Abstract
Rationale: The association between aging and idiopathic pulmonary fibrosis has been established. The associations between aging-related biomarkers and interstitial lung abnormalities (ILA) have not been comprehensively evaluated.
Objectives: To evaluate the associations among aging biomarkers, ILA, and all-cause mortality.
Methods: In the FHS (Framingham Heart Study), we evaluated associations among plasma biomarkers (IL-6, CRP [C-reactive protein], TNFR [tumor necrosis factor α receptor II], GDF15 [growth differentiation factor 15], cystatin-C, HGBA1C [Hb A1C], insulin, IGF1 [insulin-like growth factor 1], and IGFBP1 [IGF binding protein 1] and IGFBP3]), ILA, and mortality. Causal inference analysis was used to determine whether biomarkers mediated age. GDF15 results were replicated in the COPDGene (Genetic Epidemiology of Chronic Obstructive Pulmonary Disease) Study.
Measurements and Main Results: In the FHS, there were higher odds of ILA per increase in natural log–transformed GDF15 (odds ratio [95% confidence interval], 3.4 [1.8–6.4]; P = 0.0002), TNFR (3.1 [1.6–5.8]; P = 0.004), IL-6 (1.8 [1.4–2.4]; P < 0.0001), and CRP (1.7 [1.3–2.0]; P < 0.0001). In the FHS, after adjustment for multiple comparisons, no biomarker was associated with increased mortality, but the associations of GDF15 (hazard ratio, 2.0 [1.1–3.5]; P = 0.02), TNFR (1.8 [1.0–3.3]; P = 0.05), and IGFBP1 (1.3 [1.1–1.7]; P = 0.01) approached significance. In the COPDGene Study, higher natural log–transformed GDF15 was associated with ILA (odds ratio, 8.1 [3.1–21.4]; P < 0.0001) and mortality (hazard ratio, 1.6 [1.1–2.2]; P = 0.01). Causal inference analysis showed that the association of age with ILA was mediated by IL-6 (P < 0.0001) and TNFR (P = 0.002) and was likely mediated by GDF15 (P = 0.008) in the FHS and was mediated by GDF15 (P = 0.001) in the COPDGene Study.
Conclusions: Some aging-related biomarkers are associated with ILA. GDF15, in particular, may explain some of the associations among age, ILA, and mortality.
Keywords: aging, GDF15, interstitial lung abnormalities, idiopathic pulmonary fibrosis, mortality
At a Glance Commentary
Scientific Knowledge on the Subject
Gerontologic research is working to define sets of biomarkers that may help capture the biologic processes associated with normal and accelerated aging. One chronic disease for which the central role of aging, and accelerated aging, is increasingly recognized is idiopathic pulmonary fibrosis. Although the association between aging and idiopathic pulmonary fibrosis is well established, the associations between aging-related biomarkers and interstitial lung abnormalities (ILA) have not been comprehensively evaluated.
What This Study Adds to the Field
This study demonstrates that some, but not all, biomarkers associated with underlying aging processes are also associated with ILA. In particular, GDF15 (growth differentiation factor 15) may explain some of the associations among age, ILA, and mortality. These results suggest that some aging-related pathways captured by selected biomarker measures may be important in increasing the risk of ILA and, reciprocally, that ILA should be considered in gerontologic studies attempting to identify biomarkers of accelerated aging.
A central tenet of gerontologic research is that there are pathobiologic differences between normal and accelerated aging, which may help to explain why some people develop a greater burden of chronic illnesses and increased rates of mortality than would be expected on the basis of chronologic age alone (1). This research is working to define sets of biomarkers that may help capture the biologic processes associated with normal and accelerated aging (1). One chronic disease for which the central role of aging, and accelerated aging, is increasingly recognized (2–4) is idiopathic pulmonary fibrosis (IPF). Although the relative contributions of varying aging-related pathobiologic processes to pulmonary fibrosis development are incompletely understood, biomarkers of aging have been associated with IPF disease severity (5), and drugs targeting pathways such as telomere length attrition (6) and senescent cell removal (7) are currently being evaluated in patients with IPF.
Studying interstitial lung abnormalities (ILA), defined as abnormalities on chest computed tomography (CT) scans suggestive of interstitial lung disease in undiagnosed research participants, provides an opportunity to investigate some of the early developmental processes that may lead to pulmonary fibrosis. Although ILA, as currently defined, almost certainly represent a broader array of conditions than IPF alone (8), research participants with ILA have a similar syndrome that includes an increased prevalence of genetic predictors (9–11) and histopathologic findings (12), noted in patients with IPF, as well as restrictive physiologic and exercise impairments (10, 13–15), radiologic progression (16, 17), accelerated lung-function decline (16), and increased risk of death (15–19). Importantly, advanced age is among the findings most consistently associated with ILA (9–19).
On the basis of these findings, we hypothesized that some biomarkers assessed as representative endotypes of aging-related pathobiologic processes (1) are associated with ILA and explain some of the correlations among age, ILA, and all-cause mortality.
Methods
Study Design and Mortality Assessment
Protocols for participant enrollment and phenotyping in the FHS (Framingham Heart Study) and the COPDGene (Genetic Epidemiology of Chronic Obstructive Pulmonary Disease [COPD]) Study have been described previously (10, 13). The FHS is a longitudinal study originally designed to identify risk factors for cardiovascular disease in the community that now includes distinct cohorts and a range of phenotypic data (20). For this analysis, FHS refers to a subset (n = 1,463 [53%] to 1,541 [56%] participants) of the 2,764 adult men and women from the Generation 3 and Offspring cohorts who in addition participated in the FHS-MDCT2 (Multi-Detector Computed Tomography 2) study between 2008 and 2011 and in whom measures of selected biomarkers were obtained (FHS Generation 3 examination 2 and Offspring examination 8). In the FHS, survival status was ascertained using multiple strategies, including routine participant contact, local hospital surveillance, obituary reviews, and queries to the National Death Index (see Methods in the online supplement for additional details). The study was approved by the Boston University Medical Center and Massachusetts General Hospital institutional review boards, and participants signed written consent. The COPDGene Study is a multicenter longitudinal study that was designed to identify the epidemiologic and genetic risk factors for COPD in 10,364 non-Hispanic white and African American ever-smokers between the ages of 45 and 80 recruited between 2007 and 2010. This analysis is limited to a subset of the 9,292 COPDGene participants (n = 928 [10%]) previously characterized as having ILA (11), in whom measures of GDF15 (growth differentiation factor 15) were obtained, all of whom had COPD (Global Initiative for Chronic Obstructive Lung Disease [GOLD] grade ≥ 2). Enrollment in phase 1 of COPDGene was between 2008 and 2011. Participants were prospectively followed at 6-month intervals by telephone and web-based inquiry as part of the longitudinal follow-up program to determine mortality, comorbid disease events, and disease status (see Methods in the online supplement for additional details). The institutional review boards at each COPDGene participating center approved the study, and informed consent was obtained.
