The Macklin effect closely correlates with pneumomediastinum in acutely ill intubated patients with COVID-19 infection

Abstract

Purpose

Patients with COVID-19 infection are frequently found to have pulmonary barotrauma. Recent work has identified the Macklin effect as a radiographic sign that often occurs in patients with COVID-19 and may correlate with barotrauma.

Methods

We evaluated chest CT scans in COVID-19 positive mechanically ventilated patients for the Macklin effect and any type of pulmonary barotrauma. Patient charts were reviewed to identify demographic and clinical characteristics.

Results

The Macklin effect on chest CT scan was identified in a total of 10/75 (13.3%) COVID-19 positive mechanically ventilated patients; 9 developed barotrauma. Patients with the Macklin effect on chest CT scan had a 90% rate of pneumomediastinum (p < 0.001) and a trend toward a higher rate of pneumothorax (60%, p = 0.09). Pneumothorax was most frequently omolateral to the site of the Macklin effect (83.3%).

Conclusion

The Macklin effect may be a strong radiographic biomarker for pulmonary barotrauma, most strongly correlating with pneumomediastinum. Studies in ARDS patients without COVID-19 are needed to validate this sign in a broader population. If validated in a broad population, future critical care treatment algorithms may include the Macklin sign for clinical decision making and prognostication.

1. Introduction

Several studies have demonstrated that pneumomediastinum (PMD) and/or pneumothorax (PNX) occur frequently in patients with coronavirus disease 2019 (COVID-19) pneumonia,1., 2., 3., 4., 5., 6. with reported incidence of 4.2% in hospitalized patients and ranging from 15 to 20% of those receiving invasive ventilation.3., 7., 8. Notably, barotrauma has been reported to occur more frequently in patients with COVID-19-related acute respiratory distress syndrome (ARDS) as compared with patients with ARDS due to causes other than COVID-19.2., 3. It has been suggested that specific virus-induced frailty of the lung parenchyma, possibly triggered by microvascular thrombosis,9., 10. interstitial inflammation, as well as endothelial barrier disruption,10., 11. are responsible for the observed higher rates of PNX/PMD in this population. Complicating the matter, the contribution of medical treatment to the development of pulmonary barotrauma by alteration of the pulmonary parenchymal microenvironment is not well understood.12., 13. Unfortunately, development of PNX/PMD is associated with mortality rates >60% in COVID-19 ARDS patients requiring invasive ventilation.3., 4.

Several recent reports have suggested that detection of a Macklin-like radiographic sign, also known as the Macklin effect on chest computed tomography (CT) scan (defined as air in the peribronchovascular space14., 15., 16.) may be a very accurate predictor of development of PNX/PMD in patients with COVID-19-related acute respiratory distress syndrome (ARDS) requiring invasive ventilation.17., 18. However, these studies were performed in a single center by the same group and validation in an independent patient population is needed.17., 18. If validated, identification of the Macklin effect could be used to risk stratify and guide management of high-risk patients.19., 20.

Here, we investigated the association between Macklin effect and radiographically evident PMD/PNX in invasively ventilated COVID-19 positive ARDS patients admitted to a tertiary university hospital in the United States with high COVID-19 transmission during the acute phase of the pandemic.

2. Materials and methods

2.1. Patient selection

We obtained IRB approval from Columbia University (IRB IRB-AAAS9652 approved on 03/16/2020) to retrospectively study patients with COVID-19 infection. We included all patients with COVID-19 pneumonia (defined as positive COVID-19 PCR or antigen test plus signs and symptoms of pneumonia) who required hospital admission, were intubated, and had at least one chest CT available for review between 2/1/2020 and 7/1/2021. We queried our radiology record system using MModal Catalyst to identify chest CT studies during this timeframe with reports containing the search term “COVID-19”. We further refined this search by selecting cases with reports also containing the terms “intubated”, “endotracheal tube”, “endotracheal”, “ETT”, “pneumothorax”, “pneumomediastinum”, or “subcutaneous emphysema”. Subsequently duplicates and patients not meeting the inclusion criteria were removed, including those who were not invasively ventilated.

