A Common CD36 Variant Influences Endothelial Function and Response to Treatment with Phosphodiesterase 5 Inhibition

Context:

The scavenger receptor CD36 influences the endothelial nitric oxide-cGMP pathway in vitro. Genetic variants that alter CD36 level are common in African Americans (AAs), a population at high risk of endothelial dysfunction.

Objective:

To examine if the minor allele (G) of coding CD36 variant rs3211938 (G/T) which reduces CD36 level by approximately 50% influences endothelial function, insulin sensitivity (IS), and the response to treatment with the nitric oxide-cGMP potentiator sildenafil.

Design:

IS (frequently sampled iv glucose tolerance) and endothelial function (flow mediated dilation [FMD]) were determined in age- and body mass index-matched obese AA women with or without the G allele of rs3211938 (protocol 1). Effect of chronic sildenafil treatment on IS and FMD was tested in AA women with metabolic syndrome and with/without the CD36 variant, using a randomized, placebo-controlled trial (protocol 2).

Setting:

Two-center study.

Participants:

Obese AA women.

Intervention:

A total of 20-mg sildenafil citrate or placebo thrice daily for 4 weeks.

Main outcome:

IS, FMD.

Results:

G allele carriers have lower FMD (P = .03) and cGMP levels (P = .01) than noncarriers. Sildenafil did not improve IS, mean difference 0.12 (95% confidence interval [CI], −0.33 to 0.58; P = .550). However, there was a significant interaction between FMD response to sildenafil and rs3211938 (P = .018). FMD tended to improve in G carriers, 2.9 (95% CI, −0.9 to 6.8; P = .126), whereas it deteriorated in noncarriers, −2.6 (95% CI, −5.1 to −0.1; P = .04).

Conclusions:

The data document influence of a common genetic variant on susceptibility to endothelial dysfunction and its response to sildenafil treatment.

A common polymorphism in the CD36 gene contributes to endothelial dysfunction and influences the response to treatment with PDE-5 inhibition in African Americans.

African Americans (AAs) are susceptible to the development of endothelial dysfunction with multiple studies showing decreased endothelial-dependent vasodilation in response to acetylcholine (1, 2), methacholine (2), bradykinin (2), isoproterenol (3, 4), and ischemia (5). Endothelial dysfunction has been associated with reduced vascular actions of insulin, resulting in impaired peripheral glucose disposal (6, 7). Of note, AA women have reduced insulin sensitivity (IS) (8–11) and higher prevalence of type 2 diabetes mellitus (T2DM) than whites (12).

Endothelial dysfunction reflects in part reduced bioavailability of nitric oxide (NO) a vasodilator with additional antiinflammatory effects (13). Blood vessels are actively preserved by multiple signaling programs that include the regulated endothelial production of NO by endothelial NO synthase (14, 15). NO activates soluble guanylate cyclase (sGC) in vascular smooth muscle producing cGMP and vasodilation (16).

CD36 is a widely expressed scavenger receptor with diverse ligands, lipid (eg, long-chain fatty acids, native and oxidized lipoproteins) and nonlipid (eg, thrombospondin 1 [TSP1] and collagen). This receptor is abundant in vascular endothelial cells, especially in heart, skeletal muscle, and adipose tissue (17), where it facilitates high-affinity tissue fatty acid (FA) uptake (18, 19) and cellular FA metabolism (20). TSP1 binding to CD36 (21, 22) or its other receptor CD47 (23) suppresses NO-cGMP signaling in cultured endothelial cells, but whether CD36 contributes to NO-mediated vascular effects in vivo remains unknown. In the present study, we tested the hypothesis that the G allele of CD36 coding single nucleotide polymorphism (SNP) rs3211938 (G/T), which is 10 times more common in AA (genotype frequency of ∼20%) as compared with Caucasians and results in approximately 50% reduced CD36 expression (24) would associate with altered endothelial function (protocol 1).

