Introduction

Liver transplantation (LT) is the treatment of choice for acute/chronic end-stage liver disease1. In the postoperative period several risk factors such as intraoperative fluid administration, total transfusion volume, large graft size, ascites, and pleural effusion negatively affect cardiac and pulmonary function2,3 prolonging the time of dependence from mechanical ventilation.

(MV). In addition, intra-operative factors such as bilateral transection of the abdominal muscles, pain stimulation and the presence of a mesh affect the recovery of spontaneous respiratory function4. In pediatric liver recipients diaphragmatic dysfunction is associated with prolonged ventilation and pediatric intensive care unit (PICU) stay5,6. During weaning from MV, an ideal support mode should improve patient-ventilator interaction while minimizing asynchronies and detrimental hemodynamic effects induced by rising intrathoracic pressure7. Pressure support ventilation (PSV) is one of the most adopted modes of assisted spontaneous breathing to efficiently unload respiratory muscle8,9. However, PSV delivers a fixed level of assistance that is associated to higher occurrence of patient-ventilator asynchronies10,11. Neurally Adjusted Ventilatory Assist (NAVA) is a mode of assisted spontaneous ventilation that delivers positive pressure in proportion to electrical activity of the diaphragm (EAdi) thus optimizing ventilator cycling12 and reducing incidence of patient-ventilator asynchronies as compared to PSV13,14. Moreover, synchronization between positive pressure during inspiration and electrical diaphragm activity (Eadi) during NAVA leads to a reduction of pleural pressure and reduces the negative effects on cardiovascular function15. Advanced hemodynamic monitoring is essential to evaluate hemodynamic status, especially during the perioperative period of high-risk surgery16,17. Echocardiography is the most common noninvasive method to assess cardiac function18. The combination of cardiovascular and respiratory effects during NAVA is not well established, especially after major abdominal surgery as liver transplantation where several factors negatively affect respiratory system mechanics and diaphragm function. To the best of our knowledge the physiologic effects of proportional assisted mode of MV compared to PSV have not been investigated in pediatric liver recipients. The aim of this study is to evaluate patient ventilator interaction and cardiac function at the first transition to spontaneous assisted breathing of pediatric patients who underwent liver transplantation. We hypothesized that, compared to PSV, NAVA minimizes patient-ventilator asynchronies and improves cardiac performance.

Methods

Study ethics

The Use of Neurally Adjusted Ventilator Assist versus Pressure Support Ventilation During Weaning from Mechanical Ventilation in Pediatric Patients After Liver Transplantation (NAVIGATE) protocol was approved by institutional research board at Bambino Gesù Children’s Hospital (document 1695_OPBG_2018, March 19, 2019).

All research was performed in accordance with relevant guidelines/regulations, and written informed consent was obtained from parents/legal guardian. The study has been performed in accordance with the Declaration of Helsinki. No organs/tissues were procured from prisoners.

This is a single center, randomized, no profit, physiologic cross-over controlled trial comparing NAVA with PSV in children underwent liver transplantation. This study follows CONSORT recommendations.

The study was registered on ClinicalTrials.gov (Clinical Trial Number: NCT04792788, Registration date: 2021-03-11).

Protocol

All the enrolled patients were admitted to PICU in Bambino Gesù Children’s Hospital, Rome (Italy). Inclusion criteria were: (1) age between one month to 10 years of age; (2) liver recipients; (3) invasive mechanical ventilation. Exclusion criteria were: (1) neurological impairment; (2) hypotonia (by neuromuscular, mitochondrial, metabolic, or chromosomal diseases); (3) lesions of medulla; (4) hemodynamic instability requiring inotropes/vasopressors (dopamine > 6 mcg/kg/min, norepinephrine > 0.1 mcg/kg/min, epinephrine > 0.1 mcg/kg/min, dobutamine > 6 mcg/kg/min, milrinone > 0.35 mcg/kg/min) or almost one volume bolus (crystalloids/colloids > 20 ml/kg) during the past 6 h; (5) congenital cardiovascular disease; (6) patient extubated; (7) respiratory instability (paO2/FiO2 < 200; SpO2 < 90% with FiO2 0.4); (8) need of controlled mechanical ventilation; (9) intravenous infusion of benzodiazepines or propofol; (10) pneumonia, pneumothorax, massive pleural effusion; 11) patient placed on extracorporeal circuit (continuous renal replacement therapy, extracorporeal membrane oxygenation, apheresis); 12) contraindications to insert nasogastric tube; 13) not expected to survive beyond 24 h; 14) parental/legal guardian refusal.

