Annual change in FEV1 in elderly 10-year survivors with established chronic obstructive pulmonary disease.

Long-term decline in lung function is generally considered to be progressive in individuals with established chronic obstructive pulmonary disease (COPD), despite the presence of intersubject variation. We hypothesized that the annualized rate of decline in forced expiratory volume in 1 second (FEV1) would not be constant among different time periods in the natural history of established COPD. We compared the annual change rates in FEV1 during the first 5 years and the last 5 years, estimated separately using a linear mixed-effects model in 10-year survivors (n = 110). The subjects were classified into three FEV1 decline groups, based on the 25th and 75th percentile values in each time period. The rates of FEV1 changes, calculated from the first 5 years and the last 5 years, did not correlate with each other among 10-year survivors; the subjects of each FEV1 decline group during the first 5 years did not consistently remain in the same FEV1 decline group during the last 5 years. Smoking status and exacerbation frequency were not associated with decline in FEV1. In conclusion, the disease activity, which is often expressed as annualized change in FEV1, might be changeable either way over years in patients with established COPD.

Introduction

Chronic obstructive pulmonary disease (COPD) is a leading cause of morbidity and mortality worldwide[1]. Therefore, better understanding of the long-term natural history of COPD is needed to facilitate better management and intervention strategies for the disease. Since the pioneering studies by Fletcher and Peto and coworkers[2,3], forced expiratory volume in 1 second (FEV1) has been thought to deteriorate progressively in patients with COPD; notably, FEV1 determines disease severity based on airflow limitation[1]. In addition to disease severity, the concept of disease activity is important for characterizing COPD. Disease activity in COPD comprises several aspects, including frequent exacerbations or loss of quality of life; the rate of decline in FEV1 should be considered one of the most vital aspects of disease activity because airflow limitation is essential to the nature of COPD. Such progression of airflow limitation actually involves contribution of various pathological processes, such as inflammation in both small and large airways as well as emphysema progression. Indeed, annualized change in FEV1 has often been used as vital outcome measures in several landmark, large-scaled, randomized controlled trials[4,5].

Several observational cohort studies have recently shown wide interindividual variability with respect to changes in FEV1 among patients with COPD; notably, some patients have exhibited sustained lung function over several years[6-8]. Furthermore, the Lung Health Study and several general population-based studies have shown that FEV1 does not decrease in linear fashion; in contrast, its trajectories are variable[9-13]. However, it is unclear whether the long-term annualized rate of change in FEV1 is relatively constant or variable in individuals with established COPD that is appropriately managed. Thus far, there have been no studies that involved intraindividual comparison of the annualized rate of FEV1 decline across different time periods. We hypothesized that the annualized rate of FEV1 decline (i.e., the disease activity characterized by the rate of lung function decline) does not remain constant across different time periods during the clinical course of COPD.

In this study of 10-year survivors, we compared annual change rates in FEV1 during the first 5 years and the last 5 years; we used the data from the Hokkaido COPD cohort study, which was prospectively planned and carefully conducted with a high follow-up rate[8,14-16]. Our goal with regard to focusing on 10-year survivors was that we anticipated characterization of long-term individual lung function decline in patients with established COPD that was appropriately managed. In addition, annualized changes in FEV1 were calculated using separate spirometric data for 5-year’ periods in order to increase the accuracy of estimates for the annualized rate. Because exacerbation events were infrequent and were not associated with annualized decline patterns in FEV1 in the Hokkaido COPD cohort study[15], this cohort is suitable for examination of the natural history of lung function without modification by exacerbation events.

Results

Ten-year survivors in the Hokkaido COPD cohort study

Figure 1 shows the flow chart of subject selection for this study. Of 279 subjects with spirometry-confirmed COPD in the first year, 265 (95%) had mortality data during the 10-year follow-up period; 153 (55%) survived and 112 (40%) died. Among the 153 10-year survivors, 110 subjects (72%) had at least three valid spirometric measurements during the last 5 years (10-year survivors with good follow-up), whereas 43 subjects (28%) had less than three valid spirometric measurements (10-year survivors lost to follow-up). In this study, 10-year survivors with good follow-up (110 subjects) were analyzed; their characteristics, in comparison with other groups, are shown in Table 1. Ten-year survivors were significantly younger, had a higher body mass index (BMI), better lung function, and a lower CT emphysema score than non-survivors, regardless of spirometric follow-up status (Table 1). Exacerbation frequency during the first 5 years did not statistically differ among the three groups (Table 1).

