Blood cell ratio biomarkers of systemic inflammation in chronic obstructive pulmonary disease

Abstract

Background

Blood eosinophil count is an accepted biomarker for type 2 inflammation in COPD. However, the majority of COPD patients are characterized by non-type 2 inflammation. We aimed to test readily obtainable immune cell ratios as biomarkers for clinical phenotypes in the broad COPD population and to determine pathways represented by these ratios using multi-omics data.

Methods

Using complete blood counts with differential collected at the Phase 2 (5-year) visit in the COPDGene Study, we calculated three immune cell ratios previously described in COPD and other diseases: the neutrophil–lymphocyte ratio (NLR), the platelet-lymphocyte ratio (PLR), and the Systemic Immune-Inflammation Index (SII = NLR*platelets). We tested for associations with COPD outcomes, including lung function, chest CT scan phenotypes, and exacerbations. Blood RNA-sequencing and proteomics data were used to identify genes, proteins and pathways associated with the ratios.

Results

In univariate analyses, the three biomarkers were associated with COPD severity measures. In zero inflated Poisson regression models, all three were associated with increased odds of having an exacerbation but were not associated with exacerbation counts. Conversely, the three biomarkers were generally associated with prospective exacerbation counts, but not the zero-inflation term. In logistic regression models, the three biomarkers were significantly associated with having two or more exacerbations in the prior year; however, receiver operating characteristic analyses did not lead to clear cutoff values. Complement and PI3K signaling pathways were enriched across more than one ratio in both the RNA-sequencing and proteomics results. Other inflammatory pathways relevant in COPD appeared in different enrichment sets in either omics data type.

Conclusions

Higher levels of three easily obtained blood cell ratios were associated with COPD severity and exacerbations outcomes; however, there are not clear thresholds which would be required for clinical application. Blood RNA-sequencing and proteomics identified inflammatory pathways associated with the three biomarkers, including targets for COPD therapies currently in human trials.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12931-026-03632-3.

Background

Chronic obstructive pulmonary disease is a heterogeneous condition, involving multiple cellular and molecular features, [1, 2] yet there are limited biomarkers available to define the different inflammatory pathways [3]. Blood eosinophil count, a biomarker for type 2 inflammation, is associated with exacerbation risk [4] and response to inhaled corticosteroids, [5] and has been used as an entry criteria into clinical trials of biologic therapies targeting type 2 inflammation in COPD [6, 7]. Fractional exhaled nitric oxide and immunoglobulin E levels are other biomarkers of type 2 inflammation that are used in clinical care for asthma and have been applied in research studies of COPD [7, 8]. Despite the clinical relevance of type 2 inflammation in COPD, this mechanism is likely important only in a minority of COPD patients. Studies have shown that 20–40% of COPD patients have evidence of type 2 inflammation, indicated by elevated blood eosinophil counts [4, 9]. There are no available biomarkers for other inflammatory pathways in the full COPD population.

Complete blood count with white blood cell differential is a commonly obtained clinical laboratory measurement. Both blood neutrophil counts and platelet counts are elevated in inflammatory states and are associated with COPD severity [10, 11]. Several blood cell ratios have been developed to capture systemic inflammation, including the neutrophil to lymphocyte ratio (NLR), the platelet to lymphocyte ratio (PLR) and the systemic inflammatory-immune index (SII = NLR * platelets). These ratios have been predominantly applied in cancer studies, but are increasingly shown to be relevant for COPD outcomes including exacerbations and mortality [12–15]. These indices are purported to reflect systemic inflammation, but the specific inflammatory pathways represented are unknown.

Our hypothesis is that these blood cell ratios can serve as biomarkers of non-type 2 inflammation. We executed two aims to test this hypothesis. First, we sought to confirm the associations with respiratory outcomes, especially exacerbations, in a large population of current and former smokers with and without COPD in the Genetic Epidemiology of COPD Study (COPDGene). Second, we aimed to identify the biological pathways represented by the indices using blood RNA-sequencing and proteomics data in COPDGene.

