Endothelin antagonism reduces hemoglobin A1c in patients with pulmonary hypertension

Abstract

Our lab recently reported that the blockade of endothelin-1 (ET-1) receptors attenuates insulin resistance in obese mice; therefore, we hypothesized that patients taking ET-1 receptor antagonists (ERAs) will have improved glycemic control. University of Mississippi Medical Center (2013–2020) electronic health record (EPIC) data were extracted from patients ≥18 years old with a clinical diagnosis of pulmonary hypertension (Food and Drug Administration indication for ERA use) and at least two clinical visits within 2 years. Patients prescribed ERAs (n = 11) were similar in age (61 ± 14 years vs. 60 ± 14 years), body mass index (BMI) (34 ± 8 kg/m ² vs. 35 ± 11 kg/m²), diabetes prevalence (73% vs. 80%, p = 0.59), and follow-up time (209 ± 74 days vs. 283 ± 180 days) compared with patients not taking ERAs (n = 137). There was a small but similar decrease in BMI at follow-up in the ERA (−1.9 ± 3 kg/m²) and control patients (−1.6 ± 5 kg/m²). At follow-up, hemoglobin A1c (HbA1c) significantly decreased −12% ± 11% of baseline in patients taking ERAs, while this did not occur in the control patients (2% ± 20% increase in HbA1c). In the whole population, baseline HbA1c and ERA prescription predicted the fall in HbA1c, while there was no significant association with demographics, diabetes prevalence, and diabetic treatment. These data suggest a potential role of ET-1 in promoting insulin resistance and warrant further investigation into using these drugs for glycemic control.

Introduction

Insulin resistance and type 2 diabetes mellitus (T2DM) are highly associated with obesity (Smith and Smith 2016). The percentage of people with T2DM in the United States is increasing, and more than 80% of T2DM is attributable to obesity and being overweight. Further, the number of drugs available and currently in development for improving glycemic control has increased significantly in recent years, yet there are still a high percentage of patients with uncontrolled or not well-controlled blood glucose (Fang et al. 2021). Antidiabetic drugs that improve glucose control significantly improve cardiovascular morbidity and all-cause mortality. Hemoglobin A1c (HbA1c), or level of glycated hemoglobin, is a reliable measure of chronic blood glucose and correlates well with chronic diabetic complications (Fang et al. 2021). Indeed, reducing HbA1c by just 1% lowers cardiovascular mortality by 30% (Richter et al. 2018).

Recently, our laboratory demonstrated an antidiabetic effect of endothelin-1 (ET-1) receptor antagonists (ERAs) in obese mice (Rivera-Gonzalez et al. 2021). ET-1 is a vasoactive peptide that is released mostly by vascular endothelial cells. ET-1 is upregulated in adipose tissue of obese mice (Speed et al. 2011), as well as liver and vasculature (unpublished observations by our laboratory). In humans, high circulating ET-1 is associated with obesity and insulin resistance (Irving et al. 2001; Weil et al. 2011), and circulating ET-1 levels decrease following significant weight loss (Jenkins et al. 2019). It is thought that high adiposity creates a state of hypoxia, activating hypoxia-inducible factor (HIF1-a), which stimulates ET-1 production from endothelial cells (Rivera-Gonzalez et al. 2021). ET-1 then acts on two receptor subtypes, ET_A and ET_B, both of which are expressed in all insulin-sensitive cell types, including hepatocytes, adipocytes, and myocytes. ET-1, in turn, has been shown to directly reduce insulin sensitivity and induce hyperglycemia in animals (Juan et al. 1996).

Given the significant effect of atrasentan (ET_A receptor antagonist) and bosentan (dual ET_A/ET_B antagonist) to improve fasting blood glucose, insulin, and glucose and insulin tolerance in high-fat fed mice (Rivera-Gonzalez et al. 2021), we hypothesized that patients treated with ET-1 antagonists would have improved blood glucose levels compared to before treatment. ET-1 antagonists are currently Food and Drug Administration (FDA) approved to treat pulmonary hypertension; therefore, we performed a retrospective analysis of de-identified clinical data from the University of Mississippi Medical Center (UMMC) Research Data Warehouse (RDW) to determine the effect of ET-1 receptor antagonism on HbA1c levels.

