Clinical and experimental validation of a capnodynamic method for end-expiratory lung volume assessment

Monitoring CO2 kinetics as a marker of cardiopulmonary efficiency

PURPOSE OF REVIEW

To describe current and near future developments and applications of CO2 kinetics in clinical respiratory and cardiovascular monitoring.

RECENT FINDINGS

In the last years, we have witnessed a renewed interest in CO2 kinetics in relation with a better understanding of volumetric capnography and its derived parameters. This together with technological advances and improved measurement systems have expanded the monitoring potential of CO2 kinetics including breath by breath continuous end-expiratory lung volume and continuous noninvasive cardiac output. Dead space has slowly been gaining relevance in clinical monitoring and prognostic evaluation. Easy to measure dead space surrogates such as the ventilatory ratio have demonstrated a strong prognostic value in patients with acute respiratory failure.

SUMMARY

The kinetics of carbon dioxide describe many relevant physiological processes. The clinical introduction of new ways of assessing respiratory and circulatory efficiency based on advanced analysis of CO2 kinetics are paving the road to a long-desired goal in clinical monitoring of critically ill patients: the integration of respiratory and circulatory monitoring during mechanical ventilation.

Clinical validation of a capnodynamic method for measuring end-expiratory lung volume in critically ill patients

RATIONALE

End-expiratory lung volume (EELV) is reduced in mechanically ventilated patients, especially in pathologic conditions. The resulting heterogeneous distribution of ventilation increases the risk for ventilation induced lung injury. Clinical measurement of EELV however, remains difficult.

OBJECTIVE

Validation of a novel continuous capnodynamic method based on expired carbon dioxide (CO2) kinetics for measuring EELV in mechanically ventilated critically-ill patients.

METHODS

Prospective study of mechanically ventilated patients scheduled for a diagnostic computed tomography exploration. Comparisons were made between absolute and corrected EELVCO2 values, the latter accounting for the amount of CO2 dissolved in lung tissue, with the reference EELV measured by computed tomography (EELVCT). Uncorrected and corrected EELVCO2 was compared with total CT volume (density compartments between - 1000 and 0 Hounsfield units (HU) and functional CT volume, including density compartments of - 1000 to - 200HU eliminating regions of increased shunt. We used comparative statistics including correlations and measurement of accuracy and precision by the Bland Altman method.

MEASUREMENTS AND MAIN RESULTS

Of the 46 patients included in the final analysis, 25 had a diagnosis of ARDS (24 of which COVID-19). Both EELVCT and EELVCO2 were significantly reduced (39 and 40% respectively) when compared with theoretical values of functional residual capacity (p < 0.0001). Uncorrected EELVCO2 tended to overestimate EELVCT with a correlation r2 0.58; Bias - 285 and limits of agreement (LoA) (+ 513 to - 1083; 95% CI) ml. Agreement improved for the corrected EELVCO2 to a Bias of - 23 and LoA of (+ 763 to - 716; 95% CI) ml. The best agreement of the method was obtained by comparison of corrected EELVCO2 with functional EELVCT with a r2 of 0.59; Bias - 2.75 (+ 755 to - 761; 95% CI) ml. We did not observe major differences in the performance of the method between ARDS (most of them COVID related) and non-ARDS patients.

CONCLUSION

In this first validation in critically ill patients, the capnodynamic method provided good estimates of both total and functional EELV. Bias improved after correcting EELVCO2 for extra-alveolar CO2 content when compared with CT estimated volume. If confirmed in further validations EELVCO2 may become an attractive monitoring option for continuously monitor EELV in critically ill mechanically ventilated patients.

TRIAL REGISTRATION

clinicaltrials.gov (NCT04045262).

Capnodynamic end-expiratory lung volume assessment in anesthetized healthy children

BACKGROUND

Capnodynamic lung function monitoring generates variables that may be useful for pediatric perioperative ventilation.

