Introduction

Ventilation with low tidal volumes of 6–8 ml/kg ideal body weight (IBW) has become standard of care in ARDS patients [1]. This concept seemingly helps to decrease the risk of baro- and volutrauma on the affected lung and apparently even affects outcome favorably [2]. But today these principles are challenged: an IBW-based low Vt-strategy might not solely be sufficient to avoid tidal hyperinflation [3, 4]. Derived from these principles the concept of ultraprotective ventilation has been developed [5, 6]. However, a recently introduced concept stresses that no single variable (e.g. tidal volume or applied plateau- or delta pressure), but the mechanical energy inflicted on the lung corresponds to VILI (ventilator induced lung injury) [7]. The inevitable risk of consecutive hypercarbia might be accepted to a certain amount, but the effects of hypercarbia on whole-body physiology including the lung itself should not be neglected. Nevertheless, it has been shown that CO2 may also have a beneficial effect [8]. Side effects of hypercarbia include academia, increased shunt, shifted oxyhemoglobin dissociation curve [9], but also cardiac depression and impaired coronary artery perfusion [10]. As a consequence, further Vt-reduction might require additional extracorporeal support or dead space reductions to cope with these effects [4, 11,12,13].

The concept of extracorporeal CO2 removal was developed decades ago in the 1970s [14], but it required the growing knowledge on mechanisms of ventilator induced lung injury, recent technical developments and advanced options of dealing with complications [15, 16] to finally become clinically useful. Nowadays, different ECLA-devices (extracorporeal lung assist) are in clinical use. Other than active high-flow devices (extracorporeal membrane oxygenation, ECMO), mid-flow (about 1 l/min) and low-flow (about 500 ml/min) devices are readily used. Additionally, passive devices utilizing arteriovenous pressure difference are used.

In critically ill patients ARDS and renal failure (ARF) are commonly present simultaneously [17] and mortality remains unacceptably high [18]. Thus, both continuous renal replacement therapy (CRRT) and extracorporeal CO2-elimination can be necessary. With a large bore catheter for CRRT in place, allowing for higher flow rates than regular multi-lumen catheters the mental leap to add CO2-removal is obvious. Difficulties arise, because CO2-removal is flow dependent. Different concepts have been strived to facilitate CO2 removal in low flow rates, some of them based on the Henderson–Hasselbalch-equation to convert CO2/H + to bicarbonate, which is then dialyzed [19,20,21,22,23,24,25,26,27,28,29,30,31], others on electrical dialysis [32]. But none of these concepts has made the transition to clinical usefulness.

In this light, a device integrating both processes into one single extracorporeal circuit (CRRT-ECCO2R) offers potential advantages. Although the low blood flow that is applied during CRRT cannot provide sufficient oxygenation and only limited CO2 elimination, it might be enough to facilitate lung protective mechanical ventilation [13]. Based on this rational we aimed to test, whether minimal-flow CRRT-ECCO2R allows for a viable increase in lung protective ventilation without the undesirable side effects of hypercarbia or respiratory acidosis.

During standardized conditions and application of volume-controlled mechanical ventilation the potential to reduce Vt by addition of ECCO2R has been shown [13]. The effect of ECCO2R during pressure controlled or pressure support ventilation, that allows or assists spontaneous breathing are less defined and might vary. We purposely did not control for the ventilator mode and settings applied, moreover treating physicians were blinded to whether CO2-removal was active or not and ventilator settings were left to their judgment.

We hypothecated that a minimal-flow CRRT-ECCO2R is able to remove enough CO2 to facilitate lung protective ventilation in mechanically ventilated patients.

Methods

The local ethics committee approved the study prior to patient enrollment. We included adult patients (age > 18 years) in need for mechanical ventilation and CRRT; hypercarbia was considered neither inclusion nor exclusion criterion. Patients were not eligible for the study if they were hemodynamically unstable (doses of norepinephrine or epinephrine ≥ 0.1 µg/kg/min), if they were in severe acute respiratory distress (PaO2/FiO2 < 100), if they were thrombopenic (platelets < 50,000/µl), if they had an acute bleeding (> 200 ml/h) or if further therapy was futile due to bad prognosis or patients will. Assessed patient characteristics consisted of age, sex, weight, height, diagnoses, SAPS II score, ICU-stay, ventilator times, and duration of renal replacement therapy.

