Article Text

Download PDFPDF
Role of ECMO in congenital diaphragmatic hernia
  1. Merrill McHoney,
  2. Philip Hammond
  1. Paediatric Surgery, Royal Hospital for Sick Children Edinburgh, Edinburgh, UK
  1. Correspondence to Merrill McHoney, Royal Hospital for Sick Children Edinburgh, University of Edinburgh, Edinburgh EH1 1LF, UK; merrillmchoney{at}


Congenital diaphragmatic hernia (CDH) is typified morphologically by failure of diaphragmatic development with accompanying lung hypoplasia and persistent pulmonary hypertension of the newborn (PPHN). Patients who have labile physiology and low preductal saturations despite optimal ventilatory and inotropic support may be considered for extracorporeal membrane oxygenation (ECMO). Systematic reviews into the benefits of ECMO in CDH concluded that any benefit is unclear. Few randomised trials exist to demonstrate clear benefit and guide management. However, ECMO may have its uses in those that have reversibility of their respiratory disease. A few centres and networks have demonstrated an increase in survival rate by post hoc analysis (based on a difference in referral patterns with the availability of ECMO) in their series. One issue may be that of careful patient selection with regard to reversibility of pathophysiology. At present, there is no single test or prognostication that predicts reversibility of PPHN and criteria for referral for ECMO is undergoing continued refinement. Overall survival is similar between cannulation modes. There is no consensus on the time limit for ECMO runs. The optimal timing of surgery for patients on ECMO is difficult to definitively establish, but it seems that repair at an early stage (with careful perioperative management) is becoming less of a taboo, and may improve outcome and help with either coming off ECMO or decisions on withdrawal later. The provision of ECMO will continue to be evaluated, and prospective randomised trial are needed to help answer question of patient selection and management.

  • Congenital Diaphragmatic Hernia
  • Pulmonary Hypertension
  • Ecmo

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Congenital diaphragmatic hernia (CDH) is typified morphologically by failure of development of the diaphragm of the newborn resulting from incomplete fusion of elements giving rise to the diaphragm. Other developmental associations result in pathophysiological consequences which are remote from those simply associated with the herniation of the abdominal viscera into the hemithorax. Lung hypoplasia and abnormal development of the pulmonary vasculature with hyper-reactivity leads to persistent pulmonary hypertension of the newborn (PPHN). Episodes of hypoxia and hypercapnia can exacerbate the PPHN. This vicious positive cycle can lead to severe morbidity and mortality.

In patients who continue to have labile physiology and low preductal saturations despite optimal ventilator, inotropic and pulmonary vasodilatory support, where available, the next intervention considered in management is extracorporeal membrane oxygenation (ECMO). We present a review of the literature in this complex patient group.

Pathophysiology in CDH and the need for ECMO

Total lung volume in patients with CDH is significantly smaller to that of controls, and the antenatal identification of severe lung hypoplasia can be a significant determinant of postnatal course and the need for ECMO.1 Reduced surface area available for gas exchange causes severe hypoxaemia and hypercarbia, and is one of the main pathological abnormalities that can determine the consideration of ECMO. One difficulty is that severe lung hypoplasia is not currently reversible in the short term, and can make the possibility of weaning from ECMO difficult or impossible.

PPHN complicates the lung hypoplasia by causing a right-to-left shunt and persistent fetal circulation which further exacerbates hypoxaemia and hypercarbia. PPHN is a significant predictor of, and cause of morbidity, need for ECMO and mortality in CDH.2 Although there have been attempts at correlating antenatal pulmonary artery diameter and dynamics with the development of postnatal PPHN, these have not yet been proven to be predictive.3 Therefore, as yet, there are no known antenatal markers which predict the degree of PPHN in CDH. Indeed, there is no correlation between lung volume and development of PPHN.1 Therefore, patients with either good or bad prognostication based on antenatal lung volumes and ratios can have a discrepant clinical course. Postnatal clinical course is the only marker of severity of PPHN, with oxygenation index (OI) on day 1 predictive of outcome.4 5 PPHN management is a stepwise progression of ventilatory and pharmacological therapy.6 7 The institution of effective gentle ventilation, either by conventional ventilation or high frequency oscillatory ventilation (HFOV), is crucial. Inhaled nitric oxide (iNO) can improve oxygenation and reduce the acute need for ECMO while other modalities are instituted. iNO reduces requirement for ECMO in newborns with other causes of PPHN, but not conclusively in CDH,8 but it may be a temporising manoeuvre. A recent Cochrane review of randomised trials for iNO use demonstrated an overall benefit in term and near-term patients with respect to need for ECMO; outcomes of infants with CDH were not improved; outcomes were slightly, but not significantly, worse with iNO (moderate-quality evidence).9 Response to iNO should be confirmed by echocardiography. Other therapies used in PPHN are used to treat the condition in CDH with varying success.7