AIMS

Establish normal values for end-expiratory lung volume CO2 in healthy children undergoing anesthesia and to compare these values to previously published values obtained with alternative end-expiratory lung volume methods. The secondary aim was to investigate the ability of end-expiratory lung volume CO2 to react to positive end-expiratory pressure-induced changes in end-expiratory lung volume. In addition, normal values for associated volumetric capnography lung function variables were examined.

METHODS

Fifteen pediatric patients with healthy lungs (median age 8 months, range 1-36 months) undergoing general anesthesia were examined before start of surgery. Tested variables were recorded at baseline positive end-expiratory pressure 3 cmH2 O, 1 and 3 min after positive end-expiratory pressure 10 cmH2 O and 3 min after returning to baseline positive end-expiratory pressure 3 cmH2 O.

RESULTS

Baseline end-expiratory lung volume CO2 was 32 mL kg-1 (95% CI 29-34 mL kg-1 ) which increased to 39 mL kg-1 (95% CI 35-43 mL kg-1 , p < .0001) and 37 mL kg-1 (95% CI 34-41 mL kg-1 , p = .0003) 1 and 3 min after positive end-expiratory pressure 10 cmH2 O, respectively. End-expiratory lung volume CO2 returned to baseline, 33 mL kg-1 (95% CI 29-37 mL kg-1 , p = .72) 3 min after re-establishing positive end-expiratory pressure 3 cmH2 O. Airway dead space increased from 1.1 mL kg-1 (95% CI 0.9-1.4 mL kg-1 ) to 1.4 (95% CI 1.1-1.8 mL kg-1 , p = .003) and 1.5 (95% CI 1.1-1.8 mL kg-1 , p < .0001) 1 and 3 min after positive end-expiratory pressure 10 cmH2 O, respectively, and 1.2 mL kg-1 (95% CI 0.9-1.4 mL kg-1 , p = .08) after 3 min of positive end-expiratory pressure 3 cmH2 O. Additional volumetric capnography and lung function variables showed no major changes in response to positive end-expiratory pressure variations.

CONCLUSIONS

Capnodynamic noninvasive and continuous end-expiratory lung volume CO2 values assessed during anesthesia in children were in close agreement with previously reported end-expiratory lung volume values generated by alternative methods. Furthermore, positive end-expiratory pressure changes resulted in physiologically expected end-expiratory lung volume CO2 responses in a timely manner, suggesting that it can be used to trend end-expiratory lung volume changes during anesthesia.

Effect of an Individualized Lung Protective Ventilation on Lung Strain and Stress in Children Undergoing Laparoscopy: An Observational Cohort Study

BACKGROUND

Exaggerated lung strain and stress could damage lungs in anesthetized children. The authors hypothesized that the association of capnoperitoneum and lung collapse in anesthetized children increases lung strain-stress. Their primary aim was to describe the impact of capnoperitoneum on lung strain-stress and the effects of an individualized protective ventilation during laparoscopic surgery in children.

METHODS

The authors performed an observational cohort study in healthy children aged 3 to 7 yr scheduled for laparoscopic surgery in a community hospital. All received standard protective ventilation with 5 cm H2O of positive end-expiratory pressure (PEEP). Children were evaluated before capnoperitoneum, during capnoperitoneum before and after lung recruitment and optimized PEEP (PEEP adjusted to get end-expiratory transpulmonary pressure of 0), and after capnoperitoneum with optimized PEEP. The presence of lung collapse was evaluated by lung ultrasound, positive Air-Test (oxygen saturation measured by pulse oximetry 96% or less breathing 21% O2 for 5 min), and negative end-expiratory transpulmonary pressure. Lung strain was calculated as tidal volume/end-expiratory lung volume measured by capnodynamics, and lung stress as the end-inspiratory transpulmonary pressure.

