The Value of FENO Measurement for Predicting Treatment Response in Patients with Acute Exacerbation of Chronic Obstructive Pulmonary Disease

Background

Exacerbation is an important life-threatening event for patients with chronic obstructive pulmonary disease (COPD).^1,2 It can accelerate deterioration of the disease,^3,4 and lead to a heavy economic burden.⁵ It is reported that hospital expenditure accounts for roughly 45–50% of the total direct cost generated by COPD patients.⁶ Identifying biomarkers for predicting the prognosis of AECOPD patients is helpful for COPD management.

Fractional exhaled nitric oxide (FENO) measurement has been considered to be a surrogate for airway eosinophilic inflammation, which is widely used to predict the response to inhaled corticosteroids (ICS) in asthma.^7–10 As we all know, AECOPD shows obvious heterogeneity in airway inflammation: some cases are neutrophil dominated, whereas others are eosinophil dominated, mixed granulocytic or paucigranulocytic.¹¹ It has been shown that high eosinophil levels correlate with a better response to ICS in COPD.¹² As an alternative to the eosinophil level, it has been demonstrated that FENO is a good measurement for discriminating eosinophilic COPD.¹³ Nevertheless, the usefulness of FENO measurements in AECOPD patients in clinical practice is still unclear.

There is no gold standard for defining responders in clinical practice. In some studies, changes in FEV₁ are always monitored as the main outcome and used to assess the response to treatment in stable COPD.^14,15 An increase in FEV₁ over 200 mL is regarded as a significant improvement in health status, and these patients are usually defined as responders.¹⁶ Meanwhile, some studies use patient-reported health status to evaluate treatment response.¹⁷ The health status data are usually obtained after treatment, ranging from “much better” to “much worse”. The responders are considered to be those who report much better and slightly better. In this study, we use both methods to assess treatment response to determine the role of FENO measurement in predicting the overall therapeutic effect in patients with AECOPD.

Methods

This research was approved by the local Ethics Committee of the Second Xiangya Hospital of Central South University and First Affiliated People’s Hospital of Shaoyang College and was conducted in accordance with the Declaration of Helsinki and its amendments. All subjects gave written informed consent to participate in the study. The study was registered in the Chinese Clinical Trial Registry (ChiCTR-ROC-16,009,087; http://www.chictr.org.cn/).

Study Subjects

Patients with clinician-diagnosed AECOPD were recruited from the Second Xiangya Hospital of Central South University and First Affiliated People’s Hospital of Shaoyang College from September 2016 to December 2019. Patients with a primary diagnosis of AECOPD and aged over 40 years were recruited into the study. Those who had a history of asthma or other respiratory diseases (lung cancer, pulmonary thromboembolism, bronchiectasis, interstitial lung disease), severe heart failure (myocardial infarction, New York Heart Association stage IV) or malignant comorbidities were excluded from the study. The Global Initiative for Asthma (GINA) 2014 document was used to differentiate COPD from asthma and asthma-COPD overlap.¹⁸ In addition, patients with COPD exacerbation who had received systemic corticosteroids prior to hospitalization were also excluded. All diagnoses were established by the clinicians and independently verified by physicians specializing in respiratory medicine. The symptom scores at the second visit were used to classify patients into different groups (A, B, C, D) according to GOLD guideline.¹⁹ GOLD stages were established by the spirometry data obtained at the second visit. Exacerbation of COPD was defined as increased dyspnea, cough, or sputum (quality or quantity) that resulted in the subjects seeking medical care. Smokers were those who smoked tobacco every day and ex-smokers were those who stopped smoking at least 6 months prior to recruitment.

Study Design

Demographic data were collected at admission. FENO, lung function, blood gases, CAT, and mMRC were carried out at two time points: at hospital admission and on day 7, respectively. If the hospital stay was less than 7 days, the second round of measurements was performed at discharge. All measurements were obtained at the same period of the day. Treatment during hospitalization was determined by the clinicians based on the GOLD guidelines.¹⁹ In order to reduce the bias, almost all patients received similar treatment without violation of the GOLD document, all the patients received nebulized inhalation of ipratropium bromide and budesonide, but the type of antibiotic was determined by the clinicians. Treating clinicians were not directly involved in the study and were blinded to the results. Baseline measurements were performed before performing treatment for all patients. At the second visit, patient health status was divided into five groups based on a five-point Likert scale ranging from 1 to 5, representing “much better”, “slightly better”, “no change”, “slightly worse”, and “much worse”, respectively. Responders, based on patients-reported outcomes, were defined as those having a global rating of change in COPD since the last visit of “much better”, or “slightly better”; non-responders were defined as having a global rating of change in COPD since last visit of “no change,” “slightly worse” or “much worse”.^20,21 Treatment response, assessed by lung function, defined responders as those who presented with an increase in FEV₁ of more than 200 mL.¹⁶