Previous studies showed that chronic inhibition of phosphodiesterase 5 (PDE-5), which potentiates NO function by preventing cGMP hydrolysis and improves endothelial function in subjects with T2DM and IS in subjects with prediabetes (25). In protocol 2, we tested the hypothesis that chronic treatment with sildenafil would improve IS and endothelial function in AA and that the CD36 coding SNP rs3211938 might impact the response to treatment.

Materials and Methods

Protocol 1: Impact of CD36 rs3211938 G allele on FAs, cGMP levels, and endothelial function in obese AA women

Subjects

Recruitment of unrelated lean and obese AA women was conducted at Vanderbilt University and at Washington University. We used the same recruitment method at both sites; unrelated AA women that met the same eligibility criteria (exclusion and inclusion) were recruited from the community through printed and email advertisement. At both sites, eligibility included African ancestry and only subjects who reported both parents of the same race. Exclusion criteria included diabetes as defined by fasting glucose levels more than or equal to 126 mg/dL and hemoglobin 1AC (HbA_1c) more than or equal to 6.5 (26), use of nicotinic acid or statins, history of cholecystectomy, and history of alcohol or drug abuse. Subjects completed a medical history and physical examination. Venous blood was collected after an overnight fast for DNA extraction, glucose, insulin, lipid profile, and free FA. The combined cohort was used for analyzing the effect of the coding SNP on plasma metabolic parameters and for testing allele frequency in a large cohort that is metabolically similar to the smaller one used for our clinical study (protocol 2). The effect of the CD36 SNP on CD36 protein expression on monocytes, platelets, and on plasma fatty acids were determined in participants recruited at both sites.

Evaluation of endothelial function including cGMP measurements and assessment of flow-mediated dilation (FMD) were performed on a subset of the subjects recruited at Vanderbilt University after discontinuing for 2 weeks antihypertensive or lipid-lowering medications, vitamins, and antioxidants. Measurements were performed early in the morning after a minimum 8-hour fast, in a quiet, temperature-controlled room (22°C–23°C). Subjects were kept in the supine position for 20 minutes before starting FMD measurements as previously described (27). Blood samples were obtained for fasting free FA, glucose, insulin, and estradiol levels. All studies adhered to the principles of the Declaration of Helsinki and Title 45, United States Code of Federal Regulations, part 46, Protection of Human Subjects. All subjects provided informed consent. Institutional Review Boards at Vanderbilt and Washington Universities approved the studies.

Protocol 2: Effect of PDE-5 inhibition on endothelial function and metabolic parameters in AA women with metabolic syndrome and with vs without the G allele of rs3211938

Subjects

Recruitment was conducted at Vanderbilt University. Eligibility included obesity (body mass index [BMI], 30–50 kg/m²), African ancestry based on 4 grandparents, either elevated fasting serum insulin levels (≥13 μU/mL) or diagnosis of metabolic syndrome (MetS) according to the National Cholesterol Education Program, Adult Treatment Plan III (28). Subjects were excluded if they had severe chronic illness, T2DM, cardiovascular disease other than hypertension, and if they reported use of nitrates, adrenergic α-blocking agents or protease inhibitors known to affect sildenafil's metabolism.

The study was a randomized, double-blind, placebo-controlled, 2-arm parallel design (NCT01334554). Participants reported to the Vanderbilt Clinical Research Center for screening that included a medical history, physical examination, laboratory analyzes (blood cell count, lipid profile, metabolic panel, and fasting insulin) and body composition evaluation (dual energy x-ray absorptiometry, Luna iDXA; GE Healthcare). Subjects were randomly assigned (1:1) to sildenafil citrate 20 mg (Revatio; Pfizer. Inc) or placebo, both taken thrice daily (blinded medication supplied by Pfizer through an investigator-initiated research grant). The scheme was random permuted block, stratified by fasting insulin levels (<13 or >13 μU/mL). Subjects and investigators were blinded to assigned treatments until the study's end. Subjects were evaluated at baseline and 4 weeks after receiving the assigned treatment. Genotyping for rs3211938 was performed at the end of the intervention.