After enrollment, the standard nasogastric tube of each patient was replaced with a specific nasogastric tube (Edi catheter) with an array of eight bipolar electrodes mounted at its distal end (Getinge Critical Care, Solna, Sweden). The description of verification of nasogastric tube placement is explained in Additional Material S1.

Hemodynamic monitoring was performed by mathematical analysis of the arterial waveform analyzed through an arterial catheter placed in the radial, brachial, or femoral artery and recorded according to the PRAM (Pressure Recording Analytical Method) pressure analysis method (MostCare Vygon, Vytech, Italy). The PRAM is an algorithm that processes arterial pressure waveforms with high temporal resolution (1000 Hz) and does not require external calibration. This method derives cardiac index and other advanced hemodynamic parameters by evaluating the interplay between vascular resistance and arterial compliance in real time. It enables continuous and accurate monitoring of cardiovascular dynamics19. Transthoracic echocardiogram performed by two expert pediatric cardiologists completed the hemodynamic monitoring during the period study. The images of the transthoracic echocardiograms of all enrolled patients were recorded and stored on a dedicated computer.

The randomization process consisted of a computer-generated random listing of treatment allocation using block sizes of 4 (Three Randomization Plan Generators. Available from URL: https://gdallal.pages.tufts.edu/random_block_size.htm). The randomization started when the enrolled patients started spontaneously triggering the mechanical ventilator. Patients were ventilated in pressure regulated volume control before starting assisted spontaneous breathing. To initiate weaning from MV, the patient had to mantain a PaO2/FiO2 ratio > 200 and SpO2 > 90% with FiO2 < 0.4 and PEEP 4–5 cm H2O. Each patient was randomized to a ventilation mode sequence (PSV/NAVA/PSV or NAVA/PSV/NAVA).

Each patient was studied for a duration of 2 h, divided in three trials of 40-minutes (Fig. 1). The first 30-minutes of each trial was considered for washout of the previous ventilatory mode. The results were recorded only in the following ten minutes of each trial.

Fig. 1
figure 1

Randomisation scheme and study phases. Eligible patients were randomised into Group 1 and Group 2 according to the study sequence.

Full size image

Considering PSV trial: the initial PEEP was set on 4–5 cm H2O and PS level was set to reach tidal volume of 6–9 ml/kg, a reduction of respiratory rate to physiological values for age20, and absence of clinical signs of increased work of breathing (chest retractions, diaphragm paradox). The flow trigger was set to the maximum sensitivity level not causing autotriggering. The clinician set the expiratory cycling off to achieve the best synchronization according to the flow/pressure tracings, and the fastest pressure ramp time. During application of NAVA, Eadi trigger was at default value of 0.5 µV. To determine the corresponding NAVA level, able to achieve a similar inspiratory mean airway pressure to that obtained in PSV, a dedicated function called NAVA Preview was used. Periods of coughing/suctioning were excluded from the analysis. All measurements were performed if stable traces were correctly displayed. The measurements were recorded and analyzed after the patients were discharged from the PICU. Two blinded researchers (G.S. and R.C.) independently analyzed the recorded data. In case of disgreement between the two, a third blinded investigator (G.C.) reviewed the recordings and resolved the discrepancies.”

Two blinded pediatric cardiologists with more than ten years of experience in critically ill children performed an echocardiogram during the last ten minutes of each trial. Arterial line was connected to MostCareUp monitor values were recorded in the last ten minutes of each trial.

