Treatment with Sildenafil Improves Insulin Sensitivity in Prediabetes: A Randomized, Controlled Trial

Sildenafil increases insulin sensitivity in mice. In humans, phosphodiesterase 5 inhibition improves disposition index, but the mechanism of this effect has not been elucidated and may depend on duration. In addition, increasing cyclic GMP without increasing nitric oxide could have beneficial effects on fibrinolytic balance.

Objective:

The objective was to test the hypothesis that chronic phosphodiesterase 5 inhibition with sildenafil improves insulin sensitivity and secretion without diminishing fibrinolytic function.

Design:

This was a randomized, double-blind, placebo-controlled study.

Setting:

This trial was conducted at Vanderbilt Clinical Research Center.

Participants:

Participants included overweight individuals with prediabetes.

Interventions:

Subjects were randomized to treatment with sildenafil 25 mg three times a day or matching placebo for 3 months. Subjects underwent a hyperglycemic clamp prior to and at the end of treatment.

Main Outcome Measures:

The primary outcomes of the study were insulin sensitivity and glucose-stimulated insulin secretion.

Result:

Twenty-one subjects completed each treatment arm. After 3 months, the insulin sensitivity index was significantly greater in the sildenafil group compared to the placebo group by 1.84 mg/kg/min per μU/mL*100 (95% confidence interval, 0.01 to 3.67 mg/kg/min per μU/mL*100; P = .049), after adjusting for baseline insulin sensitivity index and body mass index. In contrast, there was no effect of 3-month treatment with sildenafil on acute- or late-phase glucose-stimulated insulin secretion (P > .30). Sildenafil decreased plasminogen activator inhibitor-1 (P = .01), without altering tissue-plasminogen activator. In contrast to placebo, sildenafil also decreased the urine albumin-to-creatinine ratio from 12.67 ± 14.67 to 6.84 ± 4.86 μg/mg Cr. This effect persisted 3 months after sildenafil discontinuation.

Conclusions:

Three-month phosphodiesterase 5 inhibition enhances insulin sensitivity and improves markers of endothelial function.

Type 2 diabetes affects an estimated 366 million adults worldwide (1). Insulin resistance precedes the development of type 2 diabetes, and diabetes ensues when insulin secretory capacity fails to compensate for increased insulin requirements. Weight loss and increased activity reduce progression from prediabetes to diabetes but are difficult to maintain (2). Metformin and thiazolidinediones also reduce incident diabetes, but the use of thiazolidinediones has been limited by concerns regarding adverse effects (2,–5).

Increasing cyclic GMP (cGMP) signaling also has the potential to decrease incident diabetes in high-risk patients. cGMP increases insulin sensitivity in muscle by promoting the translocation of glucose transporter 4 to the cell membrane (6). Acute increases in cGMP enhance glucose-stimulated insulin secretion (GSIS) in islets, although prolonged increases in cGMP may decrease insulin secretion (7, 8).

Pharmacological strategies to increase cGMP include increasing nitric oxide (NO), activating soluble guanylate cyclase directly, or decreasing the degradation of cGMP by giving a phosphodiesterase 5 (PDE5) inhibitor. Increasing NO can have detrimental cGMP-independent effects, however, such as decreasing fibrinolysis by nitrosylating proteins involved in exocytosis of tissue-plasminogen activator (t-PA) (9,–11). Thus, increasing cGMP by preventing its degradation by PDE5 may be a better strategy to improve glucose homeostasis while avoiding detrimental effects on fibrinolysis.

Rodent models provide evidence for a favorable effect of PDE5 inhibition on insulin sensitivity. In high fat-fed mice, 12-week treatment with sildenafil and L-arginine improves insulin sensitivity and muscle glucose uptake (12). Similarly, 8-week treatment with sildenafil improves endothelial function and hyperglycemia after an oral glucose load in fructose-fed rats (13).