Methods

Study patients

The Patient Cohort Explorer is a clinical databank at the UMMC containing >40 million electronic health records from approximately 1 million patients (UMMC 2020). All data are de-identified and are extracted from the institution’s electronic health record system (EPIC). These data are exempt from IRB approval (UMMC). The RDW data from 2013 to 2021 were searched for patients meeting inclusion criteria. Patients aged ≥18 years with a positive pulmonary hypertension diagnosis were included if they were (i) prescribed the selective ET-1 type A (ET_A) receptor antagonist, ambrisentan (5–10 mg), or the nonselective dual antagonist macitentan (5–10 mg); and (ii) if they had HbA1c measurements within 1 year before being prescribed an ERA and a measurement within 1 year after the ERA prescription. The drug bosentan was also queried but did not return any eligible patients. Additionally, control patients were included if (i) they had a clinical diagnosis of pulmonary hypertension and (ii) they had two consecutive HbA1c measurements greater than 2 months apart and within a 2-year span. Also, control patients were excluded if baseline HbA1c was >9%. BP measurements consisted of office sphygmomanometer methods. Diagnoses of pulmonary hypertension and all other comorbidities were based on ICD-10 codes.

Statistical analysis

Baseline characteristics were summarized as means, standard deviations, and 95% confidence intervals for continuous measures. Categorical factors are summarized as counts and percentages. Baseline characteristics were compared using the nonparametric Mann–Whitney test and follow-up values were compared to baseline using paired Wilcoxon’s t test. Fisher’s exact test was used for categorical factors between control and ERA groups. These are reported as relative risk with 95% confidence limits. Multivariate linear regression models were used to evaluate the associations between the change in HbA1c with initial treatment (ERA vs. control). This association was further evaluated with models adjusting for age, gender, race (white vs. black), baseline HbA1c, and incidence of diabetes or antidiabetic medication. All analyses were performed with GraphPad Prism 8 (La Jolla, CA). Probability was based on two-tailed tests of significance, and significance was considered as p < 0.05.

Results

In the RDW database, there were 133 pulmonary hypertensive patients that were prescribed ambrisentan (n = 90), bosentan (n = 47), and macitentan (n = 26). Of these, only 11 had 2 consecutive HbA1c measurements before and after ERA prescription (no bosentan patients met criteria). Additionally, there were 8672 total patients that were diagnosed with pulmonary hypertension and not prescribed an ERA. General clinical characteristics and pulmonary hypertensive parameters of the overall population are included in the supplement. In brief, both control and ERA patients were associated with clinically high right ventricular pressure at the end of systole (54 ± 21 mmHg, normal ~25 mmHg), right ventricular internal diameter during diastole (3.4 ± 1 cm, normal ~1–2 cm), and right atrial pressure (10 ± 2 mmHg, normal ~1–5 mmHg) (Supplementary Table S1). There were 137 control patients that had two consecutive HbA1c measurements within a 1-year window. Patients that were prescribed ERAs (n = 11) were not statistically different in age (61 ± 14 years vs. 60 ± 14 years), body mass index (BMI) (34 ± 8 kg/m ² vs. 35 ± 11 kg/m²), hypertension prevalence (95% vs. 72%), diabetes prevalence (73% vs. 80%), and follow-up time (209 ± 74 days vs. 283 ± 180 days) as compared to control patients (Table 1). Despite the high prevalence of these diseases, there was a relatively low treatment prevalence at baseline for hypertension (47% and 36%, respectively), dyslipidemia (45% and 18%, respectively), and diabetes (64% and 36%, respectively), in control and ERA groups (Table 2).

Table 1.

Baseline characteristics in pulmonary hypertensive patients with and without ET-1 antagonist prescribed.

	Control	ERA	p
N	137	11
Age (year)	60 ± 14	61 ± 14	0.76
%Female	66	73	0.64
%Black	74	64	0.72
BMI (kg/m2)	35 ± 11	34 ± 8	0.76
SBP (mmHg)	132 ± 28	138 ± 16	0.32
DBP (mmHg)	73 ± 15	73 ± 11	1
HR (bpm)	84 ± 17	80 ± 14	0.38
Weight (kg)	98 ± 31	95 ± 30	0.75
Baseline HbA1c	6.4 ± 1.1	6.33 ± 0.6	0.96
ΔChange HbA1c	0.1 ± 1	−0.8 ± 0.8	0.004
%Change HbA1c	3% ± 19%	−12% ±11%	0.002
ΔWeight (kg)	−4 ± 12	−5 ± 8	0.66
ΔBMI (kg/m2)	−1.6 ± 5	−1.9 ± 3	0.82
ΔSBP (mmHg)	−4.5 ± 30	−5.5 ± 21	0.90
Follow-up (days)	283 ± 180	209 ±174	0.20

Table 2.

