The Proceedings of the American Thoracic Society 2:391-393 (2005)
© 2005 The American Thoracic Society
Challenges and Opportunities for Combination Therapy in Chronic Obstructive Pulmonary Disease
Stephen I. Rennard and
Julie A. Stoner
Pulmonary and Critical Care Medicine Section, Department of Internal Medicine, University of Nebraska Medical Center, Omaha, Nebraska
Correspondence and requests for reprints should be addressed to Stephen I. Rennard, M.D., F.C.C.P., Pulmonary and Critical Care Medicine Section, Department of Internal Medicine, 985125 Nebraska Medical Center, Omaha, NE 68198-5125. E-mail: srennard{at}unmc.edu
 |
ABSTRACT
|
|---|
Advances in the understanding of chronic obstructive pulmonary disease have presented a number of novel therapeutic opportunities. More extensive use of drug combinations is likely, but the development of these therapies presents a number of challenges. In clinical trials, a combination must be tested not only against placebo but also against each of its components, and the false-positive error rate increases rapidly with multiple comparisons. Thus, more groups of subjects must be studied, and more individuals within each group must be studied, in order to ensure statistical significance. Another challenge is that the relatively slow progression of chronic obstructive pulmonary disease and the lack of specificity of commonly used outcome variables complicate the evaluation of all therapies, including combinations. In analogy to genomics and proteomics (evaluation of the pattern of expression of many things simultaneously), it may be more useful to adopt an approach that is here dubbed "clinicomics": consideration of multiple features of chronic obstructive pulmonary disease that are evaluated routinely, for example, with a well-done history and physical examination. The use of a truly multidimensional outcome parameter promises an entirely novel paradigm for the assessment of novel combinations of therapies.
Key Words: biostatistics • clinical trials • drug development • outcome measures
Chronic obstructive pulmonary disease (COPD) is a highly prevalent and easily diagnosed disorder. It is currently the fourth leading cause of death in the United States, and it is estimated to become the third leading cause of death in the United States and worldwide by the year 2020 (1). Despite its importance as a public health problem, COPD has attracted relatively little interest from either the public or the research communities compared with other major public health problems. This complacency toward COPD likely stems, in part, from the significant contribution that smoking makes to the pathogenesis of COPD and the resulting attitude that the illness is, to a large part, self-inflicted. In addition, diagnostic and therapeutic nihilism toward COPD has stemmed from the incorrect attitude that the illness is a relentlessly progressive condition for which little can be done.
These attitudes are changing. Evidence that COPD is both preventable and treatable has led to new initiatives to increase disease awareness (2, 3). In addition, advances in the understanding of the pathogenesis and pathophysiology of COPD have presented a number of novel therapeutic opportunities, which are being explored aggressively (2, 4). The future treatment of patients with COPD is almost certain to involve more extensive use of combinations of pharmacologic and nonpharmacologic therapies. The number of plausible combinations, moreover, is relatively large and increasing rapidly. Although the development of new therapeutic modalities is good news for the patient with COPD, the large number of potential therapies poses very real problems for their development and implementation. These problems are likely to challenge the statistical, regulatory, and clinical assumptions and procedures that have served so well for the development and assessment of treatment modalities over the last century.
|
STATISTICAL PROCEDURES
|
|---|
One of the major advances in biomedical research over the last century has been the application of rigorous statistical methodology to clinical problems. The early history of the development of these methods has been reviewed by David Salsburg in his elegant book The Lady Tasting Tea: How Statistics Revolutionized Science in the Twentieth Century (5). Salsburg relates the famous anecdote of an Englishwoman who claimed that tea tasted different if tea were added to milk rather than if milk were added to tea. In response to skeptics who doubted this claim, she offered to taste samples whose makeup was unknown (blinded) to her. The statistical methodologic question was how many samples must she test to "prove" the point. She would, of course, have a 50% chance of choosing correctly. By random chance she could choose two answers in a row correctly one time in four, three answers in a row one time in eight, and so on. The odds of choosing correctly for any number of choices can be calculated. If, in a given series of tests, the number of correct choices observed would have been expected, on a purely random basis (assuming a 50% chance of choosing correctly), to be 1 time in 20 (5% of the time, the "p value") or fewer, the results were regarded as "significant." In this example, the p value should be interpreted as a conditional probability; it is the probability of observing x or more correct answers under the null hypothesis that there is a 50% chance of choosing correctly. This definition of significance, and the associated assumption that p < 0.05 is a reasonable threshold, underlies most of the testing of clinical interventions in the last 100 years (although there have been many sophisticated modifications). A treatment effect is significant if the treatment results in an improvement in an outcome, relative to a control group outcome, that is unlikely to have arisen if the treatment and control outcomes were the same. This strategy has been extremely useful in identifying effective treatments.
