Inflammatory phenotype drives different immunosuppressive response in COPD exacerbations

Abstract

Background

COPD is heterogeneous, with immunosuppression contributing to immune dysfunction. We hypothesized that the immunosuppressive response varies by inflammatory phenotype during both stable disease (SCOPD) and acute exacerbations (AECOPD).

Methods

COPD patients were monitored during AECOPD hospitalization and subsequent clinical stability. Serum key pro- and anti-inflammatory molecules were measured. Circulating Th17 and regulatory T (Treg) cells, early-myeloid-derived suppressor cells (e-MDSC), monocytic-MDSC (M-MDSC), and plasmacytoid dendritic cells (pDCs) were analyzed by flow cytometry. Sputum was tested for viruses and bacteria. Eosinophilic inflammation was defined by a blood eosinophil count of ≥ 2%.

Results

The study included 78 COPD patients and 17 controls. 49% of patients consistently had a non-eosinophilic (Non-Eos) phenotype, and 18% an eosinophilic (Eos) phenotype across both AECOPD and SCOPD. Interestingly, 28% of patients exhibited Non-Eos-AECOPD followed by eosinophilia in SCOPD, while 5% showed the opposite. Immunosuppressive cell frequency increased in both Eos-AECOPD and Non-Eos-AECOPD. While Tregs increases were comparable, the M-MDSC increase in Non-Eos-AECOPD was 1.6 times greater than in Eos-AECOPD. During the stable phase, the Eos phenotype associated with more Tregs and fewer M-MDSCs. Cytokine levels rose significantly during Non-Eos-AECOPD but not in Eos-AECOPD. Critically, within the Non-Eos-AECOPD group, those who later developed eosinophilia in SCOPD showed the most pronounced increases in M-MDSC frequency and cytokine levels.

Conclusions

Immunosuppressive cell distribution in SCOPD and triggered immunosuppression pathways during AECOPD are dependent on the inflammatory profile. Understanding the COPD inflammatory phenotype might help predict immune responses and lead to more targeted treatments.

Trial registration

Not applicable.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12890-025-04020-1.

Introduction

Chronic obstructive pulmonary disease (COPD) is a complex inflammatory lung condition driven by both innate and adaptive immune responses [1, 2]. While neutrophilic inflammation and pro-inflammatory mediators are commonly observed, a significant subset (10–40%) of stable COPD (SCOPD) patients exhibit eosinophilic inflammation [3–5]. This eosinophilic endotype is notable for its association with increased exacerbation risk [6] and improved response to inhaled corticosteroids [7].

Acute inflammation is a self-limited physiological process that requires a proinflammatory (first) and an immunosuppressive (later) response that limits tissue damage restoring tissue homeostasis [8]. If it is not satisfactorily resolved, then it turns into chronic inflammation, characterized by an abnormal and prolonged immune response [9].

Immunosuppression is implicated in the immune dysfunction observed in COPD. Consequently, immunosuppression in COPD has garnered significant interest in recent years. However, studies have reported conflicting results [10–13], particularly concerning acute exacerbations of COPD (AECOPD).

Immunosuppressive cells, notably regulatory T (Treg) cells, have been extensively studied in both SCOPD and AECOPD [10–13]. Tregs play a crucial role in maintaining immune homeostasis by preventing excessive inflammation and autoimmunity. In the context of AECOPD, the majority of studies indicate a decreased proportion of Tregs in patients compared to healthy controls [14, 15], a finding also observed in SCOPD [16]. Nevertheless, some studies have reported contrasting results [15, 17]. Therefore, the distribution of Tregs in COPD, particularly during AECOPD, remains a subject of considerable debate.

Myeloid-derived suppressor cells (MDSCs) are innate immune cells known for their capacity to modulate T-cell responses and suppress inflammation [18, 19]. MDSCs secrete interleukin-10 (IL-10) and transforming growth factor (TGF)-β1, recognized as potent immunosuppressive cytokines [19]. Smoking has been shown to upregulate and activate circulating MDSCs in both current smokers with normal lung function and patients with COPD [20]. Research in COPD has predominantly focused on granulocytic-MDSC and early-myeloid precursors (e-MDSC) rather than on monocytic-MDSC (M-MDSC), despite evidence that chronic infections can stimulate the production of M-MDSC [21]. Both Treg cells and M-MDSC, as immunosuppressive cell populations, can inhibit the maturation of dendritic cells (DC), the primary cells responsible for initiating immune responses. This inhibition leads to the induction of a tolerogenic phenotype in DC, thereby contributing to a state of immunosuppression [22]. However, findings regarding these mechanisms in COPD remain inconclusive [23, 24].

We hypothesized that the induction of immunosuppressive cells during AECOPD must be conditioned by the inflammatory phenotype (eosinophilic or non-eosinophilic). To test this hypothesis, we analyzed various lymphoid and myeloid cell populations, and a panel of molecules involved in the inflammatory and anti-inflammatory responses during AECOPD and at subsequent clinical stability, stratified by inflammatory phenotype.

Methods

Study design

This is a prospective and observational study. Patients were included in the study at the time of their hospitalization due to an AECOPD. Patients were followed up for at least 90 days until their recovery, until they reached stability.

