Disease progression and changes in physical activity in

Transcription

1 Page 1 of 64 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC Disease progression and changes in physical activity in patients with COPD Benjamin Waschki 1,2, Anne M. Kirsten 1, Olaf Holz 3, Kai-Christian Mueller 2, Miriam Schaper 1, Anna-Lena Sack 1, Thorsten Meyer 4, Klaus F. Rabe 2, Helgo Magnussen 1, Henrik Watz 1 1 Pulmonary Research Institute at LungenClinic Grosshansdorf, Airway Research Center North (ARCN), Member of the German Center for Lung Research, Grosshansdorf, Germany 2 LungenClinic Grosshansdorf, Airway Research Center North (ARCN), Member of the German Center for Lung Research, Grosshansdorf, Germany 3 Fraunhofer Institute for Toxicology and Experimental Medicine, BREATH, Member of the German Center for Lung Research, Hannover, Germany 4 Institute for Epidemiology, Social Medicine and Health System Research, Hannover Medical School, Hannover, Germany Corresponding author and reprint requests: Dr. med. Benjamin Waschki, LungenClinic Grosshansdorf, Woehrendamm 80, Grosshansdorf, Germany. Phone: , Fax: , b.waschki@lungenclinic.de. 1

2 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC Page 2 of 64 Author contributions: BW and HW contributed to conception and study design, acquisition, analysis and interpretation of data, and wrote the article. AMK, OH, and KCM contributed to conception, acquisition and interpretation of data and revision prior to submission. MS and ALS contributed to acquisition of data and revision prior to submission. TM contributed to analysis and interpretation of data and revision prior to submission. KFR contributed to interpretation of data and revision prior to submission. HM contributed to conception and study design, interpretation of data, and revision prior to submission. Funding source: This study was supported by the German Center for Lung Research and the German Statutory Pension Insurance (Deutsche Rentenversicherung Nord). For former cross-sectional analyses, some baseline measurements were supported by an unrestricted research grant from AstraZeneca Germany. Short running title: COPD progression and changes in physical activity Descriptor number: 9.9 (COPD: General) Total word count: 4048 words 2

3 Page 3 of 64 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC At a Glance Commentary Scientific knowledge on this subject: Cross-sectional studies have shown that physical activity of patients with COPD is associated with pulmonary and systemic disease components. What this study adds to the field: Longitudinally, physical activity decreases across all severity stages of COPD, along with a coexisting worsening of airflow obstruction and health status. In addition, a sustained low level of physical activity over time is associated with an accelerated progression of exercise intolerance and muscle depletion. 3

4 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC Page 4 of 64 ABSTRACT Background: Little is known about the role of physical activity in the course of COPD. The aim of this study was to assess changes in physical activity in COPD in relation to severity stages and changes in other disease components, and to evaluate the longitudinal association between sustained physical inactivity and disease progression. Methods: In this prospective cohort study, we measured physical activity (multisensory armband), airflow obstruction (FEV 1 ), health status (St. George s Respiratory Questionnaire [SGRQ]), exercise capacity (6-minute walk distance [6MWD]), muscle mass (fat-free mass [FFM]), and systemic inflammation (fibrinogen and high-sensitivity C-reactive protein) over a 3-year period in 137 patients with COPD and 26 with chronic bronchitis (normal spirometry). Results: Independent of baseline disease severity, steps per day, total daily energy expenditure, and (daily) physical activity level (PAL) decreased by 393, 76 kcal, and 0.04 per year, respectively. The decline in PAL was significantly associated with a decline in FEV 1 and an increase in SGRQ total score. Changes in 6MWD, FFM, and inflammatory markers were not associated with changes in PAL. Independent of FEV 1, sustained physical inactivity (i.e., PAL T0andT1 <1.40) was related to a greater decline in 6MWD and FFM compared to that in patients with some level of activity (i.e., PAL T0and/orT1 1.40; difference, 17 m/yr and 0.87 kg/yr, respectively). Conclusions: Over time, physical activity substantially decreases across all severity stages of COPD, and this decline is paralleled by a worsening of lung function and 4

5 Page 5 of 64 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC health status. Sustained physical inactivity is associated with a progression of exercise intolerance and muscle depletion. Abstract word count: 252 words Key words (MeSH): pulmonary disease, motor activity, exercise, musculoskeletal diseases, quality of life 5

6 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC Page 6 of 64 INTRODUCTION Chronic obstructive pulmonary disease (COPD) is a debilitating disease characterized by airflow limitation that is usually progressive. Exertional dyspnoea, one of the cardinal symptoms of COPD, leads to limitations of physical activity in daily life of COPD patients, which may crucially influence the development of disability 1-3. Cross-sectional analyses have shown that physical activity is reduced early in the course of the disease 4-9. This is an important observation, because physical inactivity is an important predictor of hospitalization and mortality and might be a key link to comorbidities in COPD 16. Indeed, cross-sectional studies revealed that not only reductions in lung function but also several systemic disease components of COPD, such as cardiac dysfunction, systemic inflammation, and muscle weakness, are related to reduced physical activity in patients with COPD 6;17. A recent study in patients with mild COPD demonstrated that physical inactivity is more strongly associated with the presence of comorbidities than is airflow obstruction 9. However, prospective cohort studies that evaluate the association of objectively measured physical activity with other disease components over time are not available so far. In particular, it needs to be evaluated, whether changes of certain disease components parallel changes in physical activity or whether the changes of these disease components might be subsequent to a sustained low level of activity over time. The aim of this cohort study was to assess (1) changes in physical activity depending on baseline severity stages of COPD, (2) changes in physical activity in relation to changes in airflow obstruction, health status, exercise tolerance, skeletal muscle status, and systemic inflammation, and (3) the longitudinal association 6

7 Page 7 of 64 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC between sustained physical inactivity and the progression of these COPD components. Some of the study results have previously been reported in the form of an abstract 19. METHODS Patients and study design In this prospective observational cohort study we investigated 170 outpatients with COPD (Global Initiative for Chronic Obstructive Lung Disease [GOLD] stages I IV) and 30 patients with chronic bronchitis and normal spirometry (formerly considered at risk for COPD) at the time of enrollment at the Pulmonary Research Institute at the LungenClinic Grosshansdorf (Germany). The patients with chronic bronchitis did not differ significantly from patients with COPD in smoking history and socioeconomic factors 4. The study was designed to analyze the role of extrapulmonary manifestations of COPD in disease severity and disease progression. The objective measurement of physical activity was a central focus. Further details are described in the online supplement and elsewhere 4;17.The study was approved by the Ethics Committee of the Medical Association of Schleswig-Holstein, Bad Segeberg, Germany [III/EK 116/05(I); 185/08(I)], and all participants gave written informed consent. The follow-up assessment was performed 2 to 3.5 years after the baseline assessment (online supplement). The prognostic value of baseline measurements for survival status after a median follow-up of 4 years (also confirmed by telephone 7