Cardiac dysfunction resulting from the physiological derangements, PPHN or any associated congenital structural cardiac abnormality can complicate the clinical course. Poor cardiac output and impaired tissue oxygenation can ensue, and can influence the need for ECMO. Low systemic pressure exacerbates any right-to-left shunt and hence maintaining systemic arterial pressure is crucial, and is achieved with a combination of fluid administration and/or inotropic support effectively guided by echocardiography. The management of the cardiac dysfunction in those with structurally abnormal hearts can be challenging, and these patients have a worse outcome. Suprasystemic pulmonary artery pressures with a right-to-left shunt on echocardiography may be managed with the prostaglandin analogue to open the ductus arteriosus thereby ‘venting’ of the right side of the heart10 (as well as potentially reducing the afterload as a vasodilator of the pulmonary vessels).

Indications for and outcome of ECMO

Antenatal prediction of the need for ECMO has hinged on two main antenatally identifiable measures: lung volumes and the presence of liver herniation. Antenatal liver herniation in left-sided CDH is predictive of postnatal ECMO use in many studies.1 11 Mortality is therefore also significantly increased in those with left CDH with liver up on antenatal scans.

The prediction of need for ECMO and mortality using lung volumes has evolved significantly since first described. There is some correlation between lung-to-head ratios (LHR) and postnatal outcome, although some initial reports highlighted inconsistency in the predictive value of ultrasound (US)-measured LHR.12 13 Initial differences in methodology of measuring LHR, and timing of measurement, may have affected the prognostic value of US-measured LHR. Therefore, techniques to avoid interoperator variation and to unify antenatal data have been suggested14; using the anteroposterior diameter at mid-clavicle or a tracing method (the latter is suggested to be most reproducible) should allow better correlation. LHR changes with gestational age,14 as lung growth is four times that of head growth in the third trimester,15 and alters interpretation. Therefore, observed-to-expected LHR (O/E LHR) seems more accurate. O/E LHR and MRI total lung volume (TLV) have much more reproducibility and reliability than isolated LHR. The Antenatal-CDH-Registry Group measured the O/E LHR and demonstrated that this largely eliminates the effect of gestational age.16 The O/E LHR is lower in fetuses with CDH compared with normal fetuses, and lower still in babies who die with CDH than those who survive.14 There was however some overlap in values between survivors and non-survivors. The survival for left-sided lesion related to O/E LHR with liver down was: ≤25%:30% survival; 26%–35%:62% survival; 36%–45%:75% survival; 46%–55%: 90% survival and >55%: 85% survival.16 O/E LHR using MRI may therefore be the gold standard for those offering antenatal intervention. It is worth noting that although LHR is predictive of mortality in CDH, it has not yet been strongly correlated with morbidity or PPHN.1

TLV on fetal MRI (at 24 and 34 weeks gestation) may also have a role in providing more specific information to aid prognostic decision. Lung volume on MRI strongly correlates with lung area measured on US, and is predictive of outcome in left-sided CDH,17 with the correlations of predictions seeming stronger earlier in pregnancy. Although not proven to be better than US, fetal MRI may yield additional information (eg, % of liver herniation) and give better receiver operator curves for prediction.18 Overall O/E TLV obtained by MRI scan correlates with US-derived LHR, but without the operator-dependent nature of measurements and maternal and fetal motion artefacts.17 18 A recent meta-analysis confirmed the predictive nature of both US and MRI markers of lung development to predict the need for ECMO.1

Ideally, ECMO is suited to supporting infants who can demonstrate sufficient pulmonary function to achieve adequate gas exchange initially, but who may have a reversible clinical setback. The usual indications quoted for referral for consideration of ECMO in CDH are19:

  • low preductal saturations despite maximum ventilatory and pharmacological management, with inotropic support and pulmonary vasodilation;

  • OI >40;

  • rising lactates despite maximum treatment;

  • severe hypoxaemia and hypercapnea.

The Scottish Congenital Diaphragmatic Hernia Network20 suggests the following criteria for consideration of ECMO:

  • Inability to maintain preductal saturations >85% despite optimal ventilation and management of pulmonary hypertension.

  • Arterial pCO2 above the target range with respiratory acidosis (pH <7.15) despite optimal ventilation.

  • Peak inspiratory pressure (PIP) consistently above 25 cm H2O (on continuous mechanical ventilation) to achieve ventilatory goals, or failure to improve following conversion to HFOV.