Subjects

Patients older than 40 years of age of both sexes, current and former smokers of at least 10 pack-years, with post-bronchodilator FEV₁/FVC < 0.7 who were admitted due to respiratory worsening of any etiology were included in the study. Subjects with autoimmune disorders, malignancy, and lung disease other than COPD were excluded. None of the patients included in the study were taking systemic corticosteroids, immunosuppressants, immunomodulators, or biologics. A patient was considered stable in the absence of a worsening of respiratory symptoms or a new episode of exacerbation in the 4 weeks prior to the stable visit. Additionally, control subjects including smokers and ex-smokers (CS), with normal lung function, and free of any respiratory disease were recruited for this study. The study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of the Arnau de Vilanova Hospital, Valencia, Spain. All participants gave their informed consent.

Stratification of patients

Inflammatory phenotype of patients during AECOPD and later in SCOPD was used to categorize COPD patients. To this end, a blood eosinophil threshold of ≥ 2% was used [22, 25]. Thus, eosinophilic (Eos) inflammation was defined as one in which the blood eosinophil count was ≥ 2%, while a count of < 2% identified non-eosinophilic (Non-Eos) inflammation.

Sample collections and determinations

After admission, a nasopharyngeal swab was obtained from patients, and the presence of virus was analyzed with the FilmArray Respiratory Panel a multiplex PCR system (bioMérieux, Madrid, Spain). Whenever possible, a spontaneous sputum sample was collected, and the presence of bacteria was performed by classical culture techniques.

Blood was collected upon admission to the hospital, and later during stability. Peripheral blood nucleated cells (MNC) were isolated by Lymphoprep density gradient centrifugation and then stained to analyze Tregs, Th17 cells, M-MDSC, e-MDSC and pDC by flow cytometry (Additional File 1).

Serum samples kept at −80 °C until processed were used to determine cytokine concentrations. IL-6 and IL-10 were determined by ProQuantum™ Human IL-6 Immunoassay Kit, and ProQuantum™ Human IL-10 Immunoassay Kit, respectively (Invitrogen, Thermo Fisher Scientific, bender MedSystems GmbH, Vienna, Austria). IL-17A was quantified by Milliplex MAP human Th17 from Millipore (Merck KGaA, Darmstadt, Germany), using Luminex xMAP Technology and Luminex 200 system from Merck Millipore. Free and total TGF-β1 were quantified using the Legend Max ELISA Kits (Biolegend, San Diego, CA). C reactive protein (CRP) was determined by Immunoturbidimetry (Tina-quant^® C- reactive protein, Roche Diagnostics, Barcelona, Spain). AECOPD sampling was taken before patients received systemic corticosteroids or antibiotics. Stability sampling was performed without interrupting maintained pharmacological treatment, but without systemic glucocorticoids or antibiotics for at least 4 weeks prior the sampling.

Statistical analysis

Results were expressed as mean ± standard deviation. COPD subjects were stratified according to blood eosinophil count. Paired comparisons between two groups were analyzed using the paired Student’s t-test or the Wilcoxon signed-rank test. For multiple comparisons, ANOVA followed by Dunnett’s post hoc test, or Kruskal-Wallis test followed by Dunn test with Bonferroni adjustment were used. The Chi-square test was used to assess the association between categorical variables.

Correlations were performed with the Spearman rank correlation coefficient test. Statistical analyses were performed using GraphPad Prism v5.0 (San Diego, CA, USA). The differences were considered statistically significant if p < 0.05.

Results

Inflammatory phenotype

The study included 78 COPD patients and 17 controls (CS). Among the COPD patients experiencing an AECOPD, 18 (23%) presented with an eosinophilic phenotype (Eos-AECOPD), while the remaining 60 (77%) exhibited a non-eosinophilic phenotype (Non-Eos-AECOPD). Notably, no significant clinical differences or variations in sex, age, smoking history, or maintenance therapy were observed between the Eos-AECOPD and Non-Eos-AECOPD groups. However, the hospital stay was significantly shorter in the eosinophilic exacerbation group (Table 1).

Table 1.

Characteristics of COPD patients and control smokers used in the study

Characteristics	Eos- AECOPD (n = 18)	Non-Eos- AECOPD (n = 60)	CS (n = 17)	Eos- AECOPD vs. Non-Eos-AECOPD p- value
Sex:
- Male, n (%)	16 (89)	49 (82)	11(65)	ns
- Female, n (%)	2 (11)	11 (18)	6 (35)
Age, years	68 ± 11	71 ± 8	61 ± 13	ns
Pack-years of smoking	56 ± 28	56 ± 28	43 ± 24	ns
Smoking history:
-Current smokers, n (%)	6 (33)	21 (35)	9 (53)	ns
-Ex-smokers, n (%)	12 (67)	39 (65)	6 (35)	ns
-Non-smokers, n (%)	0 (0)	0 (0)	2 (12)	ns
FEV1 (L)	1.58 ± 0.69	1.22 ± 0.54	2.93 ± 0.67	ns
FEV1 (% predicted)	52 ± 16	48 ± 17	95 ± 15	ns
FEV1/FVC (% predicted)	52 ± 12	49 ± 12	79 ± 7	ns
GOLD I, n (%)	1 (5.6)	1 (1.67)	NA	ns
GOLD II, n (%)	6 (33.3)	24 (40.0)	NA	ns
GOLD III, n (%)	10 (55.6)	27 (45.0)	NA	ns
GOLD IV, n (%)	1 (5.6)	8 (13.3)	NA	ns
Mean hospital stay (days)	4.4 ± 1.9	6.7 ± 4.1	NA	0.03
NIV use, n (%)	6 (33.3)	0 (0)	NA	ns
ICU admission	0 (0)	0 (0)	NA	NA
90 day readmission (%)	17	18	NA	ns
Maintenance treatment:
- LAMA monotherapy, n (%)	1 (5.6)	1 (1.7)	NA	ns
- LABA + LAMA, n (%)	6 (33.3)	15 (25)	NA	ns
- LABA + ICS, n (%)	1 (5.6)	2 (3.3)	NA	ns
- LAMA + LABA + ICS	8 (44.4)	38 (63.3)	NA	ns
- ICS (any combination), n (%)	9 (50)	41 (68.3)	NA	ns

