Standardization of the Single-Breath Diffusing Capacity in a Multicenter Clinical Trial*

Transcription

1 CHEST Standardization of the Single-Breath Diffusing Capacity in a Multicenter Clinical Trial* Original Research Robert A. Wise, MD, FCCP; John G. Teeter, MD; Robert L. Jensen, PhD; Richard D. England, MD, FCCP; Pamela F. Schwartz, PhD; Donald R. Giles, RRT; Richard C. Ahrens, MD; Neil R. MacIntyre, MD, FCCP; Richard J. Riese, MD; and Robert O. Crapo, MD, FCCP PHYSIOLOGIC TESTING Background: Standardization of the measurement of single-breath diffusing capacity of the lung for carbon monoxide (DLCO) is difficult to implement in multicenter trials as differences in equipment, training, and performance guidelines have led to high variability between and within centers. The safety assessment of inhalable insulin required the standardization of measurement of single-breath DLCO in multicenter clinical trials to optimize test precision. Methods: This was an open-label, 24-week, parallel-group, outpatient study of inhaled human insulin in participants with type 1 diabetes who were randomly assigned to receive treatment with daily premeal inhaled or subcutaneous (SC) insulin for 12 weeks, followed by SC insulin for 12 weeks. Monitoring of single-breath DLCO using standardized methodology was performed. Standardization included uniform instrumentation, centrally trained study coordinators, and centralized data monitoring and review of quality control. Sites received feedback within 24 h for any tests of unacceptable quality with recommendations for improvement. Results: A total of 226 study participants at 33 sites completed 11,335 DLCO efforts during 4,797 test sessions; 3,607 (75.2%) and 4,581 (95.5%) of all testing sessions yielded two American Thoracic Society-acceptable efforts that varied by < 1 and 2 ml/min/mm Hg, respectively. Only 65 sessions produced one or fewer acceptable efforts. The root mean square intrasubject coefficient of variation in DLCO at the end of the comparative dosing phase was 6.01%. Conclusions: The standardized methodology employed in this study demonstrates the feasibility of collecting high-quality single-breath DLCO data in the setting of a multicenter clinical trial with reliability that is comparable to spirometry. (CHEST 2007; 132: ) Key words: clinical trials; diffusing capacity; inhaled human insulin; methodology; respiratory function tests Abbreviations: ATS American Thoracic Society; BHT breathhold time; CI confidence interval; CV coefficient of variation; Dlco diffusing capacity of the lung for carbon monoxide; ERS European Respiratory Society; EXU inhaled human insulin; IVC inspiratory vital capacity; PFT pulmonary function test; RMSCV root mean square coefficient of variation; SC subcutaneous The large absorptive area of the pulmonary vascular bed and the narrow interface between air and blood within the alveolar compartment make the lung an ideal portal for systemic drug delivery. With the technological advances that have occurred in the past few years in the field of aerosol drug delivery, inhaled proteins, polypeptides, and small molecules are now in development for treatment of a variety of systemic diseases. 1 The inhalation of foreign proteins and other excipients raises questions regarding the risk of lung toxicity that require careful safety assessment and monitoring. Comprehensive and sensitive measurements of lung function, including pulmonary gas exchange, provide one means of assessing the safety of aerosol drug delivery. Such measurements are important outcome measures in clinical trials 2 assessing the safety of inhalation drug delivery. Clinical trials using spirometry as a primary or key outcome measure have used highly standardized methodology for data acquisition. 3,4 These methods CHEST / 132 / 4/ OCTOBER,

2 have included consistency of equipment, centralized training, certification of technical staff, real-time data quality prompts, centralized review, and feedback to investigators. Such standardization has reduced within-subject variance, resulting in increased statistical precision. 