Analgesia was provided according to the PICU protocol for patient admitted after liver transplantation (morphine 10 mcg/kg/h after intravenous bolus of 50–100 mcg/kg).

Perioperative variables were collected as follows: gender, age, weight, body mass index, Status classification, Pediatric End Liver Disease (PELD), primary disease, cirrhosis, PICU admission before transplantation, Pediatric Index of Mortality 3 (PIM3), living liver donor, liver ischemia time, total volume of transfusion per weight, cumulative fluid balance (according to formula described by Goldstein et al.21, mesh, surgery duration, graft-to-recipient weight ratio (GRWR, graft weight in gram/recipient weight in kg × 10), if the patient needed noninvasive ventilation post transplantation (NIV), ventilator-free days, length of stay (LOS) in PICU and hospital, mortality at 28 days.

Patient-ventilator interaction parameters. To estimate the asynchrony rate, we calculated the asynchrony index (AI), which is the ratio between the number of asynchronous events and the total respiratory rate, expressed as percentage. An AI > 10% was considered a high rate of asynchrony11. The asynchronies observed and analysed were wasted efforts (defined as a patient inspiratory effort not assisted by the ventilator), auto-triggering (defined as a mechanical insufflation in absence of a patient inspiratory effort), late cycling (defined as a cycle with the mechanical inspiratory time greater than twice the patient’s neural time) and double triggering, defined as two mechanical breaths separated by a short expiratory time during the same inflection in Edi signal (< half of neural expiratory timing). Other variables collected were: tidal volume, peak airway pressure, mean airway pressure, oxygenation index, and PaO2/FiO2 ratio, inspiratory trigger delay, expiratory trigger delay, time of synchrony defined as the time during which patient’s inspiratory effort and ventilatory assistance are in phase, and time during which respiratory effort and ventilator assistance were synchronous. The amount of inspiratory effort was calculated as the Pressure Time Product of Edi per breath and per minute (PTPEdi/breath and PTPEdi/min) defined as the area under the Edi trace from the neural inspiration to the end of the neural expiration. The method of sampling patient-ventilator interaction parameters is described in Additional Material S2.

Echocardiographic evaluation. Variables were collected using Philips CX 50 echocardiography device with S5/S8 probes. Measurements: TAPSE (tricuspid annular plane systolic excursion), RVFAC (right ventricular fractional area change), IT (tricuspid valve insufficiency), TDI (Tissue Doppler Echocardiography to measure systolic velocities in right ventricle), LV EF (left ventricle ejection fraction), RV-PSV (right ventricle peak systolic velocity).

Hemodynamic parameters. Collected hemodynamic measurements were heart rate, systolic blood pressure, diastolic blood pressure, mean arterial pressure, cardiac index, systemic vascular resistance index, and central venous pressure.”

Primary outcomes

Primary outcomes were: (1) To compare the magnitude of patient ventilator asynchronies between PSV and NAVA in patients undergoing liver transplantation. Evaluate the variation in asynchronies (AI) between patient and ventilator during weaning from MV with the use of NAVA (compared with PSV); (2) To investigate the physiological effects of each ventilation mode on cardiac function.

Secondary outcomes

Evaluate the differences between the two ventilation modes in terms of pulmonary function (PaO2/FiO2, oxygenation index, mean airway pressure, tidal volume), patient ventilator interaction (auto triggering, late cycling, double triggering, wasted efforts, inspiratory and expiratory trigger delay, pressurization time, time of synchrony, PTPEadi/breath and PTPEadi/min), cardiac and hemodynamic parameters mentioned above.

Statistical plan

Collected data were presented as count and proportions (categorical data) or median and interquartile range (continuous data).

A bivariate quantile regression analysis was applied to estimate the effect of NAVA compared to PSV on each metabolic, ventilation and hemodynamic variable. Estimates of the outcome variables for change in a covariate were reported as medians and standard error.

