Neuropathology of Congenital Heart Disease in an Inpatient Autopsy Cohort 2000-2017.

Background As a result of medical and surgical advancements in the management of congenital heart disease (CHD), survival rates have improved substantially, which has allowed the focus of CHD management to shift toward neurodevelopmental outcomes. Previous studies of the neuropathology occurring in CHD focused on cases preceding 1995 and reported high rates of white matter injury and intracranial hemorrhage, but do not reflect improvements in management of CHD in the past 2 decades. The purpose of this study is therefore to characterize the neuropathological lesions identified in subjects dying from CHD in a more-recent cohort from 2 institutions. Methods and Results We searched the autopsy archives at 2 major children's hospitals for patients with cyanotic congenital cardiac malformations who underwent autopsy. We identified 50 cases ranging in age from 20 gestational weeks to 46 years. Acquired neuropathological lesions were identified in 60% (30 of 50) of subjects upon postmortem examination. The most common lesions were intracranial hemorrhage, most commonly subarachnoid (12 of 50; 24%) or germinal matrix (10 of 50; 20%), hippocampal injuries (10 of 50; 20%), and diffuse white matter gliosis (8 of 50; 16%). Periventricular leukomalacia was rare (3 of 50). Twenty-six subjects underwent repair or palliation of their lesions. Of the 50 subjects, 60% (30 of 50) had isolated CHD, whereas 24% (12 of 50) were diagnosed with chromosomal abnormalities (trisomy 13, 18, chromosomal deletions, and duplications) and 16% (8/50) had multiple congenital anomalies. Conclusions In the modern era of pediatric cardiology and cardiac surgery, intracranial hemorrhage and microscopic gray matter hypoxic-ischemic lesions are the dominant neuropathological lesions identified in patients coming to autopsy. Rates of more severe focal lesions, particularly periventricular leukomalacia, have decreased compared with historical controls.

Clinical Perspective

What Is New?

Classical lesions observed in children with congenital heart disease, such as periventricular leukomalacia, are becoming increasingly rare, even in children who die of their disease.

Neurological disability in these patients is increasingly attributed to more‐subtle lesions, which are more difficult to detect on imaging and therefore pose a greater diagnostic challenge.

What Are the Clinical Implications?

Studies of novel serum and imaging biomarkers for neuronal injury will be of critical importance for predicting and ultimately improving neurological outcomes in these children.

CHD congenital heart disease

PVL periventricular leukomalacia

Introduction

Congenital heart disease (CHD) is the most common birth defect, affecting 1 in every 110 infants born in the United States.1, 2 Advances in the medical and surgical management of CHD have dramatically increased survival rates, and an increasing proportion of infants born with CHD are surviving into adulthood.3 Two‐thirds of those living with CHD are now adults, and, as a result, CHD is increasingly viewed as a lifelong disease.4 Consequently, focus has shifted to the long‐term morbidities associated with CHD, including neurodevelopmental outcomes. It is now recognized that survivors of CHD have a wide range of neurological deficits, including impairments in cognition, executive function, memory, language, and disturbances in gross and fine motor function.5, 6 The burden of these neurological deficits is significant; 25% to 50% of patients with CHD have neurological sequelae with effects spanning into adulthood.7, 8

The classic pattern of acquired neuropathology in infants with CHD is cerebral white matter injury, ranging from diffuse gliosis to periventricular leukomalacia (PVL).8 The reported incidence of PVL in earlier neuropathological studies has ranged from 11% to 61%.9, 10, 11 Acute neuronal injury of the brainstem, hippocampus, and thalamus were also frequently encountered lesions (present in >50%).9 However, existing neuropathology data come from autopsy studies of patients dying predominantly in the 1980s and 1990s. We therefore sought to characterize the neuropathology of CHD in a recent cohort of patients at 2 major children's hospitals with active CHD programs.

