Altered lung physiology in two cohorts after COVID-19 infection as assessed by computed cardiopulmonography

Abstract

The longer-term effects of COVID-19 on lung physiology remain poorly understood. Here, a new technique, computed cardiopulmonography (CCP), was used to study two COVID-19 cohorts (MCOVID and C-MORE-LP) at both ∼6 and ∼12 mo after infection. CCP is comprised of two components. The first is collection of highly precise, highly time-resolved measurements of gas exchange with a purpose-built molecular flow sensor based around laser absorption spectroscopy. The second component is estimation of physiological parameters by fitting a cardiopulmonary model to the data set. The measurement protocol involved 7 min of breathing air followed by 5 min of breathing pure O₂. One hundred seventy-eight participants were studied, with 97 returning for a repeat assessment. One hundred twenty-six arterial blood gas samples were drawn from MCOVID participants. For participants who had required intensive care and/or invasive mechanical ventilation, there was a significant increase in anatomical dead space of ∼30 mL and a significant increase in alveolar-to-arterial Po₂ gradient of ∼0.9 kPa relative to control participants. Those who had been hospitalized had reductions in functional residual capacity of ∼15%. Irrespectively of COVID-19 severity, participants who had had COVID-19 demonstrated a modest increase in ventilation inhomogeneity, broadly equivalent to that associated with 15 yr of aging. This study illustrates the capability of CCP to study aspects of lung function not so easily addressed through standard clinical lung function tests. However, without measurements before infection, it is not possible to conclude whether the findings relate to the effects of COVID-19 or whether they constitute risk factors for more serious disease.NEW & NOTEWORTHY This study used a novel technique, computed cardiopulmonography, to study the lungs of patients who have had COVID-19. Depending on severity of infection, there were increases in anatomical dead space, reductions in absolute lung volumes, and increases in ventilation inhomogeneity broadly equivalent to those associated with 15 yr of aging. However, without measurements taken before infection, it is unclear whether the changes result from COVID-19 infection or are risk factors for more severe disease.

Keywords: COVID-19; human; laser absorption spectroscopy; lung volumes; respiratory dead space.

Graphical abstract

Figure 1.

Precision and stability of laser spectroscopy systems for analysis of respired gas every 10 ms. A, C, and E: typical absorbance spectrum taken within a single 10-ms measurement period for O₂, CO₂, and water vapor, respectively. Symbols illustrate data, and curves illustrate the Voigt profiles fitted to the data. B, D, and F: analyses every 10 ms during 20-s periods of steady gas flows for the O₂, CO₂, and water vapor analyzers, respectively. For O₂ and water vapor, the SD for the measurement is <1/1,000th of the mean. For CO₂, it is <3/1,000th of the mean.

Figure 2.

Schematic of model and parameter estimation process. The plant is comprised of 3 core models, 1 for the blood gas dissociation curves, 1 for the inhomogeneity in the lungs, and 1 for the circulation and the gases dissolved in the body tissues. Metabolism drives the model and determines (along with body size) the cardiac output. Total respiratory flow and inspired gas composition are as recorded during the measurement. The model is embedded in a nonlinear least-squares optimization algorithm. This progressively adjusts the parameters of the model so as to minimize the sum of squared deviations between the model and the data for the molar flows for each of the individual gas species during expiration.

Figure 3.

Example fit of model to gas exchange data. A: calculated lung volume against time. FRC, functional residual capacity. B: measured and model-simulated flows of individual gas species at the mouth. The model responses are essentially hidden as they are overlaid by the data. C: cumulative residuals (measured minus simulated) illustrating the residual error in the model fit for the gas exchange data.

Figure 4.

Expirograms for CO₂, O₂, and N₂ fractions (F) for a breath during the air-breathing phase and for a breath early and late in the N₂ washout phase, for an example data set. A–C: CO₂ expirograms. D–F: O₂ expirograms. G–I: N₂ expirograms. A, D, and G: breath during air-breathing phase. B, E, H: breath during early N₂ washout phase. C, F, and I: breath during late N₂ washout phase.

Figure 5.

Calculated blood gas contents (C) derived within the model for an example data set. A: systemic mixed venous and arterial contents for CO₂. B: systemic mixed venous and arterial contents for O₂. C: systemic mixed venous and arterial contents for N₂.

Figure 6.

Influence of selected physical characteristics of participants on selected parameter estimates. A–C: MCOVID cohort (112 participants). D–F: C-MORE-LP cohort (66 participants). A and D: influence of height on end-inspiratory anatomical dead space (Vdi) in liters (BTPS). B and E: influence of body mass index (BMI) on functional residual capacity (FRC) % predicted. C and F: influence of age on the SD for the natural logarithm of the standardized lung compliance (σlnCl). MCOVID cohort: ●, Control group; △, Community group; □, Ward group; ○, ICU group. C-MORE-LP cohort: ●, Basic/O₂ group; △, HFNC/CPAP group; □, IMV group. After controlling for other factors, the level of statistical significance for each significant relationship illustrated is given in Table 6.

Figure 7.

Comparison between the MCOVID and C-MORE-LP cohorts of the values for selected parameters at the different severity levels for prior COVID-19 infection. A: lung diffusing capacity for CO corrected for hemoglobin (Dl_COc). B: alveolar volume (Alv Vol). C: end-inspiratory anatomical dead space (Vdi). D: functional residual capacity (FRC). E: standard deviation for the natural logarithm of standardized lung compliance (σlnCl). Values have been corrected to those for a standard participant who is male, nonsmoking and with an age of 47.5 yr, height of 1.76 m, and body mass index (BMI) of 29.5 kg·m⁻². Filled bars, MCOVID cohort; open bars, C-MORE-LP cohort. Error bars are 1 SE. % pred, % predicted.