Pulmonary Rehabilitation and Clinical Trial Design in Patients with Severe COPD

Abstract

Pulmonary rehabilitation confers significant benefit to patients with COPD. The beneficial effects are seen in dyspnea and health-related quality of life. These measures are often considered in clinical trials as study endpoints. Pulmonary rehabilitation may affect the study population and study enrollment in clinical trials of other therapies in COPD, so it is important to anticipate the effects of pulmonary rehabilitation when designing and performing clinical trials. We review pulmonary rehabilitation and clinical trial design in patients with severe COPD. No publications were found that specifically address the impact of pulmonary rehabilitation on study design. No consensus statements, guidelines, or pharmaceutical studies in COPD were found that addressed pulmonary rehabilitation and study design. Pre-operative pulmonary rehabilitation is almost universal with lung transplantation and lung volume reduction surgery (LVRS). The National Emphysema Treatment Trial (NETT) included a mandatory pre-operative pulmonary rehabilitation program. The NETT reported that the mandatory pre-operative program reduced their study population by 32%. The NETT pre-operative program is required for Medicare reimbursement for LVRS, restricting the surgery to the subgroup of patients completing the program. Pulmonary rehabilitation can affect selection of the study population, study enrollment, outcome measures, and influence whether the results can be generalized to other patients. Clinical trials with medications, devices, or surgery for severe COPD should consider study designs that anticipate the significant effects of pulmonary rehabilitation.

Key words: :

• Rehabilitation
• Pulmonary Disease (Chronic Obstructive)
• Therapeutic Human Experimentation

INTRODUCTION

Clinical study designs have to account for factors that can confound the study group selected and the results of the clinical trial. Pulmonary rehabilitation is an important development in the care of patients with chronic lung diseases, particularly COPD, and significantly improves many symptoms of the disease. Thus, whether study participants have received, or are in the process of receiving pulmonary rehabilitation would be expected to be an important consideration in interpreting clinical study results. We noted that rehabilitation has often been required in surgical trials immediately before major pulmonary surgery for severe COPD but not in pharmaceutical studies with similar patient populations. The role of pulmonary rehabilitation in studies of investigational medical device treatments of severe COPD has not been established. A review of the impact of pulmonary rehabilitation in clinical study design for patients with COPD is the goal of this paper.

The focus of this review is largely on pulmonary rehabilitation with COPD and severe emphysema since this area has seen clinical advances with both lung transplantation and lung volume reduction surgery (LVRS) in recent decades. Also, the National Emphysema Treatment Trial (NETT) was a landmark surgical study of LVRS and had mandatory pulmonary rehabilitation prior to the procedure ([1]). This review is not intended to be comprehensive for either pulmonary rehabilitation or COPD clinical study design as separate entities.

LUNG SURGERY AND PULMONARY REHABILITATION

Most lung transplantation programs have followed the program established by Cooper and the Toronto Lung Transplant Group and enroll patients in ongoing pulmonary rehabilitation while the patient awaits assignment of an organ ([2]). Similarly, a key feature of the modern era of LVRS initiated in the 1990s by Cooper et al. ([3]) at Washington University was a requirement to complete a program of pulmonary rehabilitation. Their pulmonary rehabilitation program for LVRS used a comprehensive multi-disciplinary team, prescribed treadmill exercise 5–7 days a week and aimed at 30 continuous minutes of exercise at target heart rate ([4]). Whether post-operative complications are reduced by pre-operative pulmonary rehabilitation has not been studied and rehabilitation was not uniformly applied at all sites performing LVRS ([5]) before initiation of the NETT. The NETT required a mandatory pre-operative pulmonary rehabilitation program with 18–20 sessions over 6–8 weeks in their study design. The NETT rationale was to facilitate early postoperative mobilization and to provide a baseline of optimized preoperative exercise capacity for comparison with postoperative exercise capacity ([1]). Surgery was compared to maximal medical therapy which also included pulmonary rehabilitation and there were additional goals of pulmonary rehabilitation such as assurance that the subjects provided informed consent for an invasive surgical procedure from the best possible “baseline” condition ([1]).