Figure 1

Flow chart of subject selection.

Table 1

Characteristics of 10-year survivors and non-survivors.

	10-year survivors with good follow-up	10-year survivors lost to follow-up	10-year non-survivors	P value
Number of subjects Baseline variables	110	43	112
Age, years	66 ± 7*	68 ± 9*	74 ± 5	<0.001a
Female sex, N (%)	7 (6)	4 (9)	3 (3)	0.16c
BMI, kg/m2	23 ± 3*	23 ± 3*	21 ± 3	0.001a
Current smoker at entry, N (%)	30 (27)	15 (35)	28 (25)	0.46c
Smoking index at entry, pack-years	66 ± 32	69 ± 33	58 ± 25	0.07a
Lung function
Post-BD FEV1, L	1.91 ± 0.65*	1.81 ± 0.72	1.54 ± 0.61	<0.001a
Post-BD FEV1, % predicted	68 ± 20*	69 ± 27*	59 ± 21	0.004a
Post-BD FVC, % predicted	103 ± 16	103 ± 25	98 ± 18	0.15a
Post-BD FEV1/FVC	0.54 ± 0.13*	0.53 ± 0.13	0.48 ± 0.12	0.001a
Reversibility of FEV1, %	12 ± 11	9 ± 10	12 ± 11	0.25a
Reversibility of FEV1, mL	167 ± 110†	118 ± 96	135 ± 105	0.01a
DLco, % predicted	81 ± 23	80 ± 26	73 ± 26	0.07a
Kco, % predicted	70 ± 23*	63 ± 22	57 ± 24	<0.001a
Patient-reported outcomes
Chronic bronchitis, N (%)	12 (11)	5 (12)	12 (11)	1.00c
mMRC dyspnea score ≥ 2, N (%)	51 (46)	24 (56)	68 (61)	0.10c
SGRQ total score	30 ± 16	31 ± 18	34 ± 18	0.13a
Laboratory values
Blood neutrophil count, cells/mm3	3394 (2629–4038)	3479 (2967–4135)	3505 (2736–4054)	0.65b
Blood eosinophil count, cells/mm3	176 (101–308)	159 (116–244)	160 (87–250)	0.34b
Serum total IgE, IU/mL	82 (23–223)	65 (20–119)	63 (19–180)	0.28b
CT emphysema score	1.0 (0.5–1.7)*	1.3 (0.5–2.2)	1.5 (0.8–2.3)	<0.001b
Comorbidities
Any cardiovascular disease	22 (20)	8 (19)	29 (26)	0.50c
Ischemic heart disease	8 (7)	2 (5)	7 (6)	0.89c
Diabetes	5 (5)	3 (7)	5 (4)	0.80c
Charlson risk index	0(0–1)	0(0–0)	0(0–1)	0.29b
Longitudinal variables (first 5 years) Exacerbation frequency, events/y	0 (0–0.20)	0(0–0)	0 (0–0.40)	0.09b

14 subjects whose 10-year survival data were absent were not included in this table.

Data are shown as means ± SD, median (interquartile range), or number (%).

Post-BD = post-bronchodilator; DLco = carbon monoxide diffusion capacity; Kco = carbon monoxide transfer coefficient.

Reversibility of FEV1, % refers to % change in FEV1 by inhalation of bronchodilator.

aOne-way analysis of variance, bKruskal-Wallis test, cFisher’s exact test.

*p < 0.05 vs. 10-year non-survivors, †p < 0.05 vs. 10-year survivors lost to follow-up (Tukey’s HSD test or Mann-Whitney U test).