8 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC Page 8 of 64 contacts) is described elsewhere 12. For the current analysis, only those patients who were available for a follow-up assessment were included. Measurement of physical activity Physical activity was measured via a multisensory armband (SenseWear Pro 2 and InnerView Professional Software version 5.1; BodyMedia; Pittsburgh, Pennsylvania; USA) at baseline and at follow-up over a period of one week as previously described (online supplement) 4;17. Validity of the accelerometer and reliability of the measurement period have been demonstrated 4; Accelerometer outputs and calculated physical activity variables were the total daily energy expenditure (TDEE [kcal/d]), activity energy expenditure (AEE [kcal/d]: TDEE [kcal/d] minus sleeping energy expenditure [kcal/d]), the physical activity level (PAL: TDEE [kcal/d] divided by sleeping energy expenditure [kcal/d]), minutes of moderate activity per day (i.e. time above 3 metabolic equivalents; MMA [min/d]) and steps per day (online supplement) 4;17. The PAL was categorized twice. First, we classified the PAL status at baseline and follow-up based on the cut-off 1.40 as the lower limit of a sedentary or light activity lifestyle according to the FAO/WHO/UNU expert consultation 25. We defined the corresponding PAL groups as sustained physical inactivity (i.e., PAL <1.40 at baseline [t0] and at follow-up [t1]) and some level of activity (i.e., PAL 1.40 at least at one assessment; online supplement). This categorization was used to analyze the association of sustained physical inactivity with changes in other disease manifestations. A secondary categorization of the PAL was based on the magnitude of change, in order to illustrate the results of the ANCOVA calculations (see below). 8

9 Page 9 of 64 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC Measurement of lung function and systemic disease manifestations We assessed post-bronchodilator spirometry, 6-minute walk distance (6MWD), health status (St. George s Respiratory Questionnaire [SGRQ]), muscular status (bioelectrical impedance analysis), and systemic inflammation (fibrinogen and highsensitivity C-reactive protein [hs-crp]) as stated in the online supplement and as previously described 4;17. Fat-free mass (FFM) and muscle depletion were calculated by using an established formula and cut-off values (online supplement) 26;27. Statistical analyses Descriptive data were reported as mean and standard deviation (SD) or median and interquartile range (IQR) (continuous variables) and number and percent (dichotomous variables) depending on scale and distributional characteristics. Skewed data, such as hs-crp, were log-transformed to yield a normal distribution. Cross-sectional comparisons between groups were performed with the use of twotailed t test, analysis of variance (ANOVA) or χ 2 test for dichotomous variables. Longitudinal differences between measurements at baseline and follow-up were tested by paired t test (normal distributed variables). The rate of decline in physical activity was estimated by separate analyses of covariance (ANCOVA) for repeated measures with each physical activity parameter at baseline and follow-up as dependent variables after adjustments for potential confounders, i.e. age, sex, smoking status, frequent exacerbations, baseline FEV 1, time to follow-up, and differences in duration of daylight time as a proxy for seasonality (definition of daylight time and exacerbations see online supplement). 9

10 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC Page 10 of 64 The impact of baseline GOLD stage on the change in physical activity was analyzed in a separate model replacing baseline FEV 1 as a covariate. The association between changes in physical activity and changes in FEV 1, SGRQ, 6MWD, FFM, fibrinogen, and hs-crp was analyzed by separate ANCOVAs for repeated measures with PAL at baseline and PAL at follow-up as the dependent variables and the difference in each independent variable as the covariate, respectively. All variables were given as annual change : (follow-up value minus baseline value) / years to follow-up. Each ANCOVA was corrected for confounders as stated above. Parameter estimates B of each covariate for the decline in PAL were calculated by ANCOVAs with the annual decline in PAL as the dependent variable. In addition, variables with a significant association were depicted according to the magnitude of changes in PAL. Therefore, the change in PAL was secondary classified by every change into five categories, based on 0 ±0.025 as the reference: (1) very fast decline, i.e., greater decrease than 0.125, (2) fast decline, i.e., to 0.124), (3) slow decline, i.e., to 0.074), (4) little change, i.e., a decrease of to an increase of 0.025, and (5) increase, i.e., > To analyze the longitudinal association between sustained physical inactivity and changes in disease manifestations, we tested the factor sustained physical inactivity (i.e., PAL T0+T1 <1.40) for the change in each indicator of disease manifestation over time using ANCOVA for repeated measures. Estimated means per group and p values were adjusted for age, sex, smoking status, frequent exacerbations, time to follow-up and baseline FEV 1. The effect size of the main covariates of all ANCOVAs, was calculated by partial η 2 (eta squared), which is the 10

11 Page 11 of 64 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC proportion of explained variance, according to Cohen (0.01 = small effect, 0.06 = moderate effect, 0.14 = large effect) 28. Additional adjustments for comorbidities according to patient history were performed for all significant models, if there was at least a trend for a bivariate association (i.e., p<0.2; frequency of comorbidities see table E2 online supplement). Further confirmatory analyses were performed based on the physical activity variables steps per day and MMA (online data supplement). RESULTS Baseline characteristics A total of 200 patients (170 with COPD and 30 with chronic bronchitis) were studied at baseline. The follow-up assessment was completed for 163 patients (81.5%) (137 with COPD and 26 with chronic bronchitis). During a median follow-up of 2.8 years (IQR, years), 21 patients died, 14 withdrew from the study, and 2 were lost to follow-up. Table 1 presents patients characteristics at baseline according to their availability for follow-up assessment. Patients who completed the follow-up had better lung function and a higher PAL than those who dropped out (table 1). Changes in physical activity over time according to disease severity at baseline On average, the daily level of all physical activity parameters decreased significantly during the 3 years of follow-up (table 2). The rate of change from baseline to followup after adjusting for age, sex, smoking status, exacerbations, time to follow-up, 11

12 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC Page 12 of 64 baseline FEV 1, and delta daylight time is given in table 2. The corresponding annual rate of decline in (daily) level of TDEE, AEE, PAL, MMA, and steps per day was 76 kcal, 59 kcal, 0.04, 10 min, and 393, respectively (p <0.001). The decline in each physical activity parameter did not differ significantly between patients with COPD and patients with chronic bronchitis (table 2). Among the adjusted confounders, only delta daylight time revealed a significant association for TDEE, AEE, PAL, and MMA. Changes in body weight did not have a significant effect on the decline in physical activity. The decline in TDEE, PAL, and steps per day during follow-up was not effected by baseline severity stages (unadjusted/adjusted p=0.99/0.87, 0.99/0.68, and 0.82/0.85, respectively [figure 1A C]). Analyses of AEE and MMA showed similar results (data not shown). Overall, there was a decrease in PAL in 116 patients (71%) and an increase in 47 (29%). Relation between changes in physical activity and changes in other disease manifestations over time FEV 1, 6MWD, FFM, fibrinogen, and hs-crp changed significantly during the followup period, but SGRQ total score did not (table 3). After correcting for confounders, the annual decline in FEV 1 in our cohort of patients with COPD was 61 ml (±9.1 [SE]) and did not differ from that in patients with chronic bronchitis (59 ml; p=0.90). Nine patients (35%) with chronic bronchitis at baseline developed COPD during follow-up. After adjustment for confounders, both, a higher annual decrease in FEV 1 and a higher annual increase in SGRQ total score, were significantly associated with the decrease in PAL (p=0.034 and 0.007, respectively [table 4]). Figure 2 depicts the 12