Chest CT Analysis
CT scans were sequentially read by up to three readers (including radiologists and pulmonologists) who were blinded to participant information or prior imaging interpretations, as previously described (10, 13, 18). ILA have been previously defined as nondependent changes affecting more than 5% of any lung zone, including any combination of nondependent ground-glass or reticular abnormalities, diffuse centrilobular nodularity, nonemphysematous cysts, honeycombing, or traction bronchiectasis. The definition of ILA for this manuscript uses the updated definition of ILA adopted by the Fleischner Society (21) (i.e., excluding those with ILA limited to centrilobular nodules alone on the basis of imaging, genetic, and longitudinal outcome data demonstrating that this subset of ILA should be viewed as a distinct phenotype) (11, 17). CT scans with focal or unilateral ground-glass attenuation, focal or unilateral reticulation, or patchy ground-glass abnormality (<5% of the lung) were considered indeterminate. In exploratory analysis in the FHS, we examined subtypes of ILA. Further subtyping of ILA (e.g., fibrotic vs. nonfibrotic ILA, definite/possible usual interstitial pneumonia [UIP] pattern vs. other UIP pattern) was performed by a consensus of at least three readers as previously described (10, 11).
Selection and Measurement of Aging Biomarkers
We selected aging-related biomarkers for analysis in the FHS on the basis of those prioritized by a geroscience working group for measurement in clinical trials of aging and age-related disease that met the following criteria: measurement reliability and feasibility; biologic relevance to aging; robust and consistent ability to predict all-cause mortality, clinical, and functional outcomes; and responsiveness to intervention (1). The final list of biomarkers meeting the panel’s selection criteria included 12 measures: IL-6, CRP (C-reactive protein), TNF (tumor necrosis factor α), TNFR (TNF receptor II), GDF15, cystatin-C, HGBA1C (Hb A1C), insulin level, IGF1 (insulin-like growth factor 1), IGFBP1 (IGF binding protein 1) and IGFBP3, and NT-proBNP (N-terminal B-type natriuretic peptide). Of these 12 biomarkers, 10 (excluding TNF and NT-proBNP) were measured in a sufficiently large and similar number (n = 1,463–1,538 per biomarker) in blood plasma in the FHS to include in our analysis. The median time between biomarker measures and the subsequent MDCT2 chest CT scan in the FHS ranged between 1 and 7 years (see Table E1 in the online supplement). In the COPDGene Study, measures of GDF-15 were collected from blood plasma coincidentally with the initial chest CT examination.
In the FHS blood plasma, IL-6 was measured by using a quantitative ELISA (intraassay coefficient of variation [CV], 3.12%; R&D Systems). CRP was measured by using particle-enhanced immunonephelometry (CV, 3.20%; Dade Behring BN 100). TNFR was measured by using a quantitative ELISA (CV, 2.25%; R&D Systems). HGBA1C was measured by using a turbidimetric immunoassay (CV 0.9%; Roche Diagnostics). The insulin level was measured by using an ELISA (Linco Research, Inc.) with Roche reagents (R&D Systems), with intraassay CVs of 2–3% across waves of FHS cohorts. A modified ELISA sandwich approach, multiplexed on a Luminex xMAP platform (Sigma‐Aldrich) (22), was used to measure IGF1 (CV, 5.5–9.1%), IGFBP1 (CV, 2.5%), IGFBP3 (CV, 3.9–4.4%), GDF15 (CV, 5.0–5.4%), and cystatin-C (CV, 3.1–3.2%). In COPDGene blood plasma, GDF15 was measured via an ELISA, as previously described (CV, 2.3%; R&D Systems) (23).
Covariates
Covariates were included in models a priori on the basis of their prior consistent associations with either ILA or aging biomarkers. In the FHS, covariates included age, sex, body mass index (BMI), pack-years of smoking, and current smoking. We determined whether coronary arterial calcium (CAC), calculated using the Agatston method (24), and clinical coronary heart disease (CHD) could be potential confounders by first calculating their association with ILA (Tables E2 and E3). After adjustment for age, sex, BMI, current smoking, and smoking pack-years, ILA were not associated with CHD and were only associated with CAC scores > 400. Subsequently, we included CAC and CHD as potential confounders in an additional sensitivity analysis. In the COPDGene Study, we adjusted for age, sex, BMI, pack-years of smoking, and current smoking, with the additional inclusion of race, GOLD stage, and emphysema (percentage < −950 Hounsfield units), to account for selection based on COPD status in the COPDGene cohort. We also determined whether ILA were associated with CAC (Table E4) and then adjusted for CAC as a potential confounder in a sensitivity analysis.
Statistical Analysis
All analyses in the FHS were performed using generalized estimating equation (GEE) clustering on family membership to account for familial correlation (25). To determine whether biomarkers were strongly correlated to each other in the FHS, we calculated pairwise Pearson correlations between biomarkers (Table E5). Descriptive statistics across age categories were computed, and differences between groups were assessed using GEEs in the FHS and using the Fisher’s exact test for categorical variables and the Kruskal-Wallis test for continuous variables in the COPDGene Study. Primary analyses compared only those with and without ILA. For analyses in which having ILA was the primary outcome, we excluded participants whose CT scans were indeterminate for ILA. However, for mortality analyses adjusting for ILA as a covariate, we included participants with scans indeterminate for ILA. To approximate a normal distribution for modeling, we natural log–transformed all biomarkers, with the exceptions of HGBA1C and cystatin-C. Associations between biomarkers and ILA were determined with a logit link (logistic regression) within the GEE, with biomarkers as the independent variable and ILA as the dependent variable. If biomarkers were significantly associated with ILA at this stage, a sensitivity analysis was performed in addition, adjusting for CAC and CHD.
Associations among ILA, biomarkers, and all-cause mortality were assessed with Cox proportional hazard models after confirming the proportionality assumption. To determine the extent to which the associations between individual biomarkers and mortality were attenuated by, or independent of, ILA, additional mortality analyses were performed including ILA as a covariate. All models accounted for familial correlation and were adjusted as above, with a following sensitivity analysis adjusting for CAC and CHD. If biomarkers were significantly associated with mortality after adjusting for CAC and CHD, additional adjustment was performed for mean arterial pressure, high-density lipoprotein, low-density lipoprotein, fasting glucose, alcohol consumption (drinks/wk), and prevalent cancer. The survival period began with the date of the chest CT scan and was censored at death, the end of the follow-up period, or loss to follow-up.
