Tidal Volume Lower than 6 ml/kg Enhances Lung Protection: Role of Extracorporeal Carbon Dioxide Removal

Patients admitted from July 2006 to September 2007 in the intensive care units of the S. Giovanni Battista-Molinette hospital (University of Turin, Turin, Italy) were enrolled. Inclusion criteria were age of at least 18 yr and diagnosis of ARDS.11Exclusion criteria were more than 3 days since they met ARDS criteria, pulmonary artery occlusion pressure greater than 18 mmHg, history of ventricular fibrillation, tachyarrhythmia, unstable angina, or myocardial infarction within preceding month, chronic obstructive pulmonary disease, chest wall abnormalities, chest tube, abdominal distension, body mass index greater than 30, pregnancy, intracranial abnormality.12 

The institutional review board (Comitato Etico Interaziendale, Regione Piemonte, Turin) approved the study. If the patient was incompetent at study entry, consent was delayed, the family was informed of the study (although not required), and the study was performed. Written permission for using collected data were hence obtained from the patient (if competent) or from the family (in case of death or if the patient remained incompetent).13 

All consecutive patients who met inclusion and exclusion criteria were treated for 72 h according to the ARDSNet  strategy.2A detailed description of the protocol can be found in the Supplemental Digital Content (see text file, Supplemental Digital Content 1, https://links.lww.com/ALN/A542).

After 72 h of ventilation according to the ARDSNet  strategy,2P^PLATwas recorded for a period of 1 h with 0.5-s inspiratory pause keeping ventilator setting constant and abolishing spontaneous respiratory muscles activity by (1) reaching a Ramsay score of sedation 5 (midazolam up to 0.15 mg · kg⁻¹· h⁻¹, morphine up to 0.03 mg · kg⁻¹· h⁻¹, and propofol up to 2 mg · kg⁻¹· h⁻¹)14; (1) increasing, if required before measurements, doses of midazolam (up to 10 mg/h) and/or propofol (150 mg/h increments every 10 min).15,16 

In the patients who had 25 < P^PLAT< 28 cm H²O, the ARDSNet  strategy2was maintained at least for the subsequent 72 h (fig. 1). In patients who had 28 ≤ P^PLAT≤ 30 cm H²O, the following strategy was implemented and maintained for at least the next 72 h (fig. 1): (1) V^Twas stepwise reduced (1 ml/kg of PBW every 4 h) until 25 < P^PLAT< 28 cm H²O; (2) to manage the reabsorption atelectasis that may occur during mechanical ventilation with low V^T,17,18PEEP-Fio²combination was set according to the “higher PEEP” arm of the ALVEOLI study12; (3) respiratory rate was increased up to 40 breaths/min, and bicarbonate was infused up to 20 mEq/h; (4) if pH was no more than 7.25, extracorporeal carbon dioxide removal was initiated by using a modified continuous veno-venous hemofiltration system equipped with a membrane lung with a total membrane surface of 0.33 m²(Decap®, Hemodec, Salerno, Italy)19(Lower ARDSNet /Carbon Dioxide Removal; fig. 2).

Femoral vein was accessed via  a double lumen catheter (14 F; Arrow International Inc. Reading, PA) inserted with the Seldinger technique and connected with the extracorporeal circuit. Blood flow was driven through the circuit by a roller nonocclusive low-flow pump (0–500 ml/min) through a membrane lung (Polystan SAFE; Maquet, Rastatt, Germany) that was connected to a fresh gas flow source delivering 100% oxygen at a constant rate of 8 l/min. Exiting the membrane lung, blood was driven to a hemofilter (Medica D200, Medolla, Italy). The resulting plasmatic water was recirculated through the membrane lung by a peristaltic pump (0–155 ml/min).

The membrane lung and the hemofilter were coupled in series to: (1) increase the pressure inside the membrane lung by adding the downstream resistance exerted by the hemofilter and therefore reduce the risk of air bubble formation20; (2) minimize the need for heparin by diluting the blood entering the membrane lung by recirculating the plasmatic water separated by the hemofilter21; (3) enhance the performance of the extracorporeal device extracting the carbon dioxide dissolved in the plasmatic water separated by the hemofilter and recirculated through the membrane lung22(fig. 2).

