Extracorporeal carbon dioxide removal for patients with

acute hypercapnic respiratory failure

Lorenzo Del Sorbo, Lara Pisani, Claudia Filippini, Vito Fanelli, Luca Fasano,

Pierpaolo Terragni, Andrea Dell’Amore, Rosario Urbino, Luciana Mascia,

Andrea Evangelista, Camillo Antro, Raffaele D’Amato, Maria José Sucre,

Umberto Simonetti, Pietro Persico, Stefano Nava, and V. Marco Ranieri

Online Data Supplement

Methods

We conducted a matched cohort study with historical control that involved patients older than

18 and younger than 90 years treated with NIV for acute hypercapnic respiratory failure due

to exacerbation of COPD of two Italian ICUs. Review boards approved the protocol, and

written consent was obtained from the patients.

NIV-plus-ECCO2R

All patients admitted for exacerbation of COPD and treated with NIV for acute hypercapnic

respiratory failure were included (May 2011-November 2013). ECCO2R was added to NIV in

patients that were considered as being “at risk of failure of NIV” when, after at least two

hours of continuous application of NIV, arterial pH was ≤7.30 with a blood pressure of CO2

(PaCO2) >20% of the baseline value, and one of the following was observed: respiratory rate

≥30 breaths/min; use of accessory muscles or paradoxical abdominal movements (1).

Exclusion criteria were mean arterial pressure <60 mmHg despite infusion of fluids and

vasoactive drugs, contraindications to anticoagulation (i.e. any of the following: platelet count

<30,000/mm3; prothrombin time-international normalized ratio (INR) >1.5; history within the

previous 3 months of stroke or severe head trauma or intracranial arterio-venous

malformation, or cerebral aneurysm, or central nervous system mass lesion; patient has an

epidural catheter in place or who is anticipated to receive an epidural catheter during the

study; history of congenital bleeding diatheses such as hemophilia; gastrointestinal bleeding

within the 6 weeks prior to study entry; patient with known esophageal varices, chronic

jaundice, cirrhosis, or chronic ascites; trauma patient), body weight >120 Kg, contraindication

to continuation of active treatment, failure to obtain consent. The institutional ethics

committee approved collection and report of data at ICU admission and hospital release of

patients that did not provide consent to treatment with ECCO2R use.

ECCO2R was implemented using a modified continuous veno-venous hemofiltration system

(Decap® Smart, Hemodec, San Pietro Vimiano (Padova), Italy) (2). Blood flow is driven

through the circuit by a roller non-occlusive low-flow pump (0– 450 ml/min) through a

membrane lung (Euroset; Medolla, Modena, Italy) that is connected to a fresh gas flow source

delivering 100% oxygen at a constant rate of 8 l/min. Exiting the membrane lung, blood is

driven to a hemofilter (Medica D250, Medolla, Italy). The resulting plasmatic water is re-

circulated through the membrane lung by a peristaltic pump (0–155 ml/min). Pressure

developed by the roller pump (arterial pressure) is measured and limited to 120–150 mmHg.

Re-infusion pressure (venous pressure) and the pressure gradient across the membrane lung

are also measured. Detectors of leaks and bubbles are inserted within the circuit. The circuit,

including the membrane lung, is primed with a volume of saline ranging between 140 and 160

ml. A starting dose of heparin (80 IU/kg bolus and 18 IU/kg/h infusion) was delivered by

using a syringe pump included in the system. Continuous infusion of heparin was hence

titrated to maintain the activated partial thromboplastine time (aPTT) ratio to approximately

1.5. This latter was checked approximately every 2-3 hours. The femoral vein was accessed

via a double lumen catheter (14 F; JOLINE GmbH & Co. KG) inserted with the Seldinger

technique and connected with the extracorporeal circuit (2).

ECCO2R was interrupted and patients reverted to the “NIV-only” treatment when all of the

following were achieved for at least 12 hours: respiratory rate <25 breaths/min; pH>7·35;

PaCO2 <20% of the baseline value; absence of use of the accessory muscles or paradoxical

abdominal movements.

