Abstract
Patients with acute exacerbation of chronic obstructive pulmonary disease (AECOPD) are at increased risk of major adverse cardiovascular events (MACE) and death. Growth differentiation factor-15 (GDF-15), a marker of cellular stress and inflammation, and Syndecan-1, a marker of endothelial dysfunction, have been suggested as prognostic biomarkers in plasma for MACE. We aimed to assess their association with a combined outcome of MACE or all-cause mortality over a 5-year period. This sub-study was embedded within the randomized controlled trial CORTICOsteroid reduction in COPD (CORTICO-COP), which investigated the effects of eosinophil-guided corticosteroid therapy in patients hospitalized with acute exacerbation of chronic obstructive pulmonary disease (AECOPD). A total of 299 patients hospitalized with AECOPD were included in this analysis. Baseline plasma concentrations of growth differentiation factor 15 (GDF-15) and Syndecan-1 were measured and stored in a biobank for later analysis. The primary outcome was MACE or all-cause mortality, secondary outcomes included heart failure, re-AECOPD, and all-cause mortality. Hazard ratios (HRs) between low and high biomarker levels were adjusted for age, smoking, sex, GOLD class, and kidney insufficiency. The area under the receiver operating curve (AUC) was reported for each model after 6 months and two years respectively. Among the 299 hospitalized AECOPD patients included in this sub-study of the randomized controlled trial CORTICOsteroid reduction in COPD (CORTICO-COP), higher baseline concentrations of GDF-15 were associated with an increased risk of the combined outcome of MACE or all-cause mortality (hazard ratio [HR] 1.68, 95% confidence interval [CI] 1.16–2.44, p = 0.007), as well as all-cause mortality alone (HR 1.5, 95% CI 1.07–2.19, p = 0.02). GDF-15 showed moderate discriminative ability for survival, with an AUC of 64% at 6 months and 60% at 2 years. No significant associations were observed between GDF-15 and heart failure or hospital re-admission due to respiratory disease. Syndecan-1 concentrations were not associated with the combined endpoint or any of the secondary outcomes. GDF-15 may identify AECOPD patients at risk of MACE and all-cause mortality. Syndecan-1 has no predictive value in AECOPD patients.
Supplementary Information
The online version contains supplementary material available at 10.1038/s41598-025-27988-6.
Keywords: Biomarker, COPD, AECOPD, Cardiovascular Risk, GDF-15, Syndecan-1
Subject terms: Biomarkers, Diseases
Introduction
Chronic obstructive pulmonary disease (COPD) is not only a major respiratory burden, but also a critical contributor to cardiovascular morbidity and mortality1–3. The risk is high during acute exacerbations of COPD (AECOPD), with a nearly four-fold increase in major adverse cardiovascular events (MACE) compared with periods without exacerbations4,5.
This risk is most pronounced within the first 30 days post-exacerbation but remains elevated throughout the following year6. These findings underscore the importance of integrated cardiopulmonary care strategies and highlight the need for developing prognostic tools to stratify cardiovascular risk and guide clinical decision-making.
Two potential biomarkers are of interest in this context: Growth differentiation factor-15 (GDF-15), a stress-responsive cytokine, and Syndecan-1, a marker of endothelial dysfunction. Syndecan-1 is a novel inflammatory marker and may be associated not only with lung function and AECOPD but also with the risk of mortality following MACE7,8. GDF-15 has been suggested as a prognostic marker for MACE and all-cause mortality9,10 and elevated levels have also been associated with increased risks of exacerbations and mortality in COPD11. This study aims to determine the association between plasma concentrations of GDF-15 and Syndecan-1 and the combined outcome of MACE or all-cause mortality over a 5-year period in patients hospitalized with AECOPD. We hypothesize that individuals with a higher concentration of GDF-15 and/or Syndecan-1 during AECOPD are at a greater risk of experiencing MACE or all-cause mortality compared with those with lower levels.
Material & methods
Study design and cohort
This study population was derived from the randomized controlled trial CORTICOsteroid reduction in COPD (CORTICO-COP) NCT02857842 (Registration date 2016–08-02), which investigated the effect of eosinophil-guided corticosteroid therapy on clinical outcomes in patients hospitalized with acute exacerbations of COPD (AECOPD). From this RCT, a prospective cohort of 318 patients admitted with AECOPD in the Capital Region of Denmark was identified. Of these, 299 patients who fulfilled the specific inclusion criteria were enrolled in the present , see Fig. 112.
Fig. 1.
Flow diagram of participant inclusion in the study.
