The study findings were published in the peer reviewed Journal of Applied Physiology.
https://journals.physiology.org/doi/abs/10.1152/japplphysiol.00436.2022
This study applied a new technological approach that has been named computed cardiopulmonography (CCP) to investigate lung physiology in two cohorts of post COVID-19 patients. One cohort, MCOVID, was constructed from patients of working age. Apart from patients whose infection had been managed in hospital, this cohort also included both control participants and patients whose infection had been managed in the community.
It therefore covered a wide range of severities for the prior acute infection. Many of these patients also provided an arterial blood gas sample while breathing through the MFS during the CCP study protocol. The second cohort, C- MORE-LP, was older and more representative of the general population. It did not include either a control group or patients managed in the community, and there were no blood gas samples associated with it.
Pathophysiological findings
For the groups that had been most severely affected by COVID-19 (ICU group for MCOVID, IMV group for C-MORE-LP), computed cardiopulmonography (CCP) revealed an enlarged anatomical deadspace. The mean arterial PO2 for the ICU group was significantly lower than for the control participants and the (iA – a)PO2 gradient was significantly higher.
The radiology for these particular patients commonly revealed a degree of fibrosis. An interesting, but unanswered question is whether these changes result directly from COVID-19 pneumonia or whether they arise as a result of invasive mechanical ventilation.
Indeed, this is in keeping with a study of tissue from patients who had previously had COVID-19, and who were undergoing a partial lung resection for unrelated pathologies. In these patients, the histology of the parenchyma surrounding the lesion was found to be normal (8).
In apparent contrast to this conclusion, there has developed a large literature, e.g. (9-14), surrounding clinical lung function measurements to suggest that significant reductions in DLCO persist in patients following COVID-19 pneumonia, and that this may relate to parenchymal lung disease. – (Values for DLCOc % predicted and alveolar volume % predicted decreased in a highly significant manner as COVID severity increased).
In these studies, the imaging sequence was designed so that the signal was very heavily weighted towards red blood cells and tissue that were in close apposition to the alveoli, and so the results suggest some parenchymal abnormality. The second of these studies (16) found a close correlation between the reduction in this ratio and the reduction in DLCO.
As acute COVID-19 pneumonia is associated with pulmonary vascular endothelialitis and microthrombi (17), one possibility is that the reductions in DLCO and in red blood cell to tissue Xe ratio were caused through some loss of small vessels and capillaries in the lung. Such changes, however, would not necessarily result in any abnormality under resting conditions in the efficiency of pulmonary gas exchange or in the arterial blood gas values.
DLCO itself is the product of two primary measures. The first is KCO, which is the rate at which CO disappears from the lung under the standardised condition of a breath-hold at total lung capacity (TLC). In essence, this may be seen as a ‘density’ for the alveolar surface area at that volume. The
second primary measure is the alveolar volume, which is a single breath estimate (usually by He dilution) of TLC (this typically will underestimate true TLC by 5-10%). In our study, and in a number of others, see (9-14), it is this alveolar volume rather than KCO that is low post COVID-19.
The interpretation of this is not straightforward. While we did not detect a significant effect of BMI on this volume from the DLCO measurements, others have detected some reduction in total lung capacity with increasing BMI (18). If some part of the reduction in alveolar volume were related to obesity, then a ‘normal’ value for KCO should be seen as inappropriate, as the Global Lung Initiative reference value project found that the predictions for DLCO were essentially unaffected by whether or not data for obese individuals were included (19).
Thus for obese individuals with normal lungs, any decrease in alveolar volume below normal would need be compensated for by an increase in KCO to a supra-normal value.
Estimates for functional residual capacity (FRC) were obtained as part of the computed cardiopulmonography (CCP) methodology. These values were divided by predicted values based on the individuals’ age, height and sex to produce values expressed as % predicted. However, the predictive equations do not include a term for BMI, and as FRC is known to vary substantially with BMI (18), we included the log of the BMI as a covariate in the linear mixed- effects model.
While this term was significant for both cohorts, it did not eliminate a significant association between COVID severity and the reduction in FRC. Fig. 7 illustrates the reduction in FRC with severity for a hypothetical participant at a standardised BMI of 29.5 kg.m-2, and these were strikingly similar to the reductions observed in alveolar volume from the DLCO measurements.
There are a number of possible causes for the low FRC and for the low alveolar volume from the single breath DLCO washout. The first is that some fraction of the alveolar volume is obstructed or unavailable to inspired gas, and is therefore not detected by washout methodologies.
However, if there were any pulmonary blood flow associated with such a volume, then this would form a shunt blood flow (very low ventilation to perfusion ratio) which would not be consistent with the normal arterial blood gas values that were observed for most participants.
A second possibility is that hospital admission for COVID-19 pneumonia is associated with a loss of lung parenchyma. A third possibility is that small lungs are an additional risk factor for developing severe COVID-19 pneumonia. These last two possibilities really cannot be distinguished, and this illustrates very clearly the difficulties in interpretation that arise when there is an absence of measurements before infection to act as the control values.
