Background

Chronic obstructive pulmonary disease (COPD) is a chronic airway disease defined by persistent respiratory symptoms and irreversible airflow limitation [1,2,3]. Patients with COPD present with various symptoms, such as cough, sputum, and dyspnea, and these symptoms are closely related to the quality of life and prognosis [4, 5]. The global initiatives for chronic obstructive lung disease (GOLD) reports emphasize treatment based on patient history and symptoms, such as exacerbation history, the modified medical research council dyspnea scale (mMRC), and COPD assessment test (CAT) [6]. Forced expiratory volume in one second (FEV1) is still used to grade the severity of airflow obstruction, but the 'refined ABCD assessment tool' excludes FEV1 from the criteria for evaluating the 'ABCD' group. This is because the FEV1 value is weakly correlated with the patient's symptoms and health status [7, 8]. However, pulmonary function tests (PFT) are still important tests for diagnosing and treating COPD in the clinical field. Therefore, we want other PFT factors related to the patient's symptoms and health status rather than FEV1. Several studies have shown that the diffusing capacity of the lung for carbon monoxide (DLCO) among the various values of PFT is closely related to patient symptoms, prognosis, and oxygen demand in COPD [9, 10]. In addition, there was a recent opinion that DLCO should be added to multidimensional tools assessing COPD [11]. This study aimed to compare FEV1 and DLCO through the prognosis of severe acute exacerbations of COPD (AECOPD).

Method

Study population

We retrospectively analyzed the medical records of 342 patients admitted to Korea University Guro Hospital from January 2011 to May 2017. We searched our electronic medical records database with the keywords “COPD” and “Acute exacerbation.” This study was approved by the Institutional Review Board of Korea University Guro Hospital (KUGH16131-002). The requirement for informed consent from the patients was waived due to the retrospective nature of this study by the institutional review committee.

All patients included only patients who were followed up for more than 1 year in our hospital under the diagnosis of COPD. COPD and airflow limitation were diagnosed by synthesizing patient-reported respiratory symptoms, PFT (the ratio of FEV1 to forced vital capacity (FVC) was less than 70% in post-bronchodilator spirometry), chest image, and patient`s history (smokers with at least ten pack-years of tobacco exposure, etc.) by an experienced pulmonologist [6]. AECOPD was defined as worsening of the patient’s respiratory symptoms beyond normal day-to-day variation. Severe AECOPD was defined as ‘if the patient needs hospitalization due to AECOPD.’ The spirometry data used in the analysis was previously performed in the outpatient clinic during the stable period. Spirometry value that was measured within 1 year from the hospitalization day were used. Patients were excluded with the following criteria: (1) the cause of admission was not AECOPD; for example, acute heart failure, acute pulmonary edema, acute pulmonary embolism, pneumothorax, and arrhythmia (These diseases were excluded through cardiac enzyme, electrocardiogram, echocardiogram and chest image.), (2) the patient had undergoing active cancer treatment, (3) the patient received a major operation within 3 months, (4) the patient had an acute coronary syndrome, brain hemorrhage, or brain infarction within 3 months, (5) the patient had previously been diagnosed with asthma, and (6) the patient had no DLCO results. All patients were 40 years old or older. We retrospectively analyzed the charts by two experienced pulmonologists to exclude various exclusion factors. "events" is synonymous with "patients" in this study.

We classified 342 severe AECOPD patients by severity of DLCO and FEV1 (≤ vs. > 50% predicted). When the DLCO value is more than 50 (% of predicted value), it is defined as the 'DLCO normal group' and when it is 50 (% of predicted value) or less, it is defined as the 'DLCO impaired group' [11]. Likewise, when the FEV1 value is more than 50 (% of predicted value), it is defined as the ‘FEV1 normal group' and when it is 50 (% of predicted value) or less, it is defined as the ‘FEV1 impaired group' (Fig. 1).

Fig. 1
figure 1

Study design

Full size image

Data collection

We tested the association of FEV1 and DLCO with the following outcomes: in-hospital mortality, need for mechanical ventilation, need for intensive care unit (ICU) care. When the patient was hospitalized more than once, only the first hospitalized events were included, and the others were excluded. The following medical data were analyzed: age, sex, smoking history, comorbidities, baseline spirometry, inhaler and oral medication before admission, length of hospital stay, hospital mortality, experience of mechanical ventilation, and experience of ICU care in hospital.