A number of issues arise, however, when treatments are used in combination. Several combinations of two medications have been approved for use in patients with COPD. Food and Drug Administration requirements for approval of these combinations generally specify that the combination be more effective than either of the component medications and that each contribute to efficacy. This reasonable requirement poses an important statistical constraint. A clinical trial of a two-component combination that compares it not only with placebo, but also with each component, changes the odds that one of them will show a difference as a result solely of random chance. Specifically, as the number of comparisons increases, the chance increases that a difference will be deemed significant, when in reality there is no difference (6). The number of potential combinations, and the statistical false-positive error rate, rise rapidly as more components are included. Although there are several approaches to this problem, the following "conservative" analysis highlights the issues.
When considering multiple medications, each of which could be included or not included, the total number of possible medication combinations is 2N, where N is the number of components. The number of comparisons with the complete combination, therefore, will be 2N – 1. For a combination with 7 components, there will be 127 possible comparisons. This large number of comparisons requires that a more rigorous standard be imposed to ensure that the results are not likely to have arisen as a result of random chance alone. For example, when comparing two medications given alone or in combination relative to a control therapy, a threshold for the false-positive error rate (
level) of 1.67% may be used instead of a 5% threshold, so that only more extreme differences are described as "significant." Evaluation of combinations, therefore, poses major practical problems. First, more groups of subjects must be studied as more comparisons must be made. Second, more individuals within each group must be studied to ensure statistical significance in the presence of a lower threshold for the false-positive error rate. As a result, rigorous evaluation of combinations requires studies that are large and potentially impractical.
|
CLINICAL OUTCOME MEASURES
|
|---|
The clinical features of COPD also complicate the assessments of combinations. Limitation of expiratory airflow, most commonly measured by the forced expiratory volume in 1 second (FEV1), is both the defining characteristic of COPD and the most commonly assessed clinical outcome measure (2, 7). Several distinct anatomic lesions contribute to reduced FEV1 (8, 9), including the loss of lung elastic recoil consequent to tissue destruction that characterizes emphysema and the fibrosis and narrowing of small airways that characterize bronchitis. The fixed airflow limitation that results from these lesions progresses over a time frame of decades (10). Smoking cessation early in the course of COPD can slow the rate at which lung function is lost (11, 12). It is unknown, however, if any other therapeutic interventions have similar effects. Demonstration that smoking cessation was of benefit in reducing the rate at which lung function declined, assessed as FEV1, required a study of 6,000 subjects and a follow-up of 5 years (12, 13). Demonstration by an intention-to-treat analysis that smoking cessation was of benefit required 11 years of follow-up (11). Studies of other interventions that have effects of similar magnitude would require studies of similar size, so the lack of clinical information for other interventions should not be surprising.
For bronchodilators, improvement in FEV1 likely results from reduction in the small degree of tone present in airway smooth muscle in patients with COPD. For inhaled glucocorticoids, however, improved airflow may result from a reduction in airway inflammation (2). This results in an interesting paradox: although currently available medications for treating COPD improve FEV1, which is the defining characteristic of COPD, they improve it by actions that are not relevant to the primary lesions that are the major causes of the physiologic abnormalities in the disease, that is, lung destruction with loss of elastic recoil and peribronchial fibrosis. The relatively slow progression of COPD and the lack of specificity of the most commonly used physiologic measures will complicate the assessment of all therapies, including combinations.
