Can preoperative cranial ultrasound predict early neurodevelopmental outcome in infants with congenital heart disease?
Abstract
Aim
To determine the role of preoperative cranial ultrasound (cUS) in predicting neurodevelopmental outcome in infants undergoing bypass surgery for congenital heart disease (CHD).
Method
Prospective cohort study on 77 infants (44 males, 33 females) operated before 3 months of age (median age at surgery 10d [range 3–88d]) who received at least one preoperative cUS. Outcome at 1 year was assessed with a standardized neurological examination and the Bayley Scales of Infant Development II (mental developmental index [MDI]; psychomotor developmental index [PDI]).
Results
Abnormalities on cUS were detected in 22 (29%) infants and consisted of diffuse brain oedema (n=12, 16%), periventricular white matter injury (n=5, 6%), ventricular dilatation (n=3, 4%), and intraventricular haemorrhage (IVH) (n=2, 3%). Infants undergoing balloon-atrial septostomy (BAS) had a higher rate of subsequent brain oedema than those without BAS (p=0.006). cUS abnormalities were not related to neurodevelopmental outcome.
Interpretation
Preoperative cUS findings in infants undergoing bypass surgery for CHD occur rather frequently, consisting of mild lesions such as brain oedema or white matter changes. These findings, however, do not correlate with early neurodevelopmental outcome.
What this paper adds
- Infants with congenital heart disease are at risk for preoperative cerebral abnormalities.
- Preoperative cranial ultrasound findings occur in around 30% of infants.
- Abnormalities consist of diffuse oedema or mild white matter injury.
- Infants undergoing balloon-atrial septostomy are at risk for cerebral oedema.
- Cranial ultrasound is not related to early neurodevelopmental outcome.
Abbreviations
-
- BAS
-
- Balloon-atrial septostomy
-
- CHD
-
- Congenital heart disease
-
- cUS
-
- Cranial ultrasound
-
- IVH
-
- Intraventricular haemorrhage
-
- MDI
-
- Mental developmental index
-
- PDI
-
- Psychomotor developmental index
-
- PVL
-
- Periventricular leukomalacia
Survival rates after bypass surgery for congenital heart disease (CHD) have increased in the past few decades,1 but neurodevelopment and quality of life is often impaired.2-4 Studies have shown that infants may manifest neurobehavioural abnormalities before surgery and that these abnormalities are highly predictive of later outcome.5, 6 Recently, cerebral abnormalities have been detected before surgery in almost half of all newborn infants with complex CHD using magnetic resonance imaging (MRI).7-10 Yet, in clinical practice cranial ultrasound (cUS) is the routinely used bedside tool to assess preoperative cerebral abnormalities in the term CHD population.11, 12 In the preterm population, cUS has been shown to have a high sensitivity for cerebral injuries such as periventricular leukomalacia (PVL) and intraventricular haemorrhage (IVH), with a good predictive validity for major neurodevelopmental impairments.13, 14 A recently published study on a large cohort of infants with CHD, however, demonstrated that cUS poorly correlated with MRI findings.15 In contrast, for infants with CHD undergoing bypass surgery, the validity of preoperative cUS in predicting neurodevelopmental outcome has not yet been determined. Thus, the aim of this study was to determine the association between preoperative cUS and neurodevelopmental outcome in infants undergoing bypass surgery.
Method
Design
This is a prospective cohort study performed at the University Children's Hospital of Zurich, a paediatric tertiary care centre with 350 cardiac surgeries. All children in the study were recruited before the first open-heart surgery. The study was approved by the local ethics committee and written informed consent was obtained from the parents. Children were enrolled between August 2004 and July 2006, and between October 2007 and May 2008. Infants eligible for this analysis had their first bypass surgery at an age younger than 3 months of age. Surgical risk was determined using the Risk Adjustment for Congenital Heart Surgery-1 scoring system.16 Genetic comorbidities were screened for based on clinical suspicion. Demographic and perioperative variables were prospectively collected and entered into an SPSS database.
