Disproportionally Impaired Diffusion Capacity Relative to Airflow Limitation in COPD

Abstract

Forced expiratory volume in 1 s (FEV₁) is a standard physiological index of chronic obstructive pulmonary disease (COPD), but reflects emphysema and vascular abnormalities less sensitively than diffusion capacity for carbon monoxide (D_LCO). This study tested whether a disproportionally impaired D_LCO relative to FEV₁ (FEV₁ z-score>-3 and D_LCO z-score≤-3) is a common functional COPD phenotype associated with distinct clinical and structural features and the prognosis of two cohorts. The cross-sectional analyses of the Korea COPD Subgroup Study (KOCOSS) cohort (multicenter study in Korea) included 743 males with COPD whose D_LCO was available. The cross-sectional and longitudinal analyses of the Kyoto University Cohort (single-center study in Japan) included 195 males with COPD who were prospectively followed for 10 years. A disproportionally impaired D_LCO relative to FEV₁ was observed in 29% and 31% of patients in the KOCOSS and Kyoto University cohorts, respectively. In the multivariable analysis, the disproportionally impaired D_LCO was associated with worse symptoms, shorter 6-minute walking distance, paraseptal and centrilobular emphysema on computed tomography, and reduced arterial oxygen and carbon dioxide pressures compared to the reference (FEV₁ z-score>-3 and D_LCO z-score>-3). In the multivariable Cox proportional hazard model, a higher long-term mortality was observed in the disproportionally impaired D_LCO group than in the reference group (hazard ratio [95% confidence interval] = 3.09 [1.52–6.29]) and similar to the D_LCO z-score≤-3 and FEV₁ z-score≤-3 group. The disproportionally impaired D_LCO relative to FEV₁ is common and associated with increased symptoms, emphysema, arterial blood gas abnormalities, and increased long-term mortality in patients with COPD.

Keywords:

• Chronic obstructive pulmonary disease
• emphysema
• airway
• computed tomography
• pulmonary function
• prognosis

Introduction

Chronic obstructive pulmonary disease (COPD) is a leading cause of death worldwide [1]. While the diagnosis of COPD is simply based on airflow limitation on spirometry [2], spirometry is insufficient to capture the heterogeneous structural alterations underlying the clinical manifestations, including airway disease, emphysema, and vascular abnormalities [3].

The single-breath lung diffusion capacity for carbon monoxide (D_LCO) is a noninvasive, repeatable physiological measure of the capacity of gas exchange in the alveolar space of the lungs.[4,5]. D_LCO is closely associated with emphysema measured on histology[6,7] and computed tomography (CT) [8], as well as vascular abnormalities on CT [9–11]. Moreover, a lower D_LCO is associated with a lower arterial partial pressure of oxygen (PaO₂), exercise capacity, and poor prognosis in patients with COPD [12–14]. Even in smokers with normal spirometry, D_LCO may be decreased, and the decreased D_LCO is associated with more severe symptoms and impaired exercise capacity [15] and predicts the future development of COPD [16]. Furthermore, Balasubramanian et al. [17] recently proposed the categorization of patients with COPD based on a combination of forced expiratory volume in 1 s (FEV₁) on spirometry and D_LCO, and showed that an impaired D_LCO (≤50% of predicted) has negative effects on symptoms, exercise capacity, and exacerbation frequency, even in patients without a substantial reduction in FEV₁ (>50% of predicted). These findings suggest that functional phenotyping based on FEV₁ and D_LCO may improve clinical COPD management. However, the detailed structure–function relationships and even long-term prognosis in relation to this phenotyping remain to be explored.

A disproportionally impaired D_LCO relative to FEV₁ was hypothesized to be a common functional phenotype associated with the distinct clinical manifestations, structural changes, and prognosis of COPD. This study aimed to identify patients with COPD presenting a disproportionally impaired D_LCO relative to FEV₁ in two observational cohorts: the Kyoto University Cohort (single-center study in Japan) [8,18] and the Korea COPD Subgroup Study (KOCOSS) Cohort (multicenter study in Korea) [19,20]. Furthermore, this study tested whether this functional phenotype was associated with impairments in patient-reported outcomes and exercise capacity in the KOCOSS Cohort, and with a greater severity of emphysema on CT, abnormal arterial oxygen and carbon dioxide pressures, and increased long-term mortality in the Kyoto University Cohort.

