Prognostic value of exercise capacity among men undergoing pharmacologic treatment for erectile dysfunction: The FIT Project

Abstract

Background

Vascular erectile dysfunction (ED) has been identified as a potentially useful risk factor for predicting future cardiovascular events, particularly in younger men. Because these men typically score more favorably on traditional cardiovascular disease risk assessment tools, there exists a gap in knowledge for how to most appropriately identify those men who would benefit from more aggressive treatments. To date, no studies have examined the impact of fitness on cardiovascular outcomes in men with ED. This study sought to examine the prognostic impact of maximal exercise capacity on cardiovascular‐related outcomes in men ages 40 to 60 years being treated for ED.

Hypothesis

We hypothesized that there would be an independent association between higher baseline fitness level and lower cardiovascular events.

Methods

We analyzed 1152 men with pharmacy claims file–confirmed active pharmacologic treatment for ED from the Henry Ford Exercise Testing (FIT) Project (1991–2009). All patients were free of coronary heart disease and heart failure, and underwent clinician‐referred exercise stress testing, with fitness measured in metabolic equivalents of task (METs). Multivariable Cox proportional hazard models adjusted for traditional cardiovascular risk factors were used to study the association between fitness and all‐cause mortality, major adverse cardiovascular events (MACE) (defined as myocardial infarction or revascularization), and incident type 2 diabetes mellitus.

Results

The mean age of the population was 53 years, with 39% African Americans. In multivariable analysis, each 1 MET of fitness was associated with a 16% lower risk of death (hazard ratio [HR]: 0.84, 95% confidence interval [CI]: 0.76‐0.94, P = 0.002), and a nonsignificant reduction in MACE (HR: 0.89, 95% CI: 0.79‐1.003, P = 0.048), and incident diabetes (HR: 0.92, 95% CI: 0.85‐1.01, P = 0.129).

Conclusions

Higher baseline fitness is associated with improved cardiovascular prognosis in a population of middle‐aged men treated for ED.

1. INTRODUCTION

Vascular erectile dysfunction (ED) is common in the general adult male population, with some estimates as high as 15% of the male population ages 40 to 60 years old.1 ED is particularly common among men with cardiovascular disease (CVD) due to shared risk factors.2 Studies show symptoms of ED precede clinically evident coronary artery disease by 2 to 3 years and cardiovascular events by 3 to 5 years.3, 4, 5 ED is an independent predictor of CVD events as noted in several studies, and has therefore been proposed as a novel cardiovascular risk factor.6 In fact, a recently published cohort study found that ED was associated with a 25% increased risk of cardiovascular disease.7 Based on those results, erectile dysfunction (or its treatment) has been added to the QRISK3 risk prediction model for cardiovascular disease used by the National Health Service. Although ED is not incorporated in most other CVD risk assessment guidelines, it has generally been suggested that men with ED be considered for aggressive preventive therapies before they develop overt CVD.8, 9

Previous studies among men with ED show that lifestyle modifications, including increased physical activity, are effective in improving both overall vascular health and symptoms of ED.3 Fitness is directly associated with improved survival both in the healthy population and among those with comorbidities.10, 11, 12, 13, 14, 15, 16 Poor fitness is linked with incident diabetes mellitus, a major risk factor for ED.17, 18 However, to our knowledge, no studies have evaluated the relationship between fitness and risk of cardiovascular‐related outcomes in men with ED. The assessment of fitness among patients with ED may enable clinicians to further risk stratify the subset of men who may benefit from more aggressive risk factor control. Therefore, we aimed to examine the effect of fitness on the risk of death, major adverse cardiovascular events (MACE), and incident diabetes in a group of middle‐aged men requiring pharmacologic treatment for ED.