Patients hospitalized with an AECOPD were eligible for inclusion if enrolled within 24 h of hospital admission, between August 2016 and September 2018. Patients were recruited at three respiratory departments in the Capital Region of Denmark: Copenhagen University Hospital Herlev and Gentofte, Copenhagen University Hospital – Bispebjerg and Frederiksberg, and Copenhagen University Hospital – Amager and Hvidovre. Plasma samples were collected at baseline (on the day after admission) and at 30-day follow-up from patients aged ≥ 40 years with severe or very severe COPD (GOLD stage E), see appendix Table A1. GOLD stage E refers to patients with a history of frequent exacerbations (≥ 2 moderate or ≥ 1 severe in the past year), consistent with the 2023 GOLD definition of frequent exacerbators. Of the 318 patients included, 299 had available baseline plasma samples for biomarker analysis. For this study, only participants with baseline measurements of GDF-15 and Syndecan-1 were included. The follow-up period was extended to five years to assess the occurrence of MACE. Biomarker concentrations were categorized into high and low groups. High concentrations were defined as values in the highest quartile (≥ 75th percentile), while low concentrations comprised values in the lower three quartiles (< 75th percentile). The 75th percentile was selected as the cut-off for defining high biomarker concentrations to support clinical risk stratification, with the aim of identifying patients at potentially elevated risk of major adverse cardiovascular events. This threshold was chosen a priori based on its use in prior studies, where similar percentile-based approaches have been applied to delineate clinically meaningful risk groups and enhance interpretability in translational settings13. This categorization was applied consistently across all analyses, and comparisons were made between high and low biomarker concentrations.
Outcome
The primary outcome was a composite of MACE (stroke, acute myocardial infarction, unstable angina pectoris) or all-cause mortality within 5 years from baseline.
Secondary outcomes included (i) hospitalization for heart failure (GDF-15) within 5 years, (ii) new diagnosis with/hospitalization caused by respiratory disease within 5 years (SPD), and (iii) all-cause mortality within 5 years (GDF-15).
All outcomes were identified using the Danish National Patient Registry (DNPR) which contains information on all hospital contacts (inpatient and outpatient) and diagnoses since 1977, coded in ICD-10, see appendix Table A2.
Statistical analyses
Descriptive analyses were used to summarize demographic and clinical characteristics. Categorial variables were reported as counts and percentages, while continuous variables were presented as mean and standard deviations (SD), if normally distributed, or as median and interquartile range (IQR) if not.
The primary analysis assessed the association between biomarker levels (high vs. low) and the risk of MACE or all-cause mortality within 5 years, applying a Cox proportional hazards model, adjusted for age, sex, smoking status (pack years), cumulative corticosteroid dose received during hospitalization before blood sample, GOLD class and kidney insufficiency. Since blood sampling occurred on day 2, we were unable to adjust for systemic corticosteroids during the entirety of the initial hospitalization. The proportional hazards assumptions were evaluated using Schoenfeld residuals. Results were reported as hazard ratios (HR) with 95% confidence intervals (CI) and were assessed for all covariates. Receiver Operating Characteristic (ROC) curve analysis was used to assess the discriminative ability of the biomarkers. The area under the ROC curve (AUC) was reported for each biomarker model, accounting for the covariates mentioned above, to quantify their sensitivity and specificity in predicting MACE at 6 months and at 2 years. Kaplan–Meier survival curves were rendered to compare survival free of MACE across high and low biomarker groups. The log-rank test was applied to assess statistical differences between groups. All statistical analyses and illustrations were performed using: “RStudio”, R version 4.3.3 (2024–02–29 ucrt) – “Angel Food Cake”. Copyright © 2024 The R Foundation for Statistical Computing. Platform: x86_64-w64-mingw32/× 64. The R-packages survival14,15, ggsurvfit16, and timeROC17 were used for respectively model fitting, visualization of the survival probabilities, and estimation of area under the curve. We considered a p-value ≤ 0.05 statistically significant.
Ethics approval and consent to participate
The sub-study was conducted in accordance with the Declaration of Helsinki and approved as a supplementary protocol to the CORTICO-COP RCT by the Ethics Committees of all participating sides (H-15012207), the Danish Medicines Agency (EudraCT no 2015–003441-26), and the Danish Data Protection Agency (HGH-2015–038 and I-Suite number 04014). It was monitored according to Good Clinical Practice (GCP) by the GCP unit of the Capital Region of Denmark. Informed consent was obtained from all subjects involved in the study. No financial incentive was provided to the investigators or participants.
Results
Among the 299 AECOPD patients with available biomarker data, participants were categorized based on GDF-15 and Syndecan-1 concentrations into high (upper quartile) and low (lower three quartiles) groups. For GDF-15, 222 patients were classified into the low and 74 into the high concentration groups, respectively. A histogram showing the distribution of the biomarker is available in the appendix (GDF-15 missing for three patients), See Figure A1. The high GDF-15 group was older (mean age 79.8 vs. 73.6 years) and had a slightly lower proportion of males (41% vs. 46%). Mean GDF-15 concentrations were 5203 pg/mL (SD 2392) in the high group and 1843 pg/mL (SD 700) in the low group. Cumulative OCS dose was evenly distributed across concentration groups, with approximately 20% receiving 37.5 mg or less, 42% receiving 100 mg, and 35% receiving 137.5 mg on the first day of admission. GOLD 2 and 4 were the most common severity grades in both groups, with similar distributions. Smoking history (pack-years), eosinophil count and comorbidities including type 2 diabetes, hypertension, and ischemic heart disease were comparable between groups. Hypertension was the most frequent comorbidity (~ 40%). Kidney insufficiency was more common in the high GDF-15 group (19% vs. 3%), See Table 1. For Syndecan-1, 200 patients had low concentrations, while 67 had high levels; 32 patients had missing Syndecan-1 measurement. The mean age was approximately 75 years in both groups, with a predominance of females. Mean (SD) Syndecan-1 concentrations were 1807 (738) µg/mL in the low group and 5206 (2280) µg/mL in the high group, See Table 1. In the low group, 41% had GOLD-class 4 and 47% had GOLD-class 2. In the high group, 54% were GOLD-class 4 and 37% were GOLD-class 2. Smoking pack-years were similar across groups (mean 46 vs. 45), and so were baseline eosinophil counts and distribution of intervention groups.