We in addition sought to determine whether biomarkers attenuated the association of age with ILA to provide more evidence for possible causality in our observational data. To do so, we conducted causal mediation analysis in a subsample of unrelated FHS participants (n = 743–794 participants per biomarker) with the R mediate function (R Foundation for Statistical Computing). Although the effect size of each biomarker is dependent on the ages selected for comparison, the P value for the mediation effect is robust to change in the age comparison. Subsequently, we emphasize the P values for each biomarker in the mediation analysis but also report the percentage of attenuation of the age coefficient. All causal mediation analyses in the FHS were adjusted for age, sex, BMI, pack-years of smoking, and current smoking.
Similar logistic and survival analyses with GDF-15 were performed in the COPDGene Study (without adjustment for familial correlation) for validation. Casual mediation analysis in the COPDGene Study was performed with the PROC CAUSALMED procedure in SAS (SAS Institute Inc.) with the same parameter settings used for the R mediate function. All analyses in the COPDGene Study adjusted for age, sex, BMI, pack-years of smoking, current smoking, race, GOLD stage, and the percentage of emphysematous lung at <−950 Hounsfield units (percent emphysema). In COPDGene, we performed an additional sensitivity analysis adjusting for CAC.
In the FHS, we adjusted for multiple comparisons, such that a two-sided P value < 0.005 (0.05/10 biomarkers) was used to determine statistical significance. In the COPDGene Study, given that GDF15 was the only biomarker assessed for replication, a two-sided P value < 0.05 was considered to indicate statistical significance.
Results
Population Samples by Age Group
All correlations between biomarkers, adjusted only for familial relationship, were <0.3, except for the correlations between GDF15 and IL6 (0.31), HGBA1C (0.36), and Cystatin-C (0.64) (Table E5). Descriptive statistics of the baseline characteristics of FHS and COPDGene participants by age group are presented in Table 1. Of the 1,541 FHS participants, 176 (11%) had ILA, with an increasing prevalence by age group. Of the 928 COPDGene participants, 34 (4%) had ILA, with the greatest prevalence noted among those ≥70 years of age but more variability in the other age groups. In both cohorts, older participants were more likely to have a greater cumulative smoke exposure but were less likely to be actively smoking. Older participants were more likely to be women in the FHS versus men in the COPDGene Study, a difference likely related to the respective recruitment strategies (community-based recruitment for the FHS vs. recruitment based on a greater smoking history for the COPDGene Study). In the FHS, older age was associated with a greater burden of CAC. In the COPDGene Study, older age was associated with a greater percent emphysema and CAC. In the COPDGene Study, African Americans were underrepresented among the older age groups. Sex was not significantly associated with differences in ILA between age groups in either cohort.
Table 1.
Baseline Characteristics in the Framingham Heart Study and COPDGene Cohort Stratified by Age Group
| Age Group |
P Value* | |||
|---|---|---|---|---|
| Age < 60 yr [Number (%) or Median (IQR)] | Age ≥ 60 yr and <70 yr [Number (%) or Median (IQR)] | Age ≥ 70 yr [Number (%) or Median (IQR)] | ||
| Framingham Heart Study | 1,017 (66) | 332 (22) | 192 (12) | — |
| ILA | 44 (4) | 41 (12) | 91 (47) | <0.0001 |
| Sex, F | 473 (47) | 189 (57) | 99 (52) | 0.43 |
| BMI, kg/m2 | 28 (25–32) | 28 (25–32) | 28 (25–31) | <0.0001 |
| Pack-years smoking | 0 (0–8) | 5 (0–20) | 8 (0–24) | <0.0001 |
| Current smoking | 71 (7) | 16 (5) | 3 (2) | <0.0001 |
| CAC score† | 0 (0–19) | 48 (0–270) | 246 (40–725) | <0.0001 |
| COPDGene Study‡ | 272 (29) | 410 (44) | 246 (27) | — |
| ILA | 10 (4) | 10 (2) | 14 (6) | <0.0001 |
| Sex, F | 128 (47) | 171 (42) | 97 (39) | 0.09 |
| BMI | 27 (23–31) | 27 (23–31) | 27 (24–31) | 0.74 |
| Pack-years smoking | 41 (32–57) | 51 (40–74) | 56 (39–76) | <0.0001 |
| Current smoking | 148 (54) | 111 (27) | 26 (11) | <0.0001 |
| CAC score† | 2 (0–102) | 86 (3–291) | 193 (29–543) | <0.0001 |
| African American race | 85 (31) | 59 (14) | 19 (8) | <0.0001 |
| GOLD stage | 0.01 | |||
| 2 | 114 (38) | 131 (30) | 88 (34) | |
| 3 | 99 (33) | 138 (31) | 95 (37) | |
| 4 | 84 (28) | 172 (52) | 77 (30) | |
| Percent emphysema (<−950 HU)§ | 7 (2–23) | 16 (5–29) | 16 (7–26) | 0.004 |
Definition of abbreviations: BMI = body mass index; CAC = coronary arterial calcium; COPDGene = Genetic Epidemiology of Chronic Obstructive Pulmonary Disease; GOLD = Global Initiative for Chronic Obstructive Lung Disease; HU = Hounsfield units; ILA = interstitial lung abnormalities; IQR = interquartile range.
P values represent the between-group differences. In the Framingham Heart Study, P values are from generalized estimating equations to adjust for familial correlation. In the COPDGene Study, P values are from Fisher exact tests for categorical variables and Kruskal-Wallis tests for continuous variables.
The CAC score is calculated using the Agatston scoring method (24). There were 158 COPDGene participants missing a CAC score in this subset.
All COPDGene participants in this subset had a GOLD grade ≥ 2 for chronic obstructive pulmonary disease.
Percent emphysema is calculated as the percentage of the lung below −950 Hounsfield units. There were 37 COPDGene participants missing percent emphysema in this subset.
Association of Aging Biomarkers with ILA in the FHS
In the FHS, after adjusting for age, sex, BMI, smoking pack-years, and current smoking, age was associated with ILA. For each 10-year increase in age, the odds ratio (OR) (95% confidence interval [CI]) was 3.5 (95% CI, 2.8–4.2) (P < 0.0001). Biomarker sample sizes and unadjusted raw values by ILA status in the FHS are shown in Table E6. Although most biomarkers were associated with ILA in analyses limited to adjustment for familial correlation, after additional adjustment for age, sex, BMI, smoking pack-years, and current smoking, higher levels of four biomarkers (GDF15, TNFR, IL-6, and CRP) remained significantly associated with increased odds of ILA (P < 0.005) (Table 2). Additional adjustment for CAC and CHD did not substantially change the association of these five biomarkers with ILA (Table 2). Although underpowered among the 174 cases of ILA, there were no significant differences in the association of biomarkers with a specific subtype of ILA. For example, the odds of fibrotic ILA versus nonfibrotic ILA for a higher natural log–transformed GDF15 were 0.69 (95% CI, 0.22–2.14) (P = 0.52), and the odds of definite/possible UIP pattern versus another UIP pattern were 0.39 (95% CI, 0.15–1.0) (P = 0.05) (see Results in the online supplement).