The primary outcomes were IS assessed with the frequently sampled iv glucose tolerance test and endothelial function assessed by FMD. Secondary outcomes included pulse wave amplitude during reactive hyperemia by peripheral artery tonometry (EndoPAT 2000 device; Itamar Medical, Inc), blood pressure (BP), and heart rate (HR) using the VITAL-GUARD 450c monitor (Ivy Biomedical Systems), HbA_1c, the acute insulin response to glucose (AIRG), and disposition index (DI) estimated from frequently sampled iv glucose tolerance test. At the end of the 4-week treatment, blood sildenafil concentration was determined to assess treatment compliance.

Flow-mediated dilation

Vascular images were obtained using a 9.3-MHz linear array vascular probe attached to a high-resolution ultrasound machine (iU22; Phillips). The right brachial artery of the dominant arm was scanned over a longitudinal section of 3–5 cm above the elbow and a pneumatic cuff was placed at 5 cm distal to the elbow. The brachial artery diameter was measured at baseline over 30 seconds. Afterwards, the cuff was inflated to a pressure of 50 mm Hg above the systolic BP (SBP) for 5 minutes to provide the ischemic stimulus. The cuff was deflated, and images of the artery were obtained continuously for 180 seconds after cuff release and analyzed using a continuous edge detection and wall tracking software (Brachial Analyzer 5.0; Medical Imaging Applications LLC). The analyst (J.E.C.) was blinded to the intervention.

Frequently sampled iv glucose tolerance test

Two iv accesses were placed in both arms. Baseline samples were collected at t = −15 minutes and −5 minutes for measuring glucose and insulin. At t = 0, a bolus of 300-mg glucose/kg body weight in 20% dextrose was administered over 1 minute into a large vein with continuous monitoring to reduce the risk of infiltration. At t = 20 minutes, 0.02-units/kg body weight of regular insulin (Actrapid; Novo Nordisk) was administered iv. Blood samples were collected at different times up to 180 minutes after dextrose infusion for measuring glucose and insulin. Metabolic parameters (IS, AIRG, and DI) and homeostasis model assessment of insulin resistance (HOMA-IR) were calculated using a modified version of minimal model (MINMOD Millennium) formulas (29).

SNP genotyping and CD36 expression

Genomic DNA was extracted from peripheral blood with a salting-out precipitation (Gentra Puregene Blood kit) and CD36 SNP rs3211938 was detected using a predesigned TaqMan SNP Genotyping Assay (Applied Biosystems) on a 7900HT instrument. CD36 expression was determined as previously described (24).

Blood analyzes

Plasma glucose was measured with a glucose analyzer (YSI Life Sciences), insulin and cGMP by RIA, respectively, from Millipore and Amersham (Pharmacia Biotech AB). Free FA was measured using NEFA-HR (2) (Wako, Sopachem BV), total cholesterol (Diagnostics Chemicals), and triglycerides (Roche Diagnostics Corp) by enzymatic colorimetry. F2-isoprostanes were assayed in plasma using negative ion gas chromatography mass spectroscopy as previously described (30). Sildenafil levels were assessed by liquid chromatography-tandem mass spectrometry on silica column with aqueous-organic mobile phase as previously reported (31).

