Overview of the study
6-min walk distance; BV5/TBV: the ratio of pulmonary blood volume in vessels with a cross-sectional area <5 mm2 to the total blood vessel volume; CRP: C-reactive protein; CT: computed tomography; FEV1: forced expiratory volume in 1 s; N:L ratio: neutrophil to lymphocyte ratio; PA: pulmonary artery; Pi10: standardised airway wall thickness based on an ideal airway 10 mm in internal perimeter; WBC: white blood cell count.
Abstract
Background
Frailty, a state of reduced physiologic reserve, is associated with comorbid lung disease. Its relationship with lung function change over time and with lung computed tomography (CT) measurements is less clear.
Methods
COPDGene, a multicentre cohort study, enrolled adults aged 45–80 years with a smoking history of a minimum of 10 pack-years. This study included participants with normal baseline 6-min walk distance (6MWD), spirometry at all three site visits and frailty assessment. The primary exposures were baseline quantitative CT measurements (airway wall thickening (Pi10), log (% emphysema), pulmonary artery (PA) enlargement and the ratio of pulmonary blood volume in vessels with a cross-sectional area <5 mm2 to the total blood vessel volume (BV5/TBV)) and spirometric decline over the 10-year study period. The primary outcomes were frailty and prefrailty at the 10-year follow-up (defined using a modified Fried Frailty Phenotype). Secondary analyses evaluated associations between inflammatory markers and frailty and for mediation of these associations by spirometric decline.
Results
Of 2354 participants, 1200 (51%) were prefrail and 297 (13%) were frail at the 10-year follow-up. Baseline Pi10, log (% emphysema) and PA enlargement were significantly associated with increased odds of frailty (adjusted ORs, respectively: 1.52 (95% CI 1.30–1.77; p<0.001), 1.38 (95% CI 1.14–1.67; p<0.001) and 2.49 (95% CI 1.41–4.41; p=0.002)), as was lower BV5/TBV, an indicator of pulmonary vascular pruning (OR 0.59, 95% CI 0.48–0.72; p<0.001). Forced expiratory volume in 1 s (FEV1) decline was associated with increased odds of frailty (OR 1.47 (95% CI 1.30–1.65; p<0.001) per 1%-per-year drop). White blood cell count (WBC) and neutrophil to lymphocyte (N:L) ratio were associated with frailty; this was partially mediated by spirometric decline.
Conclusion
In a population with a cigarette smoking history, lung function decline and CT measurements were associated with prospective frailty risk.
Shareable abstract
In people with a history of smoking, findings on lung CT scan (emphysema, airway wall thickening and changes in the blood vessels in the lung) and decline in lung function are associated with risk of frailty at 10 years https://bit.ly/42Q1VwQ
Introduction
Frailty is a syndrome of increased vulnerability to stressors; the Fried Frailty Phenotype is a frequently used assessment of physical frailty [1, 2]. Among individuals with a history of smoking, frailty (with and without comorbid COPD) is associated with respiratory exacerbations and mortality [3–10].
While studies have identified associations between impaired baseline spirometry and the development or progression of frailty [11–13], and between baseline frailty and increased rate of subsequent spirometric decline [14, 15], less is known about how lung function decline relates to prospective frailty risk. Decline in peak expiratory flow has been associated with incident frailty, even among individuals without baseline obstructive lung disease [15, 16]. In a population-based study of spirometric trajectory clusters, frailty was more common in the cluster with persistently low forced expiratory volume in 1 s (FEV1) (than in those with persistently normal FEV1), but not in clusters with accelerated spirometric decline [17]. Comprehensive evaluations of how spirometric changes relate to prospective frailty risk are needed.
The relationship between quantitative lung computed tomography (CT) measurements and frailty remains underexplored. We have previously observed a cross-sectional association between airway wall thickening and frailty [3], but this has not been evaluated with respect to prospective frailty, nor have associations between other quantitative CT measurements and frailty.
In population studies, systemic inflammatory markers (particularly C-reactive protein (CRP)) have been associated with frailty; a meta-analysis re-demonstrated this association cross-sectionally, but did not find an association between CRP and frailty incidence in adjusted analyses [18]. CRP has been also associated with lung function impairment and lung function decline [19–21]. Given these links between systemic inflammation and impaired lung function, and between frailty and both inflammation and lung disease, we hypothesise that pulmonary traits (spirometry decline and airway wall thickening on CT) may mediate the relationship between systemic inflammation and frailty risk in people with a history of cigarette smoking.
This study aimed to assess how spirometric decline and lung quantitative CT measurements relate to prospective frailty risk and to evaluate for the presence of mediation of the relationship between systemic inflammation and frailty by pulmonary pathology (spirometry decline and airway wall thickening).
Methods
Study population
COPDGene is a longitudinal cohort study that enrolled individuals aged 45–80 years with a smoking history of a minimum of 10 pack-years [22]. Our study included COPDGene participants with spirometry available at all three visits (baseline (Phase 1), year 5 (Phase 2) and year 10 (Phase 3)) and a frailty assessment at Phase 3. Although the Fried Frailty Phenotype was not available at Phase 1, a relevant baseline cohort was identified by excluding individuals with a low baseline 6-min walk distance (6MWD), which was a component of the frailty phenotype (supplementary methods and figure 1) [3, 23]. All clinical centres underwent institutional review board approval, and all participants provided written informed consent.
FIGURE 1.
Consolidated Standards of Reporting Trials (CONSORT) diagram. LVRS: lung volume reduction surgery; 6MWD: 6-min walk distance.
Study design
This study evaluated the associations of quantitative CT measurements (Phase 1) and lung function decline assessed via spirometry (Phase 1 to Phase 3) with the outcomes of frailty and prefrailty (Phase 3) (figure 2). Secondary analyses evaluated the relationship between systemic inflammatory markers (Phase 2) and frailty and prefrailty, and assessed for mediation of this relationship by lung function decline and airway wall thickening.
FIGURE 2.
Study design. a) Data collection flow chart. Shows the timing of spirometry measurements, inflammatory marker measurement, lung computed tomography and frailty assessment during this 10-year longitudinal study. b) Directed acyclic graphs used for mediation analyses. FEV1: forced expiratory volume in 1 s; FVC: forced vital capacity; AEC: absolute eosinophil count; WBC: white blood cell count; CRP: C-reactive protein; N:L: neutrophil:lymphocyte; Pi10: airway wall thickness metric (millimetres).
Measurements
Smoking history, comorbidities, education level and respiratory exacerbations were assessed by survey questionnaire (supplementary methods). 6-min walk tests were performed in accordance with American Thoracic Society (ATS) standards [24]. Spirometry was measured at all three phases using ndd EasyOne Spirometers (ndd Medical Technologies, Andover, MA). The diffusing capacity of the lung for carbon monoxide (DLCO) was measured at Phase 2 (ndd Medical Technologies, Andover, MA) [25].
