The first row in Figures 1A–C shows the time course of Crs, PEEP, and Pplat in two representative patients: survivors (S, closed symbols) and non-survivors (NS, open symbols). In both patients, Crs decreased over time to a similar extent, although the Crs values in NS were lower (Panel A). A clear dissociation is seen in panel B, as PEEP increased in NS while decreasing in S. Panel C reports the Pplat values that increased over time in NS but remained essentially steady in S.

Panel D in the second row of Figure 1 shows the pressure-volume relationship of the respiratory system (Agostoni and Mead, 1964), which is expressed as a percentage of the maximum. The maximum volume decreased with decreasing Crs, reflecting a decrease in inflatable alveolar units (IAU). If the mechanical properties of the residual IAU remained unchanged, the curve in Figure 1D reflects the specific compliance of the IAU. Based on this assumption, the ordinate can be used to express the corresponding degree of lung distension of the IAU as a function of the alveolar pressure. The light grey area includes the portion of the pressure-volume curve with the highest specific compliance, extending from 45% to 70% lung distension, corresponding to a range of alveolar pressures from approximately 10–20 cmH₂O. In this range of pressures, the process of unfolding of the alveolar surface takes place on inspiration, reflecting the existence of a “reserve” surface area of the corrugated alveolar cells (Weibel, 2015). As the unfolding process develops, the parenchymal stretch gradually increases (from the light to the darker grey area), indicating lung overdistension. Under physiological conditions at rest, an increase in tidal volume during spontaneous breathing is achieved by an increase in transpulmonary pressure of approximately 5 cmH₂O; accordingly, the same tidal volume in mechanical ventilation would be achieved by an alveolar pressure of approximately 5 cmH₂O, corresponding to 30% lung distension, well below the saturation of the unfolding zone. Panels E and F show the degree of lung distention at PEEP and Pplat, respectively, for the two subjects. In the case of NS, lung distension at PEEP falls in the light grey area, whereas at Pplat, lung distension falls in the overdistension zone for both patients.

Panel A in Figure 2 shows that the time course of P/F significantly decreased in both patients. As shown in Panel B, no significant differences in SatO₂ were observed. Panels C and E again show a divergence in the time courses of FIO₂ and PaCO₂, despite displaying a similar time course for PaO₂ and VE/kg (Panels D and F, respectively).

Figure 3 (Panel A) shows the relationships of P/F plotted vs. Crs; for a decrease in Crs, P/F increased in patient S but decreased in NS. Panel B shows a decrease in P/F with increasing PEEP in patient NS; conversely (Panel C), P/F increased in S with decreasing PEEP.

Figure 4 shows the time course of tidal volume (Vt) and respiratory rate (RR) in S and NS patients (Panels A and B). By decreasing PEEP (Figure 1B), a higher driving pressure can increase Vt in S patient. Conversely, the opposite occurred in NS patient due to an increase in PEEP (Figure 1B), particularly considering the greater decrease in lung compliance (Figure 1A). Furthermore, it should be noted that in NS patients, the Vt approaches the anatomical dead space. The RR (Panel B) remained high for both subjects.

Concerning the ventilatory ratio, a clear dependence on ventilation is observed, albeit in opposite directions, considering S and NS patients (panel C). Panel D shows the striking dependence of the ventilatory ratio on PaCO₂; in the case of patient S, PaCO2 remained within the physiological range, while it increased remarkably in patient NS.

An increase in alveolar pressure leads to a decrease in capillary blood volume (Vc), owing to the squeezing of capillaries (Figure 5D) caused by an increase in parenchymal stretching (Glazier et al., 1969; Brower et al., 1985; Nieman et al., 1988; Koyama and Hildebrandt, 1991; Miserocchi et al., 2008). This decrease was found to vary remarkably among subjects, depending on the individual morphofunctional assembly of the alveolar capillary unit. The latter is characterised by the ratio of Vc to the diffusion capacity of the alveolar membrane (Vc/Dm), which essentially compares inter-individual differences in the extension of the pulmonary alveolar capillary network to the alveolar size (Miserocchi et al., 2008). The present available data do not allow for the estimation of inter-individual differences in Vc/Dm among patients. Accordingly, Figure 6 shows three cases of Vc ranging at Functional Residual Capacity (FRC) from 150 to 300 mL (corresponding to different Vc/Dm ratios). The figure reports the expected decrease in Vc with increasing lung distension with a PEEP of 5 and 15 cmH₂O (Miserocchi et al., 2008): clearly, the decrease in Vc (in absolute terms) is larger the greater the Vc value at FRC (Figure 5D).

