Fibrinogen is a promising biomarker for chronic obstructive pulmonary disease: evidence from a meta-analysis

Chronic obstructive pulmonary disease (COPD) is an escalating public health problem that affected more than 174.5 million adults in 2015 worldwide [1]. At present, COPD ranks as the third leading cause of mortality [2], and is responsible for approximately 3 million deaths annually [2,3]. Given that COPD can impair the quality of life by causing progressive reduction in lung function, loss of exercise capacity, increase in hospital admissions, and premature mortality [4]. COPD prevention represents an important public health goal. Because COPD is a progressive lung disease, the identification of biomarkers for early detection of COPD may help to improve future respiratory health.

It is widely recognized that systemic inflammation is an important clinical feature of COPD [5,6]. Much attention has been focused on the potential implication of circulating inflammatory biomarkers in the development of COPD. One of the most widely evaluated inflammatory biomarkers is fibrinogen, a key modulator of inflammation and fibrosis development, as well as tissue injury [7]. The association between circulating fibrinogen and COPD risk has been investigated by a large number of studies, with inconsistent and inconclusive findings. For example, some researchers have reported a significantly higher concentration of circulating fibrinogen in COPD patients than healthy controls [8–10], whereas others found that circulating fibrinogen concentration was comparable between the two groups [11], and even a significantly higher concentration in controls [12,13]. In a recent umbrella review of meta-analyses, Bellou and colleagues synthesized observational data on environmental factors and biomarkers in possible association with COPD, and found that circulating fibrinogen was a promising clinical marker of COPD, yet with very large heterogeneity [14]. Despite the significant heterogeneity, plasma fibrinogen this year has been qualified as a COPD biomarker for severity assessment in the United States by the Food and Drug Administration (FDA) [15]. Kirkpatrick and Dransfield have written an excellent review, and demonstrated that race may influence COPD susceptibility and progression [16]. It is hence reasonable to speculate that such very high heterogeneity is likely attributed to racial differences. Besides, other possible reasons such as different study designs and individually underpowered studies also account for this issue. However, the reasons behind inconsistence and heterogeneity when assessing the relationship between circulating fibrinogen and COPD thus far remains unexplored.

To fill this gap in our knowledge and generate more information for future studies, we prepared a comprehensive meta-analysis of published observational studies to test the hypothesis that circulating fibrinogen is a promising biomarker for COPD, and if this hypothesis is confirmed, we further examined whether circulating fibrinogen is associated with the severity of COPD in a concentration-dependent fashion.

We conducted this meta-analysis of observational studies in compliance with the requirements of the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) statement [17]. The PRISMA checklist is provided in Supplementary Table S1.

A systematic electronic search of PubMed, EMBASE (Excerpt Medica Database), Cochrane Central Register of Controlled Trials, and Web of Science was conducted from inception to 25 June 2019 for publications that assessed the relationship between circulating fibrinogen and COPD. The key terms used for search included ‘fibrinogen’, ‘FIB’, ‘chronic obstructive pulmonary disease’, and ‘COPD’.

Literature search was restricted to articles published in the English language from peer-reviewed journals. Additional articles were obtained via manually scanning the reference lists of relevant reviews and major original articles. Literature search was independently completed by two authors (B.Z. and W.N.), and divergences were resolved by discussion until a consensus was reached.

Articles were included if they were observational case–control studies, and if they had mean or median values of circulating fibrinogen concentration in both COPD patients and controls, along with standard deviation or standard error or interquartile range or whole range. Articles were excluded if they lacked control groups, or if they published in form of reviews, case reports, case series, conference abstracts, letter to the editor, comments, or editorials.

Two authors (B.Z. and W.N.) independently screened the titles and abstracts of all retrieved articles, and if necessary the full texts to assess their eligibility. If more than one article was published from the same or part of study participants, the article with the largest sample size was retained in the analysis. Any disagreement was discussed, and when necessary, adjudicated by a third author (Z.Z.).

Data from qualified articles were independently extracted by two authors (B.Z. and W.N.), including surname of first author, year of publication, country where the study was conducted, sample size, study type, COPD and its stages, circulating concentrations of fibrinogen in both patients and controls, and necessary baseline characteristics, if available. Extracted data from the two authors were checked for consistency using κ statistic, and any divergences were resolved by a third author (Z.Z.).

The quality assessment tool, the 9-star Newcastle–Ottawa Scale, was employed [18]. This scale ranges from 0 (the worst) to 9 stars (the best). The following three major study components were judged: selection (0–4 stars), comparability (0–2 stars), and exposure/outcome (0–3 stars).

The STATA software Release 14.1 (Stata Corp, College Station, TX) was used for statistical analyses. The difference in circulating fibrinogen concentration between COPD patients and healthy controls is expressed as weighted mean difference (WMD) and 95% confidence interval (95% CI) under the random-effects model using the DerSimonian and Laird method [19]. Heterogeneity is judged by the χ² test and quantified by the inconsistency index (I²) statistic, which ranges from 0 to 100%. Heterogeneity is statistically significant if the probability of χ² test is less than 10% or I² is over 50%. To explore potential sources of heterogeneity, subgroup analyses were conducted according to COPD course (stable and acute exacerbations), COPD severity (Global Initiative for Obstructive Lung Disease (GOLD) I, GOLD II, GOLD III, and GOLD IV), smoking habit in controls, region, sample size and study type, and meta-regression analyses were further conducted to assess the confounding impact of age and gender.