Methods

Data Availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Patient Selection

We searched the autopsy archives at the University of Iowa Hospitals and Clinics (including the Stead Family Children's Hospital) and the Children's Hospital of Philadelphia from 2000 to 2017 and 2000 to 2014, respectively. Inclusion criteria were: (1) critical cardiac malformation, defined as right heart obstructive lesions, including tetralogy of Fallot, tricuspid atresia, critical pulmonary stenosis, or pulmonary atresia; left heart obstructive lesions, including hypoplastic left heart syndrome, interrupted aortic arch/coarctation of the aorta, and critical aortic stenosis; and mixing lesions, including transposition of the great arteries, total anomalous pulmonary venous return, and truncus arteriosus12; and (2) complete autopsy report, including gross and microscopic examinations of the heart and brain. Although some of these lesions (eg, coarctation of the aorta, truncus arteriosus) commonly present without cyanosis, we included them given that they lead to altered cerebral blood flow and relative hypoxic blood flow to the fetal brain, have the potential to cause cyanosis, and are conventionally classified as critical lesions with the potential to cause cyanosis. They are also included in the major historical studies (eg, Kinney et al), such that excluding them would preclude any conclusions about historical trends in the pathology of CHD. Exclusion criteria were: (1) cases with definite acyanotic heart defects; (2) restricted autopsy; (3) no histological sections of brain submitted; (4) marked maceration limiting assessment according to the original pathology report; and (5) clinical history of extracorporeal membrane oxygenation. Patients with a clinical history of extracorporeal membrane oxygenation were excluded because we sought to focus on the neuropathology of CHD rather than the known neurological effects of extracorporeal membrane oxygenation.13 Patients were not excluded based on gestational age. For comparisons of neuropathological lesion rates by gestational age, prematurity was defined as birth before 37 weeks. This study was reviewed by the University of Iowa Institutional Review Board (Project 201706772). Given that all subjects were deceased at the time the research was conducted and appropriate autopsy consents had been obtained as part of clinical care at the time of death, this study was determined to not constitute human subjects research under applicable regulations and was therefore exempt from institutional review board review.

Autopsy and Histology

All cases underwent both a gross examination of the central nervous system and a microscopic examination of representative brain sections by a neuropathologist at the time of autopsy. During the time covered by this study, all cases were signed out by a small group of experienced pediatric neuropathologists at each institution, including one of the authors (P.A.K.) who signed out all cases at the University of Iowa. Tissue sampling was at the discretion of the originating pathologist and was formalin fixed and paraffin embedded in the usual fashion. Most diagnoses were made on hematoxylin and eosin–stained sections, but, where necessary, additional stains were undertaken by the original pathologist. For statistical purposes, acute neuronal injury was defined as acute neuronal necrosis and/or hypereosinophilia. Chronic neuronal injury was defined as neuronal loss or gliosis.

Record Review

All autopsy reports were systematically reviewed by one of the authors (L.A.R.) under the supervision of an experienced pediatric neuropathologist (M.M.H.). Where necessary, the original slides were retrieved and reviewed to clarify ambiguous diagnoses. Demographic and clinical data about the patient's underlying diagnosis and treatment course were abstracted. Autopsy reports were then reviewed to identify acquired neuropathological lesions. These included: periventricular leukomalacia, white matter gliosis; hippocampal neuronal injury; large territory infarctions; acute neuronal necrosis, loss and gliosis of the thalamus, basal ganglia, cerebellum, midbrain, medulla, pontine tegmentum, basis pontis, and spinal cord; olivary gliosis; various forms of intracranial hemorrhage (germinal matrix, subdural, intradural, epidural, subarachnoid, and isolated intraparenchymal); intravascular thrombi; findings consistent with meningitis and/or encephalitis; and venous sinus thrombosis. Because of the nature of the study, the clinical and demographic data available are limited to that present in the autopsy reports. We were unable to obtain medical records for a more‐detailed review of operative and physiological data.

