Individualized Flow-controlled versus Pressure-controlled Ventilation in Cardiac Surgery: A Randomized Controlled Trial

Abstract

Background:

Patients undergoing on-pump cardiac surgery are at high risk for perioperative lung injury and a hyperinflammatory state associated with postoperative complications. The authors investigated the hypothesis that flow-controlled ventilation (FCV) reduces the inflammatory stimulus compared to conventional pressure-controlled ventilation (PCV) in this patient cohort. FCV has the unique feature of controlling airway flows during inspiration and expiration and the potential to reduce mechanical power of invasive ventilation.

Methods:

In this single-center randomized controlled trial, 140 adult patients undergoing cardiac surgery with cardiopulmonary bypass were allocated 1:1 to FCV or PCV from August 10, 2020, to November 16, 2022. Participants received perioperatively either individualized FCV with a compliance-guided positive end-expiratory pressure (PEEP) and a compliance-guided driving pressure (ΔP) or PCV with a compliance-guided PEEP and ΔP for tidal volumes of 6 to 8 ml/kg predicted body weight. Postoperative plasmatic interleukin 8 (IL-8) levels 6 h after cardiopulmonary bypass were defined as the primary endpoint. Explorative secondary outcomes included incidences of postoperative pulmonary and extrapulmonary complications and hospital length of stay.

Results:

Median postoperative IL-8 levels did not differ significantly between FCV and PCV (FCV, 3.08 vs. PCV, 3.60; β, 0.08 pg/ml; 95% CI, −0.17 to 0.33; P = 0.573). ΔP values and tidal volumes were higher in the FCV group but FCV yielded lower respiratory rates and minute volumes required for normocapnia. As a result, the FCV approach reduced the perioperatively applied mechanical power by 55%. After FCV, incidences of single postoperative pulmonary complications (e.g., confirmed pneumonia, moderate and severe hypoxemia) and any postoperative extrapulmonary complication were lower and the hospital stay shorter.

Conclusions:

FCV did not reduce plasmatic IL-8 levels at the predefined timepoint 6 h after cardiopulmonary bypass. However, the reduction of mechanical power during individualized FCV application and the findings of the explorative secondary study outcomes justify future trials.

This randomized trial in cardiac surgery patients compared flow-controlled ventilation with a compliance-optimized driving pressure strategy allowing individually titrated tidal volumes, low airflows, respiratory rates, and minute volumes with a conventional pressure-controlled strategy utilizing tidal volumes of 6 to 8 ml/kg predicted body weight. The primary outcome was plasma interleukin-8 levels at 6 h after cardiopulmonary bypass and several exploratory secondary outcomes. Although no significant differences occurred in the primary outcome, perioperatively applied mechanical power was reduced by 55% in the flow-controlled ventilation group. Exploratory secondary outcomes demonstrated possible outcome benefits, suggesting that future larger studies may be indicated.

Editor’s Perspective

What We Already Know about This Topic

• Flow-controlled ventilation (FCV) is a newer alternative positive pressure ventilation mode allowing control of airway flows during both inspiration and expiration

• It has the potential to reduce applied mechanical power and minimize energy dissipation during invasive ventilation, which may be beneficial to the lungs

• In patients undergoing cardiac surgery with cardiopulmonary bypass, a hyperinflammatory state may result predisposing the patient to pulmonary complications

What This Article Tells Us That Is New

• This randomized trial in cardiac surgery patients compared FCV with a compliance-optimized driving pressure strategy allowing individually titrated tidal volumes, low airflows, respiratory rates, and minute volumes with a conventional pressure-controlled strategy utilizing tidal volumes of 6 to 8 ml/kg predicted body weight

• The primary outcome was plasma interleukin-8 levels at 6 h after cardiopulmonary bypass and several exploratory secondary outcomes

• Although no significant differences occurred in the primary outcome, perioperatively applied mechanical power was reduced by 55% in the FCV group

• Exploratory secondary outcomes demonstrated possible outcome benefits, suggesting that future larger studies may be indicated

In cardiac surgery with cardiopulmonary bypass (CPB), lung injury is inevitable and comprises three main mechanisms: (1) surgical trauma by the intrathoracic procedure, (2) volutrauma including regional overdistension resulting from mechanical ventilation in an open chest condition together with atelectrauma caused by repetitive phases of lung collapse and reinflation, and (3) lung ischemia/reperfusion injury of distal airways. All three mechanisms trigger a hyperinflammatory state that is aggravated by the contact of blood with the exogenous surfaces of the CPB circuit.¹ Different lung-protective ventilation strategies in the setting of cardiac surgery have reported an impact on the systemic and pulmonary inflammatory response.^2,3 Clinically, lung injury manifests in postoperative pulmonary complications in more than 50% of cardiac surgical patients, ranging from mild hypoxemia to the acute respiratory distress syndrome (ARDS).⁴

Barnes et al. originally described flow-controlled ventilation (FCV) as an alternative positive pressure ventilation mode for minimizing energy dissipation in controlled invasive respiration.⁵ A unique feature of FCV is the control of constant and continuous airway flows during both inspiration and expiration. Patient-individualized FCV application implies ventilation with a compliance-optimized driving pressure (ΔP) and thus individually titrated tidal volumes, which contrasts sharply with conventional, uniformly low tidal volume ventilation strategies. However, this approach provides the opportunity to use low airway flows and low respiratory rates, with a low minute volume being necessary for sufficient oxygenation and in particular decarboxylation. Thus, patient-individualized FCV has the potential to reduce the applied mechanical power during controlled ventilation as reported in preclinical and small clinical studies.^6–8

Mechanical power is an evolving key concept in ventilator-induced lung injury and in addition to airway pressures and tidal volumes accounts for airway flows and respiratory rates.^9,10 High mechanical power is associated with severe postoperative pulmonary complications in the perioperative setting and an increased mortality in intensive care unit (ICU) patients with acute respiratory failure.^11–14

The aim of this trial was to compare not only two ventilator modes but also two contrasting perioperative ventilation strategies in adult cardiac surgery. As the intervention, FCV with a compliance-optimized ΔP and individually titrated tidal volumes was applied in order to utilize low airway flows, low respiratory rates, and a low minute volume. The comparator was conventional pressure-controlled ventilation (PCV) with ΔP values for controlled tidal volume ventilation of 6 to 8 ml/kg predicted body weight (PBW). The trial was designed to test whether individualized FCV compared to PCV reduces postoperative inflammation, as measured by postoperative plasmatic interleukin 8 (IL-8) levels. Additionally, we aimed to explore the effects of the two perioperative ventilation strategies on respiratory mechanics, postoperative lung function, and patient-relevant clinical outcomes as explorative secondary endpoints.

