Study to Explore the Association of the Renin-Angiotensin System and Right Ventricular Function in Mechanically Ventilated Patients

Background: Right ventricular (RV) dysfunction is associated with pulmonary vasoconstriction in mechanically ventilated patients. Enhancing the activity of angiotensin-converting enzyme 2 (ACE2), a key enzyme of the renin-angiotensin system (RAS), using recombinant human ACE2 (rhACE2) could alleviate RAS-mediated vasoconstriction and vascular remodeling. Methods: This prospective observational study investigated the association between concentrations of RAS peptides (Ang II or Ang(1–7)) and markers of RV function, as assessed by echocardiography (ratio of RV to left ventricular end-diastolic area, interventricular septal motion, and pulmonary arterial systolic pressure (PASP)). Results: Fifty-seven mechanically ventilated patients were enrolled. Incidence rates of acute cor pulmonale (ACP) and pulmonary circulatory dysfunction (PCD) were consistent with previous studies. In the 45 evaluable participants, no notable or consistent changes in RAS peptides concentration were observed over the observation period, and there was no correlation between Ang II concentration and either PASP or RV size. The model of the predicted posterior distributions for the pre- and post-dose values of Ang II demonstrated no change in the likelihood of PCD after hypothetical dosing with rhACE2, thus meeting the futility criteria. Similar results were observed with the other RAS peptides evaluated. Conclusions: Pre-defined success criteria for an association between PCD and the plasma RAS peptides were not met in the mechanically ventilated unselected patients.


Introduction
The role of the renin-angiotensin system (RAS) has been well-described in left ventricular (LV) function, but its role in pulmonary circulatory dysfunction (PCD) and right ventricular (RV) function is not well-known. RV dysfunction may result from excessive increases in RV preload or afterload or decreased contractility, as a result of injury [1]. RV afterload, secondary to increased pulmonary vascular resistance in patients with acute respiratory distress syndrome (ARDS) [2], may result in PCD [3]. PCD, particularly its severe form acute cor pulmonale (ACP), has been associated with poor outcomes in critically ill patients, including those with ARDS [4,5]. Patients with ACP exhibit a higher incidence of shock, increased heart rate, lower arterial pressures, and higher hospital mortality than patients with moderate, or no PCD [4].
Current management approaches for acute RV failure include efforts to reduce RV afterload [1], which is achieved by decreasing pulmonary vascular resistance and increasing pulmonary vasodilation [19]. RAS-mediated vasoconstriction and vascular remodeling may be improved through the dual action of ACE2 by simultaneously degrading Ang II and forming Ang (1)(2)(3)(4)(5)(6)(7). Increasing ACE2 activity/concentrations may, therefore, represent a therapeutic strategy for reducing the likelihood of PCD.
This study was planned to investigate the association of RAS peptides and markers of RV function in participants requiring mechanical ventilation, in order to predict whether the manipulation of the RAS by administration of recombinant human ACE2 (rhACE2) protein would benefit this population. To evaluate the estimated effect of administering rhACE2 on RV outcomes via its effect on RAS peptides, a model was generated, using data from two studies [20,21] to predict the probability of RV dysfunction for pre-determined concentrations of RAS peptides.

Objectives
The primary objective of this study was to evaluate the association between plasma Ang II concentration and RV function in mechanically ventilated participants. Secondary objectives were to: (1) define the incidence of PCD in this cohort and (2) evaluate the association between plasma Ang(1-7) concentration, Ang II/Ang(1-7) ratio, and RV function.

Study Design
This was a prospective, observational, low-interventional study conducted at two centers in France from June 2018 to July 2019 (GlaxoSmithKline plc. study 205821). Participants were enrolled in the study following intubation and mechanical ventilation. Participants were evaluated over a three-day period using standard of care investigations, including transthoracic echocardiography (TTE), and/or transesophageal echocardiography (TOE). Additional investigations were limited to blood samples for RAS peptide analysis ( Figure 1). No investigational product was administered. Clinical interventions were based on local standard of care and international guidelines, including fluid resuscitation, use of vasopressor drugs and renal support, and other biomarkers that were not part of standard of care. Figure 1. Study design and key study features. h, hours; RAS, renin-angiotensin system.