Statistical analysis

Baseline characteristics are presented as the mean ± SD for protocols 1 and 2. Student's t test was used to examine mean differences in free FA, cGMP, and time to peak. Multiple linear regression was used to examine differences in mean FMD% between genotype groups (G/T vs T/T), after adjusting for weight differences (protocol 1). For protocol 2, multiple linear regression was used to examine whether there would be differences in the mean of each parameter and outcomes between sildenafil and placebo groups or between genotype groups (G/T vs T/T), after adjusting for the baseline value of each outcome (protocol 2). Analyzes for IS, AIRG, and DI included age, fat mass, and fat-free mass as covariates in addition to the baseline value. The interaction between sildenafil and genotype groups was examined to determine whether the sildenafil effect would differ depending on genotype. The final model did not include the interaction in all analyzes except for those of endothelial function. Log-transformed variables (to reduce skewness) for IS and AIRG were included in the models. Analyzes were performed with STATA 11.0 (StataCorp) and the programming language R version 3.2.1 (R Development Core Team. R: A Language and Environment for Statistical Computing. Vienna, Austria, 1–1820, 2005.EF).

The calculated sample size for protocol 2 was 19 subjects per study arm. This sample size was based on a 0.8 power to detect a 50% increase in IS. We recruited 46 obese AA women to account for approximately 13% dropout rate. A previous study showed a significant difference in IS between sildenafil treatment and placebo with a similar sample size (25).

Results

Protocol 1: Impact of rs3211938 G allele on free fatty acids, cGMP levels, and endothelial function

We enrolled a total of 103 AA women at Vanderbilt University and Washington University. The same recruitment method was used at both centers and included unrelated AA women that met the same eligibility criteria (exclusion and inclusion as described under Materials and Methods). There were no differences in the subjects' characteristics when stratified by center (data not shown); 73% were obese and hypertensive, 26 individuals were carriers of the G allele, 2 were homozygous for the G allele, and 75 were noncarriers. There were no differences in age, BMI, total cholesterol, triglycerides, or fasting insulin and glucose levels among carriers vs noncarriers (Table 1). The allelic frequency of the G variant, meaning the percent of G allele related to the total number of alleles (total chromosomes) present in the population was 15% in our cohort similar to that previously reported for the general African population (9%–10%) (32, 33). Genotype frequency, meaning the percent of subjects carrying the G allele in the population was 27% in our cohort thus involving a quarter of the population tested. Genotype frequency is higher because organisms are diploid as reflected in the following formula: f = p2 + 2pq + q2, where f = frequency, p = frequency of the major allele T, and q = frequency of the minor allele G, with p + q equal to 1. The presence of the G allele, as expected (32), results in about 50% reduction of CD36 protein expression in monocytes and platelets (Supplemental Figure 1A). Fasting free FA levels were 20% higher in G allele carriers, P = .01, as compared with T/T controls consistent with CD36's role in the clearance of circulating FA (Supplemental Figure 1B).

Table 1.

Protocol 1. Demographic Characteristics of individuals Recruited at Vanderbilt and Washington Universities

Parameters	CD36 rs3211938		P Value
	T/T (n = 75)	G/T, G/− (n = 28)
Age, years	42 ± 8.90	43 ± 10.60	0.62
BMI, kg/m2	35 ± 6.80	37 ± 8.17	0.32
Cholesterol, mg/dL	194 ± 31.00	184 ± 41.70	0.41
Triglycerides, mg/dL	89 ± 62.80	75 ± 38.90	0.28
Glucose, mg/dL	96 ± 18.75	93 ± 10.92	0.38
Insulin, μU/mL	13 ± 10.88	16 ± 12.42	0.27
HOMA-IR	3 ± 2.80	4 ± 3.26	0.31

Values expressed are mean ± SD.

Plasma cGMP levels were determined in a subgroup of AA women (G allele carriers (n = 12) and noncarriers (n = 31) recruited at Vanderbilt University. cGMP levels were 30% lower, P = .01, in G allele carriers as compared with noncarriers (Figure 1A), suggesting reduced activity of the NO-cGMP pathway.

Figure 1.

Protocol 1.