Baseline prevalence of comorbid conditions and adjunctive therapies in pulmonary hypertensive patients with and without ET-1 antagonist.

	Control	Endothelin receptor antagonist	RR
	(n = 137)	(n = 11)	(95% CI)	p
Comorbidities
Hypertension	94.9	72.0	0.74 (0.42–0.95)	0.03
Diabetes	79.6	72.7	0.97(0.80–1.07)	0.59
Stroke	8.8	0.0	0.92 (0.44–1.22)	0.31
Myocardial Infarction	19.7	0.0	0.91 (0.66–1.04)	0.1
Chronic Kidney Disease	60.6	27.3	0.90 (0.79–0.99)	0.03
Lung disease	30.7	27.3	0.99 (0.90–1.13)	0.81
Heart failure	76.6	45.5	0.88 (0.73–0.99)	0.02
Dyslipidemia	47.4	36.4	0.97 (0.87–1.07)	0.48
Treatments
Endothelin Receptor Antagonist	0	100
Soluble Guanylate Cyclase Inhibitor	0	0
Phosphodiesterase-5 inhibitor	6	27	1.30 (1.04–2.17)	0.04
Antihypertensives	47.4	36.4	0.97 (0.87–1.07)	0.48
Antilipid	45.3	18.2	0.92 (0.83–1.01)	0.08
Glucocorticoid	21.2	9.1	0.95 (0.87–1.10)	0.34
Antidiabetic	64.2	36.4	0.91 (0.80–1.01)	0.07

At follow-up, there was a trend for weight and BMI to decrease in the control group (−4 ± 12 kg, p = 0.10 and −1.6 ± 5 kg/m², p = 0.06, respectively) and in the ERA group (−6 ± 8 kg, p = 0.11 and −2 ± 3 kg/m², p = 0.09, respectively), but this did not reach statistical significance (Supplementary Fig. S1). Despite the tendency for decreased BMI, patients not prescribed an ERA were associated with a 2 ± 20% of baseline increase in HbA1c (Fig. 1A), while HbA1c significantly decreased −12% ± 11% of baseline in patients prescribed an ERA (Table 1 and Fig. 1B).

Fig. 1.

(A and B) Hemoglobin A1c, (C and D) body mass index, and (E and F) body weight in kg in patients with pulmonary hypertension. Left panels are control and right panels represent patients before and after treatment with an ET-1 receptor antagonist. Data on the right-hand side of each panel represent change from baseline visit.

Multivariable linear regression models were created to show independent relationships with the change in HbA1c adjusted for confounders. After adjusting for demographics, diabetes, and baseline HbA1c, ERA prescription significantly predicted the percentage change in HbA1c at follow-up (Table 3). Additionally, after also controlling for follow-up time, other comorbidities, and antidiabetic medication, only baseline HbA1c and ERA treatment were significantly associated with a fall in HbA1c (Model 2, Table 3).

Table 3.

Multivariate linear regression models for the percentage decrease in HbAlc at follow-up.

Model 1				Model 2
Factor	Estimate	Standard error	p	Factor	Estimate	Standard error	p
Baseline HbA1c	−0.02	0.01	0.08	Baseline HbA1c	−0.03	0.01	0.03*
ERA	−0.13	0.05	0.02*	ERA	−0.12	0.05	0.02*
Age	−0.001	0.001	0.27	Follow-up time	0.0001	0.0001	0.24
Female	−0.003	0.03	0.93	Age	−0.001	0.001	0.20
White	−0.004	0.03	0.91	Female	−0.005	0.03	0.86
Diabetes	0.04	0.04	0.35	White	−0.007	0.03	0.83
				Diabetes	0.05	0.04	0.22
				Stroke	−0.06	0.06	0.28
				MI	0.04	0.04	0.28
				CHF	−0.04	0.03	0.24
				Dyslipidemia	0.03	0.03	0.32
				Antilipids	0.02	0.03	0.63
				Antidiabetics	−0.01	0.03	0.69

Discussion

In 2001, the first ERA, bosentan, was approved by the FDA for the treatment of pulmonary hypertension. Since then, the ERAs, ambrisentan and macitentan, were approved in 2007 and 2013, respectively. Apart from the treatment of pulmonary hypertension, ERAs have been shown to lower blood pressure and slow kidney disease progression in patients with diabetic nephropathy; however, the mechanisms are unclear (de Zeeuw et al. 2014; Heerspink et al. 2019). To our knowledge, no endpoints related to improvement of insulin sensitivity or glucose control have been assessed in any clinical trial. One mechanism by which the drugs could improve outcomes in diabetes is through improvements in glucose control as suggested by the current study. Our data indicate that patients associated with obesity and diabetes treated with an ERA had lower HbA1c compared to pretreatment levels, suggesting an important contribution of ET-1 to the development of obesity-induced insulin resistance and hyperglycemia.