In a proof-of-concept study, our group reported that 3-week administration of tadalafil significantly improved β-cell function in individuals with the metabolic syndrome who underwent frequently sampled iv glucose tolerance tests (IVGTTs) and reduced the disposition index (DI; a composite measure of insulin sensitivity and secretion) in the women studied (14). Recently, Ho et al (15) reported that 12-week treatment with tadalafil improved DI and oral DI and tended to improve the insulinogenic index in insulin-resistant subjects undergoing oral glucose tolerance tests (OGTTs). In addition, tadalafil improved insulin sensitivity among severely obese participants (15). These intriguing results suggest that, with longer duration therapy, PDE5 inhibition may have beneficial effects on insulin sensitivity as well as insulin secretion.

The present study tested the hypothesis that 3-month treatment with sildenafil increases insulin secretion and improves tissue insulin sensitivity in patients with prediabetes. We utilized hyperglycemic clamps to measure GSIS and estimate insulin sensitivity. Hyperglycemic clamps were chosen over IVGTT because this technique allows assessment not only of first-phase GSIS, but also late-phase GSIS. We further examined the effect of 3-month sildenafil administration on fibrinolysis and other markers of endothelial function.

Subjects and Methods

Subjects

Overweight volunteers (body mass index [BMI] ≥ 25 kg/m 2 ) with prediabetes, defined by impaired fasting glucose, impaired glucose tolerance, or glycated hemoglobin (HbA1C) criteria, were recruited between September 2011 and August 2013. The study protocol was approved by the Vanderbilt Institutional Review Board in accordance with the Declaration of Helsinki. All volunteers provided written informed consent and underwent history and physical examination, morphometric measurements, electrocardiogram, and laboratory assessment. Impaired fasting glucose was defined as fasting plasma glucose of 100–125 mg/dL. Impaired glucose tolerance was defined as plasma glucose of 140–199 mg/dL 2 hours after a 75-g OGTT or HbA1C of 5.7–6.4%. Volunteers with diabetes mellitus, anemia, impaired hepatic function, renal insufficiency (defined as an estimated glomerular filtration rate < 60 mL/min/1.73 m 2 using the Modification of Diet in Renal Disease equation), cardiovascular disease (other than hypertension and/or left ventricular hypertrophy), and pulmonary or neurological disorders were excluded. Volunteers taking PDE5 inhibitors, nitrates, potent CYP3A4 inhibitors, anticoagulants, metformin, lithium, or chronic glucocorticoid treatment were excluded. Volunteers who had participated in a weight reduction program or had lost more than 2 kilograms in the 6 months preceding the study were excluded. Women of childbearing potential were required to use a contraceptive method during the study. Pregnancy was excluded by urine β-human chorionic gonadotropin at each glucose tolerance test and on study days.

Study protocol

Volunteers participated in a randomized, double-blind, placebo-controlled trial (Supplemental Figure 1) ( <"type":"clinical-trial","attrs":<"text":"NCT01409993","term_id":"NCT01409993">> NCT01409993) from September 2011 to August 2013. Volunteer screening was done in a stepwise fashion. During the first screening visit, volunteers were screened for impaired fasting glucose and eligibility. Volunteers were invited to return for a second screening visit, during which they underwent measurement of body composition and resting energy expenditure, measurement of HbA1C, and a 75-g OGTT.

On each clamp study day, participants reported to the Vanderbilt General Clinical Research Center (CRC) in the morning after an overnight fast. Subjects collected their urine for 8 hours before each visit for measurement of sodium and creatinine. Upon arrival at the CRC, an additional urine sample was collected for measurement of urine albumin and creatinine. Volunteers were studied in the supine position. Hyperglycemic clamps were completed using the protocol outlined below. Blood samples were collected at baseline for measurement of plasminogen activator inhibitor-1 (PAI-1), t-PA, cGMP, NO metabolites, F2-isoprostanes, cytokines, and aldosterone. Blood pressure and heart rate were measured in triplicate at baseline and every 15 minutes throughout the study using an automated brachial sphygmomanometer (Dinamap; Critikon).