Table 2.
Association of Aging Biomarkers with ILA in the Framingham Heart Study
| Biomarker | Unadjusted Odds of ILA |
Adjusted Odds of ILA* |
Adjusted Odds of ILA† |
||||
|---|---|---|---|---|---|---|---|
| OR (95% CI) | P Value | OR (95% CI) | P Value | OR (95% CI) | P Value | ||
| GDF15, ln | 12.2 (7.8–19.2) | <0.0001 | 3.2 (1.7–5.9) | 0.0002 | 3.4 (1.8–6.4) | 0.0002 | |
| TNFR, ln | 8.0 (4.8–13.1) | <0.0001 | 2.4 (1.3–4.3) | 0.004 | 3.1 (1.6–5.8) | 0.0004 | |
| IL-6, ln | 2.3 (1.9–2.8) | <0.0001 | 1.8 (1.4–2.2) | <0.0001 | 1.8 (1.4–2.4) | <0.0001 | |
| CRP, ln | 1.5 (1.3–1.7) | <0.0001 | 1.5 (1.3–1.8) | <0.0001 | 1.7 (1.3–2.0) | <0.0001 | |
| Insulin, ln | 1.4 (1.1–1.9) | 0.01 | 1.6 (1.1–2.2) | 0.01 | 1.7 (1.4–2.0) | 0.01 | |
| HGBA1C, 1% | 1.5 (1.2–1.9) | 0.0003 | 1.2 (0.9–1.5) | 0.16 | — | — | |
| Cystatin-C, 0.1 mg/L | 1.3 (1.2–1.5) | <0.0001 | 1.1 (1.0–1.2) | 0.21 | — | — | |
| IGFBP1, ln | 1.3 (1.1–1.5) | 0.0006 | 1.0 (0.8–1.3) | 0.79 | — | — | |
| IGF1, ln | 1.0 (1.0–1.0) | 0.02 | 1.0 (1.0–1.0) | 0.37 | — | — | |
| IGFBP3, ln | 1.0 (1.0–1.0) | 0.13 | 1.0 (1.0–1.0) | 0.66 | — | — | |
Definition of abbreviations: CI = confidence interval; CRP = C-reactive protein; GDF15 = growth differentiation factor 15; HGBA1C = Hb A1C; IGF = insulin-like growth factor; IGFBP = IGF binding protein; ILA = interstitial lung abnormalities; ln = natural log transformed; OR = odds ratio; TNFR = tumor necrosis factor α receptor II.
All models are adjusted for age, sex, body mass index, current smoking, smoking pack-years, and familial correlation. Each biomarker is included in its own model. ORs depict an ln increase in biomarkers, as noted.
Additionally adjusted for coronary arterial calcium score and adjudicated clinical coronary heart disease.
Association of Aging Biomarkers and ILA with All-Cause Mortality in the FHS
In the FHS, in an initial model without biomarkers that predicted death using ILA, age, sex, BMI, smoking pack-years, and current smoking, ILA and older age were associated with increased mortality. Compared with those without ILA, those with ILA had a 92% greater hazard of death (hazard ratio [HR], 1.9; 95% CI, 1.2–3.2; P = 0.01). Each 10-year increase in age was associated with increased mortality (HR, 3.6; 95% CI, 3.0–4.5; P < 0.0001). Most biomarkers were associated with increased mortality in models accounting only for familial relationships (Table 3). After adjustment for ILA, age, sex, BMI, smoking pack-years, and current smoking, GDF15 and TNFR remained associated with greater mortality (P < 0.005) (Table 3). In sensitivity analyses adjusting in addition for CAC and CHD, GDF15 (P = 0.02) and TNFR (P = 0.05) were not associated with mortality after accounting for multiple comparisons (Table 3). Additional adjustment for mean arterial pressure, high-density lipoprotein, low-density lipoprotein, fasting glucose, alcohol consumption, and prevalent cancer did not significantly alter the results, except in the case of HGBA1C, which was no longer significantly associated with mortality (data not shown). In the final models predicting death from familial relationships, each individual biomarker (entered in separate models), ILA, age, sex, BMI, current smoking, smoking pack-years, CAC, and CHD, definite ILA were consistently associated with increased mortality, CHD was nearly always associated with increased mortality, and CAC was not associated with increased mortality (Table E7).
Table 3.
Association of Aging Biomarkers with Mortality in the Framingham Heart Study
| Biomarker | Unadjusted Hazards of Death |
Adjusted Hazards of Death* |
Adjusted Hazards of Death† |
|||
|---|---|---|---|---|---|---|
| HR (95% CI) | P Value | HR (95% CI) | P Value | HR (95% CI) | P Value | |
| GDF15, ln | 7.7 (5.3–11.4) | <0.0001 | 2.1 (1.3–3.5) | 0.002 | 2.0 (1.1–3.5) | 0.02 |
| TNFR, ln | 10.3 (6.7–15.7) | <0.0001 | 2.3 (1.3–4.0) | 0.003 | 1.8 (1.0–3.3) | 0.05 |
| IL-6, ln | 2.0 (1.6–2.4) | <0.0001 | 1.4 (1.1–3.0) | 0.02 | 1.3 (1.0–1.8) | 0.07 |
| HGBA1C, 1% | 1.5 (1.3–1.6) | <0.0001 | 1.3 (1.0–1.6) | 0.02 | 1.3 (1.1–1.6) | 0.007 |
| IGFBP1, ln | 1.4 (1.2–1.6) | <0.0001 | 1.3 (1.1–1.6) | 0.02 | 1.3 (1.1–1.7) | 0.01 |
| Cystatin-C, 0.1 mg/L | 1.2 (1.2–1.3) | <0.0001 | 1.1 (1.0–1.2) | 0.02 | 2.0 (0.9–4.4) | 0.08 |
| Insulin, ln | 1.4 (1.1–1.9) | 0.009 | 1.0 (0.7–1.4) | 0.86 | — | — |
| CRP, ln | 1.2 (1.0–1.4) | 0.04 | 1.1 (0.9–1.3) | 0.35 | — | — |
| IGFBP3, ln | 1.0 (1.0–1.0) | 0.08 | 1.0 (1.0–1.0) | 0.87 | — | — |
| IGF1, ln | 1.0 (1.0–1.0) | 0.14 | 1.0 (1.0–1.0) | 0.36 | — | — |
Definition of abbreviations: CI = confidence interval; CRP = C-reactive protein; GDF15 = growth differentiation factor 15; HR = hazard ratio; HGBA1C = Hb A1C; IGF = insulin-like growth factor; IGFBP = IGF binding protein; ln = natural log–transformed; TNFR = tumor necrosis factor α receptor II.
