Respiratory muscle dysfunction in long-COVID patients

While lung, kidney and the vascular system appear to be the main sites of acute Severe acute respiratory syndrome coronavirus type 2 (SARS-CoV-2)-related complications [1, 2], early sequelae of coronavirus disease (COVID)-19 are reported by the vast majority of convalescent patients [3, 4]. Sequelae persisting for longer than 4 weeks are now phenotypically summarized under the umbrella term “Long COVID” [4]. The most commonly reported symptoms include persistent dyspnea and fatigue in up to 51% and 63% of cases, respectively, which are also among the longest lasting sequelae [3, 5]. As recently reported, exertional intolerance and dyspnea can also be observed in Long COVID patients with preserved lung function [6]. In this light, in addition to a growing body of evidence regarding pulmonary parenchymal and cardiac sequelae, exercise intolerance in Long COVID patients might have additional causes related to respiratory muscular dysfunction.

In a cross-sectional approach, we have therefore prospectively investigated respiratory drive and effort in Long COVID-19 patients with self-reported exercise intolerance presenting to our outpatient department (OPD).

Sixty-seven adult convalescent COVID-19 patients (30 female, 37 male, mean age: 49 years, baseline characteristics are given in Table 1) presenting after mild to critical disease (according to World Health Organization (WHO) classification) completed general symptom, activity and productivity (modified Work Productivity and Activity Impairment (WPAI) score) questionnaires before undergoing complete pulmonary function testing (PFT), including spirometry, body plethysmography, capillary blood gas analyses (CBG) at rest and immediately after performing a 6 min walk test (6MWT). Assessment of dyspnea intensity at rest and during the 6MWT using the Modified BORG Dyspnea Scale (Borg CR10) was performed. In addition, respiratory muscle testing to assess respiratory drive and effort was conducted following current guidelines [7, 8]. All adult patients with persisting symptoms for ≥ 4 weeks after COVID-19 with a proven record of SARS-CoV-2 infection (positive PCR for SARS-CoV-2 or presence of SARS-CoV-2-specific nucleocapsid antibodies) were eligible after informed consent. Patients < 18 years or without proven SARS-CoV-2 infection were excluded. Patients were recruited via the Post COVID Clinics of the Divisions of Pneumology and Infectious Diseases at the University Medical Center Hamburg Eppendorf. Eligible patients categorized as hospitalized had to be hospitalized due to COVID-19.

PRISM 9 (GraphPad Inc, San Diego, CA) and R for macOS version 4.0.3 (https://​cran.​r-project.​org) with RStudio 1.3 (RStudio, Boston, MA) were used for the following statistical analyses: one-sample t test and Pearson correlation analysis for normally distributed data (D’Agostino-Pearson Test); Mann–Whitney, Spearman correlation and Fisher’s exact test for nonparametric data; corrplot 0.84 library was utilized for principal component clustering.

At the time of presentation to our OPD (median of 152 days, IQR: 65–260 after onset of acute symptoms), patients initially hospitalized due to COVID-19 (55% of cohort) showed reduced PFT parameters compared with nonhospitalized patients. In addition, initially hospitalized patients walked 92.3 m (15.2%) less in the 6MWT and showed a more pronounced decrease in the arterial partial pressure of oxygen (P_aO₂) during the 6MWT (median: + 1.5 mmHg vs. − 7.8 mmHg). No differences were found in dyspnea perception, functional impairment, daily activity or productivity. While hospitalized patients were older, had a higher body mass index and more comorbidities which were associated with more severe acute disease, history of lung disease was rare and did not differ between hospitalized and nonhospitalized patients (Table 1).