Evidence of endothelial dysfunction in individuals carrying the G allele of CD36 rs3211938. A, Differences in cGMP between noncarriers (T/T) and carriers of the minor G allele (G/T) for the coding variant rs3211938. B, Changes in brachial artery diameter at baseline and postcuff deflation after 5-minute ischemia. Brachial artery diameters were measured 20 seconds after cuff release to allow for measurements of reactive hyperemia. C, Brachial artery FMD in T/T controls and G/T carriers of the G allele after adjusting for weight differences. D, Time to peak in seconds after cuff deflation in T/T and G/T subjects showing the delayed time to peak in G allele carriers. Data presented are mean ± SEM.

Measurements of vascular reactivity were conducted in a group of 18 obese AA women: G allele carriers (n = 8), and noncarriers (n = 10), who were matched by age (44 ± 11.3 vs 40 ± 11.2 y, carriers and noncarriers, respectively; P = .435), BMI (38 ± 7.5 vs 39 ± 4.1 kg/m²; P = .994), SBP (132 ± 16 vs 128 ± 11 mm Hg; P = .538), diastolic BP (DBP) (82 ± 14 vs 82 ± 10 mm Hg; P = .924), HR (73 ± 12 vs 66 ± 11 bpm in noncarriers; P = .181), and estradiol levels (245 ± 135 vs 239 ± 139 pg/mL; P = .920). Mean changes in brachial artery diameter at baseline and postcuff release are presented in Figure 1B. Endothelial function was impaired in G allele carriers as compared with noncarriers (5.8 ± 3.4 vs 8.5 ± 3.8%; P = .035) (Figure 1C) even after adjusting for differences in weight. Vascular compliance as measured by the time to peak after ischemic occlusion was prolonged in G allele carriers compared with noncarriers (41 ± 7.6 vs 34 ± 6.5 s; P = .047) (Figure 1D). We also determined differences in oxidative stress as measured by F2-isoprostanes. Despite the presence of endothelial dysfunction, there was a tendency for G allele carriers to have reduced F2-isoprostanes compared with noncarriers (0.03 ± 0.007 vs 0.05 ± 0.023 ng/mL; P = .06).

Protocol 2: Response of endothelial function to sildenafil therapy is impacted by rs3211938

The CONSORT flow diagram is shown in Supplemental Figure 2. All subjects had a MetS diagnosis or hyperinsulinemia. Baseline characteristics were similar in the 2 study groups (Table 2). FMD measurements were obtained in 30 subjects, the vascular assessment in 10 subjects could not be analyzed because the subject moved, or due to anatomical variations (2 brachial arteries) or loss of brachial artery window after cuff release. Of the 40 subjects who completed the clinical trial, 34 genotypes were validated and included in the analysis of the effect of the CD36 variant. Of these, 26 were noncarriers (T/T) and 8 were carriers of the minor allele G, similar to the frequency of the G/T genotype in the initial cohort. About 23% of participants in protocol 1 were enrolled in protocol 2.

Table 2.

Protocol 2. Baseline Characteristics of Participants Recruited at Vanderbilt University Who Participated in The Clinical Trial

Parameters q21	Sildenafil Citrate	Placebo	P Value
Age, years	42 ± 8.6	43 ± 10.4	.89
Height, cm	164 ± 5.4	162 ± 4.8	.53
Weight, kg	105 ± 17.9	102 ± 16.4	.63
BMI, kg/m2	39 ± 5.7	39 ± 5.7	.80
Waist circumference, cm	112 ± 10.7	108 ± 10.4	.26
Cholesterol, mg/dL	180 ± 34.8	176 ± 43.5	.48
LDL, mg/dL	118 ± 35.7	110 ± 38.0	.51
HDL, mg/dL	43 ± 9.3	45 ± 11.5	.48
Triglycerides, mg/dL	97 ± 55.3	81 ± 49.3	.33
Glucose, mg/dL	89 ± 10.1	97 ± 11.7	.03a
Insulin, μU/mL	14 ± 7.7	14 ± 10.0	.96
SBP, mm Hg	132 ± 13.0	130.0 ± 12.9	.61
DBP, mm Hg	84 ± 8.0	81 ± 7.3	.15
HR, bpm	74 ± 9.3	78 ± 12.3	.27
Fat mass, kg	52 ± 15.4	50 ± 18.3	.73
Fat free mass, kg	54 ± 6.8	54 ± 6.0	.81
Percentage body fat,%	48 ± 3.3	47 ± 5.2	.50

Values expressed are mean ± SD.