All models adjusted for interstitial lung abnormalities, age, sex, body mass index, current smoking, smoking pack-years, and familial correlation. Each biomarker is included in its own model. HRs depict an ln increase in biomarkers, as noted.
Additionally adjusted for coronary arterial calcium score and adjudicated clinical coronary heart disease.
Attenuation of the Association of Age with ILA by Aging Biomarkers in the FHS
We conducted causal mediation analysis (Figure 1), with all models adjusted for age, sex, BMI, smoking pack-years, and current smoking, to determine whether the association of age with ILA was mediated by aging-related biomarkers. In the causal diagram (Figure 1), the total effect of age and a specific biomarker on ILA is denoted c. The average direct effect of age on ILA is denoted c’. The subtraction c − c’ yields the average causal mediation effect of the biomarker, ab. The significance of the average causal mediation effect reveals whether the biomarker explains some of the association of age with ILA. We found that TNFR (P = 0.002) and IL-6 (P < 0.0001) significantly mediated the association of age with ILA and accounted for 8% and 9%, respectively, of the association of age with ILA. There was a suggestion that the association of age with ILA was mediated by GDF15 (P = 0.008), accounting for 22% of the association of age with ILA, but it did not meet statistical significance after adjustment for multiple comparisons. All other biomarkers did not show significant mediation (P > 0.05).
Figure 1.
Schematic for causal mediation analysis. Lowercase letters represent the effect of one variable on another variable, with c − c’ = the average causal mediation effect of the biomarker. a = the effect of age on biomarker; b = the effect of biomarker on ILA; c = total effect of age and biomarker on ILA; c’ = average direct effect of age on ILA; ILA = interstitial lung abnormalities.
Replication of GDF15 Results in the COPDGene Study
In the COPDGene Study, after adjustment for age, sex, BMI, smoking pack-years, current smoking, race, GOLD stage, and percent emphysema, higher GDF15 was associated with increased odds of ILA (OR per natural log–transformed GDF15, 8.1 [3.1–21.4]; P < 0.0001), which remained significant after additional adjustment for CAC (OR, 8.7 [2.6–29.2]; P = 0.0004). Similarly, in mortality models adjusted for age, sex, BMI, smoking pack-years, current smoking, race, GOLD stage, and percent emphysema, GDF15 was associated with an increased rate of mortality (HR, 1.6 [1.1–2.2]; P = 0.01), which remained significant after additional adjustment for CAC (HR, 1.7 [1.1–2.7]; P = 0.02). GDF15 also significantly mediated the association of age with ILA (P = 0.001) in a causal inference model, accounting for 58% of the association of age with ILA.
Discussion
Our study demonstrates that the odds of ILA increase with higher GDF15, TNFR, IL-6, and CRP. The inclusion of ILA as a covariate in traditional noncausal survival models also attenuates many previously documented associations between these biomarkers and mortality. We show by causal modeling that GDF15, TNFR, and IL-6 partially account for the statistical association of age with ILA in these study samples. Moreover, associations between GDF15 and ILA were robust in two independent cohorts. Our findings, in conjunction with those previously published (26), implicate a potential role for GDF15, in particular, in the early developmental processes of pulmonary fibrosis and suggest that some of this association may be driven by facets of biologic aging captured by this measurement.
Our study adds to the growing number of studies that have demonstrated evidence for overlapping associations between selected biomarker measures in patients with IPF (26–29) and in research participants with undiagnosed, early stages of pulmonary fibrosis (30–32). It is important to note that in this study, and others (30, 32), biomarkers were measured many years before the detection of visually apparent imaging abnormalities. Although we cannot rule out the possibility that imaging abnormalities may have been present if chest imaging were performed coincidentally with biomarker measurement, these findings suggest that, similar to genetic markers (11, 33), peripheral blood biomarkers may be helpful in detecting not only groups at risk of having coexisting ILA but also groups that may be at an increased risk to develop these imaging abnormalities.
These findings have implications for gerontology research as well. Numerous efforts to identify ideal biomarkers of accelerated aging have focused on analytes reproducibly associated with mortality and various clinical outcomes, even after adjusting for multiple common disease comorbidities (1, 34). Biomarkers meeting these criteria are pleiotropic and likely represent distinct molecular mechanisms with varying links to known disease states. Our findings demonstrate that the previously documented associations between some biomarkers (e.g., IL-6, CRP, and insulin) and mortality appear to be substantially attenuated by the inclusion of ILA in the models. Moreover, ILA remained associated with mortality in the presence of these biomarkers, whereas CAC did not. These findings suggest that undiagnosed ILA could explain some associations between selected biomarkers and mortality. Gerontologic studies may want to consider both the importance of identifying ILA in future studies of biomarker selection and the possible role that ILA could play as independent imaging biomarkers of aging. Incorporating these measurements into causal models may further clarify their interrelationships and dependencies.
The consistent associations among GDF15, ILA, and mortality are notable and warrant further mention. GDF15 (also referred to as macrophage inhibitory cytokine-1) is a member of the TGF-β (transforming growth factor β) superfamily but shares only 15–29% of its sequence homology with other TGF-β members (35). GDF15 levels are known to increase with age (36, 37) in response to numerous cellular stresses (e.g., mitochondrial dysfunction [38], oxidative stress [39], and cellular senescence [40]) and with metformin administration (41), which may explain the known consistent associations among GDF15, numerous disease states (42), and mortality (43–45). Recently, GDF15 was identified as the most significantly upregulated protein in response to type 2 epithelial cell senescence, and elevated levels were found in the lungs and blood of patients with IPF (26). Our findings add to this literature and suggest that elevated GDF15 levels can also be found in the blood of those with imaging abnormalities suggestive of early pulmonary fibrosis and may even precede their development. The evidence that GDF15 concentrations mediate the association of age with ILA suggests that accelerated aging, or more specific biologic processes captured by GDF15 measurement, may play an important part in, or are correlated with, pulmonary fibrosis development. Reciprocally, these findings suggest that ILA may, in part, represent a visual manifestation of accelerated pulmonary aging.