Materials and Methods

The study “Perioperative individualized FLOW-controlled versus conventional pressure-controlled VENTilation IN on-pump HEART SURGery (FLOWVENTIN HEARTSURG)” was a single-center randomized controlled clinical trial with two parallel groups at University Hospital Center in Bochum, Germany. The Ethics Committee of the Medical Faculty, Ruhr University Bochum, approved the final study protocol, patient information, and all study amendments (registration No. 19-6740-§ 23b). The trial was prospectively registered (German Clinical Trials Register No. DRKS00018956, available at: https://drks.de/search/en/trial/DRKS00018956/entails, accessed December 17, 2025; principal investigator, Simon Becker; registration date, June 12, 2020), and the complete and peer-reviewed study protocol was published following the "Standard Protocol Items: Recommendations for Interventional Trials" (SPIRIT) guidelines.¹⁵

Participants

We screened patients aged 18 yr or older for study participation who were scheduled for cardiac surgery with CPB excluding ventricular assist device implantation and heart transplantation. Patient-related exclusion criteria were the inability to independently consent to study participation and involvement in another perioperative interventional trial. For the reason of potentially confounding postoperative IL-8 levels, patients were not included with

• preoperative immunosuppressive drug(s) intake including antibiotics,

• suspected or confirmed endocarditis,

• suspected or confirmed pneumonia.¹⁵

Members of the trial management committee explained the study purpose and the trial protocol to eligible patients at least on the day before surgery. Participants had to sign informed consent before study enrollment.

Randomization and Masking

On the day of surgery, the principal investigator assigned participants to either perioperative FCV or PCV with the online software REDCap (Vanderbilt University, USA). Randomization assignments were concealed in REDCap until the moment of 1:1 allocation to FCV or PCV following computer-generated permuted blocks with lengths of four and six. While outcome assessors were blinded only after the interventions due to the use of different perioperative ventilators for FCV and PCV, patients remained blinded during the study up to 6 months postoperatively. We masked data analysts until fixed datasets of each outcome were available.

Procedures

After anesthesia induction, patients in both groups received a conventional endotracheal tube and in the FCV group an additional conventional endotracheal tube adapter (Ventinova Medical, The Netherlands). FCV (EVONE, Ventinova Medical) or PCV (Primus and Perseus A500, Dräger Medical Deutschland GmbH, Germany) was applied directly after orotracheal intubation (fig. 1A). General anesthesia was sustained with continuous sufentanil and propofol infusions except during the CPB run with systemic administration of sevoflurane through the membrane oxygenator of the heart–lung machine.

Fig. 1.

Characteristics and perioperative ventilation strategies in the FCV and PCV groups. After postoperative admission to the ICU, including upright patient positioning of 30 to 45°, the incremental compliance-guided PEEP trial was repeated in both groups (B), the driving pressure trial only in the FCV group (C). (D) Ventilation strategies after completion of the individualization protocol in the FCV and PCV groups. ΔP, driving pressure; PBW, predicted body weight; PEEP, positive end-expiratory pressure; PIP, peak inspiratory pressure for FCV and positive inspiratory pressure for PCV.

Ventilation in both groups started at a positive end-expiratory pressure (PEEP) of 5 cmH₂O, a ΔP to achieve tidal volumes of 6 to 8 ml/kg PBW, and respiratory rates for an end-tidal carbon dioxide partial pressure of 35 to 45 mmHg. Unlike in volume-controlled ventilation, no plateau airway pressure is displayed by the ventilator for FCV. This is due to the constant and continuous airway flow resulting in a linear increase of airway pressures during the entire inspiratory time (fig. 1A). The respiratory rate in FCV is adjusted by increasing or decreasing airway flows. At similar respiratory system mechanics and ventilator settings regarding airway pressures, a higher flow increases and a lower flow decreases the respiratory rate. Initial airway flows during inspiration and expiration in the FCV group were 8 to 12 l/min with an inspiratory to expiratory ratio of 1:1. In the PCV group, the initial inspiratory to expiratory ratio was 1:1.9, but this was individually adapted to ensure an end-expiratory airway flow of zero. Inspiratory and expiratory airway flows were not adjusted during PCV, and the ramp was 0.2 s by default.

In both groups, an incremental PEEP trial from 5 to 12 cmH₂O followed in order to determine the PEEP level with the highest dynamic compliance by calculating $\frac{Tidalvolume}{\Delta \text{P}}$ (fig. 1B). Every PEEP level was maintained until a steady state of the dynamic compliance was reached. The PEEP trial was terminated before completion if dynamic compliance decreased during three consecutive PEEP level changes and/or if the mean arterial pressure dropped more than 15 to 20% at higher PEEP levels.

Only in the FCV group, a secondary incremental ΔP trial was performed by increasing ΔP levels by 1 cmH₂O without adjustments of airway flows (fig. 1C). For every ΔP level, dynamic compliance was only determined if a steady state was reached. This trial was stopped at a respiratory rate of 5/min or if dynamic compliance decreased during two consecutive ΔP increments.

Final ventilation strategies included (fig. 1D)

• in both groups, individualized PEEP with the highest dynamic compliance plus 1 to 2 cmH₂O to prevent perioperative atelectasis formation and respiratory rates for an end-tidal carbon dioxide partial pressure of 35 to 45 mmHg,

• in the PCV group, adjusted ΔP for tidal volumes of 6 to 8 ml/kg PBW,

• in the FCV group, individualized ΔP with the highest dynamic compliance plus or minus 1 to 2 cmH₂O to compensate for perioperative changes in respiratory system compliance.¹⁵

After ICU transition, the PEEP trial was repeated in an upright patient position of 30 to 45° in both groups, whereas the ΔP trial was performed again only in the FCV group (fig. 1, B and C).