NIV-only

A dataset for matched cohort analysis was created using patients treated with “NIV-only” for

COPD and acute hypercapnic respiratory failure enrolled in two previous studies performed in

the same institutions where the present investigation was carried out (3, 4). Patients were

considered for matching if considered as being “at risk of failure on NIV” using the same

criteria as for the cases i.e. when at least two continuous hours of NIV led to an arterial pH ≤

7.30 with a value of PaCO2 higher than 20% of the baseline value after and at least one of the

following was observed: respiratory rate ≥30 breaths/min; use of accessory muscles or

paradoxical abdominal movements (1). Exclusion criteria were the same criteria as for the

cases: mean arterial pressure <60 mmHg despite infusion of fluids and vasoactive drugs,

contraindications to anticoagulation (i.e. any of the following: platelet count <30,000/mm3;

prothrombin time-international normalized ratio (INR) >1.5; history within the previous 3

months of stroke or severe head trauma or intracranial arterio-venous malformation, or

cerebral aneurysm, or central nervous system mass lesion; epidural catheter in place or plan to

insert an epidural catheter during the study; history of congenital bleeding diatheses such as

hemophilia; gastrointestinal bleeding within the 6 weeks prior to study entry; patient with

known esophageal varices, chronic jaundice, cirrhosis, or chronic ascites; trauma patient),

body weight >120 Kg, contraindication to continuation of active treatment.

Study end-points

Primary end-point was the cumulative incidence of endotracheal intubation during the 28 days

after ICU admission. Decision to intubate was taken by the attending clinicians not involved

in the study when two of the followings occurred for a least two hours: respiratory frequency

>35 breaths/min; arterial pH <7.25; PaCO2 >60 mmHg; arterial PO2 (PaO2) <60 mmHg with

a fraction on inspiratory O2 (FiO2) >60%; respiratory arrest, signs of patient distress with

accessory muscle recruitment and paradoxical abdominal or thoracic motion. In addition,

intubation was performed when any of the following was observed (5): hemodynamic

instability defined as 80–90 mmHg increase or a 30–40 mmHg decrease in systolic blood

pressure relative to the baseline value or need for inotropic drugs to maintain systolic blood

pressure higher than 85 mmHg or electrocardiogram evidence of ischemia or significant

ventricular arrhythmias; need for sedation for major agitation; decreased alertness defined as a

Glasgow Coma Score <9; cardiac arrest. Patients were followed until hospital discharge or

death. Secondary endpoints were in-hospital mortality and ICU and hospital length of stay.

Potential adverse events related to ECCO2R were recorded and classified as mechanical

(membrane lung failure, clots in the circuit, air in the circuit, pump malfunction, tubing

rupture, catheter displacement, system leaks, all affecting the correct functioning of the

system) and patient-related (vein perforation at cannula insertion, 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 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 CO2 removal of creatinine greater than

1.5 mg/dl], infectious complications [i.e., occurrence after initiation of CO2 removal of

culture proven new infection], metabolic [i.e., occurrence after initiation of CO2 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 CO2 removal of cerebral infarction, or

clinical seizure, or cerebral hemorrhage or cerebral edema]) (2).

Statistical analysis

The probability of receiving ECCO2R conditionally to a number of observed covariates

presumed to be associated with the decision to use ECCO2R was assessed using a

multivariable logistic regression analysis with ECCO2R treatment as the dependent variable.

The a priori selected variables that have been associated to risk of NIV failure (1, 6) were: (1)

age; (2) forced expiratory volume in the 1st second; (3) Charlson comorbidity index (7); (4)

simplified acute physiological score (SAPS II) (8); (5) pH values before institution of

mechanical ventilation. Except for FEV1 (taken from the most recent pulmonary function

test), all variables were obtained at ICU admission.

The genetic matching method without replacement (GenMatch) was used to match

patients treated with “NIV-plus-ECCO2R” and “NIV-only” (9, 10). GenMatch is a

multivariate matching technique that aims to make the distribution of baseline characteristics

between “control” and “intervention” as similar as possible combining propensity score

matching with multivariate matching. The automated search algorithm maximizes the balance

of the covariates included in the analysis (i.e. makes the multivariate distribution of covariates

in the matched “control” and “intervention” groups as similar as possible) by: (a) estimating

the relative weight of the propensity score and of the selected individual; (b) iteratively

checking the balance and directing the search toward matches that optimize balance; (d)

selecting those weights that give the best covariate balance in the matched samples (9, 10).

The balance statistics was performed using the Wilcoxon Mann-Whitney (9, 10).