Methods

Study design

This study consisted of the following datasets from two independent cohorts: the cross-sectional data from the KOCOSS Cohort and the cross-sectional and longitudinal data from the Kyoto University Cohort. The KOCOSS Cohort was obtained from a multicenter prospective observational study conducted at 48 tertiary referral hospitals in the Republic of Korea beginning in 2011 [19,20]. The study protocol was approved by the Institutional Review Board of Konkuk University Medical Center (Institutional Review Board No. 177 KHH1010338), and all the hospitals obtained approval from the Institutional Review Board committee. The Kyoto University Cohort is a single-center prospective observational study that has been conducted at the Kyoto University Hospital in Japan since 2006 using a single CT scanner with the fixed scanning conditions described below [8,18,21]. The study was performed in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of Kyoto University (approval Nos. E182 and R1660-1). All participants in both the Kyoto University and KOCOSS cohorts provided written informed consent. The collaborative analysis of the two cohorts was further approved by the Ethics Committee of Kyoto University (approval No. R2033).

The inclusion criteria of this study as follows: (1) age 40–85 years with a smoking history of at least 10 pack-years, (2) a physician’s diagnosis of COPD based on patient-reported respiratory symptoms and the presence of airflow limitation confirmed by a postbronchodilator FEV₁/forced vital capacity (FVC) ratio below the lower limit of normal (LLN), and (3) availability of postbronchodilator spirometry and D_LCO. D_LCO was adjusted by the blood hemoglobin level according to a previous report [22]. The LLN of FEV₁/FVC and z-scores and reference values of FEV₁ and FVC were obtained based on the “other” ethnic group data provided by the Global Lung Function Initiative (GLI) 2012 [23]. The z-scores and reference values of D_LCO were also calculated using the GLI calculation system [4]. Patients with a history of lung resection surgery or other lung diseases, such as interstitial lung disease and those with alpha-1 antitrypsin deficiency, were excluded. Because the majority (> 90%) of patients enrolled in the two cohorts were male, female patients were also excluded.

In the KCOSS Cohort, patient-reported outcomes, including the mMRC dyspnea scale, COPD assessment test (CAT), and St. George's Respiratory Questionnaire (SGRQ) [24, 25], and exercise capacity as assessed by the 6-minute walking distance (6MWD) were cross-sectionally evaluated.

In the Kyoto University Cohort, the residual volume (RV), RV to total lung capacity (TLC) ratio (RV/TLC), mMRC, emphysema and airway diseases on inspiratory CT and arterial blood gases measured in room air, including PaO₂ and partial pressure of carbon dioxide (PaCO₂) at baseline, were cross-sectionally evaluated. The CO transfer coefficient (Kco) that corresponds to D_LCO divided by alveolar volume (V_A) was also measured. Furthermore, longitudinal follow-up survival data available as of October 2019 from the Kyoto University Cohort were evaluated.

Chest CT

All subjects in the Kyoto University Cohort underwent full-inspiratory CT with a peak kilovoltage of 120, a 0.5-second exposure time, and autoexposure control using an Aquilion 64 scanner (Cannon Medical, Tokyo, Japan). Images with a 0.5-mm slice thickness were reconstructed with a high-spatial frequency algorithm (FC56). Using a SYNAPSE VINCENT volume analyzer (FUJIFILM Medical, Tokyo, Japan), the percentage of low attenuation regions less than −950 HU to the total lung regions (LAV%) was calculated to evaluate emphysema [26–28]. The wall area percentage (WA%), which was defined as the percentage of the wall area relative to the sum of the wall and lumen areas, was measured for the right apical and posterior basal segmental bronchus and averaged to evaluate airway disease [29–31]. Mild and substantial paraseptal emphysema (PSE), and mild and substantial (moderate to advanced) centrilobular emphysema (CLE) were visually identified based on the Fleischner Society classification system [32]. The inter-rater variability of two pulmonologists (NT and HS) was excellent (kappa = 0.80 and 0.76 for the PSE and CLE evaluations). Substantial PSE and CLE were considered to indicate the presence of PSE and CLE in this study. In addition, the ratio of the pulmonary artery diameter to the aorta diameter (PA/Ao) was obtained by manually measuring the pulmonary and aorta diameters [33].