3 Although guidelines for the measurement of the single-breath diffusing capacity of the lung for carbon monoxide (Dlco) have been periodically published by the American Thoracic Society (ATS) and the European Respiratory Society (ERS), 5 8 to our knowledge, such a standardized approach to the measurement of Dlco has never been performed in a multicenter clinical trial. Studies 9 have shown that the measurement of Dlco can vary markedly between laboratories. Because of this imprecision, multicenter clinical trials have not heretofore used Dlco as a primary outcome measure. Inhaled human insulin (EXU) [Exubera (of insulin human recombinant DNA origin) Inhalation Powder; Pfizer Inc; New York, NY; and Nektar Therapeutics; San Carlos, CA] is a dry-powder formulation of recombinant human insulin that has bee approved *From the Johns Hopkins University School of Medicine (Dr. Wise), Baltimore, MD; Pfizer Global Research and Development (Drs. Teeter, England, Schwartz, and Riese), New London, CT; LDS Hospital and University of Utah (Drs. Jensen and Crapo), Salt Lake City, UT; Ferraris Respiratory (Mr. Giles), Louisville, CO; University of Iowa Carver College of Medicine (Dr. Ahrens), Iowa City, IA; and Duke University Medical Center (Dr. MacIntyre), Durham, NC. Sections of this report have been previously published as an abstract at the Annual Meeting of the American Thoracic Society (ATS), San Diego, CA, May 20 to 25, This study was sponsored by Pfizer Inc and has a contract with Ferraris Respiratory to provide centralized pulmonary function test monitoring for clinical trials. Editorial support was provided by Susanne Vondracek-Klepper of PAREXEL and was funded by Pfizer Inc. Dr. Wise served on an advisory board for Pfizer studies on inhaled insulin and was paid for that work. The terms of the arrangement are managed by Johns Hopkins University in accordance with its conflict-of-interest policy. Dr. MacIntyre served as a consultant for Pfizer studies on inhaled insulin and Viasys Health Care. Dr. Crapo served on an advisory board for the Pfizer studies on inhaled insulin and was paid for that work. Dr. Jensen has served as a consultant for Pfizer and owns Pfizer stock. Dr. Riese has been an employee of Pfizer Inc since December Dr. Schwartz has been an employee of Pfizer since June Dr. Ahrens has served as a consultant to Pfizer for studies on inhaled insulin and was paid for that work. Mr. Giles is an employee of Ferraris Respiratory, Inc, which is paid by Pfizer for pulmonary function test equipment and monitoring in clinical trials. Dr. England has been an employee of Pfizer Inc since 1994 and holds Pfizer stock and stock options. Dr. Teeter has been a full-time employee of Pfizer Inc since 2000 and holds stock and stock options. Manuscript received February 21, 2007; revision accepted June 22, Reproduction of this article is prohibited without written permission from the American College of Chest Physicians ( org/misc/reprints.shtml). Correspondence to: Robert A. Wise, MD, FCCP, Johns Hopkins Asthma & Allergy Center, 5501 Hopkins Bayview Circle, Baltimore, MD 21224; rwise@jhmi.edu DOI: /chest in the United States and European Union for the treatment of type 1 or type 2 diabetes mellitus in adults However, changes in Dlco of 0.4 to 1.2 ml/min/mm Hg were observed with EXU in studies using routine clinical pulmonary function laboratory testing. The present study was conducted to define the effect of EXU treatment on Dlco using a standardized approach designed to improve test precision. The resultant improvement in test reliability is the focus of this report. Some of the results of this study have been previously reported in abstract form. 14 The primary efficacy and safety results of this study have been reported elsewhere. 15,16 Study Design Methods and Materials This was an open-label, 24-week, parallel-group trial, approved by the institutional review board of each participating center. All subjects provided written informed consent. Participants were patients with type 1 diabetes 17 (age range, 25 to 65 years) who were using subcutaneous (SC) insulin least twice daily. Exclusion criteria were as follows: history of active lung disease; smoking within 6 months of study enrollment; abnormal lung function; inability to perform pulmonary function tests (PFTs); or major organ system disease. The prediction equations of Miller et al, 18 Crapo et al, 19 and Hankinson et al, 20 were used respectively for Dlco, total lung capacity, and FEV 1. A 12% adjustment in predicted values for total lung capacity and Dlco was applied for African-American participants. This study consisted of a 3-week run-in phase, a 12-week comparative phase, and a 12-week washout phase. Participants were randomized to receive either premeal EXU 12 plus longacting insulin or SC insulin during the comparative phase. During the washout phase, all participants received SC insulin. Monitoring and Measurements Dlco measurements were undertaken at screening and during study weeks 2, 1, 0, 1, 2, 3, 4, 6, 8, 12, 14, 16, 20, and 24. Testing