The rate of lung function decline was defined as (Vbaseline–Vfollow-up)/(Vbaseline×follow-up time (years)) for both FEV1 and forced vital capacity (FVC). Analyses anchored on this annual percent decline (% per year). Secondary analyses using absolute lung volume decline (millilitres per year) are reported. Exploratory analyses evaluated early (over the first 5 years of the study, i.e. Phase 1–Phase 2) and late (the last five years, i.e. Phase 2–Phase 3) lung function decline.
The lung CT acquisition protocols utilised in the COPDGene study have been previously reported [22]. Thirona software (Thirona, Nijmegen, the Netherlands) was used to assess log-transformed % emphysema and the square root of the wall area of a hypothetical airway 10 mm in internal perimeter (Pi10) (supplementary methods) [26–28]. Pulmonary artery to aorta (PA:A) ratio (continuous and binarised as ≥1) and the ratio of pulmonary blood volume in blood vessels with a cross-sectional area <5 mm2 to the total blood vessel volume (BV5/TBV) were calculated as described previously [29, 30]. Determination of airway and emphysema disease axis “membership” based on principal component analysis has been described [31].
The blood inflammatory markers assessed were white blood cell count (WBC), absolute eosinophil count (AEC), the derived neutrophil to lymphocyte ratio (N:L ratio) and CRP. CRP was selected a priori as it has been associated with both frailty and lung disease; plasma protein concentrations were measured using the SomaScan® version 4.0 platform (SomaLogic, Boulder, CO) (supplementary methods) [32]. Phase 2 measurements were utilised to maximise sample size; Pearson correlation with Phase 1 measurements (where available) was assessed.
Frailty assessment was conducted at the 10-year follow up (Phase 3) visit using a modified Fried Frailty Phenotype as previously described [3]. Frailty was defined based on the five components: shrinking, fatigue, slowness, low activity level and weakness. The number of components present determined participant classification as robust (0), prefrail (1–2) or frail (≥3).
Statistical analysis
Continuous data are reported as mean (standard deviation) and categorical data as numbers (percentage) unless otherwise indicated. Crude associations with frailty category were tested using Kruskal–Wallis test, Pearson's chi-squared test or Fisher's exact test. For modelling, continuous predictors were scaled (supplementary methods). A two-sided p-value ≤0.05 was considered statistically significant unless otherwise indicated.
Univariate logistic regression of baseline characteristics (including spirometry and quantitative CT measurements) on the outcomes of frailty and prefrailty (compared with robustness) was performed (supplementary methods). Univariate odds ratios (ORs) and 95% confidence intervals (CIs) are reported.
Multivariable modelling
Multivariable logistic regression evaluated the association of lung function decline (FEV1 % per year and FVC % per year) with the outcomes of frailty and prefrailty. Sequentially complex models were evaluated using goodness-of-fit testing. Covariates for the final model were chosen based on a stepwise backward selection algorithm (supplementary methods); these were baseline age (third-order polynomial), sex, self-identified race, body mass index (BMI) (third-order polynomial), education, smoking status and smoking pack-years. Additional analyses, including evaluation of an interaction term between BMI and lung function decline, are described in the supplementary methods.
Multivariable logistic regression of the association between baseline quantitative CT measurements (Pi10, log (% emphysema), BV5/TBV, PA:A ratio (continuous and binarised as ≥1), airway axis membership and emphysema axis membership) and the outcomes of frailty and prefrailty was performed (supplementary methods). The increase in Pi10 was also evaluated as a predictor. Logistic regressions of systemic inflammatory markers (WBC, AEC, N:L ratio, CRP) on the outcomes of frailty and prefrailty were performed in a similar manner (supplementary methods).
Mediation analyses
Causal mediation analyses of the relationship between inflammatory markers and frailty were performed to test for the presence of mediation by lung function decline or Pi10 (exposure X=inflammatory marker, outcome Y=frailty, mediator M=lung function decline or Pi10; figure 2 and supplementary methods) [33]. Continuous variables were centred and scaled prior to analysis. The average direct effect and average causally mediated effect (ACME) were reported.
Sensitivity and secondary analyses
Given the concern that individuals with COPD at baseline might not be free of frailty (even in the absence of low 6MWD), modelling of the primary study outcomes was conducted in the subgroup of individuals with normal baseline spirometry. Given the potential effects of time-varying smoking status and lung function trajectories, sensitivity analyses restricted to the subgroup of individuals who were not current smokers throughout the study (“former smokers”) were similarly performed. Other sensitivity and secondary analyses (including partial proportional odds modelling of frailty category as an ordinal outcome) are described in the supplementary methods.
Missing data
In the case of missing data, a complete-case analysis was performed. Characteristics of participants with missing data or loss to follow-up are in supplementary table 1.
Analyses were conducted in R 4.3.0. Packages used include MASS and mediation.
Results
Baseline characteristics
Of 2354 participants (mean age at baseline 59.1 (sd 8.1) years, 48% male), 857 (36%) were robust, 1200 (51%) were prefrail and 297 (13%) were frail at the 10-year follow-up. Baseline (Phase 1) characteristics of the study population categorised by 10-year (Phase 3) frailty status are displayed in table 1 and supplementary table 2. The mean total follow-up time was 10.8 (sd 1.1) years; the mean time to the Phase 2 visit was 5.5 (sd 0.7) years.
TABLE 1.