Participants
The study population comprised adults aged 18-80 years with a body mass index (BMI) of 18.0 to 38.0 kg/m 2 receiving invasive mechanical ventilation (duration of ventilation ≤ 48 h) and capable of giving signed informed consent. If participants were not capable of giving signed informed consent, an emergency consent procedure was followed. Participants were excluded from the study if they were moribund, their clinical condition was deteriorating rapidly, or the investigator did not consider there to be a reasonable expectation that the participant would be able to complete the three days of observation in the study. Patients with any of the following were excluded: chronic obstructive pulmonary disease requiring long-term oxygen treatment or home mechanical ventilation, undergoing elective surgery, pre-existing chronic pulmonary hypertension, massive pulmonary embolism or shock, pulmonary vasculitis or pulmonary hemorrhage, lung transplantation in the previous six months, or cardiopulmonary arrest during concurrent illness. Self-reported use of RAS modulators, including angiotensin-converting enzyme (ACE) type 1 inhibitors, renin inhibitors, and angiotensin receptor blockers, within 4 days or 5.5 half-lives, whichever was longer, and a do-not-resuscitate status, also excluded patients from the study.
Endpoints were assessed up to and including Day 3 of observation. Participants were assessed by echocardiography within 48 h of mechanical ventilation (Day 1) and every 24 h for a further two days (Days 2 and 3). TTE was conducted on all participants, while TOE was conducted using a multiplane esophageal probe, when adequate windows could not be obtained using TTE (in 17/57 patients). All echocardiographies were performed by

Participants
The study population comprised adults aged 18-80 years with a body mass index (BMI) of 18.0 to 38.0 kg/m 2 receiving invasive mechanical ventilation (duration of ventilation ≤48 h) and capable of giving signed informed consent. If participants were not capable of giving signed informed consent, an emergency consent procedure was followed. Participants were excluded from the study if they were moribund, their clinical condition was deteriorating rapidly, or the investigator did not consider there to be a reasonable expectation that the participant would be able to complete the three days of observation in the study. Patients with any of the following were excluded: chronic obstructive pulmonary disease requiring long-term oxygen treatment or home mechanical ventilation, undergoing elective surgery, pre-existing chronic pulmonary hypertension, massive pulmonary embolism or shock, pulmonary vasculitis or pulmonary hemorrhage, lung transplantation in the previous six months, or cardiopulmonary arrest during concurrent illness. Self-reported use of RAS modulators, including angiotensin-converting enzyme (ACE) type 1 inhibitors, renin inhibitors, and angiotensin receptor blockers, within 4 days or 5.5 half-lives, whichever was longer, and a do-not-resuscitate status, also excluded patients from the study.
Endpoints were assessed up to and including Day 3 of observation. Participants were assessed by echocardiography within 48 h of mechanical ventilation (Day 1) and every 24 h for a further two days (Days 2 and 3). TTE was conducted on all participants, while TOE was conducted using a multiplane esophageal probe, when adequate windows could not be obtained using TTE (in 17/57 patients). All echocardiographies were performed by trained operators (competent in advanced critical care echocardiography). Images were stored in a digital format on-site, and a computer-assisted consensual interpretation was performed offline by at least two trained senior investigators. The following parameters were used to assess RV function: RV size (measurement of the end-diastolic RV/LV area ratio on a 4 chamber-view), interventricular septal motion in the short-axis view of the heart (presence or absence of dyskinesia), and systolic eccentricity index (the ratio of the LV short-axis diameter parallel to the septum to the LV short-axis diameter perpendicular to the septum) [24]. PASP was estimated using tricuspid regurgitation velocity and application of the modified Bernoulli equation [25].
Blood samples for RAS peptides were collected at the time of the echocardiogram on each day. Additional parameters (including ventilator settings and arterial blood gases) were collected at the time of intubation and daily, as per standard of care. Daily assessment of mechanical ventilation and mortality continued until hospital discharge or Day 28, whichever was sooner. Safety assessments included serious adverse event collection, safety laboratory assessments, and recording of vital signs. A participant was considered to have completed the study if he/she completed Day 3 assessments.
The evaluable population, defined as all participants for whom PCD, Ang II, and Ang(1-7) data were recorded for at least one study time point, was the population used for the primary analyses and majority of the statistical modeling. A population of participants 'at risk' of classification for PCD and ARDS, defined as participants who had a recorded PCD and/or ARDS assessment during the study period, was used for the secondary analysis of the incidences of PCD (and its subtypes) and ARDS. The safety population, defined as all participants for whom at least one echocardiograph and/or blood sample was taken, was the population used for all safety analyses.