Measurements

Measurement of FENO

In both hospitals, the FENO measurement was performed using a SUNVOU-CA2122 breath analyzer from SUNVOU company (Wuxi, China). SUNVOU-CA2122 has been widely used in China for several years. FENO levels measured by SUNVOU-CA2122 are closely related to that detected by NIOX VERO, which is used worldwide. To make sure the accuracy of FENO measurements and eliminate external interference, patients were asked to abstain from alcohol, coffee, chocolate, nitrate-rich foods (such as sausage, animal innards, lettuce, and spinach), strenuous exercise, tobacco and bronchodilators for 1 hour preceding the test. The testing process was conducted strictly in accordance with the guidelines recommended by the American Thoracic Society/European Respiratory Society and the manufacturer’s instructions. Briefly, the patients were asked to inhale deeply through the mouth to total lung capacity (TLC) and then to exhale through the mouthpiece guided by a visual animation at a constant flow rate of 50 mL/s. The required exhalation time was approximately 6 ∼ 10 s. The device displayed the results (in ppb) automatically once acceptable values were obtained.

Lung Function Tests

Lung function parameters were measured by a professional technician using an electronic spirometer (CHESTGRAPH HI-101), according to American Thoracic Society guidelines.

Measurement of Blood Gas Analysis

Blood gas parameters were determined using a blood gas analyzer (GEM Premier 3000), with 2 mL of arterial blood taken from the radial artery. Measurements were performed by a respiratory nurse on patients who had stopped inhalation of medical oxygen for 30 min.

Measurement of CAT and mMRC

These questionnaires were explained to the patients by clinicians and independently completed by the patients. The detailed protocol was discussed in our previously published paper.¹⁷

Statistical Analysis

Patient clinical characteristics were summarized descriptively. Continuous variables are presented as mean and standard deviation (if data were normally distributed) and median and interquartile range (IQR) values (if data were not normally distributed). Categorical variables were described as frequency rates and percentages. Means for continuous variables were compared by paired t-tests or ANOVA tests. Pearson, Spearman’s rank correlation coefficient and partial correlation method were used to analyze the relationship between different measurements. A receiver operating characteristic curve (ROC) was used to determine the cut-off point for predicting responders according to Youden index maximization principle. SPSS software version 25.0 and Medcalc were used for statistical analysis.

Results

Patient Demographic Characteristics

A flow chart of the study is shown in Figure 1. Initially, a total of 257 subjects with a primary diagnosis of AECOPD were screened in the study. Amongst these 257 subjects, 66 were excluded because of pneumonia (n = 28), lung cancer (n = 5), asthma (n = 15), interstitial lung disease (n = 2), bronchiectasis (n=4), severe heart failure (n = 7) or accepting systemic corticosteroids before admission (n = 5). During exacerbation, 9 patients failed to complete the second measurements of lung function. Among these 9 patients, 7 failed to complete the second lung function because of severe dyspnea and 2 dropped out due to unwillingness to perform the tests again. Thus, 182 patients were recruited to the final analysis. Demographic and clinical data are presented in Table 1. The mean age of the patients was 66.57 ± 12.09 years, with 92.90% males and 7.10% females. Among the included patients, 48 were current smokers and 134 were ex-smokers. Almost all patients received similar treatment (antibiotic + theophylline + nebulized inhalation of bronchodilators and corticosteroids) based on the Global Initiative of Chronic Obstructive Lung Disease (GOLD) document. Among 182 subjects, 13 were grouped into A and C, 169 were classified into B and D. Most of the patients were in the high symptoms group (CAT > 10, 97.26%; mMRC > 1, 97.80%) and were observed to be GOLD stage III or IV (58.79% and 29.12%, respectively). The mean hospital stay was 8.37 ± 2.66 days (Table 1).

Table 1 Demographic and Clinical Characteristics of Patients

Figure 1 Study flow chart. A total of 257 subjects with a primary diagnosis of AECOPD were screened; 66 patients were excluded because of pneumonia (n = 28), lung cancer (n = 5), asthma (n =15), bronchiectasis (n =4), interstitial lung disease (n = 2), severe heart failure (n = 7) and accepting systemic corticosteroids before admission (n = 5). During exacerbation, 9 patients failed to complete the second measurements of lung function. 182 subjects were recruited to the final analysis.

Changes in Main Parameters Between the Two Visits

All the main parameters improved significantly. The levels of FEV₁, FEV₁%, and PaO₂ improved from 0.93 ± 0.35 L to 1.07 ± 0.41L; 37.51 ± 14.72% to 43.14 ± 17.17%; and 66.05 ± 11.87 mmHg to 75.73 ± 10.95 mmHg, respectively. The CAT, mMRC and PaCO₂ scores at the second visit all decreased significantly when compared with the baseline at admission; Statistical differences were observed between the two visits for all the above parameters (Table 2).