Baseline characteristics of the study population
| Characteristic | Participants (N) | Robust (n=857) | Prefrail (n=1200) | Frail (n=297) | p-value# |
|---|---|---|---|---|---|
| Age, years, mean (sd) | 2354 | 58.0 (7.4) | 59.3 (8.1) | 61.1 (9.5) | <0.001¶,+,§ |
| Sex | 2354 | 0.47 | |||
| Male | 418 (48.8) | 563 (46.9) | 150 (50.5) | ||
| Female | 439 (51.2) | 637 (53.1) | 147 (49.5) | ||
| Race | 2354 | <0.001¶,+ | |||
| Non-Hispanic white | 712 (83.1) | 874 (72.8) | 212 (71.4) | ||
| African American | 145 (16.9) | 326 (27.2) | 85 (28.6) | ||
| No post-secondary education | 2354 | 194 (22.6) | 365 (30.4) | 116 (39.1) | <0.001¶,+,§ |
| Smoking status | 2354 | 0.25 | |||
| Former | 510 (59.5) | 671 (55.9) | 167 (56.2) | ||
| Current | 347 (40.5) | 529 (44.1) | 130 (43.8) | ||
| Smoking pack-years, mean (sd) | 2354 | 36.8 (19.1) | 40.5 (22.1) | 49.6 (26.2) | <0.001¶,+,§ |
| GOLD grade | 2354 | <0.001¶,+,§, ƒ | |||
| Normal spirometry | 533 (62.2) | 645 (53.8) | 118 (39.7) | ||
| 1 | 94 (11.0) | 135 (11.3) | 17 (5.7) | ||
| 2 | 130 (15.2) | 200 (16.7) | 70 (23.6) | ||
| 3 | 28 (3.3) | 65 (5.4) | 48 (16.2) | ||
| 4 | <5 (<0.6) | 8 (0.7) | 12 (4.0) | ||
| PRISm | 70 (8.2) | 147 (12.3) | 32 (10.8) | ||
| GOLD 2–4 | 160 (18.7) | 273 (22.8) | 130 (43.8) | <0.001+,§ | |
| FEV1, L, mean (sd) | 2354 | 2.73 (0.78) | 2.46 (0.76) | 2.05 (0.79) | <0.001¶,+,§ |
| FVC, L, mean (sd) | 2354 | 3.76 (0.95) | 3.44 (0.93) | 3.15 (0.87) | <0.001¶,+,§ |
| FEV1/FVC, mean (sd) | 2354 | 0.73 (0.10) | 0.71 (0.12) | 0.65 (0.17) | <0.001+,§ |
| BMI, kg·m−2, mean (sd) | 2354 | 27.7 (5.0) | 28.9 (5.6) | 31.3 (6.7) | <0.001¶,+,§ |
| BMI category, kg·m−2 | 2354 | <0.001¶,+,§ | |||
| ≤21 | 51 (6.0) | 60 (5.0) | 11 (3.7) | ||
| 21–25 | 225 (26.3) | 239 (19.9) | 50 (16.8) | ||
| 25–30 | 331 (38.6) | 452 (37.7) | 76 (25.6) | ||
| 30–35 | 182 (21.2) | 298 (24.8) | 66 (22.2) | ||
| >35 | 68 (7.9) | 151 (12.6) | 94 (31.6) | ||
| Diabetes | 2354 | 52 (6.1) | 116 (9.7) | 57 (19.2) | <0.001¶,+,§ |
| Hypertension | 2354 | 266 (31.0) | 466 (38.8) | 134 (45.1) | <0.001¶,+ |
| Hyperlipidaemia | 2354 | 341 (39.8) | 498 (41.5) | 132 (44.4) | 0.36 |
| Heart disease## | 2353 | 60 (7.0) | 121 (10.1) | 42 (14.1) | <0.001¶,+ |
| Osteoarthritis | 2354 | 123 (14.4) | 232 (19.3) | 74 (24.9) | <0.001¶,+ |
| % emphysema, median (IQR) | 2245 | 1.9 (0.6–4.9) | 1.7 (0.5–4.8) | 2.2 (0.5–8.2) | 0.046§ |
| % gas trapping, median (IQR) | 2015 | 11.5 (5.5–20.3) | 11.0 (5.1–21.4) | 15.8 (6.8–36.1) | <0.001+,§ |
| Pi10 (mm), mean (sd) | 2245 | 2.1 (0.5) | 2.2 (0.5) | 2.4 (0.6) | <0.001¶,+,§ |
| Segmental wall area, %, mean (sd) | 2245 | 47.0 (7.5) | 48.5 (7.9) | 52.6 (8.5) | <0.001¶,+,§ |
| PA:A ratio, mean (sd) | 2248 | 0.81 (0.11) | 0.82 (0.12) | 0.85 (0.13) | <0.001¶,+,§ |
| PA enlargement ¶¶ | 2248 | 42 (5.1) | 75 (6.5) | 36 (12.9) | <0.001+,§ |
| BV5/TBV, mean (sd) | 2005 | 0.62 (0.06) | 0.60 (0.07) | 0.58 (0.08) | <0.001¶,+,§ |
| Airway axis membership##, mean (sd) | 1992 | −0.58 (0.84) | −0.34 (0.87) | 0.18 (0.88) | <0.001¶,+,§ |
| Emphysema axis membership##, mean (sd) | 1992 | −0.08 (0.82) | −0.09 (0.86) | 0.01 (0.97) | 0.52 |
Data are presented as n (%) unless otherwise stated. Baseline (Phase 1) demographic, clinical and quantitative computed tomography (CT) characteristics by frailty status at the 10-year follow-up. Participants were categorised as robust, prefrail or frail according to a modified Fried Frailty Phenotype. Bold indicates that the p-value is ≤0.05 for the global test across groups (Kruskal–Wallis rank sum tests, Pearson's chi-squared tests or Fisher's exact test). Footnotes indicate that the adjusted p-value is ≤0.05 for the Dunn or Fisher post hoc comparison between groups. GOLD: Global Initiative for Chronic Obstructive Lung Disease; PRISm: preserved ratio with impaired spirometry; FEV1: forced expiratory volume in 1 s; FVC: forced vital capacity; BMI: body mass index; IQR: interquartile range; Pi10: standardised airway wall thickness based on an ideal airway 10 mm in internal perimeter; PA: pulmonary artery; A: aorta; BV5/TBV: the ratio of pulmonary blood volume in vessels with a cross-sectional area <5 mm2 to the total blood vessel volume. #: Global p-values were calculated by Kruskal–Wallis rank sum test, Pearson's chi-squared test or Fisher's exact test using the Monte Carlo method with 2000 simulations; ¶: p≤0.05 for robust versus prefrail post hoc comparison; +: p≤0.05 for robust versus frail post hoc comparison; §: p≤0.05 for prefrail versus frail post hoc comparison; ƒ: Fisher's exact test (global); ##: defined in the supplementary methods; ¶¶: defined as a PA:A ratio of ≥1. Bold indicates statistical significance (p<0.05).
Lower baseline spirometric volumes (FEV1 and FVC) and FEV1/FVC ratio were associated with increased crude risk of prefrailty and frailty at the 10-year follow-up. Individuals with frailty at follow-up were more likely to have had baseline moderate–severe COPD than individuals who were robust at follow-up (44% versus 19%; univariate odds ratio compared with normal spirometry: 3.67, 95% CI 2.71–4.99; p<0.001). Baseline preserved ratio with impaired spirometry (PRISm) was also associated with univariate frailty risk (PRISm (compared with normal spirometry) odds ratio for frailty: 2.06, 95% CI 1.29–3.26; p=0.002).
Mean DLCO was lower among individuals who were frail at Phase 3 than among those who were robust (73% versus 89%; p<0.001). The negative association between antecedent DLCO and frailty persisted among the subgroup with normal spirometry at baseline, despite low emphysema in this group (table 2 and supplementary table 3).
TABLE 2.
