| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Medicine, College of Physicians and Surgeons, Columbia University, New York, New York; 2 Department of Medicine, University of Alabama, Birmingham, Alabama; 3 Scottish Pulmonary Vascular Unit, Western Infirmary, Glasgow, United Kingdom; 4 Vera Moulton Wall Center for Pulmonary Vascular Disease, Department of Medicine, Stanford University School of Medicine, Stanford, California; 5 Department of Medicine, University of Colorado, Denver, Colorado; and 6 Department of Epidemiology, Joseph L. Mailman School of Public Health, Columbia University, New York, New York
Correspondence and requests for reprints should be addressed to Steven M. Kawut, M.D., M.S., Division of Pulmonary, Allergy, and Critical Care Medicine, Columbia University College of Physicians and Surgeons, PH 8, Room 101, 622 W. 168th Street, New York, NY 10032. E-mail: sk2097{at}columbia.edu
ABSTRACT
Pulmonary arterial hypertension is a rare and often devastating disease, although various effective therapies are now available. Clinical trials have used hemodynamic, cardiac imaging, laboratory, and exercise measurements as surrogate and intermediate end points in pulmonary arterial hypertension. Yet, based on the current literature, it is difficult to surmise which of these (if any) have been definitively validated. In addition, investigators have advocated the use of combined clinical end points in future clinical trials. The dependence of clinical trials and clinical management on such end points warrants a review of their use.
Pulmonary arterial hypertension (PAH) is a rare disease characterized by generally poor outcomes. In recent years, advances in our understanding of PAH have led to randomized clinical trials (RCTs) of novel treatments aimed at reducing morbidity and mortality. Regulatory agencies have approved several therapies and clinicians manage patients based on the results of these RCTs. The Food and Drug Modernization Act allows drug approval in serious diseases such as PAH when a given drug has shown an "effect on a surrogate end point that is reasonably likely, based on epidemiologic, therapeutic, pathophysiologic, or other evidence, to predict clinical benefit or on the basis of an effect on a clinical end point other than survival or irreversible mortality" (1). RCTs of therapies for PAH have therefore mainly focused on intermediate and surrogate end points as primary outcomes. Although a variety of candidate end points exist, none are unequivocally validated and data informing their use in clinical practice are lacking.
We will review the definition and criteria of valid surrogate end points, and several possible candidates, including hemodynamics, cardiac imaging, plasma biomarkers, and exercise, will be discussed. The advantages and disadvantages of composite clinical end points in PAH will also be detailed. Symptom scales and measures of quality of life are addressed elsewhere in this issue (2).
DEFINITIONS AND CRITERIA
The ultimate clinical end point in PAH is survival. Other important end points include hospitalization for PAH-related events, transition to parenteral prostacyclin analogs, atrial septostomy, and lung transplantation. Although definitive in their implications, these end points are often difficult to study in PAH, because many take years to occur and some may be postponed or avoided altogether by institution of salvage therapy. Ideally, a surrogate end point serves as a substitute for these clinically meaningful end points and may be more easily studied during a clinical trial, and ultimately used to manage patients. Alternatively, an intermediate end point is a metric or event that itself is important to a patient, whether or not it is associated with long-term outcome. For example, New York Heart Association (NYHA) functional class is itself likely to be important to patients in terms of the ability to perform activities of daily living (intermediate end point) but may or may not predict long-term morbidity or mortality (surrogate end point).
A surrogate end point must meet certain criteria. First, a surrogate must be reliable. The measure must not be highly variable within a patient and must demonstrate intra- and interobserver reproducibility. Without adequate reliability, an end point cannot be valid. Second, a surrogate is ideally integral to the causal pathway of the disease, and the intervention being tested should act on the disease pathway that is represented by the surrogate (3). For example, a plasma biomarker of endothelial dysfunction may serve as an appropriate surrogate for a drug targeting the pulmonary vasculature but not, perhaps, for a drug that improves right ventricular (RV) function. In this instance, improved outcomes due to RV remodeling might not be reflected by changes in the biomarker. Third, ample epidemiologic evidence should link the surrogate end point to outcome. Multiple studies in distinct populations are necessary to meet this criterion. Fourth (and most important), changes in the probability of reaching clinical end points effected by therapy should be reliably reflected by concomitant changes in the surrogate marker (3). This requires adequately powered RCTs of interventions administered for sufficient duration to impact on ultimate clinical outcomes, which have been difficult to achieve in PAH (4–9).
