PATIENTS AND METHODS

Children with DGS associated cardiac defects (truncus arteriosus, interrupted aortic arch, pulmonary atresia with ventricular septal defect (VSD), Fallot’s with right aortic arch) and children with cardiac lesions associated with other features of DGS (typical dysmorphism, thymic aplasia, hypocalcaemia, family history of DGS) managed at the John Radcliffe Hospital were investigated for their 22q11 status. It was established from the Churchill Department of Clinical Genetics database and the John Radcliffe Hospital Paediatric Cardiology Database that a total of 17 children with the 22q11 mutation had been seen in the units over the period 1993–98. Of these, four had died at ages between 5 days and 2 years. Parents of the remaining 13 were invited to participate in the study, of whom 10 gave consent. Table 1⇓ provides details about the subjects with 22q11 deletion who died, or whose parents declined for them to be tested.

The John Radcliffe Hospital paediatric cardiology database was used to identify two control cases for each 22q11 deletion child. The matching process was as follows: where possible (that is, for commoner lesions such as Fallot’s tetralogy and VSD) the child of the same sex with the nearest date of birth to the index child and with an identical cardiac lesion was identified; where no such match was available, a child with a haemodynamically equivalent cardiac lesion and preoperative haemodynamic state, and requiring surgical repair of comparable duration and complexity, was identified (for example, interrupted aortic arch with VSD matched with coarctation with aortopulmonary window or coarctation with transposition of the great arteries). Table 2⇓ shows clinical details of cases and controls. None of the controls have any dysmorphic features of DGS; some with cardiac lesions suggestive of DGS, or a family history of congenital heart disease, have been tested for the 22q11 deletion using fluorescence in situ hybridisation (FISH) and have been normal.

For each child the hospital notes were examined for the following: presence of hypocalcaemia, immune deficits, type of operation, duration of cardiopulmonary bypass, pre- and perioperative hypotensive and hypoxic episodes, perioperative neurological abnormality, and length of postoperative ventilation.

The neurodevelopmental status of the children was examined (MM) using the Ruth Griffiths Abilities of Babies and Young Children. This assessment was performed “open” to the 22q11 status of the children at the time of assessment, because there is no realistic way of blinding an experienced paediatrician to the dysmorphic features of DGS.

Statistical analysis was carried out using Statgraphics Plus and CIA software. A normal distribution could not be assumed for most variables and therefore the Mann-Whitney U test and the Wilcoxon rank sum test were used for intergroup comparisons and the comparisons of paired data respectively. Multivariate correlation analysis was used to identify factors correlating with intellectual outcome, and MANOVA was used to test for interaction between these factors, although the degrees of freedom only allowed two factors to be tested at a time. Local research ethics committee approval was obtained.

RESULTS

Examination of the control group alone (n = 20) shows that there was no correlation between overall developmental quotient (DQ) and age at presentation, age at first operation, or age at testing. Duration of cardiopulmonary bypass and duration of ventilation were each negatively correlated with DQ (r = −0.50, p = 0.03; and r = −0.74, p = 0.0005 respectively). Duration of cardiopulmonary bypass was most significantly negatively correlated with speech and language (r = −0.65, p = 0.005) and practical reasoning (r = −0.58, p = 0.02), but showed no significant correlation with eye-hand, performance, gross motor, or personal-social subscale scores. Prolonged ventilation, a marker for a difficult postoperative course, shows a negative correlation with all the subscale scores except gross motor and personal-social (r = −0.54 to −0.82, p = 0.047 to 0.006).

When the smaller DGS group was studied alone, the effect of duration of cardiopulmonary bypass does not reach significance, and the effect of duration of ventilation approaches significance for DQ and most subscale scores (p = 0.054 to 0.059), but only achieves it for eye-hand (r = −0.69, p = 0.038) and practical reasoning (r = −0.98, p = 0.016).

Multiple analysis of variance was used to study the combined case and control groups. This shows a clear negative effect on DQ by the presence of the 22q11 deletion (p = 0.004), hypocalcaemia (p = 0.009), and immunodeficiency (in the form of quantitative or qualitative T cell abnormalities) (p = 0.007); the latter two features are confounded by their strong association with the 22q11 deletion. Abnormal neurological status perioperatively (p = 0.035), increasing cardiopulmonary bypass time (p = 0.008), and postoperative ventilation time (p = 0.0003) also had significant negative effects on DQ. The effects of these factors were independent of 22q11 deletion status.

Table 3⇓ summarises the comparative clinical data for the 22q11 deletion group (n=10) and the matched controls (n = 20). There were no significant differences in age at presentation, age at first operation, duration of cardiopulmonary bypass, duration of postoperative ventilation, or age at testing. Hypocalcaemia and immune deficiency were, as expected, significantly more common in the group with 22q11 deletion (p = 0.003 and p = 0.0002 respectively). There was no difference in the incidence of perioperative acidosis (2/10 and 3/20 respectively).

When the neurodevelopmental outcome of the two groups was compared (table 4⇓), the DGS group mean DQ score (71, 95% CI 47 to 95) was substantially lower than the mean DQ score of the controls (113, 95% CI 108 to 118; p = 0.0001). The DGS group scored significantly worse than controls for all Griffiths subscales (range p = 0.0001 to 0.01). It is of note that although six of the 10 DGS children had DQs within the normal range, scores were below 100 in four of the six, and that the remaining four DGS children had a very poor neurodevelopmental outcome (DQs ranging from 3 to 56). The controls had DQs above 100 in all but two cases.