In addition to exercise intolerance reported by all patients, the most frequent symptoms were persistent exertional dyspnea (95.5%) and fatigue (83.6%, Fig. 1A). These symptoms were associated with alterations in respiratory drive and effort. Both hospitalized and nonhospitalized patients had increased total respiratory muscle strain (= occlusion pressure at 0.1 s (P_0.1)/ maximal inspiratory pressure (PI_max) > 0.02); 97.2 vs. 87.1%, P_0.1/PI_max range: 3–25%, p = 0.0005 and p = 6.6E-08, Fig. 1B) at the time of presentation to the OPD. Hospitalized patients showed a trend towards more pronounced respiratory muscle strain (P_0.1/PI_max: 0.05 vs. 0.06, p = 0.056). Inspiratory muscle strength (as determined by the peak value of maximum inspiratory mouth pressure measured from residual volume (PImax_peak RV) = PI_max) was decreased below six and age-specific cutoffs in 88% of patients (Fig. 1C, vertical bar), predominantly in patients previously hospitalized due to COVID-19 (p = 0.0108, female and p = 0.0079, male; Fig. 1C). In addition, inspiratory muscle weakness was more frequent in women (96.4 vs. 79.3%, p = 0.0088, Fisher’s exact test). Neuroventilatory activity as determined by P_0.1 > 0.3 kPa (~ 3.1 cmH₂O) was elevated in 56% of patients (mean P_0.1: 0.36 and 0.37 kPa, p = 0.0291 and p = 0.0029, nonhospitalized and hospitalized, respectively, Fig. 1D), which was independent of hospitalization status (p = 0.64).

Clinically, alterations in respiratory drive and effort after COVID-19 were associated with reduced distance (6MWD) in the 6MWT (P_0.1: 595.5 vs. 529.3 m, p = 0.0219; PI_max: 600.1 vs. 537.6 m, p = 0.0599; P_0.1/PI_max: 659.3 vs. 548.5 m, p = 0.0162; Fig. 1E).

While no patient was hypoxemic at rest, convalescent COVID-19 patients with elevated P_0.1 showed a significant decrease in arterial oxygen partial pressure (P_aO₂) during the 6MWT (ΔP_aO₂: − 6.6 mmHg, p = 0.0134; Fig. 1F). In all patients with exertional deoxygenation, pulmonary thromboembolic disease was ruled out by subsequent ventilation/perfusion scans.

Patients with elevated P_0.1 after COVID-19 reported increased dyspnea during the 6MWT, as informed by a larger difference (Δ) in BORG scores at rest and upon exercise (+ 1.3 vs. + 2.1, p = 0.0299; larger = worse, Fig. 1G). In addition, patients with elevated P_0.1 > 0.3 kPa also reported less daily activity and productivity due to persisting symptoms (modified WPAI score, 6.3 vs. 9.8, p = 0.0471; higher = larger impairment, Fig. 1H) as well as increased overall functional impairment as determined by the Post-COVID functional status (PCFS, [9]) scale (1 vs. 2, p = 0.0058; higher = larger impairment, Fig. 1I).

In univariate regression analysis, P_0.1 was associated with functional residual capacity (FRC, r = − 0.27, p = 0.046), ΔP_aO₂ (r = − 0.30, p = 0.007), number of comorbidities (r = 0.27, p = 0.044) and P_0.1/PI_max (r = 0.30, p = 0.007). PI_max was correlated with the diffusing capacity of carbon monoxide (DLCO, r = 0.37, p = 0.006), 6MWD (r = 0.33, p = 0.014), Carbon monoxide transfer coefficient (KCO, r = 0.36, p = 0.006) and P_0.1/PI_max (r = − 0.54, p = 2.1E-05). P_0.1/PI_max was associated with KCO (r = − 0.33, p = 0.015), ΔBORG score (r = 0.33, p = 0.013), age (r = 0.26, p = 0.05) and number of comorbidities (r = 0.40, p = 0.003) (Fig. 1J).

In a principal component-based multivariate analysis, P_0.1 and P_0.1/PI_max clustered with age, body-mass-index (BMI), number of comorbidities, FEV1/FVC, time from diagnosis and CBG while PI_max, did not clearly cluster with any of the parameters (Fig. 1J).

Comorbidities were not associated with pathologically altered P_0.1, PI_max or P_0.1/PI_max (all p > 0.05). Patients with a history of asthma were less likely to show pathological P_0.1/PI_max (χ² = 5.41, p = 0.020).

In our cross-sectional pilot study of convalescent COVID-19 patients with persistent exercise intolerance, we identified a high prevalence of impaired respiratory muscle function and upregulated neuroventilatory activity ~ 5 months after diagnosis. Functionally, this was associated with reduced 6MWD and daily activity/productivity in connection with exercise-induced deoxygenation.