Pressure developed by the roller pump (arterial pressure) was measured and limited to 120–150 mmHg. Reinfusion pressure (venous pressure) and the pressure gradient across the membrane lung and the hemofilter (drop pressure = premembrane lung pressure – venous pressure) were also measured. Detectors of leaks and bubbles were inserted within the circuit. The circuit, including the membrane lung was primed with saline with a volume that ranged between 140 and 160 ml.

A starting dose of heparin (80 IU/kg bolus and 18 IU · kg⁻¹· h⁻¹infusion) was delivered by using a syringe pump included in the system. Heparin continuous infusion was hence titrated to maintain the activated partial thromboplastine time ratio to approximately 1.5.

After 72 h of Lower ARDSNet /Carbon Dioxide Removal strategy, the following weaning trial was conducted once a day: flow through the circuit was set to the lowest value (50 ml/min), V^Twas increased to 6 ml/kg PBW, and PEEP-Fio²combination was set according to the conventional ARDSNet  strategy.2If P^PLATwith these ventilator settings remained for more that 3 h at less than 28 cm H²O, extracorporeal carbon dioxide removal was interrupted, and conventional ARDSNet  ventilatory strategy reestablished.

Underlying disease responsible for ARDS, clinical status, and laboratory data on admission (the worst value within 24 h after admission) were recorded for calculation of the Simplified Acute Physiology Score II.23 

Clinical variables (P^PLAT, V^T, respiratory rate, PEEP, minute ventilation, Pao²/Fio²ratio, Paco², pH, continuous cardiac output [Vigileo system; Edwards LifeScience, Irvine, CA], heparin doses, and activated partial thromboplastine time ratio) were prospectively collected at the following scheduled times: after 72 h of ventilation according to the conventional ARDSNet  strategy and, in those patients who had 28 ≤ P^PLAT≤ 30 cm H²O, after lowering V^Tand before initiating carbon dioxide removal (baseline) and after 60–90 min (T^1.5), 24 h (T²⁴), 48 h (T⁴⁸), and 72 h (T⁷²) after initiation of carbon dioxide removal. Blood flow during carbon dioxide removal was recorded at T^1.5, T²⁴, T⁴⁸, and T⁷².

Potential complications during the procedure were prospectively classified as mechanical (cannula problems, membrane lung failure, clots in the circuit, air in the circuit, pump malfunction, tubing rupture, catheter displacement, system leaks) and patient-related (vein perforation, significant bleeding [i.e. , any bleeding event that required the administration of 1 unit of packed red cells], hemodynamic instability [i.e. , 80–90 mmHg increase or a 30–40 mmHg decrease in systolic blood pressure relative to the baseline value or need for inotropic drugs for at least 2 h to maintain systolic blood pressure higher than 85 mmHg or electrocardiogram evidence of ischemia or significant ventricular arrhythmias] ischemic/gangrenous bowel, pneumothorax, renal complications [i.e. , occurrence after initiation of carbon dioxide removal of creatinine greater than 1.5 mg/dl], infectious complications [i.e. , occurrence after initiation of carbon dioxide removal of culture proven new infection], metabolic [i.e. , occurrence after initiation of carbon dioxide removal of glucose of at least 240 mg/dl or hyperbilirubinemia], thromboembolic complications [i.e. , occurrence after initiation of deep venous thrombosis or pulmonary embolus], and neurologic complications [i.e. , occurrence after initiation of carbon dioxide removal of cerebral infarction, or clinical seizure, or cerebral hemorrhage or cerebral edema]), and their occurrence was recorded during the period that the Lower ARDSNet /Carbon Dioxide Removal strategy was recorded.