Surgical management
Cardiac surgery was performed by two paediatric cardiac surgeons. Alpha-stat blood gas management was routine. In most cardiac surgeries normothermic or mild hypothermic cardiopulmonary bypass surgery (rectal temperature <32°C) was performed with a pump flow rate at 100 to 150ml/kg per minute. Norwood Stage 1 procedure, aortic arch surgery, or repair of truncus arteriosus communis were performed under moderate hypothermia (nasopharyngeal temperature 22–28°C) with pump flow rate maintained at 30 to 50ml/kg per minute with a target mean arterial pressure >25mmHg, measured in the right radial artery.
Cranial ultrasound
cUS was obtained before the first open-heart surgery in all infants, but after a balloon-atrial septostomy (BAS) procedure if this procedure was necessary. cUS was performed in the six coronal and five sagittal planes with an Acuson Sequoia scanner (Siemens, Mountain View, CA, USA) using 8MHz vector and 15MHz linear-array transducers. In our institution, cUS is recommended for all infants undergoing bypass surgery before the age of 3 months. One experienced paediatric radiologist (CK) who was unaware of the clinical course of the infants reviewed all cUS to classify the findings as described. It was not deemed necessary to review the scans that were considered normal, as they had been reviewed by a senior paediatric radiologist as part of the clinical routine. cUS findings were classified as follows: white matter injuries (periventricular leukomalacia [PVL]) according to de Vries et al.,17 intraventricular haemorrhage (IVH) according to Papile,18 ventricular dilatation and/or ventricular asymmetry, and diffuse brain oedema. Diffuse brain oedema was defined as generalized increase in cerebral echogenicity with slit-like lateral ventricles and effacement of sulci.19 In infants with more than one abnormal finding, the most severe finding was coded. Eight infants (10%) had repeat cUS examinations (maximum 4). cUS findings were classified into normal and abnormal. A first categorization was made where all findings were coded as abnormal ‘any cUS abnormality’. A second categorization was made where only PVL and cerebral oedema were considered abnormal.
Neurodevelopmental assessment
Infants were examined by two developmental paediatricians from the Child Development Centre of the University Children's Hospital Zurich (BL, AD). Examiners were not aware of the cUS findings, but were aware of the CHD diagnosis and clinical course. A standardized neurological assessment was performed before surgery if the infant was haemodynamically stable and at 1 year of age, at least 4 to 6 months after surgery. The assessment was modified after Prechtl and Beintema20 and resulted in a neurological severity score, which has been applied in infants with CHD.7 The following domains were graded from 0 to 3: posture, general movements, tone, primitive and muscle stretch reflexes, cranial nerves and reactivity/behaviour, and a summary score ranging from 0 to 18 was created. At 1 year of age, the Bayley Scales of Infant Development II providing a mental developmental index (MDI) and a psychomotor developmental index (PDI)21 was administered.
Statistics
Results are presented as median and ranges. Univariate analyses were performed using non-parametric tests. Comparisons between median MDI and PDI to test norms were performed using the Wilcoxon signed rank test. Correlation between variables was performed with the Spearman rank sign test. Associations between risk factors and outcome were analyzed using a univariate regression analysis with MDI, PDI and neuroscore as the dependent continuous variable. As genetic comorbidity was the strongest determinant of outcome, we excluded infants with a genetic comorbidity in the analysis of risk factors for poorer neurodevelopmental outcome. To determine the independent influence of risk factors on outcome, we performed a multiple linear regression analysis. We included variables based on their significant correlation with outcome in the univariate analysis or based on reports from the literature (sex, cyanotic CHD). Based on the research question of this article, we also included the variable preoperative cUS abnormalities. Thus, included variables were: any cUS abnormality; preoperative neuroscore; age at surgery; sepsis; length of ICU stay; male sex; and cyanotic heart defect. Analyses were performed using SPSS 19 (SPSS Inc., Chicago, IL, USA).