Table 1.
Baseline characteristics stratified by high and low biomarker concentrations.
| GDF-15 | Syndecan-1 | |||
|---|---|---|---|---|
| Low | High | Low | High | |
| n | 222 | 74 | 200 | 67 |
| Age, mean (SD) | 73.6 (8.4) | 79.8 (9.5) | 75 (8.9) | 76 (9.2) |
| Male, n (%) | 103 (46.4) | 30 (40.5) | 86 (43.0) | 31 (46.3) |
| Syndecan-1 µg/ml, mean (SD) | 2650 (2047.20) | 2688 (1745.06) | 1807 (738.6) | 5207 (2280.2) |
| GDF-15 pg/ml, mean (SD) | 1843 (699.8) | 5203 (2392.1) | 2619 (1875.6) | 3038 (2067.9) |
| OCS cumulative dose. n (%) | ||||
| 37.5 mg or less | 46 (20.7) | 19 (25.7) | 43 (21.5) | 16 (23.9) |
| 100 mg | 96 (43.2) | 31 (41.9) | 85 (42.5) | 29 (43.3) |
| 137.5 mg | 80 (36.0) | 24 (32.4) | 72 (36.0) | 22 (32.8) |
| Current smokers, n (%) | 77 (34.7) | 22 (29.7) | 66 (33.0) | 19 (28.4) |
| Pack years, mean (SD) | 47 (20.39) | 45 (16.60) | 46 (18.89) | 45 (21.21) |
| GOLD class, n (%) | ||||
| 2 | 26 (12.7) | 9 (14.3) | 23 (12.6) | 5 (8.5) |
| 3 | 85 (41.5) | 29 (46.0) | 85 (46.7) | 22 (37.3) |
| 4 | 94 (45.9) | 25 (39.7) | 74 (40.7) | 32 (54.2) |
| Eosinophil count, mean (SD) | 0.16 (0.23) | 0.17 (0.23) | 0.15 (0.22) | 0.18 (0.23) |
| Type 2 Diabetes, n (%) | 15 (6.8) | 22 (29.7) | 26 (13.0) | 8 (11.9) |
| AFib or AFL, n (%) | 37 (16.7) | 17 (23.0) | 33 (16.5) | 17 (25.4) |
| Hypertension, n (%) | 85 (38.3) | 30 (40.5) | 88 (44.0) | 17 (25.4) |
| Kidney insufficiency, n (%) | 7 (3.2) | 14 (18.9) | 14 (7.0) | 6 (9.0) |
| IHD, n (%) | 24 (10.8) | 10 (13.5) | 27 (13.5) | 5 (7.5) |
| Osteoporosis, n (%) | 42 (18.9) | 13 (17.6) | 38 (19.0) | 12 (17.9) |
| Heart Failure, n (%) | 17 (7.7) | 10 (13.5) | 21 (10.5) | 5 (7.5) |
GDF-15, growth differentiation factor-15; OCS, Oral corticosteroids cumulative dose during first day of admission, before GDF-15 measurement; GOLD, global initiative for chronic obstructive lung disease; AFib, atrial fibrillation; AFL, atrial flutter; IHD, ischemic heart disease.
Comorbidities such as kidney insufficiency, type 2 diabetes, and heart failure were evenly distributed. Atrial fibrillation or flutter was more common in the high group (25% vs. 16.5%). Hypertension and ischemic heart disease were more common in the low group (44% vs. 25% and 14% vs. 8%, respectively).
GDF-15 and risk of MACE or all-cause mortality
We found a significant increase in the risk of MACE or all-cause mortality in patients with GDF-15 concentrations in the high group compared with the low group, with a hazard ratio (HR) of 1.68 (95% CI: 1.16–2.44), p = 0.007, number of events = 158. Age was also associated with increased risk: Compared with patients under 69 years of age, those aged 69–74 had a HR of 1.94 (95% CI: 1.18–3.18), p = 0.009; those aged 75–81 had a HR of 2.93 (95% CI: 1.83–4.71), p < 0.001; and patients over 82 had nearly a four-fold increase in hazard, with a HR of 3.68 (95% CI: 2.09–6.46), p < 0.001, see Figure A2 in the appendix.
Higher COPD severity was also linked to higher risk: Patients in GOLD-class 4 had a HR of 2.8 (95% CI: 1.74–5.57), p < 0.001, compared with those in GOLD-class 2. Smoking status, OCS use, sex, and renal insufficiency showed no significant effect on risk (Table 2). The proportionality test did not indicate any model misspecifications. Using the timeROC package to assess the ability of the model to predict survival times resulted in an AUC value equal to 64% at 6 months after inclusion and 60% after 2 years, see Figure A3. ROC curves are included in the appendix.
Table 2.
