How Closely Do Clinical Trial Participants Resemble “Real-World” Patients with Groups 2 and 3 Pulmonary Hypertension? A Structured Review

To the Editor:

Randomized controlled trials (RCTs) of pulmonary vasodilators in the most common forms of pulmonary hypertension (PH)—those secondary to left heart disease (group 2 PH) and chronic lung disease (group 3 PH)—have failed to show consistent benefits, and some have shown a signal of harm (1, 2). Pulmonary vasodilators are costly; none are approved by the U.S. Food and Drug Administration for use in groups 2 or 3 PH; and guidelines recommend against their routine use in this context (3, 4). Despite this, use in “real-world” practice is common and rising over time (5, 6).

Guideline-discordant use of pulmonary vasodilators in some patients with groups 2 and 3 PH may be driven by a perception that nontrial patients may experience a more favorable balance of benefits to harms from treatment than observed in RCTs. Understanding how characteristics of patients with groups 2 and 3 PH in nontrial settings compare with RCT participants may help anticipate the risk/benefit ratio of using pulmonary vasodilators in groups 2/3 PH in a real-world population. We sought to compare baseline characteristics of participants in sentinel groups 2 and 3 PH RCTs with those of patients in the Veterans Health Administration, the largest national integrated healthcare system in the United States. We hypothesized that nontrial patients would have a higher burden of comorbid conditions than RCT participants.

Methods

We selected RCTs evaluating use of pulmonary vasodilators in PH due to left heart disease or lung disease. RCTs were eligible for comparison if they met the following criteria: 1) studied the target population of group 2 or 3 PH, 2) were full-length articles, 3) included at least 50 study participants, and 4) reported baseline comorbidities. From the selected RCTs, we abstracted study details (first author, study drug, key exclusion criteria, endpoint studied, and primary results) and baseline participant characteristics (age, sex, race and ethnicity, and pertinent comorbid conditions). We contacted authors to obtain baseline characteristics if not reported. We calculated pooled estimates for each of the variables using available data. For comparison, we created a national cohort of patients with groups 2 and 3 PH in nontrial settings by linking patient-level data from the Veterans Health Administration and Centers for Medicare and Medicaid Services, as previously described (5). The Edith Nourse Rogers Memorial Veterans Hospital Institutional Review Board approved this study.

Results

We identified eight RCTs meeting inclusion criteria, with a pooled sample size of 938 patients. Table 1 shows trial details, including the study drug, population studied, key exclusion criteria, endpoints studied, and primary results. Many trials excluded patients with common conditions, including kidney or liver impairment, ischemic heart disease, and atrial fibrillation. Three group 2 PH trials excluded participants with lung disease, and two group 3 PH trials excluded participants with heart disease. We identified 136,670 patients with groups 2/3 PH in our nontrial cohort; 2,813 (2.1%) received pulmonary vasodilators. In our nontrial cohort, 21.6% had group 2 PH only, 8.4% had group 3 PH only, and 70.0% had conditions associated with both groups 2 and 3 PH. Patients in the nontrial cohort were elderly and had high rates of comorbid illnesses (Table 2). Compared with groups 2 and 3 PH trial participants, nontrial patients with groups 2/3 PH were older (76.8 vs. 67.1 yr) and more racially diverse. Nontrial patients with PH had a greater burden of comorbid illnesses, including a higher prevalence of diabetes (49.7% vs. 29.4%), hypertension (92.9% vs. 61.7%), hyperlipidemia (81.7% vs. 33.7%), chronic kidney disease (40.1% vs. 6.6%), and arrhythmia (63.1% vs. 34.1%) at the time of PH diagnosis.

Table 1.