Our study has a number of important limitations. First, although we include results from >1,400 participants for 10 of the 12 biomarkers recently identified as reproducibly associated with accelerated aging by a gerontologic working group (1), not all possible aging-associated biomarkers were analyzed. Second, although we present significant findings of association in the FHS for biomarkers measured in participants at a point during a range of years preceding their chest CT, we cannot rule out the possibility that some analyses could have been affected by variances in these temporal relationships. Notably, in the COPDGene Study, GDF15 was measured coincidentally with chest CT examination, and associations were similar to those in the FHS. Third, although this study demonstrates the associations between multiple biomarkers associated with accelerated aging and ILA, it is important to remember that most of these biomarkers are pleiotropic and known to be associated with other disease states (46–48). Fourth, some selection bias may have occurred because of availability of biomarker measurements; for example, the prevalence of ILA was higher in the FHS and lower in the COPDGene Study than that previously reported using larger baseline samples (18). Fifth, we were underpowered to detect significant associations between biomarkers and specific subtypes of ILA. Sixth, residual differences in the structure of the age data or their correlated variables between the FHS and the COPDGene Study may have influenced the percentage of attenuation of the age coefficient, which could explain why GDF15 accounted for 22% of the age effect in the FHS and 58% of the age effect in the COPDGene Study. Finally, we cannot exclude residual confounding.
An underlying tenet of geroscience is identifying and leveraging a broad, age-associated risk profile to target aging for pleiotropic risk reduction. Our findings add further weight to the evidence that ILA may also represent an important manifestation of accelerated aging (10, 13, 18). Our findings also suggest that some, but not all, biomarkers associated with accelerated aging may be helpful in identifying ILA. Further work is necessary to identify the specific molecular pathways and cellular stress responses that lead to an increased risk of developing ILA with increasing age.
Conclusions
This study demonstrates that some, but not all, biomarkers associated with underlying aging processes are also associated with ILA. In particular, GDF15 may explain some of the associations among age, ILA, and mortality. These results suggest that some aging-associated pathways captured by selected biomarker measures may be important in increasing the risk of ILA and, reciprocally, that ILA, which are imaging abnormalities in research participants not known to have interstitial lung disease, should be considered in gerontologic studies attempting to identify biomarkers of accelerated aging. Although geroscience clinical trials targeting aging mechanisms with the hope of reducing risk of multiple diseases are underway (49), these findings raise the possibility that some selected interventions may also reduce the risk of developing the early stages of pulmonary fibrosis. Future studies hoping to reduce the risk of ILA progression in high-risk groups may benefit from targeting some specific aging-associated pathways.
Footnotes
J.L.S. is supported by NIH grant T32 HL007633. R.K.P. is supported by NIH grant K08 HL140087. J.M.M. is supported by NIH grants R01 AG067457 and R01 HL141434. M.N. is supported by NIH grant R01 CA203636. D.L.L. is supported by NIH grants ZIA HL006001 and R35 GM134885. G.R.W. is supported by NIH grants U01 HL146408 and R01 HL122464. J.L.C. is supported by NIH grant R01 HL144718 and a grant from the Veterans Health Administration (I01 CX000911). R.P.B. is supported by NIH grants R01 HL137995, R01 152735, and R21 HL140376. G.T.O’C. is supported by NIH grant OT2 OD026553. G.M.H. and this work are supported by NIH grants R01 HL111024, R01135142, and R01130974. The FHS (Framingham Heart Study) of the NHLBI of the NIH and Boston University School of Medicine has been funded in whole or in part with federal funds from the NHLBI, NIH, Department of Health and Human Services, under contract numbers N01-HC25195, HHSN268201500001I, and 75N92019D00031. Additional support is from 1R01 HL64753, R01 HL076784, 1R01 AG028321, 2R01 HL092577, 2U54HL120163, R01 AG031287, and R01 HL077477. The COPDGene (Genetic Epidemiology of Chronic Obstructive Pulmonary Disease [COPD]) Study is supported by NIH grants R01 HL089897 and R01 HL089856. The COPDGene Study (NCT 00608764) is also supported by the COPD Foundation through contributions made to an industry advisory board composed of representatives from AstraZeneca, Boehringer Ingelheim, GlaxoSmithKline, Novartis, Pfizer, Siemens, and Sunovion.
Author Contributions: Conception and design: J.L.S., J.D., and G.M.H. Acquisition, analysis, and interpretation of data: J.L.S., R.K.P., J.D., H.X., J.M.M., T.A., M.N., E.J.B., D.L.L., V.S.R., G.R.W., J.L.C., C.M.F., R.P.B., H.H., G.T.O’C., and G.M.H. Drafting of the work and critical revisions: J.L.S., R.K.P., J.D., H.X., J.M.M., T.A., M.N., E.J.B., D.L.L., V.S.R., G.R.W., J.L.C., C.M.F., R.P.B., H.H., G.T.O’C., and G.M.H. Accountability for the work: J.L.S., J.D., and G.M.H.
This article has an online supplement, which is accessible from this issue’s table of contents at www.atsjournals.org.
Originally Published in Press as DOI: 10.1164/rccm.202007-2993OC on October 20, 2020
Author disclosures are available with the text of this article at www.atsjournals.org.
