Capnodynamic monitoring of lung volume and blood flow in response to increased positive end-expiratory pressure in moderate to severe COVID-19 pneumonia: an observational study

The selection of level of positive end-expiratory pressure (PEEP) during invasive mechanical ventilatory support for COVID-19 pneumonia remains debated. A low level PEEP (≤ 10 cm H₂O) for COVID-19-related acute respiratory failure was supported by a third of experts in a recent Delphi method consensus statement with half of the panel members remaining neutral without agreement on PEEP titration [1]. In contrast, the updated Surviving Sepsis Campaign guidelines for COVID-19 provided a strong recommendation to use high level of PEEP (> 10 cm H₂O) in moderate to severe acute respiratory distress syndrome associated with COVID-19 (C-ARDS) [2]. The static compliance of the respiratory system was highly variable in a large international cohort study of mechanically ventilated COVID-19 patients [3]. This highlights the importance of adequate monitoring to allow for an individualised PEEP strategy as an identical level of PEEP might either lead to lung recruitment or overdistention depending on the compliance state. The gas exchange abnormalities in C-ARDS result from a range of ventilation/perfusion inequalities with further complexity added by diverse changes over time and in different lung regions [4, 5]. The responses to different levels of PEEP can be expected to be equally diverse and should ideally be guided by assessment of lung recruitability and lung perfusion. The effects of incremental PEEP on aerated lung tissue in C-ARDS have been investigated using computed tomography [6, 7], electrical impedance tomography [8, 9] and lung ultrasound [10]. These techniques require both specialised equipment and procedural expertise. In contrast, capnodynamic monitoring of lung volume and perfusion may be integrated with standard ventilators at the bedside and provides continuous measurements without special respiratory manoeuvres or interruptions [11‐14]. This study aimed to assess the feasibility of capnodynamic monitoring of the responses to increased PEEP on gas exchange in mechanically ventilated patients with moderate to severe C-ARDS. It was hypothesised that in patients responding with increased arterial oxygen tension to inspired fraction of oxygen ratio at high PEEP, capnodynamic monitoring would identify an increase in lung volume with increased or preserved pulmonary blood flow.

This pragmatic, prospective, observational open study was approved by the South Western Sydney Local Health District Human Research Ethics Committee (2020/ETH00778) and registered on ClinicalTrials.gov (NCT05082168). Patients admitted to Liverpool Hospital ICU between September 2021 and February 2022 were screened for eligibility with verbal consent from the patient’s person responsible. The study is reported as per the STROBE guidelines for observational studies [15] (Additional File 1: Table S1).

A total of 94 patients with moderate or severe C-ARDS were invasively ventilated in ICU during the study time period with 29 patients enrolled in the study. The study protocol was abandoned in one patient at a P_aO₂/F_iO₂ ratio of 69 with decision to proceed to veno-venous ECMO. Severe hypercarbia precluded achieving an appropriate cyclic ET-CO₂ change for capnodynamic measurements in one patient. Hence, 27 patients were investigated and included in this report. The P_aO₂/F_iO₂ ratio increased > 20 mm Hg from PEEP_low to PEEP_high in 15 patients (responders), while such an improvement was absent in 12 patients (non-responders). The patient characteristics are reported in Table 1. No significant differences between P_aO₂/F_iO₂ responders and non-responders were noted at PEEP_low for gas exchange and pulmonary mechanics (Table 2). The PEEP_high manoeuvre increased P_aO₂ in responders (mean difference 33 [22–44] mm Hg, p < 0.001) with an increase in P_aO₂/F_iO₂ ratio (mean difference 57 [36–78] mm Hg, p = 0.001) (Table 2). The C_rs was greater (mean difference 8.5 [3.2–16] mL/cm H₂O, p = 0.01) in responders at PEEP_high with a corresponding decrease in P_dr (mean difference 6.5 [5.5–11] cm H₂O, p = 0.005) (Table 2).