Hazard ratios for MACE or all-cause mortality stratified by GDF-15 levels.
| Hazard Ratio | Confidence interval | p-value | |
|---|---|---|---|
| GDF-15 low group | Ref | Ref | Ref |
| GDF-15 high group | 1.68 | (1.16—2.44) | 0.007** |
| Female | Ref | Ref | Ref |
| Male | 1.05 | (0.75—1.46) | 0.77 |
| Age < 69 | Ref | Ref | Ref |
| Age 69–74 | 1.94 | (1.18—3.18) | 0.009** |
| Age 75–81 | 2.93 | (1.83—4.71) | < 0.001*** |
| Age > 82 | 3.68 | (2.09–6.46) | < 0.001*** |
| OCS dose 37.5 mg or less | Ref | ||
| 100 mg | 1.22 | (0.79–1.87) | 0.37 |
| 137.5 mg | 1.10 | (0.70—1.72) | 0.69 |
| Pack years < 30 | Ref | Ref | Ref |
| 30–44 | 0.98 | (0.62—1.55) | 0.94 |
| 45–56 | 1.19 | (0.75–1.89) | 0.46 |
| > 56 | 0.84 | (0.54—1.33) | 0.46 |
| GOLD class 2 | Ref | Ref | Ref |
| GOLD class 3 | 1.41 | (0.80—2.48) | 0.24 |
| GOLD class 4 | 2.80 | (1.57–4.98) | < 0.001*** |
| Kidney insufficiency | 1.75 | (0.97—3.15) | 0.06 |
GDF-15, growth differentiation factor-15; OCS, Oral corticosteroids mean daily dose during admission, before GDF-15 measurement; GOLD, global initiative for chronic obstructive lung disease. Upper quantile of GDF-15 refers to levels above or equal to 3209.4 pg/ml. The model was adjusted for age, sex, smoking status (pack years) and comorbidities. Follow-up time was 5 years.
Syndecan-1 and risk of MACE or all-cause mortality
No significant increase in the risk of MACE or all-cause mortality was observed among patients in the high group of Syndecan-1 concentrations compared with those in the low group (HR 1.10; 95% CI: 0.74–1.64; p = 0.63, number of events = 144). In contrast, age was strongly associated with increased risk. Compared with patients younger than 69 years, those aged 69–74 had a hazard ratio (HR) of 2.17 (95% CI: 1.30–3.76; p = 0.004), those aged 75–81 had an HR of 2.93 (95% CI: 1.75–4.91; p < 0.001), and those older than 82 had an HR of 4.45 (95% CI: 2.47–8.03; p < 0.001).
COPD severity also influenced risk: patients in GOLD stage 4 had a significantly increased risk compared with those in GOLD stage 2 (HR 2.58; 95% CI: 1.36–4.90; p = 0.004). In contrast, smoking status, sex, and renal insufficiency were not significantly associated with MACE or all-cause mortality, consistent with findings for GDF-15 (Table 3).
Table 3.
Hazard ratios for MACE or all-cause mortality stratified by Syndecan-1 concentrations.
| Hazard Ratio | Confidence interval | p-value | |
|---|---|---|---|
| Syndecan-1 low group | Ref |
Ref (0.74—1.64) |
Ref 0.63 |
| Syndecan-1 high group | 1.10 | ||
| Female | Ref | Ref | Ref |
| Male | 0.93 | (0.65—1.32) | 0.67 |
| Age < 69 | Ref | Ref | Ref |
| Age 69–74 | 2.17 | (1.28—3.68) | 0.004** |
| Age 75–81 | 2.93 | (1.75—4.91) | < 0.001*** |
| Age > 82 | 4.45 | (2.47—8.03) | < 0.001*** |
| OCS dose 37.5 mg or less | Ref | ||
| 100 mg | 1.17 | (0.74–1.83) | 0.51 |
| 137.5 mg | 1.05 | (0.66—1.68) | 0.83 |
| Pack years < 30 | Ref | Ref | Ref |
| 30–44 | 0.90 | (0.55—1.48) | 0.68 |
| 45–56 | 1.37 | (0.85—2.20) | 0.19 |
| > 56 | 0.97 | (0.61–1.57) | 0.91 |
| GOLD class 2 | Ref | Ref | Ref |
| GOLD class 3 | 1.37 | (0.74—2.56) | 0.32 |
| GOLD class 4 | 2.58 | (1.36–4.90) | 0.004** |
| Kidney insufficiency | 2.03 | (1.10—3.73) | 0.023* |
GOLD, global initiative for chronic obstructive lung disease; Oral corticosteroids mean daily dose during admission, before Syndecan-1 measurement; Upper quantile of Syndecan-1 refers to levels above or equal to 3008.81 µg/ml. The model was adjusted for age, sex, smoking status (pack years) and comorbidities. Follow-up time was 5 years.
Proportionality test did not suggest any model misspecifications. Using timeROC package to assess the model’s ability to predict survival times, resulted in an AUC value equal to 57% at 6 months after inclusion and 54% after 2 years. ROC curves are included in the appendix, see Figure A4.
The Kaplan–Meier plot illustrating the probability of survival free without MACE, stratified by GDF-15 levels, demonstrated a significant difference between low See Fig. 2 (log-rank test, p < 0.001). Patients with high GDF-15 concentrations had a higher incidence of MACE or mortality. The mean time to death was 1,103 days, corresponding to approximately 3 years (Figs. 1, 2).