Statistical Methods
Instead of modeling the association of Ang II concentrations with the dichotomized RV dysfunction response variable and losing the information from the continuous variables, a Bayesian repeated measures mixed-effects linear regression model was used to model the association of Ang II concentration with PCD. PASP and RV size ratios were fitted as bivariate outcome variables, with Ang II as the explanatory variable (see additional statistical methods in Supplementary Materials).
The model was used to make a prediction of PASP and RV size ratio for two reference values of Ang II, i.e., 30 pg/mL and 8 pg/mL. These values were selected as concentrations that could hypothetically occur before and after dosing with rhACE2 in this patient population, based on RAS peptide analysis by Hemnes et al. [20] and Khan et al. [21]. An estimate for the probability of PCD was derived for each of the given Ang II concentrations. The difference in predicted probabilities was used to assess whether there was sufficient evidence for an association between PCD and Ang II using a priori decision criteria. A percentage point difference >30 between the probabilities was considered successful, percentage point differences between 15 and 30 were considered uncertain, and a percentage point difference <15 was deemed futile. It was proposed that a PCD risk reduction of at least 30%, due to therapeutic intervention with rhACE2, is a large enough effect size to have a reasonable chance of translating to a detectable mortality improvement in future pivotal studies.
The secondary analyses of the association of Ang(1-7) concentration with PCD and of Ang II/Ang(1-7) concentrations with PCD were both modelled per the primary analysis. Hypothetical 'pre-dose' and 'post-dose' reference values for Ang(1-7) concentrations were 2 and 30 pg/mL, respectively, and based on the RAS peptide analysis from the same studies [20,21] as for the Ang II concentrations.
The planned overall sample size of 150 participants was chosen to provide a precision of ±5% points (defined as the half-width of the 95% Wald confidence interval [CI] [26]) around a point estimate for ACP incidence (based on an estimated incidence rate of 10%). However, it was possible to conduct the primary analysis (association between Ang II and PCD, measured by PASP and RV size ratio) without reaching the target precision of ±5% points, when~50 participants had evaluable data. Further information on statistical methods can be found in the Supplementary Materials.

Study Population
During study progression, rhACE2 development was terminated, leading to early study close. A total of 57 participants were enrolled; there were no screen failures ( Figure 2). One participant was withdrawn, due to a protocol deviation (violation of the inclusion criteria for BMI and age). Demographics and clinical characteristics of the participants were similar between patients with and without PCD (Table 1). Most participants (n = 40/57; 70%) were managed with vasopressors, inotropes, and other vasoactive agents at baseline. The most common reasons for intubation were acute respiratory failure (n = 25/57; 44%), followed by sepsis (n = 10/57; 18%) and impaired neurological status or post-surgical management (n = 16/57; 28%). Baseline echocardiography did not reveal significant chronic RV dysfunction. Respiratory variables and catecholamine support at each visit are reported in Supplemental Tables S2 and S3, respectively, in Supplemental Material.