Study design details of groups 2 and 3 pulmonary hypertension randomized clinical trials

Trial	Study Drug	Population Studied	Exclusion Criteria	Primary Endpoint	Result
Group 2 PH clinical trials
Bonderman (11) (n = 201)	Riociguat	HFrEF (EF, ≤40%)	1. Other PH causes	Mean PAP	No change
			2. Cardiac decompensation in preceding 30 d
			3. Baseline systolic blood pressure <100 mm Hg
			4. Severe renal impairment (GFR, <30 ml/min)
			5. Cardiac ischemia with planned percutaneous coronary intervention or bypass surgery
Bermejo (1) (n = 200)	Sildenafil	Corrected valvular heart disease	1. Persistent significant valvular dysfunction	Composite clinical score*	Worsening in treated group
			2. Myocardial infarction, stroke, or life-threatening arrhythmia in preceding 6 mo
			3. Baseline systolic blood pressure <90 mm Hg
			4. Severe renal impairment (GFR, <30 ml/min) or liver dysfunction
			5. Life expectancy <2 yr
Kaluski (12) (n = 87)	Bosentan	HFrEF (EF, <35%), NYHA FC IIIb–IV	1. Baseline systolic blood pressure <100 mm Hg	Systolic PAP	No change; more frequent adverse events in treated group requiring drug discontinuation
Vachiéry (13) (n = 63)	Macitentan	HFpEF (EF, ≥30%), NYHA FC II–III	1. Baseline blood pressure >180/100 mm Hg or systolic blood pressure <90 mm Hg	Composite of fluid retention or worsening NYHA FC	Increased fluid retention in treated group
			2. Uncontrolled heart rate from atrial fibrillation
			3. Severe renal impairment (GFR, <30 ml/min) or liver dysfunction
			4. Unstable coronary artery disease or myocardial infarction within 6 mo
			5. Severe obstructive or moderate to severe restrictive lung disease
			6. Oxygen saturation <90% on room air
			7. Anemia (hemoglobin <10 g/dl)
			8. Other PH causes
Hoendermis (14) (n = 52)	Sildenafil	HFpEF (EF, ≥45%), NYHA FC II–IV	1. Severe noncardiac exercise limitation	Mean PAP	No change
			2. Other PH causes
			3. Myocardial infarction or coronary ischemia in preceding 6 mo
			4. Blood pressure <90/50 mm Hg
			5. Significant mitral or aortic valvular dysfunction
			6. Severe liver dysfunction
Group 3 PH clinical trials
Nathan (2) (n = 147)	Riociguat	Idiopathic interstitial pneumonia	1. Systolic blood pressure <95 mm Hg	6MWD	Stopped early; increased harm in treated group
			2. Forced vital capacity <45%
			3. Active smoking
Goudie (15) (n = 120)	Tadalafil	COPD	1. Left ventricular systolic dysfunction (EF, <45%)	6MWD	No change
			2. Systolic blood pressure <90 mm Hg
			3. Recent stroke or unstable angina
			4. COPD exacerbation within 1 mo
Raghu (16) (n = 68)	Ambrisentan	Idiopathic pulmonary fibrosis	1. NYHA FC III–IV	Disease progression†	Unfavorable trend; more hospitalizations in treated group
			2. EF <40%
			3. Coexisting obstructive airflow or emphysema on computed tomography
			4. Hospitalization or respiratory infection within 60 d

Definition of abbreviations: 6MWD = 6-minute-walk distance; COPD = chronic obstructive pulmonary disease; EF = ejection fraction; GFR = glomerular filtration rate; HFpEF = heart failure with preserved ejection fraction; HFrEF = heart failure with reduced ejection fraction; NYHA FC = New York Heart Association functional class; PAP = pulmonary arterial pressure; PH = pulmonary hypertension.

^*Combination of death, hospitalization for heart failure, change in NYHA FC, and patient global self-assessment.

^†Composite endpoint of 1) decline in functional vital capacity and diffusing capacity of the lung for carbon monoxide, 2) respiratory hospitalization event, and 3) death of any cause.

Table 2.

Comparison of baseline characteristics of nontrial patients with participants enrolled in select group 2 and group 3 pulmonary hypertension randomized clinical trials