Fig. 2.
Probability of survival without MACE, stratified by GDF-15 concentrations. Kaplan–Meier curve showing significantly Lower probability of survival without mace probability of survival without mace in patients with high (≥ 3,209.4 pg/mL) vs. low (< 3,209.4 pg/mL) GDF-15 concentrations. Mean time to event: ~ 1,103 days. Log-rank test: p < 0.001. Follow-up was 5 years.
The Kaplan–Meier plot illustrating of the probability of survival without MACE, stratified by Syndecan-1 concentrations (Fig. 3), showed no difference in all-cause mortality between the low and high Syndecan-1 groups (log-rank test, p = 0.6). To address concerns about the follow-up time we additionally preformed a 2-year follow-up analysis of the primary outcome which did not alter the results (See appendix Tables A3-A4 and Figure A5).
Fig. 3.
Probability of Survival without MACE, stratified by Syndecan-1 status. Kaplan–Meier curve showing no significant difference in probability of survival without MACE between patients with low (< 3008.81 µg/mL) and high (≥ 3008.81 µg/mL) Syndecan-1 concentrations. Mean time to event: ~ 1,100 days. Log-rank test: p = 0.60. Follow-up was 5 years.
Association between GDF-15 status and risk of all‑cause mortality, heart failure, and respiratory admission
The high group of GDF‑15 concentrations had a significantly higher risk of all‑cause mortality (HR 1.53, 95% CI 1.07–2.19; p = 0.02), but not heart failure (HR 1.80, 95% CI 0.88–3.71; p = 0.11) or respiratory admission (HR 0.71, 95% CI 0.49–1.03; p = 0.07) (Table 4).
Table 4.
Hazard ratios for all-cause mortality, heart-failure, and respiratory admissions according to level of GDF-15.
| Follow-up time was 5 years | High vs. low GDF-15 group | ||
|---|---|---|---|
| Hazard ratio | Confidence interval | p-value | |
| All-cause mortality | 1.53 | (1.07–2.19) | 0.02 |
| Heart failure | 1.80 | (0.88–3.71) | 0.11 |
| Re-admission with respiratory disease | 0.71 | (0.49–1.03) | 0.07 |
Model was adjusted for age group, smoking status, sex, pack years, GOLD class and kidney insufficiency. Death was handled as competing risk. *refers to values of p < 0.05.
In contrast, Syndecan‑1 concentrations showed no associations with all‑cause mortality (p = 0.15), heart failure (p = 0.23), or respiratory admission (p = 0.74) (Table 5).
Table 5.
Hazard ratios for all-cause mortality, heart-failure, and respiratory admissions according to Syndecan-1 levels. Follow-up time was 5 years.
| Follow-up time was 5 years | High vs. low Syndecan-1 group | ||
|---|---|---|---|
| Hazard ratio | Confidence interval | p-value | |
| All-cause mortality | 1.33 | (0.91—1.95) | 0.15 |
| Heart-failure | 0.61 | (0.27–1.37) | 0.23 |
| Re-admission with respiratory disease | 1.06 | (0.76–1.49) | 0.74 |
The model was adjusted for age group, smoking status, sex, pack years, GOLD class and kidney insufficiency. Death was handled as competing risk.
Discussion
Among patients hospitalized with AECOPD, those with high concentrations of GDF-15 had an increased risk of MACE or all-cause mortality. A similar association was not found for Syndecan-1. Neither heart failure or future AECOPD were associated with GDF-15 or Syndecan-1.
Our findings are consistent with previous studies suggesting that GDF-15 may hold prognostic value in patients with AECOPD, most likely reflecting its role as a stress-induced cytokine closely linked to cardiovascular pathology11,18. The lack of association between GDF-15 and future events of AECOPD may indicate that this biomarker is more specific to systemic or cardiovascular outcomes rather than pulmonary-related exacerbations. Indeed, the observed associations with MACE and all-cause mortality may not be unique to AECOPD, but rather reflect the broader function of GDF-15 as a marker of cardiovascular risk. GDF-15 is a stress-responsive cytokine belonging to the transforming growth factor-β (TGF-β) superfamily, upregulated by oxidative stress, inflammation, hypoxia, and mitochondrial dysfunction. In cardiovascular pathology, GDF-15 is secreted in response to myocardial stretch and ischemic injury and has been shown to modulate inflammatory signalling, apoptosis, and tissue remodelling. Elevated GDF-15 concentrations are consistently associated with worse outcomes in patients with heart failure, myocardial infarction, and atrial fibrillation, reflecting both disease severity and systemic stress responses. These mechanisms support our finding that GDF-15 in patients with AECOPD primarily captures systemic cardiovascular risk rather than pulmonary-specific outcomes19. This interpretation is supported by its established links with age, chronic kidney disease, and other comorbidities. Although our models adjusted for these factors and the associations with MACE and mortality persisted, residual confounding cannot be excluded. Taken together, these findings suggest that GDF-15 may serve as a systemic biomarker of cardiovascular stress with potential utility for general risk stratification in patients with AECOPD, while its role in predicting COPD-specific outcomes such as exacerbations remains uncertain given the currently limited and inconclusive evidence.