RESULTS

The authors studied 20 children. Before capnoperitoneum, mean lung strain was 0.20 ± 0.07 (95% CI, 0.17 to 0.23), and stress was 5.68 ± 2.83 (95% CI, 4.44 to 6.92) cm H2O. During capnoperitoneum, 18 patients presented lung collapse and strain (0.29 ± 0.13; 95% CI, 0.23 to 0.35; P < 0.001) and stress (5.92 ± 3.18; 95% CI, 4.53 to 7.31 cm H2O; P = 0.374) increased compared to before capnoperitoneum. During capnoperitoneum and optimized PEEP, children presenting lung collapse were recruited and optimized PEEP was 8.3 ± 2.2 (95% CI, 7.3 to 9.3) cm H2O. Strain returned to values before capnoperitoneum (0.20 ± 0.07; 95% CI, 0.17 to 0.22; P = 0.318), but lung stress increased (7.29 ± 2.67; 95% CI, 6.12 to 8.46 cm H2O; P = 0.020). After capnoperitoneum, strain decreased (0.18 ± 0.04; 95% CI, 0.16 to 0.20; P = 0.090), but stress remained higher (7.25 ± 3.01; 95% CI, 5.92 to 8.57 cm H2O; P = 0.024) compared to before capnoperitoneum.

CONCLUSIONS

Capnoperitoneum increased lung strain in healthy children undergoing laparoscopy. Lung recruitment and optimized PEEP during capnoperitoneum decreased lung strain but slightly increased lung stress. This little rise in pulmonary stress was maintained within safe, lung-protective, and clinically acceptable limits.

EDITOR’S PERSPECTIVE

Capnodynamic monitoring of lung volume and pulmonary blood flow during alveolar recruitment: a prospective observational study in postoperative cardiac patients

Alveolar recruitment manoeuvres may mitigate ventilation and perfusion mismatch after cardiac surgery. Monitoring the efficacy of recruitment manoeuvres should provide concurrent information on pulmonary and cardiac changes. This study in postoperative cardiac patients applied capnodynamic monitoring of changes in end-expiratory lung volume and effective pulmonary blood flow. Alveolar recruitment was performed by incremental increases in positive end-expiratory pressure (PEEP) to a maximum of 15 cmH2O from a baseline of 5 cmH2O over 30 min. The change in systemic oxygen delivery index after the recruitment manoeuvre was used to identify responders (> 10% increase) with all other changes (≤ 10%) denoting non-responders. Mixed factor ANOVA using Bonferroni correction for multiple comparisons was used to denote significant changes (p < 0.05) reported as mean differences and 95% CI. Changes in end-expiratory lung volume and effective pulmonary blood flow were correlated using Pearson's regression. Twenty-seven (42%) of 64 patients were responders increasing oxygen delivery index by 172 (95% CI 61-2984) mL min-1 m-2 (p < 0.001). End-expiratory lung volume increased by 549 (95% CI 220-1116) mL (p = 0.042) in responders associated with an increase in effective pulmonary blood flow of 1140 (95% CI 435-2146) mL min-1 (p = 0.012) compared to non-responders. A positive correlation (r = 0.79, 95% CI 0.5-0.90, p < 0.001) between increased end-expiratory lung volume and effective pulmonary blood flow was only observed in responders. Changes in oxygen delivery index after lung recruitment were correlated to changes in end-expiratory lung volume (r = 0.39, 95% CI 0.16-0.59, p = 0.002) and effective pulmonary blood flow (r = 0.60, 95% CI 0.41-0.74, p < 0.001). Capnodynamic monitoring of end-expiratory lung volume and effective pulmonary blood flow early in postoperative cardiac patients identified a characteristic parallel increase in both lung volume and perfusion after the recruitment manoeuvre in patients with a significant increase in oxygen delivery.Trial registration This study was registered on ClinicalTrials.gov (NCT05082168, 18th of October 2021).

Effects of apparatus dead space on volumetric capnograms in neonates with healthy lungs: a simulation study

BACKGROUND

Volumetric capnography in healthy ventilated neonates showed deformed waveforms, which are supposedly due to technological limitations of flow and carbon dioxide sensors.

AIMS

This bench study analyzed the role of apparatus dead space on the shape of capnograms in simulated neonates with healthy lungs.