ERAs are categorized into either ET_A or dual ET_A/ET_B antagonists. Currently, on the market, there are selective ET_A antagonists (ambrisentan) and dual ET_A/ET_B antagonists (bosentan and macitentan). Selective ET_B antagonism is contraindicated due to significant salt-sensitive hypertension produced (Speed et al. 2011). In the current study, patients were either prescribed ambrisentan or macitentan, with no difference between the two, although this analysis may have been limited due to the sample size (data not shown). Interestingly, both ET_A receptor antagonism and dual ET_A/ET_B antagonism have similar improvements in fasting blood glucose and insulin sensitivity in mice following 10 weeks of high-fat diet (Rivera-Gonzalez et al. 2021). Given that mechanisms of how ET-1 improves glucose handling are unknown, it is unclear whether diabetic patients would benefit more from a selective ET_A antagonist or a dual antagonist.

Although not FDA indicated, previous studies have shown that there are possible clinical benefits for using ERA in diabetic patients. Results from the Reducing Residual Albuminuria in Subjects with Diabetes and Nephropathy trial show that total cholesterol, low-density lipoprotein cholesterol, and triglyceride levels were significantly decreased in diabetic patients treated with atrasentan (de Zeeuw et al. 2014). More recently, the Study of Diabetic Nephropathy with Atrasentan trial indicated that atrasentan reduced adverse renal outcomes in patients with type 2 diabetes and chronic kidney disease, increasing ERA’s clinical significance and proving its need for further investigation (Heerspink et al. 2019). These data further suggest a future for ERAs in diabetes treatment and growing importance in light of the increasing obese population.

While this study represents an important step in supporting future investigation into ERAs in obesity-induced insulin resistance, there are several limitations in the current data that need to be addressed. First, the study is retrospective and uses electronic health record data, yielding highly variable length of follow-up and various conditions that may affect acute measurements of certain variables such as circulating lipids and glucose. However, follow-up time was not shown to play a significant role in the HbA1c change, and HbA1c is a very stable marker that represents glucose control over several months. Factors such as limited routine medical care and poor socioeconomic conditions in the surrounding area of Jackson, MS, most likely played a role in the current findings, in particular the high prevalence of diabetes (~80%) and hypertension (~90%). The retrospective design also does not assure that all UMMC clinics followed the same standardized measurement methods for variables such as blood pressure. Diagnoses of the comorbidities in the current study were dependent on the ICD-10 codes from the corresponding physician. Finally, while some clinical evidence suggests that ERAs improve dyslipidemia in diabetes (de Zeeuw et al. 2014), our analysis did not find significant changes at follow-up (Supplementary Fig. S2). The small sample size of the ERA group may have limited the power to detect effects for these variables as well as others (e.g., changes in body size).

Obesity is a growing epidemic worldwide and increases the risk of hypertension, dyslipidemia, and insulin resistance leading to type II diabetes (Leggio et al. 2017). It is well established that ET-1 is upregulated in people with obesity and animal models of diet-induced obesity, making this population an ideal target for the use of ERAs to improve several risk factors associated with obesity, including insulin resistance. To date, no clinical trials have reported ERA responses in an obese population. ERAs have shown efficacy to improve insulin sensitivity in rodents (Rivera-Gonzalez et al. 2021). The current retrospective data support the idea that blocking the ET-1 system in obesity-induced insulin resistance improves long-term glucose control. Given the overall efficacy of ERAs to delay the onset of diabetic nephropathy (Heerspink et al. 2019), improve dyslipidemia (Farrah et al. 2019), and reduce blood pressure (Yuan et al. 2017), we believe that the current study supports the need for a proper placebo-controlled clinical trial to determine the impact of blocking ET-1 in insulinresistant populations.

Data availability

All raw data were derived from the University of Mississippi Medical Center Research Data Warehouse.