Interestingly, a decrease in pulmonary blood volume has been documented in post COVID-19 through Dual-energy CT scan not only in opacification areas but also in parenchyma of normal appearance in acute (Aydin et al., 2021; Ball et al., 2021) and post-acute phase (Mohamed et al., 2023) (Figure 5E), suggesting a potential limitation to perfusion.

We have developed a model to estimate the dependence of alveolar gas exchange resulting from the functional coupling of blood capillary flow with gas diffusion flows (Beretta et al., 2019; Miserocchi et al., 2022; Miserocchi, 2023b). Our present aim is to rely on this model to compare two distinct conditions: hyperoxia and normocapnia in S patients, against hyperoxia and hypercapnia in NS patients. We shall briefly summarize the principles of the biophysical model.

Based on the gas mass conservation notion (Piiper and Scheid, 1981) and an exponential kinetics of the equilibration process, the alveolar-capillary equilibration for gas exchange reached at the exit of the blood from the capillary is mathematically defined as: L e q = e −   T t τ (1) being Tt blood capillary transit time (also known as “capillary residence time” or “blood contact time”), and τ is the time constant of the exponential kinetics. Tt is the key parameter to switch from volumes to flows and can be estimated as the ratio of pulmonary blood capillary volume (Vc) to cardiac output ( Q ˙ ): T t = V c Q ´ (2)

The kinetics of the equilibration is defined by the time constant given by: τ = β V c D O 2  for  O 2 (3) and τ = α V c D C O 2  for  CO 2 (4) being DO ₂ and DCO ₂ , the respective diffusive capacities, while β and α include gas solubility and transport capacity in blood.

Leq can vary from 0 (perfect equilibration) to 1 (total lack of equilibration) (Beretta et al., 2019; Miserocchi et al., 2022; Miserocchi, 2023b).

This approach provided supplementary information to the classic V ˙ A / Q ˙ distribution (Wagner, 2008; Glenny and Robertson, 2011; Hopkins, 2020). Defining the kinetics of gas exchange equilibration that includes the estimate of the blood capillary transit time allows to develop the concept of “shunt-like effect” reflecting the decrease in V c Q ´ ratio. The latter was found to vary considerably among subjects, reflecting the heterogeneity (Miserocchi et al., 2008; Miserocchi, 2023b) of inborn morpho-functional arrangement of the air blood barrier as well as the individual response to functional conditions (lung stretching, hypoxia, increase in oxygen demand) (Miserocchi and Beretta, 2023).

The key issue on comparing S with NS is that both groups had a SatO₂>90%; however, while the former remained normocapnic, the latter developed hypercapnia.

Diffusion limitation is a specific case occurring for oxygen, due to its low solubility-diffusion coefficients (Eq. 3).

Figure 7 shows that under physiological conditions, Leq = 0 at the exit of the pulmonary capillaries.

The development of interstitial and severe alveolar edema (Figures 5B, C) represent the obvious case of diffusion limitation for O₂ uptake, due to the decrease in DO ₂ . As shown in Figure 7, an exponential loss of equilibration capacity occurred with a 5-times decrease in DO ₂ , compatible with the observed average decrease in Crs in patients relative to a physiological value of approximately 100 mL/cmH₂O. The loss of O₂ equilibration capacity for the whole lung simulates a “shunt-like effect.”

Oxygen diffusion limitation is commonly compensated for by an increase in F_IO₂. One shall report that for FIO₂ > 0.7 (Aggarwal et al., 2018), cellular (Kistler et al., 1967; Weibel, 1971) and tissue (Chow et al., 2003; Kallet and Matthay, 2013) damage in the lungs were reported, leading to increased alveolar permeability (Matalon and Egan, 1981; Kolliputi et al., 2010).

Conversely, a diffusion limitation is hardly conceivable for CO₂ exchange considering its high solubility-diffusion coefficients. Accordingly, one can develop the hypothesis of perfusion limitation.

The aim of our study is to find a cause-effect relationship for developing hypercapnia. Figure 8 presents a computational (Eq. 1) estimate of the exponential increase in perfusion limitation for CO₂ removal for Tt < 1 s by decreasing Vc (Eq. 2) (Figures 5D, E).

Besides a decrease in Vc due to pulmonary stretching, a further factor arises from tissue compression in developing oedema (Figures 5B, C), that could actually lead to complete vessel closure (Mazzuca et al., 2019), functionally equivalent to the case of thrombosis (Figure 5E). Furthermore, studies on ECMO have confirmed that CO₂ removal is hampered by low blood flow (Karagiannidis et al., 2017; Giraud et al., 2021; Zanella et al., 2022).