Methods

The genetic epidemiology of COPD study (COPDGene)

COPDGene is an observational study of over 10,000 current and former smokers (at least 10 pack years) with and without COPD, along with additional non-smoking controls; [16] however, non-smokers were excluded from this analysis. One-third of COPDGene subjects are African Americans. Subjects were ages 45–80 at the initial visit. Subjects with cancer within the last 5 years and with other lung diseases except asthma were excluded. The baseline study visit included questionnaires, spirometry, a 6-min walk test, and a high-resolution chest computed tomography (CT) scan with computerized analysis for emphysema and airway disease. Five- and ten-year follow-up visits (Phases 2 and 3) used similar phenotype assessments, with the addition of a complete blood count (CBC) with differential, which was assayed in each center’s clinical laboratory. COPD exacerbations in the past year are queried at each study visit, and prospective exacerbations are recorded using web or telephone-based longitudinal follow-up questionnaires every 6 months [17]. COPD exacerbations were defined by use of antibiotics and/or systemic steroids. Severe exacerbations required emergency department visits or hospitalization. Frequent exacerbators were defined as two or more exacerbations in the prior year. All subjects gave written informed consent. COPDGene was approved by the Institutional Review Boards at all participating centers.

RNA-sequencing and proteomics

At the Phase 2 visit, blood samples were collected for RNA-sequencing and proteomics; methods have been previously published [18, 19]. Briefly, whole blood was collected in PaxGene RNA tubes (BD Biosciences). RNA was extracted and used to generate stranded total RNA libraries with ribosomal reduction. Illumina sequencing used 75 bp paired-end reads. Blood proteomics was generated using SomaScan v4.0 which uses aptamers to quantify approximately 5000 proteins. Whole genome sequencing is available through the NHLBI Trans-Omics in Precision Medicine (TOPMed) program [20]. Duffy null phenotype was determined by minor allele homozygosity for a single nucleotide polymorphism in the atypical chemokine receptor 1 gene (ACKR1; rs2814778) [21].

Statistical analysis

All statistical analyses were performed in R (v.4.3.2) in R Studio Pro Server (v2025.05.1 + 513). Unless otherwise stated, to test the univariate associations of NLR, PLR, and SII with clinical variables, we used the 2-tailed unpaired Student’s t test for categorical variables, and correlation for quantitative variables. To test the association between three immune cell ratios and exacerbations counts, zero-inflated Poisson regression analyses were performed using the pscl (v.1.5.9) package in R. Zero-inflated Poisson models were adjusted for age, sex, race, current smoking status (vs. former), pack-years, FEV₁ percent predicted and Duffy null phenotype, the major determinant of neutrophil count primarily in populations of African descent [22]. Scale function was applied to the predictors, which generally improves model stability and convergence, especially for predictors with large values or different scales. To test the association between the three immune cell ratios and binary exacerbation outcomes, logistic regression analyses were performed using stats (v.4.2.0) package in R. Logistic models adjusted for age, sex, race, smoking status (vs. fomer), pack-years, FEV₁ percent predicted and Duffy null phenotype. Models for prospective exacerbations were additionally adjusted for the number of exacerbations in the prior year.

We performed differential gene expression and proteomic analysis using Limma (v3.54.2) in R. Linear models were adjusted for age, sex, race, smoking status (current vs former), pack-years, and FEV₁ percent predicted. As per previous COPDGene analyses, proteomics were additionally adjusted for clinical center, [19] and RNA-seq was also adjusted for library batch [18]. We used Sigora (v3.1.1) to identify the underlying function of gene sets. Four annotated gene sets – hallmark gene sets, [23] KEGG, [24] Gene Ontology [25] and Reactome [26] v2025.1 – were chosen for the reference gene sets. An FDR adjusted p-value < 0.05 was set as the cutoff criteria. Volcano plots were drawn with the EnhancedVolcano package.

Results

Neutrophil–lymphocyte ratio (NLR), platelet-lymphocyte ratio (PLR), and systemic immune-inflammation index (SII) were derived from Phase 2 CBC data for 5,652 current and former smoking subjects, including 1,970 with COPD, defined as post-bronchodilator FEV₁/FVC < 0.7 and FEV₁ < 80% predicted, corresponding to Global Initiative for Chronic Obstructive Lung Disease (GOLD) stages 2–4 (Table 1). Even within the COPD subjects, there is heterogeneity in phenotypes, such as exacerbation history. The three measures were highly correlated: NLR-PLR (Pearson) r = 0.66, NLR-SII r = 0.89, SII-PLR r = 0.76, in all subjects.

Table 1.