Cohort Characteristic	Nontrial Cohort	Trial cohort*	Group 2 PH Trials						Group 3 PH Trials
			Bonderman (11)	Bermejo (1)	Kaluski (12)	Vachiéry (13)	Hoendermis (14)		Nathan (2)	Goudie (15)	Raghu (16)
n	136,670	938	201	200	87	63		52	147	120	68
Age, yr, mean (SD)	76.8 (9.5)	67.1†	58.1‡	70.0‡	69.5 (9.4)	70.0‡		74.0 (9.9)	68.5 (8.0)	69.0 (7.6)	68.0 (6.1)
Female sex	2.5%	42.3%	14.4%	77.0%	28.7%	65.1%		71.2%	35.4%	31.7%	30.9%
Race/ethnicity
White	85.3%	95.1%	—	100%	98.9%	—		96.2%	85.7%	100%	86.8%
Black	9.4%	1.2%	—	0%	1.1%§	—		3.8%§	2.7%	0%	1.5%
Hispanic	1.9%	0.9%	—	0%	—	—		—	3.4%	0%	0%
Comorbid conditions
Diabetes	49.7%	29.4%	43.3%	29.0%	36.8%	42.9%		34.6%	29.9%	8.3%	—
Smoking	41.8%	34.6%	—	6.5%	—	—		—	4.1%	100%	67.6%
Hypertension	92.9%	61.7%	—	64.0%	66.7%	90.5%		90.4%	55.8%	34.2%	—
Hyperlipidemia	81.7%	33.7%	—	42.5%	54.0%	7.9%		51.9%	14.3%	—	—
Chronic kidney disease	40.1%	6.6%	0%ǁ	0%ǁ	—	39.7%		—	8.2%	9.2%	—
Cerebrovascular disease	17.4%	8.5%	—	—	—	—		15.4%	6.8%	7.5%	—
Heart disease	67.9%	20.7%	0%ǁ	13.5%	72.4%	—		32.7%	34.7%	7.5%	—
Arrhythmia	63.1%	34.1%	12.9%	77.0%	20.7%	73.0%		61.5%	10.2%	5.0%	—

Definition of abbreviations: PH = pulmonary hypertension; SD = standard deviation.

Data are presented as percentages unless otherwise noted; data not available from authors is designated as —.

^*Pooled estimate calculated on the basis of available data.

^†Pooled SD unable to be calculated because SD was not provided for all studies.

^‡SD not provided.

^§Maximum value possible based on prevalence of other races.

^ǁAssumed on the basis of trial exclusion criteria.

Discussion

Many groups 2 and 3 PH RCTs had extensive exclusion criteria, limiting patients with common comorbid conditions and resulting in an overall healthier population. At least 70% of our nontrial cohort would have been ineligible for inclusion in these trials because they carried diagnoses associated with both groups 2 and 3 PH. This high rate of ineligibility is not unique to PH; the elderly and those with significant comorbidities are groups frequently excluded from clinical trials of cardiovascular disease (7, 8).

Patients with groups 2 and 3 PH in real-world practice were older and had a higher burden of comorbid disease than trial participants. Importantly, key differences between our nontrial cohort and trial participants, such as higher prevalence of chronic kidney disease and arrhythmias, may further reduce both the anticipated efficacy and safety profile of these drugs in nontrial settings. Although no RCT has directly evaluated outcomes of pulmonary vasodilators in older, sicker patients with groups 2/3 PH, a subgroup analysis in a trial of vasodilators in group 2 PH (1) suggested that older patients and those with worse functional class may experience an even greater harm/benefit ratio. In addition, evidence in group 1 PH suggests that patients with cardiovascular risk factors such as hypertension, diabetes, or atrial fibrillation experience reduced efficacy and higher rates of adverse events compared with those without such risk factors when treated with pulmonary vasodilators (9, 10).

Our study has limitations. Many of the trials did not report complete baseline characteristics, and, despite our efforts, we were unable to obtain these missing data from the authors. Thus, our comparisons may be valid only for the studies with complete data. Our cohort of veterans with PH does not represent all real-world practice settings. In addition, the granularity of our data did not allow comparisons of functional class or hemodynamics between the trial and nontrial cohorts. As such, we were unable to discern the role of disease severity in the decision to prescribe outside of guideline recommendations.

This comparison of RCT participants with groups 2 and 3 PH with nontrial patients with PH reveals significant differences in these populations, with the latter experiencing a greater degree of multimorbidity. RCTs demonstrate that pulmonary vasodilators offer limited benefit and potential for harm among the younger, healthier participants with groups 2 and 3 PH enrolled in these studies, and it is even more likely that the more medically complex patients seen in real-world practice will experience harm. Thus, our findings support the call in guidelines to limit the use of pulmonary vasodilators in patients with groups 2 and 3 PH in clinical practice.