Recently, published PFT data of COVID-19 patients show reduced TLC and DLCO up to 6 months after infection, which occurred more often in patients with severe disease [5]. This is in line with our data showing that patients initially hospitalized for COVID-19 had significantly lower PFT parameters, including TLC and DLCO, up to 5 months after infection. This was also associated with reduced exercise capacity in hospitalized patients after COVID-19 as measured by 6MWD. Our study extends these findings, as we report a high prevalence of increased respiratory drive and impaired respiratory muscle capacity in convalescent, persistently symptomatic COVID-19 patients.

In our cohort, patients requiring hospitalizations, including ICU treatment, also had impaired respiratory muscle strength as demonstrated by reduced PI_max, which is consistent with recently reported findings of fibrotic diaphragm remodeling in patients who died due to COVID-19-related ARDS [10].

Elevated P_0.1, as found in the majority of our patients, is strongly associated with heightened dyspnea perception [11]. This was also the case in our cohort, as shown by elevated BORG-CR scores and everyday activity, productivity and COVID-related functional impairment (PCFS). Strikingly, this was not only the case in hospitalized patients where elevated P_0.1 might be a consequence of reduced inspiratory muscle strength PI_max but also in nonhospitalized patients.

Therefore, our data support that, pathophysiologically, elevated P_0.1 might be a function of exercise-induced deoxygenation in convalescent, persistently symptomatic COVID-19 patients. While pulmonary thromboembolic disease was not detected by V/Q scan (as described above), six patients showed signs of ground-glass opacity and (mostly minor) fibrotic changes and exercise-induced deoxygenation was associated with lower DLCO (Fig. 1J). Systematic analysis of these changes, however, was out of the scope of the present study, which is a limitation. Additionally, due to unavailability of data in some patients, we cannot exclude pre-existing changes in respiratory drive and effort sustained from before SARS-CoV-2 infection. Additional limitations include putatively biased patient selection, as most patients reported to our OPD with persistent symptoms after COVID-19, with very few patients referred for routine follow-up after COVID-19. Patients and staff were also not blinded to the overall testing, possibly inserting additional bias in the measurement as does lack of historical PFT data. Particularly for ICU patients, muscular deconditioning associated with intensive care might contribute to respiratory muscle impairment. Although it was not possible to differentiate inspiratory muscle impairment from generalized muscle weakness or postinfection myopathy, in our cohort, creatine kinase and myoglobin serum levels did not differ between patients with normal or abnormal respiratory muscle function (p = 0.202 and p = 0.075, respectively). In addition, pre-existing conditions/comorbidities did not correlate with abnormal respiratory muscle function in our cohort. Also, inspiratory muscle weakness also occurred frequently in nonhospitalized patients (65%). We also cannot specifically attribute the detected changes in respiratory drive and inspiratory muscle function to SARS-CoV-2, as we cannot rule out a general effect of viral infections. Regardless of SARS-CoV-2 specificity, the high prevalence in our pilot study points toward a relevant healthcare burden given the pandemic nature of COVID-19.

As there is strong evidence that chronic fatigue syndrome (CFS) is associated with COVID-19 [3, 5], it is compelling to speculate to what extent heightened neuroventilatory activity, as documented by P_0.1 in our cohort, contributes to COVID-19-CFS. Particularly, the inability to adequately increase respiratory effort upon increased respiratory drive is known to worsen respiratory distress [11]. Therefore, more invasive techniques such as twitch interpolation might help to further characterize dysregulation of respiratory drive and effort in Long COVID patients.

We were able to detect increased respiratory drive as well as inspiratory muscle dysfunction in persistently symptomatic patients approx. 5 months after COVID-19. Notwithstanding the small sample size, our findings reveal a previously unidentified neuromuscular component of COVID-19 sequelae.

Given the wide accessibility of respiratory muscle testing as a relatively low-cost approach (in particular in comparison with imaging and immunological laboratory studies), we strongly advocate for systematic respiratory muscle testing in the diagnostic workup of persistently symptomatic, convalescent COVID-19 (Long COVID) patients.