^a

P < 0.05.

Sildenafil treatment did not improve metabolic parameters (Table 3). We did not observe significant difference in IS: means difference 0.12 (95% confidence interval [CI], −0.33 to 0.58; P = .550); AIRG: mean difference −0.04 (95% CI, −0.26 to 0.18; P = .705); or DI: mean difference 0.08 (95% CI, −0.45 to 0.62; P = .754) between interventions (sildenafil vs placebo) after adjusting for baseline values. Adjusting for age, fat mass, fat-free mass in addition to baseline values did not change the outcomes: IS (P = .572), AIRG (P = .727), and DI (P = .715). No differences were observed in IS or β-cell function measured by HOMA-IR (data not show). However, treatment with sildenafil significantly decreased HbA_1c levels (5.64 ± 0.40 baseline to 5.49 ± 0.39% posttreatment; P = .026), vs no effect with placebo (5.67 ± 0.33 baseline to 5.66 ± 0.35% posttreatment). There was no effect of CD36 rs3211938 on IS between the 2 intervention groups (P = .851).

Table 3.

Protocol 2. Effect of Sildenafil Citrate on Metabolic Parameters

Parameters	Sildenafil Citrate		Placebo
	Baseline	Intervention	Baseline	Intervention
IS, min/pmol · mL per 10−5	4.0 ± 3.5	3.4 ± 3.0	2.9 ± 2.1	3.2 ± 2.4
AIRG, μU/mL per time	778 ± 786.0	822 ± 938.0	716 ± 870.0	700 ± 836.0
DI	1737 ± 1332.0	1549 ± 1149.0	1387 ± 1652.0	1347 ± 953.0
HbA1c	5.6 ± 0.4	5.41 ± 0.3a	5.67 ± 0.3	5.69 ± 0.4
HOMA-IR	1.6 ± 1.4	2.1 ± 2.1	2.2 ± 1.6	2.4 ± 1.7

Data presented as mean ± SD.

^a

P < .05 vs baseline.

Treatment with sildenafil citrate did not affect endothelial function measured as the FMD %; means difference 0.46 (95% CI, −1.58 to 2.49; P = .649) adjusted for baseline values (Table 4). However, there was a significant interaction between sildenafil treatment and the rs3211938 G/T genotype (P = .018) (Figure 2). Sildenafil decreased FMD in T/T carriers (means difference −2.6, the 95% CI: −5.1 to −0.1; P = .040) and tended to improve it in G/T carriers (means difference 2.9, the 95% CI: −0.9 to 6.8; P = .126). Reactive hyperemia index significantly increased when adjusted for both baseline value and genotype (means difference 0.45, 95% CI: −0.01 to 0.91; P = .05). SBP or DBP did not differ between the sildenafil and placebo groups. Compliance to treatment assessed from plasma sildenafil concentration at 4 weeks showed good compliance as all subjects randomly assigned sildenafil had detectable levels in plasma (Supplemental Figure 3). Side effects associated with the treatment are listed in Supplemental Table 1.

Figure 2.

Protocol 2.

The rs3211938 genotype influences FMD response to sildenafil citrate treatment. Changes in endothelial function measured by FMD after a 4-week treatment with sildenafil citrate or placebo in a cohort of obese AA women with the metabolic syndrome. There was a significant interaction between treatment with sildenafil citrate and CD36 SNP rs3211938 (*, P = .018). Sildenafil treatment improved endothelial function in G allele carriers who have partial (∼50%) CD36 deficiency (change in FMD = 3 ± 1.14%) but not in noncarrier (change in FMD = 0.18 ± 1.25%). Data presented are mean ± SEM.