Longitudinal traits by frailty category
| Participants (N) | Robust (n=857) | Prefrail (n=1200) | Frail (n=297) | p-value# | Prefrailty OR (95% CI) | p-value | Frailty OR (95% CI) | p-value | |
|---|---|---|---|---|---|---|---|---|---|
| Longitudinal spirometric changes (Phase 1–Phase 3), mean (sd) | |||||||||
| FEV1 decline | |||||||||
| % per year | 2354 | 1.5 (1.2) | 1.6 (1.4) | 2.1 (1.8) | <0.001 | 1.05 (0.98–1.13) | 0.15 | 1.38 (1.25–1.53) | <0.001 |
| mL per year | 2354 | 39 (28) | 37 (31) | 40 (33) | 0.20 | 0.90 (0.78–1.04) | 0.16 | 1.03 (0.82–1.28) | 0.79 |
| FVC decline | |||||||||
| % per year | 2354 | 1.2 (1.1) | 1.2 (1.2) | 1.6 (1.6) | <0.001 | 1.01 (0.94–1.09) | 0.78 | 1.36 (1.22–1.52) | <0.001 |
| mL per year | 2354 | 44 (40) | 41 (42) | 53 (54) | 0.002 | 0.92 (0.83–1.03) | 0.13 | 1.26 (1.09–1.46) | 0.002 |
| Longitudinal clinical changes (Phase 1–Phase 3) | |||||||||
| Additional smoking pack-years, mean (sd) | 2354 | 2.2 (3.7) | 2.5 (3.6) | 2.6 (4.0) | 0.03 | 1.08 (0.99–1.18) | 0.09 | 1.09 (0.96–1.24) | 0.15 |
| Quit smoking¶, n (%) | 1006 | 143 (41.2) | 173 (32.7) | 41 (31.5) | 0.02 | 0.69 (0.52–0.92) | 0.01 | 0.66 (0.43–1.00) | 0.05 |
| Annual respiratory exacerbation rate+, mean (sd) | 2327 | 0.14 (0.29) | 0.23 (0.46) | 0.40 (0.59) | <0.001 | 1.35 (1.20–1.53) | <0.001 | 1.98 (1.68–2.34) | <0.001 |
| Diffusion capacity (Phase 2), mean (sd) | |||||||||
| DLCO % predicted (GLI) | 2179 | 89.1 (19.5) | 82.3 (20.9) | 72.7 (22.7) | <0.001 | 0.70 (0.64–0.77) | <0.001 | 0.44 (0.37–0.51) | <0.001 |
| DLCO % predicted (GLI) (Hb- and altitude-adjusted) | 2158 | 88.7 (19.2) | 82.4 (20.6) | 73.3 (22.9) | <0.001 | 0.72 (0.65–0.79) | <0.001 | 0.46 (0.39–0.54) | <0.001 |
| Quantitative CT metrics (Phase 2) | |||||||||
| % emphysema, median (IQR) | 2238 | 1.4 (0.5–4.1) | 1.34 (0.4–4.1) | 1.9 (0.4–9.0) | 0.008 | – | – | – | – |
| log (% emphysema), mean (sd) | 2238 | 0.11 (0.68) | 0.08 (0.74) | 0.24 (0.88) | 0.008 | 0.95 (0.86–1.04) | 0.27 | 1.20 (1.04–1.38) | 0.01 |
| Pi10 (mm), mean (sd) | 2238 | 2.06 (0.47) | 2.17 (0.52) | 2.42 (0.59) | <0.001 | 1.28 (1.16–1.42) | <0.001 | 1.93 (1.67–2.23) | <0.001 |
| Δ Pi10 (mm), mean (sd) | 2140 | −0.02 (0.30) | 0.00 (0.35) | 0.01 (0.39) | 0.18 | 1.05 (0.96–1.16) | 0.28 | 1.10 (0.95–1.28) | 0.21 |
| Inflammatory markers (Phase 2) | |||||||||
| White blood cell count, k·μL−1, mean (sd) | 2330 | 6.70 (1.86) | 7.08 (2.13) | 7.57 (2.14) | <0.001 | 1.22 (1.11–1.35) | <0.001 | 1.56 (1.36–1.79) | <0.001 |
| Eosinophils, k·μL−1, mean (sd) | 2317 | 0.17 (0.13) | 0.18 (0.16) | 0.20 (0.13) | 0.001 | 1.09 (1.00–1.20) | 0.07 | 1.21 (1.05–1.39) | 0.009 |
| Neutrophil:lymphocyte ratio, mean (sd) | 2318 | 2.15 (1.04) | 2.30 (1.34) | 2.53 (1.35) | <0.001 | 1.14 (1.04–1.26) | 0.006 | 1.41 (1.22–1.63) | <0.001 |
| CRP (adjusted)§, mean (sd) | 2209 | −0.13 (0.82) | 0.06 (0.78) | 0.16 (0.68) | <0.001 | 1.26 (1.15–1.38) | <0.001 | 1.51 (1.30–1.77) | <0.001 |
| CRP quartileƒ, n (%) | 2209 | <0.001 | 1.20 (1.11–1.30) | <0.001 | 1.35 (1.19–1.53) | <0.001 | |||
| 1 | 247 (30.2) | 258 (23.1) | 48 (17.6) | – | – | – | – | ||
| 2 | 221 (27.0) | 271 (24.2) | 60 (22.1) | 1.17 (0.92–1.51) | 0.21 | 1.40 (0.92–2.13) | 0.12 | ||
| 3 | 179 (21.9) | 285 (25.5) | 88 (32.4) | 1.52 (1.18–1.97) | 0.001 | 2.53 (1.70–3.80) | <0.001 | ||
| 4 | 172 (21.0) | 304 (27.2) | 76 (27.9) | 1.69 (1.31–2.19) | <0.001 | 2.27 (1.51–3.44) | <0.001 | ||
Longitudinal characteristics (lung function decline, changes in smoking status and respiratory exacerbation rate over the 10-year study period (Phase 1–Phase 3)) and intermediate characteristics (at the 5-year follow-up (Phase 2 visit)) by frailty category at the 10-year follow-up (Phase 3 visit). Univariate odds ratios (ORs) with 95% confidence intervals (CIs) for 10-year frailty and prefrailty (compared with robustness) are reported. Lung function decline was scaled as follows: Δ1 unit of mL·year−1 decline=50 mL·year−1 decline, Δ1 unit of % per year decline=1% per year decline. Other continuous predictors were scaled such that Δ1 unit=Δ1 sd. FEV1: forced expiratory volume in 1 s; FVC: forced vital capacity; DLCO: diffusing capacity of the lung for carbon monoxide; GLI: Global Lung Function Initiative; Hb: haemoglobin; CT: computed tomography; IQR, interquartile range; Pi10: standardised airway wall thickness based on an ideal airway 10 mm in internal perimeter; CRP: C-reactive protein. #: p-values were calculated by Kruskal–Wallis rank sum test (continuous variables) or Pearson's chi-squared test (categorical variables). ¶: Quit smoking during the study period was assessed in the subgroup of individuals who reported current smoking at the baseline visit (details in the supplementary methods). +: Annual respiratory exacerbation rate defined in the supplementary methods. §: CRP concentrations are derived from log-normalised relative fluorescence units on the SomaScan proteomics platform after adjustment for clinical centre, white blood cell count and platelet count. ƒ: ORs for quartiles for inflammatory markers are reported for both 1) treating the quartiles as ordinal (first row) and 2) treating the lowest quartile as the reference. ΔPi10 was defined as the increase in Pi10 (mm) from Phase 1 to Phase 2. Bold indicates statistical significance (p<0.05).