Ventilator settings could be modified according to surgical demands by reducing PEEP and/or ΔP in 1-cmH₂O steps.¹⁵ During CPB, ventilation was suspended, and no continuous airway pressure was applied. Recruitment maneuvers up to peak airway pressures of 30 cmH₂O were used either before aortic declamping in cases of open left heart surgery or at the discretion of the anesthesiologist in charge but were not standardized in the ventilation protocol.

Postoperatively, patients were transferred to the ICU under controlled group-specific ventilation with continuous propofol infusions. In the ICU setting, FCV was continued by means of the EVONE ventilator, whereas PCV was applied as biphasic positive airway pressure ventilation due to institutional standards for patient safety and comfort (Evita Infinity V500, Dräger Medical Deutschland GmbH).

Prerequisites for weaning from controlled ventilation in both groups were (1) time (at least 3 h after CPB), (2) no new hemodynamic instabilities or excessive chest drain loss, and (3) no pathologies requiring interventions according to the postoperative x-ray film. Weaning started with an opioid bolus (piritramide 0.1 to 0.15 mg/kg) and cessation of continuous propofol administration. As the FCV ventilator and the narrow tubing for FCV does not allow spontaneous breathing, patients in the FCV and also PCV group were changed to pressure support ventilation (Evita Infinity V500, Dräger Medical Deutschland GmbH) after the return of spontaneous breathing efforts. Extubation criteria for both groups were (1) a PEEP of 5 to 6 cmH₂O, (2) a pressure support of 6 cmH₂O, and (3) a sufficiently conscious patient.

For plasmatic IL-8 quantification, whole blood was collected via a central line. Before anesthesia (baseline IL-8), only an arterial line was available for blood sampling. After centrifugation within 1 h, plasma aliquots were stored at −80°C until one blinded assay operator performed batched enzyme-linked immunosorbent assay quantification according to the manufacturer’s instructions (Ultrasensitive IL-8 Human ELISA Kit, Thermo Fisher Scientific, Germany).

Blinded study assessors visited participants at least on postoperative days 1, 3, 5, and 8 and at hospital discharge and performed follow-up phone calls after 6 months. In case of transfer to other hospitals after the surgical inpatient stay, referral hospitals were asked about the additional hospital length of stay.

Outcomes

The primary study outcome was the plasmatic IL-8 concentration 6 h after CPB as a surrogate biomarker of postoperative pulmonary and systemic inflammation.

As predefined secondary endpoints,¹⁵ we evaluated

• Ventilation data during controlled, group-specific ventilation at six perioperative timepoints, including calculated mechanical power according to the surrogate formulae $\frac{\text{Minute Volume}·(\text{Peak Inspiratory Pressure}+\text{PEEP}+\frac{InspiratoryFlow}{6})}{20}$ for FCV and $0.098·\text{Respiratory Rate}·\text{Tidal Volume}·(\text{PEEP}+\Delta \text{P})$ for PCV¹⁰

• Plasmatic IL-8 concentrations at six additional timepoints

• Oxygenation indices (arterial oxygen tension [Pao₂]/inspired oxygen fraction [Fio₂]) at nine timepoints

• Arterial carbon dioxide tension (Paco₂) during controlled, group-specific ventilation at five timepoints

• Weaning time from controlled ventilation and duration of invasive ventilation

• Incidences of postoperative pulmonary complications (PPCs) during the first 7 postoperative days including hypoxemia, postextubation respiratory acidosis, noninvasive ventilation, reintubation, bronchoscopy, pleural effusions requiring intervention, pneumothorax requiring intervention, suspected pneumonia, confirmed pneumonia, and ICU-readmission due to PPCs

• Incidences of postoperative extrapulmonary complications (PEPCs) during the first 7 postoperative days including systemic inflammatory response syndrome, pericardial tamponade with intervention, other bleeding with intervention, infection with antibiotic treatment, coronary intervention, delirium, acute kidney injury, and ICU-readmission due to PEPCs

• Postoperative lengths of stay (intensive/intermediate care unit and in hospital)

• Hospital readmission and mortality rate 6 months after surgery

Fig. 2 displays the timeline of the perioperative primary and secondary outcome measurements.

Fig. 2.

Timeline figure of the primary and secondary outcome measurements. CPB, cardiopulmonary bypass; ∆P, driving pressure; Fio₂, inspired oxygen fraction; Paco₂, arterial carbon dioxide tension; Pao₂, arterial oxygen tension; PEEP, positive end-expiratory pressure; PEPCs, postoperative extrapulmonary complication; POD, postoperative day; PPC, postoperative pulmonary complication.

Sample Size Calculation and Statistical Analysis

Based on the study hypothesis and previous studies on protective ventilation strategies in the setting of cardiac surgery,^2,3 a moderate effect size of the intervention was estimated on postoperative plasmatic IL-8 levels (Cohen’s d = 0.5) as being appropriate and relevant. To ensure feasibility in a complex, single-center randomized controlled trial, only one cytokine and one postoperative timepoint could be defined as the primary study outcome. For a power of 0.8 in a two-sample t-test with a two-sided significance level α = 0.05 and an estimated dropout rate of 10%, the required sample size was 140 patients (70 per group). As dropouts, specific study stop criteria were predefined like the inability to apply the allocated ventilation protocol and/or major surgical complications independent from the ventilation protocol (e.g., more than one CPB run and intraoperative death).¹⁵

All primary and secondary outcome analyses were per protocol, and FCV as the interventional group was compared with PCV as the control group using the R programming language (https://www.r-project.org/, version 4.3.3, accessed February 15, 2024). Regression models appropriate for each type of outcome data were used to analyze the effect of the treatment assignment with the European System for Cardiac Operative Risk Evaluation (EuroSCORE II) as an additional independent covariate in every model. For evaluation of the primary study endpoint, the plasmatic IL-8 level before the induction of anesthesia as baseline was an additional independent covariate. In this particular analysis, missing baseline values (FCV n = 1, PCV n = 1) were imputed using multiple imputations based on a Bayesian linear regression as implemented in the “mice” R package. Baseline values below the detection limit of 0.12 pg/ml (FCV n = 14, PCV n = 9) were additionally imputed using the method of Lubin et al., which is specifically designed to handle data in the presence of known detection limits.¹⁶ In total, 10 imputed datasets were used. We additionally performed a sensitivity analysis to assess the effect of this imputation method. In all other analyses, only the complete cases were analyzed.

