| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Division of Pulmonary, Critical Care, and Sleep Medicine, Tufts Medical Center, Boston, Massachusetts
Correspondence and requests for reprints should be addressed to Nicholas S. Hill, M.D., Division of Pulmonary, Critical Care, and Sleep Medicine, Tufts Medical Center, 800 Washington Street, #257, Boston, MA 02111. E-mail: nhill{at}tuftsmedicalcenter.org
ABSTRACT
Results from clinical trials serve as the basis for approval of therapies by regulatory agencies as well as for treatment decisions by clinicians. But these findings are relevant only to patients who are similar to the ones enrolled into the trials. This is germane to clinical trials on pulmonary arterial hypertension (PAH) because the disease is uncommon but highly heterogeneous and results can easily be misapplied. The characteristics of patients entering the trials are largely determined by inclusion/exclusion criteria, with the result that most participants have idiopathic, connective tissue disease– and congenital heart disease–related PAH. Earlier trials enrolled patients mainly with New York Heart Association functional class III and IV disease and with severe pulmonary hemodynamic abnormalities. Because it has been the major outcome variable in most of the trials, eligibility is also dependent on six-minute-walk distance, ensuring that patients are moderately but not too severely functionally impaired, thereby maximizing the likelihood of detecting a favorable response to therapy. More recent trials have enrolled less ill patients, with more patients with New York Heart Association functional class II disease, less severe hemodynamic abnormalities, and more stability over time. This reflects, in part, ethical concerns about enrolling sicker patients into placebo-controlled trials. Trials have mainly enrolled white females in their 40s and 50s and have consistently excluded non-WHO group 1 forms of pulmonary hypertension. These characteristics must be carefully considered when applying the findings of pulmonary hypertension trials in clinical practice.
Key Words: randomized controlled trial • prostacyclin • endothelin receptor antagonist • phosphodiesterase inhibitor
Clinicians rely on evidence from clinical trials to guide them in selecting the best therapies for their patients. However, it is important to consider that the results of clinical trials apply only to patients who are similar to those enrolled in the trials. Therefore, if the characteristics of patients in a given trial do not resemble those of most patients seen in clinical practice, these results may have very limited applicability. For this reason, understanding the characteristics of patients enrolled into trials is important.
Many considerations come into play when clinical trials are designed. The designers must identify the target population as clearly as possible and exclude patients whose comorbidities might add to variability, or whose safety might be threatened by inclusion. Characteristics of the target population depend partly on the end points selected; clinical researchers and particularly sponsors want end points that are likely to show favorable outcomes and are acceptable to approval agencies such as the U.S. Food and Drug Administration (FDA). The study population is then targeted to optimize the likelihood of showing a favorable response to the main outcome variable.
These considerations are highly relevant to the design of pulmonary arterial hypertension (PAH) clinical trials. An orphan disease, PAH affords a relatively small patient population for study, rendering outcomes such as survival difficult to attain. Thus, easier to achieve functional outcomes have been the primary variables in most of the PAH trials to date. This means that patients entering the trials are selected for a certain baseline functional capacity to maximize the likelihood of achieving significant benefit. In this article, we will expand on these concepts, discuss principles of patient recruitment, examine inclusion and exclusion criteria, and analyze the clinical characteristics of the patients who are actually enrolled. Using these metrics as a guide, we will assess key randomized trials for PAH that have led to approval of currently available drugs (and some that failed to get approval). Understanding the composition of these target populations will inform our decision making when selecting current PAH pharmacotherapies for individual patients.
TARGET POPULATION: INCLUSION AND EXCLUSION CRITERIA
Defining the target population is a key part of clinical trial design. Ideally, the target population of the study reflects the characteristics of most patients with the disease, but this may not always be the case. The inclusions aim to limit heterogeneity as much as possible. They also attempt to maximize the likelihood of a favorable response to key outcome variables. Exclusions also limit heterogeneity and eliminate patients who are inappropriate for reasons such as pregnancy or those who might be harmed by the therapy. In the end, an overly specified target population may render results applicable to only a minority of affected patients. The goal is to create a study cohort that is reflective of the disease population, yet homogenous enough to ensure that outcomes truly reflect the impact of the intervention, rather than differences between individual subjects.