Ventilation inhomogeneity can be viewed as the unevenness with which the lung inflates and deflates during inspiration and expiration, and our parameter, σlnCL, provides a standard deviation for this property across the volume of the lung. In the MCOVID cohort, the values for σlnCL were higher for the patients than for the controls.
This reached statistical significance for both the Community and the Ward groups, although not for ICU group where the number of patients studied was considerably fewer. For the C-MORE-LP cohort, there were no significant differences between severity groups, but this may simply reflect the absence of a control group that had no prior infection with COVID-19.
Indeed, once the values from both cohorts had been corrected for the effects of sex, age, height and BMI to reflect values for the standard participant (Fig. 7), the mean values for the non-IMV patients in the C-MORE-LP cohort were extremely similar to those for the MCOVID patients, and some ~0.05 above the value for the MCOVID control group. This is not a large effect when compared with effect sizes associated with chronic airways disease (1, 20).
Nevertheless, in the MCOVID cohort σlnCL increases significantly with age by 0.004 /year (Table 6 E, also see Fig. 6 C), and in relation to this specific measure, infection with COVID-19 could be considered to have aged the lung by ~15 years.
Computed cardiopulmonography (CCP)
This study is the first use of the full CCP methodology (that includes the model of the circulation and body gas stores) in patients. Apart from FRC, there are no equations to provide predicted values for the model parameters based on a person’s physical characteristics. To control for such variations, we initially included a factor for sex, linear terms for age and height and a log linear term for BMI into the mixed effects models, and then subsequently removed them if they were not significant.
This exercise demonstrated that a number of these characteristics were important, in particular height for deadspace, BMI for FRC % predicted and age for σlnCL (Fig. 6).
A particular difficulty with conducting respiratory measurements is that the mouthpiece and nose clip used commonly cause a degree of hyperventilation. The model of the circulation and body gas stores contained within CCP compensates for this, which in turn enables the respiratory data to be used for fitting from the very beginning of the test (although the participant should spend some time sitting quietly before starting the test to ensure that the body’s gas stores are reasonably close to a steady state).
The effects of hyperventilation can be seen in Tables 7 and 8 from the higher values for the directly measured respiratory exchange ratio (RER) for air breathing as compared with the respiratory quotient (R) estimated for the tissues from the model. The difference arises from the extra CO2 eluting from the body’s gas stores.
CCP was developed as a non-invasive technology, and a penalty associated with this is the absence of blood-side measurements relating to gas exchange. In relation to this, Mountain et al (1) noted for the lung that, while the identification of airways parameters (VD, VA, σlnCL, σVD, see Table 1 for definitions) was straightforward, some of the blood-side parameters were much harder to identify. Indeed, Mountain et al adopted the approach of estimating the blood gas concentrations for the pulmonary arterial inflow and the standard deviation for the log distribution for standardised pulmonary vascular conductance, σlnCd, and then estimating a cardiac output based on the oxygen consumption and assuming a value from the literature for the correlation for the bivariate log- normal distribution of compliance and vascular conductance in the lung.
We have followed this approach in the current study, except that the estimates for the blood gas concentrations flowing into the lung have been replaced by the estimates for V·O2, R and PiCO2.
The considerable number (126) of blood gas measurements made during the MCOVID protocol has now provided an opportunity to validate how well this approach to the blood-side parameters worked. From Table 9, it can be seen that mean error between the data and the model estimate for PaCO2 is negligible (<0.1 kPa) showing that the estimate is unbiased, but this difference is somewhat larger for PaO2 (~0.5 kPa).
Similarly, the SD for the difference between measurement and model for PaCO2 was ~0.4 kPa, but was much higher for PaO2 at ~1.6 kPa. The arterial blood gas samples were not drawn slowly (and therefore did not provide a proper average over a number of respiratory cycles) and there is also some uncertainty as to how best to time align the sample with the model data, and both of these factors will contribute to the SD. Nevertheless, the large SD for the PaO2 difference suggests that the uncertainties in the estimates for the blood-side parameters has degraded somewhat the quality of the model’s prediction for the arterial blood gases.
In conclusion, a novel technique, CCP, has been applied to two cohorts of patients that underwent follow-up post COVID-19 infection. The technique revealed an elevated anatomical deadspace in patients who had been admitted to the ICU and/or had undergone IMV as compared with a control group of participants who had either no, or milder, prior infection with COVID-19.
The technique indicated that FRC was lower in patients who had been hospitalised with COVID-19 or who had had more severe disease. Finally, CCP also revealed a greater level of ventilation inhomogeneity (irrespective of severity of acute disease) in patients who had had COVID-19 compared with controls, which was broadly the equivalent of that associated with 15 years of lung aging.
The absence of parameter values prior to infection means it is not possible to discern whether these findings form a set of lung-related risk factors for developing more severe disease or whether they are the direct result of infection with COVID-19 and/or its management.