COPDGene Phase 2 subject characteristics. N(%) or mean (SD) are shown

	All Subjects (n = 5652)	COPD GOLD 2–4 (n = 1970)
Female sex	2799 (49.5%)	898 (45.6%)
Non-Hispanic white	3930 (69.5%)	1493 (75.%8)
African American	1722 (30.5%)	477 (24.2%)
Current smoker	2194 (38.8%)	677 (34.4%)
Age, years	65.26 (8.66)	67.76 (8.24)
Pack-years of smoking	43.70 (24.09)	51.66 (25.06)
FEV1% predicted, post-bronchodilator	77.81 (24.54)	52.27 (17.22)
GOLD stage
0	2423 (42.9%)	NA
1	542 (9.6%)	NA
2	1131 (20.0%)	1131 (57.4%)
3	590 (10.4%)	590 (29.9%)
4	249 (4.4%)	249 (12.6%)
PRISM	717 (12.7)	NA
SGRQ total score	23.47 (21.31)	36.32 (21.03)
6 min walk distance, ft	1290.99 (436.86)	1113.35 (434.12)
BODE score	1.68 (2.15)	3.32 (2.48)
BMI (kg/m2)	28.99 (6.40)	28.27 (6.45)
% emphysema (LAA950)	5.42 (9.02)	11.86 (12.43)
Perc15	84.81 (30.36)	66.18 (29.25)
Pi10	2.28 (0.58)	2.69 (0.56)
Airway wall thickness of segmental airways	1.04 (0.22)	1.13 (0.23)
Wall area % of segmental airways	50.29 (8.43)	54.77 (7.98)
White blood cell count	7.18 (2.34)	7.56 (2.31)
Neutrophil count	4.33 (1.80)	4.76 (1.95)
Lymphocyte count	2.06 (1.12)	1.95 (0.87)
Monocyte count	0.56 (0.21)	0.6 (0.22)
Eosinophil count	0.19 (0.15)	0.2 (0.16)
Exacerbations in the prior year
0	4652 (82.3%)	1336 (67.8%)
1	581 (10.3%)	350 (17.8%)
2 or more	419 (7.4%)	284 (14.4%)
Severe exacerbation in the prior year	532 (9.4%)	361 (18.3%)

Abbreviations: BODE Body mass index, Obstruction, Dyspnea, Exercise Capacity, BMI Body mass index, FEV₁ Forced expiratory volume in 1 s, GOLD Global Initiative for Chronic Obstructive Lung Disease, LAA950 Low attenuation area at −950 Hounsfield Units, Perc15 15th percentile of lung density histogram, Pi10 Square root wall area of hypothetical airway with 10 mm internal perimeter, PRISM Preserved Ratio Impaired Spirometry (FEV₁ < 80% predicted with FEV₁/FVC ≥ 0.7) [27], SGRQ St. George’s Respiratory Questionnaire

In univariate analyses, all three biomarkers were significantly associated with demographic and clinical variables such as sex, race, smoking status, age, lung function (FEV₁% predicted), disease-related quality of life (St. George’s Respiratory Questionnaire [SGRQ] total score), exercise capacity (6-min walk distance), and chest CT scan emphysema measures in all subjects and in COPD only (Tables 2 and 3); overall, the correlations were weak. NLR and SII, but not PLR, were associated with chest CT scan airway metrics, in all subjects and in COPD only. Weak negative correlations were found between blood eosinophil counts and NLR (r = −0.03) and PLR (r = −0.07); SII was not significantly correlated (r = −0.009). Moderate repeated measures correlations were observed between Phase 2 and Phase 3 values for NLR (r = 0.34, n = 2844), PLR (r = 0.53, n = 2826), and SII (r = 0.35, n = 2826).

Table 2.

Univariate associations of NLR, PLR, and SII with clinical variables. Categorical variables (mean values are shown)

Variable	All subjects			COPD (GOLD 2–4)
	NLR	PLR	SII	NLR	PLR	SII
Male	2.63**	132.1	594.1	3.12**	141.9	715.9
Female	2.29	134.0	590.7	2.66	142.9	967.8
White race	2.70**	136.9**	648.6**	3.12**	145.2*	763.0**
African American race	1.90	124.0	461.7	2.24	133.3	533.1
Current smoker	2.19**	121.9**	539.4**	2.49**	126.8**	614.3**
Former smoker	2.63	140.1	625.8	3.12	150.4	756.1

Abbreviations: GOLD Global Initiative for Chronic Obstructive Lung Disease, NLR neutrophil to lymphocytes ratio, PLR platelet to lymphocyte ratio, SII systemic immune-inlammation index

^*p < 0.05 **p < 0.001. P-values are for t-test

Table 3.

