We collected data on clinical and laboratory variables in the postnatal and perioperative period: sex, gestational age, preterm birth, anthropometric measurements, weight and height circumference percentiles and z scores, type of surgical repair, age at time of surgery, scores in surgical mortality risk scales, duration of extracorporeal circulation (ECC) and genetic disorders diagnosed after birth. Every newborn underwent a postnatal ultrasound examination to confirm the prenatal diagnosis. Infants were classified into 4 groups based on the cardiac anatomy and physiology (univentricular [UV] or biventricular [BV]) and whether there were lesions obstructing systemic blood (systemic obstruction: SO vs no SO). They were also classified in 3 groups based on the degree of cyanosis: (1) no cyanosis, (2) moderate neonatal cyanosis (cyanotic defect other than transposition of the great arteries [TGA]) and (3) cyanosis associated with TGA, as previously done by other authors.5,8,10

We collected variables pertaining to postoperative care, received treatment and postoperative complications including ANEs. We documented the mortality risk score at admission to the intensive care unit (ICU), the duration of mechanical ventilation, the use of vasoactive drugs, the length of stay in the ICU, the length of stay in hospital, the need for extracorporeal membrane oxygenation (ECMO) after surgery, complications in the postoperative period and postoperative mortality.

We defined ANE as intracranial bleeding, stroke, venous sinus thrombosis, seizures (clinical, electrographic or both) or brain death. The diagnosis of ANE was based on the findings of imaging tests ordered as the care team considered necessary. We did not include temporal changes in the definition of ANE. When available, diagnostic images were compared with previous imaging to verify the timing of development of the radiological features.

We documented suspected or confirmed postnatal diagnoses of genetic disorders and their association with neurodevelopmental outcomes, whether or not the association had been described in the previous literature.

We analysed the levels of 2 biomarkers of brain injury: S100 calcium-binding protein B (S100B) and neuron-specific enolase (NSE). Blood samples were collected at 4 time points: (1) immediately after birth (cord blood), (2) immediately before surgery, (3) at admission to the ICU after surgery and (5) at 24 h from admission to the ICU after surgery. The samples were collected from central venous or arterial catheters already in place, extracting the minimum necessary volume (EDTA collection tube, 0.5 mL). Samples were processed by separating the plasma through centrifugation at 1400 g for 10 min at 4 °C, and frozen immediately after for storage at −80 °C until testing. Enzyme-linked immunosorbent assays (ELISAs) were performed in duplicate with commercial kits (BlueGene Biotech CO. Ltd, Shanghai, China). The minimum detection limits in the assays were 50 pg/mL for S100B and 0.5 ng/mL for NSE.

We analysed the scores in the neurodevelopmental assessment conducted at age 2 years (Bayley-III). The Bayley-III scales assess 5 developmental domains: cognition, language, social-emotional, motor and adaptive behaviour. It yields scores based on the answers or behaviour of the child that are transformed into scaled and composite scores for age. The results are scaled to a metric a mean of 100 points, and a standard deviation (SD) of 15 points and a range of 40–160 points. For each domain, values of less than 85 points are indicative of developmental delay and are classified as mild (1–2 SDs below mean), moderate (2–3 SDs below mean) and severe (<3 SDs below mean).

Table 3 presents the results for the overall sample and by CHD anatomy group.

Six patients (7.1% of the total) were born preterm, and all had a biventricular anatomy. There were no differences in the sex distribution, mean gestational age, weight or head circumference between the groups of patients classified according to heart anatomy. Most patients (69.1%) were operated within 28 days post birth.

In the group with univentricular anatomy, a higher proportions of patients required vasoactive drugs or postoperative ECMO (11.8% vs 0%) and the mortality was higher (29% vs 0%), and the subset with SO had significantly longer stays in the ICU (38.5 days [23−60] vs 19 days [9−29.5]; P = .04), increased need of vasoactive support (maximal VIS, 61.5 [20−66] vs 19 [10–32]; P = .01), with a higher proportion requiring ECMO after surgery (16.7% vs 2.4%; P = .04) and a higher mortality (66.67% vs 5.95%; P < .01) (Table 2). Ten patients (12% of the total) received a genetic syndrome diagnosis after birth, either confirmed or highly suspected and awaiting confirmation.