2. METHODS

We analyzed data from the Henry Ford Exercise Testing (FIT) Project. The FIT Project is a retrospective cohort study aimed at investigating the long‐term association of maximal exercise capacity and clinical events. In summary, the study combined measured clinical exercise data with medical history, treatment data, and outcome data derived from electronic medical records and administrative claims files. The FIT Project included 69 885 patients who underwent a physician‐referred exercise stress test (Bruce protocol) within the Henry Ford Medical System between 1991 and 2009. The indications for stress test referral were categorized as chest pain, rule out ischemia, shortness of breath, risk stratification for coronary artery disease, palpitations, research screening, preoperative testing, and other. Full study details have been previously published.19

Our study cohort consisted of 1152 men from the FIT cohort ages 40 to 60 years old without prior history of congestive heart failure or cardiovascular disease who were undergoing pharmacologic treatment with phosphodiesterase inhibitors (sildenafil, tadalafil, and vardenafil). Medication use was verified through pharmacy claims files. Given the retrospective nature of this study, systematic diagnosis of vascular ED with penile Doppler, and characterization of ED symptoms and severity through participant questionnaire, were not available. In the FIT Project, vital status was determined via linkage with the Social Security Death Master File, whereas nonfatal outcomes were ascertained via linkage to administrative claims files, with censoring at the time of last known contact with the Henry Ford Health System.19 Patients were followed for death, qualifying MACE (defined as myocardial infarction or coronary revascularization occurring greater than 30 days after stress testing), and incident diabetes (defined as new documentation of a diagnosis of diabetes (International Classification of Diseases, Ninth Revision 250.XX) on at least 3 separate encounters). In analyses of diabetes incidence, only patients free of diabetes at baseline were considered.

Fitness was estimated based on the maximal speed and grade achieved and expressed in absolute metabolic equivalents of task (METs) and categorized as <8, 8 to 11, and ≥12. A lower METs group of <8 was selected to illustrate differences in level of exercise capacity between groups. We diverged from the a priori defined lower absolute METs categories of <6 and 6 to 10 METs groups used in prior FIT Project publications due to the low number of patients achieving lower fitness in this otherwise healthy middle‐aged male cohort. Atherosclerotic cardiovascular disease (ASCVD) risk was calculated using the Pooled Cohort Equations.20

Baseline characteristics were presented by METs category as mean ± standard deviation for continuous variables and as the total number and the percent of subjects with that certain attribute for categorical variables. To assess the differences in distributions between patients with different exercise capacities, we used t tests for continuous variables and χ² tests for categorical variables. Kaplan–Meier curves were used to describe event‐free survival by METs category and log‐rank tests to test the equality of survival functions. After confirming the proportionality assumption, multivariable Cox proportional hazard regression models were used to study the association between METs achieved and each outcome. Models were adjusted for potential confounders in accordance with previously published FIT articles. Model 1 was adjusted for age and race (white, black, other). Model 2 was additionally adjusted for obesity, resting systolic blood pressure, resting diastolic blood pressure, prior diagnosis of hypertension, diabetes status, smoking history, hyperlipidemia, family history of coronary heart disease, antihypertensive therapy, lipid‐lowering therapy, chronic lung disease medication use, as well as the reason for referral for stress testing.

We conducted additional sensitivity analyses to ensure consistency of our results. We excluded patients taking pulmonary disease medications (along with the existing exclusion of heart failure) to eliminate the potential uncommon use of phosphodiesterase inhibitors for symptomatic pulmonary arterial hypertension. In addition, we conducted additional analyses, removing patients on antidepressant medications to account for the possibility of psychogenic ED in our dataset. All statistical analyses were done using Stata 13 (StataCorp, College Station, TX) statistical software.