For this trial we used a combined CRRT-ECCO2R device (EQUA-smart®, Hemodec (MEDIA), Modena; Italy), basic settings were employed following manufacturer guidelines.

Ventilation was conducted using conventional ICU ventilators (Evita 4, Dräger, Lübeck, Germany or Servo-i, Maquet Critical Care, Solna Sweden). Peak inspiratory pressure (Pinsp) was limited to 30 ccmH2OO, ventilation modes used were either biphasic positive inspiratory airway pressure ventilation (Evita 4: BiPAP mode, Servo-i: BIVENT) or pressure support ventilation, depending on individual patient needs. ABGs were conducted frequently and used to modify ventilator settings aiming to keep paCO2 and pH as constant as possible during the preparatory phase of the trial.

CRRT was conducted using the CVVHDF mode, extraction dose 35 ml/kg, blood flow 300 ml/min, O2-flow 10 l/min, and all parameters were kept constant throughout. Patients were randomized to receive either pre- or postdilution.

CRRT was started as indicated without CO2 elimination. Then CO2 elimination was added and closely monitored for 6 h keeping blood- and gas flow constant throughout the study period. After the course of 6 h of CRRT plus ECCO2R we switched back to CRRT only. Treating physicians were blinded to the phase of the trial, i.e. if CO2 removal was active or just CRRT. The following parameters were collected just before starting CO2-removal as well as 1, 3, and 6 h during ECCO2R: arterial blood pressure, heart rate, mode of ventilation, respiratory rate, tidal volume, FiO2, PEEP, peak inspiratory pressure (Pinsp), and ABG (paO2, paCO2, pH, SaO2). Data were collected using the standard PDMS system in the unit (ICCA®, Philipps) and the integrated data recorder of the EQUAsmart® device. Data were extracted from both sources and stored in a Microsoft® Excel® Worksheet.

We analyzed changes in minute ventilation, tidal volume, respiratory rate, power applied to the lung, and paCO2 as well as CO2 removal.

Furthermore, we compared CO2 removal using pre- vs. postdilution. To quantify CO2 removal, we calculated the difference in CO2 contained in the afferent and efferent limb of the circuit. The calculation was performed using the equation described by Douglas et al. [33].

All parameters were tested for normal distribution using the Kolmogorov–Smirnov test, nevertheless due to small sample size we used statistical methods not reliant on normal distribution. Parameters were tested using the Kendall-test first and, if significance was found, additionally with the Wilcoxon-test for linked samples, we only report the Wilcoxon p value unless otherwise stated. CO2 removal pre- vs. postdilution was tested using Mann–Whitney-U-test for unrelated samples. Statistical calculations were done utilizing SPSS® Software Version 24, level of significance was assumed to be p < 0.05.

Results

We enrolled 14 patients, 3 of them had to be excluded from analysis due to incomplete datasets, leaving 11 patients for further analysis. Pre- and postdilution was equally distributed between the groups; patient characteristics and initial ventilator settings are shown in Tables 1 and 2. Most of our patients were ventilated in assist-control mode (9 out of 11), two were ventilated in pressure support mode. All patients were ventilated in the spirit of a lung protective ventilation strategy using low tidal volumes; mean tidal volume at baseline was 6.36 ± 0.67 ml/kg ideal bodyweight.

Table 1 Patient characteristics
Full size table
Table 2 Initial ventilator settings and oxygenation
Full size table

At baseline minute ventilation appeared to be 10.1 l/min (7.0–12.9 l/min), during CRRT-ECCO2R therapy a significant reduction in minute ventilation was observed to 9.2 l/min at 1 h (5.9–12.4 l/m, p < 0.05), at 3 h there was still a reduction compared to baseline to 8.4 l/min (5.1–14.2 l/min), which was not significant (p = 0.052), at 6 h there was still a reduction compared to baseline to 9.6 l/min (7.5–15.3 l/min), again not significant (p = 0.450) (Fig. 1).