POTENTIAL SURROGATE END POINTS IN PAH
Hemodynamics
The use of hemodynamic measurements as surrogate end points is appealing given (1) the apparent standardization of such measures, (2) the causal pathway of PAH, and (3) epidemiologic evidence showing consistent associations between hemodynamics (specifically heart function) and survival (10–15). However, the relationships between hemodynamic changes, treatment, and clinical outcomes have proven less predictable in the era of treatment for PAH (15–21).
Recent studies show that there may be significant variability in the conditions and techniques by which hemodynamic testing is performed (22). For example, some centers measure cardiac output by the Fick method, whereas others use thermodilution. Even in the most rigorous settings (23), there may be a difference of 1.0 L/minute between the two techniques. Variability in the measurement of oxygen consumption and agents used for acute vasoreactivity testing may create further within- or between-center differences. A standardized approach for cardiac catheterization in patients with PAH would minimize these sources of error.
It is well established that significant cardiac dysfunction and more severe PAH confer a higher risk of death in patients with PAH (15, 17, 20, 24). McLaughlin and colleagues (15) demonstrated that improvement in cardiac index and reduction in mean pulmonary artery pressure after epoprostenol predicted better survival. Similarly, persistently elevated pulmonary vascular resistance (PVR) after several months of epoprostenol (16) or bosentan (19) is associated with an increased risk of death.
Hemodynamic measurement has been validated against survival in a single RCT (25). In a 12-week RCT of intravenous epoprostenol compared with conventional therapy, Barst and colleagues (25) demonstrated significant improvements in pulmonary arterial pressure, cardiac index, and PVR, as well as increased survival in the treatment group. Unfortunately, follow-up RCTs of prostacyclin analogs have shown less robust hemodynamic effects and no significant impact on lung transplantation or survival (26–28). Similarly, clinical trials of endothelin receptor antagonists and sildenafil have demonstrated promising improvements in hemodynamic parameters, but such improvements have not translated to survival or other ultimate clinical end points (29–31). Whether the statistically significant, but relatively small, hemodynamic effects of these PAH therapies will reliably translate to clinically meaningful outcomes in the long term is unknown.
Because many of these trials were of brief duration or underpowered to detect differences in survival or other "hard" end points, the validity of hemodynamic surrogates in PAH remains unclear. The use of inotropes in congestive heart failure, supported by early work demonstrating favorable effects on hemodynamics, actually increased mortality, providing a valuable lesson in careful end point selection (32, 33). Accordingly, the U.S. Food and Drug Administration does not accept hemodynamic measures as adequate surrogates of outcome for phase III RCTs in PAH. In addition, the value of a single set of measurements taken at rest is unlikely to fully capture disease severity, which may better be assessed during usual activity or exercise. The invasiveness of cardiac catheterization reduces the attractiveness of hemodynamic end points for RCTs and routine clinical care.
Plasma Biomarkers
Plasma biomarkers are commonly used in many cardiovascular diseases, but are less well studied in PAH. The potential for serial measurement with a relatively noninvasive methodology (e.g., phlebotomy) makes circulating biomarkers potentially attractive surrogates. Such biomarkers may be recognized as sentinel (those that may reflect the early development or provide prognosticators of a disease process) or integral (those that are of known pathophysiologic relevance to the disease condition and that are potential drug targets).