Although some studies have proposed its potential utility in predicting AECOPD as well, results remain limited in size and significance20.Interestingly, Syndecan-1 concentrations did not reveal associations with any of the outcomes in our cohort. Although some studies have reported modest predictive value of Syndecan-1 in cardiovascular disease8 as well as in COPD patients, where it has been shown to predict lung function and systemic inflammation, these findings have led to suggestions that Syndecan-1 could be a useful prognostic tool for future AECOPD (7). However, our data do not support this potential association.
A reason for our lack of significant findings with Syndecan, in comparison to GDF-15 could perhaps be due to the biological stability of Syndecan21. It is possible that Syndecan-1, a marker of endothelial shedding, is more dynamic and requires repeated or time-sensitive measurements, not only at baseline, to capture its prognostic relevance or that it is more specific to conditions causing great endothelial damage such as sepsis22.
OCS use during the initial hospitalization occurs according to the study intervention and could potentially affect the risk of MACE after discharge. However, administration of corticosteroids post randomization occurs after baseline, and thus adjusting for this could potentially lead to immortal time bias, and over adjustment when inferring the effect of the biomarkers. Since administration of OCS according to intervention occurs after the blood samples, it acts as a mediator between the biomarkers and events of interest rather than as a confounder and thus might delude the effect of the biomarkers if included in the model. Although it certainly would be interesting to investigate the association between OCS and MACE, this lies beyond the scope of the current paper.
Future studies should explore multimarker strategies, combining GDF-15 with established cardiac or pulmonary biomarkers such as troponin23 or N-terminal pro b-type natriuretic peptide (NT-proBNP)24–26 to improve predictive performance and enable tailored risk stratification.
Strength and limitations
A strength of this study is its origin as a substudy of a multicentre RCT, utilizing standardized and timely biomarker collection27. Recruitment from three centralized hospitals and the use of standardized protocols for data and sample handling reduce variability and enhance internal validity. While most baseline characteristics were similar across biomarker strata, some differences, such as age and CKD, were present. These were accounted for in the multivariable analyses, minimizing potential confounding. However, the study has limitations.
First, not all patients had available and measured biomarkers, resulting in a somewhat smaller sample size which may have been insufficient to power the statistical analysis of Syndecan-1 and risk of MACE. Secondly, possible biases have been adjusted for in main analyses. However, we cannot rule out the fact, that residual confounding due to unadjusted variables may have impacted the interpretation of our results. Thirdly, we acknowledge that pre-existing cardiovascular conditions such as prior MACE, respiratory failure, and pulmonary hypertension represent important prognostic factors that may influence outcomes. These variables were, however, not systematically collected in the CORTICO-COP cohort and could therefore not be incorporated into our multivariable models. While this constitutes a limitation, it is reassuring that baseline characteristics, including history of cardiovascular disease and other major comorbidities, were generally well balanced across strata of GDF-15 and Syndecan-1. This reduces, but does not entirely eliminate, the risk of residual confounding and should be considered when interpreting our findings. Fourth, a limitation of our study is that GDF-15 was measured only at baseline during hospitalization for AECOPD. In COPD patients with multiple comorbidities, clinical status and risk factors may change over time, potentially affecting the relevance of baseline biomarker levels for events occurring several years later. Nevertheless, the majority of adverse events in our cohort occurred within the first 1–2 years, indicating that baseline GDF-15 provides a meaningful early risk signal. Future studies with longitudinal biomarker measurements may further refine risk prediction and provide insights into temporal changes in prognosis. Furthermore, the relatively small cohort size and lack of both internal and external validation represent additional limitations that may affect the generalizability and robustness of our findings. Lastly, the discriminative ability of GDF-15 was only modest, but it may still provide incremental prognostic value when combined with established clinical variables or other biomarkers such as troponin or NT-proBNP.
Conclusion
In this prospective cohort study, higher plasma concentrations of GDF-15 were associated with an increased risk of MACE or all-cause mortality, and all-cause mortality among patients hospitalized with AECOPD, suggesting a value as a prognostic biomarker in this population. No association was found between Syndecan-1 concentrations and MACE, all-cause mortality, heart-failure or future AECOPD. These findings highlight GDF-15 as a promising marker in plasma for cardiovascular risk stratification in AECOPD patients. Further work is needed to define clinically meaningful thresholds and to explore how GDF-15 might integrate into existing clinical workflows, either as a triage tool during hospitalization or to guide long-term cardiovascular monitoring in AECOPD survivors.
Supplementary Information
Acknowledgements
We wish to thank patients, relatives, clinical staff and research staff at all trial sites. Without their support the CORTICO-COP trial would never have been completed.
Author contributions
Conceptualization, P.S. and J.-U.S.J.; methodology, P.S., A.K.V. and J.-U.S.J.; software, P.S. and J.-U.S.J.; validation, P.S. and J.-U.S.J.; formal analysis, A.K.V and P.S.; investigation, P.S.; resources, P.S. and J.-U.S.J.; data curation, P.S. and J.-U.S.J.; writing—original draft preparation, D.A.A P.S., A.K.V and J.-U.S.J writing—review and editing, D.A.A., A.K.V., P.S. and J.-U.S.J.; visualization, A.K.V.; supervision, J.-U.S.J.; project administration, P.S. and J.-U.S.J.; funding acquisition, P.S. and J.-U.S.J.; All authors have read and agreed to the published version of the manuscript.