was conducted between 6:00 am and 10:00 am to reduce diurnal variation 21 before the first dose of insulin. All study sites used the same lung function analyzer (Collins CPL; Ferraris Respiratory; Louisville, CO), which was factory tested with a Dlco simulator 22 prior to deployment. Analyzers were calibrated and tested for leaks daily. Standardized protocol-specific software provided real-time quality-control prompts. Technicians, many with no prior experience in performing lung function testing, underwent a 2-day group training session. After training, technicians were required to pass a written test and to perform acceptable quality testing on five volunteers before testing study participants. Test Performance Up to five efforts were allowed per session with a goal of obtaining two acceptable and reproducible efforts. Efforts meeting the following criteria 6 were acceptable: inspiratory vital capacity (IVC) of 90% of the FVC; inspiratory time of 2 s; a breathhold time (BHT) of 9 to 11 s; and an expiratory 1192 Original Research

3 The intrasubject variability of repeated measurements over the course of the study was assessed by the root mean square coefficient of variation (RMSCV [the square root of the average squared coefficient of variation for tests performed at baseline and at week 24]). The Spearman rank-order correlation test was used to assess the correlation between the average Dlco reported for a session and the difference between acceptable efforts contributing to the average. Figure 1. Mean number of Dlco efforts per test session. time of 4 s. Efforts associated with an IVC/FVC ratio of 85% were allowed if all other criteria were met. Discard volumes and alveolar sampling volumes were set at 750 and 500 ml, respectively. Test data were transmitted to the reading center (Ferraris Respiratory), where they were reviewed within 24 h for quality and feedback to the technician. The mean of two within-session Dlco efforts that met the acceptability criteria was reported as the value for that session. If only one acceptable effort was achieved, that value was reported as the study value. If no acceptable effort was achieved, the effort with the largest IVC was used as the study value. Bimonthly biological control testing was performed at all study sites. Values that varied by 10% of the running average, indicative of potential system malfunction, were queried by the central reading facility, and preventive maintenance was performed as needed. Statistical Analysis Within-session repeatability was calculated using the highest lowest acceptable within-session efforts. The reliability coefficient 23 was determined using values collected during the run-in period. Results Subject Characteristics A total of 353 participants were screened, and 226 participants were randomized, with 196 participants completing the study. Of the 127 participants who failed the screening, 38 failed to meet PFT inclusion criteria and 11 were unable to perform PFTs of acceptable quality. The baseline demographic and clinical characteristics of participants have been reported elsewhere. 16 The mean Dlco at baseline were 27.2 ml/min/mm Hg (93.9% predicted) and 26.9 ml/min/mm Hg (94.3%), respectively, for subjects treated with EXU and SC insulin. DLCO Data Collection A total of 226 randomized participants at 33 sites performed a total of 11,335 Dlco efforts during 4,797 Dlco test sessions. The average number of Dlco efforts performed decreased during the run-in period from 2.8 at the week 3 screening to 2.4 at the week 0 screening, suggesting a training effect (Fig 1). Thereafter, the average number of efforts Figure 2. Difference among within-session acceptable Dlco efforts (n 4,732). CHEST / 132 / 4/ OCTOBER,

4 per session remained fairly constant. The mean Dlco values from all sessions ranged from 13.3 to 50.4 ml/min/mm Hg (n 4,797). The reliability coefficient for run-in values obtained prior to randomization (ie, weeks 2, 1, and 0) was 0.988, indicating a very high degree of intersession consistency within participants prior to randomization. The reliability coefficient did not change (0.990) when the week 3 measurements were included in the assessment. A total of 4,797 test sessions (95.5%) produced at least two acceptable Dlco efforts that could be used to determine the reported average. Of these, 3,607 test sessions (75.2%) and 4,581 test sessions (95.5%), respectively, were reproducible to within 1 and 2 ml/min/mm Hg (Fig 2). Only 65 test sessions (1.4%) produced one or fewer efforts. The magnitude of the intrasession test difference