Natriuretic peptides are produced by the atria and ventricles in response to myocyte stretch and reflect neurohormonal activation in heart failure. Circulating levels of brain natriuretic peptide (BNP) and N-terminal pro-BNP are increased in PAH and are correlated with disease progression (34, 35). Higher levels of BNP and N-terminal pro-BNP also predict mortality in PAH (20, 36). In a small RCT comparing the addition of sildenafil or bosentan to conventional therapy, BNP levels significantly decreased in the sildenafil arm, but neither BNP levels nor other clinical end points differed across the two drug groups (37).
Endothelial injury and dysfunction are also integral to the causal pathway of PAH. Von Willebrand factor, endothelin-1, eicosanoid metabolites, and other markers have been linked to the disease mechanism and, in certain cases prognosis, of PAH (24, 38–42). A given surrogate biomarker may vary within and across patients and may be directly affected by drug therapy, weakening the association with outcomes. Going forward, biomarkers must be routinely incorporated into RCTs and their relationship to clinically meaningful end points investigated before their validity as surrogate end points can be confirmed.
Cardiac Imaging
Much of the morbidity and mortality in PAH occurs as a result of RV failure. It is therefore plausible that imaging of cardiac structural changes may provide useful surrogate end points. Such morphologic changes are more likely to reflect chronic consequences of the disease process and may be more reliable than a highly dynamic surrogate, such as hemodynamics. Echocardiography, radionuclide angiography, and magnetic resonance imaging (MRI) are the three best-studied imaging techniques to date.
Various echocardiographic parameters have been associated with outcome in PAH (14, 20, 21, 43–45). The size of the right atrium and degree of diastolic septal shift, or the eccentricity index, predicts progression to death or transplantation (43). Similarly, the Doppler right ventricular index (i.e., Tei index) globally assesses RV function and has been independently associated with poor outcome (44). Tricuspid annular plane systolic excursion (TAPSE) is a reproducible measure of RV function that tracks with other echocardiographic and hemodynamic surrogates and is an independent predictor of survival in PAH (45). The presence and larger size of a pericardial effusion are associated with worse survival in PAH (14, 20, 43, 45, 46). It is worthwhile noting that estimation of RV systolic pressure by echocardiogram fails to predict outcomes in patients with PAH (14, 43, 44).
Two studies have suggested that echocardiography may provide useful surrogate end points. In a subgroup of patients from the phase III RCT comparing bosentan with placebo, significant improvements in echocardiographically determined RV and left ventricular size and function and inferior vena cava diameter were shown in the active therapy group (21). RV size, eccentricity index, and maximum tricuspid regurgitant jet velocity were significantly improved in the 12-week RCT of intravenous epoprostenol that also demonstrated a survival benefit (47). Although these results are promising, the usefulness of echocardiography in assessing long-term response to treatment as well as acute changes in the clinical status of patients with PAH is unknown. Specifically, the reliability of echocardiographic parameters may be undermined by patient habitus and inter- and intraoperator and interpreter variability.
Other Cardiac Imaging Modalities
Radionuclide angiography and MRI are able to accurately measure the RV ejection fraction and assess changes in ventricular mass, dimensions, and wall motion (48–51). In a retrospective cohort study of patients with PAH, RV ejection fraction as measured by radionuclide angiography was an independent predictor of death (24). In a recent prospective study of 64 patients with PAH (50), several MRI-derived indices of RV size and function, including stroke volume index and RV end-diastolic volume index, were associated with survival. Increases in stroke volume index and reductions in RV end-diastolic volume index over 1 year of treatment were associated with better survival independent of indexed PVR at baseline or over time (50).
MRI measures of RV diastolic dysfunction have been correlated with other markers of disease severity and improved with sildenafil therapy (51). Wilkins and colleagues (37) examined changes in RV mass by MRI in their small study comparing bosentan and sildenafil. Although no differences existed between the study groups, a decrease in RV mass was demonstrated in both groups and reached statistical significance in the sildenafil group at 16 weeks. The accuracy of MRI for interrogating RV morphology and function may ultimately establish this modality as an important source of surrogate end points. However, the expense, time needed, and technical feasibility remain limitations to wide-scale adoption of this technique at this time.