Statistics

The data are reported as means ± SD, unless indicated otherwise. Statistical analyses were performed with the R program [34]. A p-value less than .05 was considered statistically significant. Based on the z-scores of FEV₁ and D_LCO [35], the patients were categorized into the following four groups: (1) FEV₁ z-score > −3 and D_LCO z-score > −3 (reference), (2) FEV₁ z-score > −3 and D_LCO z-score ≤ −3 (disproportionally impaired D_LCO), (3) FEV₁ z-score ≤ −3 and D_LCO z-score > −3 (disproportionally impaired FEV₁), and (4) FEV₁ z-score ≤ −3 and D_LCO z-score ≤ −3 (mixed-impaired). Tukey’s method was used to compare the variables among the four groups. Multivariable linear regression and Cox proportional hazard models were constructed and adjusted for age, height, weight, and smoking pack-years to examine the effects of the disproportionally impaired D_LCO, disproportionally impaired FEV₁, and mixed-impaired groups on the clinical measures and long-term outcome in comparison with the reference group. Furthermore, similar analyses were performed by defining the four groups using a cutoff of 50% for the % of predicted FEV₁ and D_LCO.

Results

Figure 1 shows patient flowcharts for the two cohorts. In the KOCOSS Cohort, 743 male patients whose hemoglobin-adjusted D_LCO was available and FEV₁/FVC was below the LNN were included in the cross-sectional analysis. In the Kyoto University Cohort, of the 253 stable patients with COPD enrolled from January to December 2012, 195 male patients with an FEV₁/FVC below the LNN were included in the cross-sectional and longitudinal analyses. Table 1 shows the basic clinical data of the two cohorts.

Figure 1. Patient flowcharts.

A. The KOCOSS Cohort was cross-sectionally analyzed. B. The Kyoto University Cohort was cross-sectionally and longitudinally analyzed.

Table 1. Demographics of patients in the two cohorts.

	KOCOSS	Kyoto University
N	743	195
Age (years)	68.9 (7.6)	69.9 (8.5)
Male (%)	100%	100%
Smoking pack-years	47.2 (24.2)	67.2 (34.7)
Height (cm)	165.4 (5.6)	164.5 (6.1)
Weight (kg)	62.8 (9.6)	59.8 (9.3)
Body mass index	22.9 (3.1)	22.1 (2.9)
FEV1 (% predicted)	62.4 (19.2)	61.6 (19.7)
FVC (% predicted)	101.9 (19.2)	101.0 (18.4)
FEV1 (z-score)	–2.2 (1.1)	–2.2 (1.1)
FVC (z-score)	0.1 (1.3)	0.0 (1.3)
FEV1/FVC	0.48 (0.11)	0.48 (0.12)
DLCO (% predicted)	58.1 (19.1)	58.0 (21.9)
DLCO (z-score)	–3.0 (1.6)	–3.0 (1.9)
mMRC ≥ 2 (%)	34%	25%

Data are reported as means (SD). FEV₁ = forced expiratory volume in 1 s. FVC = forced vital capacity. D_LCO = diffusion capacity for carbon monoxide. mMRC = modified MRC dyspnea scale.

As shown in Figure 2, approximately 47%, 30%, 5%–10%, and 16% of the patients were categorized into the reference, disproportionally impaired D_LCO, disproportionally impaired FEV₁, and mixed-impaired groups in both cohorts, respectively (n = 351/212/62/118 in the KOCOSS Cohort, and n = 89/62/10/34 in the Kyoto University Cohort).

Figure 2. Distributions of FEV₁ and D_LCO in the two cohorts.

A. KOCOSS Cohort. B. Kyoto University Cohort. Patients were categorized into four groups: (1) FEV₁ z-score > –3 and D_LCO z-score > –3 (reference, red), (2) FEV₁ z-score > –3 and D_LCO z-score ≤ –3 (disproportionally impaired D_LCO, green), (3) FEV₁ z-score ≤ –3 and D_LCO z-score > –3 (disproportionally impaired FEV₁, blue), and (4) FEV₁ z-score ≤ –3 and D_LCO z-score ≤ –3 (mixed-impaired, purple).

The cross-sectional analysis of the KOCOSS Cohort showed that age, smoking pack-years, mMRC ≥ 2, CAT, and the SGRQ scores were higher, while the BMI and 6-minute walking distance were lower in the disproportionally impaired D_LCO group, as shown in Table 2. In the multivariable analysis shown in Figure 3, compared to the reference group, the disproportionally impaired D_LCO was significantly associated with higher mMRC, CAT, and SGRQ scores and a lower 6MWD.