Pharmaceutical companies, which sponsor most of the clinical trials on PAH, desire end points that are clinically meaningful, but they must be acceptable to approval agencies such as the FDA in the United States or similar agencies in other countries. They must also lead to approval for potentially effective pharmacotherapies as quickly and economically as possible. Naturally, these considerations influence selection of inclusion and exclusion criteria. As an orphan disease, the pool of patients with PAH is relatively small and approval agencies "discount" end points, permitting outcomes besides survival that would not be acceptable in trials on more common diseases, such as coronary artery disease. Thus, most PAH trials use a functional outcome as the primary variable; the distance walked in 6 minutes (also referred to as the six-minute-walk test). Accordingly, inclusion criteria target patients who are most likely to improve functionally; they can walk a minimal distance (>100–150 m) but not far enough (<400–475 m) to make detection of improvement unlikely due to a possible "ceiling effect" (1).
Ethical considerations are also important, in that the availability of effective alternative therapies makes it difficult to recruit patients into placebo-controlled trials. This favors the enrollment of patients who are likely to be stable for the duration of the study, usually 12 to 16 weeks for PAH trials. Even when not explicitly stated in the inclusion/exclusion criteria, this influences investigators to enroll patients who have been stable over the prior several months. Rather than entry into the study, less stable patients are usually offered an approved therapy previously shown to be effective.
RECRUITMENT AND ENROLLMENT
Once the target population is identified, the challenge from the standpoint of conducting a clinical trial becomes recruitment and enrollment. As shown in Figure 1, not all targeted patients are screened to enroll in a trial, perhaps because personnel are not always available to enroll them or they are deemed to be too unstable; thus, the potential number of enrollees drops off. Another drop occurs when patients are screened, because some do not meet inclusion criteria or have exclusions, further reducing the number of patients who are deemed to be eligible. Among those patients who meet entry criteria and are offered enrollment, some may decline the invitation to participate in a research study or live too far away. In the end, only a portion of the screened patients are enrolled, known as the "recruitment fraction" (2). This is important to report, because it reflects how well the enrolled population represents the targeted one.
|
Some studies have very low recruitment fractions. This can be a problem if the inclusion criteria are very narrow or there are many exclusion criteria. A study of inhaled nitric oxide to treat acute respiratory distress syndrome had a recruitment fraction of only 5.6%, raising serious questions about its applicability to the greater population with acute respiratory distress syndrome (3). How much this problem applies to PAH trials is unknown, because none have reported recruitment fractions. On the other hand, if inclusion criteria are too broad and exclusions too few, the recruitment fraction is increased at the expense of greater heterogeneity of enrolled patients, making it difficult to apply trial results to each subgroup within the study population. This is relevant to PAH trials, which commonly enroll patients with potentially diverse pathophysiologic states, such as idiopathic, connective tissue disease– and congenital heart disease–associated forms of PAH. Overall benefit in the trial could be driven by a populous subgroup, obscuring lack of benefit or even harm in a smaller subgroup. We encourage the reporting of recruitment fractions in future clinical trials on PAH, to better characterize the population enrolled.
