Diaphragm Ultrasound Distinguishes Exacerbation From Stable Status in Chronic Obstructive Pulmonary Disease

Tai Joon An Catholic University of Korea Yeouido Saint Mary's Hospital Yeun Jie Yoo Catholic University of Korea Saint Vincent's Hospital Jeong Uk Lim Catholic University of Korea Yeouido Saint Mary's Hospital Wan Seo Catholic University of Korea Yeouido Saint Mary's Hospital Chan Kwon Park Catholic University of Korea Yeouido Saint Mary's Hospital Chin Kook Rhee Catholic University of Korea Seoul St. Mary`s Hospital Hyoung Kyu Yoon (  cmcyhg@gmail.com ) Catholic University of Korea Yeouido Saint Mary's Hospital


Background
Chronic obstructive pulmonary disease (COPD) is a common chronic airway disease characterized by chronic airway in ammation with persistent air ow limitation 1,2 . The diaphragm muscle is key in the respiration process, and it is important to understand lung exercise physiology and the mechanics of COPD 3,4 . Previous studies have shown that atrophy or dysfunction of the diaphragm is related to poor COPD outcomes 5 .
An evaluation of the diaphragm is necessary for COPD patients but is di cult to achieve. The gold standard for evaluating diaphragm function is measuring trans-diaphragmatic pressure using an electromyogram during phrenic nerve stimulation or via maximal static inspiratory pressure, which is an invasive and time-consuming technique 6 . By contrast, diaphragm ultrasound (DUS) is an emerging alternative technique for evaluating the diaphragm muscle [7][8][9][10][11][12][13] . It is a non-invasive, real-time, and intuitively understandable method for evaluating various aspects of the diaphragm 10,11 . The method and effects of DUS in COPD patients have been well described by previous studies 7,8 . However, few studies have compared DUS ndings between patients with a stable status and those with acute exacerbation of COPD 9,10 . Also, there are no data about the value of diaphragm markers for distinguishing exacerbation from a stable status. We designed this study to analyze differences in diaphragm markers according to COPD status and identify potential exacerbation factors.

Study Subjects
COPD patients were retrospectively recruited between March 2020 and November 2020 at Yeouido St.
Mary's Hospital. They were diagnosed with COPD by pulmonologists. The patients were ≥ 40 years of age and satis ed the spirometry de nition of persistent air ow limitation, such as a post-bronchodilator forced expiratory volume in 1 sec/forced vital capacity ratio (FEV 1 /FVC) < 0.70.
Among them, patients who underwent DUS were enrolled. DUS was performed in the stable COPD group when they visited the outpatient clinic for a regular follow-up. DUS was performed within 48 hours of admission in patients with acute exacerbation. Exacerbation was de ned by an acute change in respiratory symptoms requiring a medication change, such as a systemic steroid or antibiotics. Patients who required hospitalization had the following indications: 1) acute respiratory failure, 2) cyanosis or edema, 3) very severe symptoms, such as dyspnea at rest or mental change, 3) severe comorbidities, such as cardiovascular disease, and 4) need for refractory to acute management. Patients who did not undergo DUS or underwent DUS after 48 hours of admission were excluded. Patients with confounding factors of diaphragm function, such as hemiplegia, quadriplegia, sequelae from an abdominal or thoracic operation, or diaphragm paralysis due to phrenic nerve palsy, were also excluded. All patients enrolled in this study completed the modi ed Medical Research Council (mMRC) scale, COPD assessment test (CAT), and history taking for comorbidities.

Comorbidities
Histories of medication and comorbidities were collected during the DUS exam. Electrical medical records were reviewed to con rm the comorbidities of the patients. The modi ed Charlson Comorbidity Index (mCCI), in which the chronic pulmonary disease categories are removed, was calculated to predict prognosis and mortality based on the ICD-10 diagnosis for the COPD patients 14 .
Diaphragm ultrasound protocol All DUS exams were performed with a single high-resolution ultrasound machine (A niti 70, Phillips, Inc., Best, the Netherlands). The exams were conducted by a respiratory physician who specializes in DUS.
DUS ndings are well established in many studies 7,8,13 . Patients were placed in a supine position, and the tests were performed at the right hemidiaphragm. A linear ultrasound probe (5-12 MHz) was used to measure the thickness of the diaphragm. Diaphragm excursion was measured with a convex ultrasound probe (1-5 MHz). B-mode was used to measure the thickness of the diaphragm (DT), and diaphragm excursion (DE) was measured in M-mode. These measurements were repeated three times in the same position and the mean value was used as the representative value.
Diaphragm thickness and diaphragm excursion DT was measured in the zone of opposition in the right hemithorax over the mid-axillary line between the eighth and eleventh intercostal spaces in longitudinal intercostal view. DT was de ned by the distance between the diaphragmatic pleura and the peritoneal membrane. First, it was measured at the end of expiration, which is correlated with functional residual capacity. Then, it was measured at the end of the inspiration during both quiet tidal breathing and maximal deep breathing.
DE was measured at the anterior subcostal margin of the right hemidiaphragm. A convex probe was positioned below the costal margin at the mid-clavicular line. The incidence angle of the ultrasound beam was perpendicular to the posterior third of the diaphragm, or the so-called DE line. The DE was the diaphragm inspiratory amplitude during respiration measured at the DE line in M-mode. DE was measured during quiet tidal breathing and maximal deep breathing (DE max ) (see Supplemental Figure 1).