Figure 3. Associations of D_LCO, FEV₁, and both impairments with patient-reported outcomes and exercise capacity in a multivariable analysis of the KOCOSS cohort.

Patients (n = 743) were categorized into four groups: (1) FEV₁ z-score > –3 and D_LCO z-score > –3 (reference, n = 351), (2) FEV₁ z-score > –3 and D_LCO z-score ≤ –3 (disproportionally impaired D_LCO, n = 212), (3) FEV₁ z-score ≤ –3 and D_LCO z-score > –3 (disproportionally impaired FEV₁, n = 62), and (4) FEV₁ z-score ≤ –3 and D_LCO z-score ≤ –3 (mixed-impaired, n = 118). A dot with an error bar indicates the least-square mean (LS mean) with the 95% CI. * p<.05 compared to the reference group in the multivariable models. Each model was adjusted for age, pack-years of smoking, height and weight. 6MWD = six-minute walking distance. CAT = COPD assessment test. SGRQ = St. George's Respiratory Questionnaire. 6MWD, CAT, and SGRQ data were available for 641, 717, and 395 patients, respectively.

Table 2. Clinical characteristics of the four groups in the KOCOSS Cohort.

	Reference	Disproportionally impaired DLCO	Disproportionally impaired FEV1	Mixed-impaired	P
N	351	212	62	118
Age (years)	69.0 (7.6)	70.4 (7.6)	65.7 (7.1) ^†	67.6 (7.3)	<.01
Smoking PY	44.2 (22.7)	51.3 (25.6)^†	45.3 (20.2)	49.4 (26.9)	<.01
BMI	23.9 (2.8)	22.1 (3.0)^†	23.3 (3.1)	21.2 (3.0)^†	<.01
FEV1 (z-score)	–1.7 (0.8)	–1.9 (1.2)^†	–3.4 (0.4)^†	–3.6 (0.4)^†	<.01
FVC (z-score)	0.4 (1.1)	0.5 (1.1)	–1.0 (1.2)^†	–0.9 (1.2)^†	<.01
DLCO (z-score)	–1.8 (0.8)	–4.3 (1.2)^†	–2.2 (0.7)^†	–4.7 (1.2)^†	<.01
mMRC ≥ 2 (%)	21%	34%^†	55%^†	59%^†	<.01
6MWD* (m)	431 (93)	404 (106)^†	417 (77)	361 (108)^†	<.01
CAT*	12.9 (7.2)	15.6 (8.4)^†	17.8 (7.4)^†	19.6 (8.1)^†	<.01
SGRQ total*	27.2 (17.0)	40.4 (18.5)^†	39.4 (18.5)^†	44.6 (20.1)^†	<.01
Symptom*	35.9 (18.1)	48.3 (20.1)^†	51.6 (19.9)^†	49.7 (20.8)^†	<.01
Activity*	38.3(21.9)	52.9 (21.3)^†	52.1 (22.3)^†	59.3 (23.5)^†	<.01
Impact*	18.4 (17.7)	30.9 (20.9)^†	28.3 (19.8)^†	34.5 (22.1)^†	<.01

Data are presented as means (SD). All subjects were male. Smoking PY = smoking pack-years. BMI = body mass index. FEV₁ = forced expiratory volume in 1 s. FVC = forced vital capacity. RV/TLC = ratio of residual volume to total lung capacity. D_LCO = diffusion capacity for carbon monoxide. 6MWD = six-minute walking distance. CAT = COPD assessment test. SGRQ = St. George's Respiratory Questionnaire. Symptom, Activity, and Impact were the domains of the SGRQ score. *6MWD, CAT, and SGRQ data were available for 641, 717, and 395 patients, respectively. P = p-value. ^†p < .05 compared to the reference group based on Tukey’s multiple comparison or multiple Fisher’s exact tests followed by Bonferroni correction.

The cross-sectional analysis of the baseline data from the Kyoto University Cohort presented in Table 3 showed that age, an mMRC ≥ 2, the prevalence of visual CT findings of CLE and PSE, and LAV% were higher, while the PaO₂ and PaCO₂ were lower in the disproportionally impaired group than in the reference group. WA% and PA/Ao on CT did not significantly differ among the groups. In the multivariable analysis shown in Figure 4, the rates of both PSE and CLE were higher and PaO₂ and PaCO₂ were lower in the disproportionally impaired D_LCO group than in the reference group. In contrast, the rates of PSE and PaCO₂ in the disproportionally impaired FEV₁ and mixed-impaired groups did not significantly differ from those in the reference group.