3. RESULTS

Descriptive baseline characteristics of the study cohort are presented in Table 1. The cohort included 1152 participants (55% white, 39% black) of mean (standard deviation) age 53 (5) years who were enrolled in the FIT Project from 1998 to 2009. The prevalence of cardiovascular risk factors was high, with hypertension in 72%, dyslipidemia in 58%, smoking history in 53%, family history of CHD in 47%, and diabetes mellitus in 31%. The most common indications for stress testing were to evaluate chest pain (45%), rule out ischemia (14%), risk stratification for coronary artery disease (11%), and shortness of breath (97%). Most participants fell in a broad intermediate risk category (50%) defined as 5% to 15%, 10‐year estimated ASCVD risk. Compared to lower METs groups, men who achieved higher METs were more likely to have a BMI <30, be nonsmokers, and normotensive.

Table 1.

Baseline population characteristics by METs

Characteristics	Total Cohort, N = 1152	<8 METs, N = 259	8–11 METs, N = 524	≥12 METs, N = 369	P Value
Demographic Data
Age, y	53 ± 5	54 ± 4	53 ± 5	52 ± 5	<0.001
Race, no (%)					<0.001
White	636 (55)	116 (45)	293 (56)	227 (62)
Black	453 (39)	136 (53)	199 (38)	118 (32)
Other	63 (6)	7 (3)	32 (6)	24 (7)
Physiologic measurements, mean ± SD
Resting systolic blood pressure, mm Hg	131 ± 17	135 ± 19	132 ± 16	128 ±15	<0.001
Resting diastolic blood pressure, mm Hg	83 ± 10	83 ± 11	83 ± 10	81 ± 9	0.02
Resting heart rate, bpm	73 ± 12	77 ± 13	74 ± 12	68 ± 11	<0.001
Peak systolic blood pressure, mm Hg	185 ± 25	187 ± 32	186 ± 24	183 ± 21	0.14
Peak diastolic blood pressure, mm Hg	83 ± 14	85 ± 15	83 ± 13	82 ± 13	<0.01
Peak heart rate, bpm	152 ± 15	144 ± 18	151 ± 14	159 ± 12	<0.001
Total cholesterol	202 ± 42	202 ± 45	203 ± 41	203 ± 40	0.92
HDL‐C	46 ± 13	44 ± 12	44 ± 12	48 ± 13	<0.001
LDL‐C	124 ± 38	122 ± 38	123 ± 37	128 ± 39	0.09
Medical history, no (%)
Hypertension	824 (72)	217 (84)	389 (74)	218 (59)	<0.001
Dyslipidemia	662 (58)	161 (62)	314 (60)	187 (51)	<0.01
Cigarette smoking history	615 (53)	160 (62)	281 (54)	174 (47)	0.001
Diabetes mellitus	353 (31)	119 (46)	161 (31)	73 (20)	<0.001
Family history of CHD	539 (47)	100 (39)	247 (47)	192 (52)	0.011
Obesity	412 (36)	149 (58)	205 (39)	58 (16)	<0.001
Lung disease medication use	140 (12)	49 (19)	53 (10)	38 (10)	0.002
Indication for stress test, no (%)
Chest pain	519 (45)	106 (41)	248 (47)	165 (45)
Rule out ischemia	160 (14)	34 (13)	71 (14)	55 (15)
Risk factor stratification	121 (11)	25 (10)	54 (10)	42 (11)
Shortness of breath	81 (7)	33 (13)	29 (6)	19 (5)
Pooled cohort equation risk categories, no (%)
AHA/ACC ASCVD risk score					<0.001
0%–5%	158 (13.8)	15 (5.9)	57 (10.9)	86 (23.4)
5%–15%	572 (49.9)	91 (35.6)	264 (50.6)	217 (59.0)
≥15%	416 (36.3)	150 (58.6)	201 (38.5)	65 (17.7)

Abbreviations: AHA/ACC, American Heart Association/American College of Cardiology; ASCVD, atherosclerotic cardiovascular disease; CHD, coronary heart disease; HDL‐C, high‐density lipoprotein cholesterol; LDL‐C, low‐density lipoprotein cholesterol; METs, metabolic equivalents of task; SD, standard deviation.