Studies, however, suggest the utility of a number of other clinical outcomes. Dynamic hyperinflation, for example, is now recognized as an important physiologic descriptor in patients with COPD (14, 15). In addition, there is increasing recognition that COPD affects patients in many ways besides altering airflow. Exacerbations, for example, are major features of the disease for many patients (16, 17), and current therapies can both prevent and treat exacerbations (2, 7). Patients with COPD, moreover, suffer from a number of systemic effects (18). Although incompletely described, these include skeletal muscle weakness (19), evidence of systemic inflammation that may contribute to an increased risk for cardiovascular complications (20), increased depression (21), and increased osteoporosis (22), among others. The skeletal muscle weakness, which may be associated with weight loss that carries a particularly poor prognosis (23), may be a better correlate of exercise capacity than is lung function (24). Similarly, exercise capacity more closely determines health status, as assessed by disease-related "quality of life" instruments such as the St George's Respiratory Questionnaire, than does FEV1 (25). Importantly, muscle weakness appears to interact with ventilatory limitations in compromising patient performance. Bronchodilator medications, for example, were found to be ineffective at improving patient performance in the presence of muscles that fatigued easily (26). Such interactions, of course, suggest the need to develop combination therapies that are able to target distinct physiologic systems as well as the need to measure clinical outcomes in addition to FEV1.
The need for combination therapies is further emphasized by advances in knowledge of the biology underlying COPD. The complex signaling pathways that control inflammation and tissue remodeling are increasingly well defined (27, 28), which presents a large number of potential therapeutic opportunities. While "strategic points" in these signaling networks may be highly amenable to treatment, an alternative approach is the simultaneous targeting of multiple components of a network. How to assess such treatment strategies, however, remains problematic.
In current combinations, each component is, generally, individually effective. This permits statistically tractable dose-ranging studies of individual components, using the current rules for significance. It is completely plausible, however, that an agent that is effective in a combination will be entirely without activity when used alone. In such a case, it will be impossible to do dose-ranging studies of the drug alone.
|
THE NEED FOR NEW PARADIGMS
|
|---|
A daunting, if not impossible, hurdle is created when the application of "classic" paradigms for drug development to a disease requires that large numbers of subjects be studied for extended periods of time. In this context, it is important to recognize that the statistical approaches that have been so important over the last century are not the only means by which knowledge can be acquired. Combinations are readily assessed, for example, by chefs, composers, and many other artists. These individuals routinely make and adjust complex mixtures of many components, gauging success by nonstatistical methods, often with a battery of well-defined outcomes. In this, there may be a lesson for biomedical research.
Our "traditional" approach has been to use a single rigorously defined outcome measure. However, it is becoming clear that COPD, like most diseases, is characterized by a multiplicity of effects. Multidimensional instruments that assess COPD have been successful (25, 29). Health status instruments, for example, integrate a number of domains of patient-reported outcomes (25). The BODE (body mass index, airflow obstruction, dyspnea, and exercise capacity) index integrates both symptoms and physiologic measures (29). Both have proved useful in a number of settings. However, both collapse multidimensional information into a single numerical score. More sophisticated approaches are possible and may be helpful in assessing COPD clinical response.
Advances in the ability to quantify the simultaneous expression of thousands of genes have led to the field of genomics. Expression of sets of related genes often characterizes conditions much better than does expression of a single gene. The ability to quantify the simultaneous expression of many proteins has led to the analogous field of proteomics. Similar advances in the studies of lipids and metabolites has led to lipidomics and metabonomics. The overall concept for "omics" is that evaluating the pattern of expression of many things simultaneously provides insight into complex disorders. There should be a clinical correlate of this approach: the multiple features that are routinely evaluated in a clinical assessment of a patient, for example, with a well-done history and physical examination, could be considered as "clinicomics." The use of a truly multidimensional outcome parameter promises an entirely novel paradigm for the assessment not only of novel therapies, but of novel combinations of therapies. The mathematical and statistical approach to the multiple "omics" fields is rapidly advancing. Applying these methods to the clinical assessment of outcomes is an important opportunity for the development of novel therapies for complex disorders such as COPD. The promise of "omics" has attracted tremendous interest as it may lead to "markers" of disease that potentially could be used as surrogates for clinical outcomes. This approach could include clinical parameters that are at present largely ignored.