Ten-year lung function change in subjects with COPD

Among 10-year survivors with good follow-up, the median number of spirometric assessments per subject was 16 (IQR: 15-16) during the entire 10-year period, 11 (IQR: 11-11) during the first 5 years, and 5 (IQR: 5-5) in the last 5 years, respectively. Their calculated annual change in post-bronchodilator FEV1 was −31 ± 21 mL/year during the entire 10-year period (“rapid decliners”: −57 ± 15 mL/year, “slow decliners”: −31 ± 9 mL/year, “sustainers”: −6 ± 11 mL/year), −28 ± 25 mL/year during the first 5 years (rapid decliners: −61 ± 11 mL/year, slow decliners: −29 ± 10 mL/year, sustainers: 4 ± 11 mL/year), and −30 ± 28 mL/year during the last 5 years (rapid decliners: −62 ± 16 mL/year, slow decliners: −31 ± 9 mL/year, sustainers: 3 ± 22 mL/year) (Fig. S1A–C). Of note, there was no significant correlation between annual changes in post-bronchodilator FEV1 during the first 5 years and the last 5 years (r = 0.04, p = 0.66) (Fig. 2A). In addition, the subjects of each FEV1 decline group during the first 5 years were recategorized to any of the three decline groups during the last 5 years (Fig. 2B). Chronologic changes in the mean FEV1, and those expressed as percent changes from baseline in each FEV1 decline group during the 10-year period, are shown in Fig. 3 (FEV1 decline group during the first 5 years) and Fig. S2 (FEV1 decline group during the last 5 years). In particular, rapid decliners during the first 5 years showed much slower FEV1 decline after 5 years; their FEV1 reached levels similar to those of slow decliners after 8 years of follow-up (Fig. 3). When GOLD airflow limitation grade at baseline was considered, rapid decliners during the first 5 years who were GOLD 2 or more at baseline showed slower FEV1 decline after 5 years; those who were GOLD 1 continued to exhibit FEV1 decline after 5 years (Fig. S3). These results clearly indicate that the rate of lung function decline can change in either direction, regardless of GOLD stage, in the natural history of COPD.

Figure 2

Correlations between annual changes in FEV1 calculated from the first 5 years and the last 5 years among 10-year survivors with good follow-up (n = 110). (A) Correlation between annual changes in FEV1 of the first 5 years (0–5 y) vs. the entire 10 years (0–10 y). (B) Bar plots of the number of subjects in each FEV1 decline group for the first 5 years and the last 5 years. The bars on the left show the number of subjects in each FEV1 group (rapid decliners, slow decliners, and sustainers) for the first 5 years. The bars on the right show the respective numbers of subjects in each FEV1 group for the last 5 years.

Figure 3

10-year annual change in FEV1 among FEV1 decline groups for the first 5 years. (A) Mean post-bronchodilator FEV1 (with SEM) expressed as absolute values. (B) Mean post-bronchodilator FEV1 (with SEM) expressed as percent changes from baseline.

Next, we investigated factors associated with the rate of FEV1 decline. Among 10-year survivors with good follow-up, rapid FEV1 decline during the entire 10-year period was associated with lower BMI, lower diffusion capacity, and a lower blood eosinophil count at baseline (Table S1). Notably, the degree of FEV1 decline during the last 5 years was not associated with any variables at the 5-year follow-up (Table S2). Smoking status and exacerbation frequency during the first 5 years were similar among the FEV1 decline groups in each condition (Tables S1 and S2).

Discussion

Progressive airflow limitation (reduction of FEV1) is an essential feature of COPD, and the percent predicted value for FEV1 determines disease severity as a result of small airway disease and emphysema[1]. Assessment of FEV1 as an index of disease severity is important for predicting the future clinical course of COPD patients, including morbidity and mortality at the population level[1]. One of the primary treatment goals for COPD patients is effective prevention of disease progression. To achieve this goal, assessment of disease activity is very important. Both disease severity and activity are associated with pathological processes but they are distinct concepts, and disease severity cannot be a surrogate for disease activity[14]. The concept of disease activity is essential for the management of chronic inflammatory diseases, such as rheumatoid arthritis and inflammatory bowel diseases[15,16]; however, individual chronological changes in disease activity based on the rate of lung function decline in established COPD patients have not been fully investigated. Because the rate of FEV1 decline is regarded as a vital measure of COPD disease activity, annualized changes in FEV1 were compared during different time periods in the same individuals with COPD. In the present study, annualized changes in FEV1 were calculated using separate spirometric data for 5-year periods, in order to increase the accuracy of estimates for the annualized rate. Notably, measurement of the true rates of FEV1 decline requires numerous time points over many years of follow-up, due to the relatively high variability in FEV1. Furthermore, the inter-subject variation in calculated rates of annual decline in FEV1 is smaller with longer follow-up periods[17]. In the present study, spirometric measurements were performed every 6 months during the first 5 years and every year afterward, based on scheduled visits; indeed, the median number of spirometric assessments was 16 during the 10-year period. Furthermore, we confirmed that the subjects were clinically stable at each visit.

