- Email: [email protected]
Venovenous extracorporeal membrane oxygenation in neonatal respiratory failure: does routine, cephalad jugular drainage improve outcome?
Venovenous Extracorporeal Membrane Oxygenation in Neonatal Respiratory Failure: Does Routine, Cephalad Jugular Drainage Improve Outcome? By Erik D. Skarsgard, Douglas R. Salt, Shoo K. Lee, and the Extracorporeal Life Support Organization Registry Vancouver, British Columbia
Background/Purpose: Extracorporeal membrane oxygenation (ECMO), may be life saving for infants with severe respiratory failure, and when possible, veno-venous bypass through a jugular double lumen cannula, can be expected to provide satisfactory support for most patients. Some ECMO centers favor routine placement of a cephalad jugular cannula for the theoretical benefits of augmented (desaturated) venous return, reduction of atrial recirculation, and cerebral venous decompression. The purpose of this study was to querie the ELSO registry for patients who had undergone VV-ECMO and compare outcomes for patients with a double lumen cannula only (VVDL), with those who had both a double lumen and cephalad jugular cannula (VVDL ⫹ V). Methods: With institutional review board (IRB) approval, the Extracorporeal Life Support Organization (ELSO) registry (Ann Arbor, MI) was queried from January 1, 1989 to December 31, 2001, and all “neonatal respiratory” patients undergoing VV-ECMO via either the VVDL or VVDL ⫹ V modes were identified. Group comparisons by age, diagnosis, hours on bypass, mean flow rates (Q) at 4 and 24 hours, mean airway pressures (MAP) at initiation and at 24 hours of bypass, complications (including neurologic and cannulaspecific), need for conversion to veno-arterial (VA) ECMO, and survival were performed. A similar analysis was performed on a congenital diaphragmatic hernia (CDH) patient subgroup. Student’s t tests were used to compare means
T
HE USE of extracorporeal membrane oxygenation (ECMO) in the support of neonates with profound respiratory failure associated with a variety of conditions is well established. When hemodynamically feasible, the preferred technique of bypass involves use of a double lumen cannula placed through the right internal jugular vein into the right atrium, which provides both venous From the Departments of Surgery, Pediatrics, and Perfusion Services, British Columbia Children’s Hospital; the Centre for Health Innovation and Improvement; and the University of British Columbia, Vancouver, British Columbia, Canada. Presented at the 35th Annual Meeting of the Canadian Association of Paediatric Surgeons, Niagara-on-the-Lake, Ontario, Canada, September 18-21, 2003. Address reprint requests to Erik D. Skarsgard, MD, B.C.’s Children’s Hospital, KO-123, 4480 Oak St, Vancouver, B.C. V6H 3V4, Canada. © 2004 Elsevier Inc. All rights reserved. 0022-3468/04/3905-0004$30.00/0 doi:10.1016/j.jpedsurg.2004.01.033 672
between groups, with P values of less than .05 considered significant. Results: The querie generated a total of 2,471 patients: 2,379 (96.3%) VVDL, and 92 (3.7%) VVDL ⫹ V. The groups were comparable with the only significant differences being a higher mean airway pressure at 24 hours of bypass and a more frequent use of inotropes during extracorporeal life support (ECLS) in the VVDL ⫹ V group. Comparison of a CDH patient subset (280 from the VVDL group and 25 from the VVDL ⫹ V group) showed the following significant differences: more frequent use of inotropes, higher MAP at 24 hours, and higher mean flow rates at 4 and 24 hours, all in the VVDL ⫹ V group. Patient outcomes, including survival, complications, and rates of conversion to VA bypass were comparable between like groups. Conclusions: The theoretical benefits of routine placement of a cephalad jugular cannula during VV-ECMO via a jugular double lumen cannula are not substantiated by critical analysis of ELSO data. J Pediatr Surg 39:672-676. © 2004 Elsevier Inc. All rights reserved. INDEX WORDS: Extracorporeal membrane oxygenation, cephalad cannula, internal jugular vein, recirculation, intracranial hemorrhage.