Cumulative analyses were done to assess the impact of first published study on subsequent studies and evolution of accumulated estimates over time. Influential analyses were used to assess the contribution of single study to overall estimate.

Publication bias was evaluated by the Begg’s funnel plots and the Egger’s tests at a significance level of 10% [20]. The trim and fill method was used to predict the number of potentially missing studies and derive the ‘unbiased’ estimates.

In total, 298 potentially relevant articles were identified after literature search for observational studies on circulating fibrinogen and COPD. On the basis of titles and abstracts, 187 articles were excluded with obvious reasons, leaving 111 articles for further evaluation in full texts. Finally, there were 45 articles in this meta-analysis [8–13,21–59], involving 5586 COPD patients and 18604 controls, and all articles were published from the year 1997 to 2019. Because two articles respectively provided data in COPD patients as a whole [46] and by COPD severity (stable and acute exacerbations) [9] based on the same study participants, we combined results from the two articles as a single study. So, a total of 44 studies were synthesized in this meta-analysis. The specific reasons for exclusion during article selection are provided in Figure 1.

Shown in Table 1 are the characteristics of all qualified studies. There were 14 studies enrolling patients with stable COPD, 11 studies with COPD exacerbations, and 9 studies with different COPD stages according to the GOLD guidelines. Eleven studies enrolled smokers as controls, and ten studies enrolled non-smokers as controls. The mean or median age of study participants ranged from 55 to 75 years. The sample size ranged from 16 to 1755.

Using the 9-star Newcastle–Ottawa Scale system, the total score of qualified case–control studies ranged from 6 to 9 stars (mean: 7.71; standard deviation: 0.76) (Supplementary Table S2).

Pooling the results of 44 qualified studies observed a significantly higher concentration of circulating fibrinogen in COPD patients than in controls (WMD: 84.67 mg/dl, 95% CI: 64.24–105.10, P<0.001), and there was strong evidence of between-study heterogeneity (I²: 97.2%, P<0.001) (Figure 2).

In view of the strong evidence of heterogeneity in overall analyses, it is necessary to explore possible causes by grouping studies according to COPD course, COPD severity, smoking habit in controls, sample size, and region, respectively.

By COPD course (Figure 3A), the degree of increased circulating fibrinogen in patients with acute exacerbations of COPD (AECOPD) relative to controls (WMD: 182.59 mg/dl, 95% CI: 115.93–249.25, P<0.001, I²: 94.5%) tripled when compared with patients with stable COPD (WMD: 56.12 mg/dl, 95% CI: 34.56–77.67, P<0.001, I²: 79.9%). By COPD severity (Figure 3B), there was a graded increase in circulating fibrinogen concentration with the increased severity of COPD relative to respective controls (for GOLD I, WMD: 13.91 mg/dl, 95% CI: 7.70–20.11, P=0.014, I²: 62.3%; for GOLD II, WMD: 29.19 mg/dl, 95% CI: 17.43–40.94, P<0.001, I²: 84.4%; for GOLD III, WMD: 56.81 mg/dl, 95% CI: 39.20–74.41, P<0.001, I²: 74.6%; for GOLD IV, WMD: 197.42 mg/dl, 95% CI: −7.88 to 402.73, P<0.001, I²: 97.1%).

By smoking habit in controls (Figure 4A), the increase in circulating fibrinogen concentration was significantly greater in COPD patients relative to non-smoking controls (WMD: 77.07 mg/dl, 95% CI: 48.30–105.83, P<0.001, I²: 89.3%) than smoking controls (WMD: 22.26 mg/dl, 95% CI: −0.24 to 44.77, P<0.001, I²: 87.9%). By sample size, the difference in circulating fibrinogen concentration between COPD patients and controls was larger in studies with sample size <50 (the median value) (WMD: 116.78 mg/dl, 95% CI: 73.84–159.72, P<0.001, I²: 94.1%) than in studies with sample size ≥ 50 (WMD: 62.31 mg/dl, 95% CI: 37.49–87.13, P<0.001, I²: 98.0%) (Figure 4B).

Subgroup analyses by region are presented in Supplementary Figure S1.

Other sources of heterogeneity were further explored through meta-regression analyses by modeling age, female percentage, and pack-year of COPD patients, and none of these factors reached statistical significance (all P>0.05) (Supplementary Figure S2).

Cumulative analyses revealed no significant impact from first published study on subsequently studies in overall analyses (Supplementary Figure S3). In influential analyses, the impact of any single study on overall effect-size estimates was non-significant (Supplementary Figure S4).