After completion of the baseline hyperglycemic clamp, participants were randomized in a 1:1 ratio to receive sildenafil 25 mg three times a day (for the first 18 subjects, Revatio; Pfizer; for the reminder after sildenafil went off patent, Greenstone LLC) or matching placebo using a permuted-block randomization algorithm. Randomization was stratified by gender. The Vanderbilt Investigational Drug Service was responsible for the storage, preparation, and labeling of the study drug. The first dose of study medication was given immediately after the first hyperglycemic clamp. Subjects were provided a 90-day supply of study drug, a medication diary to record intake of new concurrent medication or over-the-counter medication, and a safety bracelet indicating that they could be taking a PDE5 inhibitor. Subjects were contacted by phone 1, 3, 7, 30, and 60 days after initiation of study medication to monitor for adverse events. Participants underwent a second hyperglycemic clamp after 3 months of treatment with study drug. The study drug was then discontinued, and 3 months later subjects returned for a follow-up history/physical examination and OGTT.

OGTT procedures

Subjects reported to the Vanderbilt CRC in the morning after an overnight fast. An antecubital venous catheter was placed. Baseline blood samples were collected for measurements of HbA1C, TSH, free fatty acids, glucose, and insulin. A 75-g glucose solution (Trutol Glucose Tolerance Test Beverage; Thermo Fisher Scientific Inc) was then given to the volunteers by mouth. Blood samples for glucose and insulin were collected every 30 minutes for 120 minutes.

Hyperglycemic clamp procedures

Participants reported to the Vanderbilt CRC after an overnight fast. Two iv catheters were placed while volunteers rested in the supine position. Baseline samples were collected 10 minutes apart for measurement of glucose and insulin. A priming dose of 9.622 mg/m 2 of 20% dextrose monohydrate (Hospira) was then infused according to previously published protocols (16, 17). The priming dose of dextrose was administered for 14 minutes to raise the plasma glucose level to 200 mg/dL to suppress gluconeogenesis. Thereafter, the dextrose infusion rate was adjusted every 5 minutes based on the negative feedback principle (Supplemental Figure 2) (16). At 120 minutes, 5 g of L-arginine was concurrently infused as previously described (18). Blood samples for glucose, insulin, and C-peptide were collected 2.5 to 5 minutes apart. Plasma glucose was measured by glucose oxidase method with Biochemistry Analyzer (YSI 2700 Select; YSI Inc).

Calculations

First-phase GSIS was calculated as the difference between the maximum plasma insulin level in the first 10 minutes and the average of the two baseline insulin measurements. Late-phase GSIS was calculated as the difference between the maximum plasma insulin level from 90 to 120 minutes and average baseline insulin. Plasma insulin response to fixed hyperglycemia (I, μU/mL), which indicates the β-cell response to glucose, was calculated as the average concentration of insulin from 90 to 120 minutes. Metabolized glucose (M), which serves as an indicator of glucose tolerance, was determined by the average glucose infusion rate (GIR; mg/kg*min) from 90 to 120 minutes. The insulin sensitivity index (ISI) was calculated as GIR/I (mg/kg/min per μU/mL*100). ISI results were further adjusted to account for urinary glucose loss as previously described (16). DI was calculated as the acute GSIS multiplied by the ISI (DI, mg/kg/min). Maximum insulin release was calculated as the change between peak insulin level after stimulation with L-arginine and the average baseline insulin.