An estimate of the Starling pressure gradient controlling the lung fluid balance, particularly alveolar pressure, is presented in Figure 9. We accounted for the hydraulic and colloidal osmotic pressure, reflection coefficient, and alveolar surface tension (Beretta et al., 2021).

Data in Figure 9 show that, for Palv>5 cmH₂O, the transendothelial Starling gradient favours microvascular filtration and thus is an edemagenic factor (Miserocchi et al., 1993).

Lung overdistension has been found to increase microvascular filtration (Miserocchi et al., 1991), a finding confirmed by a computational model showing stress-dependent leak progression through an epithelial monolayer (Hamlington et al., 2016; Hamlington et al., 2018).

Notably, a decrease in Tt, resulting in an increase in blood velocity, leads to an increase in shear rate (Miserocchi et al., 2022), which in turn causes the increase in microvascular and protein permeability, thus favouring oedema (Sill et al., 1995; Lakshminarayanan et al., 2000; Mazzag et al., 2003; Barakat et al., 2006; Kang et al., 2014; Kolářová et al., 2014; Miserocchi and Beretta, 2023).

We have to acknowledge some study limitation. We did not have any specific exclusion criteria. However, we did not have information on the screening data of ICU admission, but this is a convenient sample size of patients admitted to ICU of the Arequipa Hospital in Perù with a clinical diagnosis of respiratory failure with a positive PCR confirmation of Sars-CoV2 infection that were enrolled from April 2020 to March 2021. This study is not designed to evaluate independent association of clinical variables on outcome by using multivariable models of association but aims at exploring the physiopathology of gas exchange in the air-blood barrier during acute respiratory failure. We relied on established computational physio-pathological models.

This paper deals with diffusion and perfusion limitation to alveolar gas exchanges in mechanically ventilated COVID-19 patients.

In the representative NS patient, P/F decreased in the first week of ICU stay (Figure 2A), clearly reflecting oxygen diffusion limitation compensated by an increase in PEEP and FIO₂ (Figures 1B, 2C, respectively). This was in accordance to the PEEP:FIO₂ tables (Brower et al., 2004) and the guidelines (Fan et al., 2017), the leading idea being to increase alveolar recruitment and ventilation to favor oxygen uptake, although a steady SatO₂>90%. However, the increase in PEEP may contribute to a progressive increase in PaCO₂ (Reazoagli and Bellani, 2022), due to perfusion limitation, hindering CO₂ removal (Figure 2E) and leading to a remarkable increase in the ventilatory ratio – which -in turn – is a marker per se of the respiratory failure severity (Figure 4D).

In representative S patient, the increase in P/F over the first week (Figure 2A) led to the decision (Brower et al., 2004) to decrease both PEEP and FIO₂ (Figures 1B, 2C). Consequently, this allowed ventilation of patient S with a progressively lower mean airway pressure, favoring CO₂ removal, notwithstanding the potential for alveolar de-recruitment.

In NS group, the ventilatory strategy led to SatO₂ >90% coupled with severe hypercapnia. The latter has been considered as a biomarker of increased dead space due to perfusion limitation, a condition associated independently with a high risk of mortality (Nuckton et al., 2002).

In fact, survivors or non-survivors are separated by a faint border considering that in both groups, lung compliance was decreased by the disease to approximately 1/5 of normal, meaning that the total number of alveolar units assuring gas exchange was decreased from a physiological value of approximately 500 (Ochs et al., 2004) to 100 million. One cannot exclude an overestimate of the decrease in Crs in the study population considering the possible presence of auto-PEEP, which may occur with an increasing respiratory rate (Marini, 2011). In severe cases of mechanically ventilated patients, a 70%–80% reduction in DLCO relative to the expected normal value was reported at 5–12 months (Krueger et al., 2023; van Willigen et al., 2023). Radiological pulmonary abnormalities have been described more than 100 days after the diagnosis of COVID-19 (Sonnweber et al., 2021). It appears reasonable to relate these decreased variables to pulmonary fibrosis development (Figure 5).

Several parameters obviously impact on the efficiency of gas exchanges in the air blood barrier.

The time dependent analysis that we performed allows an integrated view coherent with the parameters considered, highlighting the time dependence of gas exchanges in the air blood barrier, being diffusion limited for O₂ and perfusion limited for CO₂.