In addition to baseline lung disease, leading univariate predictors of frailty were obesity, diabetes, PA enlargement and airway axis membership (supplementary table 4). supplementary figure 2 demonstrates the nonlinear relationship between BMI and frailty risk.
Among participants who smoked at the start of the study, quitting smoking during the study period was associated with reduced univariate odds of prefrailty (OR 0.69, 95% CI 0.52–0.92; p=0.01), with a similar trend for frailty (OR 0.66, 95% CI 0.43–1.00; p=0.054) (table 2).
Lung function decline and frailty risk
Figure 3 demonstrates transitions in Global Initiative for Chronic Obstructive Lung Disease (GOLD) grade over the study period. Excluding individuals with PRISm, 17% of individuals whose GOLD grade increased during the study had frailty at the 10-year follow-up, compared with 11% of individuals whose GOLD grade improved or remained the same (p<0.001).
FIGURE 3.
Sankey diagram of Global Initiative for Chronic Obstructive Lung Disease (GOLD) grade at baseline (Phase 1), year 5 (Phase 2) and year 10 (Phase 3) among individuals who were robust, prefrail and frail at the 10-year (Phase 3) follow-up. PRISm: preserved ratio with impaired spirometry.
Summary statistics of lung function decline are reported in supplementary table 5 and supplementary figure 1. The mean decline in FEV1 was 2.1% per year (sd 1.8% per year) among those who were frail at the 10-year follow-up, and 1.5% per year (sd 1.2% per year) among individuals who were robust at the 10-year follow-up (p<0.001) (table 2).
Model-building of the relationship between antecedent lung function decline and frailty is shown in supplementary table 6. In adjusted models, for every additional 1% drop per year in FEV1 over the study period, the odds of frailty increased (OR 1.47, 95% CI 1.30–1.65; p<0.001), but the odds of prefrailty did not significantly change (figure 4 and supplementary table 7). Similar results were observed for FVC decline. In exploratory analyses, early lung function decline (during the first half of the study period) was consistently associated with increased odds of frailty and prefrailty.
FIGURE 4.
Multivariable modelling of lung function decline on odds of frailty and prefrailty. Odds ratios (ORs) and 95% confidence intervals (95% CIs) of antecedent lung function decline on frailty and prefrailty are shown. Frailty was assessed at the 10-year follow-up (Phase 3). Lung function decline was assessed in decline in forced expiratory volume in 1 s (FEV1) and forced vital capacity (FVC) in annualised percent decline (% per year) or annualised decline in millilitres (mL·year−1). Lung function decline over the entire 10-year study period (Phases 1–3 (P1–3)) and over the first 5 years of the study period (Phases 1–2 (P1–2)) are shown. For interpretability, lung function decline was scaled as follows: Δ1 unit of mL·year−1 decline=50 mL·year−1 decline; Δ1 unit of % per year decline=1% per year decline. Covariates were baseline age (third-order polynomial), race, sex, body mass index (third-order polynomial), education, smoking status, smoking pack-years and centred length of follow-up time.
We identified an interaction effect between baseline BMI and lung function decline on frailty risk (supplementary table 8). While obesity and lung function decline were both associated with increased frailty risk, baseline obesity attenuated the magnitude of the effect estimate for lung function decline on frailty.
A complementary analysis of univariate odds ratios for lung function decline on frailty stratified by BMI category (supplementary table 9) supported this finding.
Quantitative CT measurements and frailty risk
Baseline airway wall thickening and pruning of the pulmonary vessels (lower BV5/TBV) had crude associations with frailty category (table 1), with similar trends observed across spirometric subgroups (supplementary table 2). Baseline emphysema was associated with frailty category in the overall study population and among those with moderate to severe COPD (supplementary table 2); however, among the group with normal spirometry, lower % emphysema was associated with crude frailty risk.
Associations between baseline quantitative CT measurements and frailty risk persisted in multivariable modelling of the full study population (odds ratios for scaled continuous predictors: Pi10, 1.52 (95% CI 1.30–1.77; p<0.001); log (% emphysema), 1.38 (95% CI 1.14–1.67; p<0.001); BV5/TBV, 0.59 (95% CI 0.48–0.72; p<0.001); and PA:A ratio, 1.30 (95% CI 1.09–1.53; p=0.003) (table 3). PA enlargement (PA:A ratio ≥1) was also associated with frailty risk in adjusted analyses (OR 2.49, 95% CI 1.41–4.41; p=0.002).
TABLE 3.
Adjusted odds ratios of quantitative CT markers on frailty and prefrailty
| Prefrailty | Frailty | |||
|---|---|---|---|---|
| OR (95%) CI | p-value | OR (95% CI) | p-value | |
| Baseline (Phase 1) | ||||
| Pi10 (mm) | 1.16 (1.05–1.29) | 0.004 | 1.52 (1.30–1.77) | <0.001 |
| Airway axis membership | 1.25 (1.12–1.39) | <0.001 | 2.10 (1.74–2.54) | <0.001 |
| log (% emphysema) | 1.03 (0.92–1.15) | 0.62 | 1.38 (1.14–1.67) | <0.001 |
| Emphysema axis membership | 1.00 (0.90–1.12) | 0.95 | 1.23 (1.02–1.48) | 0.03 |
| BV5/TBV | 0.83 (0.73–0.95) | 0.006 | 0.59 (0.48–0.72) | <0.001 |
| PA:A ratio (continuous) | 1.08 (0.98–1.20) | 0.14 | 1.30 (1.09–1.53) | 0.003 |
| PA enlargement | 1.16 (0.77–1.75) | 0.48 | 2.49 (1.41–4.41) | 0.002 |
| 5-year follow-up (Phase 2) | ||||
| Pi10 (mm) | 1.17 (1.05–1.30) | 0.003 | 1.56 (1.32–1.83) | <0.001 |
| log (% emphysema) | 1.02 (0.92–1.14) | 0.68 | 1.43 (1.19–1.73) | <0.001 |
| ΔPi10 | 1.04 (0.94–1.15) | 0.39 | 1.04 (0.88–1.23) | 0.66 |
Odds ratios (ORs) of (scaled) quantitative computed tomography (CT) metrics on the odds of prefrailty and frailty at the 10-year follow-up. Models were adjusted for age (third-order polynomials), sex, race, education, body mass index (third-order polynomials), smoking status, smoking pack-years, CT scanner type and centred length of follow-up time. Predictors were scaled such that Δ1 unit=Δ1 sd. Airway and emphysema axis membership definitions are discussed in the supplementary methods. Pulmonary artery (PA) enlargement was defined as a PA to aorta (PA:A) ratio of ≥1. ΔPi10 was defined as the increase in Pi10 (mm) from Phase 1 to Phase 2. Pi10: standardised airway wall thickness based on an ideal airway 10 mm in internal perimeter; log (% emphysema): log10 of the per cent low attenuation area (< −950 Hounsfield units) on inspiratory CT; BV5/TBV: ratio of pulmonary blood volume in blood vessels with cross-sectional area <5 mm2 to the total blood vessel volume. Bold indicates statistical significance (p<0.05).