If large departures from a normal distribution were observed by examination of histograms and boxplots, continuous outcome data was log-transformed before linear regression analyses (plasmatic IL-8 concentrations only). For binary outcomes and for outcomes that are counts, logistic and Poisson regression models were employed, respectively. The Firth logistic regression was applied in all analyses of postoperative pulmonary and extrapulmonary complications to circumvent issues of nonconvergence due to few event counts. The Kaplan–Meier method and Cox proportional hazards regression models were utilized for time-to-event outcomes such as postoperative lengths of stay. β Coefficients are displayed for linear and Poisson regression models as well as odds ratios and hazard ratios for logistic regression models and Cox models, respectively. All model coefficients, odds ratios, and hazard ratios were calculated in reference to the FCV group. When analyzing secondary endpoints, we did not adjust P values for multiple comparisons due to the exploratory nature of the multiple secondary outcomes in this study. Therefore, significant results of secondary endpoint data are hypothesis-generating only.

Results

Participants

From August 10, 2020, to November 16, 2022, a total of 140 patients were recruited and enrolled after signed informed consent. Six patients in the FCV group and four patients in the PCV group were defined as dropouts meeting prespecified stop criteria of the study protocol.¹⁵ A complete summary of the excluded participants is provided in the Supplemental Digital Content table 1 (https://links.lww.com/ALN/E292). All cases were due to major surgical complications and/or had no evidence of causality attributable to the ventilation protocol. However, inclusion would have significantly biased primary and secondary outcomes, especially in cases with more than one CPB run. As a result, 130 patients were included in the final analysis (FCV n = 64, PCV n = 66; fig. 3).

Fig. 3.

Trial profile. FCV, flow-controlled ventilation; PCV, pressure-controlled ventilation.

Preoperative patient baseline characteristics were similar between groups except for more patients with obesity and diabetes mellitus in the PCV group (table 1). As we properly randomized participants with REDCap, these group differences arose by chance.

Table 1.

Demographics and Preoperative Baseline Characteristics

	FCV n = 64	PCV n = 66
Sex, No. (%)
Male	50 (78)	46 (70)
Female	14 (22)	20 (30)
Age, yr, mean ± SD	65.5 ± 9.7	66.7 ± 10.2
Body mass index, kg/m2, mean ± SD	28.2 ± 4.3	29.4 ± 4.7
Predicted body weight*, kg, mean ± SD	70.1 ± 10.0	67.2 ± 11.3
EuroSCORE II†, %, mean ± SD	2.25 ± 2.06	2.05 ± 1.36
Preoperative physiologic parameters, mean ± SD
Left ventricular ejection fraction, %	52.6 ± 7.6	53.0 ± 8.0
Hemoglobin‡, g/dl	14.0 ± 1.7	14.0 ± 1.7
Pao2 on room air‡, mmHg	81.7 ± 12.1	79.4 ± 10.3
pH‡	7.45 ± 0.03	7.45 ± 0.04
ASA Physical Status classification, No. (%)
III	15 (23)	13 (20)
IV	49 (77)	53 (80)
Medical history, No. (%)
Chronic obstructive pulmonary disease	6 (9)	7 (11)
Respiratory infection within 1 month preoperatively	0 (0)	1 (2)
Current smoker	16 (25)	14 (21)
Pulmonary hypertension	5 (8)	5 (8)
Arterial hypertension	51 (80)	58 (88)
Diabetes mellitus	12 (19)	23 (35)
Obesity (BMI ≥ 30 kg/m2)	17 (27)	32 (48)
NYHA class ≥ III	23 (36)	20 (30)
Coronary artery disease	44 (69)	48 (73)
Previous myocardial infarction	7 (11)	7 (11)
Previous stroke	4 (6)	6 (9)
Surgical procedure, No. (%)
Isolated CABG	26 (41)	33 (50)
Isolated open valve(s) repair/replacement	11 (17)	6 (9)
Isolated minimal invasive valve repair/replacement	10 (16)	10 (15)
CABG plus valve(s) repair/replacement	13 (20)	11 (17)
Bentall procedure§	3 (5)	4 (6)
Other combined procedures (≥ 2)	1 (2)	2 (3)

^*Calculated according to the formulas 45.5 + 0.91 × (height in centimeters − 152.4) for women and 50 + 0.91 × (height in centimeters − 152.4) for men.

^†EuroSCORE II ranging between 0 and 100%, indicating the risk for postoperative in-hospital death.

^‡Missing data for one patient in the PCV group.

^§Procedure with replacement of aortic valve, root, and ascending aorta including reinsertion of coronary arteries.

ASA, American Society of Anesthesiologists; BMI, body mass index; CABG, coronary artery bypass grafting; EuroSCORE II, European System for Cardiac Operative Risk Evaluation; NYHA, New York Heart Association; Pao₂, arterial oxygen tension.

Perioperative Ventilation Interventions

Compliance-guided airway pressure optimization resulted in a perioperative aggregated mean PEEP that was 0.77 cmH₂O higher in the PCV group, whereas the ∆P was 2.2 cmH₂O higher in the FCV group. Consequently, patients in the FCV group were ventilated with tidal volumes 3.6 ml/kg larger and exhibited dynamic compliances 12 ml/cmH₂O greater than those in the PCV group. During individualized FCV, respiratory rates and minute volumes required to maintain normocapnia were lower by 11 breaths and 3.6 l per minute, respectively. The combined effect of higher ∆P and tidal volumes, but lower PEEP, minute volumes, and particularly respiratory rates, resulted in a more than 55% (9.4 J/min) reduction in calculated mechanical power. Perioperative ventilation data are presented in table 2, figure 4, and Supplemental Digital Content figures 1 to 6 (https://links.lww.com/ALN/E292). Surgical characteristics, intraoperative fluid administration, and catecholamine support until ICU admission (tables 1 and 2) did not differ significantly between groups. The onset of weaning from controlled ventilation was not significantly different with respect to CPB end. After FCV, both the duration of weaning and the total time of invasive ventilation were, on average, 62 and 106 min shorter, respectively (table 2).