INCLUSION/EXCLUSION CRITERIA IN RANDOMIZED CONTROLLED PAH TRIALS
Table 1 outlines the inclusion and exclusion criteria used in the key RCTs used by the FDA to approve currently available PAH therapies, as well as some of the more recent combination trials. These trials have a number of entry criteria in common. Virtually all of them (with one exception noted below) have included only patients with PAH (World Health Organization [WHO] group 1; see Table 2). With the exception of a few recent trials that have added new therapies to existing ones, the studies excluded patients on other PAH therapies. All required that patients undergo right heart catheterizations, either at the time of enrollment in the earlier trials or within the previous few years in some of the more recent trials. Most used pulmonary hemodynamic criteria to define PAH based on those used in the National Institutes of Health (NIH) registry, which collected data during the early 1980s (4). These include a mean pulmonary artery pressure (
) of 25 mm Hg or greater, a pulmonary vascular resistance (PVR) of 3 or more Wood units, and a pulmonary arterial occlusion pressure of 15 mm Hg or less.
|
|
The next RCT (Table 1), the subcutaneous treprostinil trial (6), ostensibly a double-blind trial, was effectively unblinded due to the frequent occurrence of pain at the infusion site in the active treatment group. This trial added inclusion criteria seen in most subsequent trials—an age range, inclusion of patients with connective tissue and congenital heart disease—and excluded those with portopulmonary hypertension. It also included NYHA class II patients rather than only patients with class III and IV disease. Another novel feature was to include limits on the six-minute-walk distance, although these were broad. This was done to ensure that enrolled patients could walk, but not too far, thus increasing the likelihood of detecting differences in the main outcome variable.
The largest bosentan RCT, BREATHE 1 (Bosentan Randomized Trial of Endothelin Receptor Antagonist Therapy for Pulmonary Hypertension), enrolled NYHA class III and IV patients (7). However, the class IV patients were also "required to have a sufficiently stable clinical status to enable them to participate in a placebo-controlled trial." This proviso raises the question as to whether the NYHA class IV patients in this trial were comparable to those in other trials or even to those seen in everyday clinical practice. Patients with connective tissue disease were included, but not those with congenital heart disease. Patients had to have a minimal initial six-minute-walk distance of 150 m, slightly more than the earlier studies (Table 1), with a maximum of 450 m.
The aerosolized iloprost (AIR) trial (8), performed in Europe, was unique in that it enrolled not only patients with idiopathic and anorexigen-associated PAH, but also those with WHO group 3 chronic thromboembolic pulmonary hypertension, unlike other trials. It had a higher upper limit for the six-minute-walk distance (500 m) and slightly different inclusion criteria for hemodynamics. These included a relatively high minimal of greater than 30 mm Hg and a cardiac index (CI) between 1.5 and 4 L/minute/m2, although the requirement for a pulmonary arterial occlusion pressure of 15 mm Hg or less was similar to other trials.
More recently published RCTs (Table 1) include the beraprost study group trial (9), which failed to show a statistically significant improvement in the six-minute-walk distance after a year. This drug has not been submitted for approval to the FDA in the United States, but is commercially available in Japan. The trial included similar patients to those enrolled in the subcutaneous treprostinil trial, except that it excluded NYHA class IV patients as well as patients who had taken other PAH drugs within a month. Another RCT that failed to show statistically significant improvement in the main outcome variable was STRIDE 1 (Sitaxsentan to Relieve Impaired Exercise in Pulmonary Hypertension) for sitaxsentan (10), which used the change in maximal O2 uptake (MvO2) as determined by cardiopulmonary exercise testing instead of the six-minute-walk distance. The inclusion criteria reflected this choice, requiring an initial MvO2 between 25 and 75% of predicted. Even though the six-minute-walk distance significantly improved in this trial, the drug has not yet been approved by the FDA (although it has been approved in Europe, Canada, Australia, and New Zealand), partly because of failure to achieve a statistically significant benefit in the major outcome variable.
The SUPER 1 (Sildenafil Use in Pulmonary Arterial Hypertension) trial (11) had inclusion criteria similar to the subcutaneous treprostinil and BREATHE 1 trials, except that patients with any NYHA functional class disease were eligible and those with congenital heart disease had to have underwent surgical repair at least 5 years previously. The RCT for ambrisentan (12), an oral endothelin receptor antagonist that was approved by the FDA in June 2007, used eligibility criteria similar to those of the subcutaneous treprostinil trial except that it excluded congenital heart disease and included patients with HIV.