  • Inadequate tissue oxygen delivery as evidenced by metabolic acidosis pH <7.15 and lactate >5 mmol/L.

  • Systemic hypotension resistant to fluid and inotropes with a urine output <0.5 mL/kg/hour over 12–24 hours.

  • A period of adequate oxygenation/ventilation should be demonstrated with optimal management (eg, preductal SpO2 >85%, pCO2 <8 kPa).

Current contraindications are:

  • gestational age <approximately 34 weeks

  • weight <approximately 2 kg

  • bleeding disorders (ongoing bleeding or uncorrectable coagulopathy)

  • intraventricular haemorrhage (≥grade 2).

Similar work on developing national guidelines that may help with decision making for ECMO in CDH is available and being developed (eg, The Canadian Pediatric Surgery Network Congenital Diaphragmatic Hernia Evidence Review Project).21

Park et al used the formula [PaO2-PaCO2] from blood gas in the period after birth to predict need for ECMO or death.22 Best OI on day 1 is predictive of outcome,4 and serial or mean OI on day 1 is even more so.5

The difficulty with predicting the need for and outcome of ECMO in CDH is identifying neonates with potentially reversible disease, compared with those that are irreversible from the outset.23 ECMO in neonates with PPHN associated with other more reversible lung disease carries a much better outcome than in CDH.24 Two centres compared their outcome of a cohort of patients in the same era, one using HFOV as rescue25 and the other using ECMO as rescue.26 They concluded that in that time period, rescue with HFOV produced an overall survival in CDH equivalent to rescue with ECMO.

Few randomised studies of ECMO in CDH has been performed, with a paucity of good evidence to inform best practice. In one large review in the UK,27 survival benefit of ECMO in CDH could not be established with referral criteria used at that time. A meta-analysis of randomised trials also failed to show a long-term benefit (late mortality was similar in ECMO and non-ECMO CDH patients).28 A Cochrane review of the randomised controlled trials of ECMO concluded that the benefit of ECMO for babies with CDH is unclear.29

However, ECMO may have its uses in those that have reversible respiratory disease,30 31 and has been shown to confer benefit in some patients with CDH.32 One recent study did not identify any significant overall increased survival rate after the introduction of ECMO31; however, it was associated with an increase in survival rate by post hoc analysis (based on a difference in referral patterns). One issue may be that of careful patient selection alluded to previously. At present, there is no single test or prognostication that predicts reversibility of disease and criteria for referral for ECMO is undergoing continued refinement.

The presence of liver herniation is associated with the need for ECMO1; it has also been suggested that early repair of left liver-up CDH before ECMO may result in improved survival.33 Using multivariate models to assess risk for ECMO at 1 hour of life, 95% repaired in the first 60 hours and before institution of ECMO survived; whereas of those who had ECMO before CDH repair, only 65% survived. This was, however, a retrospective non-randomised analysis.

Adverse outcomes and complications are increasingly recognised with prolonged ECMO. The need for inotropes while on ECMO independently doubles mortality.24 The morbidity in survivors is high34 and costly. Complications included intracranial infarct or bleed, major extracranial bleeding, seizures and infection. Historically around one-fifth had severe neurodevelopmental problems27; although this may be improving.35 Neonates requiring prolonged ECMO historically had a low survival to discharge (approximately 25%), but this is also increasing (approximately 50%).36 Overtime, selection criteria and management strategies have changed. The changes in outcome and survival rate are difficult to compare as there are few randomised and good prospective studies with sufficient patient numbers to ascertain what changes in outcome are affected by differing selection criteria used and management strategies.

Ex utero intrapartum therapy to ECMO

The ex utero intrapartum therapy (EXIT) procedure is occasionally used in fetuses with obstructing neck masses or known severe respiratory disease diagnosed antenatally that have a high risk of mortality or need for ECMO. Originally used for fetuses with CDH that had antenatal plugging, it is now also used by some centres for severe CDH with high risk of mortality and need for ECMO. EXIT procedure allows maintenance of uteroplacental blood flow and gas exchange until airway intubation and ventilation are successfully achieved. In EXIT to ECMO, placental circulation is continued until ECMO cannulation is obtained.

In patients with less than 15% predicted lung volume on antenatal scan who underwent EXIT to ECMO, Buchmiller’s group identified no advantage in either survival37 or long-term morbidity35 in their small cohort. Although these patients were not randomised, results led to the suggestion of little benefit for routine use. There are no randomised studies on EXIT to ECMO.