Table 2.

Perioperative Characteristics of Ventilation, Surgery, and Hemodynamic Management

	FCV n = 64	PCV n = 66	P Value
Ventilation data*, mean ± SD
PEEP, cmH2O	7.1 ± 1.6	7.9 ± 1.7	0.009
ΔP†, cmH2O	13.9 ± 2.2	11.8 ± 2.1	< 0.001
Mean airway pressure, cmH2O	14.0 ± 2.1	12.1 ± 2.1	< 0.001
Tidal volume, ml/kg predicted body weight	10.9 ± 1.5	7.4 ± 0.3	< 0.001
Dynamic compliance‡, ml/cm H2O	55.8 ± 10.6	43.5 ± 8.6	< 0.001
Respiratory rate, breaths/min	6.9 ± 1.2	17.8 ± 2.2	< 0.001
Minute volume, l/min	5.1 ± 0.8	8.7 ± 1.6	< 0.001
Paco2§, mmHg	40.2 ± 3.2	42.0 ± 3.9	0.003
Resistance, cm H2O · l–1 · s–1	9.0 ± 2.2	11.5 ± 2.1	< 0.001
Calculated mechanical power∥, J/min	7.7 ± 2.0	17.1 ± 5.1	< 0.001
Surgery time, min, mean ± SD
CPB time	99.6 ± 42.5	106.1 ± 42.4	0.296
Cross-clamp time	67.4 ± 33.4	73.4 ± 35.1	0.234
Duration of surgery	234.8 ± 60.9	239.6 ± 66.8	0.557
Intraoperative fluid administration, mean ± SD
Total, l	3.40 ± 1.21	3.47 ± 0.96	0.568
Erythrocyte transfusion, units	1.11 ± 1.74	1.10 ± 1.29	0.665
Platelet transfusion, units	0.06 ± 0.24	0.11 ± 0.40	0.297
Norepinephrine administration, No. (%)
Intraoperatively	64 (100)	66 (100)	—
At ICU admission	64 (100)	64 (97)	—
Inotropic drugs administration#, No. (%)
For CPB weaning off	24 (38)	26 (39)	0.766
At ICU admission	13 (20)	9 (14)	0.582
Weaning from controlled ventilation, min, mean ± SD
Weaning start after CPB	231.0 ± 58.4	269.4 ± 152.1	0.063
Weaning time**	153.6 ± 63.2	212.7 ± 160.4	0.005
Duration of invasive ventilation**	614.1 ± 107.7	717.0 ± 230.0	0.001

^*Ventilation data show aggregated means encompassing six perioperative time points as presented in more detail in figure 4 and Supplemental Digital Content figures 1 to 6. Ventilation data are missing for two patients in the PCV group.

^†Calculated: peak inspiratory pressure minus PEEP.

^‡Calculated: tidal volume/ΔP.

^§Encompassing five measurements at five time points (Supplemental Digital Content fig. 4).

^∥Calculated according to the surrogate formulae $\frac{\text{Minute Volume}·(\text{Peak Inspiratory Pressure}+\text{PEEP}+\frac{InspiratoryFlow}{6})}{20}$ for FCV and $0.098·\text{Respiratory Rate}·\text{Tidal Volume}·(\text{PEEP}+\Delta \text{P})$ for PCV as reported by Chiumello et al.¹⁰ #Including milrinone, dobutamine, levosimendan, or combination.

^**Exclusion of data for one patient in the PCV group due to perioperative stroke with postoperative intervention.

CPB, cardiopulmonary bypass; ΔP, driving pressure; ICU, intensive care unit; Paco₂, arterial carbon dioxide tension; PEEP, positive end-expiratory pressure.

Fig. 4.

Perioperative positive end-expiratory pressures, driving pressures, tidal volumes, and respiratory rates. Boxplots show data of six perioperative timepoints during group-specific ventilation. Outliers are not shown. As ventilation data are missing for two patients in the PCV group, n = 64 for the FCV and PCV groups. The effect of the treatment assignment at single timepoints was analyzed with linear regression models including the European System for Cardiac Operative Risk Evaluation (EuroSCORE II) as an additional independent covariate. *P < 0.05. **P < 0.01. ***P < 0.001. CPB, cardiopulmonary bypass; FCV, flow-controlled ventilation; PCV, pressure-controlled ventilation.

Primary and Secondary Outcomes

Six hours after CPB, log-transformed plasmatic IL-8 levels were not significantly different in the FCV and PCV groups (table 3; Supplemental Digital Content fig. 7, https://links.lww.com/ALN/E292; additional Supplemental Digital Content referenced in this section can be accessed via this link). This result was not affected by the applied imputation methods as displayed in Supplemental Digital Content table 2. However, evaluating IL-8 at five more perioperative timepoints as secondary outcomes, log-transformed plasma levels were significantly lower in the FCV group directly after CPB (β, 0.39 pg/ml; 95% CI, 0.08 to 0.70; P = 0.015) and also 24 h after CPB (β, 0.26 pg/ml; 95% CI, 0.02 to 0.49; P = 0.032; Supplemental Digital Content table 3 and fig. 7).

Table 3.