13 Page 13 of 64 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC unadjusted mean annual change in FEV 1 and SGRQ according to the magnitude of change in PAL. The largest decline in FEV 1 and the largest increase in SGRQ were observed in patients with a fast or very fast decline in PAL. Patients with increased PAL showed the smallest decline in FEV 1 and a decline in SGRQ (figure 2). The change in 6MWD, FFM, and inflammatory markers was not associated with a decrease in PAL (table 4). The association between the changes in PAL and changes in FEV 1 and SGRQ could be confirmed by analyses based on the activity variables AEE and MMA (table E3 online supplement). Changes in steps per day were significantly associated with changes in SGRQ, but not with lung function decline (table E3 online supplement). Sustained physical inactivity and the progression of COPD manifestations Next, we studied the association of sustained physical inactivity with changes in FEV 1, SGRQ total score, 6MWD, FFM, fibrinogen, and hs-crp levels during followup. We classified patients according to their PAL from baseline to follow-up assessment as sustained physical inactivity (i.e., PAL T0andT1 <1.40; n=39) and some level of activity (i.e., PAL T0and/orT1 >1.40; n=124). After adjustment for confounders, there were significant differences in the baseline values of FEV 1, SGRQ and 6MWD between these groups, whereas the baseline values of FFM, fibrinogen and hs-crp levels did not differ significantly (table 5). The difference in daylight time between baseline and follow-up did not differ between the groups (p = 0.30). Sustained physical inactivity was not related to the change in FEV 1 over time (table 5). The changes in 6MWD and FFM over the 3 years showed clear differences between groups after adjusting for confounders (p = and <0.001 [table 5). The crude annual decline for both groups is given in figure 3A B. The corresponding 13

14 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC Page 14 of 64 adjusted annual rate of decline for 6MWD and FFM for patients with sustained physical inactivity is 35 m/yr and 1.20 kg/yr. Patients with some level of activity had a corrected annual decrease in 6MWD of 18 m/yr and a decrease in FFM of 0.33 kg/yr (table 5). The unadjusted differences in the increase in fibrinogen and hs-crp levels between groups were of borderline significance (p=0.060 and [table 5 and figure 3C D]). After correction for confounder, the corresponding p values were 0.11 and 0.13 (table 5). However, the increases resulted in significant differences between groups in the corresponding serum levels at the follow-up assessment (p = and 0.011, respectively). The associations between sustained physical inactivity based on PAL (i.e., PAL T0andT1 <1.40) and the accelerated decline in FFM and in 6MWD could be confirmed by defining sustained physical inactivity based on steps per day or MMA (table E4 E5 online supplement). DISCUSSION The main finding of our study is that physical activity substantially decreases over time independent of the baseline severity of COPD. The decline in physical activity is paralleled by a worsening of airflow obstruction and health status but not by a worsening of exercise intolerance and muscle depletion. Moreover, a sustained low level of physical activity is related to a progression of exercise intolerance and muscle depletion in all patients with COPD. Interestingly, sustained physical inactivity was not associated with a progression of airflow obstruction or a worsening of health status. 14

15 Page 15 of 64 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC Decrease in physical activity over time So far, there is little information about the natural course of physical activity over time in COPD, even though it is commonly assumed, based on clinical impressions, that physical activity declines as the disease progresses 2;29. However, to the best of our knowledge, there are only two studies available that investigated objectively measured physical activity in patients with COPD longitudinally 30;31. The first study by Agarwal and colleagues demonstrated a decrease of 22 % of vector magnitude units (i.e. movement counts; assessed by a triaxial accelerometer) in 18 patients with moderate COPD during slightly more than 1.5 years of follow-up 30. In another study, Durheim and colleagues evaluated the mean number of steps per day which were measured by an uniaxial accelerometer on two consecutive days at baseline and after 16 weeks in 326 COPD patients 31. They found a significant decline in patients with GOLD D (59% of the total cohort) by approximately 350 steps per day, whereas no decrease of physical activity could be observed in less severe stages during that 16-week period 31. In our study, covering the whole range of disease severity, we observed a significant decline in physical activity in all severity stages of COPD during a follow-up of about 3 years. We found no differences in the absolute rate of decline in physical activity, neither between patients with chronic bronchitis and patients with COPD, nor between the baseline severity stages of COPD. However, because of the lower baseline values in patients with severe to very severe COPD, the relative decline in all physical activity parameters was substantially higher in these stages. Patients with GOLD stage IV nearly halved their number of steps per day from 3073 at baseline to 1602 at follow-up corresponding to a corrected annual decline of 461 steps. They had a mean PAL of 1.23 after 3 years of follow-up, which 15

16 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC Page 16 of 64 is close to the PAL of bed-bound patients 25. Nevertheless, even patients with mild to moderate COPD had a significant decline in physical activity. Furthermore, the proportion of physically very inactive patients with a PAL of <1.40 doubled from 15 to 32% in GOLD stages I and II. This appears to be a clinically relevant observation, considering the fact that this degree of physical inactivity is generally not compatible with long-term health 25 and may accelerate the progression of exercise intolerance and muscle depletion early in the course of the disease. Compared to existing data on the decline in objectively measured physical activity in healthy elderly subjects 32-35, our results indicate that the decline in physical activity in patients with COPD is about two- to fourfold higher. In line with this, Vaes and coworkers recently found a decline in self-reported physical activity to be more likely in patients with COPD than in subjects without COPD 15. Physical activity and changes in lung function The longitudinal relationship between self-reported physical activity and changes in lung function over time has been evaluated in two population-based cohorts, which found a reduced decline in FEV 1 or FEV 0.75 by 5 to 10 ml/yr in physically more active smokers compared with physically less active smokers 36;37. Applying an objective measurement of physical activity in our cohort, we found that the decrease in physical activity was significantly associated with the decrease in FEV 1 over time. In contrast, a sustained low level of physical activity (i.e., PAL T0andT1 <1.40) was not associated with lung function decline in our cohort. This observation doesn t support the theory that physical inactivity may contribute to lung function 16