A pulmonary computed tomography (CT) scan of the whole lung was performed in all patients after study enrollment. The CT scan was repeated after approximately 72 h of ventilation with the Lower ARDSNet /Carbon Dioxide Removal strategy in those patients who had 28 ≤ P^PLAT≤ 30 cm H²O and after approximately 72 h of ventilation with the conventional ARDSNet  strategy in some of the patients who had 25 < P^PLAT< 28 cm H²O (fig. 1). (See text file, Supplemental Digital Content 1, which is a detailed description of the methods used, https://links.lww.com/ALN/A542).3 

Weight and volume of the entire lungs and of the nonaerated, poorly aerated, normally aerated, and hyperinflated compartments were assessed.13,24Protected tidal inflation and tidal hyperinflation were defined as the volume of the normally aerated and hyperinflated compartment at end-inspiration minus the volume of the normally aerated and hyperinflated compartment at end-expiration, respectively. Tidal recruitment of the nonaerated compartment was defined as the volume of the nonaerated compartment at end-expiration minus the volume at end-inspiration. All were expressed as percent of the total tidal inflation-related change in CT lung volume.3,24 

A bronchoalveolar lavage was performed in all patients after study enrollment.25The bronchoalveolar lavage was repeated after approximately 72 h of ventilation with the Lower ARDSNet /Carbon Dioxide Removal strategy in the patients who had 28 ≤ P^PLAT≤ 30 cm H²O and after approximately further 72 h of ventilation with the conventional ARDSNet  strategy in some of the patients who had 25 < P^PLAT< 28 cm H²O (fig. 1). Interleukin 6 (IL-6), IL-8, IL-1b, and IL-1 receptor antagonist (IL-1Ra) were measured. (See text file, Supplemental Digital Content 1, which is a detailed description of the methods used, https://links.lww.com/ALN/A542).25 

Data are expressed as mean ± SD of the mean or median and interquartile (25th and 75th percentile) range as appropriate. To evaluate differences between the two groups, the Fisher exact test for categorical variables and the t  test with unequal variance for continuous variables were used. Data at different times during carbon dioxide removal were compared by analysis of variance (ANOVA) for repeated measures by using a Bonferroni correction. If significant (P ≤ 0.05), the values at baseline, T^1.5, T^24,T⁴⁸, and T⁷²were compared with those obtained after 72 h of ventilation according to the ARDSNet  strategy by using a paired t  test as modified by Dunnett. All tests were two tailed. Analysis was carried out by using the SPSS software package (SPSS Inc. Chicago, IL).

Of the 32 patients who matched study criteria, 22 patients had 25 < P^PLAT< 28 cm H²O and 10 patients had a 28 ≤ P^PLAT≤ 30 cm H²O. Patient characteristics are shown in table 1. Age, gender, Simplified Acute Physiology Score II, and underlying diseases responsible for ARDS did not differ between the two groups; Paco²:Fio²ratio was lower in patients with 28 ≤ P^PLAT≤ 30 cm H²O than in patients with 25 < P^PLAT< 28 cm H²O (P < 0.01).

Lungs were heavier, and the extent of the hyperinflated, nonaerated, and poorly aerated CT lung compartments was larger and the extent of the normally aerated compartment was smaller in patients with 28 ≤ P^PLAT≤ 30 cm H²O than in patients with 25 < P^PLAT< 28 cm H²O (all P < 0.001) (table 2). Protected tidal inflation was smaller and tidal hyperinflation was larger in patients who had 28 ≤ P^PLAT≤ 30 cm H²O than in the patients who had 25 < P^PLAT< 28 cm H²O (19 ± 6 vs.  81 ± 6 and 67 ± 5 vs.  11 ± 4% of the total tidal inflation-associated change in CT lung compartments, respectively; P < 0.01). Tidal recruitment of the poorly aerated compartment did not differ between the two groups of patients.

In the patients who had 28 ≤ P^PLAT≤ 30 cm H²O, V^Twas decreased from 6.3 ± 0.2 to 4.2 ± 0.3 ml/kg PBW and P^PLATdecreased from 29.1 ± 1.2 to 25.0 ± 1.2 cm H²O (all P < 0.001). Despite the increase in respiratory rate (from 31.2 ± 2.3 to 37.0 ± 1.9 breaths/min; P < 0.001) and bicarbonate infusion (20.2 ± 0.8 mEq/h), the reduction in minute ventilation (from 12.03 ± 2.77 to 9.03 ± 1.18 l/min; P < 0.001) increased Paco²(from 48.4 ± 8.7 to 73.6 ± 11.1 mmHg; P < 0.001) and decreased pH (from 7.36 ± 0.03 to 7.20 ± 0.02; P < 0.001). The increase in PEEP (from 12.1 ± 2.5 cm H²O to 15.2 ± 0.8 cm H²O, P < 0.001) attenuated the reduction in Pao²/Fio²(from 135 ± 30 to 124 ± 29; P < 0.01) associated to the lowering of V^T(fig. 3). All patients met pH criteria for carbon dioxide removal and were therefore connected to the veno-venous bypass.