Primary and Explorative Secondary Outcomes

	FCV n = 64	PCV n = 66	β* or OR† or HR‡ (95% CI)	P Value
Primary outcome
IL-8 levels 6 h after CPB, pg/ml	3.08	3.60	0.08*	0.573
Median (IQR)	(2.01, 4.94)	(2.67, 5.25)	(−0.17 to 0.33)
Secondary outcomes
Pao2/Fio2, mmHg, mean ± SD
After individualization of ventilation	421 ± 65	419 ± 86	−1* (−27 to 26)	0.945
After CPB weaning	342 ± 116	305 ± 133	−34* (−77 to 9)	0.116
After ICU transition	378 ± 71	338 ± 87	−41* (−69 to −13)	0.004
12 h after CPB	343 ± 129	298 ± 127	−46* (−91 to −2)	0.041
24 h after CPB	337 ± 112	271 ± 97	−67* (−104 to −31)	< 0.001
Patients with any PPCs within the first 7 postoperative days, No. (%)	57 (89)	59 (91)	1.23† (0.41 to 3.72)	0.715
Incidences of single PPCs within the first 7 postoperative days, No. (%)
Mild hypoxemia§	30 (47)	16 (25)	0.38† (0.18 to 0.79)	0.010
Moderate hypoxemia∥	23 (36)	36 (55)	2.16† (1.07 to 4.39)	0.033
Severe hypoxemia#	0 (0)	7 (11)	16.4† (1.08 to 248)	0.044
Postextubation respiratory acidosis**	11 (17)	14 (21)	1.28† (0.54 to 3.06)	0.574
Noninvasive ventilation	7 (11)	14 (21)	2.15† (0.82 to 5.66)	0.121
Reintubation	3 (5)	3 (5)	1.08† (0.23 to 5.09)	0.924
Bronchoscopy	1 (2)	3 (5)	2.51† (0.36 to 17.5)	0.352
Pleural effusions needing intervention	4 (6)	5 (8)	1.15† (0.32 to 4.15)	0.827
Pneumothorax needing intervention	4 (6)	2 (3)	0.51† (0.11 to 2.46)	0.404
Suspected pneumonia††	3 (5)	8 (12)	4.3† (0.9 to 19.6)	0.064
Confirmed pneumonia‡‡	0 (0)	6 (9)	38.6† (1.18 to 1264)	0.040
Intensive care readmission due to PPCs	1 (2)	0 (0)	0.31† (0.01 to 6.84)	0.457
Patients with any PEPCs within the first 7 postoperative days, No. (%)	44 (70)	56 (86)	2.65† (1.11 to 6.37)	0.029
Incidences of single PEPCs within the first 7 postoperative days, No. (%)
SIRS§§	40 (64)	51 (79)	2.10† (0.96 to 4.59)	0.062
Pericardial tamponade needing intervention	0 (0)	3 (5)	6.2† (0.4 to 95.1)	0.188
Other bleeding needing intervention	2 (3)	4 (6)	2.15† (0.41 to 11.2)	0.364
Infection with antibiotic treatment	4 (6)	1 (2)	0.30† (0.05 to 1.89)	0.199
Coronary intervention	1 (2)	2 (3)	1.48† (0.21 to 10.3)	0.694
Delirium∥∥	6 (9)	10 (15)	1.71† (0.59 to 4.9)	0.321
Acute kidney injury##	0 (0)	2 (3)	4.1† (0.3 to 60.0)	0.305
Intensive care readmission due to PEPCs	4 (6)	3 (5)	0.78† (0.18 to 3.30)	0.731
Hospital lengths of stay, days, median (IQR)
Surgical intensive and intermediate care unit	3 (2, 5)	3 (2, 5)	0.76‡ (0.53 to 1.09)	0.136
Surgical hospital stay	14 (11, 20)	17 (13, 24)	0.54‡ (0.33 to 0.87)	0.011
Hospital readmissions within 6 months, No. (%)	16 (25)	17 (26)	1.10‡ (0.49 to 2.47)	0.821
Death within 6 months, No. (%)	0 (0)	2 (3)	—	—

^*Beta coefficient. †Odds ratio. ‡Hazard ratio.

^§Defined as Pao₂/Fio₂ < 300 but ≥ 200 mmHg.

^∥Defined as PaO₂/Fio₂ < 200 but ≥ 100 mmHg. #Defined as Pao₂/Fio₂ < 100 mmHg. Only the most severe hypoxemia category is reported in each participant. Fio₂ estimations for Pao₂/Fio₂ calculations in spontaneously breathing patients according to oxygen flow rates and devices are specified in the Supplemental Digital Content table 3.

^**Defined as pH ≤ 7.3 and arterial carbon dioxide tension > 45 mmHg.

^††Defined as new pulmonary infiltrates and two or more of the following: temperature > 38.5° or < 35.5°C, leukocytes > 12 or < 4/nl, purulent secretions with antibiotic therapy.

^‡‡Defined as pulmonary infiltrates and microbiologic germ proof in tracheal/bronchial secretions.

^§§Defined as two or more of the following: temperature > 38° or < 36°C, leukocytes > 12 or < 4/nl, and heart rate > 90/min. SIRS data were missing for one patient in the FCV and PCV group.

^∥∥Diagnosed by the Confusion Assessment Method for the ICU.

^##Defined as Kidney Disease: Improving Global Outcomes stage ≥ 2.

CPB, cardiopulmonary bypass; Fio₂, inspired oxygen fraction; ICU, intensive care unit; Pao₂, arterial oxygen tension; PEPC, postoperative extrapulmonary complication; PPC, postoperative pulmonary complication; SIRS, systemic inflammatory response syndrome.

Oxygenation capacities with respect to Pao₂/Fio₂ were similar during controlled group-specific ventilation before and up to 1 h after CPB. After ICU transition, oxygenation was significantly better in the FCV group. Additionally, Pao₂/Fio₂ ratios were significantly higher in the FCV group 12 h and 24 h after CPB, when 93% and 99% of patients were extubated and spontaneously breathing again, respectively (table 3). Oxygenation indices of the complete nine perioperative timepoints are on display in Supplemental Digital Content table 3 and figure 8.

Regarding PPCs within the first 7 postoperative days and hypoxemia categories defined by the ranges of Pao₂/Fio₂, patients in the FCV group had more mild hypoxemias but fewer moderate and no severe hypoxemia (table 3). Although incidences of suspected pneumonia only showed a trend toward FCV, confirmed pneumonia cases with microbiologic germ proof numbered 0 of 64 in the FCV group and 6 of 66 in the PCV group (table 3). Analyses of all pneumonia cases during the primary inpatient stay confirmed these results (Supplemental Digital Content table 4). Single incidences of further PPCs such as drained pneumothorax showed no significant differences between the groups (table 3). In addition, single incidences of PEPCs were similar but patients with any PEPC within the first 7 postoperative days were fewer in the FCV group (table 3).