Results
Patients
During the study period, 99 infants were younger than 3 months at the time of the surgery. Of those, 77 had at least one preoperative cUS. Infants with a preoperative cUS had a higher surgical risk and were younger at surgery than those without a cUS (Table 1). cUS was obtained at a median age of 3 days (range 0–83d).
| Preoperative cUS n=77 | No preoperative cUS n=21 | p-value | |
|---|---|---|---|
| n (%) | n (%) | ||
| Sex (male) | 44 (57) | 12 (57) | 1.0 |
| Cyanotic CHD | 55 (71) | 8 (38) | 0.01 |
| Univentricular CHD | 11 (14) | 0 (0) | 0.12 |
| Genetic defect | 19 (25) | 5 (24) | 1.0 |
| Prenatal diagnosis | 21 (27) | 1 (5) | 0.04 |
| Median (range) | Median (range) | p-value | |
| Age at surgery (d) | 10 (3–88) | 63 (5–91) | <0.001 |
| Surgical risk (RACHS-1) | 3 (2–6) | 2 (2–3) | <0.001 |
| Apgar 1 | 8 (2–9) | 8 (4–7) | 0.07 |
| Apgar 5 | 9 (5–10) | 9 (7–10) | 0.45 |
| Apgar 10 | 9 (3–10) | 10 (8–10) | 0.002 |
| Arterial cord blood pH | 7.27 (7.01–7.44) | 7.25 (7.08–7.44) | 0.55 |
| Gestational age (wks) | 38.7 (32.2–41.7) | 39.3 (35.7–41.6) | 0.44 |
| Birthweight (g) | 3180 (1560–4350) | 3110 (2200–5430) | 0.79 |
- Statistics. Categorical variables: χ2 Fisher's exact test. Continuous variables: Mann–Whitney U test. cUS, cranial ultrasound; CHD, congenital heart disease; RACHS-1, risk adjusted classification for congenital heart surgery.
Surgery was performed at a median age of 10 days, 33 (43%) infants were operated within the first 7 days of age, 22 (28.5%) between 8 days and 28 days of age, and 22 (28.5%) were older than 28 days at the time of the surgery. cUS was performed at a median age of 1 day (0–5d) in the first group, at a median age of 3.5 days (0–19d) in the second group and at 33 days (1–83d) in the third group. Infants who had surgery within 1 month were all admitted to the neonatal intensive care unit and not discharged.
Nine infants died postoperatively, one family moved away, and one child was placed in foster care and was not brought back for the 1-year examination. Thus, follow-up examination was achieved for 66 of 68 surviving infants (97%) who had a preoperative cUS. Infants with preoperative cUS abnormalities were not more likely to have undergone a preoperative neurological examination (χ2 p=0.53). No difference in the frequency of cUS abnormalities was found between those followed and those without a follow-up (p=1.0). Ten infants were born prematurely with a gestational age ranging from 33.9 to 36.7 gestational weeks. Prematurity was not associated with a higher rate of cUS abnormalities (p=0.6). No child presented with clinical seizures. Nineteen children were diagnosed with a genetic disorder (25%), three had a trisomy 21, four had 22q11.2 deletion syndrome, and 12 had another genetic disorder. Cardiac diagnoses were categorized using the common technique of single ventricle versus two ventricles, with or without arch obstruction22 and are presented in Table 2. Fifty-five infants had a cyanotic CHD, 30 a transposition of the great arteries, and seven a hypoplastic left heart syndrome.
| n (%) | |
|---|---|
| Class I | 50 (64.9) |
| Class II | 15 (19.5) |
| Class III | 4 (5.2) |
| Class IV | 8 (10.4) |
- Class I, two-ventricle heart without arch obstruction; class II, two-ventricle heart with arch obstruction; class III, single ventricle heart without arch obstruction; and class IV, single ventricle heart with arch obstruction.
Preoperative cUS
Cranial ultrasound was performed at a median age of 3 days. cUS findings were detected in 22 infants (29%) (Fig. S1). Findings were mostly mild. The majority of infants had diffuse brain oedema (n=12), five infants had a PVL, of those two infants had a cystic PVL grade II and none had higher grade PVL. One infant with cystic PVL grade II had a ventricular septal defect (38.6wks gestational age) and one a hypoplastic left heart syndrome (40wks gestational age). Both infants had PDI and MDI scores below 85 at 1 year of age. Three infants had a ventricular dilatation and two had an IVH grade I. Variables potentially associated with an abnormal preoperative cUS are listed in Table 3. None of them were associated with cUS findings. We also examined whether there were any differences in these variables if only PVL and cerebral oedema were considered abnormal. There was no difference in any variable except for preoperative BAS. Those infants requiring BAS were more likely to manifest subsequent brain oedema than those without BAS intervention (9/28 [32.1%] vs 3/49 [6.1%]; p=0.006). The occurrence of cerebral oedema was not associated with worse Apgar scores or blood gases (both p>0.1).