The general complications and the ANEs developed by the patients are presented in Tables 4 (entire cohort) and 5 (by CHD anatomy group). We found no differences between groups in the variables under study, except for the use of insulin for management of hyperglycaemia, which was more frequent in the group with biventricular anatomy without SO.

The postoperative mortality was 5.9%, corresponding to 5 patients, all of who had univentricular anatomy, and 4 of who had SO.

As regards the neurologic complications detected in the cohort, 13 ANEs were detected in the cohort. Two of them were diagnosed in 2 different patients before surgery.

Eleven ANEs were diagnosed after surgery in 9 patients: 5 ischaemic strokes, 3 episodes of seizures, 1 venous sinus thrombosis, 1episode of severe extra-axial haemorrhage and 1 case of severe hypoxic-ischaemic encephalopathy resulting in brain death). Of the 9 patients that experienced postoperative ANEs (10.7%), 6 were operated in the neonatal period (9.5 days [7.5–12.5]) and 1 at 36 days post birth.

There were no differences in the incidence of ANE between anatomical groups; on the other hand, postoperative ANEs were more frequent in the moderate cyanosis group (26.1% vs 10.7%; P = .04). Two patients who experienced ANEs died.

During the follow-up, 5 patients required antiepileptic drugs, 3 experienced postoperative ischaemic strokes and 1 was diagnosed with a genetic syndrome characterised by heart disease, epilepsy and neurodevelopmental delay. All 5 had a biventricular anatomy.

The incidence of ANEs was higher patients who required ECMO after surgery compared to all other patients (100% vs 9.76%; P = .01), and both of these patients died.

Of the patients that experienced and survived cardiac arrest in the postoperative period (n = 6), 3 had ANEs and 1 died from the neurologic complication.

The incidence of ANE was greater in patients with moderate cyanosis (no TGA), patients who required ECMO after surgery and patients who had cardiac arrest in the immediate postoperative period. We did not find significant differences in incidence based on gestational age, birth weight or head circumference, a high STAT score, the minutes of ECC, the maximal VIS at 24 h, the Pediatric Risk of Mortality score (PRISM III) or other variables.

In the analysis of the levels of the two biomarkers, we found detectable levels in cord blood (S100B, 48.27 [16.89–171.12]; NSE, 0.46 [0.42−0.53]). The levels of S100B increased after surgery compared to the preoperative period, with greater values in the immediate postoperative period and a decreasing trend observed by 24 h after surgery (136.19 [72.68−211.14] vs 169.90 [67.5–284.07] vs 162.75 [62.12−278.10]; P ≤ .01). The levels of NSE did not increase after surgery compared to before (0.88 [0.73−1.06] vs 0.86 [0.70−0.98] vs 0.62 [0.54−0.86]; P = .241).

The levels of the biomarkers at the different time points by CHD group can be found in Table 6. Levels of S100B increased significantly in the immediate postoperative period in patients with SO (in both the BV and UV groups). We did not find significant differences at the other perioperative time points or between the other CHD categories.

The levels of S100B were higher in the immediate postoperative period in patients that experience ANEs after surgery compared to patients who did not (284.07 [167.74–483.50] vs 164.54 [67.50–273.01]; P = .20), but the difference was not statistically significant; the levels of NSE were similar in patients who had ANEs and the rest of the patients (0.75 [0.75−0.76]) vs 0.87 [0.70−0.98]; P = .578).

Five of the 84 patients died during the follow-up, and another 5 were lost to follow-up. Of the 74 remaining patients, 55 (74%) were evaluated at 2 years with the Bayley-III. The median scores in each of the 5 domains of the test were all within the normal range (85−115) (Fig. 2).

Figure 2.

Graphic representation of the scores in the Bayley-III scales in the overall sample, by heart anatomy group, by genetic diagnosis and based on the occurrence of acute neurologic events. Dx, diagnosis; TGA, transposition of the great arteries.

(0.42MB).