Other, more recent trials include those that have added new therapies to existing ones, termed "add on" trials (Table 1). The STEP (Iloprost Inhalation Solution Safety and Pilot Efficacy Trial in Combination with Bosentan for Evaluation in Pulmonary Arterial Hypertension) trial (13) added inhaled iloprost in patients who had received bosentan for at least 4 months. Inclusions and exclusions were similar to previous trials, except that patients had to have NYHA class III or IV disease and the upper limit for the six-minute-walk distance was slightly less (425 m) in an attempt to increase the likelihood of detecting a significant increase. The PACES trial (14) added sildenafil, 80 mg three times daily, to the regimen in patients who had been receiving intravenous epoprostenol for at least 3 months. Inclusion criteria were otherwise similar to those of most previous trials, except that patients with a history of anorexigen use and those with NYHA class II disease were specifically included. More recently, the EARLY (Endothelin Antagonist Trial in Mildly Symptomatic PAH Patients) trial (15) of bosentan limited inclusion to those with NYHA class II disease. Because the patients were less ill, a six-minute-walk distance up to 500 m or 80% of predicted was permissible as long as the exertional Borg score was greater than 2. To eliminate patients with other underlying lung pathologies, patients with a total lung capacity of less than 80% of predicted values were excluded.
In summary, the RCTs on current PAH drugs have been fairly consistent in targeting WHO group 1 patients; although the predominant subgroups represented are those with idiopathic PAH or PAH associated with either connective tissue disease or congenital heart disease. The studies have included patients with a broad range of ages, although most have excluded the very elderly and small children. Most have also excluded patients taking other PAH drugs. Because the six-minute-walk distance has been the most commonly used major outcome variable, most studies have set limits on admissible walk distances. More recently, the availability of effective PAH pharmacotherapies has led to the exclusion of NYHA class IV patients because of the ethical concern that these patients could be harmed by being enrolled in a placebo group. This has been accompanied by a trend to enroll less ill patients by limiting enrollment to patients with NYHA class II or III disease and with slightly greater six-minute-walk distances than in earlier trials. More recent trials have also begun adding therapy to preexisting PAH therapies partly to see if benefit can be enhanced by combination therapy and partly because this addresses some of the ethical concerns about placebo-controlled trials.
DEMOGRAPHICS OF PATIENTS IN PAH TRAILS
Although the inclusion and exclusion criteria determine the target population of a clinical study, ultimately the characteristics of patients actually enrolled into the study determine the relevant patient population. In the epoprostenol trial (5), patients were mainly women of late child-bearing age, 25% had NYHA class IV disease, and their six-minute-walk distance was severely limited (Table 3). Among the other earlier trials, females continued to predominate, but age increased slightly and six-minute-walk distance was slightly longer, reflecting the inclusion of NYHA class II patients in some of these trials (6). The subsequent studies also began including patients with connective tissue and congenital heart disease, broadening enrollment criteria. Despite the greater baseline six-minute-walk distance in the AIR trial compared with the epoprostenol trial, the proportion of NYHA class IV patients was actually greater in the AIR trial (41 vs. 28%). This may reflect the well-known inaccuracies inherent in both parameters (16) rather than any actual baseline difference between the groups.
|
Among the add-on studies, the STEP trial targeted NYHA class III and IV patients, but enrolled almost exclusively class III patients (Table 3). The PACES trial targeted patients of any NYHA functional class using intravenous epoprostenol, but similarly enrolled mainly patients with class III disease. The EARLY trial enrolled almost exclusively NYHA class II patients, consistent with the inclusion criteria. Six-minute-walk distances reflected these NYHA functional class distributions, with the STEP trial matching almost exactly the baseline six-minute-walk distance of BREATHE 1, which also enrolled mainly class III patients. The baseline six-minute-walk distance of the STEP trial was closer to those of the SUPER 1 and ARIES 1 trials, which had similar NYHA class distributions, and that of the EARLY trial was longer, consistent with its inclusion of only class II patients.