Inflammatory markers and frailty risk
Phase 2 systemic inflammatory markers, including WBC, AEC and CRP, had crude associations with frailty category (table 2). In multivariable models, WBC and N:L ratio were associated with increased odds of frailty and prefrailty, with a trend for CRP (p=0.052) (supplementary table 10). In the subgroup with normal baseline spirometry, WBC remained associated with frailty and prefrailty risk (supplementary table 11). The correlation between Phase 1 and Phase 2 inflammatory markers (in the subset of participants for whom these data are available) is reported in supplementary table 12.
Mediation analyses
FEV1 decline (% per year) mediated the associations of WBC, N:L ratio and CRP with frailty (supplementary figure 4 and supplementary table 13). For example, FEV1 % per year partially mediated the relationship between WBC and log-odds of frailty (ACME 0.006, 95% CI 0.0004–0.01; p=0.03). Airway wall thickness (Pi10) was also found to partially mediate the relationship between systemic inflammatory markers and frailty (supplementary table 13).
Sensitivity and secondary analyses
Sensitivity analysis of individuals who did not smoke during the study period (self-reported former smoking at all three visits, n=1269; 159 (13%) of whom were frail) yielded similar results for lung function decline, quantitative CT findings and inflammatory markers on frailty risk (supplementary table 11). Sensitivity analyses of the subgroup with normal baseline spirometry (n=1296; 118 (9%) of whom were frail) recapitulated associations between lung function decline and frailty; of the quantitative CT predictors, airway axis membership and pulmonary vascular pruning remained associated with frailty in this subgroup (supplementary table 11). Secondary analyses of quantitative CT measurements on frailty additionally adjusted for clinical centre had similar results (supplementary table 14), as did sensitivity analyses using partial proportional odds modelling to evaluate odds of frailty category as an ordinal outcome (supplementary table 15).
Discussion
This longitudinal cohort study demonstrated prospective associations of lung function decline, quantitative CT measurements and systemic inflammatory markers with frailty risk. Higher rates of antecedent FEV1 and FVC decline were associated with increased risk of frailty at the 10-year follow-up visit; this finding persisted in sensitivity analyses of individuals with normal baseline spirometry. To our knowledge, this is the first study to identify a relationship between FEV1 and FVC decline and prospective frailty risk (although associations between peak flow decline and frailty have been previously identified) [15, 16]. Frailty screening should be considered in all patients with lung function decline, even in those without severe COPD.
This study is, to our knowledge, the first comprehensive evaluation of quantitative lung CT measurements with respect to frailty risk. Baseline airway wall thickening was associated with increased frailty and prefrailty risk. PA enlargement and reduced BV5/TBV (which represents pruning of the small pulmonary vessels) were associated with higher frailty risk. These pulmonary vascular findings may explain the associations between lower DLCO and frailty in the subgroup with normal spirometry (who had minimal emphysema). Additional investigation into whether occult pulmonary vascular remodelling relates to frailty risk (and into the dynamics between comorbid heart and lung disease in people with a history of smoking) is merited.
Using causal mediation analyses, this study demonstrated that lung function decline and airway wall thickening partially mediated associations between systemic inflammatory markers and frailty. While relationships between systemic inflammatory markers and frailty have been described previously [18, 34], mediation of this relationship by pulmonary pathology has not been previously explored. These findings suggest a role for lung pathology (including lung function decline and airway inflammation) on the pathway between systemic inflammation and frailty in individuals at risk of lung disease. Given the expanding role of targeted anti-inflammatory agents in the treatment of lung disease, how these treatments may influence frailty risk is of future interest.
Changes in cigarette smoking during the study period (quitting smoking or increasing smoking pack-years) were associated with changes in prefrailty or frailty risk. Although prior studies have shown associations between current smoking and frailty incidence and between time since smoking cessation and frailty progression, this study underscores the importance of cigarette smoking cessation in mitigating frailty and prefrailty risk [35–37]. However, the persistent associations of lung function decline and CT measurements with frailty risk among those who had already quit smoking (prior to the study) highlight the importance of frailty screening and prevention in everyone with any history of smoking.
The associations between baseline obesity and diabetes with frailty underscore the relevance of metabolic comorbidities in this population at risk of lung disease progression. While associations between obesity and frailty risk have been identified previously, this study additionally identified a negative interaction term between obesity and lung function decline on the outcome of frailty [38, 39]. Since a meta-regression had found that BMI partly moderates the relationship between inflammation (CRP) and frailty [18], we hypothesise that this finding may be related to the association with systemic inflammation that both obesity and lung function decline share.
These findings indicate a need for awareness among primary care providers and pulmonologists of the risk of frailty in individuals with a history of cigarette smoking, and they suggest a role for active monitoring and screening in individuals with lung function decline or abnormal lung CT findings, even in the absence of obstructive lung disease. Due to the potential for increased longitudinal frailty risk, providers may consider mitigation of other risk factors (such as current cigarette smoking and management of comorbidities), active monitoring of lung function change over time, evaluation for systemic inflammation and consideration of anti-inflammatory therapies where indicated, or referral for pulmonary rehabilitation. Since patients may have incidental chest CT findings (after CT was performed for other indications) or have serial lung function measurements (with their primary care provider or pulmonologist), these data suggest the availability of multiple high-risk indicators for frailty across different care settings.
The strengths of this study include the extensive phenotyping (including quantitative lung CT measurements) and longitudinal follow-up with repeated spirometry measurements. This study has several limitations. Firstly, given the absence of a frailty assessment at baseline, we are unable to formally assess frailty and prefrailty incidence during this 10-year study. For this, we employed careful baseline population selection and sensitivity analyses to support our findings. Secondly, the generalisability of this study's findings to populations without a history of cigarette smoking is unclear; however, the lung function decline rates in this study were within the range described in population studies [40]. For mediation analyses, the exposures (inflammatory markers) were measured at the midpoint (Phase 2) of the study, whereas the lung function decline mediator was obtained across the entire study period. To address this, we assessed correlations between inflammatory markers at baseline and Phase 2 (where available), and we adjusted for cigarette smoking exposure across the study period. While we hypothesise that the systemic inflammation precedes lung function decline, the direction of mediation effects is not fully determined and is an area for future research. In addition, regression to the mean is a consideration with the analysis of change variables (such as lung function change over time); future analyses incorporating additional time points (e.g., COPDGene Phase 4) and expanding this analysis to other cohorts are indicated. Lastly, survival bias is a consideration in this type of longitudinal cohort study; the characteristics of participants who were lost to follow-up are reported above.
Conclusion
This study demonstrated novel associations of lung function decline and quantitative CT measurements with prospective frailty risk. Among those with normal baseline spirometry, associations persisted for lung function decline and some quantitative CT measurements, highlighting the importance of early identification of at-risk individuals. Both lung function decline and airway wall thickening on CT partially mediated the relationships between systemic inflammatory markers and frailty, suggesting a potential role of pulmonary pathology in the development of frailty. Clinical frailty assessment should be considered in people at risk of lung disease.