17 Page 17 of 64 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC decline, but rather suggests a coexisting decline of both parameters during the natural course of the disease. Physical activity and changes in health status We found that a change in PAL was linearly associated with a change in SGRQ total score over time. Patients with a fast decrease in PAL had a significant worsening in health status, whereas patients with an increase in PAL showed an improvement in health status. In contrast, a sustained low level of physical activity was not associated with a worsening of health status during follow-up. In line with our observation, Esteban and coworkers recently found that changes in self-reported physical activity were related to changes in SGRQ 38. However, Esteban et al. also reported that a sustained low level of physical activity was related to a decline in health status, which is different from our observation. A possible explanation for this discrepancy is that the SGRQ includes items of physical activity in daily life, which might bias the relationship between self-reported physical activity and health status. Our objectively measured data only demonstrated a parallel association between physical activity decline and worsening of health status. Physical activity and changes in exercise tolerance Functional exercise capacity as assessed by the 6MWD and physical activity are closely related in several cross-sectional studies in COPD 4;6;39;40, whereas directionality is still unclear 18. Both 6MWD and physical activity decrease with an increase in disease severity. However, physical activity seems to be impaired early in the course of the disease 4-9 whereas in most patients the decrease in 6MWD occurs in the later stages of the disease 4;41. A recent study evaluated the relation between 17

18 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC Page 18 of 64 self-reported physical activity and longitudinal change in 6MWD. Frisk and colleagues observed that self-reported physical activity at baseline was a predictor for longitudinal changes in 6MWD after 3 years of follow-up 42. In our study, we observed that the decline in 6MWD in patients with a sustained very low level of physical activity was significantly faster compared to patients with some level of physical activity over time. By contrast, we did not find a parallel association between the decline in physical activity and the decline in 6MWD. Taking both of our observations into account, it seems that physical inactivity might promote exercise intolerance rather than vice versa. Physical activity and changes in muscle mass Physical inactivity due to exertional dyspnoea is believed to be one of the main contributing factors for skeletal muscle depletion and deconditioning in patients with COPD 2;43. Hopkinson and colleagues observed a loss in FFM of 0.2 kg over a period of 1 year in 64 subjects with severe COPD 44. Among the clinical variables associated with the loss of muscle mass, lung hyperinflation had the strongest association with muscle depletion 44. Interestingly, van den Borst et al. found that the longitudinal loss of lean mass in elderly patients with mild-to-moderate obstructive lung disease (mean age, 74 years) was not different from the loss observed in elderly never-smokers, even though there was a significant difference in their baseline values 45. In our cohort, with an average age of 64 years and an overall decline in FFM of 0.6 kg/yr, a sustained low level of physical activity, but not the decline in physical activity, was associated with an accelerated loss of FFM during follow-up independent of airflow obstruction. This observation supports the clinical concept that physical inactivity itself plays a crucial role in muscle depletion in COPD. 18

19 Page 19 of 64 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC Physical activity and changes in systemic inflammation Physical inactivity is a relevant condition related to systemic inflammation in the general population 46. In patients with COPD, cross-sectional studies have also shown physical inactivity to be related to systemic inflammation 6;17;47. These observations are of interest, because a lack of muscle use seems to result in systemic inflammation, and regular physical activity seems to provide antiinflammatory effects 48. We were unable to demonstrate a significant relation between sustained physical inactivity and systemic inflammation over time in our COPD cohort. This might be related to the relatively low number of patients we studied, considering the large variation in inflammatory markers 49. Limitations of our study First, we did not include a control group at baseline to compare the changes in physical activity in patients with COPD with the changes in healthy subjects. However, our detailed assessment of the decline in physical activity at each GOLD stage and the comparison with patients with chronic bronchitis but normal lung function at baseline might partially compensate for this limitation. Furthermore, even though the results for patients with sustained physical inactivity over time are controlled by patients with some level of activity, we are not able to clarify whether our observations are a COPD specific phenomenon or a general effect related to physical inactivity. Second, we performed only one follow-up assessment. Therefore, we are not able to show whether the decline in physical activity is strictly linear over time. More assessment points might better elucidate the association between changes in 19

20 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC Page 20 of 64 physical activity and disease progression over time. However, ours is the first study to longitudinally evaluate this association at all. Third, our FEV 1 decline in COPD patients of 61 ml/yr is slightly higher than previously reported values from a large randomized controlled trial, which found a decrease of 40 to 42 ml/yr in patients with an average FEV 1 of 48% of predicted at baseline 50. However, our study population differs with regard to severity stages and smoking status compared to the study of Tashkin and colleagues 50 and the observed FEV 1 decline in our study is within reported ranges of other larger COPD populations 51. In addition, we were also able to follow-up patients who developed very advanced COPD which is indicated by a relatively low proportion of patients who dropped out during the 3-year period (i.e., 8%). Such patients would maybe drop out in other large longitudinal studies. Fourth, this is a single center study and the number of patients in our cohort is relatively small. Therefore, further studies are needed to confirm our findings. Fifth, this study design does not allow any interpretation on causality. The advantage of longitudinal studies compared to cross-sectional studies is, that they are able to evaluate time-dependent associations, which represents only one criterion of causality (among several others). At its best, a longitudinal observation study may limit the effect of reverse causality. Clinical implications and conclusions Recent guidelines advocate regular physical activity for patients with COPD at all severity stages despite the fact that little COPD-specific evidence exists 1. Our data clearly support this recommendation, because a sustained low level of physical 20

21 Page 21 of 64 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC activity over time is associated with an accelerated progression of exercise intolerance and muscle depletion. Furthermore, we have demonstrated that physical activity decreases early in the course of the disease, along with a worsening of lung function and health status. Of note, our results were obtained in a cohort that for the current presentation excluded patients who had died during follow-up. For those deceased patients, physical inactivity was already shown to be the strongest predictor of mortality 12. Therefore, the role of physical inactivity in disease progression might even be underestimated in the current analysis. Acknowledgements We thank Dr. Beke Feindt, Kirsten Paasch, Ines Zimmermann and Jeanette Kotzur for their help in collecting patient data and Mary McKenney for a critical review of the manuscript. Some data represent part of the medical doctoral theses of Miriam Schaper and Anna-Lena Sack. Competing Interests The authors have no competing interests to declare. 21