Postoperative hospital stay was shorter after FCV (table 3; fig. 5). As several patients were transferred to another hospital in both groups, additional analyses of total ICU and total inpatient lengths of stay were performed but had similar results as shown in Supplemental Digital Content table 4 and figure 9. Hospital readmissions and mortality 6 months postoperatively showed no group differences (table 3).

Fig. 5.

Number of patients at risk of needing postoperative intensive/intermediate care (A) or of being inpatient (B) during the primary inpatient stay. Hazard ratios and P values were calculated using a univariable Cox model containing only the treatment group as independent variable. Censoring was in case of postoperative transfer to another ICU in a different hospital (A) or in case of postoperative transfer to a different hospital (B). FCV, flow-controlled ventilation; HR, hazard ratio; ICU, intensive care unit; IMC, intermediate care unit; PCV, pressure-controlled ventilation.

Due to more patients with obesity and diabetes mellitus in the PCV group, post hoc supplemental analyses of all primary and secondary outcome data were employed with the body mass index and the prevalence of diabetes mellitus as additional independent covariates. The results remained similar compared to the primary statistical analyses except of only a trend toward less severe hypoxemia categories in the FCV group and are on display in the Supplemental Digital Content table 5.

Discussion

FCV as an alternative mode for controlled ventilation is commercially available since 2017. Inspiration in FCV implies filling of the lungs with a set constant airway flow over the complete inspiratory time from the set PEEP to the set peak inspiratory pressure. A unique feature of FCV is the controlled emptying of the lungs with the same constant flow exiting the lungs during the complete expiratory time using the Bernoulli effect (fig. 1A).⁵ This trial represents the largest randomized controlled trial on FCV to date and included patients at high risk for perioperative lung injury and postoperative complications. By titrating the ΔP to each patient’s respiratory compliance, FCV allowed tidal volumes exceeding 10 ml/kg PBW while maintaining low mechanical power. The mechanisms to reduce mechanical power included mainly low respiratory rates but also lower PEEP levels and minute volumes being necessary to ensure normocapnia. The interventions exhibited similar levels of the primary inflammatory biomarker at the predefined timepoint. However, FCV possibly improved patient-relevant postoperative outcomes, including postoperative lung function, incidences of PPCs and PEPCs, and hospital length of stay. These explorative secondary outcomes provide potential evidence that compliance-guided FCV application aiming for low mechanical power may enhance clinical outcomes.

In the past decade, large multicenter randomized clinical trials have evaluated the effect of open-lung ventilation strategies in open abdominal surgery,¹⁷ in obese patients,¹⁸ during one-lung ventilation,¹⁹ and in cardiac surgery.⁴ Regarding secondary outcome data in all of these studies, application of higher PEEP levels led to a better intraoperative oxygenation, a reduction of applied ΔP values, and/or improved lung compliance. However, the negative effects of applying higher PEEP were increased incidences of intraoperative hypotension or the need for vasopressor therapy.^4,17,18 PPCs were defined as primary study outcomes in all of these trials, but only one trial could demonstrate the superiority of applying an individualized, perioperative open-lung ventilation strategy during one-lung ventilation.¹⁹ In FLOWVENTIN HEARTSURG, a compliance-guided individual PEEP within a range of 5 to 12 cmH₂O was applied in the interventional and control group. The rationale of this approach was to minimize the risk of hemodynamic instabilities at higher PEEP levels in this fragile patient population with strict intraoperative blood pressure targets. On the other hand, we still aimed to achieve the positive effects of an optimized PEEP strategy on oxygenation in surgical procedures with repetitive phases of apnea and a possible reduction of the applied ΔP.

Elevated ΔP values are associated with a higher mortality in patients with acute respiratory failure and distress syndrome.²⁰ The intraoperatively applied ΔP has also been shown to impact on incidences of PPCs.²¹ In this study, the ΔP in the FCV group was higher than in the PCV group. However, airway pressures in FCV, including derived parameters like ΔP, are measured in the trachea at the distal end of the endotracheal tube. This is in sharp contrast to conventional ventilators where airway pressures are measured close to or within the ventilator. Furthermore, in FCV the effective alveolar ΔP is lower than the intratracheally measured ΔP which is due to the constant and continuous flow profiles in FCV during inspiration and expiration.^7,22

Different studies could demonstrate similar results for oxygenation and PPCs when patients were conventionally ventilated with 6 versus 10 ml/kg PBW in orthopedic and noncardiothoracic/nonintracranial surgery.^23,24 Also in the setting of cardiac surgery, ventilation with 6 versus 10 ml/kg PBW did not affect the primary outcome of invasive ventilation time.²⁵ In the current study, tidal volumes exceeding 10 ml/kg PBW were applied in the FCV group. Whereas mechanical power calculations are missing in the other studies, the FCV approach in this study decoupled high tidal volumes from increased mechanical power. Dividing the aggregated means of mechanical power by the respiratory rate reveals a higher total inspiratory energy per breath of 1.1 J in the FCV group and 1.0 J in the PCV group. Thus the main mechanism of mechanical power reduction was the lower respiratory rate during individualized FCV.²⁶

Spraider et al. reported the first randomized controlled trial on FCV in the setting of on-pump cardiac surgery.⁷ FCV with compliance-guided PEEP and ΔP (n = 24) was compared with conventional PCV including a compliance-optimized PEEP and tidal volumes of 6 to 8 ml/kg PBW (n = 26). In contrast to our study, ventilation protocols were only maintained after induction of general anesthesia until postoperative admission to the ICU. The intraoperative ventilation data from the study from Spraider et al. are consistent with our data and also resulted in a distinct reduction of the intraoperatively applied mechanical power. Spraider et al. defined the Pao₂/Fio₂ ratio 15 min after chest closure as the primary study outcome, which was higher in the FCV group. The comparative study could not demonstrate significant effects on secondary clinical outcomes like PPCs and PEPCs or hospital length of stay, possibly due to a smaller group size.