Exercise Testing
Six-minute-walk test.
The six-minute-walk test (6MWT) is frequently used in clinical practice to assess progression of disease and response to therapy in PAH. As an intermediate end point, the distance walked may translate to the ability to perform activities of daily living, exercise capacity, and quality of life in patients with PAH (52). The test is easily performed, low cost, and is reasonably reliable (53). The validity of the 6MWT may be affected by patient effort, learning with repeat testing, musculoskeletal fatigue, disease state, and characteristic side effects of certain therapies (i.e., leg pain with parenteral prostacyclin analogs).
As in other cardiopulmonary conditions, higher exercise capacity may predict better survival in PAH. Baseline distance from the 6MWT (six-minute-walk distance [6MWD]) has been associated with cardiac function (54, 55) as well as survival in a recent meta-analysis (56). In addition, improvements in 6MWD after treatment are associated with increases in stroke volume and heart rate response and decreases in right atrial pressure (55). It follows that the assessment of cardiac function via exercise testing may serve as a surrogate for true clinical outcomes. In a study evaluating a variety of potential surrogates in PAH, only the 6MWT was associated with mortality (54). This finding has been documented in numerous other studies (15, 16, 19–21, 42, 43).
Sitbon and colleagues (16) reported that a 6MWD of less than 250 m at baseline and a 6MWD of less than 380 m after 3 months of epoprostenol were associated with an increased risk of death. In this same study, the change in the 6MWD did not track with outcome. In a retrospective review of patients with PAH treated with bosentan as first-line therapy, baseline 6MWD as well as absolute and incremental increase in 6MWD after 4 months of therapy were associated with survival (19).
In an RCT of epoprostenol, Barst and colleagues demonstrated an increased 6MWD from baseline in the active therapy arm as well as improved survival (25). Follow-up clinical trials of prostacyclin analogs, endothelin receptor antagonists, and phosphodiesterase-5 inhibitors have demonstrated significant increases in 6MWD but no differences in the need for lung transplantation or survival (26, 27, 29–31). A recent meta-analysis of several RCTs in PAH failed to find an association between the placebo-corrected differences in the change in 6MWD and the effect estimates for time to clinical end points (56). Therefore, the magnitude of this often-touted result of RCTs in PAH may not have a predictable relationship with long-term outcome.
These data suggest that 6MWD may be a surrogate (and likely an intermediate) end point in PAH, but questions remain. There is no consensus on whether absolute, incremental, or threshold values of 6MWD are equally (or individually) relevant. Like any physiologic disease marker, the 6MWD may have less room to improve in healthier patients with PAH (i.e., the "ceiling effect"). Moreover, unlike all other measurements of pulmonary function, 6MWD is not currently routinely indexed to height, weight, race, or sex (and it is not clear that it should be). Finally, the use of the 6MWD as an inclusion or exclusion criteria in a given clinical trial may limit its value as a surrogate, by limiting both the distribution of the values and generalizability to clinical practice.
Cardiopulmonary exercise testing.
Cardiopulmonary exercise testing (CPET) measures an individual's oxygen consumption at peak exercise (
O2max) and generates a variety of parameters to help delineate the contribution of different bodily systems to exercise limitation. Highly calibrated equipment is required for standardization and the test is subject to patients' ability and willingness to exercise maximally on a cycle ergometer or treadmill. It is biologically plausible that such a global assessment of human performance would strongly correlate with outcomes in disease states such as PAH. The relationship between O2max and outcome has been validated and CPET has been used as a surrogate end point in other cardiopulmonary diseases (57).
There is some evidence to suggest that CPET is reliable in PAH (58). O2max and other measurements of ventilatory efficiency and carbon dioxide production are abnormal in PAH and have been correlated with a variety of hemodynamic measurements (59, 60). Specifically, ventilatory equivalents for oxygen and carbon dioxide have been correlated with pulmonary arterial pressure during exercise (61). Two studies have demonstrated a significant relationship between O2max and event-free survival in PAH (17, 62). In another retrospective cohort of patients with PAH, O2max did not predict outcome (24). Lower systolic blood pressure and higher ventilatory equivalent for carbon dioxide at peak exercise, however, were significantly associated with an increased risk of death.