2.2. Data collection

Clinical and demographic data was collected by querying patients' charts in our medical record system Epic. We collected data regarding patient age, gender, steroid treatment, invasive ventilation, and in-hospital mortality. For this study, we defined barotrauma as the development of PMD, PNX, or subcutaneous emphysema.

2.3. Radiographic analysis

All cases were reviewed by an attending radiologist with 18 years of experience. Each case was analyzed for the air dissecting along the bronchovascular bundles (known as the Macklin Effect). In cases where the Macklin effect was present, the finding was characterized as “central”, “peripheral” or “both”. CT scans were also evaluated for the presence of pneumothorax (noting “left”, “right” or “bilateral”), pneumomediastinum, and tracheal lesions.

2.4. Statistical analysis

Statistical analysis was performed using an advanced statistics package in Microsoft Excel v.16.48 (Redmond, WA, USA) and the online calculator “Chi Square Calculator”.²¹ For all analyses, a p value less than or equal to 0.05 was considered significant. Of note, in any case where the data was not available for a clinical or demographic feature, the patient was excluded from that specific analysis.

3. Results

3.1. Demographic and clinical characteristics

A total of 75 patients met inclusion criteria and were enrolled in this study. Fig. 1 summarizes the study design.

Fig. 1.

Criteria for patients included in this study.

The Macklin effect on chest CT scan was identified in a total of 10/75 (13.3%) patients (Fig. 2, Fig. 3 ). Baseline characteristics and hospital mortality in patients with and without the Macklin effect are presented in Table 1 . Patients found to have the Macklin effect trended toward a younger age although the difference was not significant. The majority of patients in both groups received steroid therapy. In-hospital mortality did not differ significantly among the groups.

Fig. 2.

Representative coronal (A, C) and axial (B, D) images of patients with the Macklin Effect (arrows) with and without associated pulmonary barotrauma, respectively.

Fig. 3.

A–C are representative axial images of a patient with the Macklin effect with concomitant pneumomediastinum, pneumothoraces, and severe subcutaneous emphysema.

Table 1.

Clinical characteristics of patients with and without the Macklin effect on CT.

Clinical characteristic	No Macklin on CT N = 65	Macklin on CT N = 10	p value
Age (SD)	61.5 (14.9)	53.2 (13.7)	0.08
Male gender	32/65 (49.2%)	6/10 (60.0%)	0.53
Steroid treatment for COVID-19	51/63 (81.0%) (data unavailable for 2 patients)	9/10 (90.0%)	0.49
Invasive mechanical ventilation at time of Macklin effect detection	N/A	7/10 (70.0%)	N/A
In-hospital mortality	32/64 (50.0%) (one patient transferred to outside hospital)	5/10 (50.0%)	1.0

3.2. Correlation of the Macklin effect with pneumothorax and pneumomediastinum on chest CT

Rates and characteristics of pulmonary barotrauma events are presented in Table 2 . The Macklin effect was seen on initial imaging in 70.0% of patients who ultimately went on to demonstrate this sign; the remaining 20.0% were identified on follow-up scans.

Table 2.

Imaging features.

Radiographic characteristic	No Macklin on CT	Macklin on CT	p value
Barotrauma	25/65 (38.5%)	9/10 (90.0%)	0.0023
Pneumothorax	21/65 (32.3%)	6/10 (60.0%)	0.09
Right only	10/21 (47.6%)	1/6 (16.7%)	0.17
Left only	5/21 (23.8%)	3/6 (50.0%)	0.22
Bilateral	6/21 (28.6%)	2/6 (33.3%)	0.82
Diagnosed on chest radiograph	15/21 (71.4%)	5/6 (83.3%)	0.56
Diagnosed on chest CT	6/21 (28.6%)	1/6 (16.7%)	0.56
Pneumomediastinum	8/65 (12.3%)	9/10 (90.0%)	<0.00001
Diagnosed on chest radiograph	3/8 (37.5%)	5/10 (50.0%)	0.60
Diagnosed on chest CT	5/8 (62.5%)	4/10 (40.0%)	0.34
Tracheal lesion	2/65 (3.1%)	1/10 (10.0%)	0.29

Nine out of the 10 patients with the Macklin effect on chest CT scan had barotrauma which was significantly higher than the 32% (21/65) reported in patients without the Macklin effect (p < 0.001). A single patient with the Macklin effect did not have either pneumothorax or pneumomediastinum and died during their hospitalization. All patients with pneumothorax (60%) also had pneumomediastinum. The presence of the Macklin effect did not impact whether the patient had a unilateral or bilateral pneumothorax.