MEDICAL DEVICE STUDIES AND PULMONARY REHABILITATION

In the late 1990s many investigators started to develop non-surgical devices or methods to treat severe emphysema. Some of these non-surgical approaches that have advanced into clinical trails are bronchial valves ([6], [7]), airway bypass ([8]), and biological airway sealant methods ([9]). All use bronchoscopic approaches for the procedures and have shown lower rates of post-procedure complications and mortality compared to LVRS, so the rationale for pre-procedure pulmonary rehabilitation to reduce complications is weaker. In some of these initial trials, pre-procedure pulmonary rehabilitation has been incorporated ([7], [8]), in others it has not ([6], [9]). The stated rationale for including pulmonary rehabilitation in one trial was to facilitate comparison of results with LVRS as performed in the NETT trial ([7]), although another trial included a run-in period without required pulmonary rehabilitation using a similar rationale ([9]).

PULMONARY REHABILITATION

Pulmonary rehabilitation has significant effects on patients with COPD and recommended standards have been published ([10]). Completion of a pulmonary rehabilitation program for COPD clearly reduces dyspnea and improves health-related quality of life (HRQL). There is also evidence for improved muscle function and a reduction in health care costs ([10], [11], [12]). Consideration of pulmonary rehabilitation is recommended for all COPD patients with symptomatic disease and a GOLD grade II severity or greater ([10], [11], [12]). However, programs providing pulmonary rehabilitation are not widely available, most likely because of non-coverage decisions by insurance payers, so the estimate for the fraction of patients that have received pulmonary rehabilitation that would potentially benefit is low, likely less than 10% in the United States ([13]). Another problem for incorporating pulmonary rehabilitation into clinical trials is the travel and coordination required between the centers providing the clinical trial and procedure and the facility administering the pulmonary rehabilitation. For example, the NETT enlisted and certified over 539 satellite sites to provide pulmonary rehabilitation closer to the subjects' homes ([14]).

Onset of benefit and peak response

The onset of benefit with pulmonary rehabilitation is variable among individuals, but it may be seen as early as a few weeks after initiation ([10]). Variability of response to pulmonary rehabilitation presents an additional challenge for clinical trial design. Many studies have found that the standard deviation (SD) among subjects of outcome measures increases after a pulmonary rehabilitation program ([15], [16]). In the Griffiths trial ([15]) the coefficient of variation (SD expressed as percentage of the mean) of the SGRQ rose from 19.7% at baseline to 26.8% at the end of the program and 28.5% at the end of 1 year – a relative increase in variation of 45%. This addition of variance to the measure may prevent observing a true change or at least result in a larger sample size being needed.

The benefits of pulmonary rehabilitation increase with the duration of the program and a peak response time is not predictable. Studies have shown increasing benefit after continuous programs for 6 and 12 months ([17], [18]). But between patient differences appear to be time dependent. The data for 6-minute walking distance showed a different pattern of change in coefficient of variation to that seen with the SGRQ: baseline 67%, post-rehab 56%, 1 year 70%. This suggests that rehabilitation standardizes the patients' state measured in terms of exercise capacity for a period post-rehabilitation, but thereafter differences between patients emerge again. This presents a problem for study design since waiting for an extended period after a subject is eligible for an experimental procedure or surgery is not desirable.

Previous pulmonary rehabilitation

Patients that have already had pulmonary rehabilitation may present for a clinical study. Should such subjects be required to repeat a rehabilitation program? In the NETT, 64% of the 1,218 patients at initial evaluation had received prior pulmonary rehabilitation. Prior pulmonary rehabilitation reduced the effects of pulmonary rehabilitation in NETT ([14]). For example, repeating pulmonary rehabilitation had only a small effect on health-related quality of life: an improvement in the total SGRQ score of −2.6, which is less than the accepted minimal clinically important difference ([14]). Other improvements with repeating rehabilitation were also modest with exercise (cycle ergometry) improving 3.1 W and 6-min walk distance 76 feet. It is also noteworthy that changes after rehabilitation did not predict differential mortality or improvement in exercise (primary outcomes) by NETT treatment group ([14]).

