Plasma YKL-40 and all-cause mortality in patients with chronic obstructive pulmonary disease

Airflow limitation is a central feature of chronic obstructive pulmonary disease (COPD). The airflow limitation is irreversible, and it is recognized that localized tissue destruction in response to inflammatory processes in lung tissue due to prolonged exposure to noxious gases like tobacco smoke is associated with the development of COPD (http://​www.​goldcopd.​org – accessed 1 February, 2013). The disease is usually progressive, and it is one of the leading causes of death in the Western world [1, 2]. In addition to localized inflammation in lung tissue, systemic low grade inflammation is recognized as part of the disease spectrum in COPD [3, 4]. Basal levels of systemic inflammation could reflect disease activity and thus be a valuable tool in determining disease activity in patients with COPD.

The plasma concentration of YKL-40 (also called chitinase3-like-1 (CHI3L1)) has attracted attention as a biomarker of disease activity in a wide array of diseases hallmarked by chronic low grade inflammation, tissue remodeling, and fibrosis, e.g. cardiovascular diseases [5‐7], asthma [8], diabetes mellitus type 1 [9] and 2 [10, 11], rheumatoid arthritis [12], liver fibrosis [13‐15], and cancer [16]. Furthermore, YKL-40 levels have been shown to be a strong predictor of overall mortality in patients admitted to hospital irrespective of diagnosis [17].

The crystal structure of YKL-40 is known [18]. YKL-40 is mainly secreted by cancer cells, macrophages, and neutrophils [16, 19]. Studies suggest that YKL-40 plays a role in cell proliferation and differentiation [20], inflammation [21, 22], extracellular tissue remodeling [21], and protection against apoptosis [23]. YKL-40 also induces cancer angiogenesis both independently and through stimulation of vascular endothelial growth factor [24]. In Streptococcus pneumoniae infected CHI3L1 null mice, YKL-40 is a regulator of antibacterial responses that augment antimicrobial resistance by contributing to bacterial killing and controlling bacterial dissemination [25].

Recently, it has also become increasingly evident that YKL-40 plays a role in inflammatory lung diseases. Increased concentrations of YKL-40 in plasma and bronchoalveolar lavage fluid are found in patients with asthma [8], COPD [26], and idiopathic pulmonary fibrosis (IPF) [27]. Interestingly, high YKL-40 levels predicted short survival in 85 patients with IPF [27]. When exposed to YKL-40, macrophages from COPD patients produce elevated levels of the pro-inflammatory biomarkers IL-8, MCP-1, MIP-1α, and MMP-9 [26], and YKL-40 is secreted from alveolar macrophages when these are stimulated by TNF-α. Serum YKL-40 is positively correlated to low-attenuation area percentage a marker of the extent of lung emphysema, and negatively correlated to forced expiratory volume in 1 second (FEV₁)% predicted, a marker of disease severity, in patients with COPD [28] and patients with asthma [8]. A single nucleotide polymorphism in the promoter of the CHI3L1 gene (-131 C → G) of patients with asthma was correlated with elevated serum YKL-40, bronchial hyper reactivity, and pulmonary function [29]. Knockdown of the CHI3L1 gene in a human airway epithelia cell line protected the cell line against hypoxic cell damage [30], further substantiating the pro-inflammatory role of YKL-40 in inflammatory pulmonary disease.

In this study we investigated whether plasmaYKL-40 levels above the age-corrected 75% percentile were associated with long-term mortality in a group of patients with moderate to very severe COPD. We also examined whether there was a relationship between COPD severity and plasmaYKL-40 as previously reported. The hypothesis was that plasma concentrations of YKL-40 above the 75% age-corrected percentile reflect increased basal inflammation in patients with moderate to very severe COPD which is implied by an increased mortality rate in patients with COPD. We tested this hypothesis in 493 patients with COPD followed for 10 years.

COPD is characterized by a localized and usually progressive destruction of lung tissue, and increasing awareness has been given to the systemic effects of COPD. It is believed that the ongoing inflammation in the lungs “overspills” into the systemic circulation. Monitoring of systemic inflammatory biomarkers may reflect disease activity in patients with COPD, and would help to monitor disease progression, treatment efficacy, and identification of COPD phenotypes that would benefit from disease modifying pharmacotherapy.