Univariate associations of NLR, PLR, and SII with clinical variables. Quantitative variables (correlation coefficients are shown)

Variable	All subjects			COPD (GOLD 2–4)
	NLR	PLR	SII	NLR	PLR	SII
Age	0.20**	0.12**	0.14**	0.17**	0.10**	0.11**
Pack-years	0.12**	0.02	0.08**	0.08**	0.02	0.06*
FEV1% predicted, post-bronchodilator	−0.19**	−0.09**	−0.17**	−0.16**	−0.12**	−0.14**
SGRQ total score	0.17**	0.08**	0.16**	0.15**	0.11**	0.16**
6 min walk distance	−0.16**	−0.08**	−0.15**	−0.18**	−0.14**	−0.17**
BODE index	0.21**	0.13**	0.20**	0.19**	0.15**	0.18**
% emphysema (LAA950)	0.21**	0.14**	0.19**	0.15**	0.15**	0.15**
Perc15	−0.19**	−0.17**	−0.17**	−0.15**	−0.16**	−0.15**
Pi10	0.12**	0.01	0.11**	0.09**	0.03	0.08**
Airway wall thickness	0.14**	−0.01	0.09**	0.10**	0.01	0.06*
Wall area %	0.13**	−0.01	0.10**	0.08*	0.01	0.06*

Abbreviations BODE Body mass index, Obstruction, Dyspnea, Exercise Capacity, FEV₁ forced expiratory volume in 1 s, GOLD Global Initiative for Chronic Obstructive Lung Disease, LAA950 Low attenuation area at −950 Hounsfield Units, NLR Neutrophil to lymphocytes ratio, Perc15 15th percentile of lung density histogram, Pi10 Square root wall area of hypothetical airway with 10 mm internal perimeter, PLR Platelet to lymphocyte ratio, SGRQ St. George’s Respiratory Questionnaire, SII Systemic immune-inflammation index

^*p < 0.05 **p < 0.001, P-values are for correlation test

Exacerbations were assessed both retrospectively and prospectively. Subjects were followed for an average of 7.05 (± 2.65) years. In zero inflated Poisson regression models for exacerbations in the past year, adjusted for clinical covariates and Duffy null genotype, higher values for all three biomarkers were associated with reduced odds for being a non-exacerbator (zero inflation term in Table 4) – i.e. higher odds of having an exacerbation – but were not associated with exacerbation counts. Conversely, the three biomarkers were generally associated with prospective exacerbation counts, but not the zero inflation term. Similar results were found for prospective severe exacerbation counts (data not shown). In logistic regression models adjusted for covariates, the three biomarkers were significantly associated with having frequent exacerbations, defined as two or more in the prior year (Table 5). All three were associated with having a severe exacerbation in the past year; however only NLR and SII were associated with having a severe exacerbation in the prospective follow-up.

Table 4.

Associations with exacerbation outcomes. Zero inflated Poisson regression models for exacerbations counts (beta (SE) are shown)

Outcome	Term	All subjects			COPD (GOLD 2–4)
		NLR	PLR	SII	NLR	PLR	SII
Exacerbations in past year	Zero inflation	−0.22 (0.05)**	−0.22 (0.05)***	−0.25 (0.05)***	−0.18 (0.07)**	−0.21 (0.07)**	−0.24 (0.09)**
	Count	0.001 (0.018)	−0.001 (0.018)	0.002 (0.016)	0.02 (0.03)	−0.004 (0.03)	0.003 (0.02)
Prospective exacerbations	Zero inflation	0.14 (0.09)	0.28 (0.08)***	0.13 (0.08)	0.32 (0.21)	0.28 (0.17)	0.38 (0.21)
	Count	0.43 (0.01)***	0.37 (0.01)***	0.39 (0.01)***	0.65 (0.01)***	0.42 (0.02)***	0.54 (0.01)***

Abbreviations: GOLD Global Initiative for Chronic Obstructive Lung Disease, NLR neutrophil to lymphocytes ratio, PLR platelet to lymphocyte ratio, SII systemic immune-inflammation index

**p < 0.01, ***p < 0.001. P-values are for regression coefficients

Table 5.