Table 4.

Protocol 2. Effect of Sildenafil Citrate on Vascular Parameters

Parameters	Sildenafil Citrate		Placebo
	Baseline	Intervention	Baseline	Intervention
SBP, mm Hg	115 ± 17.0	109 ± 13.1	116 ± 14.8	111 ± 9.3
DBP, mm Hg	70 ± 12.6	68 ± 11.0	72 ± 9.9	70 ± 8.0
HR, bpm	73 ± 6.9	71 ± 6.2	68 ± 10.4	70 ± 11.0
FMD, %	4.7 ± 0.4	5.3 ± 2.8	5.3 ± 3.1	6.0 ± 3.2
Baseline diameter, mm	4.1 ± 0.5	4.1 ± 0.57	4.0 ± 0.5	3.9 ± 0.6
Peak diameter, mm	4.3 ± 0.5	4.3 ± 0.61	4.2 ± 0.5	4.1 ± 0.6
Peripheral artery tonometry	2.2 ± 0.5	2.6 ± 0.77	2.0 ± 0.7	2.1 ± 0.6

Data presented are mean ± SD.

Discussion

Endothelial dysfunction characterized by reduced NO bioavailability is an early event in vascular disease and a predictor of cardiovascular complications and T2DM. NO produced by the endothelium is a potent vasodilator in addition to mediating numerous cellular functions such as inhibiting smooth muscle cell proliferation and migration, platelet aggregation, and adhesion of leukocytes to the endothelium (22).

Our findings suggest that CD36 physiologically impacts endothelial function at least in part through regulation of the NO-sGC pathway. The reduction in cGMP levels we observed in partially CD36-deficient AA women is opposite to what would be predicted based on the in vitro inhibition of the NO-sGC pathway by TSP1-CD36 engagement. However, TSP1 can potently inhibit vascular NO signaling via its other receptor CD47, and CD36 deletion might enhance TSP1 availability for binding CD47 (22, 23, 34). Possibly CD36 binding TSP1 buffers the more potent inhibitory effect on TSP1 signaling via CD47.

The NO-induced cGMP production increases smooth muscle relaxation and consequently blood flow and would be important for regulation of tissue perfusion (21). Impaired regulation of vascular smooth muscle relaxation and tissue perfusion consequent to reduced cGMP generation could theoretically contribute to the high circulating FA levels observed in partially CD36-deficient individuals.

Our cohort in the sildenafil trial consisted exclusively of AA women who met criteria for MetS or had hyperinsulinemia. AAs have higher prevalence of cardiovascular complications as compared with Caucasians, in part reflecting higher susceptibility to endothelial dysfunction (35). AAs without MetS have reduced NO bioavailability, lower microvascular vasodilation, and increased arterial stiffness when compared with whites. In our study, AA women without MetS matched for BP and BMI had reduced vascular reactivity if they carried the G allele of rs3211938. Thus, this genotype that is exclusive and relatively common (20%–28% frequency) in AA might contribute to the ethnic differences related to susceptibility to endothelial dysfunction. Additional high-frequency CD36 genotypes that alter CD36 level or regulation (32, 36) could also impact endothelial function in this population.

Our subgroup analyzes showed that the CD36 coding SNP rs3211938 also influences the response to sildenafil treatment on endothelial function, which tended to improve in G allele carriers who have approximately 50% lower CD36 levels, whereas it worsened in noncarriers. However, contrary to our hypothesis, a 4-week treatment with sildenafil did not improve IS or endothelial function in AA women with MetS despite a modest reduction in HbA_1c.