Several putative inflammatory biomarkers in plasma or serum like C-reactive protein (CRP) [34, 35], pulmonary and activation-regulation chemokine (PARC/CCL-18) [36], and fibrinogen [37] have been tested for their ability to predict all-cause and COPD-related mortality in patients with various stages of COPD. Despite showing promise as prognostic biomarkers, serum CRP and fibrinogen are not modified by potent inflammation modifying medications [38]. A general consensus on the ability of serum CRP to predict mortality was challenged in a study which was unable to demonstrate the same predictive value of serum CRP on mortality in patients with moderate to very advanced COPD [39]. Furthermore, the repeatability of serum CRP in patients with COPD and stable disease showed a high degree of variability, suggesting that the use of serum CRP as a biomarker of basal disease activity in patients with COPD is unfeasible [40].

The reasons for these ambiguous results can be many, but a recent study found that only a subset of patients with COPD is characterized by persistent systemic inflammation and the authors propose that a clinical phenotype with persistent inflammation is the reason for this [41]. In this study patients with persistently elevated levels of a select group of inflammatory biomarkers (IL-6, CRP, fibrinogen and white blood cells) were associated with an adverse outcome. These findings are very interesting as a very recent study found that IL-6 levels increased during a three year period whereas no change was apparent in mean CRP levels. IL-6 levels correlated with six-minute walk distance and mortality further corroborating a role of IL-6 as a marker of persistent inflammation [42] an interesting finding as a fairly recent study found that IL-6, but not TNF-α, stimulates YKL-40 production. These results suggest that IL-6 could be an upstream activator of YKL-40 independent of TNF-α [43]. In the present study of patients suffering from moderate to very severe COPD, we found that a high plasma concentration of YKL-40 was an independent predictor of shorter OS. This is a novel observation in COPD patients. The study benefited from a fairly large number of 493 well-characterized patients, and within the study period of 10 years, follow-up was almost complete (99.4% complete). In addition to this, the study population carried a very high fatality rate, and more than 76% of the population died during the study period.

Our primary outcome was all-cause mortality. We did not have access to cause-specific mortality in this study. It is well known that patients suffering from COPD are subject to co-morbidities, e.g. lung cancer and cardiovascular disease, which are associated with elevated plasma concentrations of YKL-40 and increased mortality [5, 6, 16, 44]. We cannot rule out that these causes of death were a contributing factor to death in our cohort. However, the patients were excluded from inclusion into the study if they suffered from pulmonary malignancies, other pulmonary disease, or if they were suffering from advanced heart or kidney disease (Table 1). COPD as a primary cause of death is underreported, and hence a more general inquiry into causes of death would most likely underestimate COPD as cause of death [45].

Our study suffered from lack of a healthy control group. When we compared plasma YKL-40 of the patients with age-adjusted plasma YKL-40 in a large group of 3130 healthy subjects from the general population [31], we found that 35% of the patients with COPD had a plasma YKL-40 level higher than the age-corrected 75% level in healthy subjects. In the literature, the cut-off values for high/normal plasma YKL-40 in patients with cancer are often set at 90% or 95%. If these cut-offs were used, 17% of the COPD patients had a plasma YKL-40 level higher than the 90% upper normal level, and only 8% of the COPD patients had a plasma YKL-40 level higher than the 95% upper normal level. We initially decided to use the age-corrected 75% percentile because we deemed it reasonable that levels above this cut-off would indicate low- grade increased inflammatory activity in COPD. We thought it less likely that plasma YKL-40 would elevate to levels comparable to those seen in metastatic cancer.

Corticosteroids decrease plasma concentrations of YKL-40 [46]. The use of corticosteroids was not registered in our cohort of COPD patients at the time of enrollment. But the blood samples used for determination of YKL-40 were drawn at time of inclusion in the study when the patients were in a stable disease phase, and only a minor group of patients were probably treated with oral corticosteroids. Patients were excluded if they had been treated with antibiotics up to 1 week before inclusion or if they had a history of hospitalization within the last month. This should minimize the risk of patients having active infections or exacerbations present at the time of blood sampling and would give a more accurate picture of chronic inflammation in patients with COPD. We cannot exclude that we underestimated the plasma levels of YKL-40 in individual cases.