Laboratory analysis

Blood samples were collected on ice and centrifuged immediately at 5°C for 20 minutes. Serum or plasma was separated and stored at −80°C until the time of assay. Plasma insulin concentrations were determined by RIA (Millipore). Samples for C-peptide were drawn into heparinized tubes containing 250 KIU aprotinin (Trasylol; Roche Diagnostics Corporation) per milliliter of whole blood, which resulted in a final concentration of 500 KIU aprotinin/mL of serum or plasma. C-peptide was measured using RIA (Millipore). Blood for measurement of PAI-1 and t-PA was collected in Vacutainer tubes containing acidified 0.105 mol/L of sodium citrate (Becton Dickinson) (19, 20). PAI-1 antigen and t-PA antigen levels were determined using a double antibody principle (Tcoag). Aldosterone was determined using a RIA with 125 I-aldosterone (MP Biomedicals). A commercially available RIA kit was used to measure plasma levels of cGMP (Amersham Pharmacia Biotech AB). F2-isoprostanes were measured in plasma using a negative ion gas chromatography mass spectroscopy (21). Cytokines were measured in plasma using cytometric bead array (BD, Biosciences). The urine albumin-to-creatinine ratio (UACR) was determined as the ratio between microalbumin and urine creatinine in a fresh urine collection. Microalbumin was determined using a turbidimetric immunoassay with endpoint determination, and urine creatinine was determined using a kinetic alkaline picrate assay. Both urine assays were performed by the Vanderbilt Pathology Laboratory Services.

Statistical analysis

The primary endpoints were those calculated from the hyperglycemic clamp. Secondary endpoints included: 3-month fasting glucose and insulin, systolic and diastolic blood pressure, PAI-1, t-PA, urine albumin and creatinine, body weight, resting energy expenditure, plasma cGMP, aldosterone, inflammatory cytokines, F2-isoprostanes, and the incidence of side effects and of serious side effects.

Descriptive statistics of baseline characteristics and measurements are presented as median with interquartile range and as mean and standard deviation. Comparisons between treatment groups were made using Wilcoxon rank-sum test or Student’s t test for continuous variables and χ 2 test for categorical variables. Multivariable linear regression models were fitted to investigate the effect of sildenafil on the endpoints after 3-month treatment. Baseline values and BMI were included as covariates in the models for ISI, DI, and GSIS. Baseline PAI-1, BMI, race, and the interaction between race and treatment were included in the model for PAI-1. Baseline UACR, BMI, use of angiotensin-converting enzyme (ACE)/angiotensin receptor blocker (ARB) as well as the interaction between ACE/ARB and treatment were included in the model for UACR. A smooth relationship between BMI and the endpoints was considered in all the models by using restricted cubic splines with three knots. To examine the effect of sildenafil on microalbuminuria, a logistic regression model was fitted for post-treatment microalbuminuria on treatment with adjustment for baseline microalbuminuria status with exact conditional inference. A two-sided P value < .05 was considered statistically significant. All the analyses were performed using R 3.1.2.

The calculated sample size for this study was 61 subjects per study arm. This sample size was based on a 0.8 power to detect a 40% increase in acute GSIS. In March 2014, the Data Safety Monitoring Board evaluated the progress of the study and recommended stopping, based on a futility analysis that suggested a lack of effect of sildenafil on GSIS. The futility analysis was based on conditional power calculated through simulations using the data collected at the time of the interim analysis.

Results

Subject characteristics

Fifty-one subjects were randomized to the study drug, and 42 subjects completed the study per protocol ( Figure 1 ). In three subjects who were discontinued or who dropped out before the second clamp, fasting blood was collected for PAI-1 and t-PA at discontinuation, and these subjects are included in the analysis of PAI-1 and t-PA.

An external file that holds a picture, illustration, etc.
Object name is zeg9991522070001.jpg

Study flow diagram.

Baseline subject characteristics were similar in the two study groups ( Table 1 ). After 3 months of treatment, the plasma sildenafil concentration was 93.0 ± 63.5 ng/mL in the sildenafil group and 2.6 ± 9.1 ng/mL in the placebo group (P < .001).

Table 1.