3.3. Distribution of Macklin effect

The distribution of Macklin effect is presented in Table 3 . The Macklin effect was adjacent to central airways in the majority of cases. In terms of distribution, the Macklin effect was bilateral in 50% of cases and omolateral to the pneumothorax in 83.3% of cases.

Table 3.

Macklin distribution.

Distribution
Side
Right	3/10 (30.0%)
Left	2/10 (20.0%)
Bilateral	5/10 (50.0%)
Omolateral to pneumothorax	5/6 (83.3%)
Region
Central	6/10 (60.0%)
Peripheral	2/10 (20.0%)
Both	2/10 (20.0%)

4. Discussion

We found that 13.3% of COVID-19 positive patients who required invasive ventilation had the Macklin effect on chest CT scan during the course of their disease. Nine out of ten patients with the Macklin effect had pneumomediastinum, and there was a trend toward a significantly increased number of pneumothoraces. When considering pneumothoraces, the Macklin effect was found omolateral to the pneumothorax 83.3% of the time.

To the best of our knowledge, there are only two other studies investigating the role of Macklin effect in patients with COVID-19 ARDS. Both studies were performed in Italy and showed a strong association between presence of Macklin effect on chest CT scan and subsequent development of pulmonary barotrauma. This is the first study to confirm the association of the Macklin effect with pneumomediastinum and suggests that observation of the Macklin effect can predict patients at high risk for barotrauma.17., 18.

Palumbo et al. investigated 173 invasively ventilated COVID-19 positive ARDS patients and found the Macklin effect in 22% of patients. Of note, 89% of patients who developed barotrauma (pneumothorax or pneumomediastinum) had the Macklin effect on baseline chest CT scan approximately 8 days before detection of barotrauma. Using these data, authors calculated a sensitivity of 89.2% of Macklin effect in predicting barotrauma and a specificity of 95.6%.¹⁸

Paternoster et al. performed a larger study of 698 COVID-19 positive patients presenting to the emergency department in Italy. In their study, the Macklin effect was present in 33 (4.7%) patients at the first chest CT scan obtained after presentation to ED and 32 developed barotrauma within 3.2 days.¹⁹ Calculated sensitivity and specificity for predication of barotrauma in their study were 100% and 99.8%, respectively. Furthermore, all patients with the Macklin effect developed ARDS requiring ICU admission.

Our study, focusing specifically on invasively ventilated COVID-19 positive ARDS patients, is the first performed in United States and also the first performed outside Italy. Our findings confirmed that the Macklin effect is frequently seen in patients with COVID-19 ARDS and is associated with development of pneumomediastinum. It may also be associated with development of pneumothorax, although the association did not reach statistical significance in the present study.

Outside the ARDS setting, a large study of patients with blunt trauma to the chest evaluated the correlation of pneumomediastinum with the Macklin effect and reported a 39% incidence of the Macklin effect in these patients.²² A review by Murayama et al. evaluated the correlation of spontaneous pneumomediastinum and the Macklin effect, finding several case reports in the literature where the two barotrauma events occurred concurrently.¹⁴ These studies are in accord with data from our study and the previous group investigating the Macklin effect in patients with COVID-19 infection.

Here, we sought to validate the findings of a single-center study in our own population of severely ill intubated COVID-19 positive patients and demonstrated a high incidence of the Macklin effect in this population. Our study confirms that Macklin effect is frequently detected in patients with COVID-19 ARDS and that the there is an association between the Macklin effect and barotrauma. In our experience, there was a statistically significant association with pneumomediastinum and a trend toward an association between the Macklin effect and pneumothorax, although this did not reach significance.