| Any abnormality on preoperative cUS n=22 | Normal preoperative cUS n=55 | p-value | |
|---|---|---|---|
| n (%) | n (%) | ||
| Sex (male) | 13 (59.0) | 31 (56.0) | 1.00 |
| Cyanotic CHD | 17 (77.0) | 38 (70.0) | 0.58 |
| Univentricular CHD | 1 (4.5) | 10 (18.2) | 0.16 |
| Preoperative BAS | 11 (50.0) | 17 (30.9) | 0.13 |
| Genetic comorbidity | 7 (31.8) | 12 (21.8) | 0.39 |
| Prenatal diagnosis | 5 (22.7) | 16 (29.1) | 0.78 |
| Preoperative NG tube | 14 (63.6) | 32 (58.2) | 0.79 |
| Median (range) | Median (range) | p-value | |
| Age at cUS (d) | 3 (1–80) | 3 (0–83) | 0.73 |
| Age at surgery (mo) | 0.36 (0.60–2.89) | 0.23 (0.10–2.82) | 0.39 |
| Surgical risk (RACHS-1) | 3 (2–6) | 3 (2–6) | 0.16 |
| Apgar 1 | 8 (3–9) | 8 (2–9) | 0.68 |
| Apgar 5 | 8.5 (5–10) | 9 (6–10) | 0.72 |
| Apgar 10 | 9 (3–10) | 9 (6–10) | 0.77 |
| Arterial cord blood pH | 7.24 (7.09–7.35) | 7.28 (7.01–7.44) | 0.42 |
| Gestational age (wks) | 40.0 (33.9–41.2) | 38.7 (34.4–41.7) | 0.11 |
| Birthweight (g) | 3190 (1560–4030) | 3110 (1970–4350) | 0.94 |
| Head circumference at birth (cm) | 34.5 (31.0–37.0) | 34.0 (28.5–37.5) | 0.77 |
| Preoperative neuroscore | 4 (0–13) | 4 (0–14) | 0.39 |
| Preoperative lowest O2 saturation | 70 (21–93) | 77 (25–99) | 0.29 |
- Statistics. Categorical variables: χ2 Fisher's exact test. Continuous variables: Mann–Whitney U test. BAS, balloon atrial septostomy; cUS, cranial ultrasound; CHD, congenital heart disease; NG, nasogastric tube; RACHS-1, risk adjusted classification for congenital heart surgery.
Preoperative cUS and preoperative neurological examination
Sixty-four infants (86.5%) of the 77 infants with preoperative cUS also had a preoperative neurological examination. Those infants not examined were medically unstable with mechanical ventilation or irritability caused by prostaglandine infusion. However the percentage of infants receiving a neurological examination was no different between those on prostaglandine and those not on prostaglandine (p=1.0). Preoperative cUS abnormalities did not correlate with preoperative neurological abnormalities (p=0.82).
Neurodevelopmental outcome
Children were examined at a mean age of 12 months and 24 days (SD 2mo 3d). Median MDI was 89.0 (range 49–107) and median PDI was 69.0 (range 49–113) respectively. Median neuroscore was 2 (range 0–13). Infants with genetic disorders had significantly lower MDIs, PDIs, and higher neuroscores than those without (all p≤0.001). When children with genetic disorders were excluded, median MDI was 91 (50–107), median PDI was 70 (49–113) and median neuroscore was 1 (0–7). Median MDI and PDI of infants without a genetic comorbidity were significantly lower (both p<0.001) than test norms (100 [SD 15]).
Association between cUS and neurodevelopmental outcome
Cranial ultrasound findings were not related to neurodevelopmental outcome at 1 year (cUS preoperative with neuroscore at 1y: p=0.54, with MDI at 1y: p=0.82, with PDI at 1y: p=0.40). This was also the case when only PVL and cerebral oedema or only cerebral oedema were considered abnormal findings (all p>0.2).
We also correlated subgroups of cUS abnormalities with outcome. None of them (e.g. PVL, ventricular dilatation/asymmetry, IVH, or cerebral oedema) were significantly correlated with MDI, PDI, or neuroscore (results not shown).