The \({\text{EELV}}_{{{\text{CO}}_{2} }}\) was less in responders (1286 ± 347 ml) compared to non-responders (1746 ± 599 ml) at PEEP_low (mean difference 486 [88–831] mL, p = 0.01) but not at PEEP_high (1804 ± 462 mL in responders, 2052 ± 652 in non-responders, mean difference 241 [194–682], p = 0.61). These findings remained when \({\text{EELV}}_{{{\text{CO}}_{2} }}\) was indexed to body surface area (data not shown). The \({\text{EELV}}_{{{\text{CO}}_{2} }}\) indexed by PBW was 20 ± 5.7 mL/kg and 26 ± 6.5 mL/kg (mean difference 5.8 [1.0–8.5] mL/kg, p < 0.01) at PEEP_low and 28 ± 6.8 mL/kg and 31 ± 7.7 mL/kg (mean difference 2.7 [− 4.6 to 9.9] ml/kg, p = 0.73) at PEEP_high in responders and non-responders, respectively. The EBPF was not statistically different between non-responders and responders at PEEP_low (4.23 ± 1.67 L/min vs. 4.36 ± 1.59 L/min, p = 0.88) nor at PEEP_high (4.42 ± 1.61 L/min vs. 4.78 ± 1.61 L/min, p = 0.94). The concomitant changes in \({\text{EELV}}_{{{\text{CO}}_{2} }}\) and EPBF to increased PEEP in P_aO₂/F_iO₂ responders demonstrated a positive correlation (r = 0.56 [0.18–0.83], p = 0.03), i.e. recruitment of lung volume was associated with increased pulmonary perfusion (Fig. 1, left graph). In contrast, non-responders demonstrated a negative correlation (r = − 0.65 [− 0.12 to − 0.89], p = 0.02), i.e. an increased lung volume was associated with decreased pulmonary perfusion (Fig. 1, left graph). In P_aO₂/F_iO₂ responders, V_dV_t decreased from PEEP_low (0.43 ± 0.12) to PEEP_high (0.36 ± 0.10) (mean difference − 0.06 [− 0.02 to − 0.09], p = 0.001), while no statistically significant difference was observed in non-responders (PEEP_low 0.40 ± 0.08 vs. PEEP_high 0.43 ± 0.10, p = 0.38). The V_d/V_t responses were further explored by correlating changes in \({\text{EELV}}_{{{\text{CO}}_{2} }}\) and EBPF. For the 20 patients with reduced V_d/V_t, the increase in \({\text{EELV}}_{{{\text{CO}}_{2} }}\) was significantly correlated with increased or maintained EPBF (r = 0.46 [0.04–0.75], p = 0.04), while the correlation in seven patients with increased V_d/V_t failed to attain statistical significance (r = − 0.29, p = 0.53) (Fig. 1, right graph).

The change in \({\text{EELV}}_{{{\text{CO}}_{2} }}\) correlated with the ∆Vol_rec (r = 0.85 [0.69–0.93, p < 0.0001] (Fig. 2) and a positive correlation was demonstrated between the R/I ratio and the change in \({\text{EELV}}_{{{\text{CO}}_{2} }}\) from PEEP_low to PEEP_high (r = 0.87 [0.74–0.94], p ≤ 0.0001) (Fig. 3). The median R/I ratio for all 27 patients was 1.0 and the PEEP induced change in \({\text{EELV}}_{{{\text{CO}}_{2} }}\) below and above the median R/I ratio was 170 ± 198 mL and 578 ± 176 mL, respectively (mean difference 408 [236–546] mL, p < 0.0001).

In this pragmatic, observational open study of capnodynamic monitoring in mechanically ventilated patients with moderate to severe C-ARDS, an improved P_aO₂/F_iO₂ ratio in response to increased PEEP, was associated with increased end-expiratory lung volume and pulmonary perfusion. The change in end-expiratory lung volume was positively correlated with the lung volume recruited and the recruitment-to-inflation ratio. In patients without an improvement in P_aO₂/F_iO₂ ratio, PEEP increased end-expiratory lung volume with a decrease lung perfusion consistent with increased dead space. This study demonstrates the feasibility of capnodynamic monitoring to assess physiological responses to PEEP at the bedside to facilitate an individualised setting of PEEP.