Data are presented as mean ± SD. AECOPD were classified according to their inflammatory phenotype in Eos-AECOPD and Non-Eos-AECOPD. Lung function parameters and GOLD stage correspond to basal conditions before the exacerbation episode

GOLD Global Initiative for Chronic Obstructive Lung Disease, CS Control smokers, FEV₁ Forced expiratory volume in 1 s, FVC forced vital capacity, NIV Non-invasive ventilation, ICU Intensive care unit, ICS Inhaled corticosteroid, LAMA Long-acting muscarinic antagonist, LABA Long-acting β2-agonist, NA Not applicable, ns Not significant

Figure 1 illustrates the longitudinal changes in inflammatory phenotype. Upon recovery and stabilization, 46% of patients presented with an eosinophilic phenotype (Eos-SCOPD), and 54% with a non-eosinophilic phenotype (Non-Eos-SCOPD). Notably, a high proportion of patients maintained their initial exacerbation phenotype during stability: 78% of those with Eos-AECOPD and 63% of those with Non-Eos-AECOPD. The remaining patients experienced a shift in their inflammatory profile.

Fig. 1.

Flowchart of COPD patients

Compared to the stable state, AECOPD were characterized by a significant increase in leukocyte count and neutrophil percentage, alongside a decrease in lymphocyte and eosinophil percentages. When patients were stratified by exacerbation phenotype, Eos-AECOPD did not induce significant changes in blood cell composition. In contrast, the Non-Eos-AECOPD group, exhibited changes similar to the overall AECOPD group, including a significant decrease in the percentage of monocytes (Additional File 2).

Circulating immunosuppressive treg cells and inflammatory Th17 cells increase during AECOPD

During AECOPD, the proportion of immunosuppressive Treg cells (CD45⁺CD3⁺CD4⁺CD25^+/highCD127^-/low; Addtional File 3) was significantly higher (7.91 ± 3.80% of CD4⁺ T cells) compared to SCOPD (6.79 ± 2.35%, p = 0.05) (Fig. 2A). Stratifying the patient cohort by exacerbation phenotype revealed a significant decline in the proportion of circulating Treg cells in both groups upon reaching the stability (Fig. 2B, Additional File 4). Importantly, Treg levels were consistently elevated in patients with Eos-AECOPD compared to those with Non-Eos-AECOPD. This difference was observed during both the exacerbation phase (9.02 ± 4.62%, and 7.58 ± 3.50%, respectively) and the subsequent stability phase (7.74 ± 2.42% and 6.50 ± 2.27%, respectively (Fig. 2B). Despite these overall differences, a distinct pattern emerged when grouping patients based on their phenotype across both phases: in the specific subgroup that transitioned from a non-eosinophilic exacerbation to an eosinophilic stability phenotype, Treg cell levels remained similar between AECOPD (6.99 ± 2.75%) and SCOPD (6.70 ± 1.82%) (Additional File 5).

Fig. 2.

Frequency of Treg and Th17 cells, and M-MDSC during exacerbation and stable COPD. COPD patients were stratified by exacerbation phenotype and subsequently by stability phenotype. Panels A, C, and E present the frequencies of Treg and Th17 cells and M-MDSC, respectively, during exacerbation and stability, categorized by inflammatory phenotype. Each symbol denotes a single patient value. p-values were derived from multiple comparisons analyses. Panels B, D, and F illustrate individual patient values for Treg and Th17 cells and M-MDSC frequencies, respectively, based on the phenotype at exacerbation (Non-Eos or Eos) and subsequently at stability. Each symbol represents a patient, with paired samples (representing the same individual) connected by lines. p values were obtained from the analysis of paired samples within each group. Mean values ± SD are depicted. All AECOPD (n = 78), All SCOPD (n = 78), CS (n = 17); Eos-AECOPD and paired SCOPD (n = 18); Non-Eos-AECOPD and paired SCOPD (n = 60)

Similarly, the frequency of pro-inflammatory Th17 cells (CD45⁺CD3⁺CD4⁺CD194⁺CD183^-CD196⁺CCR10^-, Additional File 3) during AECOPD was 3.75 ± 1.76% of CD4⁺ T cells, which significantly decreased by 24% upon reaching stability (3.02 ± 1.39%, p = 0.05) (Fig. 2C). Analysis of the different subgroups, stratified by their phenotype in the exacerbation or across both phases of COPD, revealed that this decrease was not uniform. The Th17 cell frequency did not change in patients with an Eos-AECOPD (Fig. 2D, Additional File 5). However, the frequency significantly decreased in the groups with a Non-Eos-AECOPD (Fig. 2D), regardless of whether they transitioned to an Eos-SCOPD or maintained the non-eosinophilic phenotype (Additional File 5).

The frequency of circulating M-MDSC in COPD is higher in the non-eosinophilic phenotype

During AECOPD, the frequency of circulating M-MDSC (CD45⁺Lineage (CD3, CD56, CD19)^-CD33⁺CD14⁺CD15^-HLA-DR^-/low; Additional File 6) was significantly elevated (5.63 ± 5.35% of monocytes) compared to SCOPD (2.25 ± 1.73%; p < 0.001). (Fig. 2E).