Study Population
During study progression, rhACE2 development was terminated, leading to early study close. A total of 57 participants were enrolled; there were no screen failures ( Figure 2). One participant was withdrawn, due to a protocol deviation (violation of the inclusion criteria for BMI and age). Demographics and clinical characteristics of the participants were similar between patients with and without PCD (Table 1). Most participants (n = 40/57; 70%) were managed with vasopressors, inotropes, and other vasoactive agents at baseline. The most common reasons for intubation were acute respiratory failure (n = 25/57; 44%), followed by sepsis (n = 10/57; 18%) and impaired neurological status or post-surgical management (n = 16/57; 28%). Baseline echocardiography did not reveal significant chronic RV dysfunction. Respiratory variables and catecholamine support at each visit are reported in Supplemental  Tables S2 and S3, respectively, in Supplemental Material. Figure 2. Participant flow. * Patients were pre-screened for eligibility; however, the number of prescreened patients is unavailable. a Screened population: all participants who were screened. Safety population: all participants for whom at least one echocardiograph and/or blood sample was taken. At-risk population: participants for whom PASP and RV size ratio were recorded for all three study days and/or had a databased PCD and/or ARDS assessment during the study period. b Evaluable population: all participants for whom PASP, RV size ratio, Ang II, and Ang(1-7) data were recorded for at least one study time point. c Completed population: all participants for whom PASP, RV size ratio, Ang II, and Ang(1-7) data were recorded for all three study days. ACP, acute cor pulmonale; Ang, angiotensin; ARDS, acute respiratory distress syndrome; PASP, pulmonary arterial systolic pressure; PCD, pulmonary circulatory dysfunction; RV, right ventricle.  Safety population: all participants for whom at least one echocardiograph and/or blood sample was taken. At-risk population: participants for whom PASP and RV size ratio were recorded for all three study days and/or had a databased PCD and/or ARDS assessment during the study period. b Evaluable population: all participants for whom PASP, RV size ratio, Ang II, and Ang(1-7) data were recorded for at least one study time point. c Completed population: all participants for whom PASP, RV size ratio, Ang II, and Ang(1-7) data were recorded for all three study days. ACP, acute cor pulmonale; Ang, angiotensin; ARDS, acute respiratory distress syndrome; PASP, pulmonary arterial systolic pressure; PCD, pulmonary circulatory dysfunction; RV, right ventricle.

Association between Plasma Ang II Concentration and PCD
A total of 45 participants were included in the evaluable population for the primary analysis. Although the individual time-profile plots of Ang II demonstrated that most participants had Ang II concentrations above those seen in healthy participants (physiological concentrations of 8 pg/mL [27]), there were no notable or consistent changes in Ang II concentration over time (Figure 3). Individual PASP and RV model predictions were similar at the hypothetical Ang II concentrations before and after 'dosing' with rhACE2 (30 pg/mL and 8 pg/mL, respectively) ( Figure 4). The probability of PCD, given an Ang II concentration of 30 pg/mL (hypothetical 'pre-dose' value), was estimated to be 93.1%; the probability of PCD, given an Ang II concentration of 8 pg/mL (hypothetical 'post-dose' value), was estimated to be 95.4% ( Figure 5A). As regression gradients were flat, the difference (−2.4%) in predicted probability was likely an artefact of noise. The difference was compared with the pre-defined criteria. It did not exceed 15% and, therefore, met the futility criterion. The observed patient-level data of PASP versus RV size ratio, marked by interventricular septal motion status, are shown in Figure 5B. The statistical model was used to predict the probability of PCD for a given concentration of Ang II ( Figure 5C). Subsequently, the model-predicted posterior distributions for the hypothetical 'pre-dose' and 'post-dose' concentrations of Ang II were generated, in which a large overlap between the distributions was observed ( Figure 5D).       Figure 4A contains additional sampling points that did not have an associated Ang II concentration value. Ang, angiotensin; N, no; ND, not done; PASP, pulmonary arterial systolic pressure; PCD, pulmonary circulatory dysfunction; Pr, probability; Y, yes. Ang(1-7) Concentration, Ang II/ Ang(1-7) Ratio, and PCD Data from the 45 participants in the evaluable population were used for this analysis. Consistent with the Ang II results, the time profile of the Ang(1-7) and Ang II/Ang(1-7) ratio showed no notable or consistent changes in Ang(1-7) concentration or Ang II/Ang(1-7) ratio over the study period (Figures 3 and S1 in Supplementary Materials). The model  Figure 4A contains additional sampling points that did not have an associated Ang II concentration value. Ang, angiotensin; N, no; ND, not done; PASP, pulmonary arterial systolic pressure; PCD, pulmonary circulatory dysfunction; Pr, probability; Y, yes.

