Clinical utility of blood neutrophil-lymphocyte ratio in Japanese COPD patients

The overall design of the K-CCR has been previously published [20, 22, 23]. In brief, this study was a 3-year, prospective observational study that enrolled 572 men and women aged 40–91 years who had been diagnosed with COPD (n = 440) or as being at risk for COPD (n = 132) by pulmonary physicians from April 2010 to December 2012 [20, 22, 23]. For the purpose of this study, only data from the patients with spirometrically confirmed COPD (FEV1/forced vital capacity (FVC) <  0.7) who had undergone at least 3 blood examinations over 3 years and were not receiving any oral steroid treatments (n = 274) were selected and analyzed (Fig. 1). All patients were clinically stable and had no exacerbations and no acute infections for at least 1 month prior to recruitment [20, 22, 23]. The ethics committees of the Keio University and its affiliated hospitals approved the study protocol. Each patient provided written, informed consent to analyze and present their data. All aspects of the study conformed to the principles of the Declaration of Helsinki adopted by the 59th WMA General Assembly, Seoul, Republic of Korea, October 2008.

Complete medical and smoking histories, as well as information regarding the current pharmacologic treatments, were obtained on enrolment and annually thereafter [20, 22]. All patients were assessed by spirometry and chest computed tomography (CT) imaging. On the CT images, the extent of emphysema was quantified as the ratio of low attenuation area (LAA %) and the percentage of airway wall area (WA %) using a custom-made software (AZE Ltd., Tokyo, Japan) [22, 23]. Independent investigators judged the number and severity of exacerbations based on the reviews of physicians’ medical records, as previously reported [22, 24]. Mild COPD exacerbation was defined as worsening of symptoms that were self-managed (by measures such as an increase in salbutamol use) and resolved without systemic corticosteroids or antibiotics [24]. Moderate COPD exacerbation was defined by the requirement for treatment with systemic corticosteroids and/or antibiotics [24]. Severe COPD exacerbation was defined as one that required hospitalization, including an emergency admission for 24 h [24]. Comorbid diagnoses were established by clinical history and examination findings based on a review of available medical records [20, 21, 23, 24].

Blood samples were collected at baseline and annually thereafter, at a time when the disease was stable. The total numbers of white blood cells (WBCs), neutrophils, and lymphocytes were measured at each participating center. NLR was defined as the absolute neutrophil count divided by the absolute lymphocyte count [9]. Serum C-reactive protein (CRP) levels were measured using a Hitachi Ltd. LABOSPECT 008 devise. Serum amyloid A (SAA) is an acute-phase protein similar to CRP, and its levels were evaluated by latex agglutination immunoassay. The annual rates of change in NLR were measured over 3 years. Excel (Microsoft Inc., Redmond, WA, USA) was used to calculate the derived slope between each of the above measures and the date of data collection [22, 25].

All patients were clinically stable and had no exacerbations for at least 1 month before study enrollment and on the day of the annual examination [22]. The Japanese version of COPD assessment test (CAT) [26, 27] and the St. George’s Respiratory Questionnaire (SGRQ) in Japanese were used for the assessment of COPD-specific health status [28‐30]. All of the questionnaires were completed by the patients themselves at home [20].

Data are presented as the mean ± standard deviation (SD) or as median (interquartile range (IQR)). Data were compared between two groups using the student t-test, Mann-Whitney U test, and χ² test. Data were compared among three or more groups using Kruskal-Wallis test and Steel-Dwass test. Correlations between continuous variables were evaluated using the Pearson’s correlation coefficient. Univariate and multivariate logistic regression analyses were performed in order to assess the effects of factors on increasing NLR. Receiver operating characteristics (ROC) curves were constructed to assess the areas under the curves (AUCs). We investigated the optimal cutoff value by maximizing the Youden index. The data of NLR over 3 years were only used in the analysis of the annual rates of change in NLR, and in the comparison of the average of NLR among the three groups over 3 years. Differences in rates of change over time and in NLR among the 3 groups classified based on the severity of exacerbation were estimated using mixed-effects modeling with Bonferroni correction [22]. For all tests, two-sided p values < 0.05 were considered significant. Data were analyzed using the JMP 13 software (SAS Institute, Cary, NC, USA). A mixed-effect model was applied using SPSS 23 (IBM Corporation, Armonk, NY, USA).