Maintenance and decline of pulmonary rehabilitation benefits

If completion of a pulmonary rehabilitation program is considered mandatory before patients are randomized to receive an intervention in a trial of another therapy, and it affects the study endpoints, then continued adherence to an ongoing exercise program (maintenance) is necessary to maintain the gains achieved. Long-term studies of pulmonary rehabilitation have demonstrated declines in outcome measures ([15], [16], [17], [18]). However, the maintenance phase of pulmonary rehabilitation is usually poorly controlled and studies have largely relied on self-reporting to assess adherence ([16]). This lack of supervised maintenance is presumably because of the burden of continuing a prolonged program for the subject, study center and study sponsor.

A requirement for all study candidates to participate in a PR program before entering a clinical trial will bring subjects closer to maximum in many measures but these measures decline again after completion of pulmonary rehabilitation ([18]). In the NETT trial, only 20% of the medical therapy group did not have a decline in their exercise capacity after 6 months ([20]). A similar decline was found in the walk distance after 6 months (20, Supplementary Appendix 4). Such declines may differ between subjects, resulting in increased variability in the clinical study outcome measures, thereby reducing the power of the study to detect a pre-specified difference between treatment arms. In addition, mandatory pulmonary rehabilitation may temporarily inflate endpoint measures so that, 1) if the intervention is after pulmonary rehabilitation, the therapeutic difference between treatment and control may be due to deterioration of the control group rather than improvement in the treatment group or, 2) a true treatment effect might be overlooked because the therapeutic difference is largely attributed to the deterioration of the control arm of the study.

STUDY DESIGN

Study population selection

Requiring participation in a program like pulmonary rehabilitation can have effects on the study population. In the NETT, 32% of the enrolled patients did not return for randomization after the pre-operative pulmonary rehabilitation program ([14]). Another study found 40% or 129 of 297 suitable patients declined participation in a trial that required pulmonary rehabilitation ([21]). These trials therefore studied a subgroup that may not be representative of the entirety of the disease population under study.

Table 1 Effects of pulmonary rehabilitation on clinical research study design in patients with COPD.

Aspect of Study Design	Effect of Pulmonary Rehabilitation
Study Population	Selected population
	Adds variation to HRQL
	May standardize the population
Subject Enrollment	Decreased enrollment
Morbidity of a Surgical Procedure	Not known
Safety of a Minimally Invasive Procedure	Not known
Study Endpoints
•Dyspnea Scales	•Improvement
•HRQL	•Improvement
•PFT	•No effect
•Exercise Tests	•Improvement

In addition, the NETT found, in patients with upper-lobe predominant emphysema, the greatest differences between medical and surgical results were between those patients that had low cycle ergometry results after pulmonary rehabilitation (high versus low exercise groups). The increased difference between medical and surgical results is because of the poor medical outcome for this subgroup of patients, rather than a surgical difference compared to the other subgroups ([20]). In later follow-up of NETT subjects, low exercise after pulmonary rehabilitation remained a predictor for improved survival with surgery, but was not a predictor of major pulmonary and cardiac morbidity, or 90 day mortality after LVRS ([22]).

Thus, if a pulmonary rehabilitation program is a requirement before a clinical study, then the study results will be applicable to only those patients that can complete rehabilitation and not to the broader population of COPD patients. This broader population may have declined pulmonary rehabilitation because of access, cost, transportation, or other practical reasons.

Study population preparation

Optimization or revision of patient management before entry into a study is often part of study design. This is reasonable because an existing and approved treatment may obviate the need for a research treatment. Optimal management before a research intervention is also good study design so that treatment is not added during the study period. An added treatment during the trial could modify the estimated treatment effect for the therapy under study. Any treatment added during the start of a trial needs careful consideration for possible effects on the study. These would include onset of action, side effects, continuation of treatment, availability of treatment, and interactions with the research intervention or outcome measures.

Table 2 Summary points regarding clinical trial design and pulmonary rehabilitation in patients with COPD.

•Pulmonary rehabilitation has significant effects on common endpoints in COPD trials so this variable needs to be addressed in study design.