Technique and application

Venoarterial (VA) ECMO is usually performed with an open cut-down technique where the right common carotid artery (CCA) and right internal jugular vein (IJV) are isolated and cannulated. A heparin bolus and infusion is started. Circuit flows are gradually increased to provide about 100 mL/kg/min. Modern centrifugal ECMO circuits require smaller blood volumes for priming and less anticoagulation than traditional roller-pump systems. The appropriate cannula size is determined with reference to the baby’s weight (sizes available down to 8 Fr; hence the VA technique may be feasible for the smaller babies). VA ECMO may have some advantages in babies with cardiac dysfunction (allowing cardiac rest and maintaining good systemic output) and may be less affected by small changes in position. Extracorporeal Life Support Organization (ELSO) data show that it was the predominant mode initially and eventually offered in babies with concomitant congenital heart disease.38 It is often feasible to repair the CCA at decannulation although the rates of long-term patency are unclear (the IJV is usually ligated).

Venovenous (VV) ECMO is often performed using an ultrasound-guided percutaneous technique to cannulate the IJV (thereby preserving the CCA). Because the cannulae are dual-lumen, the smallest size used is 12 Fr which requires the baby to be greater than about 2.5 kg. A potential advantage of this technique is that hyperoxygenated blood is directed into the pulmonary artery and hence may help decrease the pulmonary artery pressure. Of course, this technique is dependent on satisfactory cardiac output and higher flows are usually required (often about 120 mL/kg/min). Recirculation of oxygenated blood up the venous lumen makes the precise catheter position more critical.

Although a systematic review39 suggested that there was no overall advantage with either the VV or VA technique, there is a difference in preferred mode of cannulation between centres. VA ECMO seem the more popular of the two modes according to the published paper and ECMO registries; presumably as VA ECMO may give the additional benefit in the presence of severe cardiac dysfunction. They reported that VA was associated with slightly higher incidence of intracranial bleeds and seizures, while VV was associated with poorer renal perfusion. Size and vascular anatomy may sometimes dictate the mode used. Overall survival was similar between modes.

Timing of surgery on ECMO

Historically, a time limit on the length of ECMO run in patients with CDH patients at between 2 and 4 weeks was suggested. There is little doubt that prolonged need for ECMO is associated with both increased morbidity and mortality.40 However, the ideal length of time on ECMO in patients with CDH is difficult to establish. In patients with severe CDH, time on ECMO can be over 4 weeks, but satisfactory pulmonary outcomes may still be achievable.41 There is no consensus on the time limit for ECMO runs.

Another dilemma is the timing of repair in those on ECMO. Surgical repair while on ECMO is associated with a higher incidence of bleeding. One study, while not dissuading repair on ECMO, identified an increased survival rate if repair could be delayed until off ECMO.42 Therefore, repair has traditionally been performed after decannulation from ECMO. Complications from bleeding have, however, been lessened by careful anticoagulation management and the use of tranexamic acid perioperatively.43 Intestinal complications from delaying surgery (ischaemia or volvulus) are possible.44 There may even be complications of prolonging the mechanical compression on the lungs and pulmonary vessels by the intestines in the thorax. Theoretically, repair on ECMO may improve respiratory function by restoring more normal anatomy. Some reports suggest that early repair on ECMO can decrease time on ECMO, decrease complications and trend towards improved survival.30 The optimal timing of surgery for patients on ECMO is difficult to definitively establish but should include all these factors. It seems that there is a developing consensus that repair at an early (within 2 weeks) stage (with careful management of perioperative risks) may help with either coming off ECMO or decisions on withdrawal later, and potentially improve outcome.


In summary, the use of ECMO in patients with CDH remains under careful review with assessment of the indication and outcome in this group of patients. Individual responses are definitely seen, but the overall outcome benefit is hard to establish. This discrepancy may be due to the difficulty in assessing reversibility of PPHN in patients with CDH, and thereby being able to select individual patients in whom it will be of benefit. The provision of ECMO will continue to be evaluated and prospective data may help answer that question of patient selection. It seems that earlier repair on ECMO is becoming less of a taboo, and may improve outcome. Consideration for ECMO in CDH is being incorporated into integrated guidelines and care pathways. There is also the need for multicentre randomised controlled trials to help answer many of the questions that still remain in the management of patients with CDH. The hope is that suitable patients will be identified who  will benefit from refinements of this intervention, although currently, this is unclear.



  • Contributors Both authors have contributed to the design and writing of the manuscript, seen and approved the final manuscript and agreed with the content of the manuscript with regard to the review works.

  • Competing interests None declared.

  • Provenance and peer review Commissioned; externally peer reviewed.