correlated weakly with the mean Dlco value (Spearman correlation coefficient, 0.23; p 0.001; n 4,732) [Fig 3]. An IVC of 90% FVC was achieved in 10,167 test sessions (89.7%) and in 85% of participants in 10,919 of all efforts (96.3%) [Fig 4]. Interestingly, there were 218 sessions (1.9% of the total number) in which at least one effort, which was associated with an IVC/FVC ratio of 85 to 90%, contributed to the average value reported. The repeatability of the data obtained from these sessions did not appear to differ from those obtained in the overall group (74.9% and 94.5% of acceptable efforts, respectively, from these 218 sessions were reproducible to within 1 and 2 ml/min/mm Hg). BHTs between 9 and 11 s, inclusive, were achieved in 11,101 of all efforts (97.9%) [Fig 5]. The majority of efforts outside of this range were due to BHTs in excess of 11 s (n 209). An inspiratory time to 90% of IVC in 2 s was achieved in 11,209 of all efforts (98.9%) [Fig 6]. Adjustment of the washout volume to ensure alveolar sampling at the time of centralized data review was needed for 1% of all test sessions. Figure 4. Distribution of IVC/FVC ratios among all efforts (n 11,335). The RMSCV in Dlco was 6.01% after 12 weeks of comparative treatment. This contrasts with the results of previous studies in patients with type 1 diabetes in which the RMSCV Dlco values using nonstandardized methods after 24 weeks of treatment were 9.99% 11 and 9.51%. 12 The mean treatment group difference in the change from baseline Dlco after 12 weeks of therapy in the present study was ml/min/mm Hg (95% confidence interval [CI], to ml/min/mm Hg). 15 These results are contrasted with historical data from within the EXU development program 11,12 in which standardized methods of Dlco measurement were not used (Fig 7). Discussion To our knowledge, this is the first time that standardized methods for measuring Dlco, analogous to those widely used for spirometry, have been employed in a multicenter clinical trial. Our results demonstrate the feasibility of obtaining reproducible, high-quality data in this setting, and provide a basis for optimizing precision and minimizing sample sizes in multicenter trials using Dlco. Figure 3. Correlation between within-session acceptable efforts and Dlco test average (n 4,732). The Spearman rank correlation is 0.23 (p 0.001). Figure 5. Distribution of BHTs among all efforts (n 11,335) Original Research

5 Figure 6. Distribution of time to achieve 90% of IVC among all efforts (n 11,334). (Does not include one effort in which the time to achieve 90% IVC was 10 s.) In addition to standardized instrumentation, a two-layered strategy was used in the present study to ensure the collection of high-quality data. In the local layer, all research coordinators underwent a standardized training session, and were required to demonstrate written and technical competency prior to conducting protocol testing. Customized software provided real-time feedback and limited data manipulation during the performance of testing. In the centralized layer, a standardized quality-control procedure was used to review all transmitted efforts within a session and to select the efforts contributing to the reported value. Finally, timely feedback to sites was provided for test results of unacceptable quality. Using this approach, 75.2% and 95.5%, respectively, of the 4,797 Dlco test sessions produced at least two acceptable efforts that varied by 1 and 2 ml/min/mm Hg. This within-session test variability is well within the range of reproducibility recommended by the ATS and ERS 8 (ie, 3.0 ml/ min/mm Hg) and is better than the variability from a single university-based pulmonary function laboratory, 24 where a within-session difference of 2.5 ml/min/mm Hg was achieved in 96% of test sessions. Because patients with higher values of Dlco had slightly greater within-session differences, we cannot attribute our better results to having patients with normal lung function (Fig 3). Moreover, the within-patient coefficient of variation (CV) in this 24-week study was 6.01%. This degree of variability compares favorably to that obtained in a highly controlled single-center 25 that assessed the performance of five commercially available PFT instruments (including the Collins instrument used in the present study) in the longitudinal measurement of Dlco. In that study, 25 the CV associated with the repetitive measurement of Dlco in 11 healthy volunteers over a 24-week period ranged from 4.93 to 9.83%, depending on the instrument used in the testing. In another single-center study, 