Risk factors for adverse neurodevelopmental outcome
A genetic comorbidity was significantly correlated with poorer neurodevelopmental outcome (genetic comorbidity p<0.001 for all three outcome parameters).
When we excluded infants with a genetic comorbidity (Table 4), older age at surgery, sepsis and longer length of hospital, and intensive care stay were associated with poorer MDI scores and higher neuroscores at 1 year of age in the univariate analysis.
| MDI | PDI | Neuroscore | ||||
|---|---|---|---|---|---|---|
| Beta | p-value | Beta | p-value | Beta | p-value | |
| Any cUS abnormality | −0.01 | 0.97 | −0.13 | 0.38 | −0.03 | 0.85 |
| Gestational age | 0.22 | 0.12 | 0.01 | 0.96 | 0.03 | 0.87 |
| Preoperative neuroscore | 0.16 | 0.33 | −0.03 | 0.84 | −0.37 | 0.05 |
| Cyanotic heart defect | −0.06 | 0.70 | −0.18 | 0.21 | −0.05 | 0.79 |
| Age at surgery | −0.29 | 0.04 | −0.25 | 0.09 | 0.34 | 0.04 |
| CPB time | 0.03 | 0.84 | 0.23 | 0.12 | 0.28 | 0.10 |
| Sepsis | −0.33 | 0.02 | −0.19 | 0.19 | 0.57 | <0.001 |
| Clinical seizures | −0.11 | 0.47 | −0.16 | 0.26 | 0.24 | 0.16 |
| Length of ICU stay | −0.55 | <0.001 | −0.23 | 0.12 | 0.58 | <0.001 |
| Total length of hospital stay | −0.49 | <0.001 | −0.23 | 0.12 | 0.65 | <0.001 |
- CPB, cardiopulmonary bypass time; cUS, cranial ultrasound; ICU, intensive care unit; MDI, mental developmental index; PDI, psychomotor developmental index of the Bayley Scales of Infant Development II.
In the multiple linear regression analysis, older age at surgery and longer length of intensive care stay were significantly associated with lower MDI (Beta −0.39, p=0.03 and Beta -0.68, p=0.002 respectively), while cyanotic CHD and older age at surgery were associated with poorer PDI (Beta −0.47, p=0.01 and Beta −0.48, p=0.02 respectively). No factor was independently associated with a neuroscore at 1 year of age.
Discussion
The aim of our study was to determine the usefulness of preoperative cUS in predicting neurodevelopmental outcome after bypass surgery in infants with CHD. Abnormalities were detected in 29% of all infants and were mostly mild, consisting of diffuse brain oedema, periventricular white matter injury and ventricular dilatation. We could not detect a relationship between preoperative cUS findings and neurodevelopmental outcome at 1 year. In addition, preoperative cUS findings neither correlated with preoperative neurological findings nor with peri- or intraoperative complications.
Cranial ultrasound is a widely used bedside tool to assess cerebral abnormalities in neonates. It has been shown to be sensitive for lesions such as haemorrhage or cystic changes in preterm born infants and somewhat less sensitive in term neonates with hypoxic-ischaemic encephalopathy.23 A recently published study examined the sensitivity of head ultrasound scans compared to cerebral MRI findings in neonates with CHD.15 In a large cohort of 167 patients, cUS only detected cerebral lesions in 3% of all neonates. In addition, those lesions poorly correlated with MRI findings with an 80% false positive rate.