Ethnicity distribution was not reported in the epoprostenol or AIR trials, but in the others African Americans constituted 2 to 7%, Hispanics 5 to 11%, and Asian Americans 2 to 8% of patients. In the STEP trial, 3% of the patients were Native Americans. Although there are no published prevalence figures for PAH by race, a recent single-center report found that 15% of patients with PAH were African American (17). Coupled with the observation that ethnicity/race may influence access to tertiary medical care (18, 19) (and thus enrollment into clinical trials), these data raise the possibility that participants in PAH clinical trials may not reflect the ethnicity of the broader PAH population in the United States (or elsewhere).
Thus, as would be anticipated, demographics of patients enrolled into RCTs for PAH reflect the inclusion/exclusion criteria, but there are clear trends for enrolling less ill patients, with more NYHA class II and fewer class IV patients. This in part reflects the ethical concern that enrolling patients into placebo-controlled trials exposes them to potential harm should they end up in the placebo group, and the concern is greater with sicker patients. This also reveals a desire to determine whether pharmacotherapies can favorably affect the natural history of the disease if started in mildly affected patients. Minorities have been underrepresented in PAH trials thus far, perhaps because of the constitution of the populations at the participating centers.
BASELINE HEMODYNAMICS OF ENROLLEES IN RCTs FOR PAH
Baseline pulmonary hemodynamics afford another way to characterize patients entering PAH trials. However, inconsistencies in the reporting of hemodynamic variables limit comparisons between trials. Some studies report cardiac output and PVR and others report CI and PVR index (PVRI). In addition, fewer patients are undergoing hemodynamic evaluations at the time of enrollment because recent trials have focused on functional, not hemodynamic, end points.
In the intravenous epoprostenol trial, enrolled patients had severe hemodynamic derangement, with marked elevation of right atrial (RA) and , a relatively low CI and severe elevation of PVR (Table 4). The subsequent trials enrolled patients with milder hemodynamic abnormalities. Compared with the epoprostenol trial, RA and pulmonary arterial pressures were slightly lower and CI was slightly higher in both the treprostinil and bosentan trials, and in the BREATHE1 trial PVR was lower. The AIR trial did not report RA pressure and provided cardiac output rather than CI, but pulmonary arterial pressure and PVR elevations were similar to those in the BREATHE 1 trial. The similarities between these measures as well as in the six-minute-walk distance in the AIR and BREATHE 1 trials are remarkable considering that NYHA class IV patients constituted 41 and 10% of the enrollees in these trials, respectively. This raises questions about differences in the criteria used to determine NYHA functional class between the two trials, as well as subjectivity of assessment by both the investigators and subjects.
|
is in the same range as most of the earlier trials, but mean RA pressure has fallen to the upper normal range, and PVR values are slightly lower than on the earlier trials; these trends are consistent with the higher proportion of NYHA class II patients in these trials. In the STEP, PACES, and EARLY trials, baseline levels are still in the severe range, RA pressures are inconsistently reported, cardiac indices are in the normal range, and PVR values are again slightly lower than in earlier trials. These trends indicate less severe derangement of pulmonary hemodynamics in more recent trials than in earlier ones. STABILITY OF THE CONTROL GROUP
The stability over time of patients entering PAH RCTs is another important characteristic. Although stability has not been used as an inclusion criterion, one would expect that, as more NYHA class II patients enter trials, less deterioration over time would occur among control subjects. Also, investigators are more likely to place patients with a history of recent deterioration on approved, effective therapies than enroll them in RCTs, creating a bias toward enrolling more stable patients. This is borne out in Table 3, which shows the change in baseline six-minute-walk distance in the placebo group over the randomized portion of the trial. As can be seen, the placebo groups of earlier trials manifested greater deteriorations in six-minute-walk distance than more recent ones. Patients in the STEP and PACES trials were receiving background therapy, which may offer another explanation for their greater stability.