Change in O2max was the primary outcome variable in a single RCT comparing 12 weeks of two doses of sitaxsentan with placebo (29). Although the high-dose sitaxsentan group had significant improvement in O2max compared with placebo, there was no treatment effect seen in other CPET parameters or in the other active treatment arm. Interestingly, 6MWT and hemodynamic parameters, both secondary outcomes in the trial, were improved over the course of therapy. This discrepancy may be partially explained by intrapatient variability, as demonstrated by improved correlation of CPET parameters with 6MWD over each patient's course during the clinical trial (63). From these data, it appears that no definitive CPET parameter has been validated as a surrogate. It is also unknown whether other types of maximal exercise testing may serve this purpose (15, 64).
Combined End Points
Clearly, no single surrogate has proven ideal in therapeutic PAH trials. These trials have been of brief duration with small sample sizes and, in many cases, low rates of definitive clinical events. The use of a composite or a combined end point in future PAH trials (e.g., time to clinical worsening) has thus been proposed.
Derived from a combination of individual end points, a composite end point is appealing for several reasons (65). First, precision (and therefore statistical power) increases with event rate. Combining the events of death, lung transplantation, institution of prostacyclin analogs, and hospitalization as a single outcome in a PAH trial, for example, would make it easier to detect a therapeutic benefit compared with analyzing each event separately without requiring an increase in sample size. This is because a higher number of events provides more power to detect any given effect size. Second, nonfatal but important morbid events in the course of disease may be incorporated, offering a more global assessment of the patient and his or her clinical state.
However, the use of combined end points in clinical trials is not without issue. An individual (nonfatal) end point may confound the composite and other, more clinically relevant, individual outcomes (i.e., mortality), especially when occurring at different rates during a trial of insufficient duration (66, 67). Confounding may lead to individual and composite outcomes that are falsely diluted and/or compounded (68–70). For example, if the first hospitalization during a patient's course is chosen as an end point, an admission for a PAH-unrelated event may "mask" more significant, future events such as institution of a prostacyclin analog or hospitalization for a PAH-related event (70). An increased rate of hospitalization in one arm of a trial could balance an increased mortality in the other, leading to an unwarranted "null" conclusion.
Incorporating an outcome such as hospitalization in a combined end point is also problematic in that it is somewhat driven by social and other nonmedical factors, which may disproportionately influence a composite also containing more direct measures of disease progression (e.g., death) (69). A composite outcome driven by individual end points with center-specific availability (i.e., lung transplantation and atrial septostomy) may pose difficulty in multicenter trials. The inclusion of volitional end points with unclear implications to the patient (e.g., an arbitrarily chosen decrement in 6MWD) in a combined end point may actually thwart the goal of constructing a "clinically relevant" composite. Last, a combined end point assumes that each of the component end points has equal implications to the patient and the physician. Clearly, the addition of a new PAH medication (one suggested component of a composite end point for PAH) is much different than death; however, each would be equally weighted in a composite end point.
The combined end point of "time to clinical worsening" has been examined as a secondary outcome variable in several RCTs in PAH. In the RCT of sitaxsetan, time to clinical worsening (defined as death, epoprostenol use, atrial septostomy, or transplantation) occurred only rarely (5% in the placebo, 0% in the 100-mg, and 2% in the 300-mg group) in this study of 178 patients (29). A similarly defined secondary outcome was examined in the 16-week RCT of bosentan (71). Active therapy significantly increased the time to clinical worsening, with each component of the composite end point occurring more frequently in the placebo arm of the trial. There was no significant difference noted in the time to clinical worsening (which also included the individual outcome of use of additional therapeutic oral and/or intravenous agents) in an RCT of sildenafil (30). The combined end point of death, transplantation, epoprostenol rescue, or greater than 25% decrease in peak oxygen consumption was chosen as the primary outcome variable in the 12-month RCT of beraprost (28). Although a greater rate of individual events occurred in the placebo group throughout the study period, a significant treatment effect was demonstrated only at the 6-month time point. Should a combined end point such as time to clinical worsening be incorporated in future PAH trials, individual end points should be clinically relevant and standardized across clinical trials.