Two previous studies showed a beneficial effect of the PDE-5 inhibitor, tadalafil, on glucose DI after 3 weeks (37) and 3 months of treatment (38) in women with MetS and morbidly obese patients, respectively. Recently, we showed in a primarily Caucasian cohort that 3-month treatment with sildenafil improved IS in subjects with prediabetes (25). The present cohort consisted exclusively of obese AA women with MetS or hyperinsulinemia, a group with the highest risk for hypertension and T2DM. The lack of IS response to sildenafil in our cohort can have several explanations. First, sildenafil potentiates cGMP action by preventing its degradation and not by enhancing its production. AAs have reduced NO production, and therefore, blocking degradation of cGMP may not suffice to restore endothelial function. An alternative approach to overcome this limitation is to use an NO substrate such as L-arginine in combination with PDE-5 inhibition. L-arginine was previously reported to ameliorate microvascular coronary endothelial dysfunction in AA (39). Second, a more prolonged treatment might be needed to reverse the vascular and metabolic abnormalities. However, arguing against this is that no trend towards improvement was observed in any of the cardiovascular and metabolic parameters after 4-week treatment, whereas previously, DI was shown to improve after 3-week treatment with tadalafil in patients with MetS (37).

The interaction we observed between the rs3211938 genotype and the efficacy of sildenafil treatment highlights clinical importance of considering genotype in therapeutic interventions. Sildenafil treatment in subjects with the MetS who had normal CD36 level (T/T genotype) showed worsened FMD, whereas those carrying the G allele (G/T) where the CD36 level is reduced had FMD improvement. Possibly the lower CD36 level might have a beneficial effect downstream of cGMP in down-regulating some of the negative vascular effects of inflammation in individuals with metabolic syndrome or hyperinsulinemia. For example, platelet activation by oxidized low-density lipoproteins correlates with CD36 expression (36) and vessel walls from CD36 null mice have less reactive oxygen species coincident with up-regulated expression of antioxidant enzymes in vascular smooth muscle cells (40). Consistent with this, our subjects with the G/T genotype as compared with the T/T genotype tended to have lower levels of F2-isoprostanes, a marker of oxidative stress suggesting lower reactive oxygen species production.

A limitation of our clinical trial is that our subjects had components of the MetS so the abnormality of FMD introduced by rs3211938 was likely to be blunted by the effect of the syndrome itself so future studies should examine the effect of the SNP in healthy individuals without MetS. The negative effects of sildenafil in obese individuals who are metabolically compromised warrants further investigation.

Acknowledgments

We thank Mrs Davalynn Johnson for her work as a community developer who was instrumental in the recruitment and retention of African American women in these studies.

This work was supported by National Institute of Health (NIH) Grants P01 HL056693, R01 HL102387, DK033301, and DK060022; the NORC Grant P30 DK056341; the DRTC Grant DK20593; the Pfizer Investigator-Initiated Research Grant WS572296; the Vanderbilt Clinical and Translational Science Award Grant UL1 RR024975 from the National Center for Research Resources and the NIH. C.A.S. was supported by the NIH Grant K23 HL103976, Faculty Development Award in Clinical Pharmacology from the PhRMA Foundation, and a Clinical Scientist Career Development Award from the Doris Duke Foundation.

Disclosure Summary: The authors have nothing to disclose.

Abbreviations

 
• AA

  African American

 
• AIRG

  acute insulin response to glucose

 
• BMI

  body mass index

 
• BP

  blood pressure

 
• CI

  confidence interval

 
• DBP

  diastolic BP

 
• DI

  disposition index

 
• FA

  fatty acid

 
• FMD

  flow-mediated dilation

 
• HbA_1c

  hemoglobin 1AC

 
• HOMA-IR

  homeostasis model assessment of insulin resistance

 
• HR

  heart rate

 
• IS

  insulin sensitivity

 
• MetS

  metabolic syndrome

 
• NO

  nitric oxide

 
• PDE-5

  phosphodiesterase 5

 
• SBP

  systolic BP

 
• sGC

  soluble guanylate cyclase

 
• SNP

  single nucleotide polymorphism

 
• T2DM

  type 2 diabetes mellitus

 
• TSP1

  thrombospondin 1.