Thirty-eight patients had normal scores (≥ 85) in the cognition, motor and language domains; 30 had normal scores in every domain in the test. Of the patients who had an abnormal score in at least one domain, 16 had a low score in only one domain. There were 9 patients with scores below the normal range in 2 or more domains. Of the 6 patients with abnormal scores in 3 or more domains, 4 had neonatal cyanosis of moderate or high degree, 2 had UV anatomy, and 1 who had abnormal scores in all 5 domains had been diagnosed with a genetic syndrome associated with abnormal neurodevelopment.

We obtained a total of 260 scores for the assessed domains in the 55 patients (all patients underwent assessment of at least 3 domains and up to a total possible maximum of 5); 43 of the 260 scores were below the normal range, of which 27 indicated mild delay, 13 moderate delay and 3 severe delay.

Patients with UV anatomy scored lower in the cognition, motor, language and adaptive behaviour domains, and the difference was statistically significant in the motor assessment (89.5 [85–97] vs 103 [94–121]; P = .02).

We did not find significant differences when we compared the scores of patients based on the degree of cyanosis (Fig. 2).

Patients who experienced an ANE had scores lower than normal in at least 1 domain of the Bayley-III (2 of these patients had low scores in 3 or more domains), although the differences compared to patients without ANEs were not statistically significant: cognition (95 [81–105] vs 110 [95–130]; P = .5; language (79 [78–94] vs 91 [86–115]; P = .5); motor (91 [85–91] vs 100 [91–118]; P = .13); socioemotional (100 [90–110] vs 100 [90–115]; P = .8) and adaptive behaviour (82 [75−25; 89.25] vs 98 [85–106]; P = .07), as can be seen in Fig. 2.

Patients with epilepsy had significantly lower scores in the motor assessment (100 [91–118] vs 53.5 [46–61]; P = .02).

Patients who survived an episode of cardiac arrest after surgery had lower scores compared to patients who did not experience cardiac arrest in 4 of the 5 domains: cognition (82.5 [67.5–101.25] vs 110 [97.5–1032.25]; P ≤ .01), language (82.5 [77.75–88] vs 95.5 [86–115]; P ≤ .01); motor (89.5 [81.25–100] vs 100 [92.5−118]; P ≤ .01) and adaptive behaviour (76.5 [71.75–81.25] vs 98 [85–106]; P ≤ .01).

Patients with a suspected or confirmed genetic disorder scored lower in all domains, with significantly lower scores in the language, motor and adaptive behaviour domains. The scores in this group were the following, compared to those of patients without a genetic diagnosis: language, 74 [57.5–83] vs 92.5 [83–115] (P = .006); motor, 76 [52–98.6] vs 100 [91–118] (P = .0057); adaptive behaviour, 59 [47–77] vs 98 [84–106] (P = .007); cognition, 77.5 [55–105] vs 110 [95–130] (P = .051); socioemotional, 85 [55–100] vs 100 [90–110] (P = .069).

Excluding the patients with genetic syndromes from the analysis, 21 patients had scores below 85 in at least one domain, with 15 scoring low in one domain, 2 in two domains and 4 in three domains. Nine patients had language delay (mild in 7, moderate in 2), 9 had delays in adaptive behaviour (mild in 5, moderate in 4), 5 cognitive delay (mild in all), 4 delay in socioemotional development (mild in all) and 3 in motor development (mild in 2, moderate in 1).

We analysed the correlation between the measured levels of the markers of brain injury at each time point and the scores of the patients in the different domains of the Bayley-III. There was a correlation between higher levels of S100B in the immediate postoperative period and lower scores in the domains of cognition (r = 0.336 [0.023−0.589]; P = .036), language (r = 0.337 [0.020−0.593]; P = .038) and motor development (r = 0.424 [0.125−0.652]; P ≤ .01). We did not find a correlation between the level of NSE at the different perioperative time points and the scores in the Bayley-III domains.

Thus, we found less favourable scores in the Bayley-III in patients who experienced ANEs in the postoperative period, although the difference with the scores of patients without ANEs was not statistically significant. In our cohort, patients with CHD operated in the first year of life had lower scores in the Bayley-III scales if they had a univentricular cardiac anatomy, received a diagnosis of genetic disorder, suffered and survived cardiac arrest in the postoperative period and had elevated levels of S100B in the immediate postoperative period.