Stratification by exacerbation phenotype revealed distinct M-MDSC frequencies. During Eos-AECOPD, the M-MDSC frequency was 2.83 ± 2.01%, significantly decreasing to 1.62 ± 1.38% at stability (Fig. 2F). In contrast, patients with Non-Eos-AECOPD exhibited a 2.3-fold higher M-MDSC frequency (6.48 ± 5.75%) compared to those with Eos-AECOPD. This frequency also significantly decreased to 2.44 ± 1.79% at stability (Fig. 2F, Additional File 4).

Interestingly, the most pronounced increase in M-MDSC frequency during exacerbation was observed in patients who presented with Non-Eos-AECOPD (7.08 ± 4.76%) but transitioned to an eosinophilic phenotype at stability (1.69 ± 1.39; p‹0.0001) (Additional File 5).

In contrast to M-MDSC, the frequency of e-MDSC (CD45⁺Lineage (CD3, CD56, CD19)^-CD33⁺CD14^-HLA-DR^-CD123^-; Additional File 6) did not differ significantly across the AECOPD, SCOPD, and CS groups (Table 2, Additional File 5).

Table 2.

Frequencies of e-MDSCs and pDCs, and HLA-DR expression on monocytes and pDCs during AECOPD and SCOPD

	e-MDSC	pDC	pDC	Monocytes
	(% MNC)		(MFI HLA-DR)
CS	0.14 ± 0.26	0.51 ± 0.28	25.3 ± 10.7	8.26 ± 3.23
All AECOPD	0.16 ± 0.20	0.16 ± 0.24	27.9 ± 13.0	7.60 ± 4.24
Eos-AECOPD	0.18 ± 0.17	0.18 ± 0.17	30.2 ± 11.1	9.18 ± 4.02
Non-Eos-AECOPD	0.16 ± 0.21	0.15 ± 0.24	27.2 ± 13.5	7.13 ± 4.22
All SCOPD	0.13 ± 0.25	0.39 ± 0.21	37.7 ± 20.1	14.7 ± 8.1
Eos-SCOPD	0.09 ± 0.09	0.41 ± 0.17	38.5 ± 22.7	15.2 ± 8.5
Non-Eos-SCOPD	0.17 ± 0.32	0.38 ± 0.24	36.9 ± 17.9	14.3 ± 7.8
Stats (p value)
All SCOPD vs CS	ns	0.056	0.02	0.002
All AECOPD vs All SCOPD	ns	‹0.00001	0.01	<0.00001
Eos- vs Non-Eos-SCOPD	ns	ns	ns	ns
Eos- vs Non-Eos-AECOPD	ns	ns	ns	ns
Eos-AECOPD	0.18 ± 0.17	0.19 ± 0.25	30.2 ± 11.1	9.2 ± 4.0
SCOPD (paired patients)	0.14 ± 0.14	0.43 ± 0.20	39.8 ± 28.9	16.2 ± 10.1
Stats (p value)
Eos-AECOPD vs paired SCOPD	ns	0.002	ns	0.004
Non-Eos-AECOPD	0.16 ± 0.21	0.15 ± 0.24	27.2 ± 13.5	7.13 ± 4.22
SCOPD (paired patients)	0.13 ± 0.27	0.38 ± 0.21	37.0 ± 16.9	14.2 ± 7.4
Stats (p value)
Non-Eos-AECOPD vs paired SCOPD	ns	<0.0001	0.0001	‹0.00001

e-MDSC and pDC were expressed as percentage of MNC. MFI value of HLA-DR was used to determine HLA-DR expression. Values are expressed as mean ± SD. Patients were classified according to their inflammatory phenotype. Comparisons between AECOPD and SCOPD groups were analysed in the full cohort and in each of the established groups, and p values are given. All AECOPD (n = 78), All SCOPD (n = 78), CS (n = 17); Eos-AECOPD and paired SCOPD (n = 18); Non-Eos-AECOPD and paired SCOPD (n = 60).

e-MDSC Early precursors of myeloid-derived suppressor cells, pDC Plasmacytoid dendritic cells, AECOPD Acute exacerbation of COPD, SCOPD Stable COPD, Eos- Eosinophilic, Non-Eos- non-eosinophilic, HLA-DR Human Leukocyte Antigen–DR, MFI Mean Fluorescence Intensity, SD Standard deviation

HLA-DR expression on antigen-presenting cells

Analysis of pDC (CD45⁺Lineage (CD3, CD56, CD19)^-CD33^-HLA-DR⁺CD123⁺, Additional File 6) revealed a roughly 2.5-fold decrease in pDC frequency during exacerbation compared to stability. This value was closer to that observed in the CS group. Notably, this trend remained consistent across both inflammatory exacerbation phenotypes (Table 2., Additional File 5).

Given that HLA-DR expression on antigen-presenting cells is crucial for initiating immune responses and high levels correlate with immune activation [26], we analyzed HLA-DR expression levels on pDC by Mean Fluorescence Intensity (MFI) (Additional File 7). HLA-DR MFI in the Eos-AECOPD group did not significantly change between exacerbation and stability. However, the Non-Eos-AECOPD group showed a significant increase in MFI from 27.2 ± 13.5 at exacerbation to 37.0 ± 16.9 at stability (p < 0.0001) (Table 2.). This same pattern was observed when patient subgroups were classified based on their phenotype in both COPD phases (Additional File 5).