Table 1 shows the baseline clinical characteristics of the study patients. The mean age of the 274 COPD patients was 72.2 ± 7.9 years, and 9.6% of them were current smokers. At baseline, 23.0%, 48.9%, 22.3%, and 5.8% were diagnosed as COPD grade 1, 2, 3, and 4, respectively based on the Global Initiative for Chronic Obstructive Lung Disease (GOLD) classification [31]. Baseline NLR significantly increased with the severity of COPD grade (1.9 (1.4–2.4) vs. 2.1 (1.5–2.6) vs. 2.3 (2.0–3.0) vs. 2.7 (1.9–5.3), p = 0.002) (Table 2).

NLR data showed a non-normal distribution. The median baseline NLR was 2.1(1.6–2.7). Baseline characteristics of patients according to the baseline NLR are shown in Table 3. The cut-off value of NLR was arbitrarily defined based on the upper quartile points as 2.7. COPD patients with NLR ≥ 2.7 (n = 66) were older (74.0 ± 7.1 years vs. 71.6 ± 8.1 years, p = 0.037), had a lower BMI (22.0 ± 3.4 vs. 23.2 ± 2.9, p = 0.005) and a lower % FEV₁ (55.8 ± 21.3% vs. 66.5 ± 20.4%, p = 0.0003) compared to patients with NLR < 2.7 (n = 199). Baseline NLR did not differ between current smokers and past smokers (2.12 ± 0.24 vs. 2.38 ± 0.08, p = 0.30). On CT images, LAA% of the patients with NLR ≥ 2.7 tended to be higher than that of the others (18.4 (6.4–28.7) % vs. 11.3 (5.0–23.3) %, p = 0.093), while WA% did not differ between the two groups (54.9 (46.4–59.8) % vs. 52.4 (47.4–57.7) %, p = 0.416). The COPD patients with NLR ≥ 2.7 exhibited a significantly higher total CAT score (13.9 ± 7.7 vs. 11.5 ± 8.0, p = 0.039) and SGRQ total score (36.2 ± 19.1 vs. 24.8 ± 18.0, p < 0.001) compared to the others.

Baseline NLR showed a weak but significant correlation with baseline CRP (r = 0.18, p = 0.003) and SAA (r = 0.34, p < 0.001). Similarly, baseline absolute neutrophil count also correlated significantly with baseline CRP (r = 0.24, p < 0.001) and baseline SAA (r = 0.31, p < 0.001), but the baseline absolute lymphocyte count did not show correlation with the other inflammatory markers. The absolute neutrophil count showed no correlation with the absolute lymphocyte count (r = 0.08, p = 0.214). These results imply that the absolute neutrophil count has a bigger impact on the value of NLR as systemic inflammation.

Univariate logistic regression analysis showed that high age, % FEV₁ < 50%, low BMI (BMI < 18.5), high CRP and high SAA were all significantly associated with NLR ≥ 2.7 (Additional file 1: Table S1). We then performed a multivariate logistic regression analysis including the associating factors that reached significance in the univariate analysis. Because of the significant correlations between SAA and CRP (r = 0.84, p < 0.001), SAA was excluded from the multivariate analysis. % FEV₁ < 50%, low BMI, and high CRP were found to be independent determinant factors of NLR ≥ 2.7 (Table 4). These results imply that NLR is multifactorial, and is independently associated with severe airflow limitation, advanced age, and low BMI. Also, it is reflective of other systemic inflammation markers.