Fig. 1
figure 1

Composite figure of the analyzed parameters; each figure shows the evolution of the parameter displayed during the course of the trial relative compared to baseline. Parameters do not shape in any consistent manner but show a trend towards reductions after 3 h of therapy. This is conclusive to the statistical measures performed. Pmean mean airway pressure, Pinsp maximum inspiratory pressure, Vt tidal volume, paCO2 arterial partial pressure of carbon dioxide

Full size image

Respiratory rate was 23.6 bpm (18–39 bpm) at baseline. It was reduced to 21.1 bpm (15–26 bpm) at 1 h, to 23 bpm (17–34 bpm) at 3 h and increased slightly to 24.8 bpm (18–36 bpm) at 6 h into therapy, none of these changes reached significance (Kendall p = 0.127).

Tidal volume was 425.5 ml (300–500 ml) at baseline. It increased to 440.6 ml (346–600 ml, p = 0.750) after 1 h and decreased significantly to 364.1 ml (258–450 ml, p < 0.05) after 3 h. After 6 h of therapy we found a decrease to 395 ml (305–550 ml) still, but this decrease was not statistically significant (p = 0.240).

There was a statistically significant decrease in paCO2 from 34.4 mmHg (29–49 mmHg) at baseline to 28.8 mmHg (20–40 mmHg, p < 0.05) at 1 h, a statistically meaningful decrease to 31.7 mmHg (27–37 mmHg, p < 0.05) was observed at 3 h, which kept constant to 31.6 mmHg (27–37 mmHg, p = 0.164) at 6 h during CRRT-ECCO2R (Fig. 2).

Fig. 2
figure 2

Combined whiskers-boxplots of studies parameters throughout the trial period. Differences in respiratory rates over time are marginal with no decrease found. Moreover, an increase is seen after 6 h. We do not have a conclusive explanation for this fact. In the first hour there was no difference in Pinsp, after 3 h the median decreased but the values also spread wider, strangely enough Pinsp increased after 6 h, we do not have an explanation for this. Counter intuitively the power applied did not decrease although CO2 was actually removed from the body, but it also did not increase. A clear decrease in arterial CO2 is seen after 1 h of therapy as expected due to extracorporeal CO2 removal. Nevertheless, CO2 is nearly back to baseline values at 3 h and stays at this level until 6 h of therapy. One explanation may be oxygenator exhaustion, but we did not have any hint towards this. paCO2 arterial partial pressure of carbon dioxide, Pinsp maximum inspiratory pressure

Full size image

We also calculated delta pressure, which was also not significantly different between data-acquisition-points (Kendall p = 0.143, data not shown).

The power transmitted to the lungs owing to ventilator settings at each data-acquisition-point was calculated. We found the values to be 18.4 J/min (11.8–34 J/min) at baseline, 17.8 J/min (8.6–29.8 J/min, p = 0.695) after 1 h, 15 J/min (6.2–32.8 J/min, p = 0.054) after 3 h, and 16.3 J/min (8.0–23.5 J/min, p = 0.175) after 6 h of therapy.

Total CO2 removal of the tested device averaged over all recordings was 20.7 ml CO2/min (18.7–23.8 ml CO2/min) regardless of the mode of dilution applied.

Regarding the mode of dilution, we found no significant difference between predilution (CO2 elimination 19.4 ml CO2/min, 15.0–26.1 ml CO2/min) and postdilution (CO2 elimination 23.1 ml CO2/min, 17.5–28.7 ml CO2/min) (p = 0.26) (Tables 3, 4).

Table 3 Displayed are the measured or calculated values for each patient at each data acquisition point
Full size table
Table 4 Displayed are the mean and standard deviation for all measured patients at the data acquisition points
Full size table

No complications or adverse events occurred during conduction of the trial.

Discussion

We used a combined approach that incorporates extracorporeal CO2 removal in a system for hemodialysis (CRRT-ECCO2R). In patients with combined renal and respiratory failure this system was in our trial capable to safely reduce minute ventilation statistically significant in the first hour of therapy. Nevertheless, it was not able to keep or even improve this result throughout the next hours of our trial. Furthermore, it is debatable if a mean reduction of 0.8 l/min although statistically significant is clinically meaningful. Especially because the statistically significant reduction in tidal volume appeared 2 h later at 3 h into therapy, when the decrease in minute ventilation was not statistically significant.