Similarly, during AECOPD, monocyte HLA-DR MFI decreased by approximately two-fold, reaching levels comparable to the CS group (Table 2.). Interestingly, both inflammatory phenotypes demonstrated a significant reduction in monocyte HLA-DR MFI during exacerbation (Table 2.), and this same behavior was observed across the different subgroups (Additional File 5). Moreover, a significant inverse correlation was observed between circulating M-MDSC frequency and monocyte HLA-DR expression in both phenotypes during both stability and exacerbation. However, the non-eosinophilic phenotype consistently exhibited a higher number of M-MDSCs for a given MFI compared to the Eos phenotype (Additional File 8).

Effect of exacerbation etiology on circulating levels of immunosuppressive cells and pDCs

Respiratory infections were associated with 8 (44.4%) of Eos-AECOPD and 36 (60%) of Non-Eos-AECOPD cases. Viruses were detected in 4 (22.2%) of Eos-AECOPD and 23 (38.3%) of Non-Eos-AECOPD, while bacteria were found in 3 (16.6%) and 7 (11.7%) cases, respectively. A combination of both was present in 1 (5.6%) Eos-AECOPD and 6 (10%) Non-Eos-AECOPD cases. Rhinovirus (39%) and influenza A (16%) were the most frequent viruses. Among bacteria, Haemophilus influenzae was the most prevalent (33%), followed by Pseudomonas aeruginosa and Streptococcus pneumoniae (both 16.6%), and Moraxella and methicillin-resistant Staphylococcus aureus (both 6%).

To assess whether exacerbation etiology influenced changes in the frequency of studied cell subpopulations, we classified exacerbations as infectious (virus and/or bacteria) or non-infectious (absence of virus and bacteria).

Overall, the etiology of the exacerbation did not induce significant changes between the two inflammatory phenotypes in any of the three cell types (Fig. 3A, C, D). However, when analyzing patients grouped by etiology and exacerbation phenotype, we observed that only the Non-Eos-AECOPD group with non-infectious etiology showed a significant increase in Treg frequency upon comparison with the stable phase (Fig. 3B).

Fig. 3.

Effect of exacerbation etiology on the frequency of Treg, Th17 and M-MDSC cells. COPD patients were classified according to the exacerbation etiology and its inflammatory phenotype. The left panels (A, C, and E) display the frequencies of Treg, Th17 and M-MDSC cells, respectively, across the different subgroups. p-values were determined by multiple comparisons analysis. The right panels (B, D, and F) show the frequencies of Treg, Th17 and M-MDSC cells, respectively, for each group during exacerbation and upon reaching stability. p-values were obtained from the analysis of paired samples within the same group. Each dot represents an individual, and paired samples are connected by lines. Mean values ± SD are depicted. Non-infectious Eos-AECOPD (n = 10), Infectious Eos-AECOPD (n = 8); Non-infectious Non-Eos-AECOPD (n = 22); Infectious Non-Eos-AECOPD (n = 38). Abbreviations: AECOPD, acute exacerbation of COPD; SCOPD, stable COPD; Eos-, eosinophilic; Non-Eos, non-eosinophilic

Interestingly, of the 22 patients with the Non-Eos-AECOPD/Eos-SCOPD phenotype who showed no change in Treg frequency during exacerbation, 16 experienced infectious exacerbation.

Furthermore, our results show that Non-Eos-AECOPD exhibited a significant increase in both Th17 cells and M-MDSC independent of etiology. Conversely, Eos-AECOPD showed a differential response: Th17 cells significantly increased in response to non-infectious etiology, while M-MDSC responded to infectious etiology (Fig. 3D, F, respectively).

Pro-inflamatory and anti-inflammatory cytokines in COPD

Overall, AECOPD led to significantly increased serum levels of pro-inflammatory cytokines CRP, IL-6, and IL-17A and the anti-inflammatory cytokine IL-10, while TGF-β1 levels remained unchanged. This pattern was also observed in Non-Eos-AECOPD group (Table 3.). Conversely, Eos-AECOPD was associated with a significant increase in free TGF-β1 levels, when compared to its paired Eos-SCOPD group in stability. Notably, the patient group with Eos-SCOPD who suffered a Non-Eos-AECOPD exhibited greater increases in serum CRP and IL-6 levels than the Non-Eos-SCOPD group (Table 3.).

Table 3.

Serum concentrations of pro- and anti-inflammatory molecules in COPD patients during AECOPD and SCOPD