Prevalence of comorbidity in the study patients according to the baseline NLR is shown in Table 5. The prevalence of arteriosclerosis obliterans, cancer and cataract were tended to be higher in the NLR ≥ 2.7 group than in the NLR < 2.7 group (arteriosclerosis obliterans; 0.5% vs. 3.0%, P = 0.096, caner; 18.4% vs. 28.7%, p = 0.072, cataract; 38.7% vs. 53.3%, p = 0.081). There were no statistical differences in the prevalence of cardiovascular disease, hypertension, diabetes. These results imply that NLR is not influenced by comorbidities in COPD patients.

Overall, time-dependent changes in NLR showed a slight increase over 3 years (p = 0.003) (Fig. 2a). The distribution of the annual rate of change in NLR (ΔNLR score/year) over 3 years is shown in Fig. 2b. The median ΔNLR score/year was 0.05 (− 0.15–0.25).

Based on the degree of exacerbations defined in our previous paper [24], the COPD subjects were classified into 3 groups: without exacerbations (n = 86), only mild exacerbations (n = 52), and moderate or severe exacerbations (n = 74). Among the 3 groups, the median baseline NLR was significantly higher in the moderate/severe exacerbation group compared to the exacerbation-free group but it was not significantly different when compared to the mild exacerbation group (2.4 (1.8–3.2) vs. 2.0 (1.5–2.4) and 2.0(1.4–2.7), p = 0.010 and 0.104, respectively) (Fig. 3a). Univariate logistic regression analysis showed that the baseline NLR ≥ 2.7 (OR; 2.89, p = 0.001), Age (OR; 1.05, p = 0.01), %FEV₁ < 50 (OR; 3.23, p = 0.0002) and LAA% (OR; 1.05, p = 0.0004) were significant predictors of moderate/severe exacerbation (Additional file 1: Table S2). Multivariate logistic regression analysis was performed using risk factors that reached significance on univariate analyses. Because of the significant negative correlations between %FEV₁ and LAA% (r = − 0.47, p < 0.0001), LAA% were excluded from multivariate analysis. The baseline NLR ≥ 2.7 was an independent risk factor for moderate exacerbations over 3 years (OR; 2.22, p = 0.025) (Additional file 1: Table S3). ROC curves showed that the optimal cutoff for the baseline NLR for predicting moderate/severe exacerbations was 2.7 (AUC 62.3, sensitivity 37.8%, specificity 83.3%) (Fig. 3b). Follow-up analysis over 3 years revealed that there was a significant difference in the average value of NLR among the three groups (p = 0.006), but the trends of NLR were comparable among the three groups over 3 years (p = 0.476) (Fig. 4).

The area under ROC for NLR to predict future exacerbation was 63.4% (95% CI 55.0–71.7) and the baseline low % FEV₁ (%FEV₁ < 50%) was 64.2% (95% CI 56.9–71.6), and both seemed to be better than the baseline absolute neutrophil and lymphocyte counts alone, CRP and SAA, implying that NLR was a superior predictor of the severity of airflow limitations and future exacerbations compared to the neutrophil and lymphocyte counts alone (Table 6).