Figure 4. Associations of D_LCO, FEV₁, and both impairments with emphysema subtypes and arterial blood gases in a multivariable analysis of the Kyoto University Cohort.

Patients (n = 195) were categorized into four groups: (1) FEV₁ z-score > –3 and D_LCO z-score > –3 (reference, n = 89), (2) FEV₁ z-score > –3 and D_LCO z-score ≤ –3 (disproportionally impaired D_LCO, n = 62), (3) FEV₁ z-score ≤ –3 and D_LCO z-score > –3 (disproportionally impaired FEV₁, n = 10), and (4) FEV₁ z-score ≤ –3 and D_LCO z-score ≤ –3 (mixed-impaired, n = 34). (A) Odds ratio for the presence of paraseptal emphysema and centrilobular emphysema on CT. A dot with an error bar indicates the regression coefficient with the 95% CI. (B) Least-square mean (LS mean) with the 95% CI for the partial pressure of oxygen (PaO₂) and partial pressure of carbon dioxide (PaCO₂). * p<.05 compared to the reference group in the multivariable models. Each model was adjusted for age, pack-years of smoking, height and weight. PaO₂ and PaCO₂ data were available for 184 patients.

Table 3. Clinical characteristics of the four groups in the Kyoto University Cohort.

	Reference	Disproportionally impaired DLCO	Disproportionally impaired FEV1	Mixed-impaired	P
N	89	62	10	34
Age (years)	68.8 (9.0)	73.6 (6.4)^†	67.5 (5.1)	66.9 (9.2)	<.01
Smoking PY	65.2 (34.8)	72.6 (35.9)	70.4 (39.4)	61.7 (30.5)	.43
BMI	23.0 (2.7)	21.6 (3.1)^†	21.3 (2.4)	20.8 (2.7)^†	<.01
FEV1 (z-score)	–1.6 (0.9)	–2.0 (0.8)^†	–3.5 (0.4) ^†	–3.7 (0.6) ^†	<.01
FVC (z-score)	0.3 (1.1)	0.4 (1.0)	–0.7 (1.4)^†	–1.2 (1.2)^†	<.01
RV/TLC (%)	39.2 (6.4)	42.3 (7.9)^†	48.0 (11.2)^†	48.8 (6.1)^†	<.01
DLCO (z-score)	–1.6 (0.9)	–4.2 (1.0)^†	–1.7 (0.9)	–5.1 (1.5)^†	<.01
Kco (z-score)	–1.0 (1.2)	–3.5 (1.0)^†	–1.1 (0.9)	–4.0 (1.5)^†	<.01
VA/TLC (%)	81.8 (5.2)	78.4 (7.1)^†	82.7 (7.4)	74.0 (7.4)^†	<.01
mMRC ≥ 2 (%)	15%	32%^†	10%	47%^†	<.01
PaO2* (mmHg)	79.6 ()	74.0 (8.1)^†	72.3 (9.2)	74.6 (7.3)^†	<.01
PaCO2* (mmHg)	39.8 (3.5)	38.0 (3.7)^†	42.4 (2.7)	40.2 (4.4)	<.01
LAV% (%)	24.0 (6.8)	32.4 (6.9)^†	31.4 (6.7)^†	39.5 (8.2)^†	<.01
WA% (%)	59.7 (5.8)	60.3 (6.1)	61.3 (5.1)	59.1 (5.9)	.68
PSE	35%	58%^†	10%	53%	<.01
CLE	29%	92%^†	60%	94%^†	<.01
PA/Ao	0.77 (0.11)	0.76 (0.10)	0.77 (0.06)	0.80 (0.13)	0.51

Data are presented as means (SD). All subjects were male. Smoking PY = smoking pack-years. BMI = body mass index. FEV₁ = forced expiratory volume in 1 s. FVC = forced vital capacity. RV/TLC = ratio of residual volume to total lung capacity. D_LCO = diffusion capacity for carbon monoxide (CO). Kco = CO transfer coefficient. V_A/TLC = ratio of alveolar volume to total lung capacity. PaO₂ = partial pressure of oxygen. PaCO₂ = partial pressure of carbon dioxide. PSE = paraseptal emphysema. CLE = moderate to severe centrilobular emphysema. LAV% = low attenuation volume percentage. WA% = wall area percentage. PA/Ao = diameter ratio of pulmonary artery to aorta. * PaO₂ and PaCO₂ data were available for 184 patients. P = p value. ^†p < 0.05 compared to the reference group based on Tukey’s multiple comparison or multiple Fisher’s exact tests followed by Bonferroni correction.