Acknowledgements
The authors would like to acknowledge the COPDGene study participants for their time and contributions.
Footnotes
Provenance: Submitted article, peer reviewed.
Ethics statement: This study was approved by institutional review boards of all clinical sites due to human subject participation with more than minimal risk (including spirometry, blood draws and computed tomography).
Conflicts of interest: K.F. Hoth has a leadership role in the COPD Foundation, an unpaid position. J. Bon received payment or honoraria for lectures, presentations, speakers’ bureaus, manuscript writing or educational events from GSK, Sanofi and Chiesi, and participates in a data safety monitoring or advisory board for ProPharma, Sanofi/Regeneron, GSK, AstraZeneca, Genentech and Verona Pharma. B. Make reports disclosures as follows: AstraZeneca – received personal fees, consulting fees, payment or honoraria for presentations, serves on a medical advisory board, and research grant funds provided to and controlled by National Jewish Health; Baystate Medical Center and Apreo – participation in a data safety and monitoring board; Boehringer Ingelheim – received personal fees and participation in a medical advisory board; GSK – advisory board member and received personal fees; Optimum Patient Care Global Limited – received personal and consulting fees, and payment or honoraria for presentations, and advisory board member; Regeneron – honoraria for disease state presentation; Spiration – received personal fees and participated in a data and safety monitoring board; Third Pole – received personal and consulting fees; Verona – received consulting fees; Wolters Kluwer Health – received royalties; and the US Department of Defense and the National Institutes of Health – research grant funds provided to and controlled by National Jewish Health. A.M. Yohannes received consulting fees from Chiesi for participation in a Delphi study and writing up a manuscript. N.A. Hanania has received honoraria for serving as a consultant/advisor for GSK, Sanofi, Regeneron, Roche (Genentech), Chiesi, AstraZeneca and Connect Biopharma; his institution received research grant support on his behalf from Sanofi, Regeneron, GSK, AstraZeneca, Amgen, Chiesi and Roche (Genentech). G. Washko has received institutional grant support from Boehringer Ingelheim and consulting fees from 35Pharma, Apogee, AstraZeneca, GSK, Montai Therapeutics, Pulmonx, Pieris Therapeutics, Intellia Therapeutics, Regeneron and Sanofi; is a co-founder and equity share holder in Quantitative Imaging Solutions, a company that provides consulting services for image and data analytics; and reports that their spouse works for Biogen. E.K. Silverman has received institutional grant support from Bayer and Northpond Laboratories. J. Crapo is the chair of the COPD Foundation board of directors, an unpaid position. In the past 3 years, D.L. DeMeo received institutional grant support from Bayer and consulting fees from AstraZeneca, and declares being an unpaid member of the Medical and Scientific Advisory Committee of the COPD Foundation and the grants review committee of the Alpha-1 Foundation. The other authors have nothing to disclose.
Support statement: E.K. Phillips is supported by 5T32HL007427-42. D.L. DeMeo is supported by US National Institutes of Health (NIH) grants K24 HL171900, R01 HG011393 and R01 HL178032. This work was supported by US National Heart, Lung, and Blood Institute grants U01 HL089897 and U01 HL089856, and by NIH contract 75N92023D00011. The COPDGene study (NCT00608764) has also been supported by the COPD Foundation through contributions made to an industry advisory committee that has included AstraZeneca, Bayer Pharmaceuticals, Boehringer Ingelheim, Genentech, GSK, Novartis, Pfizer and Sunovion. Funding information for this article has been deposited with the Open Funder Registry.
Supplementary material
Please note: supplementary material is not edited by the Editorial Office, and is uploaded as it has been supplied by the author.
Supplementary methods
00470-2026.SUPPLEMENT
Supplementary tables and figures
00470-2026.SUPPLEMENT2
References
- 1.Fried LP, Tangen CM, Walston J, et al. Frailty in older adults: evidence for a phenotype. J Gerontol A Biol Sci Med Sci 2001; 56: M146–M157. doi: 10.1093/gerona/56.3.M146 [DOI] [PubMed] [Google Scholar]
- 2.Clegg A, Young J, Iliffe S, et al. Frailty in elderly people. Lancet 2013; 381: 752–762. doi: 10.1016/S0140-6736(12)62167-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Phillips EK, Huang Y, Regan E, et al. Frailty associates with respiratory exacerbations and mortality in the COPDGene cohort. Aging (Albany NY) 2025; 17: 1590–1623. doi: 10.18632/aging.206275 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Marengoni A, Vetrano DL, Manes-Gravina E, et al. The relationship between COPD and frailty. Chest 2018; 154: 21–40. doi: 10.1016/j.chest.2018.02.014 [DOI] [PubMed] [Google Scholar]
- 5.Lahousse L, Ziere G, Verlinden VJA, et al. Risk of frailty in elderly with COPD: a population-based study. J Gerontol A Biol Sci Med Sci 2016; 71: 689–695. doi: 10.1093/gerona/glv154 [DOI] [PubMed] [Google Scholar]
- 6.Hanlon P, Lewsey J, Quint JK, et al. Frailty in COPD: an analysis of prevalence and clinical impact using UK Biobank. BMJ Open Respir Res 2022; 9: e001314. doi: 10.1136/bmjresp-2022-001314 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Gao X, Zhang Y, Saum K-U, et al. Tobacco smoking and smoking-related DNA methylation are associated with the development of frailty among older adults. Epigenetics 2017; 12: 149–156. doi: 10.1080/15592294.2016.1271855 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Hanlon P, Nicholl BI, Jani BD, et al. Frailty and pre-frailty in middle-aged and older adults and its association with multimorbidity and mortality: a prospective analysis of 493 737 UK Biobank participants. Lancet Public Health 2018; 3: e323–e332. doi: 10.1016/S2468-2667(18)30091-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Scarlata S, Finamore P, Laudisio A, et al. Association between frailty index, lung function, and major clinical determinants in chronic obstructive pulmonary disease. Aging Clin Exp Res 2021; 33: 2165–2173. doi: 10.1007/s40520-021-01878-z [DOI] [PubMed] [Google Scholar]
- 10.Osadnik CR, Brighton LJ, Burtin C, et al. European Respiratory Society statement on frailty in adults with chronic lung disease. Eur Respir J 2023; 62: 2300442. doi: 10.1183/13993003.00442-2023 [DOI] [PubMed] [Google Scholar]
- 11.He D, Yan M, Zhou Y, et al. Preserved ratio impaired spirometry and COPD accelerate frailty progression. Chest 2024; 165: 573–582. doi: 10.1016/j.chest.2023.07.020 [DOI] [PubMed] [Google Scholar]