22 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC Page 22 of 64 Reference List 1. Vestbo, J., S. S. Hurd, A. G. Agusti, P. W. Jones, C. Vogelmeier, A. Anzueto, P. J. Barnes, L. M. Fabbri, F. J. Martinez, M. Nishimura, R. A. Stockley, D. D. Sin, and R. Rodriguez-Roisin Global strategy for the diagnosis, management, and prevention of chronic obstructive pulmonary disease: GOLD executive summary. Am.J.Respir.Crit Care Med. 187: ZuWallack, R How are you doing? What are you doing? Differing perspectives in the assessment of individuals with COPD. COPD. 4: Roche, N Activity limitation: a major consequence of dyspnoea in COPD. Eur.Respir.Rev. 18: Watz, H., B. Waschki, T. Meyer, and H. Magnussen Physical activity in patients with COPD. Eur.Respir.J. 33: Troosters, T., F. Sciurba, S. Battaglia, D. Langer, S. R. Valluri, L. Martino, R. Benzo, D. Andre, I. Weisman, and M. Decramer Physical inactivity in patients with COPD, a controlled multi-center pilot-study. Respir.Med. 104: Waschki, B., M. A. Spruit, H. Watz, P. S. Albert, D. Shrikrishna, M. Groenen, C. Smith, W. D. Man, R. Tal-Singer, L. D. Edwards, P. M. Calverley, H. Magnussen, M. I. Polkey, and E. F. Wouters Physical activity monitoring in COPD: compliance and associations with clinical characteristics in a multicenter study. Respir.Med. 106:

23 Page 23 of 64 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC 7. Shrikrishna, D., M. Patel, R. J. Tanner, J. M. Seymour, B. A. Connolly, Z. A. Puthucheary, S. L. Walsh, S. A. Bloch, P. S. Sidhu, N. Hart, P. R. Kemp, J. Moxham, M. I. Polkey, and N. S. Hopkinson Quadriceps wasting and physical inactivity in patients with COPD. Eur.Respir.J. 40: Van Remoortel, H., M. Hornikx, H. Demeyer, D. Langer, C. Burtin, M. Decramer, R. Gosselink, W. Janssens, and T. Troosters Daily physical activity in subjects with newly diagnosed COPD. Thorax 68: Van Remoortel, H., M. Hornikx, D. Langer, C. Burtin, S. Everaerts, P. Verhamme, S. Boonen, R. Gosselink, M. Decramer, T. Troosters, and W. Janssens Risk factors and comorbidities in the preclinical stages of chronic obstructive pulmonary disease. Am.J.Respir.Crit Care Med. 189: Garcia-Aymerich, J., P. Lange, M. Benet, P. Schnohr, and J. M. Anto Regular physical activity reduces hospital admission and mortality in chronic obstructive pulmonary disease: a population based cohort study. Thorax 61: Garcia-Aymerich, J., E. Farrero, M. A. Felez, J. Izquierdo, R. M. Marrades, and J. M. Anto Risk factors of readmission to hospital for a COPD exacerbation: a prospective study. Thorax 58: Waschki, B., A. Kirsten, O. Holz, K. C. Muller, T. Meyer, H. Watz, and H. Magnussen Physical activity is the strongest predictor of all-cause mortality in patients with COPD: a prospective cohort study. Chest 140:

24 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC Page 24 of Garcia-Rio, F., B. Rojo, R. Casitas, V. Lores, R. Madero, D. Romero, R. Galera, and C. Villasante Prognostic value of the objective measurement of daily physical activity in patients with COPD. Chest 142: Watz, H., F. Pitta, C. L. Rochester, J. Garcia-Aymerich, R. ZuWallack, T. Troosters, A. W. Vaes, M. A. Puhan, M. Jehn, M. I. Polkey, I. Vogiatzis, E. M. Clini, M. Toth, E. Gimeno-Santos, B. Waschki, C. Esteban, M. Hayot, R. Casaburi, J. Porszasz, E. McAuley, S. J. Singh, D. Langer, E. F. Wouters, H. Magnussen, and M. A. Spruit An official European Respiratory Society statement on physical activity in COPD. Eur.Respir.J. 15. Vaes, A. W., J. Garcia-Aymerich, J. L. Marott, M. Benet, M. T. Groenen, P. Schnohr, F. M. Franssen, J. Vestbo, E. F. Wouters, P. Lange, and M. A. Spruit Changes in physical activity and all-cause mortality in COPD. Eur.Respir.J. 44: Decramer, M., W. Janssens, and M. Miravitlles Chronic obstructive pulmonary disease. Lancet 379: Watz, H., B. Waschki, C. Boehme, M. Claussen, T. Meyer, and H. Magnussen Extrapulmonary effects of chronic obstructive pulmonary disease on physical activity: a cross-sectional study. Am.J.Respir.Crit Care Med. 177: Gimeno-Santos, E., A. Frei, C. Steurer-Stey, J. de Batlle, R. A. Rabinovich, Y. Raste, N. S. Hopkinson, M. I. Polkey, H. Van Remoortel, T. Troosters, K. Kulich, N. Karlsson, M. A. Puhan, and J. Garcia-Aymerich Determinants and 24

25 Page 25 of 64 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC outcomes of physical activity in patients with COPD: a systematic review. Thorax 69: Waschki, B., M. Schaper, A. L. Sack, O. Holz, A. Kirsten, T. Meyer, K. F. Rabe, H. Magnussen, and H. Watz Decrease of physical activity in patients with COPD in the course of the disease [abstract]. Eur Respir J 40:548s. 20. Patel, S. A., R. P. Benzo, W. A. Slivka, and F. C. Sciurba Activity monitoring and energy expenditure in COPD patients: a validation study. COPD. 4: St Onge, M., D. Mignault, D. B. Allison, and R. Rabasa-Lhoret Evaluation of a portable device to measure daily energy expenditure in free-living adults. Am.J.Clin.Nutr. 85: Hill, K., T. E. Dolmage, L. Woon, R. Goldstein, and D. Brooks Measurement properties of the SenseWear armband in adults with chronic obstructive pulmonary disease. Thorax 65: Van Remoortel, H., Y. Raste, Z. Louvaris, S. Giavedoni, C. Burtin, D. Langer, F. Wilson, R. Rabinovich, I. Vogiatzis, N. S. Hopkinson, and T. Troosters Validity of six activity monitors in chronic obstructive pulmonary disease: a comparison with indirect calorimetry. PLoS.One. 7:e Rabinovich, R. A., Z. Louvaris, Y. Raste, D. Langer, H. V. Remoortel, S. Giavedoni, C. Burtin, E. M. Regueiro, I. Vogiatzis, N. S. Hopkinson, M. I. Polkey, F. J. Wilson, W. MacNee, K. R. Westerterp, and T. Troosters Validity of physical activity monitors during daily life in patients with COPD. Eur.Respir.J. 25