The rate of cUS findings in our study with 29% is significantly higher than the one reported by Rios et al.15 However, the rate was similar to those reported in studies using MRI with rates ranging from 26%7, 15 to 56%,24 even though the type of detected abnormalities was different (oedema vs periventricular abnormality).9, 10 Importantly, cerebral findings were generally mild. The sensitivity of cUS in detecting cerebral lesions can vary significantly, depending on the frequency in which it is used, the resolution of the machine and the experience of the examiner. A large interobserver variability in the assessment of cUS abnormalities has been shown for preterm infants.25 This may be the reason why Rios et al. had a rather low detection rate using cUS. In addition, the detection of small lesions such as focal white matter changes and stroke, the most frequently found lesion in neonates with CHD,9, 10 is even more difficult using cUS and has been shown to be very poor.15
Our results are in line with the study by Te Pas et al11 who found cUS abnormalities, mostly atrophy and acute ischaemic events, in 42% of neonates with CHD before surgery. van Houten et al.12 reported a higher prevalence of cUS abnormalities (59%) in a retrospective sample of 49 infants with cerebral atrophy being the most common finding. In contrast to those two studies, our cohort comprised a more recent prospective sample of infants with a higher proportion of brain oedema. This may be because of the rather early timing of the cUS (median age 3d). Infants with oedema had similar Apgar scores and cord gases as those without oedema. Thus, perinatal asphyxia seems unlikely to be the reason for cerebral oedema. Instead, BAS was associated with subsequent cerebral oedema, however, without an effect on neurodevelopmental outcome. This finding would support the view that the observed oedema may reflect postinterventional hyperperfusion rather than true brain injury. This would be supported by the fact that the resistance index was not reduced in the majority of children with cerebral oedema (results not shown).
Cerebral lesions detected on cUS correlate with adverse neurodevelopmental outcome in the population of preterm patients, however, this is only true for larger lesions and a severely abnormal outcome such as non-ambulatory cerebral palsy.23 Smaller changes such as non-cystic white matter lesions may be underestimated by cUS. Nevertheless, the advantages of cUS are significant as it is readily available, can be repeated and does not require sedation or transport to an MR suite. Despite the frequent use of cUS in infants with CHD undergoing bypass surgery,11, 12 its predictive validity for neurodevelopmental outcome has not been studied so far. In our cohort, we could not detect a relationship between preoperative cUS findings and neurodevelopmental outcome at 1 year.
Importantly, the association between neurodevelopmental outcome and preoperative cerebral MRI abnormalities has also not yet been established. In addition, neurodevelopmental problems may not be detected until school-age where more complex intellectual functions can be assessed. It is thus possible that the mild findings on cUS may be related to neurocognitive deficits detected later in life. But overall, the question arises whether cUS may not be the appropriate tool to determine the full extent of brain injury in this patient group. MRI better captures the extent and regional distribution of cerebral abnormalities. This has been shown for preterm born infants,13 term infants with perinatal asphyxia26 and currently for infants with severe CHD.15 In the recent study by Rios et al.,15 using cerebral MRI as the criterion standard, there was a high rate of false-positive findings on cUS with very little agreement between these two methods. Importantly, small white matter injuries and infarcts were not detected reliably by cUS.
Limitations
This study has limitations worth mentioning. Infants examined with cUS represented a younger population with a higher surgical risk than those without cUS. In most infants, cUS was performed only once. It has been shown that repeat cUS examinations increase the sensitivity in detecting cerebral abnormalities. However, in this patient population, repeat cUS is often not feasible. Some of the oedema may have been caused by hyperperfusion and may not be associated with brain injury. However, especially at the beginning of the study, flow measurements were not performed routinely. Further, we did not obtain cerebral MRI in these patients, which would have helped to determine whether the observed brain oedema was transient or corresponded to brain injury.
In summary, in infants undergoing bypass surgery for CHD, mild preoperative cUS findings occur rather frequently, mainly consisting of brain oedema and mild white matter changes. These abnormalities, however, do not correlate with 1-year neurodevelopmental outcome. It is unclear whether the observed oedema reflects transient hyperperfusion or actual brain injury. Cerebral MRI is clearly a better tool in determining the severity and extent of cerebral lesions, in particular small lesions, as are frequently observed in infants with CHD. Thus, in the presence of a cerebral MRI, there is clearly no need to perform a cUS. The value of the cUS lies in the quick assessment of major cerebral injuries such as bleeding or large strokes in critically ill infants, in particular when an MRI cannot be performed because of MRI-incompatible devices (pacemaker) or if infants are on extracorporeal membrane oxygenation.
Acknowledgement
We thank the parents and infants for their participation in this study. This work was supported by a grant from the Foundation Mercator, Switzerland. The foundation had no influence on the study design, data collection, analysis and manuscript preparation, and/or publication decisions. The authors have stated that they had no interests that might be perceived as posing a conflict or bias.