Funding
The CORTICO-COP trial was funded by The Danish Regions Medical Fund and the Danish Council for Independent Research. The research salary for P.S. was provided by Herlev-Gentofte University Hospital.
Data availability
It is the opinion of Copenhagen Respiratory Research (COP:RESP) that knowledge sharing leads to more and better scientific outcomes. Requests for trial information can be submitted to the Project Management team and investigators. Such requests, including study protocol with clear hypotheses should be sent to the principal investigator, and the CORTICO-COP steering committee will review such a request. If the hypothesis does comply with the informed consent supplied by the participants, and the hypothesis is judged to be valid, a data transfer agreement will be prepared, after which the data will be transferred. The study protocol and statistical analysis plan for the original study is available at www.coptrin.dk. Informed consent forms will not be available according to Danish legislation.
Declarations
Competing interests
T.B-S. reports receiving research grants from Sanofi Pasteur and GE Healthcare, is a Steering Committee member of the Amgen financed GALACTIC-HF trial, on advisory boards for Sanofi Pasteur and Amgen, and speaker honorariums from Novartis and Sanofi Pasteur. J.V. reports personal fees from GlaxoSmithKline, Chiesi Pharmaceuticals, Boehringer-Ingelheim, Novartis, and AstraZeneca; and is supported by the National Institute of Health Research Manchester Biomedical Research Centre (Manchester, UK). The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. All other authors declare no conflicts of interest.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Hurst, J.R., Gale, C.P., Hurst, J.R., Bhutani M, Bourbeau J, Han ML, et al. MACE in COPD: addressing cardiopulmonary risk. Lancet Respir. Med.12(5):345–8. https://www.thelancet.com/action/showFullText?pii=S2213260024000389. (2024).
- 2.Mkorombindo, T., Dransfield, M.T. COPD exacerbations and MACE: More than a chance encounter?. Respirology. 24 (12):1129–30. https://pubmed.ncbi.nlm.nih.gov/31432595/. (2019).
- 3.Sá-Sousa, A., Rodrigues, C., Jácome, C., Cardoso, J., Fortuna, I., Guimarães, M. et al. Cardiovascular risk in patients with chronic obstructive pulmonary disease: a systematic Review. J. Clin. Med. 13 (17). https://pubmed.ncbi.nlm.nih.gov/39274386/. (2024).
- 4.Sin, D.D., Anthonisen, N.R., Soriano, J.B., Agusti, A.G. Mortality in COPD: Role of comorbidities. Eur. Respir. J.28(6):1245–57. https://pubmed.ncbi.nlm.nih.gov/17138679/. (2006).
- 5.Reilev, M. et al. Increased risk of major adverse cardiac events following the onset of acute exacerbations of COPD. Respirology24(12), 1183–1190. 10.1111/resp.13620 (2019). [DOI] [PubMed] [Google Scholar]
- 6.Pirera, E., Di Raimondo, D., D’Anna, L., Tuttolomondo, A. Risk trajectory of cardiovascular events after an exacerbation of chronic obstructive pulmonary disease: A systematic review and meta-analysis. Eur. J. Intern. Med. (2025).
- 7.Li, D., Wu, Y., Guo, S., Qin, J., Feng, M., An, Y. et al. Circulating syndecan-1 as a novel biomarker relates to lung function, systemic inflammation, and exacerbation in COPD. Int. J. Chron. Obstruct Pulmon Dis.14 :1933–41. https://pubmed.ncbi.nlm.nih.gov/31695352/. (2019).
- 8.Wernly, B., Fuernau, G., Masyuk, M., Muessig, J.M., Pfeiler, S., Bruno R.R. et al. Syndecan-1 predicts outcome in patients with ST-segment elevation infarction independent from infarct-related myocardial injury. Sci. Rep. 9(1): 1–7. https://www.nature.com/articles/s41598-019-54937-x. (2019).
- 9.Kato, E.T., Morrow, D.A., Guo, J., Berg, D.D., Blazing, M.A., Bohula, E.A. et al. Growth differentiation factor 15 and cardiovascular risk: individual patient meta-analysis. Eur. Heart J.44 (4):293–300. https://pubmed.ncbi.nlm.nih.gov/36303404/. (2023).
- 10.Lyu, L. et al. Growth differentiation factor-15 predicts all-cause death and major adverse cardiovascular events in patients with coronary heart disease: a prospective cohort study. J. Thromb. Thrombolysis.57(7), 1109–1121 (2024). [DOI] [PubMed] [Google Scholar]
- 11.Husebø, G.R., Grønseth, R., Lerner, L., Gyuris, J., Hardie, J.A., Bakke, P.S. et al. Growth differentiation factor-15 is a predictor of important disease outcomes in patients with COPD. Eur. Respir. J.49(3). https://pubmed.ncbi.nlm.nih.gov/28298399/. (2017).
- 12.Sivapalan, P., Lapperre, T.S., Janner, J., Laub, R.R., Moberg, M., Bech, C.S. et al. Eosinophil-guided corticosteroid therapy in patients admitted to hospital with COPD exacerbation (CORTICO-COP): a multicentre, randomised, controlled, open-label, non-inferiority trial. Lancet Respir. Med.7(8):699–709. https://www.thelancet.com/action/showFullText?pii=S2213260019301766. (2019).
