--	--	--
High Systolic Blood Pressure (>131 mmHg)	50%	5.41 (0.53 – 10.30)	0.030	1.02 (−5.09 – 7.12)	0.741
Resistin * SBP category (>131mmHg vs. <131mmHg)	--	--	--	0.00045 (0.00007 – 0.0008)	0.024
LDL/HDL Ratio	2.3 (1.8 – 3.0)	--	--	--	--

Abbreviations: SBP: systolic blood pressure; ACE: angiotensin converting enzyme; BMI: body mass index; LDL: low-density lipids; HDL: high-density lipids; CI: confidence interval

¹Data regarding atherosclerosis status was available in 89% of the patients.

²Model 1 was determined based on backward selection of covariates and adjusts for sex (male vs. female), systolic blood pressure category (>131 mmHg vs. <131mmHg), and age.

³Model 2 was determined based on backward selection of covariates and adjusts for sex (male vs. female), systolic blood pressure category (>131 mmHg vs. <131mmHg), age, and interaction term of resistin*SBP category

Multivariable analysis revealed a positive relationship between resistin and PASI score (p<0.0001) after adjusting for age, sex, and SBP (Table 1, adjusted model 1), supporting evidence of an association between resistin and psoriasis severity.

Because previous findings demonstrated that psoriatic patients are more likely to develop hypertension, and that hypertensive patients have elevated resistin levels, we were also interested in exploring whether SBP influenced the resistin-PASI relationship (Zhang et al. 2017; Papadopoulos Dimitrios P. et al. 2007; Armstrong, Harskamp, and Armstrong 2013; Qureshi et al. 2009)). To determine if the PASI-resistin relationship varied between hypertensives and non-hypertensives, we included an interaction term to test for effect modification by SBP in our multivariable model:

$$\text{E}[\text{PASI}]={ß}_0+{ß}_1\text{ Resistin }+{ß}_2\text{Sex}+{ß}_3\text{Age}+{ß}_4\text{SBP}+{ß}_5\text{Resistin*SBP}$$

Covariates that remained statistically significant at the α=0.10 level were sex and the interactive variable of resistin and binary SBP (Table 1, adjusted model 2). These findings suggest that SBP is an effect modifier on the relationship between PASI and resistin in our cohort. The model indicates that among those with low/normal SBP, every 10,000 pg/L increase in resistin is associated with a 0.9-point increase in PASI score (covariates held constant). By contrast, among those with high SBP, a 10,000 pg/L increase in resistin is associated with a 5.4-point increase in PASI score (covariates held constant), a 6-fold difference.

These findings suggest that resistin is positively associated with PASI in our cohort, and that this relationship is augmented by SBP. Conversely, in patients with high blood pressure, increasing PASI severity leads to elevated resistin levels (results not shown), which may mark a population that needs additional cardiovascular attention. This demonstrates the importance of exploring interaction terms to assess whether the effect of an exposure varies across different groups (Corraini et al. 2017; Gail and Simon, 1985). Effect modification assesses heterogeneity across cohorts which may be used to inform treatments and research (Wang et al. 2007). For example, pravastatin’s efficacy in reducing coronary events was greater in patients with elevated baseline LDL (Sacks et al. 1996; Wang et al. 2007). It is particularly important for relationships involving overlapping pathways, such as that between psoriasis and cardiovascular risk (Norata et al. 2007; Makni et al. 2013). In the search for cardiovascular risk biomarkers, interaction term analyses may indicate that psoriasis patients cannot be treated as one group, revealing endotypes with different risks/needs.

Thus, in the psoriasis subset with hypertension, resistin elevation appears to associate with more intense psoriasis expression and may biomark a higher CVD risk population via adipocytokine-mediated inflammation. One limitation of our study is in its cross-sectional nature; therefore, the association of the interaction term of resistin and SBP with psoriasis disease severity warrants further research into biomarkers of CVD in psoriasis. Furthermore, these findings are purely correlative, not causative, and demand further study. The findings presented here build on prior work on CVD-psoriasis biomarkers which demonstrated that myeloperoxidase (MPO), a pro-inflammatory hemeprotein associated with CVD events, is elevated 2.5-fold in psoriasis patients (Cao et al. 2013; Dilek et al. 2016). We contend that resistin and SBP are additional biomarkers to consider for cardiovascular risk among psoriasis patients.

Data came from a psoriasis prevalence data set from University Hospitals Cleveland Medical Center and the Murdough Family Center for Psoriasis, Cleveland, Ohio with data on 276 individuals with skin disorders. Study participants completed a detailed questionnaire including socio-demographics, smoking history, current medications and treatments, psoriatic arthritis, and cardiovascular disease. Resistin, leptin, HDL, and LDL levels in serum were measured. We included only those participants with quantified resistin levels (n=100).

Participant characteristics were reported as medians with interquartile ranges for non-parametric variables and as means with standard deviations for parametric variables. Proportions were reported for binary variables. Univariate regression was performed using independent t-tests to test each covariate against PASI score at a significance level of α=0.10. Potential covariates were selected based on existing literature and backward selection using Akaike’s information criteria (AIC) to reduce overfitting. Model diagnostics included independence of residuals, heteroscedasticity, multicollinearity, and influential outliers. Based on existing literature, we were interested in potential effect modification of SBP on the relationship between PASI score and resistin. For clinical interpretability, we transformed SBP from a continuous to a binary variable.

As the study is exploratory in nature, p-values are reported without correction for multiple testing. All analyses were performed in Stata 14 (College Station, TX).