Associations with exacerbation outcomes. Logistic regression models for frequent exacerbations (2 or more in the prior year) and severe exacerbations (beta (SE) shown)

Outcome	All subjects			COPD (GOLD 2–4)
	NLR	PLR	SII	NLR	PLR	SII
Frequent exacerbations	0.11 (0.02)***	0.003 (7e-4)***	4.6e-4 (9e-5)***	0.07 (0.03)**	0.002 (8e-4)**	3e-4 (1e-4)**
Retrospective severe exacerbation	0.09 (0.02)***	0.002 (6e-4)***	3e-4 (8e-5)***	0.08 (0.02)**	0.002 (4e-4)**	3e-4 (9e-5)***
Prospective severe exacerbation	0.06 (0.02)***	5.6e-4 (5.3e-4)	1.5e-4 (7.4e-5)*	0.07 (0.03)**	0.001 (6.6e-4)	1.8e-4 (9.1e-5)*

Abbreviations: GOLD Global Initiative for Chronic Obstructive Lung Disease, NLR Neutrophil to lymphocyte ratio, PLR Platelet to lymphocyte ratio, SII Systemic immune-inflammation index

^*p < 0.05, **p < 0.01, ***p < 0.001. P-values are for regression coefficients

To assess the discrimination of frequent exacerbation status (two or more in the prior year), we next constructed receiver operating characteristic (ROC) analyses. When combined with clinical covariates, all three biomarkers showed small but significantly increased area under the ROC curve (AUROC) values compared to covariates alone in all subjects (Fig. 1, Table 6). In COPD subjects, AUROCs were slightly lower than in all subjects (Supplemental Fig. 1, Table 6); NLR and SII showed significant improvement compared to covariates alone. We examined various cutoffs to dichotomize the three biomarkers, using the 1 st, 2nd or 3rd quartile of each. ROC curve analyses did not show an optimal cutoff in all subjects or in COPD subjects only (Supplemental Table 1).

Fig. 1.

Receiver operating characteristic curves for frequent exacerbations outcome (2 or more in the prior year)

Table 6.

Area under the receiver operating characteristic curve, frequent exacerbation outcome (2 or more in the prior year)

	Covariates only	NLR	NLR  + covariates	PLR	PLR  + covariates	SII	SII  + covariates
All subjects	78.7	60.6	79.6a,b	58.2	79.5a,b	62.9	79.7a,b
COPD (GOLD 2–4)	74.6	57.1	75.4a,b	57.7	75.4b	59.7	75.6a,b

Abbreviations: GOLD Global Initiative for Chronic Obstructive Lung Disease, NLR Neutrophil to lymphocyte ratio, PLR Platelet to lymphocyte ratio, SII Systemic immune-inflammation index

^ap < 0.05 for comparison vs covariates only by Delong test

^bp < 0.0001 for comparison with biomarker only (e.g. NLR) by Delong test

We performed a sensitivity analysis to assess whether blood eosinophil counts affected these associations in all subjects and in COPD only. The results of zero inflated Poisson regression models for exacerbations in the past year and logistic regression models for frequent exacerbations (2 or more in the prior year) and severe exacerbations in the past year did not change appreciably when adjusted for blood eosinophil counts, nor did the area under the ROC curves for the frequent exacerbations outcome (data not shown).

Blood RNA-sequencing and proteomics

Using whole blood RNA-sequencing data we found that the majority of the expressed transcripts were significantly associated (false discovery rate < 0.05) with each of the three blood cell ratios: NLR 13,458 transcripts, PLR 13,380, SII 13,307 (Supplemental Tables 2– 4, Supplemental Fig. 2). A large number of plasma proteins were also significantly associated with each ratio, although these represented a lower fraction of the total proteins assayed: NLR, 1429 proteins; PLR, 1602; SII, 1807 (Supplemental Tables 5– 7, Supplemental Fig. 3). Pathway enrichment analysis of the top 300 genes in each analysis (ranked by adjusted p-value) was performed using several databases including Gene Ontology, KEGG, Reactome, and Hallmark gene sets (Table 7, Supplemental Tables 8–13). As positive controls, we employed pathways related to neutrophil degranulation (NLR and SII RNA-seq and proteomics) and platelet activation (PLR RNA-seq and proteomics). In the RNA-seq analysis, there was enrichment across more than one cell ratio in several pathways, including apoptosis, Rho GTPase, complement, and IL6/JAK/STAT3 signaling. Similarly, in the proteomics data, there was enrichment across more than one ratio in pathways related to collagen and extracellular matrix, epithelial-mesenchymal transition, interleukin signaling, complement, and PI3K signaling. Thus, the latter two pathways appeared in both RNA-seq and proteomics analyses. Other inflammatory pathways relevant in COPD appeared in different enrichment sets, including signaling pathways such as TNF-alpha, WNT, and IL-2/STAT5. Interestingly, IL-4/IL-13 signaling, relevant for T2 inflammation, was enriched in the SII RNA-seq and NLR proteomics results. A number of targets for COPD therapies currently in human trials appeared in the enriched proteomics pathways, such as neutrophil elastase, alpha-1 antitrypsin, and IL1RL1 (ST2 protein) [28, 29].