This argument gets even stronger when taking into account that CO2 decreased significantly also at that exact point in time. But there is no significant corresponding decrease in respiratory rate. Calculating the power transmitted, although the formula has not been widely used yet, shows the same result; there is a non-significant decrease in power at this exact time.

Terragni et al. using the same device [13] found significant reductions in minute ventilation and concluded that CRRT-ECCO2R is sufficient to facilitate lung-protective ventilation. In their trial, a fixed tidal volume of 6 ml/kg was aimed and reached. After achieving this goal, the team removed CO2 in the intervention group and found they were able to manage the increased CO2 load produced by decreased tidal volumes. Looking at their data it becomes obvious that only a fixed amount of CO2 was removed bringing CO2 back to baseline before tidal volumes were reduced. Our approach was different. We adopted the ventilator settings as they stood and removed CO2 first. Treating physicians were blinded to whether CO2 removal was active or not. They were not briefed about the trial but kept going with routine practice to ventilate a patient according to the ALARA principle (as-low-as-reasonably-achievable) we use in our unit. We think our approach is much less of a laboratory situation but reflects everyday life closer than the approach Terragni et al. used. The authors were able to show, that in a controlled environment, CRRT-ECCO2R enables treating physicians to decrease tidal volume and control for increased carbon dioxide via the extracorporeal circuit. In our uncontrolled situation CO2 was removed first and we observed, if lung protective ventilation resulted. In this setting, we were not able to show statistical significance, although absolute trends towards protective ventilation could be found. Data provided in the aforementioned article do not allow for calculation of applied power retrospectively.

We were able to confirm the findings of Terragni et al. in that both studies show a constant portion of CO2 being removed by this system. As expected by the minimal-flow, the amount of CO2 elimination is limited. The capacity to remove CO2 we found is likely to be insufficient in some cases. Since there is only very little experience with this system and technique, further evidence must be generated. Another drawback is the very limited amount of oxygenation at these flow rates making it feasible for patients with partial respiratory insufficiency only.

Furthermore, we evaluated the effect of pre- versus postdilution on CO2-removal. By theory, predilution would be more efficient in terms of CO2-removal than postdilution because higher efflux rates can be achieved. In spite of this theory we did not find any significant difference in CO2-removal between the two modes. One possible explanation could be a suboptimal membrane lung, so that pre- vs. postdilution should be tested using another membrane lung to verify or falsify this possibility.

There are limitations to our study. First, our sample size was small; we might have missed statistical significance for this reason, although this seems unlikely by looking at our data. Second, this was a feasibility study; we did not have a control group. Third, inclusion criteria were kept very broad so that we included only due to the need of CRRT and mechanical ventilation. All patients included were not in need for ECCO2R, they did find without it before and after. We, therefore, do not know if the system works for patients with a heavy CO2 load. On the other hand, this is strength as well, because we can largely exclude recruiting bias. From our results one can see that CO2 removal consistently was 21 ml CO2/min. This is only about one-tenth of the normal resting CO2 production in an adult. Taking this into account it is very unlikely that a significant amount of CO2 can be removed using this system in a severely hypercarbic patient who would need it the most.

After all, experience is limited with these jack-of-all-trades devices, making it hard to draw general conclusions on their effectiveness. Further research needs to be conducted with those systems to find the right patients benefiting from it, since from our results shown here one cannot conclude that hypercarbic patients will benefit from the small amount of CO2 removal we found.

Conclusion

Based on our current knowledge the least harmful approach to mechanical ventilation should be chosen to avoid ventilator-induced injury. Although extracorporeal CO2 removal using minimal-flow systems suffers from a limited efficiency, the concept of reduced invasiveness, which is enabled by vascular access with smaller cannulas, is appealing. This is especially true since fusing CRRT and ECCO2R in patients that require both treatments might enhance the benefit.

With our study we were able to find detached statistically significant reductions in minute ventilation, tidal volume, and paCO2. In our study, we found no significant decrease in applied power besides aforementioned statistically significant reductions in surrogates for lung protection during mechanical ventilation. From our trial, we need to draw the conclusion that ECCO2R using minimal flow rates with that merely small amount of CO2 removal we found is most likely not sufficient to benefit patients with heavy CO2 load.