In total, patients were followed for mean (standard deviation) of 9.0 (2.9) years for the outcome of death, 4.7 (2.8) years for a qualifying MACE event, and 4.5 (2.9) years for incident diabetes. There were 799 participants free of diabetes at baseline who were included in analyses of incident diabetes. Over the course of follow‐up, there were 71 deaths (6%), 59 qualifying MACE (5%), and 141 cases of incident diabetes (18%). Compared to those achieving <8 METs, men with ED who were able to achieve 8 to 11 and ≥12 METs had lower rates of mortality, MACE, and diabetes (Figure 1). Table 2 shows multivariable adjusted hazard ratios of events for the categorical and continuous definitions of achieved METs. Each additional MET was associated with a 16% lower risk of mortality (hazard ratio [HR]: 0.84, 95% confidence interval [CI]: 0.76‐0.94), and was borderline significant in MACE (HR: 0.89, 95% CI: 0.79‐1.003) and diabetes (HR: 0.92, 95% CI: 0.85‐1.01). In addition to adjustment with the risk factors included in model 1 and model 2, we also conducted sensitivity analyses excluding either pulmonary medication or antidepressant medication use, and found that there were no meaningful changes to our results.

Figure 1.

Kaplan‐Meier survival estimates of (A) death, (B) major adverse cardiac events, and (C) incident diabetes. P values shown are for log‐rank test for equality of survivor functions. Abbreviations: METs, metabolic equivalents of task.

Table 2.

Multivariable adjusted hazard ratio (95% confidence interval) for the relationship between METs and death, MACE, and incident diabetes

	No. of Events, %	Model 1, HR (95% CI)	Model 2, HR (95% CI)
Death	71, 6.2%
METs <8	31, 12.0%	1.0	1.0
METs 8–11	23, 4.4%	0.40 (0.23‐0.70)	0.37 (0.21‐0.66)
METs ≥12	17, 4.6%	0.44 (0.24‐0.81)	0.37 (0.19‐0.71)
P for trend		0.005	0.002
Per MET achieved		0.88 (0.80‐0.96)	0.84 (0.76‐0.94)
MACE	59, 5.3%
METs <8	20, 8.0%	1.0	1.0
METs 8–11	25, 4.9%	0.57 (0.32‐1.04)	0.61 (0.32‐1.16)
METs ≥12	14, 3.9%	0.41 (0.20‐0.82)	0.45 (0.20‐0.99)
P for trend		0.012	0.048
Per MET achieved		0.87 (0.79‐0.96)	0.89 (0.79‐1.00)
Incident diabetes	141, 17.7%
METs <8	31, 22.1%	1.0	1.0
METs 8–11	76, 20.9%	0.97 (0.62‐1.51)	1.04 (0.64‐1.67)
METs ≥12	34, 11.5%	0.48 (0.28‐0.81)	0.66 (0.37‐1.19)
P for trend		0.002	0.129
Per MET achieved		0.89 (0.83‐0.96)	0.92 (0.85‐1.01)

Abbreviations: CI, confidence interval; HR, hazard ratio; MACE, major adverse cardiac event; METs, metabolic equivalents of task.

Model 1: Adjusted for age and race. Model 2: Model 1 plus resting systolic blood pressure, resting diastolic blood pressure, hypertension, use of hypertension medications, diabetes (death and MACE models), dyslipidemia, use of lipid‐lowering medications, use of pulmonary medications, obesity, family history of coronary heart disease, smoking status, and indication for stress testing.