Table 7.

Selected results from pathway analysis of RNA-sequencing and proteomics data. Full results are available in Supplemental Tables 7–13

Pathway	RNA-sequencing			Proteomics
	NLR	PLR	SII	NLR	PLR	SII
Complement		**	**	**	**	**
PI3K signaling			**	**		**
Apoptosis	**		**
RhoGTPase	*	**	*
Epithelial-mesenchymal transition				**	**	**
Collagen/extracellular matrix				**	**	**
IL6/JAK/STAT3 signaling	**		**	**
IL2/STAT5 signaling			*			**
TNF-alpha signaling			**
WNT signaling				**
IL4/IL13 signaling			**	**

^*Bonferroni adjusted p < 0.05** Bonferroni adjusted p < 0.001. P-values are for pathway enrichment

Discussion

We examined three blood cell ratios – neutrophil–lymphocyte ratio, platelet-lymphocyte ratio, and systemic immune-inflammatory index – in a large study of current and former smokers with and without spirometry-defined COPD. We found that these ratios were either weakly negatively correlated or uncorrelated with blood eosinophil count, a currently used biomarker in COPD management. They were weakly associated with almost all COPD severity outcomes, including retrospective and prospective exacerbations, though it is not clear whether they add substantially to clinical risk factors for exacerbations. We did not find clear threshold levels which would be required for use as a clinical biomarker in COPD management. Using RNA-sequencing and proteomics obtained at the same blood draw, we showed that these ratios were associated with multiple biologic pathways, including inflammatory pathways relevant for COPD.

The three ratios, especially the NLR, have been extensively studied in multiple chronic diseases in addition to COPD. Previous studies have demonstrated associations with COPD outcomes, including acute exacerbations. Studies of NLR and PLR have been reviewed elsewhere [12, 30, 31]. Despite the multiple studies, no clear consensus has emerged on an optimal cut-off which would be required to use a ratio as a clinical biomarker. Our study largely confirmed that higher levels of these biomarkers are associated with exacerbations and similarly did not find an optimal threshold value for any of the three ratios. We controlled for multiple clinical covariates including lung function, and found a statistically significant, but unlikely clinically relevant, improvement in discriminative ability in the ROC curve analyses. Even when adjusting for lung function, we AUROCs were higher in the entire cohort than in the subset with COPD. Prior studies in COPDGene have shown that current and former smokers without airflow obstruction may also experience respiratory exacerbations, [32] which may be related to systemic inflammation.

The SPIROMICS study, another large observational study of COPD in the US, has recently reported associations between NLR and outcomes in current and former smokers with and without COPD, as well as non-smokers [33]. They found stability in NLR values at 6 weeks and 1 year (intraclass correlations 0.74 and 0.62, respectively). Subjects with NLR in the highest quartile had increased odds of an exacerbation within the following year and higher mortality. They found no difference in mortality for quartiles 1–3, but did not examine these thresholds for associations with exacerbations, as we did. Similar to our study, the SPIROMICS analyses were adjusted for Duffy null phenotype.

Our study addressed three easily obtained blood cell ratios in a large sample with and without COPD, finding consistent results regardless of disease status. The novelty of our study is the assessment of two different omics data types, blood RNA-sequencing and proteomics, allowing us to determine the biological pathways that may be indicated by these biomarkers. Reassuringly, we found strong enrichment for neutrophil and platelet-related pathways. Several inflammatory pathways were found in one of the two omics datasets. Complement and phosphatidylinositol 3-kinase (PI3K) signaling were shared by more than one biomarker in each of the two omics data types.

Proteins in the complement pathway are known to be altered in COPD [34]. Proteomics analysis of blood and bronchoalveolar lavage fluid found enrichment for proteins in the complement pathway in subjects with rapid lung function decline [35, 36]. Markers of complement activation are elevated in sputum during COPD exacerbations [37]. Hypocomplementemic urticarial vasculitis, a rare small vessel vasculitis, has been associated with COPD [38]. Furthermore, components of cigarette smoke may cause dysregulation of the complement system [39].