	CRP (ng/mL)	IL-17A (pg/mL)	IL-6 (pg/mL)	IL-10 (pg/mL)	Free-TGF-β1 (pg/mL)	Total TGF-β1 (ng/mL)
CS	NA	2.8 ± 2.6	2.09 ± 0.52	23.9 ± 14.3	36.4 ± 39.2	49.2 ± 8.8
All AECOPD	85.1 ± 105.2	2.5 ± 3.8	21.8 ± 38.4	30.8 ± 28.4	23.6 ± 18.5	47.5 ± 9.9
Eos-AECOPD	12.6 ± 11.9	1.8 ± 1.8	7.22 ± 4.31	14.4 ± 16.1	27.1 ± 16.0	49.3 ± 13.9
Non-Eos-AECOPD	109 ± 111	2.7 ± 4.2	26.0 ± 42.3	35.5 ± 29.8	22.7 ± 19.2	47.0 ± 8.7
All SCOPD	6.77 ± 8.58	1.4 ± 2.6	5.31 ± 3.19	15.4 ± 14.3	25.0 ± 23.7	49.2 ± 8.8
Eos-SCOPD	6.07 ± 7.39	1.6 ± 3.4	6.34 ± 3.18	10.9 ± 5.2	17.2 ± 17.6	48.4 ± 8.6
Non-Eos-SCOPD	7.39 ± 10.12	1.3 ± 1.6	4.23 ± 2.91	20.1 ± 18.9	32.4 ± 27.1	49.9 ± 9.3
Stats (p value)
All SCOPD vs CS	NA	0.05	0.02	ns	ns	ns
All AECOPD vs All SCOPD	<0.000001	0.02	0.05	0.03	ns	ns
Eos- vs Non-Eos-SCOPD	ns	ns	0.02	0.02	0.03	ns
Eos- vs Non-Eos-AECOPD	<0.0001	ns	ns	ns	ns	ns
Eos-AECOPD	12.6 ± 11.9	1.8 ± 1.8	7.2 ± 4.3	14.4 ± 16.1	27.1 ± 16.0	49.3 ± 13.9
SCOPD (paired patients)	5.8 ± 6.5	1.2 ± 1.6	7.6 ± 2.3	10.3 ± 7.3	17.7 ± 19.1	49.4 ± 11.6
Stats (p value)
Eos-AECOPD vs paired SCOPD	ns	ns	ns	ns	ns	ns
Non-Eos-AECOPD	109 ± 111	2.7 ± 4.2	26.0 ± 42.8	35.5 ± 29.7	22.7 ± 19.2	47.0 ± 8.7
SCOPD (paired patients)	7.2 ± 9.3	1.5 ± 2.8	4.7 ± 3.1	16.8 ± 15.5	27.2 ± 24.7	49.1 ± 8.1
Stats (p value)
Non-Eos-AECOPD vs paired SCOPD	<0.000001	0.01	0.01	0.0001	ns	ns
Eos-AECOPD	12.0 ± 10.3	2.0 ± 2.0	7.2 ± 4.3	14.4 ± 16.1	29.9 ± 16.0	49.3 ± 13.9
Paired Eos-SCOPD	6.1 ± 7.0	1.4 ± 1.8	7.6 ± 2.33	10.3 ± 7.3	14.9 ± 17.0	49.4 ± 11.6
Stats (p value)	ns	ns	ns	ns	0.01	ns
Non-Eos-AECOPD	79.7 ± 93.5	2.6 ± 3.9	8.5 ± 10.9	34.9 ± 31.2	30.0 ± 21.5	47.5 ± 10.0
Paired Non-Eos-SCOPD	7.7 ± 10.1	1.3 ± 1.7	4.2 ± 2.9	20.1 ± 18.9	32.4 ± 27.1	49.9 ± 9.3
Stats (p value)
Non-Eos paired groups	‹0.001	0.07	ns	0.002	ns	ns
Non-Eos-AECOPD	169 ± 125	3.0 ± 4.8	57.7 ± 59.6	36.8 ± 28.1	12.8 ± 7.1	46.2 ± 6.5
Paired Eos-SCOPD	6.4 ± 8.0	1.7 ± 4.3	5.4 ± 3.5	11.3 ± 3.0	18.9 ± 18.6	47.6 ± 5.8
Stats (p value)
Non-Eos/Eos paired groups	‹0.0001	0.01	0.02	0.01	ns	ns

Data are presented as mean ± SD. Patients were classified according to their inflammatory phenotype. The p values presented were obtained from multiple comparisons or between groups. The number of patients in each group was as follows: All AECOPD (n = 78), All SCOPD (n = 78), CS (n = 17); Eos-AECOPD and paired SCOPD (n = 18); Non-Eos-AECOPD and paired SCOPD (n = 60). Eos-AECOPD and paired Eos-SCOPD (n= 14); Non-Eos-AECOPD and paired Non-Eos-SCOPD (n= 38); Non-Eos-AECOPD and paired Eos-SCOPD (n=22)

Abbreviations: CS Control smokers, AECOPD Acute exacerbation of COPD, SCOPD Stable COPD, Eos- Eosinophilic, Non-Eos Non-eosinophilic, CRP C reactive protein, TGF-β1 Transforming growth factor- β1, SD Standard deviation

Finally, our results show that CRP levels significantly increased during Non-Eos-AECOPD, irrespective of the exacerbation etiology. However, within the Non-Eos-AECOPD group, we found that infectious processes were associated with significant increases in IL-6, IL-17 and IL-10, and a decrease in total TGF- β1, whereas non-infectious processes showed no change in the levels of these studied molecules.

Conversely, the etiology (infectious vs. non-infectious) of eosinophilic exacerbations did not significantly impact the levels of the studied molecules, with the exception of a significant increase in IL-17A in non-infectious Eos-AECOPD and CRP in infectious Eos-AECOPD (Table 4).

Table 4.

Effect of the exacerbation etiology on serum concentrations of pro- and anti-inflammatory molecules