4. DISCUSSION

We observed higher baseline fitness to have an inverse relationship with death, MACE, and development of diabetes in this population of middle‐aged men with ED symptoms. Despite the presence of multiple cardiac risk factors such as hypertension, dyslipidemia, smoking history, obesity, and family history of heart disease among men in our study, the significant decreased risk for all‐cause mortality and trends toward decreased MACE and development of diabetes remained after thorough risk factor adjustment. Higher baseline fitness was observed to be a top predictor of cardiovascular outcomes in our population when compared to other traditional risk factors. In addition to being the first study to examine the prognostic impact of fitness in men undergoing treatment for ED, strengths of our study included the racially diverse sample as well as extended follow‐up out to 9 years. The FIT Project also included a standard clinical assessment of exercise capacity using the Bruce protocol.

Previous studies have shown that ED predicts cardiovascular events with similar accuracy as more traditional CVD risk factors like smoking, hyperlipidemia, or family history of myocardial infarction.21, 22 As a result, researchers have identified men with ED as a unique population in whom early lifestyle intervention may be particularly valuable, as the onset of ED symptoms has been found to precede clinical symptoms of coronary artery disease.4, 5 Erectile dysfunction has recently been validated as a cardiovascular risk factor to be used as part of the QRISK3 calculator for the National Health Service.7 Esposito et al23 found that lifestyle intervention was successful in reducing the incidence of ED in an at‐risk population, and although cardiac events were not collected, the intervention group had improved total cholesterol, blood pressure, and weight after 2 years. The current study provides observational evidence supporting the concept that lifestyle modifications aimed at improving fitness may attenuate the risk of cardiovascular events in patients with ED, although further studies are needed to prove causation.

This study has a few limitations. First, our ED definition may not be 100% specific, because phosphodiesterase inhibitors may very rarely be prescribed for other diseases. However, because participants with diagnoses of heart failure and those taking pulmonary medications were excluded from analysis, it is highly unlikely that any of the included participants had pulmonary hypertension requiring treatment with phosphodiesterase inhibitors without also being excluded. By design, our ED definition is also not 100% sensitive, because we likely will have excluded men with less‐severe ED symptoms who may not have a clinical ED diagnosis and who may not require pharmacologic treatment. As a result, our sample may be biased toward men with more severe ED symptoms. Although we did not use a questionnaire to assess for ED symptoms, that method can be subject to desirability bias, and medication use is a more objective method to characterize ED. Prior studies have likewise utilized the prescription of phosphodiesterase inhibitors as a proxy for the presence of ED.7, 8, 24

Alternatively, our sample may be confounded by a treatment bias, as patients who are healthier and have higher socioeconomic status may be more likely to undergo treatment for ED. The men in our sample were also relatively fit, with the average participant able to achieve 10.2 METs. Despite the prevalence of cardiovascular risk factors in our study population, our study was underpowered to look exclusively at acute myocardial infarctions during follow‐up, likely due to the younger age of our population. Lastly, our results may not be generalizable to low‐risk asymptomatic individuals, because our patients were clinically referred for stress testing. However, we argue that because ED is common in patients at risk for CVD, our results are clinically applicable to this specialized high‐risk primary prevention population.

5. CONCLUSION

In a racially diverse sample, we are the first to observe that greater baseline fitness is associated with lower incidence of death, MACE, and development of diabetes in men undergoing pharmacologic treatment for ED symptoms. Future studies are warranted to examine whether improving fitness through lifestyle intervention can reduce risk of adverse outcomes among men with ED.

Author contributions

Robert V. Same, MD, and Mahmoud Al Rifai, MD, are co–first authors. All authors contributed equally to the generation of this article.

Conflicts of interest

Dr. Blaha reports grants from the National Heart, Lung, and Blood Institute, grants from American Heart Association, grants from the Food and Drug Administration, grants from Aetna Foundation, personal fees from Novartis, and personal fees from Amgen, outside the submitted work. Dr. Billups reports personal fees from Abbvie, outside the submitted work.

Same R. V, Al Rifai M, Feldman D. I, et al. Prognostic value of exercise capacity among men undergoing pharmacologic treatment for erectile dysfunction: The FIT Project. Clin Cardiol. 2017;40:1049–1054. 10.1002/clc.22768