CONCLUSIONS
The ability to reliably predict long-term response to therapy in PAH using surrogate end points is crucial for both scientific advancement and management of this fragile patient population. Unfortunately, the majority of RCTs thus far have not collected adequate surrogate data and have failed to demonstrate therapeutic gain in terms of definitive end points.
There is good epidemiologic evidence supporting the use of hemodynamic parameters as surrogate end points in PAH. Future work needs to address which of these parameters are relevant to outcome, how and when hemodynamics should be assessed, and whether improvement in hemodynamic parameters should be sufficient for drug approval by regulatory boards. Echocardiography and BNP are widely available, noninvasive means of assessing RV dysfunction and, at face value, seem to be logical surrogates for predicting outcome. Neither measure, however, is sufficiently validated at this point. Specific echocardiographic indices and BNP-guided treatment strategies need to be tested in future RCTs before recommendations for clinical practice can be made. Although validity has yet to be established, MRI may prove an attractive alternative to assess RV function as the technology becomes more widely available.
Perhaps the most studied and corroborated surrogate in PAH, the 6MWT is a seemingly straightforward and "low-tech" end point. Whether incremental change, absolute distance, and/or threshold in the 6MWD will prove to be clinically relevant remains to be seen. The 6MWT needs to be confirmed as a surrogate for survival in PAH and ideally studied in an unselected patient population alongside non–effort-dependent variables. The use of CPET as a primary outcome assessment should be avoided pending further confirmation of the test's validity in PAH and epidemiologic evidence correlating specific indices to PAH outcomes. Although statistically appealing, composite end points must be carefully designed, diligently applied, and accurately reported going forward. Future sponsors and trialists in PAH have an important opportunity to clarify many of the uncertainties outlined here. Larger, longer RCTs incorporating these and other potential surrogates will ultimately define valid end points, allowing us to design "smarter" clinical trials, which will lead to greater impact on outcomes in this still incurable disease.
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: C.E.V. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. R.L.B. has been reimbursed $6,000 as an honorarium for speaking for Actelion, and $2,000 as an honorarium from Gilead and United Therapeutics each. He received $200,000 from Actelion, Gilead, United Therapeutics, Lung Rx, and Encysive in multicentered clinical trials. A.J.P. has received lecture fees, consultancy fees, and assistance with meeting attendance from GlaxoSmithKline, Pfizer, Actelion, Encysive, and Myogen, and has received an unrestricted educational grant from Encysive for £60,000. R.T.Z. is a recipient of an Actelion-Entelligence Career development grant for $75,000, is on the speakers' bureau for United Therapeutics and Actelion, and is a consultant to Gilead Sciences. D.B.B. has commercial/industry relationships that are not directly relevant to the content of this paper. S.M.K. has received lecture fees, consultancy fees, and/or other support from Actelion, Pfizer, Gilead, Encysive, INO Therapeutics, CoTherix, United Therapeutics, Gerson Lehrman, and Clinical Advisors. He has received a $50,000 grant from Pfizer for an investigator-initiated clinical trial.
(Received in original form March 27, 2008; accepted in final form May 16, 2008)
REFERENCES
ska L. Detectable serum cardiac troponin T as a marker of poor prognosis among patients with chronic precapillary pulmonary hypertension. Circulation 2002;108:844–848.[CrossRef]Related articles in Proceedings of the American Thoracic Society:
This article has been cited by other articles:
|
|
 |
|
  S. D. Halpern, R. Doyle, and S. M. Kawut The Ethics of Randomized Clinical Trials in Pulmonary Arterial Hypertension Proceedings of the ATS, July 15, 2008; 5(5): 631 - 635. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