|
CONCLUSIONS
|
|---|
COPD is a devastating condition, the burden of which is increasing. Currently available treatments are helpful, but novel treatments are desperately needed. The many appealing targets that are being explored and the plausible use of agents in combination are extremely encouraging. However, more effective and efficient means for testing therapeutic strategies will have to be developed if the burden of disease is to be reduced. New paradigms for the assessment of clinical interventions are likely to be required for these goals to be met.
|
FOOTNOTES
|
|---|
Conflict of Interest Statement: S.I.R. has participated as a speaker in scientific meetings and courses under the sponsorship of GlaxoSmithKline (GSK). He has consulted with GSK with relevance to the topics noted in the present manuscript. He serves on advisory boards for Roche and Altana. He has been sponsored by GSK for several clinical trials and has received laboratory support. Over 3 years, the total for this approximates $1,800,000. He has also conducted clinical trials for Roche ($60,000) and Altana ($80,000). He has conducted both clinical trials and basic studies under the sponsorship of Centocor ($140,000). He has conducted clinical trials for Pfizer ($80,000). A patent is pending on the use of PDE4 inhibitors in repair; and he is a coinventor of the patent owned by the University of Nebraska Medical Center. J.A.S. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript.
(Received in original form April 25, 2005; accepted in final form May 25, 2005)
|
REFERENCES
|
|---|
- Murray CJ, Lopez AD. Alternative projection of mortality by cause 1990–2020: global burden of disease study. Lancet 1997;349:1498–1504.[CrossRef][Medline]
- Global Initiative for Chronic Obstructive Lung Disease. Global strategy for the diagnosis, management and prevention of chronic obstructive pulmonary disease: NHLBI/WHO workshop report. Bethesda, MD: National Heart, Lung, and Blood Institute; 2001. Update of the Management Sections, GOLD website (www.goldcopd.com). Date updated: July 1, 2003.
- Petty RL, Nett LM. COPD: prevention in the primary care setting. National Lung Health Education Program (NLHEP); 2001.
- Barnes PJ, Shapiro SD, Pauwels RA. Chronic obstructive pulmonary disease: molecular and cellular mechanisms. Eur Respir J 2003;22:672–688.[Abstract/Free Full Text]
- Salsburg D. The lady tasting tea: how statistics revolutionized science in the twentieth century. New York: Henry Holt and Company; 2001.
- Hochberg Y, Tamhane A. Multiple comparison procedures. New York: Wiley, 1987.
- Rennard SI. Treatment of stable chronic obstructive pulmonary disease. Lancet 2004;364:791–802.[CrossRef][Medline]
- Piquette CA, Rennard SI, Snider GL. Chronic bronchitis and emphysema. In: Murray JF, Nadel JA, editors. Textbook of respiratory medicine. Philadelphia, PA: W.B. Saunders, 2000. pp. 1187–1245.
- Niewoehner DE, Sobonya RE. Structure–function correlations in chronic obstructive pulmonary disease. In: Baum GL, Wlinsky E, editors. Textbook of pulmonary diseases. Boston, MA: Little, Brown and Company; 1994. pp. 973–993.
- Fletcher C, Peto R, Tinker C, Speizer F. The natural history of chronic bronchitis and emphysema. New York: Oxford University Press; 1976.