return and arterial inflow. Reported single-center and registry experience with this form of veno-venous double lumen (VVDL)-ECMO suggests that it provides extracorporeal membrane gas exchange that is equivalent to that achieved with venoarterial (VA)-ECMO, with preservation of the common carotid artery that is either ligated or reconstructed after VA-ECMO.1-3 Among a number of potential complications associated with either VVDL or VA-ECMO, intracranial hemorrhage or infarction during bypass is perhaps the most devastating. Data from the Extracorporeal Life Support Organization (ELSO) “neonatal respiratory” registry suggests that the complication rate of significant ischemic or hemorrhagic central nervous system (CNS) injury during ECMO is approximately 15%.1 The etiology of these injuries is undoubtedly multifactorial, with abrupt changes in local and systemic blood pressure, ischemia/reperfusion, systemic anticoagulation, and venous hypertension caused by distal internal jugular vein Journal of Pediatric Surgery, Vol 39, No 5 (May), 2004: pp 672-676
VV ECMO WITH CEPHALAD JUGULAR CANNULA
673
ligation all purported to play contributory roles. The hypothesized causal relationship between cerebral venous hypertension and neurologic injury has led some to advocate routine placement of a drainage catheter in the cephalad internal jugular vein at the time of cannulation. Other potential benefits to routine cephalad jugular drainage include augmented venous return and reduced atrial circulation of arterialized blood during VVDL-ECMO. Neither experimental nor clinical data clearly substantiate a benefit to routine placement of a cephalad jugular cannula during VVDL or VA-ECMO. Moreover, the presence of a second (or third, in the case of VA-ECMO) cannula and its connection into the vascular circuit may increase the risk of catheter-associated complications, including catheter or circuit thrombosis, air embolism, infection, or catheter dislodgement. For these reasons, we obtained ELSO registry data to see whether routine cephalad jugular cannula drainage was associated with improved outcome in patients undergoing VVDLECMO. MATERIALS AND METHODS With institutional review board (IRB) approval, ELSO registry data for all neonates receiving ECMO for the treatment of respiratory failure between January 1, 1989 and December 31, 2001 was obtained. Using the “mode of bypass” data field as a sorting tool, we identified 2 patient groups: those undergoing venovenous (double lumen cannula) ECMO without a cephalad jugular drainage cannula (VVDL), and those patients undergoing venovenous ECMO with a cephalad jugular drainage cannula (VVDL ⫹ V). As confirmation of sorting accuracy and to identify patients from both groups requiring conversion to VA ECMO, (which are not otherwise specifically identified by any data field), we cross referenced the VVDL and VVDL ⫹ V groups to all possible combinations of conforming cannulae (using manufacturer’s cannula identifiers) to generate the following 4 patient groups: (1) VVDL (not converted): 1 double-lumen cannula, (2) VVDL converted to VA: 1 double-lumen cannula and 1 single-lumen cannula, (3) VVDL ⫹ V (not converted): 1 double lumen cannula and 1 single lumen cannula, and (4) VVDL ⫹ V (converted): 1 double-lumen cannula and 2 singlelumen cannulae. Only those patients with a “mode of bypass” field that was reconcilable with the cannula or combination of cannulae used, were included in the analysis. Patients in the VVDL and VVDL ⫹ V groups were characterized according to gestational age at birth, age, weight, and diagnosis at cannulation; these variables allowed group comparison. The following outcomes were analyzed: survival to hospital discharge, hours on bypass, mean flow rate at 4 and 24 hours of bypass, mean airway
Fig 1. Distribution (by primary diagnosis) of ELSO registry neonatal patients treated by VVDL (n ⴝ 2,379) or VVDL ⴙ V (n ⴝ 92) venovenous ECMO (January 1, 1990 to December 31, 2001). PPHN, primary pulmonary hypertension; CDH, congenital diaphragmatic hernia; MecAsp, meconium aspiration; PTX, pneumothorax; PNA, pneumonia.
pressure at cannulation and at 24 hours of bypass, need for conversion to VA-ECMO and complications. Complications included those related to the cannulae (mechanical failure, thrombosis, embolism, dislodgement, and cannula site hemorrhage), neurologic problems (central nervous system hemorrhage or infarction, seizures— clinically or EEG determined, or brain death), cardiovascular support required on ECMO, and systemic infections on ECMO. A subgroup of patients with a diagnosis of congenital diaphragmatic hernia (International Classification of Diseases, Revision 9[ICD-9] code 756.6), were individually analyzed by the method described above. Means ⫹ SD were calculated for all groups and were the basis for comparison, with Student’s t tests used to determine significant differences between groups (P ⬍ .05).
RESULTS
From nearly 15,000 neonatal respiratory cannulations recorded in the database over the period of study, our querie generated 2,471 patients: 2,379 (96%) who were treated with VVDL-ECMO and 92 (3.7%) treated with VVDL ⫹ V ECMO. The distribution of patients by primary diagnosis is shown in Fig 1. The most common diagnosis in both groups was primary pulmonary hypertension of the newborn (PPHN), which accounted for 70% of VVDL patients and 59% of VVDL ⫹ V patients. Both groups were comparable from the perspective of
Table 1. ELSO Registry Patients Receiving Neonatal ECMO Via Either VVDL or VVDL ⴙ V Techniques
VVDL VVDL ⫹ V
No.
GA (wk)
Age (d)
Weight (kg)
Hours on Bypass
Q4h (mL/min/kg)
Q24h (mL/min/kg)
Vent MAP (t0) (mm Hg)
Vent MAP (t24) (mm Hg)*
VV to VA
Survival to DC?
2379 92
39.5 ⫾ 1.8 39.4 ⫾ 1.9
6.45 ⫾ 33 10.8 ⫾ 42
3.45 ⫾ 0.5 3.40 ⫾ 0.5
137.4 ⫾ 95 153.6 ⫾ 95
109 ⫾ 45 112 ⫾ 54
102 ⫾ 44 102 ⫾ 45
19.2 ⫾ 1.8 19.7 ⫾ 2.2
10.4 ⫾ 1.7 12.7 ⫾ 1.2
174 (7.3%) 8 (8.7%)
2067 (87%) 75 (82.4%)
*Significant difference between groups (P ⬍ .05).
674
SKARSGARD ET AL Table 2. ELSO Registry CDH Patients Receiving Neonatal ECMO via Either VVDL or VVDL ⴙ V Techniques
VVDL VVDL ⫹ V
No.
GA (wk)
Age (d)
Weight (kg)
Hours on Bypass
280 25
39.0 ⫾ 1.7 38.5 ⫾ 1.7
4.5 ⫾ 11 4.5 ⫾ 10
3.15 ⫾ 0.6 3.10 ⫾ 0.5
226 ⫾ 155 225 ⫾ 114
Q4h (mL/min/kg)*
Q24h (mL/min/kg)*
Vent MAP (t0) (mm Hg)
Vent MAP (t24) (mm Hg)*
VV to VA
Survival to DC?