References
- 1.Justice JN, Ferrucci L, Newman AB, Aroda VR, Bahnson JL, Divers J, et al. A framework for selection of blood-based biomarkers for geroscience-guided clinical trials: report from the TAME Biomarkers workgroup. Geroscience. 2018;40:419–436. doi: 10.1007/s11357-018-0042-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Wolters PJ, Collard HR, Jones KD. Pathogenesis of idiopathic pulmonary fibrosis. Annu Rev Pathol. 2014;9:157–179. doi: 10.1146/annurev-pathol-012513-104706. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Gulati S, Thannickal VJ. The aging lung and idiopathic pulmonary fibrosis. Am J Med Sci. 2019;357:384–389. doi: 10.1016/j.amjms.2019.02.008. [DOI] [PubMed] [Google Scholar]
- 4.Selman M, Martinez FJ, Pardo A. Why does an aging smoker’s lung develop idiopathic pulmonary fibrosis and not chronic obstructive pulmonary disease? Am J Respir Crit Care Med. 2019;199:279–285. doi: 10.1164/rccm.201806-1166PP. [DOI] [PubMed] [Google Scholar]
- 5.Stuart BD, Lee JS, Kozlitina J, Noth I, Devine MS, Glazer CS, et al. Effect of telomere length on survival in patients with idiopathic pulmonary fibrosis: an observational cohort study with independent validation. Lancet Respir Med. 2014;2:557–565. doi: 10.1016/S2213-2600(14)70124-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Townsley DM, Dumitriu B, Liu D, Biancotto A, Weinstein B, Chen C, et al. Danazol treatment for telomere diseases. N Engl J Med. 2016;374:1922–1931. doi: 10.1056/NEJMoa1515319. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Justice JN, Nambiar AM, Tchkonia T, LeBrasseur NK, Pascual R, Hashmi SK, et al. Senolytics in idiopathic pulmonary fibrosis: results from a first-in-human, open-label, pilot study. EBioMedicine. 2019;40:554–563. doi: 10.1016/j.ebiom.2018.12.052. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Hunninghake GM. Interstitial lung abnormalities: erecting fences in the path towards advanced pulmonary fibrosis. Thorax. 2019;74:506–511. doi: 10.1136/thoraxjnl-2018-212446. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Kropski JA, Pritchett JM, Zoz DF, Crossno PF, Markin C, Garnett ET, et al. Extensive phenotyping of individuals at risk for familial interstitial pneumonia reveals clues to the pathogenesis of interstitial lung disease. Am J Respir Crit Care Med. 2015;191:417–426. doi: 10.1164/rccm.201406-1162OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Hunninghake GM, Hatabu H, Okajima Y, Gao W, Dupuis J, Latourelle JC, et al. MUC5B promoter polymorphism and interstitial lung abnormalities. N Engl J Med. 2013;368:2192–2200. doi: 10.1056/NEJMoa1216076. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Putman RK, Gudmundsson G, Araki T, Nishino M, Sigurdsson S, Gudmundsson EF, et al. The MUC5B promoter polymorphism is associated with specific interstitial lung abnormality subtypes. Eur Respir J. 2017;50:1700537. doi: 10.1183/13993003.00537-2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Miller ER, Putman RK, Vivero M, Hung Y, Araki T, Nishino M, et al. Histopathology of interstitial lung abnormalities in the context of lung nodule resections. Am J Respir Crit Care Med. 2018;197:955–958. doi: 10.1164/rccm.201708-1679LE. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Washko GR, Hunninghake GM, Fernandez IE, Nishino M, Okajima Y, Yamashiro T, et al. COPDGene Investigators. Lung volumes and emphysema in smokers with interstitial lung abnormalities. N Engl J Med. 2011;364:897–906. doi: 10.1056/NEJMoa1007285. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Doyle TJ, Washko GR, Fernandez IE, Nishino M, Okajima Y, Yamashiro T, et al. COPDGene Investigators. Interstitial lung abnormalities and reduced exercise capacity. Am J Respir Crit Care Med. 2012;185:756–762. doi: 10.1164/rccm.201109-1618OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Podolanczuk AJ, Oelsner EC, Barr RG, Hoffman EA, Armstrong HF, Austin JH, et al. High attenuation areas on chest computed tomography in community-dwelling adults: the MESA Study. Eur Respir J. 2016;48:1442–1452. doi: 10.1183/13993003.00129-2016. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Araki T, Putman RK, Hatabu H, Gao W, Dupuis J, Latourelle JC, et al. Development and progression of interstitial lung abnormalities in the Framingham Heart Study. Am J Respir Crit Care Med. 2016;194:1514–1522. doi: 10.1164/rccm.201512-2523OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Putman RK, Gudmundsson G, Axelsson GT, Hida T, Honda O, Araki T, et al. Imaging patterns are associated with interstitial lung abnormality progression and mortality. Am J Respir Crit Care Med. 2019;200:175–183. doi: 10.1164/rccm.201809-1652OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Putman RK, Hatabu H, Araki T, Gudmundsson G, Gao W, Nishino M, et al. Evaluation of COPD Longitudinally to Identify Predictive Surrogate Endpoints (ECLIPSE) Investigators; COPDGene Investigators. Association between interstitial lung abnormalities and all-cause mortality. JAMA. 2016;315:672–681. doi: 10.1001/jama.2016.0518. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Pompe E, de Jong PA, Lynch DA, Lessmann N, Išgum I, van Ginneken B, et al. Computed tomographic findings in subjects who died from respiratory disease in the National Lung Screening Trial. Eur Respir J. 2017;49:1601814. doi: 10.1183/13993003.01814-2016. [DOI] [PubMed] [Google Scholar]
- 20.Dawber TR, Meadors GF, Moore FE., Jr Epidemiological approaches to heart disease: the Framingham study. Am J Public Health Nations Health. 1951;41:279–281. doi: 10.2105/ajph.41.3.279. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Hatabu H, Hunninghake GM, Richeldi L, Brown KK, Wells AU, Remy-Jardin M, et al. Interstitial lung abnormalities detected incidentally on CT: a position paper from the Fleischner Society. Lancet Respir Med. 2020;8:726–737. doi: 10.1016/S2213-2600(20)30168-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Ho JE, Lyass A, Courchesne P, Chen G, Liu C, Yin X, et al. Protein biomarkers of cardiovascular disease and mortality in the community. J Am Heart Assoc. 2018;7:e008108. doi: 10.1161/JAHA.117.008108. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Martinez CH, Freeman CM, Nelson JD, Murray S, Wang X, Budoff MJ, et al. COPDGene Investigators. GDF-15 plasma levels in chronic obstructive pulmonary disease are associated with subclinical coronary artery disease. Respir Res. 2017;18:42. doi: 10.1186/s12931-017-0521-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Agatston AS, Janowitz WR, Hildner FJ, Zusmer NR, Viamonte M, Jr, Detrano R. Quantification of coronary artery calcium using ultrafast computed tomography. J Am Coll Cardiol. 1990;15:827–832. doi: 10.1016/0735-1097(90)90282-t. [DOI] [PubMed] [Google Scholar]