26 the coefficient of variation associated with the repetitive measurement of Dlco in eight healthy volunteers ranged between 4.73% and 6.83%, depending on the instrument used in the testing. Our results show that an acceptable precision of Dlco is achievable in multicenter trials and is comparable to the 3 to 5% CV occurring with spirometry. Previous ATS Dlco guidelines 6 recommended an IVC of at least 90% of the largest previously measured IVC. More recently, the joint ATS/ERS guidelines 8 have recommended that the IVC be at least 85% of the vital capacity. In a previous study, 27 an inspiratory volume of at least 90% of the FVC in two acceptable efforts within a testing session was achieved in only 68% of subjects in a communitybased study. In the present study, a within-session inspiratory volume of at least 90% of the FVC was achieved in 95% of efforts, and was at least 85% in almost 100% of efforts. In the subset of patients in which at least one effort associated with an IVC/FVC Figure 7. Mean (95% CI) treatment group difference in change from baseline Dlco in present study vs historical comparisons. CHEST / 132 / 4/ OCTOBER,

6 ratio of 85 to 90% contributed to the average value reported, the general reproducibility between acceptable intrasession efforts was comparable with that obtained in the overall group. Our data therefore demonstrate that the IVC/FVC ratio of 90% previously recommended by the ATS is achievable in a large proportion of this population and suggests that the allowance of efforts with ratios down to 85 to 90% will not reduce intrasession reproducibility. Currently, the ATS/ERS guidelines 8 recommend that acceptable Dlco efforts include an inspiratory time of 4 s and a BHT of 8 to 12 s. Our study was conducted in patients with normal lung function at baseline, and used the more stringent criteria of an inspiratory time of 2sandaBHTof9to11s. Almost all participants achieved 90% IVC within 2 s, and 97.9% had a BHT of 9 to 11 s (Fig 5). Sensitive and precise methods are needed to study the small treatment group differences in Dlco that have been reported with the use of inhaled insulin. Controlled studies of EXU and other inhaled insulin formulations currently under development 28 have revealed small treatment group differences in the change from baseline Dlco after 12 to 24 weeks. However, to date, these small changes in gas transfer have not been prospectively evaluated as primary outcome measures in studies employing standardized methods of measurement. The statistical impact of our approach is illustrated by the size of the 95% CI of the treatment group difference in the change from baseline Dlco in the present study vs previous studies of EXU in which clinical laboratories were used (Fig 7). Despite a reduction in sample size of approximately 25% in the present study, the width of the CI in the difference in Dlco was considerably narrower than that in previous studies reported by Quattrin et al 11 and Skyler et al, 12 which is indicative of a more precise estimate of the impact of EXU on pulmonary gas exchange. The current study has aspects that may limit its general applicability. The study was conducted in a population of adult patients with type 1 diabetes who had normal, or nearly normal, lung function on study entry and was of limited duration, whereas patients with underlying lung disease would be expected to have greater intersession and intrasession variability of Dlco. Moreover, subjects who were unable to perform PFTs of acceptable quality at screening were excluded from participation, although this was only 3.1% of all patients screened. Nonetheless, our experience from this trial suggests that this approach to the measurement of Dlco is applicable to studies of emphysema, interstitial lung disease, or vascular lung disease in which Dlco is an important outcome measure. In summary, a highly standardized approach to the measurement of the single-breath Dlco has provided quality lung function data in a multicenter clinical trial setting, and has permitted a precise assessment of pulmonary gas exchange during EXU therapy. ACKNOWLEDGMENT: We thank Aimee Basile for her assistance with data formatting. References 1 Patton JS, Fishburn CS, Weers JG. The lungs as a portal of entry for systemic drug delivery. Proc Am Thorac Soc 2004; 1: Gilbert-McClain LI. Pulmonary update from the US FDA: the inhalation route; a novel route for systemic drug delivery. Presented at: American Thoracic Society International Conference, Session L8, 2005; San Diego, CA; May 22 25, Enright PL, Johnson LR, Connett