Two of the DGS group had particularly poor developmental outcomes (DQs 3 and 35). As these two subjects had had preoperative acidosis, preoperative hypocalcaemic fits and hypotension, and a particularly difficult postoperative course, the DGS group was also compared with the controls, with these subjects excluded. The statistically significant difference in DQ between DGS and control groups persisted when these two children were excluded from analysis.

There were four DGS children with both hypocalcaemia and perioperative fits. The two with preoperative seizures temporally associated with severe hypocalcaemia had DQs of 3 and 35, whereas the children who had brief postoperative seizures not coinciding with hypocalcaemia, had DQs of 91 and 110. No control children had pre- or perioperative fits. In both the total group (p < 0.0001) and within the DGS group (p = 0.005), hypocalcaemic fits were strongly correlated with a low DQ (correlation coefficient −0.81). It is notable that perioperative acidosis was strongly correlated with a low DQ in children with DGS (p = 0.005, correlation coefficient −0.81) but not in the control children (p = 0.41, correlation coefficient −0.20). In the case of two DGS children with DQs of 49 and 56, no predictive risk factors could be identified except the 22q11 deletion itself. Conversely there were several very sick infants with good outcomes among the controls; for example, a child who was hypotensive, acidotic, and profoundly hypoxic for 12 hours after coarctation repair, before proceeding to an emergency arterial switch operation requiring the longest bypass time in the study—165 minutes—had a DQ of 117. Another, described as “moribund and anuric” after seven hours of intensive care trying to reverse a pH of 7.05, has a DQ of 104.

DISCUSSION

While this study is small and the matching process and developmental assessment were both “open”, the group differences are very clear and indicate that abnormal neurodevelopmental status following cardiac surgery in DGS cannot be attributed to the cardiac disease and the effects of cardiac surgery per se. The three DGS children who refused to participate are all “statemented” as having special educational needs, and two of them have movement as well as cognitive difficulties; thus it seems unlikely that their reluctance to participate has biased our sample in favour of more severely affected children. If account is also taken of the 4/17 children with DGS who died at ages between 5 days and 2 years, all having complex congenital heart lesions recognised as associated with a high mortality, the poor overall prognosis of children with congenital heart disease associated with 22q11 deletion is underlined; in this regional cohort 8/17 subjects either died or have severe neurodevelopmental handicap (DQs 3 to 56), whereas in matched controls there was no child with a poor neurodevelopmental outcome.

A search for preoperative factors in the four DGS children with poor outcome which may distinguish them from the DGS children with a good outcome (that is, DQs between 81 and 110) revealed only two probable predictors—hypocalcaemic seizures (p = 0.005) and preoperative acidosis (p = 0.005). It would therefore seem reasonable to be cautious in predicting the neurodevelopmental outcome following cardiac surgery in children who have been identified as having the 22q11 deletion, particularly where these two risk factors are present. Perioperative seizures did not occur in control children in this study; while seizures in children undergoing cardiac surgery clearly occur for a variety of reasons, consideration should be given to the possibility of DGS in such children if this has not been excluded.

The reasons for the discrepancy in neurodevelopmental outcome in children matched for cardiac status, with and without the 22q11 deletion, merit further investigation. Abnormal developmental status has been identified in a variable proportion of children with DGS in a number of studies. In addition, structural brain abnormalities on imaging or at postmortem examination have been identified in some DGS children, of which some at least are congenital rather than acquired, although none have been explicitly associated with developmental outcome.^3,4,10,11 It seems likely therefore that the developmental problems experienced by a proportion of DGS children after cardiac surgery are a function of their DGS and/or an interaction between this diagnosis and their cardiac condition, rendering them more susceptible to neurological sequelae. The findings of Gerdes and colleagues⁵ (although not a cohort or case-control study) would support the former of these two explanations. However, the finding in our cohort study that preoperative acidosis appears to have a more severe impact in children with DGS does suggest the possibility that the 22q11 deletion impairs the response to cardiovascular stress, perhaps more so when functional hypoparathyroidism with hypocalcaemia is present. The notion of an increased CNS vulnerability of children with 22q11 deletion is also intriguing in the context of the high incidence of psychosis, particularly schizophrenia reported in adults with 22q11 deletion.¹² It may be that hypocalcaemia, together with immunodeficiency, are markers of a more significant chromosomal deletion and that it is the nature of this mutation which is responsible for the poorer developmental outcome. In any event it seems clear that the spectrum of developmental difficulties in this patient group cannot be laid exclusively at the feet of their cardiac illness.

Finally, this study emphasises the importance, commented on previously,^13,14 of identifying pre-existing factors (such as DGS—identifiable by FISH within 72 hours) in any evaluation of developmental outcome following cardiac surgery. It is noteworthy, in this context, that those cardiac malformations considered high risk for neurological sequelae are also those associated with DGS. In the meantime it would seem sensible, given the gloomy outcome following hypocalcaemic fits in this study, to identify and treat hypocalcaemia aggressively in this group of patients.