The original trial investigated the effects of the antibiotic azithromycin on a number of outcomes. Azithromycin is a potent antibiotic but is also an anti-inflammatory drug. This serves as a confounder when interpreting our results. The distribution of patients receiving azithromycin was comparable in patients with low or high plasma YKL-40 (Table 2). We were unable to demonstrate an effect of azithromycin on OS in the univariate model or the final multivariate model (results not shown). In addition to this, the final multivariate model was tested in a model in which we stratified into a group who had received treatment vs. one that had not. These groups were then examined using a likelihood ratio test in which we compared the two models against each other and no difference was found (results not shown). Thus we do not believe this potential confounder had any impact on the outcome of this study.

An association between mortality and increased plasma concentrations of YKL-40 does not prove causation. The finding is interesting, however, as the same observation has been made in patients suffering from IPF [27]. Increased levels of several key inflammatory mediators are secreted from macrophages in patients suffering from COPD, e.g. IL-8, MCP-1, MIP-1α and MMP-9 when stimulated by YKL-40 [26]. Intriguingly, stimulation of macrophages with TNF-α results in increased secretion of YKL-40 [26]. This potentially places YKL-40 centrally in the inflammatory cascade between upstream signaling through TNF-α and downstream signaling via IL-8, MCP-1, MIP-1α, and MMP-9. These findings were recently contested however and the role of YKL-40 in airway inflammation remains to be elucidated (see above) [43].

The notion that high plasma YKL-40 is associated with increased inflammatory response and a rapid decline of lung function is supported by our findings of short OS. Paraclinical findings in another study support this hypothesis as high plasma YKL-40 was associated with high levels of low-attenuation area percentage and a negative correlation to FEV₁% predicted [28].

Other potential markers of short OS were tested in the present study. As expected, patients with severe or very severe COPD had shorter OS than did patients with moderate COPD. We could not confirm that BMI and Charlson Comorbidity Index were prognostic parameters. The median BMI of 24 in our cohort was high, and only 100 (20%) patients had BMI < 20. Twenty-six percent of our patients had moderate COPD, and Charlson Comorbidity Index may not accurately predict mortality in patients suffering from advanced COPD. Such patients are already at an increased risk of dying, and co-morbidities may play a lesser role.

No association was found between FEV₁% predicted and plasma YKL-40 in our study. This is in contrast to two earlier small studies of 45 and 50 patients with COPD [26, 28] reporting that plasma YKL-40 was related to disease stage, i.e. related to the degree of airway obstruction measured by FEV₁% predicted. The patients in these studies suffered from a lesser degree of obstruction and were younger than our patients, which may account for some of the difference.

The parameter FEV₁% predicted represents disease staging and does not provide information about how fast the disease developed to the current level of airway obstruction, an observation that corresponds well with the recent changes in the GOLD COPD disease staging guidelines (http://​www.​goldcopd.​org – accessed 17 April, 2013) in which symptoms and exacerbations also have an influence on disease staging. This may explain why we see this discrepancy in our cohort. YKL-40 may more accurately describe disease activity, whereas FEV₁% represents the current level of lung damage. This notion is supported by our findings: a high plasma YKL-40 in patients with severe and very severe COPD predicted a worse outcome independent of the traditional disease staging levels (Figure 3B-C). This could potentially signify that these patients had a disease characterized by a higher degree of inflammation and that plasma YKL-40 is able to assign patients with advanced disease to a high- and a low-risk group independent of disease severity.

Identification of biomarkers that can predict progression of COPD remains a high priority. Currently we only have an indirect measure of COPD disease progression through the use of spirometry. A biomarker level in a blood sample able to assess the inflammatory activity in patients with COPD would prove a valuable tool in monitoring disease activity, treatment efficacy, and prognosis of patients. This could help to identify patients characterized by high inflammatory activity who might benefit most from inflammatory modifying therapies.

In conclusion, our study supports the hypothesis that plasma YKL-40 is elevated in many patients suffering from advanced COPD. Patients with the highest plasma YKL-40 had the shortest OS. The exact functions of YKL-40 in disease progression of COPD remain unknown. Prospective, longitudinal studies of patients with COPD are needed in which plasma YKL-40 is determined several times during follow-up and related to clinical characteristics, e.g. loss of FEV₁. It would also be interesting to investigate whether high plasma YKL-40 is associated with an increased susceptibility to exacerbations of COPD. Increased inflammatory activity could potentially lead to exaggerated responses to inflammatory insults to the airways, and plasma YKL-40 may be a new prognostic biomarker in patients with COPD.