- 13.Brouwers, F. P. et al. Clinical risk stratification optimizes value of biomarkers to predict new-onset heart failure in a community-based cohort. Circ. Heart Fail.7(5), 723–731 (2014). [DOI] [PubMed] [Google Scholar]
- 14.Therneau, T. M. & Grambsch, P. M. Modeling Survival Data: Extending the Cox Model (Springer New York, 2000). [Google Scholar]
- 15.Therneau, T.M. Survival analysis [R package survival version 3.8–3]. CRAN: contributed packages. https://CRAN.R-project.org/package=survival. (2024).
- 16.Sjoberg, D.D., Baillie, M., Fruechtenicht, C., Haesendonckx, S., Treis, T. Flexible time-to-event figures [R package ggsurvfit version 1.1.0]. CRAN: contributed packages. https://CRAN.R-project.org/package=ggsurvfit. (2024).
- 17.timeROC: time-dependent ROC curve and AUC for censored survival data. CRAN: contributed packages. https://CRAN.R-project.org/package=timeROC. (2012).
- 18.Verhamme, F.M., Freeman, C.M., Brusselle, G.G., Bracke, K.R., Curtis J.L. GDF-15 in pulmonary and critical care medicine. Am. J. Respir. Cell Mol. Biol. 60(6):621–8. https://pmc.ncbi.nlm.nih.gov/articles/PMC6543753/. (2019).
- 19.Schwarz, A., Kinscherf, R., Bonaterra, G.A. Role of the stress- and inflammation-induced cytokine GDF-15 in cardiovascular diseases: from basic research to clinical relevance. Rev. Cardiovasc. Med. 24, 2023.
- 20.Lv, Z., Liang, G., Cheng, M. Predictive value of GDF-15 and sST2 for pulmonary hypertension in acute exacerbation of chronic obstructive pulmonary disease. Int. J. Chron. Obstruct Pulmon. Dis.18:2431. https://pmc.ncbi.nlm.nih.gov/articles/PMC10637188/. (2023).
- 21.Teng, Y.H.F., Aquino, R.S., Park, P.W. Molecular functions of syndecan-1 in disease. Matrix Biol. 31(1):3. https://pmc.ncbi.nlm.nih.gov/articles/PMC3568394/. (2011).
- 22.Sun, T., Wang, Y., Wu, X., Cai, Y., Zhai, T., Zhan, Q. prognostic value of syndecan-1 in the prediction of sepsis-related complications and mortality: a meta-analysis. Front. Public Health.10:870065. https://pmc.ncbi.nlm.nih.gov/articles/PMC9035829/. (2022).
- 23.Kallis, C., Kaura, A., Samuel, N.A., Mulla, A., Glampson, B., O’gallagher, K. et al. The relationship between cardiac troponin in people hospitalised for exacerbation of COPD and major adverse cardiac events (MACE) and COPD readmissions. Int. J. Chron. Obstruct Pulmon. Dis.18: 2405–16. https://pubmed.ncbi.nlm.nih.gov/37955026/. (2023).
- 24.Li, N., Chen, R., Li, J., Zhao, X., Wang, Y., Zhou, J. et al. Prognostic significance of serial N-terminal pro-B-type natriuretic peptide levels in patients with acute myocardial infarction: A prospective study. Am Heart J.262:90–9. https://pubmed.ncbi.nlm.nih.gov/37116605/. (2023).
- 25.Zhang, L.J., Li, N., Li, Y., Zeng, X.T., Liu, M.Y. Cardiac biomarkers predicting MACE in patients undergoing noncardiac surgery: a meta-analysis. Front Physiol. 9:1923. https://pmc.ncbi.nlm.nih.gov/articles/PMC6346145/. (2019).
- 26.Davies, E., Mitford.,Buckley, B.J.R.,Austin, P.,Lip, G.Y.H.,Rao, A. et al. Aalborg universitet cardiac biomarkers predict major adverse cardiac events (MACE) in incident haemodialysis patients: results from a global federated database cardiac biomarkers predict major adverse cardiac events (MACE) in incident haemodialysis patients: results from a global federated database. 13:367. https://creativecommons.org/licenses/by/4.0/. (2025).
- 27.Sivapalan, P. et al. A multi-center randomized, controlled, open-label trial evaluating the effects of eosinophil-guided corticosteroid-sparing therapy in hospitalised patients with COPD exacerbations - The CORTICO steroid reduction in COPD (CORTICO-COP) study protocol. BMC Pulm. Med.17(1), 1–7. 10.1186/s12890-017-0458-7 (2017). [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
Data Availability Statement
It is the opinion of Copenhagen Respiratory Research (COP:RESP) that knowledge sharing leads to more and better scientific outcomes. Requests for trial information can be submitted to the Project Management team and investigators. Such requests, including study protocol with clear hypotheses should be sent to the principal investigator, and the CORTICO-COP steering committee will review such a request. If the hypothesis does comply with the informed consent supplied by the participants, and the hypothesis is judged to be valid, a data transfer agreement will be prepared, after which the data will be transferred. The study protocol and statistical analysis plan for the original study is available at www.coptrin.dk. Informed consent forms will not be available according to Danish legislation.