METHODS

We simulated mechanical breaths in neonates of 2, 2.5, and 3 kg of body weight using a neonatal volumetric capnography simulator. The simulator was fed by a fixed amount of carbon dioxide of 6 mL/kg/min. Such simulator was ventilated in a volume control mode using fixed ventilatory settings with a tidal volume of 8 mL/kg and respiratory rates of 40, 35, and 30 breaths per minute for the 2, 2.5 and 3 kg neonates, respectively. We tested the above baseline ventilation with and without an additional apparatus dead space of 4 mL.

RESULTS

Simulations showed that adding the apparatus dead space to baseline ventilation increased the amount of re-inhaled carbon dioxide in all neonates: 0.16 ± 0.01 to 0.32 ± 0.03 mL (2 kg), 0.14 ± 0.02 to 0.39 ± 0.05 mL (2.5 kg), and 0.13 ± 0.01 to 0.36 ± 0.05 mL (3 kg); (p < .001). Apparatus dead space was computed as part of the airway dead space, and therefore, the ratio of airway dead space to tidal volume increased from 0.51 ± 0.04 to 0.68 ± 0.06, from 0.43 ± 0.04 to 0.62 ± 0.01 and from 0.38 ± 0.01 to 0.60 ± 0.02 in the 2, 2.5 and 3 kg simulated neonates, respectively (p < .001). Compared to baseline ventilation, adding apparatus dead space decreased the ratio of the volume of phase III to VT size from 31% to 11% (2 kg), from 40% to 16% (2.5 kg) and from 50% to 18% (3 kg); (p < .001).

CONCLUSIONS

The addition of a small apparatus dead space artificially deformed the volumetric capnograms in simulated neonates with healthy lungs.

Effect of external dead space removal on CO2 homeostasis in mechanically ventilated adult Covid-19 patients

BACKGROUND

Patients with Covid-19 respiratory failure present with hypoxemia, often in combination with hypercapnia. In this prospective, observational study we examined the effect of removing external dead space (DS) on CO₂ -homeostasis in mechanically ventilated Covid-19 patients. In addition, volumetric capnography was validated for its ability to estimate external DS volume using in vitro measured DS volumes as reference.

METHODS

In total, 10 patients with acute respiratory distress syndrome from Covid-19 were included. Volumetric capnography, mechanical ventilation, and arterial blood gas data were analyzed before and after removal of external DS and analyzed for potentially significant changes in response to DS removal. Measurements of external DS were obtained in circuit using volumetric capnography and compared to actual measured DS volumes off the circuit.

RESULTS

After the removal of external DS, the alveolar minute ventilation and CO₂ elimination improved, notwithstanding unchanged respiratory rate and tidal volumes. The increase in CO₂ elimination was associated with a decrease in arterial CO₂ partial pressure (PaCO₂ )_. The volumetric capnography method for assessment of external DS showed a low bias of -9 mL (lower limit of agreement -40, 95% CI -60 to -20 mL, upper limit of agreement 21 mL, 95% CI: 1-40 mL) and a percentage error of 48% compared to absolute values measured in vitro.

CONCLUSION

Removal of external DS increased alveolar minute ventilation and CO₂ elimination in Covid-19 patients with respiratory failure in the current study. This was associated with a decrease in PaCO₂ . This may indicate a decreased CO₂ production due to decreased work of breathing and more effective gas-exchange in response to DS removal. In addition, volumetric capnography appears to be a clinically feasible method for continuous measurement of external DS in the current study and may be of value in optimizing ventilator treatment.