JE, et al. Spirometry in the Lung Health Study: 1. Methods and quality control. Am Rev Respir Dis 1991; 143: Childhood Asthma Management Program Research Group. The childhood asthma management program (CAMP): design, rationale, and methods. Control Clin Trials 1999; 20: American Thoracic Society. Single breath carbon monoxide diffusing capacity (transfer factor): recommendations for a standard technique; statement of the American Thoracic Society. Am Rev Respir Dis 1987; 136: American Thoracic Society. Single-breath carbon monoxide diffusing capacity (transfer factor): recommendations for a standard technique; 1995 update. Am J Respir Crit Care Med 1995; 152: Cotes JE, Chinn DJ, Quanjer PH, et al. Standardization of the measurement of transfer factor (diffusing capacity): report Working Party Standardization of Lung Function Tests, European Community for Steel and Coal; official statement of the European Respiratory Society. Eur Respir J Suppl 1993; 16: Macintyre N, Crapo RO, Viegi G, et al. Standardization of the single-breath determination of carbon monoxide uptake in the lung. Eur Respir J 2005; 26: Wanger J, Irvin C. Comparability of pulmonary function results from 13 laboratories in a metropolitan area. Respir Care 1991; 36: Hollander PA, Blonde L, Rowe R, et al. Efficacy and safety of inhaled insulin (Exubera) compared with subcutaneous insulin therapy in patients with type 2 diabetes: results of a 6-month, randomized, comparative trial. Diabetes Care 2004; 27: Quattrin T, Belanger A, Bohannon NJ, et al. Efficacy and safety of inhaled insulin (Exubera) compared with subcutaneous insulin therapy in patients with type 1 diabetes: results of a 6-month, randomized, comparative trial. Diabetes Care 2004; 27: Skyler JS, Weinstock RS, Raskin P, et al. Use of inhaled insulin in a basal/bolus insulin regimen in subjects with type 1 diabetes: a 6-month, randomized, comparative trial. Diabetes Care 2005; 28: Rosenstock J, Zinman B, Murphy LJ, et al. Inhaled insulin (Exubera) improves glycemic control when substituted for or added to oral combination therapy in type 2 diabetes: a randomized, controlled trial. Ann Intern Med 2005; 143: Original Research

7 14 Wise RA, Ahrens R, Jensen JL, et al. Standardized measurement of single breath diffusing capacity (Dlco) in a multicenter clinical trial [abstract]. Proc Am Thorac Soc 2005; 2:A34 15 Norwood P, Dumas R, Cefalu W, et al. Randomized study to characterize glycemic control and short-term pulmonary function in patients with type 1 diabetes receiving inhaled human insulin (Exubera). J Clin Endocrinol Metab 2007; 92: Teeter JG, Riese RJ. Dissociation of lung function changes with humoral immunity during inhaled human insulin therapy. Am J Respir Crit Care Med 2006; 173: American Diabetes Association. Diagnosis and classification of diabetes mellitus. Diabetes Care 2004; 27(suppl):S5 S10 18 Miller A, Thornton JC, Warshaw R, et al. Single breath diffusing capacity in a representative sample of the population of Michigan, a large industrial state: predicted values, lower limits of normal and frequencies of abnormality by smoking history. Am Rev Respir Dis 1983; 127: Crapo RO, Morris AH, Clayton PD, et al. Lung volumes in healthy nonsmoking adults. Bull Eur Physiopathol Respir 1982; 18: Hankinson JL, Odencrantz JR, Fedan KB. Spirometric reference values from a sample of the general US population. Am J Respir Crit Care Med 1999; 159: Cinkotai FF, Thomson ML. Diurnal variation in pulmonary diffusing capacity for carbon monoxide. J Appl Physiol 1966; 21: Jensen RL, Crapo RO. Diffusing capacity: how to get it right. Respir Care 2003; 48: Cronbach LJ. Coefficient and the internal structure of tests. Psychometrika 1951; 16: Punjabi NM, Shade D, Patel AM, et al. Measurement variability in single-breath diffusing capacity of the lung. Chest 2003; 123: Jensen RL, Teeter JG, England RD, et al. Instrument accuracy and reproducibility in measurements of pulmonary function. Chest 2007; 132: Hathaway EH, Tashkin DP, Simmons MS. Intraindividual variability in serial measurements of Dlco and alveolar volume over one year in eight healthy subjects using three independent measuring systems. Am Rev Respir Dis 1989; 140: Welle I, Eide GE, Bakke P, et al. Applicability of the single-breath carbon monoxide diffusing capacity in a Norwegian community study. Am J Respir Crit Care Med 1998; 158: Garg S, Rosenstock J, Silverman BL, et al. Efficacy and safety of preprandial human insulin inhalation powder versus injectable insulin in patients with type 1 diabetes. Diabetologia 2006; 49: CHEST / 132 / 4/ OCTOBER,