Statistical analysis

Data were analyzed using SPSS 20 software (SPSS for Windows, SPSS Inc., Chicago, IL, USA). Data are presented as average ± standard deviation or number (percentage). We performed a statistical analysis in two directions. First, two groups were classified using DLCO and FEV1 and analyzed statistically. Continuous variables were compared using the independent t-test, and categorical variables were compared using the chi-squared test. We analyzed the prognostic factors (except length of hospital stay) by multivariate analysis through logistic regression. Multivariate analysis was conducted for variables with a P value of less than 0.05 in the univariate analysis, except for baseline spirometry (DLCO and FEV1). In the case of DLCO, multivariate analysis included sex, previous TB history, cerebrovascular accident, inhaler use before admission, oral β2 adrenoreceptor agonist, roflumilast, and mucolytic agent. In the case of FEV1, multivariate analysis included age, sex, previous TB history, inhaler use before admission, roflumilast, and mucolytic agent. Multivariate analysis was conducted using a backward elimination procedure and was assessed by the Hosmer–Lemeshow test.

Second, the linear correlation between spirometry factors (DLCO and FEV1) and length of hospital stay were analyzed. In univariate analysis, the correlation coefficients between spirometry factors and length of hospital stay were analyzed using the Pearson correlation analysis. In addition, we performed a multivariate linear regression analysis that included variables with a P value of less than 0.05 in the univariate analysis, except baseline spirometry. In addition, multivariate linear regression analysis was conducted using a backward elimination procedure. In the multivariate analysis, B was the regression coefficient, and a negative sign of the regression coefficient meant that the variables were negatively associated.

Third, we used receiver operating characteristic (ROC) curve analysis to predict the sensitivity and specificity of DLCO, FEV1 and DLCO + FEV1 as prognostic markers in severe AECOPD. When analyzing the ROC curve, DLCO, FEV1 and DLCO + FEV1 were analyzed as continuous variables. A P value of less than 0.05 was considered statistically significant.

Results

Characteristics of studied subjects

Among the 342 events, the DLCO normal group comprised 227 events (the DLCO value was more than 50% of the predicted value), and 115 in the DLCO impaired group. In the FEV1 normal group (the FEV1 value was more than 50% of the predicted value), there was 173 events, and the FEV1 impaired group had 169 events. The average age was 71.5 ± 9.2 years. A total of 238 (69.6%) events were male and 104 (30.4%) were female. Sixty-three (18.4%) events were current smokers and the average pack/year history was 41.3 ± 17.1 years. A total of 225 (65.38) events were using inhalers, and 165 (48.2%) were taking respiratory-related oral medications. Averaged FEV1 was 1.3 ± 0.5 L (54.0 ± 19.3%) and DLCO was 10.6 ± 4.8 L (59.3 ± 21.4%). (Table 1) In both groups, the average length of hospital stay was 10.0 ± 5.1 days. The mortality rate was 11 (3.2%), the experience of ventilator care was 29 (8.5%), and the experience of ICU care was 39 (11.4%).

Table 1 Baseline characteristics of patients with AECOPD
Full size table

Prognostic factor analysis classified using DLCO and FEV1

When classified through DLCO, the DLCO impaired group showed a poor prognosis in all four factors by univariate analysis (Fig. 2). When classified through FEV1, the FEV1 impaired group showed a poor prognosis in three factors by univariate analysis (Fig. 3). However, there was no statistically significant mortality rate when classified as FEV1 (P value = 0.116) (Fig. 3B).

Fig. 2
figure 2

Prognosis analysis for severe AECOPD according to DLCO classification. a Length of hospital stay (days), b mortality in hospital, c mechanical ventilation, and d intensive care unit. AECOPD, acute exacerbations of chronic obstructive pulmonary disease; DLCO, diffusing capacity of the lung for carbon monoxide

Full size image
Fig. 3
figure 3

Prognosis analysis for severe AECOPD according to FEV1 classification. a Length of hospital stay (days), b mortality in hospital, c mechanical ventilation, and d intensive care unit. AECOPD, acute exacerbations of chronic obstructive pulmonary disease; FEV1, forced expiratory volume in one second

Full size image

In multivariate analyses, DLCO was associated with mortality (odds ratio = 4.408; 95% CI 1.070–18.167; P = 0.040) and need for mechanical ventilation (odds ratio = 2.855; 95% CI 1.216–6.704; P = 0.016) and ICU care (odds ratios = 2.685; 95% CI 1.290–5.590; P = 0.008). In severe AECOPD, DLCO has been shown to predict mortality rate, ventilator, and ICU possibilities. When classified as FEV1, the experience of mechanical ventilation and ICU showed statistical significance. However, there was no significant difference in mortality rate (P = 0.075) (Table 2).