107 ⫾ 48 176 ⫾ 99
104 ⫾ 43 171 ⫾ 107
17.2 ⫾ 1.8 16.0 ⫾ 2.2
10.6 ⫾ 1.7 12.8 ⫾ 2.2
21 (7.5%) 3 (12%)
201 (72%) 16 (64%)
*Significant difference between groups (P ⬍ .05).
gestational age (GA) at birth, gender, age at cannulation, weight, hours on bypass, mean airway pressure (MAP) at cannulation, flow rates at 4 and 24 hours, conversion rates to venoarterial (VA) ECMO, and survival to discharge (Table 1). The only significant difference between the groups was an observed higher MAP at 24 hours of bypass in the VVDL ⫹ V group. A total of 305 patients with a diagnosis of CDH were identified; 280 (91.8%) were treated by VVDL-ECMO, and 25 (8.2%) were treated by VVDL ⫹ V-ECMO. The patient characteristics (summarized in Table 2) included 2 significant differences between groups: CDH patients treated with VVDL ⫹ V-ECMO had a higher observed MAP after 24 hours of bypass, as well as higher flow rates per kilogram at 4 and 24 hours of bypass. VV to VA conversions occurred more frequently in CDH patients treated with VVDL ⫹ V-ECMO; however, this result did not achieve statistical significance. We reviewed complications data for a total of 13 specific complications (from a total of 48), which are summarized in Table 3, for all patients and the CDH patient subgroup. We selected those complications that were most likely to illustrate the hypothetical advantages or disadvantages of one technique compared with the other. The cephalad jugular cannula offers the potential benefits of cerebral venous decompression (neurologic injury advantages) and augmented, oxygen-desaturated venous return (mechanical flow, reduced atrial recirculation, improved organ perfusion advantages). Its princi-
pal disadvantage is the need for an additional cannula (cannula-related disadvantages), and venous inflow that is out of series with the rest of the vascular circuit (circuit-related disadvantages). We found no significant difference between the frequency of complications between like groups, although we did observe a significantly more frequent use of inotropes during ECLS for patients receiving VVDL ⫹ V-ECMO in both the “all patient” and “CDH” groups. DISCUSSION
Since its first successful application to a newborn with respiratory failure in 1975, ECMO has been used in over 13,000 neonates, with an overall survival rate of 86%.1 Although not itself a direct therapy, ECMO enables oxygenation and ventilation that is either not achievable or only achievable at the extreme cost of iatrogenic lung injury or perfusion impairment while the pathophysiologic processes that were the cause of cardiorespiratory failure are given time to reverse. Most neonates (and infants weighing less than 5 kg), with reversible cardiorespiratory failure can be supported with a double lumen cannula placed through the internal jugular vein into the right atrium (VVDL-ECMO). In contrast to venoarterial (VA)-ECMO, in which arterialized blood is pumped directly into the arterial circulation via the right common carotid artery, VVDL-ECMO does not provide “pump” support and is unavoidably associated with recirculation of a fraction of infused arterial
Table 3. Rates of Complication: VVDL Versus VVDL ⴙ V All Patients
CDH Subgroup
Complication
VVDL (n ⫽ 2,471)
VVDL ⫹ V (n ⫽ 92)
VVDL (n ⫽ 280)
VVDL ⫹ V (n ⫽ 25)
Mechanical: air in circuit Mechanical: cannula “problems” Mechanical: cannula site hemorrhage Neurologic: brain death Neurologic: clinical seizures Neurologic: EEG seizures Neurologic: CNS infarction by US/CT Neurologic: CNS hemorrhage by US/CT Perfusion: creatinine ⬎ 3.0 Perfusion: inotropes on ECLS* Pulmonary: pneumothorax req treatment Pulmonary: pulmonary hemorrhage Infection: culture proven
154 (6.5%) 312 (13.1%) 98 (4.1%) 10 (0.4%) 158 (6.6%) 8 (0.3%) 127 (5.3%) 104 (4.4%) 21 (0.9%) 208 (8.7%) 106 (4.4%) 84 (3.5%) 116 (4.9%)
3 (3.3%) 12 (13.0%) 2 (2.1%) 0 (0%) 3 (9.8%) 1 (1.0%) 2 (2.1%) 4 (4.3%) 1 (1.1%) 28 (30.4%) 2 (2.2%) 2 (2.2%) 1 (1.1%)
20 (7.1%) 46 (16.4%) 20 (7.1%) 3 (1.1%) 21 (7.5%) 2 (0.7%) 13 (4.6%) 16 (5.7%) 1 (0.4%) 48 (17%) 24 (8.6%) 24 (8.6%) 24 (8.6%)
1 (4%) 6 (24%) 0 (0%) 0 (0%) 1 (4%) 0 (0%) 1 (4%) 1 (4%) 1 (4%) 9 (36%) 0 (0%) 1 (4%) 1 (4%)
*Significant difference between groups (P ⬍ .05).