PI3K signaling is involved in fundamental cellular processes such as growth and proliferation. Several PI3K inhibitors are FDA-approved for use in different types of cancer [40]. PI3K activation leads to secretion of inflammatory cytokines and generation of reactive oxygen species by several immune cell types in COPD, including neutrophils [41]. PI3K is a potential target for COPD therapies [42]. Nemiralisib, a PI3K delta inhibitor, has shown mixed results thus far regarding efficacy, though it appears to be safe [43].

Our study has several limitations. In COPDGene, CBC with differential were obtained starting at the Phase 2 (5-year) visit, not at the baseline visit. Despite this, we still have a large sample size with an average of 7 years follow-up time. Furthermore, the RNA-seq and proteomics data were collected concurrently at Phase 2. The blood cell ratios may not fully reflect lung inflammation; bronchoalveolar lavage or sputum samples would be required for comparison and were not included in COPDGene.

Conclusions

GOLD recommends measurement of blood eosinophil count as part of the initial assessment of a person with COPD [3]. The three ratios we analyzed can be derived from the same CBC with differential that is used to measure the blood eosinophil count. Currently, COPD patients are dichotomized into type 2 and non-type 2 inflammation subtypes using the blood eosinophil count. However, COPD is a heterogeneous condition, and it is likely that both pathways, and other inflammatory pathways, are active to varying degrees in individual patients. Further studies will be needed to determine how these biomarkers can be used to determine response to existing anti-inflammatory therapies as well as novel therapies targeting other inflammatory mechanisms and how they relate to other measures of COPD progression such as lung function decline.

Abbreviations

AUROC

Area under the ROC curve

BODE

Body mass index, Obstruction, Dyspnea, Exercise Capacity

CBC

Complete blood count

CT

Computed tomography

FEV₁

Forced expiratory volume in 1 s

FVC

Forced vital capacity

GOLD

Global Initiative for Chronic Obstructive Lung Disease

LAA950

Low attenuation area at -950 Hounsfield Units

NLR

Neutrophil to lymphocyte ratio

Perc15

15^Th percentile of lung density histogram

Pi10

Square root wall area of hypothetical airway with 10 mm internal perimeter

PLR

Platelet to lymphocyte ratio

ROC

Receiver operating characteristic

SGRQ

St. George’s Respiratory Questionnaire

SII

Systemic immune-inflammation index

Authors’ contributions

Concept and design: KS, JLC, CPH Data collection: PJC, RPB, CPH Data analysis: KS Statistical support: AS, JHY, MHR, LR Manuscript writing and editing: KS, AS, JHY, MHR, ES, PJC, LR, PRB, JLC, CPH Funding: CPH, RPB.

Funding

NIH grants R01HL166231 and K24HL173667 (CPH).

COPDGene was supported by NHLBI grants U01HL089897 and U01HL089856 and by NIH contract 75N92023D00011. Proteomics data generation was supported by R01HL137995. The COPDGene study (NCT00608764) has also been supported by the COPD Foundation through contributions made to an Industry Advisory Committee that has included AstraZeneca, Bayer Pharmaceuticals, Boehringer Ingelheim, Genentech, GlaxoSmithKline, Novartis, Pfizer, and Sunovion.

Data availability

COPDGene data are available on the NCBI’s Database of Genotypes and Phenotypes (dbGaP), accessions phs000179, phs000765, phs000951.

Declarations

Ethics approval and consent to participate

COPDGene was approved by the Institutional Review Board at Mass General Brigham and all participating centers. The research was conducted in accordance with the Belmont Report.

Consent for publication

Not applicable.

Competing interests

JHY reports consulting fees from Bridge BioTherapeutics and serving on an Advisory Board for Genentech. PJC reports grants from Sanofi and Bayer, and consulting fees from Verona Pharma and Genentech. JLC reports grants from the COPD Foundation, consulting fees from AstraZeneca and participation on Data Safety Monitoring Board or Advisory Board for Novartis and Genetech. CPH reports grants from the Alpha-1 Foundation and Bayer, and consulting fees from Apogee Therapeutics, AstraZeneca, Chiesi, Genentech, Ono Pharma, Sanofi, Takeda, and Verona Pharma. None of the other authors report any conflicts of interest.