Shown in Figure 5 are the Begg’s funnel plot and filled funnel plot in overall analyses. The Begg’s funnel plot seemed apparently symmetrical, and the Egger’s test indicated a low probability of publication bias (P=0.103). As reflected by filled funnel plot, there was an estimated ten missing studies required to make the Begg’s funnel plot symmetrical. After taking these ten missing studies into account, COPD patients still had a statistically significant higher concentration of circulating fibrinogen than controls (WMD: 46.30 mg/dl, 95% CI: 21.85–70.75, P<0.001).

The aim of this meta-analysis was to examine the association of circulating fibrinogen with COPD and its severity by pooling the results of 45 published articles. The key finding of this meta-analysis suggests that a graded, concentration-dependent, significant relation exists between higher circulating fibrinogen, and more severity of COPD. To the best of our knowledge, this is thus far the first meta-analysis that has evaluated the concentration-dependent association of circulating fibrinogen with COPD severity in the literature.

In human body, fibrinogen is mainly synthesized by the liver, and it is converted into fibrin by thrombin during blood coagulation [60]. Fibrinogen is a major acute-phase reactant, and its synthesis is up-regulated in response to inflammation [61], a major clinical feature of COPD. There is evidence that fibrinogen is implicated in clot formation and is associated with advanced COPD [11]. In addition, elevated fibrinogen in circulation was found to be associated with an increased risk of acute exacerbations in COPD [62]. Furthermore, as indicated by the nationally representative NHANES III data, impaired lung function is a correlate of fibrinogen concentration and the presence of high fibrinogen concentration increases mortality risk both in the overall population and among subjects with COPD [63]. It is hence reasonable to speculate that circulating fibrinogen is a promising clinical biomarker in predicting the risk and severity of COPD. Our findings based on a meta-analysis of 45 studies reinforced this speculation by showing that there is a graded, concentration-dependent, significant relation between higher concentration of circulating fibrinogen and more severity of COPD. It is also worth noting that statistical significance retained when analysis was restricted to studies with large sample sizes and when including theoretically missing studies as estimated by the fill and trim method to control publication bias, indicating the robustness of our meta-analytical findings.

Another important finding is that there is a possible interaction between circulating fibrinogen and cigarette smoking in our subgroup analyses by smoking habit in controls, as we interestingly noticed that the difference in circulating fibrinogen concentration was more obvious when compared with non-smoking controls than with smoking controls. Cigarette smoking is an established risk factor for the development of COPD, and it is of added interest to know whether the contribution of cigarette smoking to COPD is mediated by elevated fibrinogen in circulation, which cannot be reliably investigated in the present meta-analysis due to the unavailable individual participant data. We agree further explorations on the interaction or medication impact of cigarette smoking on circulating fibrinogen are needed.

The findings of this meta-analysis are susceptible to several possible limitations. First, language selection bias is possible, as we restricted literature search to the articles written in the English language. As estimated by McAuley and colleagues, the exclusion of gray literature from a meta-analysis may result in an overestimate of an association impact by an average of 12% [64]. Second, the majority of included studies in this meta-analysis are cross-sectional in design, which might yield recall bias and preclude comments on causality. Third, in our overall and subgroup analyses, the majority of comparisons were obsessed by moderate to high degree of between-study heterogeneity, which made it difficult to draw firm conclusions and required further explorations of other possible sources for heterogeneity. Fourth, nearly all qualified studies in this meta-analysis had circulating fibrinogen concentration measured only once, and did not evaluate its long-term change in the development of COPD. Fifth, the methods to assay circulating fibrinogen are not identical across studies, which might yield measurement bias. Thereby, the jury must refrain from drawing a conclusion until future large-scale, longitudinal, well-performed studies to confirm or refuse our findings.

Despite these limitations, our findings suggest a graded, concentration-dependent, significant relation between higher circulating fibrinogen and more severity of COPD. For practical reasons, circulating fibrinogen can be proposed as a promising biomarker and an early warning sign of the susceptibility to develop COPD in the future, as issued by the FDA [15], as well as a robust predictor for the severity of COPD. Importantly, circulating fibrinogen is an easy-to-assay biomarker and can be proposed as a more practical approach toward clinical translation applications.

The authors declare that there are no competing interests associated with the manuscript.

The authors declare that there are no sources of funding to be acknowledged.

Conceived and designed the experiments: T.Y., Z.Z., W.N. Performed the experiments: B.Z., W.N. Analyzed the data: B.Z., T.Y., W.N. Contributed materials/analysis tools: S.L., D.H., K.W., Y.W., T.Y., Q.Z., Z.Z. Wrote and revised the paper: B.Z., Q.Z., W.N. All authors read and approved the final manuscript prior to submission.

AECOPD

acute exacerbations of chronic obstructive pulmonary disease

COPD

chronic obstructive pulmonary disease

FDA

Food and Drug Administration

GOLD

Global Initiative for Obstructive Lung Disease

PRISMA

Preferred Reporting Items for Systematic reviews and Meta-Analyses

WMD

weighted mean difference

95% CI

confidence interval