In agreement with our hypothesis, the rates of FEV1 changes calculated from the first 5 years and from the last 5 years were not correlated among 10-year survivors; the subjects of each FEV1 decline group during the first 5 years did not consistently remain in the same FEV1 decline group during the last 5 years (Fig. 2A,B). These findings indicate that COPD disease activity, defined by the rate of decline in lung function, is likely to change in each individual over a long-term period. Previous clinical trials showed that patients with milder airflow limitation had more rapid average decline in FEV1, compared with those with more severe airflow limitation[18]; this suggests that the loss of lung function is more accelerated in the initial phases of COPD. The present study extends this finding by showing that such variable rates of lung function decline were present in individuals at each GOLD grade (i.e., disease severity) (Fig. S3). Disease severity in each COPD patient must be a result of different lung function trajectories, including reduced lung function in early adulthood; accelerated decline in FEV1 is not an obligatory feature of COPD[12,13]. Therefore, we must consider disease severity, as well as disease activity (characterized by the rate of lung function decline), in the management of COPD.

In the present study, lower BMI, lower diffusion capacity, and lower blood eosinophil count at baseline were associated with a rapid decline in FEV1 during the entire 10-year period (Table S1), which is consistent with our previous report based on the data from the first 5 years[8]. Of note, in the 10-year survivors, none of the variables at the 5-year follow-up was associated with the rate of FEV1 decline during the last 5 years (Table S2). This observation suggests that the factors associated with the rate of lung function decline may change depending on the time point during the clinical course of COPD in 10-year survivors. For example, patients with severe emphysema who experienced rapid FEV1 reduction during the first 5 years would not exhibit further reduction during the last 5 years because of their low absolute FEV1 values. In this case, patients with severe emphysema would resemble rapid decliners during the first 5 years, and would then resemble sustainers during the last 5 years. Another explanation might be that unknown risk factors affect future lung function decline in 10-year survivors.

Importantly, exacerbation frequency was not associated with future lung function decline in the present study. The reduced exacerbation frequency in the Hokkaido COPD cohort study, compared with some large-scale clinical trials in the Western countries, may be responsible for the absence of a significant association between exacerbation frequency and lung function decline[19]. Alternatively, exacerbation may represent a different aspect of COPD disease activity, independent of natural lung function decline. We previously reported that the neutrophil elastase-alpha1-protenase inhibitor (NE-alpha1PI) complex in bronchoalveolar lavage fluid was markedly elevated in asymptomatic smokers who had subclinical emphysema on CT scans[20]; furthermore, higher levels of the NE-alpha1PI complex were associated with accelerated decline in FEV1[21]. This suggests that lung function decline could be affected by local proteolytic conditions in the lung without apparent exacerbation events. Of course, exacerbation history and symptoms are important measures for COPD disease activity, as emphasized in current GOLD guidelines[1]; however, the present study clearly indicates that the rate of lung function decline should be assessed as an important metric of COPD disease activity.

In the present study, smoking status at baseline and during the first 5 years was not associated with future changes in lung function; in contrast, the Lung Health Study showed that smoking cessation reduced FEV1 decline in smokers with mild airway limitation[9,10]. This discrepancy may be due to differences in study population characteristics. In the present study, current smokers at baseline comprised only 27% of the study population; therefore, the effect of smoking cessation on FEV1 decline would be much weaker than in the Lung Health Study.