Baseline Subject Characteristics

Characteristic Placebo Sildenafil P Values
n 26 25
Age, y 50.6 ± 9.4 (52.8, 44.6–58.5) 51.6 ± 9.3 (54.1, 45.7–60.4) .61
Gender, female, n (%) 18 (69) 17 (68) .93
Race, n
    Black American 4 7 .27
    White American 22 18
Ethnicity, n
    Hispanic 0 1 .31
BMI, kg/m 2 36.7 ± 7.0 (34.9, 31.1–41.2) 35.6 ± 6.3 (36.5, 31.2–38.0) .60
Metabolic syndrome, n 23 22 .96
Screening systolic blood pressure, mm Hg 133.9 ± 14.4 (131.5, 123.0–143.3) 129.4 ± 14.7 (128.0, 121.0–142.0) .41
Screening diastolic blood pressure, mm Hg 78.8 ± 10.4 (78.0, 69.0–85.8) 79.1 ± 9.8 (80.0, 74.0–86.0) .77
HDL, mg/dL
    Males 37.0 ± 8.8 (34.5, 30.3–43.5) 40.6 ± 7.7 (40.5, 33.5–46.5) .31
    Females 43.8 ± 8.4 (42.0, 38.0–49.0) 42.2 ± 5.3 (43.0, 39.0–46.0) .88
Triglycerides, mg/dL 117.2 ± 84.7 (94.0, 74.0–138.0) 115.4 ± 71.7 (94.0, 66.0–147.0) .99
Waist circumference, cm
    Males 112.9 ± 7.5 (110.8, 107.8- 115.8) 117.6 ± 12.3 (114.0, 110.6–127.8) .48
    Females 113.3 ± 11.3 (113.0, 104.6–117.8) 111.9 ± 14.4 (111.0, 102.0–116.0) .51
Fasting glucose, mg/dL 102.6 ± 9.4 (103.0, 97.0–108.5) 102.7 ± 12.5 (103.0, 95.0–08.0) .82
Impaired glucose tolerance, n 14 15 .66
Albumin-to-creatinine ratio 8.3 ± 12.6 (4.0, 3.0–6.0) 11.2 ± 13.8 (5.0, 3.0–14.0) .49
Concomitant medications, n
    ACE inhibitors/ARBs 11 10 .97
    Thiazides 10 5 .17
    Statins 5 9 .15
    Hormonal therapy 5 3 .52
First-phase GSIS, μU/mL 44.6 ± 26.4 (38.3, 26.1–53.9) 49.1 ± 36.6 (41.6, 20.5–59.4) .96
Late-phase GSIS, μU/mL 107.7 ± 69.1 (76.0, 52.3–180.0) 99.9 ± 71.2 (87.1, 49.9–106.4) .66
ISI adjusted, mg glucose/kg/min per μU/mL*100 6.6 ± 4.5 (5.5, 3.3–7.3) 5.5 ± 2.9 (5.6, 3.5–7.2) .85
DI, mg/kg/min 292.7 ± 266.3 (221.3, 150.6–318.6) 260.8 ± 209.1 (160.3, 118.0–387.6) .55

Abbreviation: HDL, high-density lipoprotein. Values represent mean ± SD. Numbers in parentheses represent the median and interquartile range (median, lower quartile-upper quartile) for continuous variables. P values are based on Wilcoxon rank-sum test for continuous variables and χ 2 test for categorical variables.

Effect of PDE5 inhibition on glucose homeostasis

ISI, DI, and GSIS were similar in the two groups at baseline ( Table 1 ). After a 3-month treatment, ISI was significantly greater in the sildenafil group compared to the placebo group by 1.84 mg/kg/min per μU/mL*100 (95% confidence interval [CI], 0.01 to 3.67 mg/kg/min per μU/mL*100; P = .049) ( Figure 2 ) after adjusting for baseline ISI and BMI. DI trended higher in the sildenafil group compared to the placebo group by 83.5 mg/kg/min (95% CI, −7.01 to 174.02 mg/kg/min; P = .070). In contrast, there was no significant effect of treatment on first-phase or late-phase GSIS, insulin concentrations after L-arginine infusion, or C-peptide concentrations (all P values > .30). There was no significant effect of sildenafil on weight, resting energy expenditure, or free fat mass (data not shown).