Sixty to ninety minutes of veno-venous bypass decreased Paco²to 50.4 ± 8.2 mmHg and increased arterial pH to 7.32 ± 0.03 (P < 0.001). After 72 h of extracorporeal support, Paco²and arterial pH were 47.2 ± 8.6 mmHg and 7.38 ± 0.04, respectively (P < 0.001; fig. 4). The extracorporeal veno-venous carbon dioxide removal device was used for 144 (84, 168) h. The pump-driven blood flow through the circuit ranged between 191 and 422 ml/min (5–10% of cardiac output; table 3). Heparin infusion ranging between 3 and 19 IU/kg was needed to maintain activated partial thromboplastine time ratio between 1.1 and 1.7 (table 3).

No patient-related complications were observed. Mechanical complications are reported in table 4. In three cases, the 14-French double-lumen catheter had to be replaced by two 8-French simple-lumen catheters (one for each femoral vein) due to recirculation issues (two cases) and catheter kinking (one case). None of the reported malfunctions impaired patient status during the procedure. The membrane clotting observed in three patients did not result in additional transfusion. The Lower ARDSNet /Carbon Dioxide Removal strategy did not require any increase in nursing resources. A technician with expertise in the Decap® system was present (9 am–5 pm) during the treatment of the first five patients.

Average lung density histograms of tidal changes in CT lung compartments in patients who had 28 ≤ P^PLAT≤ 30 cm H²O during the ARDSNet  strategy are shown at study entry (fig. 5A, left) and after 72 h of Lower ARDSNet /Carbon Dioxide Removal strategy (fig. 5A, right). Ventilation with the Lower ARDSNet /Carbon Dioxide Removal was associated with: (1) the reduction of lung weight and of the extent of hyperinflated, nonaerated, and poorly aerated CT lung compartments and an increase of the extent of normally aerated lung compartments (P < 0.001; table 2); (2) increase of protected tidal inflation (from 19 ± 6% to 86 ± 8% of the total tidal inflation-associated change in CT lung compartments; P < 0.01); (3) the reduction of tidal hyperinflation (from 67 ± 5% to 5 ± 4% of the total tidal inflation-associated change in CT lung compartments; P < 0.01); (4) a significant improvement of Pao²/Fio²(from 136 ± 30 to 221 ± 56; P < 0.001) (fig. 3). Same data for the patients who during the ARDSNet  strategy had 25 < P^PLAT< 28 cm H²O are shown at study entry (fig. 5B, left). In 12 of the 22 patients, CT scans were obtained after further 72 h of conventional ARDSNet  strategy (fig. 5B, right); these data were retrospectively obtained by using CT scan performed for other studies or for clinical reasons. Relative to study entry, further 72 h of conventional ARDSNet  strategy were also associated to the reduction of lung weight and of the extent of the hyperinflated, nonaerated, and poorly aerated CT lung compartments and the increase of the extent of the normally aerated lung compartments (P < 0.05; table 2) with no changes in amount of protected tidal inflation and tidal hyperinflation (fig. 5B, left); a significant improvement of Pao²/Fio²(from 185 ± 60 to 301 ± 42; P < 0.001) was also observed.

Pulmonary concentration of inflammatory cytokines at study entry were lower (P = 0.001) in patients who during the ARDSNet  strategy had 25 < P^PLAT< 28 cm H²O than in patients who had 28 ≤ P^PLAT≤ 30 cm H²O. In the former, further 72 h of conventional ARDSNet  strategy did not modify concentration of pulmonary inflammatory cytokines (data were retrospectively obtained in 15 of the 22 patients by using bronchoalveolar lavage performed for other studies). In the latter, the use of the Lower ARDSNet /Carbon Dioxide Removal strategy was associated with a significantly reduced concentration of pulmonary inflammatory cytokines (P = 0.001; fig. 6). Bronchoalveolar lavage procedure did not cause complications in any of the studied patients.