- 25.Uh HW, Wijk HJ, Houwing-Duistermaat JJ. Testing for genetic association taking into account phenotypic information of relatives. BMC Proc. 2009;3:S123. doi: 10.1186/1753-6561-3-s7-s123. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Zhang Y, Jiang M, Nouraie M, Roth MG, Tabib T, Winters S, et al. GDF15 is an epithelial-derived biomarker of idiopathic pulmonary fibrosis. Am J Physiol Lung Cell Mol Physiol. 2019;317:L510–L521. doi: 10.1152/ajplung.00062.2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Mackinnon AC, Gibbons MA, Farnworth SL, Leffler H, Nilsson UJ, Delaine T, et al. Regulation of transforming growth factor-β1-driven lung fibrosis by galectin-3. Am J Respir Crit Care Med. 2012;185:537–546. doi: 10.1164/rccm.201106-0965OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Rosas IO, Richards TJ, Konishi K, Zhang Y, Gibson K, Lokshin AE, et al. MMP1 and MMP7 as potential peripheral blood biomarkers in idiopathic pulmonary fibrosis. PLoS Med. 2008;5:e93. doi: 10.1371/journal.pmed.0050093. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Richards TJ, Kaminski N, Baribaud F, Flavin S, Brodmerkel C, Horowitz D, et al. Peripheral blood proteins predict mortality in idiopathic pulmonary fibrosis. Am J Respir Crit Care Med. 2012;185:67–76. doi: 10.1164/rccm.201101-0058OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Ho JE, Gao W, Levy D, Santhanakrishnan R, Araki T, Rosas IO, et al. Galectin-3 is associated with restrictive lung disease and interstitial lung abnormalities. Am J Respir Crit Care Med. 2016;194:77–83. doi: 10.1164/rccm.201509-1753OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.McGroder CF, Aaron CP, Bielinski SJ, Kawut SM, Tracy RP, Raghu G, et al. Circulating adhesion molecules and subclinical interstitial lung disease: the multi-ethnic study of atherosclerosis. Eur Respir J. 2019;54:1900295. doi: 10.1183/13993003.00295-2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Armstrong HF, Podolanczuk AJ, Barr RG, Oelsner EC, Kawut SM, Hoffman EA, et al. MESA (Multi-Ethnic Study of Atherosclerosis) Serum matrix metalloproteinase-7, respiratory symptoms, and mortality in community-dwelling adults. Am J Respir Crit Care Med. 2017;196:1311–1317. doi: 10.1164/rccm.201701-0254OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Hobbs BD, Putman RK, Araki T, Nishino M, Gudmundsson G, Gudnason V, et al. Overlap of genetic risk between interstitial lung abnormalities and idiopathic pulmonary fibrosis. Am J Respir Crit Care Med. 2019;200:1402–1413. doi: 10.1164/rccm.201903-0511OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Kennedy BK, Berger SL, Brunet A, Campisi J, Cuervo AM, Epel ES, et al. Geroscience: linking aging to chronic disease. Cell. 2014;159:709–713. doi: 10.1016/j.cell.2014.10.039. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Bootcov MR, Bauskin AR, Valenzuela SM, Moore AG, Bansal M, He XY, et al. MIC-1, a novel macrophage inhibitory cytokine, is a divergent member of the TGF-beta superfamily. Proc Natl Acad Sci U S A. 1997;94:11514–11519. doi: 10.1073/pnas.94.21.11514. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Brown DA, Stephan C, Ward RL, Law M, Hunter M, Bauskin AR, et al. Measurement of serum levels of macrophage inhibitory cytokine 1 combined with prostate-specific antigen improves prostate cancer diagnosis. Clin Cancer Res. 2006;12:89–96. doi: 10.1158/1078-0432.CCR-05-1331. [DOI] [PubMed] [Google Scholar]
- 37.Tanaka T, Biancotto A, Moaddel R, Moore AZ, Gonzalez-Freire M, Aon MA, et al. CHI Consortium. Plasma proteomic signature of age in healthy humans. Aging Cell. 2018;17:e12799. doi: 10.1111/acel.12799. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Fujita Y, Taniguchi Y, Shinkai S, Tanaka M, Ito M. Secreted growth differentiation factor 15 as a potential biomarker for mitochondrial dysfunctions in aging and age-related disorders. Geriatr Gerontol Int. 2016;16:17–29. doi: 10.1111/ggi.12724. [DOI] [PubMed] [Google Scholar]
- 39.Han ES, Muller FL, Pérez VI, Qi W, Liang H, Xi L, et al. The in vivo gene expression signature of oxidative stress. Physiol Genomics. 2008;34:112–126. doi: 10.1152/physiolgenomics.00239.2007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Basisty N, Kale A, Jeon OH, Kuehnemann C, Payne T, Rao C, et al. A proteomic atlas of senescence-associated secretomes for aging biomarker development. PLoS Biol. 2020;18:e3000599. doi: 10.1371/journal.pbio.3000599. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Coll AP, Chen M, Taskar P, Rimmington D, Patel S, Tadross JA, et al. GDF15 mediates the effects of metformin on body weight and energy balance. Nature. 2020;578:444–448. doi: 10.1038/s41586-019-1911-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Verhamme FM, Freeman CM, Brusselle GG, Bracke KR, Curtis JL. GDF-15 in pulmonary and critical care medicine. Am J Respir Cell Mol Biol. 2019;60:621–628. doi: 10.1165/rcmb.2018-0379TR. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Daniels LB, Clopton P, Laughlin GA, Maisel AS, Barrett-Connor E. Growth-differentiation factor-15 is a robust, independent predictor of 11-year mortality risk in community-dwelling older adults: the Rancho Bernardo Study. Circulation. 2011;123:2101–2110. doi: 10.1161/CIRCULATIONAHA.110.979740. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Rohatgi A, Patel P, Das SR, Ayers CR, Khera A, Martinez-Rumayor A, et al. Association of growth differentiation factor-15 with coronary atherosclerosis and mortality in a young, multiethnic population: observations from the Dallas Heart Study. Clin Chem. 2012;58:172–182. doi: 10.1373/clinchem.2011.171926. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Wang TJ, Wollert KC, Larson MG, Coglianese E, McCabe EL, Cheng S, et al. Prognostic utility of novel biomarkers of cardiovascular stress: the Framingham Heart Study. Circulation. 2012;126:1596–1604. doi: 10.1161/CIRCULATIONAHA.112.129437. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Wollert KC, Kempf T, Wallentin L. Growth differentiation factor 15 as a biomarker in cardiovascular disease. Clin Chem. 2017;63:140–151. doi: 10.1373/clinchem.2016.255174. [DOI] [PubMed] [Google Scholar]
- 47.Nickel N, Jonigk D, Kempf T, Bockmeyer CL, Maegel L, Rische J, et al. GDF-15 is abundantly expressed in plexiform lesions in patients with pulmonary arterial hypertension and affects proliferation and apoptosis of pulmonary endothelial cells. Respir Res. 2011;12:62. doi: 10.1186/1465-9921-12-62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Emmerson PJ, Duffin KL, Chintharlapalli S, Wu X. GDF15 and growth control. Front Physiol. 2018;9:1712. doi: 10.3389/fphys.2018.01712. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Justice J, Miller JD, Newman JC, Hashmi SK, Halter J, Austad SN, et al. Frameworks for proof-of-concept clinical trials of interventions that target fundamental aging processes. J Gerontol A Biol Sci Med Sci. 2016;71:1415–1423. doi: 10.1093/gerona/glw126. [DOI] [PMC free article] [PubMed] [Google Scholar]