Capnodynamics - noninvasive cardiac output and mixed venous oxygen saturation monitoring in children

Hemodynamic monitoring in children is challenging for many reasons. Technical limitations in combination with insufficient validation against reference methods, makes reliable monitoring systems difficult to establish. Since recent studies have highlighted perioperative cardiovascular stability as an important factor for patient outcome in pediatrics, the need for accurate hemodynamic monitoring methods in children is obvious. The development of mathematical processing of fast response mainstream capnography signals, has allowed for the development of capnodynamic hemodynamic monitoring. By inducing small changes in ventilation in intubated and mechanically ventilated patients, fluctuations in alveolar carbon dioxide are created. The subsequent changes in carbon dioxide elimination can be used to calculate the blood flow participating in gas exchange, i.e., effective pulmonary blood flow which equals the non-shunted pulmonary blood flow. Cardiac output can then be estimated and continuously monitored in a breath-by-breath fashion without the need for additional equipment, training, or calibration. In addition, the method allows for mixed venous oxygen saturation (SvO₂) monitoring, without pulmonary artery catheterization. The current review will discuss the capnodyamic method and its application and limitation as well as future potential development and functions in pediatric patients.

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

Background

The optimal level of positive end-expiratory pressure (PEEP) during mechanical ventilation for COVID-19 pneumonia remains debated and should ideally be guided by responses in both lung volume and perfusion. Capnodynamic monitoring allows both end-expiratory lung volume (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\text{EELV}}_{{{\text{CO}}_{2} }}$$\end{document}EELVCO2) and effective pulmonary blood flow (EPBF) to be determined at the bedside with ongoing ventilation.

Methods

Patients with COVID-19-related moderate to severe respiratory failure underwent capnodynamic monitoring of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\text{EELV}}_{{{\text{CO}}_{2} }}$$\end{document}EELVCO2 and EPBF during a step increase in PEEP by 50% above the baseline (PEEP_low to PEEP_high). The primary outcome was a > 20 mm Hg increase in arterial oxygen tension to inspired fraction of oxygen (P/F) ratio to define responders versus non-responders. Secondary outcomes included changes in physiological dead space and correlations with independently determined recruited lung volume and the recruitment-to-inflation ratio at an instantaneous, single breath decrease in PEEP. Mixed factor ANOVA for group mean differences and correlations by Pearson’s correlation coefficient are reported including their 95% confidence intervals.

Results

Of 27 patients studied, 15 responders increased the P/F ratio by 55 [24–86] mm Hg compared to 12 non-responders (p < 0.01) as PEEP_low (11 ± 2.7 cm H₂O) was increased to PEEP_high (18 ± 3.0 cm H₂O). The \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\text{EELV}}_{{{\text{CO}}_{2} }}$$\end{document}EELVCO2 was 461 [82–839] ml less in responders at PEEP_low (p = 0.02) but not statistically different between groups at PEEP_high. Responders increased both \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\text{EELV}}_{{{\text{CO}}_{2} }}$$\end{document}EELVCO2 and EPBF at PEEP_high (r = 0.56 [0.18–0.83], p = 0.03). In contrast, non-responders demonstrated a negative correlation (r = − 0.65 [− 0.12 to − 0.89], p = 0.02) with increased lung volume associated with decreased pulmonary perfusion. Decreased (− 0.06 [− 0.02 to − 0.09] %, p < 0.01) dead space was observed in responders. The change in \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\text{EELV}}_{{{\text{CO}}_{2} }}$$\end{document}EELVCO2 correlated with both the recruited lung volume (r = 0.85 [0.69–0.93], p < 0.01) and the recruitment-to-inflation ratio (r = 0.87 [0.74–0.94], p < 0.01).

Conclusions

In mechanically ventilated patients with moderate to severe COVID-19 respiratory failure, improved oxygenation 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. This study demonstrates the feasibility of capnodynamic monitoring to assess physiological responses to PEEP at the bedside to facilitate an individualised setting of PEEP.

Trial registration: NCT05082168 (18th October 2021).

Supplementary Information

The online version contains supplementary material available at 10.1186/s13054-022-04110-0.