Diaphragm thickening fraction
The thickening fraction of the diaphragm (TF) has been evaluated in many studies 7,8,12 . It is related to the generation of diaphragm muscle pressure. The TF was calculated with the DT value. TF was de ned as the ratio of DT changes between the end of expiration and the end of inspiration. The TF equation was [(DT at end-inspiration) -(DT at end-expiration)]/(DT at end-expiration) × 100. TF was also calculated during tidal breathing and maximal deep breathing (TF max ).

Statistical analyses
We used Student's t test and the Mann-Whitney U test for analyzing continuous variables according to the normality test results. Pearson's chi-square test or Fisher's exact test were used to compare categorical variables between groups. A receiver operating characteristic (ROC) curve analysis was performed to evaluate the usefulness of the DUS ndings for classifying the exacerbation group. The Youden index was used to identify the optimal cut-off value and compare the ROC curves. Binary univariate and multivariate logistic regression analyses were conducted to calculate the odds of being classi ed in the exacerbation group. A p-value < 0.05 was considered to indicate signi cance. Student's t test, the Mann-Whitney U test, Pearson's chi-square test, Fisher's exact test, and the logistic regression analyses were performed using IBM SPSS Statistics for Windows, version 26.0 (IBM Corp., Armonk, NY, USA). The ROC curve analyses, the Youden's index, and the comparison of the ROC curves were performed using MedCalc® Statistical Software version 19.5.6 (MedCalc Software Ltd., Ostend, Belgium; https://www.medcalc.org; 2020).

Ethics approval This study was approved by the Institutional Review Board of The Catholic University of Korea Yeouido
St. Mary's Hospital (approval no. SC20RIS0181). The need for informed consent was waived due to the retrospective nature of the study.

Demographics of study subject
Fifty-ve COPD patients were enrolled in the study. Twenty-two were classi ed as having an acute exacerbation status (AE group) and the remainder had a stable status (stable group). No signi cant differences in age, sex, mCCI, or COPD medication were observed between the groups. The AE group had a lower BMI than the stable group (20.9 vs. 24.2 kg/m 2 , p = 0.003). Scores of the symptom and dyspnea scales, such as the CAT (25.4 ± 8.2 vs. 13.7 ± 3.8, p < 0.001) and mMRC (3.1 ± 1.0 vs. 1.7 ± 1.2, p < 0.001), were signi cantly higher in the AE group than in the stable group. Smoking status and the baseline Global Initiative for Chronic Obstructive Lung Disease grouping differed between the two groups.
Lung function was different between the two groups. Absolute values and predictive percentages of FVC and FEV 1 were signi cantly lower in the AE group than in the stable group. The air-trapping index (residual volume/total lung capacity) was higher in the AE group than in the stable group (51.7 ± 9.4 vs. 43.7 ± 8.1, p = 0.019) ( Table 1).
DT at end-expiration and end-inspiration during quiet tidal breathing, and at end-inspiration during maximal deep breathing did not differ between the two groups. TF and DE during quiet tidal breathing did not differ between the AE and stable groups. TF max and DE max were signi cantly lower in the AE group than in the stable group compared to values during tidal breathing (94.8 ± 8.2% vs. 158.4 ± 83.5%, p = 0.010; 30.8 ± 11.1 mm vs. 40.5 ± 12.5 mm, p = 0.007, respectively) ( Table 1).
Diaphragm ultrasound distinguished acute exacerbation from stable status in COPD patients well ROC curve analyses of TF max and DE max for distinguishing the AE group were performed. The areas under the curve (AUC) for TF max and DE max were 0.745 and 0.721, respectively ( Figure 1 and Table 2).
The maximal Youden's index was summarized for the maximum potential effectiveness of variables with optimal cut-off values. The TF max cut-off value was 93.8% (sensitivity 68.4%, speci city 78.8%) and that of DE max was 44.9 mm (sensitivity 95.2%, speci city 44.8%) ( Table 2). No signi cant difference in TF max and DE max (p = 0.608) was observed when the ROC curves were compared (Figure 1).
We also conducted ROC curve analyses for the other variables, such as the absolute values of FVC, FEV 1 , and BMI, because these variables are well-known exacerbation factors from previous studies. These variables exhibited clinical signi cance in the ROC curve analyses (Table 2). Therefore, we conducted multiple comparisons of the ROC curves to compare the performance of the TF max ( Figure 2a Differences in demographics according to low or high TF max and low or high DE max We set new variables according to optimal TF max and DE max cut-off values. The low TF max and high TF max groups were divided by a cut-off value of 93.8%. The low DE max and high DE max groups were divided by a cut-off value of 44.9 mm. No signi cant differences in age, sex, mCCI, BMI, CAT score, or mMRC score were observed between the two groups. The low TF max and low DE max groups usually had many symptoms and exacerbation histories. Lung function was lower in the low TF max and low DE max groups than in the high TF max and high DE max groups. The percentage of AE patients was signi cantly higher in the low TF max (63.2%) and low DE max (55.6%) groups than in the high TF max (23.5%) and high DE max (7.1%) groups ( Table 3).
The low TF max and low DE max groups were associated with acute COPD exacerbation status We performed multivariate logistic regression analyses to determine whether a low TF max and low DE max could be used for distinguishing exacerbation status. The variables were entered into two models that included age, male sex, mCCI, and BMI. Model 1 included variables such as age, male sex, mCCI, BMI, and a low TF max , and model 2 included a low DE max as the variable instead of TF max . In model 1 univariate analyses, a high mCCI, low BMI, and low TF max were associated with exacerbation. In the adjusted analyses, a low BMI (odds ratio [OR] 0.70; 95% con dence interval [CI] 0.56-0.88) and low TF max (OR 8.40; 95% CI 1.55-45.56) were associated with exacerbation. Similar results were observed for model 2. A high mCCI (OR 2.68; 95% CI 1.09-6.60), low BMI (OR 0.79; 95% CI 0.64-0.97), and low DE max (OR 11.51; 95% CI 1.15-115.56) were associated with exacerbation after adjustment. Therefore, low TF max and low DE max groups were associated with an exacerbation status (Table 4).