Use of V^Tlower than 6 ml/kg PBW in patients who during ARDSNet  ventilation had 28 ≤ P^PLAT≤ 30 cm H²O was associated to a significant reduction of inflammatory and morphological markers of ventilator-induced lung injury. Respiratory acidosis consequent to further V^Treduction was effectively and safely corrected by using a modified renal replacement device incorporating a membrane lung.

These results support the use of extracorporeal carbon dioxide removal as a tool that, integrated with conventional ventilation, allows more protective ventilator settings.9However, these data can only be used as proof of principle for the following reasons. (1) The confounding effect of time on the observed improvement of physiologic, radiographic, and inflammatory parameters before and after 72 h of Lower ARDSNet /Carbon Dioxide Removal strategy cannot be ruled out because we don’t have a control group of patients with 28 ≤ P^PLAT≤ 30 cm H²O who received usual care without Lower ARDSNet /Carbon Dioxide Removal. Moreover, the study design prevents us from knowing what the respective effect of decreasing V^T, increasing PEEP, and application of carbon dioxide removal. (2) We could not assess weather the observed substantial decrease in Paco²with the limited blood flow was exclusively related to the device used in the study; measurements of extracorporeal carbon dioxide transfer and of total body carbon dioxide production were not performed. (3) Studies in experimental models suggest that unbuffered respiratory acidosis may reduce ventilator-induced lung injury.26It is therefore possible that physiologic, morphological, and inflammatory variables would have improved as much or more if we reduced V^Tand P^PLATwithout carbon dioxide removal. (4) None of the patients who had 28 ≤ P^PLAT≤ 30 cm H²O during “ARDSNet ” ventilation were able to reach the target P^PLATvalues (25–28 cm H²O) with the increase in respiratory rate and bicarbonate infusion. This may be because V^Tin three patients was reduced more than required; P^PLATin these patients before initiating extracorporeal carbon dioxide removal was 24.2, 23.3, and 24.1 cm H²O, values that are lower than the target P^PLAT. Moreover tris-hydroxymethyl aminomethane, a non-carbon dioxide-generating buffer that was recently shown to efficiently manage respiratory acidosis,27,28was not used in the current study.

Gattinoni et al.  hypothesized that, “to rest the lung,” oxygenation via  mechanical ventilation could be dissociated from decarboxylation via  extracorporeal carbon dioxide removal.5In that study, mechanical ventilation was limited to apneic oxygenation and to 3–5 sighs every minute with peak inspiratory pressure not higher than 35–45 cm H²O; PEEP ranged between 15 and 25 cm H²O. carbon dioxide removal was performed by using a pump-driven veno-venous bypass, allowing blood flow to pass through two membrane lungs (9 m²total membrane surface area). Extracorporeal blood flow was progressively raised from 200 to 300 ml/min to the selected maintenance flow (20–30% of cardiac output). Although the observed mortality rate was lower than expected, several episodes of severe bleeding were reported.5A subsequent randomized clinical trial did not confirm these data.6Concerns have been raised regarding the standard use of extracorporeal support6,29because of the high incidence of serious complications such as hemorrhage, hemolysis, and neurologic impairments.8As a result, extracorporeal carbon dioxide removal has been restricted to the sickest patients in whom all other treatments have failed30and to centers with large expertise.31 

Retrospective evaluation of the ARDSnet  database suggested that V^Treduction would have improved outcome, even in patients who already had P^PLAT< 30 cm H²O.4Moreover, physiologic32and morphological3evidences of tidal hyperinflation have been described in patients in whom the ARDSNet  strategy resulted in 28 ≤ P^PLAT≤ 30 cm H²O. These data have therefore challenged the view of extracorporeal support only as rescue therapy33,34,35and have generated the hypothesis that extracorporeal carbon dioxide removal may be incorporated in a lung protective strategy to allow further reduction of V^Tand P^PLAT.9 

The concept originally developed by Pesenti et al.  of removing “only a portion of carbon dioxide production”10has been recently implemented in new devices that may reduce side effects, complexity, and costs of extracorporeal carbon dioxide removal.19,33Bein et al.  recently reported the use of a pump-less extracorporeal device in ARDS patients.34Retrospective analysis of 90 patients demonstrated that using this device was possible to obtain physiologic values of Paco²(31–42 mmHg) and pH (7.38–7.50) despite ventilation with low V^T(320–470 ml). However, the authors reported a complication rate of 24%, including limb ischemia, compartment syndrome, and intracranial hemorrhage. In addition, continuous intravenous infusion of norepinephrine was needed to maintain the pressure gradient between arterial and venous blood.