Patients were studied about a week after their COVID-19 diagnosis with the majority developing moderate ARDS. A majority of patients in this study improved the P_aO₂/F_iO₂ > 20 mm Hg in response to increased PEEP. Compared to recent observational reports of PEEP interventions in C-ARDS, the C_rs was similarly low [7, 22] or lower [6, 23, 24] with P_dr correspondingly higher. Patients in this study were class 2 obese with half having a body mass index above 35. This suggests that the prevalence and degree of obesity leading to an increased load on the chest wall should be considered together with the reduced lung compliance associated with C-ARDS. A lung protective ventilation strategy limiting airway pressures was employed including permissive hypercapnia. The associated moderate respiratory acidosis might have aggravated pulmonary vasoconstriction. The changes in \({\text{EELV}}_{{{\text{CO}}_{2} }}\) and EPBF in response to increased PEEP should be interpreted with those characteristics of the study cohort in mind.

The \({\text{EELV}}_{{{\text{CO}}_{2} }}\) at PEEP_low (mean PEEP 8 cm H₂O) was overall similar to the range of end-expiratory lung volumes, 1000–1400 mL, at PEEP 5–8 cm H₂O reported in C-ARDS [6, 7, 22, 24] and non-COVID ARDS [25] using chest computed tomography. The EPBF, that does not include shunt flow, was numerically consistent with a normal cardiac output reflecting the inclusion criterion of haemodynamic stability prior to study procedures. The increased P_aO₂/F_iO₂ in response to PEEP_high was associated with increases in both \({\text{EELV}}_{{{\text{CO}}_{2} }}\) and EPBF and this positive correlation supports an improved ventilation/perfusion matching. The greater \({\text{EELV}}_{{{\text{CO}}_{2} }}\) is consistent with recruitment of previously non-aerated pulmonary tissue that is in line with the concomitant improvement in C_rs and decrease in P_dr. Importantly, a reduced shunt fraction would result in an increased EPBF and this plausibly explains the observed response in gas exchange to PEEP_high. A PEEP_high-induced decrease in cardiac output from the typical hyperdynamic haemodynamic state of C-ARDS [26, 27] would reduce the shunt fraction as would recruitment of previously perfused but not ventilated lung areas. The increase in EPBF could furthermore indicate a maintained or potentially increased cardiac output as PEEP_high reduced pulmonary vascular resistance along with decreased atelectases. A previous study of C-ARDS patients who underwent pulmonary artery catheterisation reported an inverse relation between P_aO₂/F_iO₂ and shunt at both low (5 cm H₂O) and high (15 cm H₂O) PEEP levels without a significant reduction in cardiac output [28]. The improved ventilation/perfusion matching is also supported by the reduced dead space observed in responders. In contrast, patients without a significant improvement of P_aO₂/F_iO₂ in response to PEEP_high demonstrated an increased \({\text{EELV}}_{{{\text{CO}}_{2} }}\) but decreased EPBF. This is consistent with overstretching the lungs, increased pulmonary vascular resistance and right ventricular strain that would reduce pulmonary perfusion. While these changes point to increased dead space, the numerical increase in V_d/V_t and the negative correlation between \({\text{EELV}}_{{{\text{CO}}_{2} }}\) and EPBF failed, however, to attain statistical significance.