Discussion
We designed this study to analyze the association between DUS ndings and COPD exacerbation status. Many signi cant differences in DUS ndings, such as the DT fraction during maximal deep breathing (TF max ) and DE during maximal deep breathing (DE max ) were observed between patients with a stable or acute COPD exacerbation status. However, no signi cant difference in DT was observed at end-expiration or end-inspiration. In this study, differences in breathing effort produced different results. The maximal effort measurements were signi cantly associated in the AE group, whereas those of quiet tidal breathing were not. Respiratory muscle reserve or contractile strength was related to the change of diaphragm thickness (TF) or change of length (DE), not by muscle mass itself (DT) in previous studies. These ndings support the results of those articles 8,9,12,13 .
The results of the ROC curve analyses suggest the usefulness of TF max and DE max in classifying the exacerbation and stable conditions. TF max and DE max were not inferior markers to each other and they have complementary roles. TF max was highly speci c (78.8%) and DE max was highly sensitive (95.2%).
Both were non-inferior to other classical factors of exacerbation, such as age, sex, FVC, FEV 1 , and BMI.
After dividing the patients into those with low or high TF max and those with low or high DE max , the low TF max and low DE max groups exhibited poorer lung function and a higher proportion of exacerbation.
After adjusting for age, sex, mCCI, and BMI, low TF max patients were classi ed into the AE group 8.40times higher than high TF max patients, and low DE max patients were classi ed into the AE group 11.51times higher than high DE max patients. These ndings suggest that DUS ndings can be used as distinguishing markers for COPD exacerbation.
These are valuable results regarding the use of DUS in patients with COPD. This study compared TF max and DE max . We rstly showed that they were non-inferior to each other and were complementary markers for detecting an acute exacerbation status, as far as we know. No previous study has examined differences in DUS ndings between quiet breathing and maximal breathing both in stable and exacerbation status. Only the maximal deep breathing ndings were different between the groups and those were associated with an exacerbation status. TF and DE should be checked together during maximal breathing effort when DUS is performed on a patient with COPD.
Another interesting nding is the usefulness of the DUS ndings. The results were associated with exacerbation status after adjustment, and the markers were not inferior to FVC, FEV 1 , and BMI for classifying exacerbation. DUS is a real-time test that can be performed immediately when exacerbation is suspected. With these results, we have identi ed novel markers for distinguishing patients with severe exacerbation who need hospitalization.
Several limitations of this study should be discussed. First, the cut-off value of the DUS ndings re ected maximal potential e ciency only. More studies on proper cut-off values are needed. However, the cut-off values in this study were not inferior to those of conventional markers for classifying COPD status. More studies should speci cally evaluate proper cut-off values. Second, we retrospectively included patients in this study. However, we compared the diaphragm marker itself without another intervention in this setting. Prospective research is required for further analyses.

Conclusions
This study describes the utility of DUS in COPD patients. This is the rst study to report the role of DUS for distinguishing exacerbation from a stable status. We also showed the value of DUS ndings with the consideration of other contributing factors, such as sex, age, and BMI. Based on the results, we suggest using DUS ndings as indicators of exacerbation status in COPD patients. Availability of data and materials: Researchers may request datasets which were used in this study to the corresponding author with reasonable request.

List Of Abbreviations
Competing interests: None of the authors have any con icts of interest

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.