VV ECMO WITH CEPHALAD JUGULAR CANNULA
blood back into the extracorporeal circuit. Nonetheless, the advantages of carotid artery preservation, improved pulmonary and coronary perfusion during bypass, and the reduced risk of circuit embolization into the systemic circulation make VVDL-ECMO the preferred technique of extracorporeal support for infants with cardiorespiratory failure, provided myocardial function is adequate. Some ECMO centers advocate routine placement of an 8F or 10F single-lumen cannula into the cephalad portion of the ligated right internal jugular vein. By means of a Y connector, the cephalad jugular catheter is connected (in parallel) to venous drainage from the double-lumen cannula and therefore provides extraatrial, desaturated venous drainage into the ECMO circuit. In addition to augmentation of venous return and reduced atrial recirculation, the placement of a drainage catheter into the cephalad jugular vein may result in reduction of jugular venous hypertension and reduced risk of intracranial hemorrhage (ICH) on ECMO. O’Connor et al4 reported that in comparison with an historical cohort without a cephalad jugular catheter, the routine use of a cephalad jugular catheter in a mixture of VA- and VVDL-ECMO patients resulted in a reduction of the rate of ICH from 35% to 6%, which they attributed to cerebral venous decompression. Kitagawa et al5 studied the effects of cephalad cannula clamping and unclamping on sagittal sinus and intracranial pressure in primates on VA bypass, and found no significant pressure difference, but did observe an augmentation of venous return when the cephalad cannula was unclamped. Weber and Kountzman6 studied 6 newborns on VA-ECMO with cephalad jugular cannulae and found that cannula closure resulted in an immediate reduction in blood flow velocity in the right middle cerebral artery (measured by duplex scanning) that was restored immediately when the cannula was opened or, if the cannula remained occluded, was restored between 8 and 24 hours after institution of bypass. Closure of the cephalad jugular cannula had no effect on left middle cerebral artery flow velocity at any time.6 In spite of anecdotal clinical evidence, which suggests that cephalad jugular venous drainage might confer neurologic protection to patients on ECMO, our review of ELSO data on patients treated by VVDL-ECMO with and without a cephalad jugular drainage cannula does not substantiate this appealing hypothesis. We did note that although comparable at cannulation, the mean airway pressure recorded at 24 hours of bypass was significantly higher in VVDL ⫹ V-ECMO patients (both all patients and CDH patients). We know that compared with VAECMO, VV-ECMO represents “partial” bypass and that pulmonary gas exchange plays a variable, but often essential, supportive role, depending on the quality of bypass and the infant’s needs. One might hypothesize
675
that if comparable tissue perfusion and gas exchanged were accomplished with lower ventilatory pressures in the VVDL group, then this group experienced a superior extracorporeal contribution to overall gas exchange compared with the VVDL ⫹ V group. However, an equally plausible explanation is that the pressure differences are caused by fundamental ventilatory practice differences that exist between treating critical care units. Similarly, the observation that VVDL ⫹ V patients were significantly more likely to have received inotropes while on ECLS does not necessarily imply perfusion disadvantages to VVDL ⫹ V bypass, because mean flow rates per kilogram at 4 and 24 hours and VV to VA conversion rates were comparable in the 2 groups. Rather, it may just reflect clinical practice differences. Certainly, the fact that the 2 groups are very disparate in size makes it more likely that subtle practice differences between groups are likely to be exaggerated; this is particularly true when the larger group represents many more contributing centers and, therefore, a much greater diversity of practice than the smaller group. We were especially interested in whether the registry data might show bypass advantages conferred by improved venous return via cephalad jugular drainage in the subgroup of patients with a diagnosis of CDH, who many feel are especially difficult to support with VVDLECMO because of anatomic deformation of the mediastinum. Although the CDH patients treated with VVDL ⫹ V had significantly higher flow rates per kilogram at 4 and 24 hours after institution of bypass, the comparable and relatively low rate of conversion to VA-ECMO suggests that both techniques enable adequate bypass. As was the case with the parent groups, there was a significantly higher observed MAP at 24 hours and a higher incidence of inotrope use during bypass among the CDH subset receiving VVDL-ECMO. Data from the ELSO registry comparing VVDLECMO patients managed with and without a cephalad jugular drainage cannula does not show a clear benefit to either technique from the perspectives of patient outcome and technical ease. Complication rates appear to be comparable, although the disparity in numbers of patients in the 2 groups may make it difficult to detect subtle differences. The observation of significantly higher mean bypass flow rates in the VVDL ⫹ V CDH patient subset may reflect a venous drainage advantage in a group of patients in whom adequate cannula position may be especially difficult to maintain because of atrial deformation and mediastinal shift. Perhaps special consideration to the use of a cephalad drainage catheter should be given to these patients. Ultimately, the technique of ECMO used should be determined by individual centers based on their own experience and preference.