Table 2 Prognosis analysis for severe AECOPD
Full size table

Correlation analysis between spirometer factors and length of hospital stay

The length of hospital stay of the DLCO normal group was 7.3 ± 5.0 days and the DLCO impaired group was 12.4 ± 13.2 days. The length of hospital stay of the FEV1 normal group was 7.7 ± 5.4 days and the FEV1 impaired group was 10.4 ± 11.4 days. In the Pearson correlation analysis, both DLCO and FEV1 showed a negative correlation. In multivariate linear regression analyses, DLCO (B = − 0.542 ± 0.121, P < 0.001) and FEV1 (B = − 0.106 ± 0.106, P = 0.006) were negatively associated with length of hospital stay. Additionally, the regression coefficient was more pronounced in the DLCO analysis (Table 3).

Table 3 Correlation analysis of length of hospital stay
Full size table

ROC curve analysis of DLCO and FEV1

When analyzing the sensitivity and specificity using the ROC curve, DLCO showed better predictive ability than FEV1 (Table 4). When analyzing three prognostic factors (mortality in hospital, mechanical ventilation, and ICU care) through ROC curve analysis, area under the curve (AUC) was greater than 0.68 in all cases of DLCO (Fig. 4). In contrast, the AUCs of FEV1 were below 0.68 in all three prognostic factors. In addition, the sensitivity and specificity of DLCO were more than 64.1%, which was generally higher than FEV1. DLCO + FEV1 showed similar values to DLCO.

Table 4 ROC curve analysis of DLCO, FEV1, and DLCO + FEV1
Full size table
Fig. 4
figure 4

ROC curve of DLCO, FEV1, and DLCO + FEV1. a Mortality in hospital, b mechanical ventilation, and c intensive care unit. ROC, receiver operating characteristics; FEV1, forced expiratory volume in one second; DLCO, diffusing capacity of the lung for carbon monoxide

Full size image

Discussion

This is the study to compare FEV1 and DLCO as prognostic markers in severe patients with AECOPD in Korea. In our study, the factors of prognosis were defined as the length of hospital stay, mortality rate in the hospital, experience of ventilation, and experience of ICU care. Classification by DLCO showed significant differences in all prognostic factors. However, classification by FEV1 did not show a statistically significant mortality rate. The number of deaths was small, so caution is needed in the interpretation about death (the 95% confidence interval of the odds ratio was large and the P value was marginal). In the correlation analysis, both DLCO and FEV1 showed a negative correlation with the length of hospital stay. The correlation coefficient was more pronounced in the DLCO classification. In addition, when analyzing the ROC curve, DLCO showed better predictive ability than FEV1. Of course, some odds ratio values were better when classified as FEV1 in our study. However, DLCO was better in various analysis methods (correlation analysis, ROC curve analysis), which was likely to be as good as or better than FEV1.

The PFT has various parameters. In general, we used FEV1 to grade COPD and select the inhaler. In addition to FEV1, DLCO is an important prognostic factor. In a study of smokers who did not show an obstruction pattern in PFT, a low DLCO group showed quickly decreased pulmonary function and COPD progression [12]. Studies have shown that DLCO is a more accurate prognostic factor than FEV1 when assessing postoperative risk [13, 14]. In addition, DLCO is known to accurately represent the actual emphysema level and performance status [15, 16]. These results suggest that DLCO can be a good predictor of early pulmonary dysfunction and prognosis.

If we know the prognosis of the patient early, we can focus on high-risk patients and improve the prognosis. The prognostic factors that can be used in the clinic are laboratory findings, scoring systems such as CAT or mMRC, and baseline spirometry [17, 18]. In some studies, high-C-reactive protein, eosinopenia, and thrombocytopenia are associated with poor outcomes in AECOPD [19,20,21]. Although various scoring systems—such as St. George's Respiratory Questionnaire, mMRC, and CAT, are useful—patients with severe symptoms may not be graded or might have similar scores, making them difficult to use. Instead, we focused on baseline spirometry and confirmed that DLCO is more accurate in evaluating the prognosis of hospitalized patients than FEV1. If a grading system that considers both DLCO and FEV1 is developed, the prognosis can be predicted more accurately.

Our study was limited because it was a retrospective single-center study. We were unable to analyze including important prognostic factors such as frequent exacerbations, obstructive sleep apnea, and body mass index. As this study is a retrospective study, data on these factors were not available or inaccurate. To compensate for this, we carefully analyzed the charts by two experienced pulmonologists. Also, we included as many factors as possible in baseline characteristics and multivariate analysis. In addition, the treatment received during the hospitalization period and the prognosis after discharge were not evaluated. Large prospective clinical studies that include information on treatment during hospitalization and post discharge may be required.

Conclusion

DLCO was likely to be as good as or better as a prognostic marker than FEV1 in severe AECOPD. Accurate classification using DLCO may help to treat severe ACEOPD patients.