Table 2 Changes in Main Parameters Between the Two Visits

The Correlation Between FENO Level and Health Status

The majority (80.22%, n = 146) of AECOPD patients reported an improved health status at the second visit. Of the 146 patients, 83 (45.60%) reported feeling much better and 63 (34.62%) reported feeling slightly better. Based on the definition of patient-reported non-responders, 36 patients were grouped into non-responders, with 16 patients reporting “no change”, 18 reporting feeling “slightly worse” and 2 reporting feeling “much worse”. Subjects who reported “much better” and “slightly better” presented with higher levels of FENO compared to those who reported “no change” and “slightly worse”; the FENO levels were 37.24 ± 16.25 ppb, 35.83 ± 20.39 ppb, 17.13 ± 8.54 ppb, and 13.61 ± 3.57 ppb, respectively (Figure 2A). A negative correlation was observed between FENO level at admission and health status (rho = −0.408, p < 0.001) (Figure 2B). The responders had a higher concentration of FENO than non-responders (36.63 ± 18.10 ppb vs 15.28 ± 6.55 ppb, p < 0.001) (Figure 2C). We did not observe significant correlation between FENO level obtained at admission and COPD severity classified by A, B, C, D groups. (rho = −0.04, p =0.60) (Figure 2D).

Figure 2 The correlation between FENO level determined at admission and health status. (A) Comparison of FENO level determined at admission among different patient-reported health statuses. (B) Correlation between FENO level obtained at admission and health status. (C) Comparison of FENO level between responders and non-responders. (D) Correlation between FENO level obtained at admission and disease severity classified by A, B, C, D groups.

The Correlation Between FENO Level and Other Lab Findings

The FENO level was positively correlated with the increase in FEV₁ and FEV₁% (r = 0.291, p < 0.001 and r = 0.208, p = 0.005, respectively) (Figure 3A and B). There was no relationship between the FENO level determined at admission and changes in PaO₂ or PaCO₂ (r = 0.07, p = 0.38 and r = 0.05, p = 0.49, respectively) (Figure 3C and D). Taking FEV₁ > 200mL as a criterion, only 57 responders were identified. The FENO concentration was higher in responders than in non-responders (41.86 ± 20.97 ppb vs 28.10 ± 15.59 ppb, p<0.001) (Figure 3E). We also analyzed the correlation between FENO values and blood eosinophil count determined at admission and found that they were positively correlated (r = 0.37, p < 0.001) (Figure 3F).

Figure 3 The correlation between FENO level and other lab findings. (A) Relationship between FENO level detected at admission and changes in FEV1. (B) Relationship between FENO level detected at admission and changes in FEV1%. (C) Correlation between FENO level measured at admission and changes in PaO2. (D) Correlation between FENO level measured at admission and changes in PaCO2. (E) Comparison of FENO level between responders and non-responders. (F) Association between FENO level measured at admission and blood eosinophil count.

Correlation Between FENO Level and Questionnaire Scores and Hospital Stay

We found that FENO concentration was inversely correlated with changes in the CAT score (r = −0.197, p = 0.008) (Figure 4A). However, no significant correlation was observed between FENO level at admission and changes in the mMRC score (r = −0.02, p = 0.57) (Figure 4B). There was a negative correlation between the length of hospital stay and FENO level at admission (r = −0.18, p = 0.02) (Figure 4C).

Figure 4 The correlation between FENO, questionnaire score and hospital stay. (A) Relationship between FENO level detected at admission and changes in CAT. (B) Relationship between FENO level detected at admission and changes in mMRC. (C) Association between FENO level measured at admission and length of hospital stay.

Cut-off Point of FENO to Differentiate Responders

The responders were assessed by both patient self-reported health status and lab findings (FEV₁). For patient-reported health status, the cut-off point of FENO to differentiate responders from non-responders was 18 ppb, with a sensitivity of 89.7% and a specificity of 88.9%, the area under the curve (AUC) was 0.922 (95% confidence interval was 0.873–0.956) (Figure 5A). Based on the criterion of an increase in FEV₁ > 200 mL, the cut-off point for FENO level was 30 ppb; sensitivity was 73.68% and specificity was 67.2%, the AUC was 0.711 (95% confidence interval: 0.640–0.776) (Figure 5B).

Figure 5 Cut-off point of FENO level measured at admission for predicting responders. (A) Cut-off point of FENO level at admission for predicting patient-reported health status identified responders. (B) Cut-off point of FENO level detected at admission for predicting responders identified by improvement in FEV1.