TRENDS OVER TIME
Figure 2 shows trends in enrollee characteristics over time. Age has been trending upward and the percentage of NYHA class II participants has been increasing. Six-minute-walk distances have increased on average, although this may also be influenced by restrictions, or lack thereof, on permissible walk distances for inclusion. Pulmonary hemodynamic derangements are becoming less severe, as mean RA pressure, , and PVR have been dropping. Once again, these trends undoubtedly reflect investigators' tendency to place sicker patients on known, effective therapies rather than in placebo-controlled RCTs.
|
Over the past dozen years, PAH clinical trials have demonstrated the efficacy and safety of six medications that have gained FDA approval, and many patients with PAH are reaping the benefits with improved functional capacity and slowing of disease progression. But what patients get into the trials and to whom do the results apply?
Enrollees in PAH clinical trials have been mainly white, with minorities being underrepresented. They have severe functional and pulmonary hemodynamic derangements, but these have been gradually getting milder over time. Earlier trials excluded patients receiving other approved PAH therapies, but more recent trials have included them and have added the new therapy to their regimen. Control groups are becoming progressively more stable. These observations are important in applying results to the larger population of individuals with pulmonary hypertension. First, even though enrollees have been confined largely to those in WHO group 1, there is considerable heterogeneity among enrollees, and one cannot assume that global efficacy from these trials necessarily applies to each subgroup of enrolled patients. Also, we should be cautious about extrapolating the results to patients in other WHO groups, even though PAH therapies are widely being given to such patients. Finally, because enrollment is confined to increasingly stable patients, study designers must be wary of relying too much on the rate of clinical worsening as a primary end point, considering that statistical significance for this end point depends on the number of events in the control group. With more healthy control subjects, studies will have to be larger and longer to have sufficient statistical power.
The pioneering studies discussed above have provided much valuable data and have served as the basis for approval for current PAH therapies, but future clinical trials must take into consideration some of the methodologic limitations. All trials should report recruitment fractions. Designers of trials should strive for consistency in describing diagnostic categories and reporting of NYHA class as well as hemodynamic variables. Efforts should be made to enroll as representative a population as possible, including diverse ethnic groups. As novel biomarkers and alternative end points are identified, these should be incorporated into trials, and can be used to better characterize the study population. Given the logistical challenges inherent in the performance of clinical trials, especially for rare diseases, investigators and clinicians need to pool their patient resources, perhaps in the form of clinical consortia or networks.
FOOTNOTES
Support for this conference, including travel for each of the authors, was provided by unrestricted educational grants from Actelion Pharmaceuticals, Pfizer, Gilead Sciences, United Therapeutics, and Lung Rx, Inc.
Conflict of Interest Statement: N.S.H. has participated as a speaker in scientific meetings or courses organized and financed by various pharmaceutical companies (Gilead, $2,000 in 2007; United Therapeutics, $3,000 in 2005–2007). He has received honoraria for participating in advisory boards (Actelion, $2,000 in 2007; Gilead, $2,000 in 2007 and $2,000 in 2008; United Therapeutics, $2,500 in 2007). He has received research grants for participating in multicenter clinical trials ($40,000 from Actelion, $100,000 from Gilead, $100,000 from United Therapeutics, $120,000 from Pfizer, $40,000 from Lilly-Icos). I.R.P. has participated as a speaker in scientific meetings or courses organized and financed by various pharmaceutical companies between 2005 and 2008 (Actelion, $3,000; Gilead, $5,000; United Therapeutics, $3,000). She has received honoraria for participating in advisory boards (from Actelion, $1,500 in 2006 and $1,500 in 2007; from Gilead, $2,000 in 2007 and $2,000 in 2008; from United Therapeutics, $1,500 in 2006 and $1,500 in 2007). She has received a research grant for multicenter clinical trials from Actelion ($20,000). K.E.R. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript.
(Received in original form March 28, 2008; accepted in final form May 29, 2008)
REFERENCES
| ||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