676
SKARSGARD ET AL
REFERENCES 1. ECLS Registry Report: International Summary. July, 2003 2. Osiovich HC, Peliowski A, Ainsworth W, et al: The Edmonton experience with venovenous extracorporeal membrane oxygenation. J Pediatr Surg 33:1749-1752, 1998 3. Heiss KF, Clark RH, Cornish JD, et al: Preferential use of venovenous extracorporeal membrane oxygenation for congenital diaphragmatic hernia. J Pediatr Surg 30:416-419, 1995 4. O’Connor TA, Haney BM, Grist GE, et al: Decreased incidence
of intracranial hemorrhage using cephalic jugular venous drainage during neonatal extracorporeal membrane oxygenation. J Pediatr Surg 28:1332-1335, 1993 5. Kitagawa H, McComb G, Ford E, et al: Proximal and distal cannulation of the internal jugular vein for ECMO in a primate. J Pediatr Surg 27:1189-1191, 1992 6. Weber TR, Kountzman B: The effects of venous occlusion on cerebral blood flow during ECMO. J Pediatr Surg 31:1124-1127, 1996
Background/Purpose: Extracorporeal membrane oxygenation (ECMO), may be life saving for infants with severe respiratory failure, and when possible, veno-venous bypass through a jugular double lumen cannula, can be expected to provide satisfactory support for most patients. Some ECMO centers favor routine placement of a cephalad jugular cannula for the theoretical benefits of augmented (desaturated) venous return, reduction of atrial recirculation, and cerebral venous decompression. The purpose of this study was to querie the ELSO registry for patients who had undergone VV-ECMO and compare outcomes for patients with a double lumen cannula only (VVDL), with those who had both a double lumen and cephalad jugular cannula (VVDL ⫹ V). Methods: With institutional review board (IRB) approval, the Extracorporeal Life Support Organization (ELSO) registry (Ann Arbor, MI) was queried from January 1, 1989 to December 31, 2001, and all “neonatal respiratory” patients undergoing VV-ECMO via either the VVDL or VVDL ⫹ V modes were identified. Group comparisons by age, diagnosis, hours on bypass, mean flow rates (Q) at 4 and 24 hours, mean airway pressures (MAP) at initiation and at 24 hours of bypass, complications (including neurologic and cannulaspecific), need for conversion to veno-arterial (VA) ECMO, and survival were performed. A similar analysis was performed on a congenital diaphragmatic hernia (CDH) patient subgroup. Student’s t tests were used to compare means
T
HE USE of extracorporeal membrane oxygenation (ECMO) in the support of neonates with profound respiratory failure associated with a variety of conditions is well established. When hemodynamically feasible, the preferred technique of bypass involves use of a double lumen cannula placed through the right internal jugular vein into the right atrium, which provides both venous From the Departments of Surgery, Pediatrics, and Perfusion Services, British Columbia Children’s Hospital; the Centre for Health Innovation and Improvement; and the University of British Columbia, Vancouver, British Columbia, Canada. Presented at the 35th Annual Meeting of the Canadian Association of Paediatric Surgeons, Niagara-on-the-Lake, Ontario, Canada, September 18-21, 2003. Address reprint requests to Erik D. Skarsgard, MD, B.C.’s Children’s Hospital, KO-123, 4480 Oak St, Vancouver, B.C. V6H 3V4, Canada. © 2004 Elsevier Inc. All rights reserved. 0022-3468/04/3905-0004$30.00/0 doi:10.1016/j.jpedsurg.2004.01.033 672
between groups, with P values of less than .05 considered significant. Results: The querie generated a total of 2,471 patients: 2,379 (96.3%) VVDL, and 92 (3.7%) VVDL ⫹ V. The groups were comparable with the only significant differences being a higher mean airway pressure at 24 hours of bypass and a more frequent use of inotropes during extracorporeal life support (ECLS) in the VVDL ⫹ V group. Comparison of a CDH patient subset (280 from the VVDL group and 25 from the VVDL ⫹ V group) showed the following significant differences: more frequent use of inotropes, higher MAP at 24 hours, and higher mean flow rates at 4 and 24 hours, all in the VVDL ⫹ V group. Patient outcomes, including survival, complications, and rates of conversion to VA bypass were comparable between like groups. Conclusions: The theoretical benefits of routine placement of a cephalad jugular cannula during VV-ECMO via a jugular double lumen cannula are not substantiated by critical analysis of ELSO data. J Pediatr Surg 39:672-676. © 2004 Elsevier Inc. All rights reserved. INDEX WORDS: Extracorporeal membrane oxygenation, cephalad cannula, internal jugular vein, recirculation, intracranial hemorrhage.
return and arterial inflow. Reported single-center and registry experience with this form of veno-venous double lumen (VVDL)-ECMO suggests that it provides extracorporeal membrane gas exchange that is equivalent to that achieved with venoarterial (VA)-ECMO, with preservation of the common carotid artery that is either ligated or reconstructed after VA-ECMO.1-3 Among a number of potential complications associated with either VVDL or VA-ECMO, intracranial hemorrhage or infarction during bypass is perhaps the most devastating. Data from the Extracorporeal Life Support Organization (ELSO) “neonatal respiratory” registry suggests that the complication rate of significant ischemic or hemorrhagic central nervous system (CNS) injury during ECMO is approximately 15%.1 The etiology of these injuries is undoubtedly multifactorial, with abrupt changes in local and systemic blood pressure, ischemia/reperfusion, systemic anticoagulation, and venous hypertension caused by distal internal jugular vein Journal of Pediatric Surgery, Vol 39, No 5 (May), 2004: pp 672-676