Monitoring Expired CO2 Kinetics to Individualize Lung-Protective Ventilation in Patients With the Acute Respiratory Distress Syndrome

Mechanical ventilation (MV) is a lifesaving supportive intervention in the management of acute respiratory distress syndrome (ARDS), buying time while the primary precipitating cause is being corrected. However, MV can contribute to a worsening of the primary lung injury, known as ventilation-induced lung injury (VILI), which could have an important impact on outcome. The ARDS lung is characterized by diffuse and heterogeneous lung damage and is particularly prone to suffer the consequences of an excessive mechanical stress imposed by higher airway pressures and volumes during MV. Of major concern is cyclic overdistension, affecting those lung segments receiving a proportionally higher tidal volume in an overall reduced lung volume. Theoretically, healthier lung regions are submitted to a larger stress and cyclic deformation and thus at high risk for developing VILI. Clinicians have difficulties in detecting VILI, particularly cyclic overdistension at the bedside, since routine monitoring of gas exchange and lung mechanics are relatively insensitive to this mechanism of VILI. Expired CO₂ kinetics integrates relevant pathophysiological information of high interest for monitoring. CO₂ is produced by cell metabolism in large daily quantities. After diffusing to tissue capillaries, CO₂ is transported first by the venous and then by pulmonary circulation to the lung. Thereafter diffusing from capillaries to lung alveoli, it is finally convectively transported by lung ventilation for its elimination to the atmosphere. Modern readily clinically available sensor technology integrates information related to pulmonary ventilation, perfusion, and gas exchange from the single analysis of expired CO₂ kinetics measured at the airway opening. Current volumetric capnography (VCap), the representation of the volume of expired CO₂ in one single breath, informs about pulmonary perfusion, end-expiratory lung volume, dead space, and pulmonary ventilation inhomogeneities, all intimately related to cyclic overdistension during MV. Additionally, the recently described capnodynamic method provides the possibility to continuously measure the end-expiratory lung volume and effective pulmonary blood flow. All this information is accessed non-invasively and breath-by-breath helping clinicians to personalize ventilatory settings at the bedside and minimize overdistension and cyclic deformation of lung tissue.

Positive end-expiratory pressure individualization guided by continuous end-expiratory lung volume monitoring during laparoscopic surgery

To determine whether end-expiratory lung volume measured with volumetric capnography (EELV_CO2) can individualize positive end-expiratory pressure (PEEP) setting during laparoscopic surgery. We studied patients undergoing laparoscopic surgery subjected to Fowler (F-group; n = 20) or Trendelenburg (T-group; n = 20) positions. EELV_CO2 was measured at 0° supine (baseline), during capnoperitoneum (CP) at 0° supine, during CP with Fowler (head up + 20°) or Trendelenburg (head down - 30°) positions and after CP back to 0° supine. PEEP was adjusted to preserve baseline EELV_CO2 during and after CP. Baseline EELV_CO2 was statistically similar to predicted FRC in both groups. At supine and CP, EELV_CO2 decreased from baseline values in F-group [median and IQR 2079 (768) to 1545 (725) mL; p = 0.0001] and in T-group [2164 (789) to 1870 (940) mL; p = 0.0001]. Change in body position maintained EELV_CO2 unchanged in both groups. PEEP adjustments from 5.6 (1.1) to 10.0 (2.5) cmH₂O in the F-group (p = 0.0001) and from 5.6 (0.9) to 10.0 (2.6) cmH₂O in T-group (p = 0.0001) were necessary to reach baseline EELV_CO2 values. EELV_CO2 increased close to baseline with PEEP in the F-group [1984 (600) mL; p = 0.073] and in the T-group [2175 (703) mL; p = 0.167]. After capnoperitoneum and back to 0° supine, PEEP needed to maintain EELV_CO2 was similar to baseline PEEP in F-group [5.9 (1.8) cmH₂O; p = 0.179] but slightly higher in the T-group [6.5 (2.2) cmH₂O; p = 0.006]. Those new PEEP values gave EELV_CO2 similar to baseline in the F-group [2039 (980) mL; p = 0.370] and in the T-group [2150 (715) mL; p = 0.881]. Breath-by-breath noninvasive EELV_CO2 detected changes in lung volume induced by capnoperitoneum and body position and was useful to individualize the level of PEEP during laparoscopy.Trial registry: Clinicaltrials.gov NCT03693352. Protocol started 1st October 2018.