An adjusted quantile regression model was applied to account for potential confounders to evaluate the effect of ventilation mode NAVA (PSV mode was the reference value) on asynchrony index, TAPSE and left ventricles strain. The determinants of AI and RVFAC for which the p-value was < 0.20 in the bivariate analysis were included in the initial multivariable logistic regression model. Subsequently, Patient baseline characteristics such as weight, graft/recipient weight ratio, cumulative fluid balance after surgery and cirrhosis with ascites (which are clinically and pathophysiologically relevant to respiratory mechanics and related to diaphragm and abdominal muscle function during spontaneous ventilation) were considered as potential confounders.

According to previous studies22, we estimated an AI of 50% in PSV and we expected NAVA to reduce it to 2%. Considering α-error equal to 0.05 and power equal to 80%, the study would have needed 16 patients to detect a 48% reduction in AI. Statistical software Stata 15.0 (StataCorp) was used for statistical analysis.

Results

From March 2021 to September 2022 thirty-seven pediatric patients were screened for inclusion. Twenty-four liver recipients were enrolled in the study, two showed hemodynamic instability after randomization, and one returned to the operating room (Fig. 2). All the remaining twenty-one patients completed measurements of ventilatory and invasive hemodynamic parameters during the three post-randomization phases. Only 12 patients were studied by echocardiography (6 for each ventilation sequence) due to the unavailability of the cardiologists overnight. Patient characteristics are shown in Table 1. No differences were detected by gender. The most frequent primary disease is biliary tract malformation (15/21, 71%). One third of the patients had cirrhosis with ascites (7/21, 33%). All patients received massive transfusions during transplantation (> 40 ml/kg in 6 h)23. Six patients had GRWR > 4% (6/21, 29%).

Fig. 2
figure 2

Screening and randomization in a study of the effect of NAVA vs PSV in pediatric liver recipients.

Full size image
Table 1 Baseline characteristics.
Full size table

In a regression model with one covariate (Table 2 and Table S1), no significant changes in respiratory gas exchange or metabolism were observed during the application of NAVA and PSV. In terms of patient-ventilator interaction, NAVA was associated with significantly lower asynchrony index (1.5 versus 6.8%, p 0.016), inspiratory delay time (80 versus 150 milliseconds, p < 0.001), peak airway pressure (7.2 vs. 10.3 cmH2O, p < 0.001). The main asynchronies (late ciclying, double triggering, and wasted efforts) were not included in Table 2 due to collinearity among variables. In particular, the strong correlation between some asynchrony indices led to their automatic omission in the regression model. PSV is associated with lower pressurization time (415 versus 615 milliseconds, p < 0.001) and synchronization time (420 versus 615 milliseconds, p 0.046).

Table 2 Comparison of metabolic, respiratory, and cardiovascular variables between PSV and NAVA ventilation modes.
Full size table

In the multivariable logistic model analysis (Table 3) NAVA is significantly associated with a reduction in patient-ventilator asynchronies (AI) compared to PSV (Coeff − 6.66, 95% CI −11.5 to −1.78, p 0.08). In contrast, NAVA did not have an effect on right/left ventricular function. These findings are confirmed in the logistic model, which accounted for covariates with a p-value < 0.2 in the bivariate analysis (Table 2s).

Table 3 Effects of ventilation mode and patient baseline characteristics on asynchrony index and right ventricles function.
Full size table

Discussion

In this trial involving pediatric liver recipients, the application of NAVA reduces asynchronies between patient and ventilator.

Improved synchrony during NAVA, as well as respiratory exchanges and reduced wasted efforts, have been shown in other studies in patients with PARDS, bronchiolitis and post-cardiac surgery24,25,26.

In the previous studies in which NAVA was compared to PSV, the median AI varied between 1.7 and 11% during the application of NAVA, in our population NAVA resulted in an AI value of 1.5%. It should be noted that even during the application of PSV, AI was lower in comparison with the above-mentioned previous studies (8.6 vs. 8.8–25%). The presence of Edi monitoring allowed to optimize the expiratory trigger.