Although the association between Macklin effect and barotrauma had already been described in several clinical contexts, its use as a potential predictor of PNX/PMD in patients with ARDS represents a novel approach. PNX/PMD associated with ARDS is difficult to manage and there is a lack of a standardized approach, as they generally occur in critically ill patients with limited cardiovascular reserve. Indeed, several studies have suggested that patients with ARDS who develop barotrauma have worse outcomes as compared with patients who do not. The possibility of anticipating barotrauma development by several days, if confirmed, would provide clinicians with a unique opportunity to alter management in advance in order to prevent the development of PMD/PNX. For example, clinicians may decide to avoid using high airway pressures for recruitment maneuvers, or use ultraprotective ventilation with or without extracorporeal membrane oxygenation (ECMO) support.23., 24., 25. Indeed, in a small pilot study, Paternoster et al. used the Macklin effect detected on chest CT scan to identify candidate patients with severe COVID-19 ARDS to administer extracorporeal membrane oxygenation (ECMO) without invasive ventilation.

Our study is unique from prior work in that it highlights the critical link between pneumomediastinum and the Macklin effect, which has been observed in other research outside of the critical care field. Compared with previous investigations, our study focused on invasively ventilated patients with invasive ventilation. The validation of Palumbo et al.'s work also suggests that the Macklin effect may be a robust predictor of pneumomediastinum and suggests further study of this relationship in patients with ARDS without COVID-19 infection. This finding is likely underpinned by a mechanistic link between the two entities involving the continually progressive dissection of air along bronchovascular bundles and underscores the potential of the Macklin effect as a radiographic biomarker. We did not find a mortality difference between patients with and without Macklin effect. However, our study was not powered to detect such difference, and whether Macklin effect could be a prognostic marker of mortality remains to be determined.

While our findings are in accord with prior literature, there were several limitations. This data is derived from a single-center tertiary care hospital with extensive critical care expertise, limiting generalizability to the community setting. As a retrospective review, data was limited to the study period and significant changes to patients' clinical course may have occurred after the end of the study. Given that the Macklin effect is seen on CT, data was obtained from timepoints when patients were scanned, however significant barotrauma related events may have occurred and resolved between imaging studies. Radiology studies were obtained at the discretion of ordering providers when clinically appropriate (for example, a change in clinical status), and therefore were not evenly distributed across the course of a patient's hospitalization. Patients with more severe disease were probably more likely to undergo chest CT scan, therefore we cannot exclude selection bias. Finally, a single radiologist reviewed images in this study; a more robust analysis could have included several readers to provide independent analysis for the presence of the Macklin effect. Despite these limitations, we believe that these data provide robust support for the association of the Macklin effect with pneumomediastinum and represent the independent validation of a new important radiographic biomarker for pneumomediastinum.

Future work will investigate the role of the Macklin effect in COVID-19 negative mechanically ventilated patients to demonstrate whether the findings presented here can apply to a broader patient population. If validated as a novel biomarker for pulmonary barotrauma, this may become an important part of critical care algorithms and may have prognostic implications.²⁰

5. Conclusions

Here, we demonstrate that the Macklin effect is frequently seen in critically ill mechanically ventilated patients with COVID-19 infection and has a significant correlation with the incidence of pneumomediastinum. Our work validates and refines prior findings of the correlation between the Macklin effect and pulmonary barotrauma. This sign may prove to be in important new radiographic biomarker that can be incorporated into treatment algorithms and aid in prognostication.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

CRediT authorship contribution statement

Valerie Maccarrone: Investigation, Writing – review & editing. Connie Liou: Investigation, Writing – review & editing. Belinda D'souza: Writing – review & editing. Mary M. Salvatore: Writing – review & editing. Jay Leb: Writing – review & editing. Alessandro Belletti: Conceptualization, Writing – review & editing. Diego Palumbo: Conceptualization, Writing – review & editing. Giovanni Landoni: Writing – review & editing. Kathleen M. Capaccione: Conceptualization, Investigation, Formal analysis, Writing – original draft, Writing – review & editing.