26 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC Page 26 of Human energy requirements: report of a joint FAO/ WHO/UNU Expert Consultation Food Nutr Bull 26: Rutten, E. P., M. A. Spruit, and E. F. Wouters Critical view on diagnosing muscle wasting by single-frequency bio-electrical impedance in COPD. Respir.Med. 104: Vestbo, J., E. Prescott, T. Almdal, M. Dahl, B. G. Nordestgaard, T. Andersen, T. I. Sorensen, and P. Lange Body mass, fat-free body mass, and prognosis in patients with chronic obstructive pulmonary disease from a random population sample: findings from the Copenhagen City Heart Study. Am.J.Respir.Crit Care Med. 173: Cohen, J Statistical power analysis for the behavioural sciences. Academic Press, New York. 29. O'Donnell, D. E. and P. Laveneziana Dyspnea and activity limitation in COPD: mechanical factors. COPD. 4: Agarwal, V., S. Tetenta, J. Bautista, R. ZuWallack, and B. Lahiri Longitudinal changes in directly measured physical activity in patients with chronic obstructive pulmonary disease: the trajectory of change. J.Cardiopulm.Rehabil.Prev. 32: Durheim, M. T., P. J. Smith, M. A. Babyak, S. K. Mabe, T. Martinu, K. E. Welty- Wolf, C. F. Emery, S. M. Palmer, and J. A. Blumenthal Six-Minute-Walk Distance and Accelerometry Predict Outcomes in Chronic Obstructive 26

27 Page 27 of 64 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC Pulmonary Disease Independent of Global Initiative for Chronic Obstructive Lung Disease 2011 Group. Ann.Am.Thorac.Soc. 12: Elia, M., P. Ritz, and R. J. Stubbs Total energy expenditure in the elderly. Eur.J.Clin.Nutr. 54 Suppl 3:S Luhrmann, P. M., R. Bender, B. Edelmann-Schafer, and M. Neuhauser- Berthold Longitudinal changes in energy expenditure in an elderly German population: a 12-year follow-up. Eur.J.Clin.Nutr. 63: Rothenberg, E. M., I. G. Bosaeus, and B. C. Steen Energy expenditure at age 73 and 78--a five year follow-up. Acta Diabetol. 40 Suppl 1:S134-S Dwyer, T., A. L. Ponsonby, O. C. Ukoumunne, A. Pezic, A. Venn, D. Dunstan, E. Barr, S. Blair, J. Cochrane, P. Zimmet, and J. Shaw Association of change in daily step count over five years with insulin sensitivity and adiposity: population based cohort study. BMJ 342:c Garcia-Aymerich, J., P. Lange, M. Benet, P. Schnohr, and J. M. Anto Regular physical activity modifies smoking-related lung function decline and reduces risk of chronic obstructive pulmonary disease: a population-based cohort study. Am.J.Respir.Crit Care Med. 175: Pelkonen, M., I. L. Notkola, T. Lakka, H. O. Tukiainen, P. Kivinen, and A. Nissinen Delaying decline in pulmonary function with physical activity: a 25-year follow-up. Am.J.Respir.Crit Care Med. 168:

28 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC Page 28 of Esteban, C., J. M. Quintana, M. Aburto, J. Moraza, M. Egurrola, J. Perez- Izquierdo, S. Aizpiri, U. Aguirre, and A. Capelastegui Impact of changes in physical activity on health-related quality of life among patients with COPD. Eur.Respir.J. 36: Pitta, F., T. Troosters, M. A. Spruit, V. S. Probst, M. Decramer, and R. Gosselink Characteristics of physical activities in daily life in chronic obstructive pulmonary disease. Am.J.Respir.Crit Care Med. 171: van Gestel, A. J., C. F. Clarenbach, A. C. Stowhas, V. A. Rossi, N. A. Sievi, G. Camen, E. W. Russi, and M. Kohler Predicting daily physical activity in patients with chronic obstructive pulmonary disease. PLoS.One. 7:e Spruit, M. A., M. L. Watkins, L. D. Edwards, J. Vestbo, P. M. Calverley, V. Pinto- Plata, B. R. Celli, R. Tal-Singer, and E. F. Wouters Determinants of poor 6-min walking distance in patients with COPD: the ECLIPSE cohort. Respir.Med. 104: Frisk, B., B. Espehaug, J. A. Hardie, L. I. Strand, R. Moe-Nilssen, T. M. Eagan, P. S. Bakke, and E. Thorsen Physical activity and longitudinal change in 6-min walk distance in COPD patients. Respir.Med. 108: Decramer, M., S. Rennard, T. Troosters, D. W. Mapel, N. Giardino, D. Mannino, E. Wouters, S. Sethi, and C. B. Cooper COPD as a lung disease with systemic consequences--clinical impact, mechanisms, and potential for early intervention. COPD. 5:

29 Page 29 of 64 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC 44. Hopkinson, N. S., R. C. Tennant, M. J. Dayer, E. B. Swallow, T. T. Hansel, J. Moxham, and M. I. Polkey A prospective study of decline in fat free mass and skeletal muscle strength in chronic obstructive pulmonary disease. Respir.Res. 8: van den Borst, B., A. Koster, B. Yu, H. R. Gosker, B. Meibohm, D. C. Bauer, S. B. Kritchevsky, Y. Liu, A. B. Newman, T. B. Harris, and A. M. Schols Is age-related decline in lean mass and physical function accelerated by obstructive lung disease or smoking? Thorax 66: Magnussen, H. and H. Watz Systemic inflammation in chronic obstructive pulmonary disease and asthma: relation with comorbidities. Proc.Am.Thorac.Soc. 6: Garcia-Aymerich, J., I. Serra, F. P. Gomez, E. Farrero, E. Balcells, D. A. Rodriguez, J. de Batlle, E. Gimeno, D. Donaire-Gonzalez, M. Orozco-Levi, J. Sauleda, J. Gea, R. Rodriguez-Roisin, J. Roca, A. G. Agusti, and J. M. Anto Physical activity and clinical and functional status in COPD. Chest 136: Handschin, C. and B. M. Spiegelman The role of exercise and PGC1alpha in inflammation and chronic disease. Nature 454: Agusti, A., L. D. Edwards, S. I. Rennard, W. MacNee, R. Tal-Singer, B. E. Miller, J. Vestbo, D. A. Lomas, P. M. Calverley, E. Wouters, C. Crim, J. C. Yates, E. K. Silverman, H. O. Coxson, P. Bakke, R. J. Mayer, and B. Celli Persistent 29

30 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC Page 30 of 64 systemic inflammation is associated with poor clinical outcomes in COPD: a novel phenotype. PLoS.One. 7:e Tashkin, D. P., B. Celli, S. Senn, D. Burkhart, S. Kesten, S. Menjoge, and M. Decramer A 4-year trial of tiotropium in chronic obstructive pulmonary disease. N.Engl.J.Med. 359: Tantucci, C. and D. Modina Lung function decline in COPD. Int.J.Chron.Obstruct.Pulmon.Dis. 7:

31 Page 31 of 64 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC Table 1 Baseline characteristics according to availability for follow-up assessment Available for follow-up Unavailable for follow-up* p value Number of patients Age, years 63.6 (6.3) 64.5 (7.4) 0.44 Men, n (%) 121 (74) 30 (81) 0.38 Current smokers, n (%) 73 (45) 18 (49) 0.67 Pack-years of smoking 52.8 (25.2) 48.9 (20.0) 0.38 Body mass index, kg/m (4.9) 25.8 (5.4) 0.35 FEV 1, L 1.92 (0.84) 1.66 (0.93) FEV 1 % predicted 64.8 (25.0) 54.1 (28.6) FEV 1 /FVC 0.53 (0.16) 0.48 (0.16) Preceding exacerbations, 2 per year, n (%) 29 (18) 8 (22) 0.59 Patients with CB / GOLD I / II / III / IV, n 26/30/51/29/27 4/4/6/14/9 N/A Physical activity level 1.57 (0.28) 1.36 (0.25) <0.001 Data are expressed as mean ± SD unless stated otherwise. Abbreviations: CB, chronic bronchitis; FEV 1, forced expiratory volume in one second; FVC, forced vital capacity; GOLD, Global Initiative for Chronic Obstructive Lung Disease stage. * During the follow-up period, 21 patients died, 14 withdrew, and 2 were lost to follow-up. The mean FEV 1 for COPD patients only was 58.3 (21.5) % predicted at baseline. 31

32 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC Page 32 of 64 Table 2 Physical activity at baseline and follow-up and the adjusted rate of change during 3 years Total cohort COPD Chronic Bronchitis Baseline assessment Follow-up assessment p value* Adjusted change Adjusted change Adjusted change p value Total daily energy expenditure, kcal 2708 (590) 2518 (536) < ( 296 to 121) 200 ( 295 to 105) 250 ( 457 to 43) 0.66 Activity energy expenditure, kcal 703 (441) 560 (384) < ( 239 to 86) 159 ( 242 to 76) 182 ( 362 to 1) 0.82 Physical activity level 1.57 (0.28) 1.48 (0.25) < ( to 0.057) ( to 0.049) ( to 0.007) 0.75 Minutes >3METs, min 135 (85) 115 (84) < ( 43 to 13) 27 ( 43 to 11) 31( 66 to 4) 0.84 Steps per day 6822 (3786) 5685 (3546) < ( 1583 to 581) 957 ( 1499 to 416) 1729 ( 2908 to 549) 0.23 Unadjusted values of the baseline and follow-up assessment for each physical activity parameter are expressed as mean ± SD. The mean adjusted decline was calculated as the difference between the marginal means calculated by an ANCOVA for repeated measures for the difference between each physical activity parameter between baseline and follow-up adjusting for age, sex, smoking status, exacerbations 2 per year, time to follow-up, delta daylight time, and baseline FEV 1. The changes are expressed as mean (95% CI). The total cohort consisted of 163 patients (137 with COPD and 26 with chronic bronchitis). * p values were calculated by paired t tests. p values were calculated for the effect of COPD vs. chronic bronchitis on the difference in each physical activity parameter calculated by ANCOVA for repeated measures adjusting for confounders. TDEE corrected for body weight decreased from 34.9 (7.3) kcal/kg to 33.0 (7.0) kcal/kg (p <0.001). AEE corrected for body weight decreased from 9.0 (5.6) kcal/kg to 7.4 (5.2) kcal/kg (p <0.001). 32

33 Page 33 of 64 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC Table 3 Lung function and systemic disease components at baseline and follow-up and the adjusted rate of change during 3 years Baseline assessment Follow-up assessment p value Adjusted change FEV 1, L 1.92 (0.84) 1.76 (0.85) < ( 216 to 121)* FEV 1 % predicted 64.8 (25.0) 60.8 (26.4) < ( 6.2 to 2.8) SGRQ, total score 41.1 (20.5) 40.9 (22.1) ( 4.5 to 0.9) 6-min walk distance, m 457 (110) 385 (146) < ( 78 to 42) Fat-free mass index, kg/m (2.4) 18.3 (2.5) < ( 0.74 to 0.33) Fat-free mass, kg 56.3 (10.4) 55.0 (10.6) < ( 2.1 to 0.9) Fibrinogen, mg/dl 424 (89) 458 (90) < (11.3 to 47.9) hs-crp, mg/l, median (IQR) 2.6 ( ) 3.0 ( ) N/A Unadjusted values of the baseline and follow-up assessment are expressed as mean ± SD unless stated otherwise. P values were calculated by paired t tests. The mean adjusted change was calculated as the difference between the marginal means calculated by an ANCOVA for repeated measures for the difference of each parameter between baseline and follow-up, adjusting for age, sex, smoking status, exacerbations 2 per year, time to follow-up, and baseline FEV 1. The change in FEV 1 was not adjusted for baseline FEV 1 as this was already one of the dependent variables. The changes are expressed as mean (95% CI). Mean change for hs-crp was not given owing to the skewness. Therefore, hs-crp was log-transformed to obtain a normal distribution. Data are given for the total cohort (n=163). There are missing values for: fat-free mass and fat-free mass index (n=7), fibrinogen (n=2), and hs-crp (n=1). Abbreviations: FEV1, forced expiratory volume in one second; hs-crp, high-sensitivity C-reactive protein; SGRQ, St. George's Respiratory Questionnaire. * The decline in FEV 1 for COPD patients only was ( 219 to 120) L, corresponding to an annual decline of 61mL/yr. Body weight decreased from 79.4 (17.8) kg to 78.3 (18.5) kg (p=0.009). 33

34 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC Page 34 of 64 Table 4 Adjusted effects of changes in lung function and systemic disease components on changes in PAL Parameter Estimate B 95% CI p value Annual change in FEV 1, per 100mL (0.005 to 0.038) 0.034* Annual change in SGRQ, per 10 point total score ( to 0.016) Annual change in 6MWD, per 10m ( to 0.007) 0.33 Annual change in FFM, per 1kg ( to 0.015) 0.89 Annual change in fibrinogen, per 1mg/dL ( to 0.001) 0.65 Annual change in log-hs-crp, per 1mg/L ( to 0.044) 0.87 P values of each row are based on separate ANCOVAs for repeated measures for the difference in physical activity level (i.e., baseline PAL and follow-up PAL as dependent variables) with the annual change in forced expiratory volume in one second (FEV 1 ), St. George's Respiratory Questionnaire (SGRQ) total score, 6-minute walk distance (6MWD), fat-free mass (FFM), fibrinogen and hs-crp level as the covariates and an adjustment for age, sex, smoking status, exacerbations 2 per year, delta daylight time, time to follow-up, and baseline FEV 1. Parameter estimates B (95% CI) are calculated by ANCOVAs with the annual decline in PAL as the dependent variable. History of coronary heart disease, diabetes mellitus, depression and osteoporosis were not associated with the changes in PAL bivariately (i.e., p>0.2). Data are given for the total cohort (n=163). There are missing values for: FFM (n=7), fibrinogen (n=2), and hs-crp (n=1). * Partial η (i.e., the annual decline in FEV 1 accounted for 2.9% variance of the decline in PAL, which is a small to moderate effect 28 ). Partial η (i.e., the annual increase in SGRQ accounted for 4.6% variance of the decline in PAL, which is a small to moderate effect 28 ). Replacing the SGRQ total score by the SGRQ activity score did not improve the association (p=0.078). 34