Association between Plasma Ang(1-7) Concentration, Ang II/Ang(1-7) Ratio, and PCD
Data from the 45 participants in the evaluable population were used for this analysis. Consistent with the Ang II results, the time profile of the Ang(1-7) and Ang II/Ang(1-7) ratio showed no notable or consistent changes in Ang(1-7) concentration or Ang II/Ang(1-7) ratio over the study period (Figure 3 and Figure S1 in Supplementary Materials). The model predicted that increasing Ang(1-7) from 2 to 30 pg/mL would reduce the probability of PCD by~5.7 percentage points ( Figure 6). Although there was a supportive trend towards a decreased likelihood of dysfunction with increased Ang(1-7) concentration, the difference in predicted probabilities between pre-and post-dose did not exceed 15% and, thus, met the futility criterion.

Incidence of PCD
The incidence rates of PCD and its subtypes (including ACP and severe ACP) were measurable in 57 participants; therefore, data from all 57 participants were used in this secondary analysis. A summary of PCD subtypes and ARDS incidence rates is presented in Table 2.
Safety data are summarized in the Supplementary Materials.

Discussion
Few studies have evaluated the association between the RAS and PCD in humans. The objective of this study was to investigate the association of RAS peptides and markers of PCD in participants requiring mechanical ventilation, as well as to further characterize this patient population. The PCD incidence rates seen in this study were in line with those observed previously in a similar population [5]. Overall, we observed no consistent changes in Ang II or Ang(1-7) concentration over the observation period. Despite a significant range of Ang II concentrations, indicating the RAS was activated in many patients, the results did not demonstrate a relationship between Ang II concentrations and either PASP or RV size. The same conclusions applied to Ang (1)(2)(3)(4)(5)(6)(7).
It is interesting to compare our results to the ones published by Khan et al. in patients with moderate ARDS [21]. While our population of 45 mechanically ventilated patients is quite different and more heterogeneous, with only 26% of ARDS, we finally found results similar to the 20 patients randomized in their control group. AngII was indeed significantly elevated, had a large distribution of values, and little change over the first days following the inclusion, as was also reported by Reddy et al. [18]. This could suggest that more than the ARDS, per se, it is the need for mechanical ventilation that mainly induces RAS peptides alterations.
Most participants had Ang II concentrations above those seen in healthy participants, and no substantial changes in Ang II concentration over time were observed. A previous study of patients with heart failure demonstrated that, despite ACE inhibitor therapy, Ang II concentrations remained persistently higher than those seen in healthy individuals in half of the study population [28]. Experimental studies suggest that Ang II plays an important role in ventricular remodeling [28]. Ang II may provide a positive feedback loop in vascular remodeling by upregulating the expression of pro-fibrotic factors, thus resulting in further production of Ang II, therefore perpetuating vascular remodeling [29].
A Bayesian repeated measures mixed-effects regression model was used to explore the association of Ang II with PCD, based on prediction of PASP and RV size ratio, according to hypothetical Ang-II concentrations pre-(30 pg/mL) and post-(8 pg/mL) dosing with rhACE2. If high Ang II concentrations were associated with PCD and/or low Ang II concentrations were associated with no PCD, this would provide a supportive rationale for the use of rhACE2 as a potential treatment in this patient population, as lowering Ang II concentrations with rhACE2 could reduce the likelihood of PCD. The model of predicted posterior distributions for the pre-and post-dose values showed a large overlap between the distributions, thus indicating no change in the likelihood of PCD as Ang II concentrations decreased from 30 to 8 pg/mL. These data suggest that a reduction in Ang II concentrations after treatment with rhACE2 would not have a clinically significant impact on PCD in this population. The probability of PCD, given an Ang II concentration at the hypothetical 'pre-dose' value, was high and similar to the 'post-dose' value. The difference between the predicted probabilities met the futility criterion, indicating that a future study to evaluate rhACE2 as a treatment in patients with severe respiratory failure would likely be unsuccessful. The high probability of PCD at the post-dose Ang II concentration was surprising. These results could suggest that the association of RAS peptide levels with PCD may have been confounded by the impact of mechanical ventilation, in particular the use of high positive end-expiratory pressure on RAS peptide concentrations. In fact, transpulmonary pressure may have a direct effect on the compression of pulmonary vessels [30]. Other factors may also be involved in PCD during ARDS, including hypercapnia [31,32]. Therapeutic dosing of Ang II has recently been evaluated as a vasopressor treatment for distributive shock, with evidence of efficacy [33] that led to approval by the US Food and Drug Administration [34], further highlighting the complex interaction between the RAS and both systemic and pulmonary vascular function. Most mechanically ventilated patients were ventilated for acute respiratory failure, and mechanical ventilation itself may have induced PCD in the study population, especially when associated with low partial pressure of oxygen (PaO 2 ). Similar results were observed with the other RAS peptides evaluated; no association between the PCD and Ang (1-7) concentrations or Ang II/Ang (1-7) ratio was observed.
Consistent with our study, a systematic review on the effects of ACE inhibitors, Ang II receptor blockers, and aldosterone antagonists in adults with congenital heart disease and RV dysfunction found that RAS inhibition did not have a beneficial effect in these patients, suggesting a lack of association between RAS peptides and RV function. However, the studies included in the review generally had low patient numbers, short follow-up periods, and evaluated surrogate endpoints [16]. Clinical studies of rhACE2 in patients with pulmonary arterial hypertension have demonstrated rapid modulation of RAS peptides but inconsistent changes in clinical outcomes, including pulmonary hemodynamics [35]. Similarly, in a meta-analysis investigating the role of RAS inhibition on RV function, the treatment arm (in which patients were treated with ACE inhibitors and/or angiotensin receptor blockers) showed a trend towards increased RV ejection fraction, compared with the control arm, suggesting a potential role of the RAS pathway in RV dysfunction. However, as with our present study, the results were not significant, indicating that further, larger prospective studies are needed to further elucidate the role of the RAS in RV function [13].