In our study, we did not detect significant changes in respiratory exchanges. No studies have compared echocardiographic function during NAVA and PSV in pediatric patients undergoing major abdominal surgery. A study in adult population submitted to abdominal surgery showed greater efficacy of NAVA (vs. PSV) in respiratory exchange and Eadi but did not evaluate asynchronies and cardiac function27.

The advantage of NAVA, during weaning from MV, becomes more relevant when analyzing same specific aspects (abdominal and not) that characterize patients undergoing major abdominal surgery. During anesthesia and surgery, the insult on chest wall due to retractor and the reduction in muscle tone lead to a cephalic elevation of the diaphragm, which is prone to atelectasis development; in the post-operative period, pain and diaphragmatic dysfunction lead to a reduction in residual functional capacity and may consequently lead induce the persistence or development of new atelectasis28. It should be added that, unlike ascites, which is drained by tubes, pleural effusion can get worse because of pulmonary and diaphragmatic dysfunction.

Abdominal factors that may alter weaning and spontaneous/assisted ventilatory activity include abdominal surgery itself, the need for postoperative intravenous analgesia, graft-to-recipient weight ratio (GRWR), and the presence of abdominal mesh. Spontaneous ventilatory activity also relies on abdominal muscle action coordinated with the diaphragm; therefore, the bilateral transection of the abdominal muscles (with the lost of physiological continuity of abdominal wall), the presence of abdominal mesh and the painful stimulus already determine respiratory adjustment during postoperative period. Grimaldi et al. pointed out that a GRWR of > 4% increases the risk of abdominal distension and vascular complications of the graft29. It is understandable that abdominal mass may alter diaphragmatic excursion and influence venous return to the right atrium. In our population, mesh was used in three (3/21, 14%) of the 6 patients (6/21, 29%) with GRWR > 4%. In both cases (simultaneous presence of abdominal mesh and GRWR > 4%) we have patients with abdominal features that may affect the effectiveness of diaphragmatic activity during weaning from MV. Despite these issues, NAVA is superior in terms of patient-ventilator.

The effect of positive cumulative fluid balance and massive transfusion (> 40 ml/Kg in 6 h) are associated with prolonged invasive MV, difficulty weaning from MV and postoperative pulmonary complications. In this study, while in all patients we detected massive transfusions and postoperative fluid balance between 10 and 20%, the use of NAVA results in lower peak and mean airway pressure proving less stress of the lungs than PSV.

In a cross-over study with an A-B-A design, the return to phase A allows for the assessment of washout effects, ensuring that any residual influence from the previous treatment in phase B does not persist. This controls for carry-over effects, which could confound results by allowing treatment effects to persist into subsequent phases. Additionally, the design evaluates the reversibility of treatment effects, enhances statistical robustness with repeated baseline measurements, and minimizes inter-individual variability by using participants as their own controls.

There are several limitations to this study. Firstly, the study was conducted in a single center. Second, the number of patients, although higher than in previous studies, is still small. Thirdly, the potential role of echocardiography in assessing the effect of the two ventilation modes on asynchronies and cardiac function could not be fully explored in all patients studied. The absence of blinding of the investigators at the patient’s bedside (with the exception of the cardiologists and researchers who analysed the ventilatory curve recordings) may introduce bias by influencing themselves.

Conclusion

Diaphragmatic dysfunction, positive cumulative balance and massive transfusions may adversely affect weaning from invasive MV in children undergoing major abdominal surgery. Confirming data from previous studies31,32, NAVA significantly improves patient-ventilator interaction, as evidenced by a reduction in AI, compared to PSV mode. This study highlights a potential application in patients undergoing spine, thoracic and cardiac surgery, where the chest would combine the surgical factors and ventilatory management; additionally, the role of the diaphragm would be involved during the postoperative period after abdominal oncological surgery, where its function may also be altered. Incorporating diaphragmatic ultrasound may provide added value as a confirmatory tool for detecting diaphragmatic dysfunction, particularly in relation to duration of mechanical ventilation and type of surgery performed. Further multicenter studies are needed to evaluate the clinical effects of NAVA on weaning from MV in post-operative period and to assess the effects of NAVA on ventricular function.