VV ECMO WITH CEPHALAD JUGULAR CANNULA
673
ligation all purported to play contributory roles. The hypothesized causal relationship between cerebral venous hypertension and neurologic injury has led some to advocate routine placement of a drainage catheter in the cephalad internal jugular vein at the time of cannulation. Other potential benefits to routine cephalad jugular drainage include augmented venous return and reduced atrial circulation of arterialized blood during VVDL-ECMO. Neither experimental nor clinical data clearly substantiate a benefit to routine placement of a cephalad jugular cannula during VVDL or VA-ECMO. Moreover, the presence of a second (or third, in the case of VA-ECMO) cannula and its connection into the vascular circuit may increase the risk of catheter-associated complications, including catheter or circuit thrombosis, air embolism, infection, or catheter dislodgement. For these reasons, we obtained ELSO registry data to see whether routine cephalad jugular cannula drainage was associated with improved outcome in patients undergoing VVDLECMO. MATERIALS AND METHODS With institutional review board (IRB) approval, ELSO registry data for all neonates receiving ECMO for the treatment of respiratory failure between January 1, 1989 and December 31, 2001 was obtained. Using the “mode of bypass” data field as a sorting tool, we identified 2 patient groups: those undergoing venovenous (double lumen cannula) ECMO without a cephalad jugular drainage cannula (VVDL), and those patients undergoing venovenous ECMO with a cephalad jugular drainage cannula (VVDL ⫹ V). As confirmation of sorting accuracy and to identify patients from both groups requiring conversion to VA ECMO, (which are not otherwise specifically identified by any data field), we cross referenced the VVDL and VVDL ⫹ V groups to all possible combinations of conforming cannulae (using manufacturer’s cannula identifiers) to generate the following 4 patient groups: (1) VVDL (not converted): 1 double-lumen cannula, (2) VVDL converted to VA: 1 double-lumen cannula and 1 single-lumen cannula, (3) VVDL ⫹ V (not converted): 1 double lumen cannula and 1 single lumen cannula, and (4) VVDL ⫹ V (converted): 1 double-lumen cannula and 2 singlelumen cannulae. Only those patients with a “mode of bypass” field that was reconcilable with the cannula or combination of cannulae used, were included in the analysis. Patients in the VVDL and VVDL ⫹ V groups were characterized according to gestational age at birth, age, weight, and diagnosis at cannulation; these variables allowed group comparison. The following outcomes were analyzed: survival to hospital discharge, hours on bypass, mean flow rate at 4 and 24 hours of bypass, mean airway
Fig 1. Distribution (by primary diagnosis) of ELSO registry neonatal patients treated by VVDL (n ⴝ 2,379) or VVDL ⴙ V (n ⴝ 92) venovenous ECMO (January 1, 1990 to December 31, 2001). PPHN, primary pulmonary hypertension; CDH, congenital diaphragmatic hernia; MecAsp, meconium aspiration; PTX, pneumothorax; PNA, pneumonia.
pressure at cannulation and at 24 hours of bypass, need for conversion to VA-ECMO and complications. Complications included those related to the cannulae (mechanical failure, thrombosis, embolism, dislodgement, and cannula site hemorrhage), neurologic problems (central nervous system hemorrhage or infarction, seizures— clinically or EEG determined, or brain death), cardiovascular support required on ECMO, and systemic infections on ECMO. A subgroup of patients with a diagnosis of congenital diaphragmatic hernia (International Classification of Diseases, Revision 9[ICD-9] code 756.6), were individually analyzed by the method described above. Means ⫹ SD were calculated for all groups and were the basis for comparison, with Student’s t tests used to determine significant differences between groups (P ⬍ .05).
RESULTS
From nearly 15,000 neonatal respiratory cannulations recorded in the database over the period of study, our querie generated 2,471 patients: 2,379 (96%) who were treated with VVDL-ECMO and 92 (3.7%) treated with VVDL ⫹ V ECMO. The distribution of patients by primary diagnosis is shown in Fig 1. The most common diagnosis in both groups was primary pulmonary hypertension of the newborn (PPHN), which accounted for 70% of VVDL patients and 59% of VVDL ⫹ V patients. Both groups were comparable from the perspective of
Table 1. ELSO Registry Patients Receiving Neonatal ECMO Via Either VVDL or VVDL ⴙ V Techniques
VVDL VVDL ⫹ V
No.
GA (wk)
Age (d)
Weight (kg)
Hours on Bypass
Q4h (mL/min/kg)
Q24h (mL/min/kg)
Vent MAP (t0) (mm Hg)
Vent MAP (t24) (mm Hg)*
VV to VA
Survival to DC?
2379 92
39.5 ⫾ 1.8 39.4 ⫾ 1.9
6.45 ⫾ 33 10.8 ⫾ 42
3.45 ⫾ 0.5 3.40 ⫾ 0.5
137.4 ⫾ 95 153.6 ⫾ 95
109 ⫾ 45 112 ⫾ 54
102 ⫾ 44 102 ⫾ 45
19.2 ⫾ 1.8 19.7 ⫾ 2.2
10.4 ⫾ 1.7 12.7 ⫾ 1.2
174 (7.3%) 8 (8.7%)
2067 (87%) 75 (82.4%)
*Significant difference between groups (P ⬍ .05).
674
SKARSGARD ET AL Table 2. ELSO Registry CDH Patients Receiving Neonatal ECMO via Either VVDL or VVDL ⴙ V Techniques
VVDL VVDL ⫹ V
No.
GA (wk)
Age (d)
Weight (kg)
Hours on Bypass
280 25
39.0 ⫾ 1.7 38.5 ⫾ 1.7
4.5 ⫾ 11 4.5 ⫾ 10
3.15 ⫾ 0.6 3.10 ⫾ 0.5
226 ⫾ 155 225 ⫾ 114
Q4h (mL/min/kg)*
Q24h (mL/min/kg)*
Vent MAP (t0) (mm Hg)
Vent MAP (t24) (mm Hg)*
VV to VA
Survival to DC?