Discussion

In this longitudinal study, we assessed the value of measuring FENO at admission of AECOPD patients. The study demonstrated that FENO concentrations at admission were related to improvements in FEV₁ and patients’ self-reported outcome, suggesting that FENO may predict the response to therapy in patients with AECOPD.

FENO has been recognized as an alternative biomarker of eosinophil, and as asthma is a disease predominated by eosinophil inflammation,²² FENO has been widely used to predict treatment responses. In fact, FENO measurement is recommended to predict asthma severity and the response to Inhaled Corticosteroids (ICS).^8,23,24 Since COPD is a heterogeneous disease, some patients are also eosinophil predominated; thus, FENO measurement has also proven useful in COPD management.^25,26 Some studies have found that FENO levels are increased in AECOPD,^27,28 but the usefulness of FENO measurements and the clinical relevance of these findings remain limited.

In this study, we used both patient-reported health status and improvements in FEV₁ to assess treatment response to determine the role of FENO measurements in predicting the overall therapeutic effect in patients with AECOPD. As in a previous study,¹⁶ we were unable to detect any consistent relationship between FENO concentration and COPD severity. However, we found that FENO levels measured at admission positively correlated with improvements in FEV₁. Interestingly, we also demonstrated that FENO values were negatively related to patient-reported health status, suggesting that patients with a higher level of FENO may have a better prognosis. This may be explained by the good correlation between FENO and eosinophil,^29,30 which is also confirmed in this study. It has been shown that patients with higher eosinophil counts experience better clinical outcomes,³¹ and this is consistent with the findings in our study.

Patient-reported outcomes had a moderate correlation with improvements in FEV₁, suggesting that both measurements can provide prognostic information for physicians to better manage AECOPD patients. However, there were only 57 responders at the second visit when taking an increase in FEV₁ > 200 mL as a criterion. Given that the mean length of hospital stay was 8.37 ± 2.66 days, obviously, most of the patients were thought to be have met the discharge criteria at the second visit, indicating that an increase in FEV₁ > 200 mL is too strict for identifying responders. For some severe patients, it is impossible to get such a huge improvement; if clinicians take this standard into their clinical work, it will dramatically increase the mean hospital stay, economic cost, and risk of nosocomial infections. Conversely, the patient-reported health status identified 146 responders at the second visit, and this is more consistent with the real world in clinical practice. In our previous study, we demonstrated that changes in CAT could evaluate the evolution of health status in patients with AECOPD.¹⁷ We also found that FENO was correlated with changes in CAT; patients with a higher FENO level had a greater improvement in CAT score, indicating that FENO is a good biomarker for predicting the treatment response of AECOPD patients not only based on objective lab findings or subjective questionnaires. This study also demonstrated that patients with higher FENO levels at the onset of the exacerbation were discharged home earlier. This is reasonable, as a better functional response and a better health status is usually associated with a more rapid clinical recovery.

The ROC curve analysis was performed to determine the cut-off point of FENO to identify responders. The cut-off point varied according to different criteria for differentiating responders—it was 18.0 ppb for responders identified by patient-reported health status and 30.0 ppb for responders determined by FEV₁. In Antus’ study, the authors used an increase in FEV₁ over 200 mL as the criterion, and the cutoff point for optimum predictive accuracy was 26.8 ppb, roughly in agreement with the findings of our study. As discussed above, it seems that the evolution of patient-reported health status may be more in line with the real clinical situation and, as such, taking 18.0 ppb as cut-off point may be more useful in clinical practice.

We acknowledge that this study has some limitations. First, we did not have the FENO baseline data before exacerbation, and this would have provided a more meaningful assessment of the role of FENO in patients with AECOPD. However, compared with the FENO level of stable COPD published in other studies, it seems that FENO was increased at exacerbations. Second, as outcome variables, patient-reported health status and hospital stay are not only influenced by treatment response but also by a number of other factors, including the economic situations of the patients and the discretion of the physician. However, we found the changes in FEV₁ were closely associated with patient status, and despite the lack of standardized criteria for hospital discharge, we still observed a negative correlation between FENO levels and the length of hospital stay. Finally, there were some biases of the sample size among each group, especially for “much worse” group, but that is reasonable, most of the patients recovered with professional treatment, only a small portion of patients would be even worse. The total number of non-responders identified by patient-reported health status is 36, it fits the minimum number for performing t-test between responders and non-responders. To better investigate the role of FENO in patients with AECOPD, a multicentre study with large sample size is worthy to be performed in the future.

Conclusion

FENO levels determined at hospital admission are potential to predict the overall treatment response in AECOPD patients, as determined by both improvements in patient-reported health status and also functional changes (FEV₁).