An external file that holds a picture, illustration, etc.
Object name is zeg9991522070002.jpg

Effect of 3-month treatment with sildenafil 25 mg three times daily (red bars) or matching placebo (gray bars) on GSIS, DI, and ISI. Data are presented as estimated marginal means after controlling for baseline values and in the case of DI and ISI, BMI. *, P < .05 vs placebo; †, P = .07 vs placebo.

Effect of PDE5 inhibition on microalbuminuria and other biomarkers

Two subjects in the placebo group and four subjects in the sildenafil group had microalbuminuria at randomization, defined as an albumin-to-creatinine ratio ≥ 17 μg/mg Cr in men or ≥ 25 μg/mg Cr in women (22). Sildenafil reduced the UACR from 12.67 ± 14.67 to 6.84 ± 4.86 μg/mg Cr ( Figure 3 ). This effect persisted 3 months after discontinuation of sildenafil. In contrast, UACR increased from 8.45 ± 14.17 to 13.41 ± 17.71 μg/mg Cr after 3 months of placebo and was 18.35 ± 54.81 μg/mg Cr 3 months after discontinuation of placebo. Thus, four subjects in the placebo group had microalbuminuria, whereas none in the sildenafil group had microalbuminuria after 3 months of treatment (P = .036 for effect of treatment) ( Figure 3 ). Sildenafil treatment significantly reduced plasma PAI-1 concentrations (P = .01) without affecting t-PA ( Figure 4 ). There was no effect of placebo on PAI-1 or t-PA. Neither treatment affected circulating cGMP, cytokine concentrations, aldosterone, or urine F2-isoprostanes (data not shown).

An external file that holds a picture, illustration, etc.
Object name is zeg9991522070003.jpg

UACRs in individuals with prediabetes after 3-month treatment with placebo or sildenafil and after washout. UACR was ln-transformed before analysis. *, P = .001 vs placebo after controlling for BMI, ACE inhibitor or ARB use, and thiazide use.

An external file that holds a picture, illustration, etc.
Object name is zeg9991522070004.jpg

Change in PAI-1 and t-PA after 3-month treatment with placebo (gray bars) or sildenafil (red bars). Data are presented as estimated means with 95% CI after controlling for baseline PAI-1 or t-PA, BMI, and race. *, P = .01 vs placebo. There was no difference between placebo and sildenafil after 3-month washout.

Hemodynamic effects of PDE5 inhibition

Sildenafil did not affect systolic/diastolic blood pressure (from 125.4 ± 15.6/70.7 ± 9.04 to 122.0 ± 17.2/71.5 ± 10.1 mm Hg in placebo-treated, and from 127.7 ± 15.0/73.5 ± 8.2 to 123.1 ± 14.6/7 0.9 ± 9.5 mm Hg in sildenafil-treated; P = .76 and .73 for systolic and diastolic blood pressure, respectively).

Safety

There were no significant differences in the rate of discontinuation due to side effects ( Figure 1 ) or in the frequency of specific side effects (Supplemental Table 1) in the two treatment groups.

Discussion

In individuals with prediabetes, 3-month treatment with the PDE5 inhibitor sildenafil improves insulin sensitivity but does not alter GSIS measured rigorously using hyperglycemic clamps. The favorable effect of sildenafil on insulin sensitivity is accompanied by an improvement in fibrinolytic balance and decreased urinary albumin excretion.