- 12.Wijnant SRA, Benz E, Luik AI, et al. Frailty transitions in older persons with lung function impairment: a population-based study. J Gerontol A Biol Sci Med Sci 2023; 78: 349–356. doi: 10.1093/gerona/glac202 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Yang X, Cheng C, Ma W, et al. Longitudinal association of lung function with frailty among older adults: the English Longitudinal Study of Ageing. Eur Geriatr Med 2023; 14: 173–180. doi: 10.1007/s41999-022-00732-3 [DOI] [PubMed] [Google Scholar]
- 14.Nishimura K, Kusunose M, Shibayama A, et al. Relationship between COPD progression and frailty progression: a five-year observation in real clinical practice. Int J Chron Obstruct Pulmon Dis 2025; 20: 1955–1964. doi: 10.2147/COPD.S512326 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Trevisan C, Rizzuto D, Maggi S, et al. Cross-sectional and longitudinal associations between peak expiratory flow and frailty in older adults. J Clin Med 2019; 8: 1901. doi: 10.3390/jcm8111901 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Wang R, Shi W, Zhou W, et al. Associations between peak expiratory flow and frailty in olderly individuals: findings from the China health and retirement longitudinal study. Front Public Health 2024; 12: 1392581. doi: 10.3389/fpubh.2024.1392581 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Bertels X, Ross JC, Faner R, et al. Clinical relevance of lung function trajectory clusters in middle-aged and older adults. ERJ Open Res 2024; 10: 00793-2023. doi: 10.1183/23120541.00793-2023 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Soysal P, Stubbs B, Lucato P, et al. Inflammation and frailty in the elderly: a systematic review and meta-analysis. Ageing Res Rev 2016; 31: 1–8. doi: 10.1016/j.arr.2016.08.006 [DOI] [PubMed] [Google Scholar]
- 19.Serban KA, Pratte KA, Bowler RP. Protein biomarkers for COPD outcomes. Chest 2021; 159: 2244–2253. doi: 10.1016/j.chest.2021.01.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Baines KJ, Backer V, Gibson PG, et al. Impaired lung function is associated with systemic inflammation and macrophage activation. Eur Respir J 2015; 45: 557–559. doi: 10.1183/09031936.00187514 [DOI] [PubMed] [Google Scholar]
- 21.Jiang R, Burke GL, Enright PL, et al. Inflammatory markers and longitudinal lung function decline in the elderly. Am J Epidemiol 2008; 168: 602–610. doi: 10.1093/aje/kwn174 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Regan EA, Hokanson JE, Murphy JR, et al. Genetic epidemiology of COPD (COPDGene) study design. Int J Chron Obstruct Pulmon Dis 2011; 7: 32–43. doi: 10.3109/15412550903499522 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Agustí A, Celli BR, Criner GJ, et al. Global initiative for chronic obstructive lung disease 2023 report: GOLD executive summary. Eur Respir J 2023; 61: 2300239. doi: 10.1183/13993003.00239-2023 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.ATS statement: guidelines for the six-minute walk test. Am J Respir Crit Care Med 2002; 166: 111–117. doi: 10.1164/ajrccm.166.1.at1102 [DOI] [PubMed] [Google Scholar]
- 25.Miller MR, Hankinson J, Brusasco V, et al. Standardisation of spirometry. Eur Respir J 2005; 26: 319–338. doi: 10.1183/09031936.05.00034805 [DOI] [PubMed] [Google Scholar]
- 26.Bhatt SP, Bodduluri S, Kizhakke Puliyakote AS, et al. Structural airway imaging metrics are differentially associated with persistent chronic bronchitis. Thorax 2021; 76: 343–349. doi: 10.1136/thoraxjnl-2020-215853 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Charbonnier J-P, Pompe E, Moore C, et al. Airway wall thickening on CT: relation to smoking status and severity of COPD. Respir Med 2019; 146: 36–41. doi: 10.1016/j.rmed.2018.11.014 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Regan EA, Lynch DA, Curran-Everett D, et al. Clinical and radiologic disease in smokers with normal spirometry. JAMA Intern Med 2015; 175: 1539–1549. doi: 10.1001/jamainternmed.2015.2735 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Wells JM, Washko GR, Han MK, et al. Pulmonary arterial enlargement and acute exacerbations of COPD. N Engl J Med 2012; 367: 913–921. doi: 10.1056/NEJMoa1203830 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Estépar RSJ, Kinney GL, Black-Shinn JL, et al. Computed tomographic measures of pulmonary vascular morphology in smokers and their clinical implications. Am J Respir Crit Care Med 2013; 188: 231–239. doi: 10.1164/rccm.201301-0162OC [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Kinney GL, Santorico SA, Young KA, et al. Identification of chronic obstructive pulmonary disease axes that predict all-cause mortality. Am J Epidemiol 2018; 187: 2109–2116. doi: 10.1093/aje/kwy087 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Gold L, Ayers D, Bertino J, et al. Aptamer-based multiplexed proteomic technology for biomarker discovery. PLoS ONE 2010; 5: e15004. doi: 10.1371/journal.pone.0015004 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Imai K, Keele L, Tingley D. A general approach to causal mediation analysis. Psychol Methods 2010; 15: 309–334. doi: 10.1037/a0020761 [DOI] [PubMed] [Google Scholar]
- 34.Velissaris D, Pantzaris N, Koniari I, et al. C-reactive protein and frailty in the elderly: a literature review. J Clin Med Res 2017; 9: 461–465. doi: 10.14740/jocmr2959w [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Kojima G, Iliffe S, Jivraj S, et al. Does current smoking predict future frailty? The English Longitudinal Study of Ageing. Age Ageing 2018; 47: 126–131. doi: 10.1093/ageing/afx136 [DOI] [PubMed] [Google Scholar]
- 36.Kojima G, Taniguchi Y, Aoyama R, et al. Association between time since smoking cessation and frailty trajectory among community-dwelling older people: English Longitudinal Study of Ageing. J Am Med Dir 2025; 26: 105328. doi: 10.1016/j.jamda.2024.105328 [DOI] [PubMed] [Google Scholar]
- 37.Kojima G, Iliffe S, Walters K. Smoking as a predictor of frailty: a systematic review. BMC Geriatr 2015; 15: 131. doi: 10.1186/s12877-015-0134-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Haapanen MJ, Mikkola TM, Kortelainen L, et al. Body composition in late midlife as a predictor of accelerated age-associated deficit-accumulation from late midlife into old age: a longitudinal birth cohort study. J Gerontol A Biol Sci Med Sci 2023; 78: 980–987. doi: 10.1093/gerona/glac233 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Ho H-E, Yeh C-J, Chu W-M, et al. Midlife body mass index trajectory and risk of frailty 8 years later in Taiwan. J Nutr Health Aging 2019; 23: 849–855. doi: 10.1007/s12603-019-1226-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Thomas ET, Guppy M, Straus SE, et al. Rate of normal lung function decline in ageing adults: a systematic review of prospective cohort studies. BMJ Open 2019; 9: e028150. doi: 10.1136/bmjopen-2018-028150 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Please note: supplementary material is not edited by the Editorial Office, and is uploaded as it has been supplied by the author.
Supplementary methods
00470-2026.SUPPLEMENT
Supplementary tables and figures
00470-2026.SUPPLEMENT2