Of the 195 male patients enrolled in the Kyoto University Cohort from 2006 to 2012, 52 had died as of October 2019. As shown in Figure 5A, the survival rate differed among the four groups. In Figure 5B, the percentages of respiratory disease-related deaths were 29%, 36%, 0%, and 67% in the reference, disproportionally impaired D_LCO, disproportionally impaired FEV₁, and mixed-impaired groups, respectively. In the multivariable Cox proportional hazard model shown in Figure 5C, the disproportionally impaired D_LCO and mixed-impaired groups had similar effects on all-cause mortality (HR [95% confidence interval (CI)] = 3.09 [1.52–6.29] and 3.53[1.56–8.03], respectively), whereas the effect of the disproportionally impaired FEV₁ on all-cause mortality was not significant (HR [95% CI] = 0.91 [0.19–4.19]). The prognostic effect of the disproportionally impaired D_LCO was detected even after adjusting for LAV% (HR [95% CI] = 2.55 [1.21–5.34]).

Figure 5. Long-term survival of patients with COPD in the Kyoto University Cohort.

(A) Kaplan–Meier curves of survival for the four groups: (1) FEV₁ z-score > –3 and D_LCO z-score > –3 (reference, n = 89), (2) FEV₁ z-score > –3 and D_LCO z-score ≤ –3 (disproportionally impaired D_LCO, n = 62), (3) FEV₁ z-score ≤ –3 and D_LCO z-score > –3 (disproportionally impaired FEV₁, n = 10), and (4) FEV₁ z-score ≤ –3 and D_LCO z-score ≤ –3 (mixed-impaired, n = 34). (B) Causes of death. (C) Multivariable Cox proportional hazard models. A dot with an error bar indicates the hazard ratio with 95% CI. * p<.05 compared to the reference group in the multivariable models. The model used for the upper panel included the group, age, pack-years of smoking, height, and weight as independent variables, and the model used for the lower panel included the group, age, pack-years of smoking, height, weight, and LAV% (a CT index of emphysema severity) as independent variables.

Furthermore, additional analyses were performed using the % predicted FEV₁ and D_LCO to categorize patients into the four groups (see the online Supplemental Figures S1 and S2). While the percentage of subjects with the disproportionally impaired D_LCO, defined using the z-scores of FEV₁ and D_LCO, was 29% and 31% in the KOCOSS and Kyoto University cohorts, the use of the % predicted value-based definition of this subtype (% of predicted FEV₁ > 50% and % of predicted D_LCO ≤ 50%) changed the percentages to 18% and 21% in the KOCOSS and Kyoto University cohorts, respectively. Nonetheless, the disproportionally impaired D_LCO relative to FEV₁ based on the % predicted value was significantly associated with an increase in MRC, CAT, and SGRQ scores in the KOCOSS cohort, and with increased odds ratio of the presence of PSE and CLE, lower PaO₂ and PaCO₂, and higher mortality in the Kyoto University Cohort.

Discussion

This study shows that a disproportionally impaired D_LCO relative to FEV₁ was common (approximately 30%) in patients with COPD in two cohorts from different countries. This functional subgroup presented an increased severity of symptoms, impaired quality of life and exercise capacity, greater PSE and CLE, and lower PaO₂ and PaCO₂ than the reference group. Furthermore, the longitudinal data collected over 10 years from the Kyoto University Cohort show that this group exhibited a higher risk of long-term mortality. These findings highlight the clinical relevance of identifying a disproportionally impaired D_LCO relative to FEV₁ in COPD management.