Two prior studies have reported a favorable effect of tadalafil on DI after 3-week (14) and 3-month (15) treatments. The finding that sildenafil tended to increase DI in the current study suggests that this is a class effect of PDE5 inhibitors. DI is a composite measure that reflects both insulin secretion and insulin sensitivity. The two prior studies differed somewhat in the underlying contributors to improvement in DI with PDE5 inhibition. After a 3-week treatment with tadalafil, Hill et al (14) observed an improvement in DI in women reflecting an overall improvement in β-cell function measured using the frequently sampled IVGTT. After 12 weeks of treatment with tadalafil, Ho et al (15) observed a significant improvement in insulin sensitivity in the severely obese subjects studied. In vitro and rodent studies support the concept that the relative effects of PDE5 inhibition on insulin secretion vs insulin sensitivity may change with prolonged therapy. For example, acutely increasing cGMP or administering the cGMP analog 8-Br-cGMP increases glucose-stimulated insulin secretion (7, 23, 24) and/or promotes antiapoptotic activity in β-cells and islets in vitro (25), whereas with prolonged exposure increasing cGMP can decrease insulin secretion (8). Conversely, in high fat-fed mice, acute administration of sildenafil and L-arginine had no effect on insulin sensitivity, whereas 12-week treatment improved insulin sensitivity and muscle glucose uptake (12).

The present study does not address the mechanism through which PDE5 inhibition improves tissue insulin sensitivity. In rodent models, sildenafil improves muscle glucose uptake through recruitment of vasculature (26) and has no effect on insulin signaling in the muscle (12). Future studies of the effect of PDE5 inhibition on insulin signaling in muscle biopsies after hyperinsulinemia are needed to address this mechanism. Of note, in the present study the effect of sildenafil on insulin sensitivity was independent of blood pressure, which did not change.

Three-month treatment with sildenafil decreased circulating PAI-1 concentrations. Both glucose (27, 28) and insulin (29) stimulate response elements in the PAI-1 promoter, leading to increased circulating PAI-1 concentrations during insulin resistance. Thus, a decrease in PAI-1 may reflect improved insulin sensitivity. In addition, inhibiting the degradation of cGMP per se might be expected to decrease circulating PAI-1 concentrations because cGMP decreases PAI-1 expression (30). Importantly, t-PA concentrations did not change, highlighting the advantage of increasing cGMP independently of NO, which suppresses endothelial t-PA release (11).

Fibrinolytic balance is one measure of endothelial function. Albuminuria may also represent generalized endothelial dysfunction (31). In addition, albuminuria may predict incident type 2 diabetes in high-risk groups (32,–34). PDE5 inhibition reduces albuminuria and glomerular injury in rodent models. PDE5 is expressed in proximal tubules, collecting ducts, and the glomerulus (35, 36). In the Otsuka Long-Evans Tokushima fatty rat model of type 2 diabetes, sildenafil reduces albuminuria, glomerular hyperfiltration, mesangial matrix expansion, and glomerulosclerosis (37). One prior study has reported that sildenafil reduces UACR and HbA1C in men with diabetes (38). In contrast, in the Effect of Phosphodiesterase-5 Inhibition on Exercise Capacity and Clinical Status in Heart Failure with Preserved Ejection Fraction (RELAX) trial, creatinine increased 0.05 mg/dL over 24 weeks in sildenafil-treated heart failure patients (39). Importantly, in the present study, we excluded patients with baseline renal dysfunction. Studies are needed to assess the effect of long-term PDE5 inhibition on albuminuria in patients who are heterogeneous with respect to renal function.

We have noted some limitations of this study. In brief, additional studies using hyperinsulinemic clamps are needed to assess the effect of PDE5 on muscle glucose and hepatic uptake and insulin signaling. The use of the hyperglycemic clamp allowed for accurate determination of GSIS and tissue insulin sensitivity but precluded the study of a large number of subjects. Larger randomized trials using clinical endpoints are needed to assess whether PDE5 inhibition can prevent the onset of diabetes in high-risk patients. In addition, the effect of PDE5 inhibition on albuminuria remains to be studied in a population with a wide range of renal function.

In conclusion, our study suggests that chronic PDE5 inhibition improves insulin sensitivity, fibrinolytic balance, and albuminuria in subjects with prediabetes. Further studies will be needed to determine whether long-term treatment with drugs that increase cGMP can prevent the onset of diabetes in high-risk patients.

Acknowledgments

The authors thank Loretta Byrne for assistance with subject recruitment, conduct of the research protocols, and data entry. The authors also thank Anthony DeMatteo and Zuofei Wang for laboratory assistance.