Although the sample size in this study was small relative to that of several previous large-scale observational cohort studies, the strongest point was that it was very carefully designed and performed; thus, it was possible to collect very accurate annual spirometric data for a 10-year period with a very low dropout rate. However, this study has some other limitations. First, all subjects were Japanese; therefore, future studies with subjects of other ethnic backgrounds are warranted. Second, exacerbation data, which might affect lung function decline, was not collected during the last 5 years. Third, subjects were at different stages during their natural histories of COPD at the time of enrollment (staggered entry). This limitation is inevitable because of the nature of the observational cohort study; however, we followed subjects for a sufficient period of time (10 years) and found that the chronological changes in the rate of lung function decline were present among all spirometric GOLD grades. Lastly, there may be a survivorship bias in the analysis of this study because only 10-year survivors with good follow-up were analyzed. We acknowledge this limitation; however, we firmly believe that focusing on 10-year survivors alone is valuable and provides clues regarding the long-term natural history of established COPD.

In summary, the disease activity, which is often expressed as annualized change in FEV1, may change in either direction in the natural history of patients with established COPD, regardless of spirometric GOLD stage. These findings challenge the current concept that COPD is a disease with a natural history characterized by relatively constant progressive decline in FEV1. In addition to disease severity, disease activity (characterized by the rate of lung function decline) should also be considered in the management of COPD, as well as within any clinical research regarding COPD.

Methods

Participants

The cohort details have been described elsewhere[8,19,22,23]. Briefly, a total of 330 Japanese patients with COPD were recruited at Hokkaido University Hospital, Sapporo, Japan, and its nine affiliated hospitals from May 2003 to May 2005. Subjects with clinically-diagnosed asthma were meticulously excluded[23]. During the first follow-up year (visits 1–3), the diagnosis was reconfirmed based on the spirometric criteria of the Global Initiative for Chronic Obstructive Lung Disease (GOLD) guidelines (a ratio of post-bronchodilator FEV1 to FVC < 0.70)[1]. As a result, a total of 279 subjects with COPD (GOLD 1, 26%; GOLD 2, 45%; GOLD 3, 24%; and GOLD 4, 5%) were eligible for subsequent follow-up. The Ethics Committee of Hokkaido University School of Medicine approved the study protocol (med02-001) and written informed consent was obtained from all participants. This study was performed in accordance with the Declaration of Helsinki.

Study protocol

Until the fifth year, information regarding COPD exacerbations was collected every month. Spirometry before and after short-acting bronchodilator inhalation was performed every 6 months after confirming withdrawal of any long-acting bronchodilators. Diffusion capacity testing, chest computed tomography (CT), and a health-related quality of life assessment using St. George’s Respiratory Questionnaire (SGRQ) were performed each year. The severity of emphysema on chest CT was assessed visually[8,22,24]. Exacerbation was defined as worsening or new-onset of respiratory symptoms that required changes in prescription treatment[19].

After the fifth year, spirometry after bronchodilator inhalation and diffusion capacity testing was performed every year for those who agreed with the extension of the scheduled follow-up program until the tenth year. Spirometry data after the diagnosis of lung cancer or any major comorbidities, which might interfere with the natural history of COPD, were not used in this study. The majority of the subjects continued to visit outpatient clinics for appropriate medical care, even if they dropped out from the scheduled follow-up program of this study. Therefore, a telephone interview and/or medical chart review to monitor annual mortality data over 10 years was required in only a minority of the study population.

Statistical analysis

Annual changes in post-bronchodilator FEV1 during the first 5 years, the last 5 years, and the entire 10-year period were estimated separately using a linear mixed-effects model. In each time period, the subjects were classified into three FEV1 decline groups based on the magnitude of the annual change in FEV1: rapid decliners, those in the less than 25th percentile group; slow decliners, those in the 25th to 75th percentile group; and sustainers, those in the greater than 75th percentile group[8]. Correlations between annual changes in FEV1 during the two time periods were analyzed using Pearson’s correlation coefficient. Differences among the groups were analyzed using one-way analysis of variance, the Tukey’s HSD test, the Kruskal-Wallis test, the Mann-Whitney U test, or Fisher’s exact test, as appropriate. Statistical significance was defined as p < 0.05. All analyses were performed using R version 3.1.2 (The R Foundation, http://www.r-project.org/).

Online data supplement