	CRP (ng/mL)	IL-17 A (pg/mL)	IL-6 (pg/mL)	IL-10 (pg/mL)	Free-TGF-β1 (pg/mL)	Total TGF-β1 (ng/mL)
Non-infectious
Eos-AECOPD	9.69 ± 10.55	2.29 ± 2.24	5.66 ± 4.50	18.6 ± 21.5	30.0 ± 15.5	51.6 ± 12.7
SCOPD (paired patients)	7.04 ± 8.33	1.03 ± 1.47	6.86 ± 1.99	7.84 ± 2.30	16.7 ± 15.8	49.3 ± 11.3
Stats (p value)
Eos-AECOPD vs. paired SCOPD	ns	0.05	ns	ns	ns	ns
Infectious
Eos-AECOPD	17.0 ± 13.4	1.20 ± 1.03	9.17 ± 3.68	9.17 ± 3.44	23.4 ± 18.1	46.3 ± 16.6
SCOPD (paired patients)	3.82 ± 1.17	1.45 ± 1.82	8.47 ± 2.66	13.4 ± 10.6	19.0 ± 25.3	49.6 ± 13.8
Stats (p value)
Eos-AECOPD vs. paired SCOPD	0.04	ns	ns	ns	ns	ns
Non-infectious
Non-Eos-AECOPD	113 ± 125	2.07 ± 3.66	12.7 ± 15.8	38.9 ± 35.7	26.9 ± 19.9	47.7 ± 8.8
SCOPD (paired patients)	5.00 ± 4.91	0.93 ± 1.27	4.93 ± 3.79	21.4 ± 23.0	21.7 ± 19.6	47.0 ± 7.0
Stats (p value)
Non-Eos-AECOPD vs. paired SCOPD	0.001	ns	ns	ns	ns	ns
Infectious
Non-Eos-AECOPD	107 ± 105	3.09 ± 4.45	34.4 ± 51.2	33.4 ± 26.3	20.1 ± 18.9	46.6 ± 8.90
SCOPD (paired patients)	17.6 ± 49.5	1.81 ± 3.41	4.53 ± 2.79	14.1 ± 8.05	29.2 ± 27.4	50.4 ± 8.65
Stats (p value)
Non-Eos-AECOPD vs. paired SCOPD	0.001	0.014	0.01	0.001	ns	0.03

Data are presented as mean ± SD. Patients were classified according to their inflammatory phenotype and the etiology of the exacerbation. The p values presented were obtained from comparisons between groups. The number of patients in each group was as follows: Non-infectious Eos-AECOPD (n = 10), Infectious Eos-AECOPD (n = 8); Non-infectious Non-Eos-AECOPD (n = 22); Infectious Non-Eos-AECOPD (n = 38)

Abbreviations: AECOPD Acute exacerbation of COPD, SCOPD Stable COPD, Eos- Eosinophilic, Non-Eos Non-eosinophilic, SD Standard deviation

Discussion

Our findings demonstrate that the frequency of circulating immunosuppressive cells increased during AECOPD. However, the proportions of Tregs and M-MDSC, as well as their capacity for induction, were dependent on the inflammatory phenotype, which remained consistent from exacerbation to recovery in the majority of patients.

In our study, 23% of AECOPD cases presented with an eosinophilic phenotype. Notably, after recovery, this phenotype was observed in 46% of patients, a finding consistent with previous reports [4, 22, 27, 28]. It is well-established that COPD exacerbations encompass distinct biological clusters or endotypes. In a seminal study, Bafadhel et al. [22]utilized unbiased clustering and factor analyses to categorize patients into four exacerbation subgroups: pro-inflammatory, associated with bacterial infections (35% of AECOPD); T2 inflammation, corresponding to eosinophilic inflammation (30% of AECOPD); T1 inflammation, linked to viral infections (24% of AECOPD); and pauci-granulocytic inflammation (11% of AECOPD). Furthermore, these authors reported that exacerbations associated with bacteria or sputum eosinophilia could be predicted from the stable state in patients with multiple exacerbations [22]. This phenomenon has been replicated in additional cohorts [28–30]. Similarly, in our study, nearly 80% of Eos-AECOPD and 65% of Non-Eos-AECOPD cases maintained the same inflammatory phenotype at the recovery stage.

Immunosuppressive response during AECOPD

Previous studies on the distribution of Treg cells in AECOPD have yielded inconsistent results [14–17]. Our findings indicate an increase in Treg frequency during AECOPD. Consistently, IL-10 levels were significantly elevated during exacerbations. In a cross-sectional study, Xiong et al. [17] also demonstrated that the proportion of Tregs and the levels of IL-10 in patients with AECOPD were significantly higher than in stable COPD patients, healthy controls, or smokers without airflow limitation. The discrepancies between studies may be attributed to: (1) the study design; our study employed a longitudinal approach with measurements taken from the same patients during AECOPD and subsequently during stability; (2) the sample size; or (3) the methodologies employed. For instance, CD4⁺CD25⁺Tregs can be identified by FoxP3 (intracellular) or CD127 (cell surface) staining. Both Xiong et al. [17]and our study utilized CD127 staining. The presence of different inflammatory phenotypes may also contribute to the observed discrepancies. Similar to Tregs, the frequency of M-MDSC was higher during AECOPD than in SCOPD. While MDSC are known to suppress T-cell responses, there is limited information regarding their role during AECOPD. Stimulation of these cells leads to the production of IL-10, as we observed. Additionally, MDSC contribute to the clearance of dead neutrophils, aiding in the resolution of lung inflammation, preventing lung injury, and ultimately restoring tissue homeostasis [31]. Therefore, the accumulation of M-MDSC may underlie the immunosuppressive response observed in AECOPD.

Differential immunosuppressive response according to inflammatory endotype

One of the key findings of our study was demonstrating that the immunosuppressive response varied according to the underlying inflammatory phenotype. The eosinophilic phenotype of AECOPD, compared to the non-eosinophilic phenotype, was characterized by a higher expression of Tregs but a lower expression of M-MDSC. Furthermore, Eos-AECOPD did not induce significant changes in serum levels of CRP, IL-6, IL-17A, or IL-10. In contrast, Non-Eos-AECOPD induced a higher expression of M-MDSC, but a similar increase in Tregs to that induced by the Eos-AECOPD group, along with significant increases in serum levels of CRP, IL-6, IL-17A, and IL-10. Tebartz et al. [32] demonstrated using TCR-transgenic OT-II mice that infection with S. aureus induced a substantial expansion of MDSCs but a lesser expansion of Tregs, and that immunosuppression was not mediated by IL-10 or TGF-β but required cell-cell proximity or the mediation of other molecules such as nitric oxide. This finding could explain the lack of Treg induction in AECOPD associated with infectious processes observed in our study.