D_LCO reflects emphysema more strongly than FEV₁ and predicts future emphysema progression and mortality [12,36]. Nonetheless, FEV₁ on spirometry has been exclusively used in clinical practice and research fields until Balasubramanian et al. [17] recently showed the utility of categorizing patients with COPD based on a combination of FEV₁ and D_LCO. The present data confirm and extend those previous findings by showing that the disproportionally impaired D_LCO is associated with worse patient-reported outcomes, an abnormal gas exchange, higher rates of PSE and CLE, and increased mortality rates in patients with COPD. In particular, the finding that the hazard ratio of mortality did not differ between the disproportionally impaired D_LCO and mixed-impaired groups is important, as it improves our ability to estimate the prognosis of patients with COPD.

The rates of both PSE and CLE were higher in the disproportionally impaired D_LCO group, while the rate of CLE, but not PSE, was higher in the disproportionally impaired FEV₁ and mixed-impaired groups than in the reference group. This result is consistent with a previous finding that a reduced in FEV₁ is associated with CLE, but not PSE [37–39]. A recent micro-CT study showed relatively milder small airway disease in PSE than CLE regions in explanted lungs from patients with COPD [40]. Collectively, the disproportionally impaired D_LCO might reflect more severe emphysema, particularly PSE, with relatively less damage to the airways in patients with COPD.

The disproportionally impaired D_LCO group showed a higher mortality than the reference group, even after adjusting for LAV%. An impaired diffusion capacity is associated with emphysema, pulmonary vascular abnormalities [6,9–11], and dysfunction of pulmonary microvascular perfusion [41], even in patients with mild COPD. Therefore, the disproportionally impaired D_LCO might reflect pulmonary vascular dysfunction and might be associated with increased mortality independent of the emphysema severity.

The present data, showing associations between the disproportionally impaired D_LCO and lower PaO₂ and PaCO₂, confirm a previous finding that the diffusion capacity is correlated with PaO₂ [12]. Additionally, the data are the first to show that a lower D_LCO is associated with a lower PaCO₂ in patients with a relatively preserved FEV₁. This result is also consistent with a previous finding that differences in alveolar-arterial oxygen levels characterized by decreases in both PaCO₂ and PaO₂ precede chronic respiratory failure in patients with COPD [13]. Therefore, PaCO₂ may be decreased in the early stage of emphysema development and D_LCO impairment, and then become increased in the late stage of the disease to eventually cause chronic hypercapnic respiratory failure.

The use of two cohorts from Japan and Korea is an advantage of this study. The two cohorts consistently showed similar frequencies in the four groups, suggesting that the disproportionally impaired D_LCO relative to FEV₁ is commonly identified in patients with COPD. Interestingly, the percentage of this functional phenotype was higher than the value documented in a previous report from the COPDGene study [17]. The discrepancy might be due to the different severity between the studies as % of predicted FEV₁ in the previous study (70%) was higher than in the present two cohorts.

FEV₁/FVC decreases with age and may cause an overdiagnosis of COPD in elderly subjects [35,42,43]. Therefore, this study defined the airflow limitation based on FEV₁/FVC < LNN, but not FEV₁/FVC < 0.7 (the Global Initiative for Chronic Obstructive Lung Disease [GOLD] criteria [2]). Indeed, as shown in Supplemental Figure S3, of 798 males with FEV₁/FVC <0.7 in the KOCOSS cohort, 55 males showed FEV1/FVC ≥ LLN, and age was higher in those with FEV₁/FVC ≥ LNN than those with FEV₁/FVC < LLN.

This study has some limitations. First, although cardiac dysfunction and pulmonary hypertension may affect D_LCO, this study did not examine the possible effects of these abnormalities using echocardiography and heart catheterization. However, PA/Ao, which is a good marker for pulmonary hypertension [44], did not differ significantly between the four groups in this study. Second, this study analyzed the data from male patients. Further studies are needed to confirm whether the findings from this study are generalizable to female subjects.

Conclusion

In this study, the data obtained from the Korean and Japanese cohorts show that a disproportionally impaired D_LCO relative to FEV₁ is a common functional phenotype in patients with COPD. The identification of this phenotype may improve our understanding of the various clinical manifestations of each individual and help noninvasively estimate the long-term prognosis of patients with COPD in daily practice.

Disclosure statement

The authors report no conflicts of interest in this work.

Additional information

Funding

The Kyoto University Cohort was partially supported by the Japan Society for the Promotion of Science (JSPS) [Grants-in-Aid for scientific research 19K08624, 16K09536, 25461156, 21590964, and 16390234]. The KOCOSS cohort was funded by Research of Korea Centers for Disease Control and Prevention [2016ER670102 and 2018ER670100].