To ascertain whether inclusion criteria influenced the base-case findings, all analyses were

repeated eliminating from both groups patients with severe neurological impairment (Kelly

score ≥3) and patients with severe hypoxemia (arterial O2 pressure to inspiratory O2 fraction:

PaO2/FIO2 ratio <150) since both are associated to high risk of failure of NIV.(1, 6)

Data are presented as mean (standard deviation), or median (inter-quartile range). Comparison

between and within groups was made using the Wilcoxon-Mann Whitney test for continuous

variables and the chi-square or the Fisher’s exact test for categorical variables as appropriate.

Results are reported as hazard ratio (HR) with 95% confidence intervals (95% CI).

Cumulative incidence of endotracheal intubation during the 28 days immediately after ICU

admission was assessed considering death as a competing event using the method of Gooley.

Adjusted HR and 95% CI were estimated using the Fine and Grey model.

All statistical tests were two-sided and P values of 0·05 or less were considered statistically

significant and were conducted using Stata software version 11·0 (Stata- Corp, College

Station, Texas), R software version 2·11 (R Foundation for Statistical Computing, http:

//www.r-project.org/) and SAS software package (SAS Institute, Cary, NC; version 8·2).

RESULTS

Characteristics of patients at risk of NIV failure but not included in the study protocol

(64 patients for the “NIV-plus-ECCO2R” group and 21 patients for the “NIV-only” group) did

not differ from those included in the final analysis (Table E1).

Values of PaCO2, arterial pH and respiratory rate at baseline and after one hour of

treatment only in patients treated with “NIV-plus-ECCO2R” and in “NIV-only” before and

after GenMatch are shown in Table E2

Before matching, application of ECCO2R during NIV decreased by 79% the risk of

intubation relative to “NIV-only” (HR 0.21; 95%CI 0.06-0.67, P=0.009). Intubation rate in

the “NIV-plus-ECCO2R” group was 12% (three patients, 95% CI 2.5-31.2%), while in “NIV-

only” was 48% (40 patients, 95% CI 36.6-58.8%, P=0.0011) before matching and 33% (seven

patients, 95% CI 14.6-57.0%, P=0.1495) after GenMatch. Analysis was repeated eliminating

patients with severe hypoxemia (PaO2/FIO2 <150) and patients with severe neurological

impairment (Kelly score ≥3). In patients without severe hypoxemia, before matching

intubation was observed in two out of 22 patients (9%, 95% CI 1.12-29.2%) in “NIV-plus-

ECCO2R” and in 28 out of 56 (50%, 95% CI 36.3-63.7%) in “NIV-only” (P<0.001). After

GenMatch, intubation was observed in two out of 22 patients (9%, 95% CI 1.1-29.2%) in

“NIV-plus-ECCO2R” and in seven out of 17 patients (41%, 95% CI 18.4-67.0%) in “NIV-

only” (P=0.0262). In patients without severe neurological impairment, before matching

intubation was observed in 1 out of 16 patients (6%, 95% CI 0.2-30.2%) in “NIV-plus-

ECCO2R” and in 24 out of 59 patients (41%, 95% CI 28.1-54.2%) in “NIV-only” (P=0.0145).

After GenMatch, intubation was observed in 1 out of 16 patients (6%, 95% CI 0.2-30.2%) in

“NIV-plus-ECCO2R” and in six out of 13 patients (46%, 95% CI 19.2-74.9%) in “NIV-only”

(P=0.026). Indications and timing of intubation are presented in Table E3.

Hospital mortality, ICU and hospital length of stay in patients treated with “NIV-plus-

ECCO2R” and in “NIV-only” before and after GenMatch are shown in Table E4.

Table E1. Characteristics and outcomes of patients at risk of intubation for NIV failure but

not included in the study protocol (64 patients for the “NIV-plus-ECCO2R” group and 21

patients for the “NIV-only” group).

Data are mean (standard deviation) or median and inter-quartile range. Abbreviations: NIV: non-invasive

ventilation; ECCO2R: extracorporeal carbon dioxide removal; FEV1: forced expiratory volume in the 1st second;

SAPS II: simplified acute physiological score; PaCO2: arterial pressure of carbon dioxide; PaO2/FiO2: arterial-

to-inspiratory oxygen fraction; ICU: intensive care unit.