Consistent with a previous report [14], in this study, we found that NLR was associated with COPD severity and exacerbations. Several studies have reported the appropriate thresholds to predict the natural history of systemic diseases other than COPD. In this study, we revealed the appropriate cut-off value of NLR as 2.7 to predict COPD severity and future exacerbations [32, 33]. The mechanisms underlying these relationships are unknown, but there could be several explanations. Firstly, it is well known that, even after smoking cessation, inflammation in the lungs continue, especially in patients with advanced COPD [34], suggesting that the persistent inflammatory response in the lungs could lead to neutrophil recruitment and activation [35]. When activated, neutrophils release a variety of serine- and metalloproteinases, which contribute to the development and progression of emphysema [36]. It has been shown that sputum neutrophilia is increased in advanced COPD and is associated with severity of airflow limitation [37]. Blood neutrophilia is a hallmark of current smokers [38] and is also a predictor of mortality in COPD patients [4]. Secondly, a relationship between bacterial colonization and exacerbations is increasingly recognized [39]. Thus, in some patients with COPD, the disturbed flora may continue to activate the innate immune responses, perpetuating lung inflammation and blood neutrophilia. We would like to emphasize that most of the participants in this study were not current smokers. Thirdly, blood lymphopenia is associated with age [40] and poor nutritional status [41], which could also be applicable to a subset of COPD patients. Taken together, NLR along with age and nutritional status could be reflective of the systemic inflammatory condition. Weak correlations between NLR and other biomarkers might be caused by multifactorial determinants of NLR.

It should be noted that this is the first study to assess the relationships between NLR and COPD comorbidities. Chronic systemic inflammation which is a characteristic of most comorbidities is believed to be a key factor that links COPD with its comorbidities [42]. In the ECLIPSE cohort, most COPD patients with heart disease had elevated IL-6, IL-8, and fibrinogen, while those with hypertension had elevated fibrinogen and those with diabetes had elevated CRP [19]. In our study, these comorbidities were not associated with NLR. It would, therefore, be unreliable to predict the presence of comorbidities by the serum inflammatory markers in COPD patients.

NLR is a new addition to the list of inflammatory markers and has got special attention in the recent years. The value for NLR is obtained by dividing the absolute neutrophil count by the absolute lymphocyte count in peripheral blood samples. Moreover, this test is quite inexpensive, usually not requiring a proper setup and can be used as a screening and drug monitoring tool in population on a large scale [43].

Strategies to prevent exacerbations include targeting risk factors, addressing comorbid conditions, and giving bronchodilator therapies which include long-acting β2-agonists (LABAs) or long-acting muscarinic antagonists (LAMAs) used alone or in combination with each other or with an inhaled corticosteroid [ICS]) [44, 45]. A number of studies have shown that the higher the blood eosinophil count, higher is the risk of exacerbation [46] and greater is the exacerbation reduction response to inhaled corticosteroids [47, 48]. In this study, although NLR was a predictor of exacerbations, blood eosinophilia was not a risk factor. This may be because the prescription rate of ICS was high in our populations than in the other cohort studies. One previous report has shown that the count and percent of blood eosinophils were not associated with the exacerbation frequency in COPD patients after carefully excluding the Asthma–COPD overlap syndrome (ACOS) patients [49]. ICS may be insufficient to prevent exacerbations in COPD patients with a high NLR, and a new prophylactic therapy should be considered. NLR may be a simple and inexpensive biomarker for detecting a neutrophilic endotype, and this endotype may be indicative of the new pharmacological strategy for treating the exacerbation. In addition to the usual care, azithromycin and roflumilast are likely candidates for prophylactic therapy in these patients. Azithromycin and roflumilast treated patients had lower sputum proline-glycine-proline (PGP) levels and showed a reduction in other markers of neutrophilic inflammation [50, 51].

Our study had several limitations. First, although NLR was associated with exacerbations over 3 years, outcomes such as hospital admissions, emergency room visits, and mortality remain to be evaluated. Sørensen et al. reported that NLR was associated with higher mortality in moderate to very severe COPD patients [52], but it is unknown whether this result can be applied to milder COPD patients in our study. Second, blood NLR may not accurately reflect the lung inflammation of COPD patients. Low sensitivity of NLR to predict future exacerbations and airflow limitation severity may be caused by this reason. It would be preferable if we could compare blood NLR and biomarkers or cell fractionation of bronchoalveolar lavage and sputum. Third, COPD patients of the KCCR cohort study had few women (6.2%) and were older compared to subjects of other clinical studies in Western countries. Therefore, the results may not apply to women and younger COPD patients.