In the current study, carbon dioxide removal was performed through a dedicated pump-driven extracorporeal veno-venous circuit with a neonatal membrane lung and a hemofilter coupled in series (fig. 2). The main elements that characterize and differentiate this system are: (1) a blood flow lower than the one used in standard carbon dioxide removal (191–422 ml/min [5–10% of cardiac output]versus  1.5–2.0 l/min [20–30% of cardiac output])5,6,36; (2) a small neonatal membrane lung (0.33 m²) instead of two large adult membrane lung (3–4.5 m²each)5,6,36; (3) the use of 14-French double-lumen catheters instead of large 21- to 28-French double- or single-lumen catheters9; (4) a priming volume smaller than currently used (140–160 ml vs.  1,500–1,800 ml)37; (5) a relatively small infusion rate of heparin (3–19 IU/kg) and less hypocoagulation than used in previous studies (activated partial thromboplastine time ratio of 1.1–1.7 vs.  2.0–2.5).5,6 

Seventy-two hours of use of this extracorporeal bypass was associated to a 33.6 ± 6.3% reduction of Paco²(from 73.6 ± 11.1 to 48.5 ± 6.3 mmHg, P < 0.001) sufficient to normalize arterial pH (from 7.20 ± 0.02 to 7.38 ± 0.04, P < 0.001) while ventilating patients with V^Tranging between 167 and 340 ml (3.7–4.6 ml/Kg PBW) and minute ventilation ranging between 8.1 and 11.9 l/min. No adverse events in terms of vein perforation, significant bleeding, hemodynamic instability, ischemic/gangrenous bowel, pneumothorax, and renal, infectious, metabolic, thromboembolic, and neurologic complications were observed during the 141 ± 69 h in which the device was in clinical use. However, the level of blood flow required to normalize arterial pH (up to 422 ml/h) was relatively high for a circuit like the one used in the current study38; therefore, the 14-French double-lumen catheter had to be replaced in three cases by two 8-French simple-lumen catheters (one for each femoral vein).

Reabsorption atelectasis may occur during mechanical ventilation with low V^T, depending on the Fio², the regional ventilation/perfusion ratios and the end-expiratory lung volume.17In a recent study, Dembinsky et al.  randomized animals with ARDS to be ventilated for 24 h with a V^Tof 3 ml/kg or with a V^Tof 6 ml/kg; consequent respiratory acidosis was managed in the former group removing carbon dioxide via  a pumpless system.18The study showed that, despite a significant reduction of P^PLAT, organ function and organ injury assessment did not reveal significant improvements when compared with conventional strategy. On the contrary, pulmonary gas exchange was impaired because of increased pulmonary ventilation/perfusion mismatch. In the current study, after 72 h of Lower ARDSNet /Carbon Dioxide Removal strategy, we observed (1) the reduction of lung weight and of the extent of the nonaerated and poorly aerated; (2) the increase of the extent of the normally aerated CT lung compartments lung (table 2); (3) a significant improvement of Pao²/Fio²(from 136 ± 30 to 221 ± 56; P < 0.001) (fig. 5). These differences may be explained by the fact that, whereas PEEP was set to 5 cm H²O in both groups in the Dembinsky et al.  study, PEEP in our study was increased from 12.1 ± 2.5 cm H²O to 15.2 ± 0.8 cm H²O (P < 0.001).12,39 

In conclusion, this study suggests that further reduction of V^Tminimizes tidal hyperinflation and attenuates pulmonary inflammation in ARDS patients who have a P^PLATranging between 28 and 30 cm H²O when ventilated with the ARDSNet  ventilation. Extracorporeal carbon dioxide removal effectively and safely managed the respiratory acidosis consequent to V^Tlower than 6 l/kg PBW and reestablished a normal arterial pH. This proof of concept study provides preliminary clinical evidence that extracorporeal lung support may integrate conventional care and allow the use of more protective ventilator settings. Additional clinical studies are required to further confirm these results.