107 ⫾ 48 176 ⫾ 99
104 ⫾ 43 171 ⫾ 107
17.2 ⫾ 1.8 16.0 ⫾ 2.2
10.6 ⫾ 1.7 12.8 ⫾ 2.2
21 (7.5%) 3 (12%)
201 (72%) 16 (64%)
*Significant difference between groups (P ⬍ .05).
gestational age (GA) at birth, gender, age at cannulation, weight, hours on bypass, mean airway pressure (MAP) at cannulation, flow rates at 4 and 24 hours, conversion rates to venoarterial (VA) ECMO, and survival to discharge (Table 1). The only significant difference between the groups was an observed higher MAP at 24 hours of bypass in the VVDL ⫹ V group. A total of 305 patients with a diagnosis of CDH were identified; 280 (91.8%) were treated by VVDL-ECMO, and 25 (8.2%) were treated by VVDL ⫹ V-ECMO. The patient characteristics (summarized in Table 2) included 2 significant differences between groups: CDH patients treated with VVDL ⫹ V-ECMO had a higher observed MAP after 24 hours of bypass, as well as higher flow rates per kilogram at 4 and 24 hours of bypass. VV to VA conversions occurred more frequently in CDH patients treated with VVDL ⫹ V-ECMO; however, this result did not achieve statistical significance. We reviewed complications data for a total of 13 specific complications (from a total of 48), which are summarized in Table 3, for all patients and the CDH patient subgroup. We selected those complications that were most likely to illustrate the hypothetical advantages or disadvantages of one technique compared with the other. The cephalad jugular cannula offers the potential benefits of cerebral venous decompression (neurologic injury advantages) and augmented, oxygen-desaturated venous return (mechanical flow, reduced atrial recirculation, improved organ perfusion advantages). Its princi-
pal disadvantage is the need for an additional cannula (cannula-related disadvantages), and venous inflow that is out of series with the rest of the vascular circuit (circuit-related disadvantages). We found no significant difference between the frequency of complications between like groups, although we did observe a significantly more frequent use of inotropes during ECLS for patients receiving VVDL ⫹ V-ECMO in both the “all patient” and “CDH” groups. DISCUSSION
Since its first successful application to a newborn with respiratory failure in 1975, ECMO has been used in over 13,000 neonates, with an overall survival rate of 86%.1 Although not itself a direct therapy, ECMO enables oxygenation and ventilation that is either not achievable or only achievable at the extreme cost of iatrogenic lung injury or perfusion impairment while the pathophysiologic processes that were the cause of cardiorespiratory failure are given time to reverse. Most neonates (and infants weighing less than 5 kg), with reversible cardiorespiratory failure can be supported with a double lumen cannula placed through the internal jugular vein into the right atrium (VVDL-ECMO). In contrast to venoarterial (VA)-ECMO, in which arterialized blood is pumped directly into the arterial circulation via the right common carotid artery, VVDL-ECMO does not provide “pump” support and is unavoidably associated with recirculation of a fraction of infused arterial
Table 3. Rates of Complication: VVDL Versus VVDL ⴙ V All Patients
CDH Subgroup
Complication
VVDL (n ⫽ 2,471)
VVDL ⫹ V (n ⫽ 92)
VVDL (n ⫽ 280)
VVDL ⫹ V (n ⫽ 25)
Mechanical: air in circuit Mechanical: cannula “problems” Mechanical: cannula site hemorrhage Neurologic: brain death Neurologic: clinical seizures Neurologic: EEG seizures Neurologic: CNS infarction by US/CT Neurologic: CNS hemorrhage by US/CT Perfusion: creatinine ⬎ 3.0 Perfusion: inotropes on ECLS* Pulmonary: pneumothorax req treatment Pulmonary: pulmonary hemorrhage Infection: culture proven
154 (6.5%) 312 (13.1%) 98 (4.1%) 10 (0.4%) 158 (6.6%) 8 (0.3%) 127 (5.3%) 104 (4.4%) 21 (0.9%) 208 (8.7%) 106 (4.4%) 84 (3.5%) 116 (4.9%)
3 (3.3%) 12 (13.0%) 2 (2.1%) 0 (0%) 3 (9.8%) 1 (1.0%) 2 (2.1%) 4 (4.3%) 1 (1.1%) 28 (30.4%) 2 (2.2%) 2 (2.2%) 1 (1.1%)
20 (7.1%) 46 (16.4%) 20 (7.1%) 3 (1.1%) 21 (7.5%) 2 (0.7%) 13 (4.6%) 16 (5.7%) 1 (0.4%) 48 (17%) 24 (8.6%) 24 (8.6%) 24 (8.6%)
1 (4%) 6 (24%) 0 (0%) 0 (0%) 1 (4%) 0 (0%) 1 (4%) 1 (4%) 1 (4%) 9 (36%) 0 (0%) 1 (4%) 1 (4%)
*Significant difference between groups (P ⬍ .05).