# values obtained after one hour on NIV.

“NIV-plus-ECCO2R” “NIV-only”

Age, years 75 (7.2) 71 (9.9)

FEV1, liters 30 (10) 29.1 (8.6)

SAPS II 36.1 (6.9) 35.8 (8.3)

Arterial pH before institution of NIV 7.25 (0.08) 7.22 (0.06)

Charlson comorbidity index 5.3 (3.4) 5.7 (3.3)

PaCO2, mmHg# 74 (64; 89) 78 (70; 88)

Arterial pH# 7.27 (7.21; 7.3) 7.25 (7.21; 7.3)

Respiratory rate, breath per min# 38 (30; 44) 35 (33; 38)

PaO2/FiO2 # 211 (138; 248) 187 (141; 245)

Endotracheal Intubation, number (%) 32 (50) 11 (52)

Hospital mortality, number (%) 11 (30) 7 (33)

ICU length of stay (days) 8 (4; 10) 8 (5; 11)

Hospital length of stay (days) 15 (10; 20) 17 (10; 27)

TABLE E2. Respiratory variables in NIV-plus-ECCO2R and NIV-only

Data are expressed as median (interquartile range). Abbreviations: NIV: non-invasive ventilation; ECCO2R:

extracorporeal carbon dioxide removal; GenMatch: genetic matching. T0: time 0, time when criteria for “risk

of NIV failure” were matched; T1: time 1, time 0 plus one hour. Comparisons between and within groups are

made using the Wilcoxon-Mann-Whitney test.

groups are made using the Wilcoxon-Mann-Whitney test

NIV-plus-ECCO2R

(N=25) NIV-only

Before matching

(N=84) P

After GenMatch

matching (N=21) P

T0 88.0 (67.0; 96.0) 82.5 (76.0; 90.5) 0.6951 82.0 (76.0; 89.0) 0.6067

PaCO2, mmHg P=0.0037 P=0.9072 P=0.7363

T1 63.0 (52.0; 84.0) 78.47 (60.2; 104.1) 0.0082 80.0 (66.0; 104.5) 0.0125

T0 7.27 (7.25; 7.28) 7.27 (7.22; 7.29) 0.5005 7.28 (7.23; 7.30) 0.2423

Arterial pH P<0.0001 P=0.2181 P=0.4979

T1 7.34 (7.32; 7.39) 7.25 (7.13; 7.33) <0.0001 7.28 (7.17; 7.30) 0.0003

T0 32 (29; 35) 32 (29; 36) 0.6280 30 (28; 32) 0.2641

Respiratory rate, breath per min P<0.0001 P<0.0001 P=0.1268

T1 22 (18; 24) 29 (26; 32) <0.0001 27 (25; 31) 0.0002

T0 168 (133; 210) 190 (144; 242) 0.2296 176 (152; 233) 0.3648

PaO2/FiO2 P=0.9493 P=0.0289 P=0.0701

T1 178 (131; 203) 233 (180; 264) 0.0007 235 (212; 262) 0.0006

Table E3. Timing and reasons of intubation in both NIV-plus-ECCO2R and NIV-only groups.

Timing (days) Reason

NIV-plus-ECCO2R

Patient #1 2 pH <7.25 and PaCO2 >60 mmHg following membrane and

circuit clotting

Patient #14 1 pH <7.25 and PaCO2 >60 mmHg following blood pump

malfunction of ECCO2R system

Patient #4 1

40 mmHg decrease in systolic blood pressure relative to the

baseline despite transfusion of two packs of red cells

(significant bleeding; retroperitoneal hematoma)

NIV-only

Patient #1 3 pH <7.25 and PaCO2 >60 mmHg

Patient #2 1 pH <7.25 and PaCO2 >60 mmHg

Patient #3 1

need for inotropic drugs (norepinephrine 18 mcg/min) and

fluids (25 mL/kg of crystalloid administered during the 6

hours interval surrounding the start of vasopressor

treatment) to maintain systolic blood pressure higher than

85 mmHg (pneumonia, septic shock)

Patient #4 1 pH <7.25, PaCO2 >60 mmHg and respiratory rate >35

breaths/min

Patient #5 1 pH <7.25, PaCO2 >60 mmHg and respiratory rate >35

breaths/min

Patient #6 1

need for inotropic drugs (norepinephrine 22 mcg/min) and

fluids (30 mL/kg of crystalloid administered during the 4

hours interval surrounding the start of vasopressor

treatment) to maintain systolic blood pressure higher than

85 mmHg (abdominal septic shock)

#7 1 pH <7.25, PaCO2 >60 mmHg and respiratory rate >35

breaths/min

TABLE E4. Study secondary end-points.