- Anthonisen NR, Connett JE, Murray RP. Smoking and lung function of Lung Health Study participants after 11 years. Am J Respir Crit Care Med 2002;166:675–679.[Abstract/Free Full Text]
- Anthonisen NR, Connett JE, Kiley JP, Altose MD, Bailey WC, Buist AS, Conway WA Jr, Enright PL, Kanner RE, O'Hara P, et al. Effects of smoking intervention and the use of an inhaled anticholinergic bronchodilator on the rate of decline of FEV1. JAMA 1994;272:1497–1505.[Abstract]
- Anthonisen N, Connett J, Friedman B, Glass M, Kilday DP, Mingo TS, Rudolphus A, Williams GW. Design of a clinical trial to test a treatment of the underlying cause of emphysema. Ann N Y Acad Sci 1991;624:31–34.
- O'Donnell DE. Assessment of bronchodilator efficacy in symptomatic COPD: is spirometry useful? Chest 2000;117:42S–47S.[Abstract/Free Full Text]
- O'Donnell DE, Revill SM, Webb KA. Dynamic hyperinflation and exercise intolerance in chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2001;164:770–777.[Abstract/Free Full Text]
- Spencer S, Jones PW. Time course of recovery of health status following an infective exacerbation of chronic bronchitis. Thorax 2003;58:589–593.[Abstract/Free Full Text]
- Seemungal TA, Donaldson GC, Paul EA, Bestall JC, Jeffries DJ, Wedzicha JA. Effect of exacerbation on quality of life in patients with chronic obstructive pulmonary disease. Am J Respir Crit Care Med 1998;157:1418–1422.[Medline]
- Wouters EF. Chronic obstructive pulmonary disease. 5. Systemic effects of COPD. Thorax 2002;57:1067–1070.[Abstract/Free Full Text]
- Gosselink R, Troosters T, Decramer M. Peripheral muscle weakness contributes to exercise limitation in COPD. Am J Respir Crit Care Med 1996;153:976–980.[Abstract]
- Sin DD, Man SF. Why are patients with chronic obstructive pulmonary disease at increased risk of cardiovascular diseases? The potential role of systemic inflammation in chronic obstructive pulmonary disease. Circulation 2003;107:1514–1519.[Abstract/Free Full Text]
- van Manen JG, Bindels PJ, Dekker FW, IJzermans CJ, van der Zee JS, Schade E. Risk of depression in patients with chronic obstructive pulmonary disease and its determinants. Thorax 2002;57:412–416.[Abstract/Free Full Text]
- Praet JP, Peretz A, Rozenberg S, Famaey JP, Bourdoux P. Risk of osteoporosis in men with chronic bronchitis. Osteoporos Int 1992;2:257–261.[CrossRef][Medline]
- Schols AM, Slangen J, Volovics L, Wouters EF. Weight loss is a reversible factor in the prognosis of chronic obstructive pulmonary disease. Am J Respir Crit Care Med 1998;157:1791–1797.
- Schols AM, Mostert R, Soeters PB, Wouters EF. Body composition and exercise performance in patients with chronic obstructive pulmonary disease. Thorax 1991;46:695–699.[Abstract]
- Jones PW, Quirk FH, Baveystock CM. The St George's Respiratory Questionnaire. Respir Med 1991;85(Suppl B):25–31; discussion 33–37.
- Saey D, Debigare R, LeBlanc P, Mador MJ, Cote CH, Jobin J, Maltais F. Contractile leg fatigue after cycle exercise: a factor limiting exercise in patients with chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2003;168:425–430.[Abstract/Free Full Text]
- Barnes PJ. Potential novel therapies for chronic obstructive pulmonary disease. Novartis Found Symp 2001;234:255–267; discussion 267–272.[Medline]
- Rennard S. Defective REPAIR in COPD: the American hypothesis. In: Pauwels RA, Postma, DS, editors. Long-term intervention in chronic obstructive pulmonary disease. New York: Marcel Dekker; 2004. pp. 165–200.
- Celli BR, Cote CG, Marin JM, Casanova C, Montes de Oca M, Mendez RA, Pinto Plata V, Cabral HJ. The body-mass index, airflow obstruction, dyspnea, and exercise capacity index in chronic obstructive pulmonary disease. N Engl J Med 2004;350:1005–1012.[Abstract/Free Full Text]
TOP
ABSTRACT
STATISTICAL PROCEDURES