VV ECMO WITH CEPHALAD JUGULAR CANNULA
blood back into the extracorporeal circuit. Nonetheless, the advantages of carotid artery preservation, improved pulmonary and coronary perfusion during bypass, and the reduced risk of circuit embolization into the systemic circulation make VVDL-ECMO the preferred technique of extracorporeal support for infants with cardiorespiratory failure, provided myocardial function is adequate. Some ECMO centers advocate routine placement of an 8F or 10F single-lumen cannula into the cephalad portion of the ligated right internal jugular vein. By means of a Y connector, the cephalad jugular catheter is connected (in parallel) to venous drainage from the double-lumen cannula and therefore provides extraatrial, desaturated venous drainage into the ECMO circuit. In addition to augmentation of venous return and reduced atrial recirculation, the placement of a drainage catheter into the cephalad jugular vein may result in reduction of jugular venous hypertension and reduced risk of intracranial hemorrhage (ICH) on ECMO. O’Connor et al4 reported that in comparison with an historical cohort without a cephalad jugular catheter, the routine use of a cephalad jugular catheter in a mixture of VA- and VVDL-ECMO patients resulted in a reduction of the rate of ICH from 35% to 6%, which they attributed to cerebral venous decompression. Kitagawa et al5 studied the effects of cephalad cannula clamping and unclamping on sagittal sinus and intracranial pressure in primates on VA bypass, and found no significant pressure difference, but did observe an augmentation of venous return when the cephalad cannula was unclamped. Weber and Kountzman6 studied 6 newborns on VA-ECMO with cephalad jugular cannulae and found that cannula closure resulted in an immediate reduction in blood flow velocity in the right middle cerebral artery (measured by duplex scanning) that was restored immediately when the cannula was opened or, if the cannula remained occluded, was restored between 8 and 24 hours after institution of bypass. Closure of the cephalad jugular cannula had no effect on left middle cerebral artery flow velocity at any time.6 In spite of anecdotal clinical evidence, which suggests that cephalad jugular venous drainage might confer neurologic protection to patients on ECMO, our review of ELSO data on patients treated by VVDL-ECMO with and without a cephalad jugular drainage cannula does not substantiate this appealing hypothesis. We did note that although comparable at cannulation, the mean airway pressure recorded at 24 hours of bypass was significantly higher in VVDL ⫹ V-ECMO patients (both all patients and CDH patients). We know that compared with VAECMO, VV-ECMO represents “partial” bypass and that pulmonary gas exchange plays a variable, but often essential, supportive role, depending on the quality of bypass and the infant’s needs. One might hypothesize
675
that if comparable tissue perfusion and gas exchanged were accomplished with lower ventilatory pressures in the VVDL group, then this group experienced a superior extracorporeal contribution to overall gas exchange compared with the VVDL ⫹ V group. However, an equally plausible explanation is that the pressure differences are caused by fundamental ventilatory practice differences that exist between treating critical care units. Similarly, the observation that VVDL ⫹ V patients were significantly more likely to have received inotropes while on ECLS does not necessarily imply perfusion disadvantages to VVDL ⫹ V bypass, because mean flow rates per kilogram at 4 and 24 hours and VV to VA conversion rates were comparable in the 2 groups. Rather, it may just reflect clinical practice differences. Certainly, the fact that the 2 groups are very disparate in size makes it more likely that subtle practice differences between groups are likely to be exaggerated; this is particularly true when the larger group represents many more contributing centers and, therefore, a much greater diversity of practice than the smaller group. We were especially interested in whether the registry data might show bypass advantages conferred by improved venous return via cephalad jugular drainage in the subgroup of patients with a diagnosis of CDH, who many feel are especially difficult to support with VVDLECMO because of anatomic deformation of the mediastinum. Although the CDH patients treated with VVDL ⫹ V had significantly higher flow rates per kilogram at 4 and 24 hours after institution of bypass, the comparable and relatively low rate of conversion to VA-ECMO suggests that both techniques enable adequate bypass. As was the case with the parent groups, there was a significantly higher observed MAP at 24 hours and a higher incidence of inotrope use during bypass among the CDH subset receiving VVDL-ECMO. Data from the ELSO registry comparing VVDLECMO patients managed with and without a cephalad jugular drainage cannula does not show a clear benefit to either technique from the perspectives of patient outcome and technical ease. Complication rates appear to be comparable, although the disparity in numbers of patients in the 2 groups may make it difficult to detect subtle differences. The observation of significantly higher mean bypass flow rates in the VVDL ⫹ V CDH patient subset may reflect a venous drainage advantage in a group of patients in whom adequate cannula position may be especially difficult to maintain because of atrial deformation and mediastinal shift. Perhaps special consideration to the use of a cephalad drainage catheter should be given to these patients. Ultimately, the technique of ECMO used should be determined by individual centers based on their own experience and preference.
676
SKARSGARD ET AL
REFERENCES 1. ECLS Registry Report: International Summary. July, 2003 2. Osiovich HC, Peliowski A, Ainsworth W, et al: The Edmonton experience with venovenous extracorporeal membrane oxygenation. J Pediatr Surg 33:1749-1752, 1998 3. Heiss KF, Clark RH, Cornish JD, et al: Preferential use of venovenous extracorporeal membrane oxygenation for congenital diaphragmatic hernia. J Pediatr Surg 30:416-419, 1995 4. O’Connor TA, Haney BM, Grist GE, et al: Decreased incidence
of intracranial hemorrhage using cephalic jugular venous drainage during neonatal extracorporeal membrane oxygenation. J Pediatr Surg 28:1332-1335, 1993 5. Kitagawa H, McComb G, Ford E, et al: Proximal and distal cannulation of the internal jugular vein for ECMO in a primate. J Pediatr Surg 27:1189-1191, 1992 6. Weber TR, Kountzman B: The effects of venous occlusion on cerebral blood flow during ECMO. J Pediatr Surg 31:1124-1127, 1996