Data are expressed as number (percentage) or median (interquartile range). Abbreviations: NIV: non-invasive

ventilation; ECCO2R: extracorporeal carbon dioxide removal; ICU: intensive care unit; GenMatch: genetic

matching. Comparisons between groups are made using Fisher’s exact test. a There were 25 patients for the NIV-plus-ECCO2R treated group

b There were 84 patients for the non ECCO2R treated group

c There were 21 patients for the non ECCO2R treated group

NIV-plus-ECCO2R a NIV-only P

Hospital mortality, number (%) 2 (8)

Before matching b 29 (35) 0.0107

After GenMatch c 8 (35) 0.0347

ICU length of stay (days) 8 (7; 10)

Before matching b 8 (4; 13) 0.7351

After GenMatch c 12 (6; 15) 0.1943

Hospital length of stay (days) 24 (21; 28)

Before matching b 20 (10; 34) 0.2176

After GenMatch c 22 (13; 36) 0.8007

References

1. Confalonieri M, Garuti G, Cattaruzza MS, Osborn JF, Antonelli M, Conti G, Kodric

M, Resta O, Marchese S, Gregoretti C, Rossi A. A chart of failure risk for noninvasive

ventilation in patients with copd exacerbation. Eur Respir J 2005;25:348-355.

2. Terragni PP, Del Sorbo L, Mascia L, Urbino R, Martin EL, Birocco A, Faggiano C,

Quintel M, Gattinoni L, Ranieri VM. Tidal volume lower than 6 ml/kg enhances lung

protection: Role of extracorporeal carbon dioxide removal. Anesthesiology 2009;111:826-835.

3. Antro C, Merico F, Urbino R, Gai V. Non-invasive ventilation as a first-line treatment

for acute respiratory failure: "Real life" experience in the emergency department. Emergency

medicine journal : EMJ 2005;22:772-777.

4. Valentini I, Pacilli AM, Carbonara P, Fasano L, Vitale R, Zenesini C, Melotti RM,

Faenza S, Nava S. Influence of the admission pattern on the outcome of patients admitted to a

respiratory intensive care unit: Does a step-down admission differ from a step-up one?

Respiratory care 2013;58:2053-2060.

5. Brochard L, Mancebo J, Wysocki M, Lofaso F, Conti G, Rauss A, Simonneau G,

Benito S, Gasparetto A, Lemaire F, et al. Noninvasive ventilation for acute exacerbations of

chronic obstructive pulmonary disease. N Engl J Med 1995;333:817-822.

6. Keenan SP, Sinuff T, Cook DJ, Hill NS. Which patients with acute exacerbation of

chronic obstructive pulmonary disease benefit from noninvasive positive-pressure

ventilation? A systematic review of the literature. Annals of internal medicine 2003;138:861-

870.

7. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying

prognostic comorbidity in longitudinal studies: Development and validation. Journal of

chronic diseases 1987;40:373-383.

8. Le Gall JR, Lemeshow S, Saulnier F. A new simplified acute physiology score (saps ii)

based on a european/north american multicenter study. JAMA : the journal of the American

Medical Association 1993;270:2957-2963.

9. Sekhon JS, Grieve RD. A matching method for improving covariate balance in cost-

effectiveness analyses. Health economics 2012;21:695-714.

10. Noah MA, Peek GJ, Finney SJ, Griffiths MJ, Harrison DA, Grieve R, Sadique MZ,

Sekhon JS, McAuley DF, Firmin RK, Harvey C, Cordingley JJ, Price S, Vuylsteke A, Jenkins

DP, Noble DW, Bloomfield R, Walsh TS, Perkins GD, Menon D, Taylor BL, Rowan KM.

Referral to an extracorporeal membrane oxygenation center and mortality among patients

with severe 2009 influenza a(h1n1). JAMA : the journal of the American Medical Association

2011;306:1659-1668.