Possible mechanistic effects of inspiratory and expiratory flow control on lung aeration and the high efficacy regarding oxygenation and decarboxylation capacities have been demonstrated in computed tomography scans and data of electrical impedance tomography. The results of lung healthy porcine models and clinical data during individualized FCV application consistently show an increased lung aeration, especially in dependent lung regions and globally well-aerated lungs.^6–8 In a porcine model of ARDS, during FCV with similar low tidal volumes and respiratory rates compared to volume-controlled ventilation, normally aerated areas were also increased in dependent lung zones.²⁷ These effects favor less alveolar dead space ventilation by an increased ventilation–perfusion match. Together with the lower respiratory rate and less anatomic dead space ventilation, a minimized functional dead space during FCV is part of the physiologic explanation for the lower Paco₂ levels observed despite decreased minute volumes compared to PCV. As a further aspect of CT scans, high tidal volumes in individualized FCV caused a similar amount of overinflated lung tissue compared to conventional low tidal volume ventilation.^6,7 Higher PEEP levels in the PCV group could be attributable to more patients with obesity in the control group. However, recruitment and a better aeration of dependent lung regions due to the inspiratory and expiratory flow profiles of FCV (fig. 1A) may reduce the need for higher PEEP levels, at least in lung-healthy patients. Lower PEEP levels were also found in the FCV group of the comparator trial in a cardiac surgical cohort.⁷

To date, only two studies of porcine ARDS models have shed light on longer-lasting effects of FCV on the lung ultrastructure.^27,28 The control of inspiratory and expiratory airway flows for 3 to 5 h reduced histopathological lung injury, as indicated by less edematous alveolar septa, migration of fewer inflammatory cells, and preserved surfactant protein A concentrations compared to uncontrolled expiration. Contrary to diseased lungs, the control of expiratory airway flows in healthy porcine lungs for up to 48 h did not result in a difference of ventilator-induced, histological lung injury despite a remarkable reduction of dissipated energy compared to conventional volume-controlled ventilation.²⁹ Further translational and mechanistic research including different advanced lung imaging techniques is warranted to fully elucidate the mechanisms underlying FCV’s apparent lung protective effects in health and disease.

Study Limitations

This trial was conducted in a single-center with specific expertise in FCV. Thus the external validity of the study results is limited. By chance, more patients in the control group suffered from a metabolic syndrome with the potential to affect postoperative outcomes such as incidences of PPCs.¹⁸ Participant imbalances were addressed in post hoc analyses but obtained similar results compared to the primary analysis plan. Double blinding during the ventilation interventions with different ventilators was impossible, and only postoperative outcome assessors, data analysts, and patients could be blinded. A biased perioperative management of the anesthesia, surgical, and intensive care team might have confounded postoperative outcomes.

A recruitment maneuver for homogenization of lung ventilation before the incremental PEEP trials was not part of the study protocol. This is particularly relevant as the prevalence of obesity was higher in the control group, and obese patients are prone to lung derecruitment. Additionally, the compliance-guided PEEP was determined in a closed thorax condition and not reevaluated after median sternotomy and CPB after collapse and reinflation. As these conditions may include marked differences in respiratory system mechanics, the applicability of the initial PEEP settings may be significantly limited.

By predefining specific study stop criteria in the published study protocol,¹⁵ primary and secondary outcome data analyses were per protocol, which might have biased the study results. Last, we assessed multiple, explorative secondary outcomes and did not adjust significant results for multiple comparisons. Thus, these results are hypothesis-generating only.

Conclusions

The introduction of FCV and its compliance-guided application is a potentially practice-changing ventilation strategy in perioperative medicine. Future clinical trials must expand to diverse surgical patient cohorts before widespread implementation and be powered for patient-relevant outcomes as primary endpoints. Very recently, a multicenter randomized controlled pilot trial has started patient recruitment in order to design a larger trial, which will evaluate the effect of individualized FCV versus conventional ventilation on PPCs in robot-assisted laparoscopic surgery.³⁰

Data Sharing

The detailed study protocol including the statistical analysis plan have been published open access (https://doi.org/10.1186/s13063-023-07201-7). The informed consent form in German and complete deidentified data sets of all primary and secondary study outcome data can be shared immediately after publication with no end date after personal request with a reasonable proposal to the corresponding author (S.B.).

Acknowledgments

Professor Dietmar Enk, M.D., Ph.D., Faculty of Medicine, University of Münster, Münster, Germany, is the inventor of flow-controlled ventilation. He made the authors familiar with the theoretical and practical considerations of energy- and power-saving invasive ventilation. The authors thank Christoph Weiss (www.innovative-ventilation.de, Bornhöved, Germany) for his excellent technical support with respect to flow-controlled ventilation throughout the study. The authors successfully collaborated with the art designer Michel Guss (www.michelguss.de, Hamburg, Germany) to create the figures. The authors would like to express their gratitude to all patients who participated in this trial. The syntax and grammar of the manuscript were enhanced with the assistance of artificial intelligence (ChatGPT, OpenAI, San Francisco, California) to ensure clarity and fluency; no modifications were made to the original content, data, or conclusions.

Research Support

The Forschungsförderung Ruhr University Bochum, Faculty of Medicine, Germany (FoRUM Program), supported this trial (grant No. K134-19 to Dr. Becker). Ventinova Medical, Eindhoven, The Netherlands, provided consumables for FCV (conventional tube adapters, cartridges, and tubing). The funders of the study had no role in study design, data collection, data analysis, data interpretation, writing of the report, or decision to submit the manuscript for publication.

Competing Interests

Dr. Becker received travel expenses from Ventinova Medical (Eindhoven, The Netherlands), for a lecture at the first FCV symposium in Rotterdam, The Netherlands, in 2022. Dr. Erdoes is an associate editor of Anesthesiology. The other authors declare no competing interests.

Reproducible Science

Full protocol available at: simon.becker@rub.de. Raw data available at: simon.becker@rub.de.

Supplemental Digital Content

Additional primary and post hoc analyses of outcome data, https://links.lww.com/ALN/E292

Supplementary Material

aln-144-582-s001.pdf ^{(1MB, pdf)}