It is important to highlight that the group comprised of patients with Non-Eos-AECOPD who presented an eosinophilic phenotype in the stable phase experienced the greatest expansion of M-MDSC but failed to induce changes in Treg expression. IL-6 is a potent inducer of M-MDSC, even in the absence of TGF-β [33]. Therefore, it is plausible that the IL-6 levels achieved during Non-Eos-AECOPD by the Eos-SCOPD group, which were seven times higher than those observed in the Non-Eos-SCOPD group, enhance M-MDSC differentiation. However, it cannot be ruled out that IL-6 favors Th17 differentiation while inhibiting Treg differentiation [34].

Eos-AECOPD in patients with an eosinophilic phenotype in stability, unlike Non-Eos-AECOPD, induced a significant increase in TGF-β1 in its active form, suggesting that TGF-β1 plays a critical role in Eos-AECOPD-mediated immunosuppression. Accordingly, it has been reported that eosinophils in the murine lamina propria express high levels of TGF-β1 mRNA and promote the differentiation of naïve CD4⁺T cells into Tregs through mechanisms involving all-trans retinoic acid and TGF-β1 [35].

Dendritic cells during AECOPD

The frequency of pDC significantly decreased during AECOPD, an observation independent of the inflammatory phenotype and etiology of AECOPD. pDC play a critical role in the antiviral response by sensing viruses via Toll-like receptor (TLR)−7and TLR-9 [36]. Furthermore, TLR9 appears necessary for an effective anti-bacterial immune response [37]. The pDC reduction, coupled with the concurrent expansion of immunosuppressive cells, strongly suggests a state of systemic immunosuppression [38], but is also consistent with the established knowledge that viral infections and inflammatory processes mobilize pDCs to the site of virus entry or inflammation [39, 40]. Therefore, the pDC decline could alternatively indicate their recruitment to the affected organ, a possibility further supported by the elevated IL-10 levels observed during the exacerbation.

It was also noted that during SCOPD, pDC and particularly HLA-DR⁺ monocytes exhibited higher expression of HLA-DR, indicating increased pDC maturity and a higher activation state of monocytes. This may contribute to the immune dysfunction present in COPD patients. Moreover, we found that for the same level of HLA-DR the Non-Eos-COPD showed a higher proportion of M-MDSC, thus suggesting that immunosuppression mediated by M-MDSC must play an important role in this inflammatory phenotype. Our study presents certain limitations. Firstly, we acknowledge that the non-eosinophilic inflammation category encompasses both neutrophilic and pauci-granulocytic inflammation. Secondly, our analysis is limited by the number of subjects within each subgroup.

Taken together, our results strongly suggest that different pathways of immunosuppression are leading in these two inflammatory phenotypes during AECOPD.

Conclusions

These findings could have implications for understanding the pathogenesis of COPD exacerbations and the mechanisms underlying persistent immune dysfunction in COPD. Consistent with findings from previous literature, our study also demonstrates that patients experiencing Eos-AECOPD had significantly shorter hospital stays. Indeed, many studies indicate that patients with Eos-AECOPD have a lower risk of mortality, shorter hospital stays, and potentially less need for ICU admission compared to patients with Non-Eos-AECOPD [41, 42]. The likely reason for this is the lower immunosuppressive activity coupled with the absence of significant changes in the expression of inflammatory molecules. Eosinophilic inflammation is a treatable trait. Therefore, identifying a patient’s inflammatory phenotype, both during stability and exacerbations, might help predict their immune response and potentially inform more targeted therapeutic interventions.

Abbreviations

COPD

Chronic obstructive pulmonary disease

AECOPD

Acute exacerbation of chronic obstructive pulmonary disease

SCOPD

Stable chronic obstructive pulmonary disease

Eos

Eosinophilic

Non-Eos

Non-eosinophilic

IL-6

Interleukin-6

IL-10

Interleukin-10

IL-17A

Interleukin-17 A

TGF-β1

Transforming growth factor-β1

CRP

C reactive protein

Authors’ contributions

MDM, JJSC, and CM: Conception and design. CM, MLN, and MCA: Recruitment of patients, collection of samples and clinical data. MDM, CM, and VDM: Performing experiments. MDM, CM, and JJSC: Analysis and interpretation of data, and manuscript writing. All authors approved the final version of the manuscript.

Funding

This work was supported by the Valencian Pneumology Society (grant No UGP-19-483), and by GlaxoSmithKline (grant No CPRES00124).

Data availability

Data is provided within the manuscript and supplementary information files.

Declarations

Ethics approval and consent to participate

The study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of the Arnau de Vilanova Hospital, Valencia, Spain. All participants gave their informed consent.

Consent for publication

Not applicable.

Competing interests

JJSC has received speaker fees from AstraZeneca, Bial, Boehringer Ingelheim, Chiesi, FAES, GlaxoSmithKline, Grifols, Menarini and Sanofi, and consulting fees from AstraZeneca, Bial, Chiesi, GSK, Grifols and Sanofi, and grants from GSK. The rest of authors declare no competing interests.