The significant correlation between the change in \({\text{EELV}}_{{{\text{CO}}_{2} }}\) and the independently measured ∆Vol_rec in response to PEEP lends support to the validity of capnodynamic monitoring of lung volumes in C-ARDS. The correlation coefficient was similar to that reported between absolute \({\text{EELV}}_{{{\text{CO}}_{2} }}\) and functional residual capacity in a porcine experimental model [29] and superior to that previously reported in anaesthetised patients [14]. Since tidal volumes in this study were kept unchanged from PEEP_low to PEEP_high, the increased \({\text{EELV}}_{{{\text{CO}}_{2} }}\) represents a true recruitment effect. In 6 patients, the \({\text{EELV}}_{{{\text{CO}}_{2} }}\) failed to increase during PEEP_high using a threshold of at least + 10% to consider random measurements error. This represents a lack of alveolar recruitment where additional PEEP contributes to increased lung stress without any benefit in gas exchange. The capacity of capnodynamic monitoring at the bedside to facilitate an individualised setting of PEEP warrants further clinical investigation to evaluate if it can contribute to minimising ventilator induced lung injury in C-ARDS and non-COVID ARDS [30].

Haemodynamic changes may affect \({\text{EELV}}_{{{\text{CO}}_{2} }}\) since CO₂ kinetics are dependent on pulmonary blood flow. Experimental observations, however, demonstrate \({\text{EELV}}_{{{\text{CO}}_{2} }}\) and EPBF as independent factors in the capnodynamic equation [29] in a wide range of cardiac output states. Within this study, three sets of observations were made for patients who progressed to veno-venous extracorporeal membrane oxygenation support (Additional File 1: Fig. S1). The \({\text{EELV}}_{{{\text{CO}}_{2} }}\) remained stable during variable pump flow and native pulmonary perfusion states that corroborates the potential to separately monitor \({\text{EELV}}_{{{\text{CO}}_{2} }}\) and EPBF by the capnodynamic algorithm.

The change in \({\text{EELV}}_{{{\text{CO}}_{2} }}\) was also significantly correlated to alveolar recruitment as indicated by the R/I ratio. The median R/I ratio of 1 was higher compared to other studies reporting a median around 0.7 [10, 31] and higher than the threshold of 0.5 previously used to differentiate poorly from highly recruitable patients in C-ARDS [23, 32] and non-COVID ARDS [19]. Recruitability in acute respiratory failure may be highly variable between patients and over time. In this study, a similar proportion (17/27; 63%) of patients would have been considered highly recruitable by an R/I ratio > 0.5 compared to what has been reported in patients intubated early after ICU admission [23] but higher than that in patients intubated late [32]. Most patients in this study demonstrated low compliance and high recruitability consistent with the high elastance (“H”) phenotype based on recruitability idiosyncratic to C-ARDS [33]. More recent studies have questioned this distinction and instead reported similar patterns in C-ARDS and non-COVID ARDS [34]. Irrespectively, capnodynamic monitoring allowed changes in functional lung volume to be continuously monitored during manoeuvres aimed at alveolar recruitment in C-ARDS.

This study has some important limitations. External validity might be limited by the relatively small sample size, non-consecutive enrolment dependent on availability of the clinical research team and a high proportion of responders to recruitment by increased PEEP. No standard comparators were included for \({\text{EELV}}_{{{\text{CO}}_{2} }}\) or EPBF since this pragmatic study was primarily designed to evaluate the feasibility of capnodynamic monitoring and validation studies have already been published [14, 29]. Levels of PEEP above the PEEP_high might be considered for alveolar recruitment but were not investigated for effects on \({\text{EELV}}_{{{\text{CO}}_{2} }}\) and EPBF. The R/I ratio was not measured from the 15–5 cm H₂O pressure drop as originally described [19] with less of a pressure difference achieved between PEEP_high and PEEP_low. A formal assessment of airway opening pressure was not performed. Visual inspection, however, confirmed progressive, steep increases in both volume and pressure curves from the start of a breath.

This pragmatic, observational open study using capnodynamic monitoring in patients with COVID-19 ARDS demonstrated associations between an improved P_aO₂/F_iO₂ ratio in response to increased PEEP and increased end-expiratory lung volume and pulmonary perfusion. The change in end-expiratory lung volume was positively correlated with independent measures of recruited lung volume and the recruitment-to-inflation ratio. Capnodynamic monitoring is feasible to assess physiological responses to PEEP at the bedside and could facilitate an individualised level of PEEP during mechanical ventilatory support.