Strengths and Limitations
We conducted this prospective, observational study with low patient intervention to determine whether an investigational product could benefit a patient population. Without exposing patients to a potentially non-beneficial investigational drug, we conceptually determined that an interventional study would not be successful. A broad population of mechanically ventilated participants in critical care were enrolled, increasing the potential generalizability of the study results. However, the number of pre-screened patients was unavailable, and investigating particular patient subgroups may have yielded more specific results on the role of RAS peptides in the pathophysiology of severe respiratory failure based on specific diagnoses. It is possible the heterogeneity of the patient population confounded the ability to observe a detectable relationship between the RAS peptide levels and PCD, as RAS is known to be influenced by physiological characteristics, such as BMI, vascular tone, renal function, and by co-medications. COVID-19-related ARDS may be a relevant target population, given the burden of lung vascular dysfunction in this subgroup [36,37]. This was a small double-centre study, which may limit the generalizability of the findings. The study was limited in its feasibility; obtaining all data from each patient was challenging. As 21 patients were excluded for missing data (because PCD, Ang II, and Ang(1-7) data were not available for at least one study time point), this may have induced selection bias in the results and precludes any definite conclusion. A further limitation was that PASP was estimated by echocardiography as a surrogate measure, rather than as a direct measurement using invasive measures, which is more robust than echocardiography.

Conclusions
The results of this study did not produce sufficient evidence to meet pre-defined success criteria regarding evidence for an association between PCD in mechanically ventilated patients and plasma RAS peptides Ang II, Ang(1-7), or Ang II/Ang(1-7) ratio. Further studies of specific patient subgroups (such as patients with ARDS and patients with a high-risk score for developing ACP) could help further elucidate the role of RAS peptides in these patients.