•Effects from pulmonary rehabilitation should be approaching a stable state before adding the research intervention or agent to be studied.

•Minimally-invasive procedures should not share with major surgery the logic that prior pulmonary rehabilitation may reduce post-operative morbidity.

•A study requirement for pulmonary rehabilitation can affect the study population enrolled and limit the generalization of the study results.

Randomization and stratification

Randomization can be expected to reduce bias between study groups regarding the effects of pulmonary rehabilitation. However, if the study endpoints are strongly affected by pulmonary rehabilitation, then a stratification scheme may be appropriate to ensure a more balanced randomization. A study using a dyspnea, exercise performance, or a HRQL measure for an endpoint may be an example. Another is when the study population is anticipated to have a significant incidence of previous pulmonary rehabilitation, particularly in the prior few months, since this significantly blunted the effect of the mandatory pre-operative program in the NETT ([14]). For example, the SGRQ improved −5.1 with no prior pulmonary rehabilitation compared to −2.6 (p < 0.01) if the patient had prior pulmonary rehabilitation.

Factorial design

A factorial design, where there is randomization for pulmonary rehabilitation followed by randomization to the primary intervention should be considered with caution, since there may be high drop-out in the group randomized to pulmonary rehabilitation. If the drop-out following rehabilitation is similar to the 32% observed in the NETT, this would adversely impact one arm, create imbalance, and may seriously compromise the validity of the entire study.

Uniform starting point and run-in period

It is an ideal to have all subjects in a uniform and stable state at the start of a study. All known variables relevant to the expected response should be in a steady state prior to randomization to study treatment. COPD patients have multiple medications available and because of variable responses, side-effects, availability and costs these medications need to be “tailored” to each patient by an experienced physician. In addition, exercise patterns and nutritional status should be in a steady state. Comprehensive pulmonary rehabilitation programs address these factors to a degree. Exercise is the best studied component of pulmonary rehabilitation, but some investigators have begun to address other factors. Depression has been reported to be a significant risk factor for drop-out compared to non-depressed patients ([23]). Nutritional interventions such as dietary supplements and anabolic steroids have also been studied as part of pulmonary rehabilitation ([24], [25]) and may add more variability to the outcomes of a pulmonary rehabilitation program. These may prove to be important interventions, but will add more complexity to study designs that involve pulmonary rehabilitation.

A run-in period several months after pulmonary rehabilitation and before an intervention can help establish a steady state regarding many of these variables. The run-in period can allow any medication changes to stabilize and help establish a stable pattern for exercise and nutrition. A run-in period will also help mitigate the effects of a COPD exacerbation on dyspnea, exercise and HRQL. It takes at least 12 weeks for most patients to return to a steady state level after a COPD exacerbation ([26]). In the NETT a hospitalization within 3 months had a large effect on the mean SGRQ: 57.9 with an admission compared to 52.2 without an admission (p = 0.001).

SUMMARY

Pulmonary rehabilitation presents many challenges for clinical study design of other therapies. Logically, it should be given before lung transplantation and LVRS to reduce the morbidity of major surgery, although the case for this is not proven. Pulmonary rehabilitation immediately before minimally-invasive procedures and medication trials does not share that rationale. In patients with severe COPD, if pulmonary rehabilitation is provided, it should be completed a number of months prior to enrollment in a clinical trial. This period would allow the patient to return closer to a steady-state before entering the trial. This follows the principle that all known variables should be in a steady-state before an experimental intervention.

Without knowing the onset of action, peak response, and level of maintenance for the components of pulmonary rehabilitation, rehabilitation in close relation to an intervention becomes very problematic for study design. It is even more problematic when a subgroup of patients receives pulmonary rehabilitation for a research intervention and the characteristics of this subgroup are not well defined compared to the larger group of patients that could benefit from treatment. Pulmonary rehabilitation can affect selection of the study population, study enrollment, outcome measures, and influence whether the results can be generalized. Clinical trials with medications, devices, or surgery for severe COPD should consider study designs that anticipate the significant effects of pulmonary rehabilitation.