35 Page 35 of 64 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC Table 5 Adjusted effects of sustained physical inactivity on changes in COPD manifestations during 3 years Sustained physical inactivity (PAL T0andT1 <1.40) Some level of activity (PAL T0and/orT1 1.40) Effect of sustained inactivity on baseline value Effect of sustained inactivity on changes Mean (95% CI) Mean (95% CI) Adjusted p* Crude p Adjusted p n = 39 n = 124 Airflow obstruction FEV 1 at baseline, L 1.10 (0.86 to 1.35) 1.84 (1.68 to 2.00) <0.001 Change in FEV 1, L ( to 0.073) ( to 0.122) Health status SGRQ at baseline, total score 57.9 (51.6 to 64.1) 44.4 (40.6 to 48.1) <0.001 Change in SGRQ, total score 4.9 ( 9.8 to 0.1) 1.0 ( 3.9 to 2.0) Exercise tolerance 6MWD at baseline, m 376 (342 to 411) 448 (427 to 467) <0.001 Change in 6MWD distance, m 96 ( 127 to 64) 50 ( 69 to 31) < $ Muscle status FFM at baseline, kg 53.5 (50.7 to 56.2) 51.4 (47.7 to 53.1) 0.17 Change in FFM, kg 3.3 ( 4.3 to 2.3) 0.9 ( 1.5 to 0.3) <0.001 <0.001 ǁ Muscle depletion at baseline, n (%)** 6 (15.4) 10 (8.4) 0.21 Muscle depletion at follow-up, n (%)** 11 (28.2) 14 (11.5) N/A N/A Systemic inflammation Fibrinogen at baseline, mg/dl 433 (398 to 468) 412 (391 to 433)

36 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC Page 36 of 64 Change in Fibrinogen, mg/dl 52 (19 to 84) 23 (4 to 43) Log-hs-CRP at baseline, mg/l 0.97 (0.55 to 1.40) 0.71 (0.46 to 0.97) 0.26 Change in log-hs-crp, mg/l 0.35 ( 0.03 to 0.73) 0.04 ( 0.19 to 0.27) Effects of sustained physical inactivity vs some level of activity on the change in FEV 1, SGRQ total score, 6MWD, FFM, fibrinogen, and hs-crp levels were calculated by six separate ANCOVA for repeated measures with the baseline and the follow-up value as the dependent variables, respectively. Each model was adjusted for age, sex, smoking status, exacerbations 2 per year, time to follow-up, and baseline FEV 1. The change in FEV 1 was not adjusted for baseline FEV 1 as this was already one of the dependent variables. Estimated marginal means (95% CI) and the corresponding mean changes (95% CI) during 3 years are given unless stated otherwise. There are missing values for FFM and muscle depletion (n=7), fibrinogen (n=2), and hs-crp (n=1). Analyses of hs-crp were performed after log transformation to obtain a normal distribution. Abbreviations: FEV 1, forced expiratory volume in one second; FFM, fat-free mass; hs-crp, high-sensitivity C-reactive protein; 6MWD, 6-minute walk distance; PAL, physical activity level; SGRQ, St. George's Respiratory Questionnaire. * p values of the parameter estimate on the baseline value corrected for confounders. p values for the effect of sustained physical inactivity on changes in each variable over time without adjustment for confounders. p values for the effect of sustained physical inactivity on changes in each variable over time adjusted for confounders. FEV 1 remained a significant covariate for the change in SGRQ and 6MWD. Both significant models (i.e. the model for the change in 6MWD and FFM) remained unchanged after further adjustment for history of comorbidities. $ Partial η (i.e., sustained inactivity accounted for 4.5% variance of the decline in 6MWD, which is a small to moderate effect 28 ). ǁ Partial η (i.e., sustained inactivity accounted for 11.5% variance of the decline in FFM, which is a large effect 28 ). ** Frequency of muscle depletion is stated without an adjustment and the corresponding p values are calculated by Chi-square test for the difference between the groups at each measurement point. 36

37 Page 37 of 64 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC Figure legends Figure 1 Unadjusted physical activity parameters during the 3-year follow-up period according to baseline severity stages. Mean total daily energy expenditure (TDEE) (A), physical activity level (PAL) (B), and steps per day (C) at baseline (grey bars) and at follow-up (dark bars) are shown. The overall declines in TDEE, PAL, and steps per day showed strong significance (p <0.001), but were not related to GOLD stages (unadjusted p = 0.99, 0.99, and 0.82, respectively). After adjustment for confounders, the estimated mean annual decline from left to right was: (TDEE, in kcal/yr) 101, 78, 73, 76, and 51 (p = 0.87); (PAL) 0.056, 0.046, 0.038, 0.031, and (p = 0.68); (steps per day) 504, 393, 291, 435, and 461 (p = 0.85). Bars and whiskers represent means ± SE. Figure 2 Unadjusted change in FEV 1 (A) and SGRQ total score (B) according to the magnitude of change in physical activity level (PAL) during the 3-year follow-up period. The change in PAL was classified by every change into five categories, based on 0 ±0.025 as the reference. The number of patients per group were: increase (i.e., >0.025) 29 (18%), little change (i.e., decrease of to increase of ) 51 (31%), slow decline (i.e., to 0.074) 35 (21%), fast decline (i.e., to 0.124) 32 (20%), and very fast decline (i.e., greater decrease than 0.125) 16 (10%). Mean baseline FEV 1 (±SD) from left to right were: 70% (±26), 56% (±24), 71% (±26), 65% (±21), and 71% (±26). For multivariate analyses on these relations please refer to table 4. Bars and whiskers represent means ± SE. 38

38 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC Page 38 of 64 Figure 3 The unadjusted annual change in systemic disease manifestations of COPD and sustained physical inactivity. The unadjusted change over time in 6MWD (A), FFM (B), fibrinogen (C), and hs-crp (D) differed between patients with sustained physical inactivity (i.e., PAL T0andT1 <1.40; n = 39) and those with some level of activity (i.e., PAL T0and/orT1 1.40; n = 124) (p <0.001, <0.001, 0.060, and 0.065, respectively). After adjustment for confounders, including baseline FEV 1 and frequent exacerbations, the contribution of sustained inactivity on the change in 6MWD and FFM remained significant (see also table 5). Hs-CRP was log-transformed to reveal a normal distribution. Bars and whiskers represent means ± SE. 39

39 Page 39 of 64 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC

40 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC Page 40 of 64

41 Page 41 of 64 AJRCCM Articles in Press. Published on 28-May-2015 as /rccm OC