Statistical Analysis

Continuous variables are described using nonparametric descriptors and expressed as median (range). Comparisons for continuous variables were performed using 2‐tailed Student t tests. Categorical variables were summarized using absolute values and percentages. Rates of neuropathological lesions were compared across groups using Fisher's exact test. All P values were corrected for multiple comparisons using Bonferroni's method, where appropriate. Corrected P<0.05 was considered statistically significant.

Results

Patient Cohort

A total of 161 patients dying from CHD were identified at both institutions. A total of 111 cases were excluded because of: (1) acyanotic heart defects (n=8); (2) restricted autopsy (n=45); or (3) no histological sections of brain tissue submitted or tissue was excessively soft or macerated at the time of autopsy prohibiting extensive microscopic examination (n=44); and (4) clinical history of extracorporeal membrane oxygenation (n=14). Fifty autopsy cases (25 females, 25 males; ranging in age from 20 gestational weeks to 46 years) met inclusion criteria. Of the 50 cases identified, 7 cases were either stillborn or an elective pregnancy termination was performed following ultrasound diagnosis of CHD. Gestational age was available for 40 patients. Median gestational age was 36.1 weeks (range, 20–40). Of the 40 patients for which gestational age was known, 55% (22 of 40) were premature whereas 45% (18 of 40) were born at term. Median postnatal age at death was 18.5 days (range, 0 days to 46 years). Most cases (82%; 41 of 50) died at age <1 year. In the remaining cases (n=9), median postnatal age at death was 14 years (range, 20 months to 46 years). All patients within the cohort represent cases dying in the hospital. Patient demographics are summarized in Tables 1 and 2.

Table 1

Patient Demographics and Presence of Acquired Neuropathologic Lesions in Autopsy Cases With History of Congenital Heart Disease

	PNA	GA (wk)	Sex	Cardiac Diagnosis	Genetic Diagnosis	Acquired Neuropathology
1	0 d	20 1/7	F	TAPVR, HLHS	None	No
2	0 d	21	F	HLHS	Trisomy 13	No
3	0 d	21	F	TOF	Chromosome 13 deletion	No
4	0 d	21 4/7	F	TA	Trisomy 18	GMH, SAH
5	0 d	22 4/7	F	CoA	None	No
6	0 d	22 5/7	M	TAPVR	VACTERL syndrome	No
7	0 d	23	M	TOF	Femoral‐facial syndrome	No
8	0 d	30 4/7	M	TGA	Right laterality sequence anomaly	No
9	0 d	39	M	TOF	Trisomy 13	No
10	1 d	36 1/7	M	TOF	VACTERL syndrome	No
11	1 d	39 2/7	F	HLHS	None	GMH, IDH, IPH
12	1 d	37	F	TGA	None	WMG, thromboemboli
13	1 d	32 2/7	F	TOF	del(1p36)	No
14	1 d	36	M	PA	Schinzel–Giedion syndrome	SAH
15	2 d	23	M	HLHS	None	GMH
16	2 d	39 1/7	F	IAA	Chromosome 22 deletion	ANN
17	3 d	35	F	TGA, TA, CoA	None	ANN, SAH, GMH
18	4 d	26	M	TAr	None	GMH
19	4 d	38 2/7	M	TOF	None	ANN
20	6 d	38 1/7	M	TGA	Pyruvate dehydrogenase deficiency	WMG, ANN, SAH, GMH
21	7 d	37 4/7	F	TAr	None	No
22	8 d	40 4/7	M	PA	del(13q)	WMG, ANN, GMH
23	10 d	37	M	HLHS	None	No
24	10 d	32 6/7	F	TAPVR, HLHS	None	PVL,WMG, ANN
25	16 d	38 2/7	M	TGA, DORV	None	SAH
26	21 d	36	F	TAPVR	Strickler's syndrome	ANN, infarct
27	26 d	30 1/7	M	HLHS	None	PVL, SAH, GMH
28	32 d	33 2/7	M	TAr	None	SAH, GMH
29	6 wk	…	M	TAPVR	Heterotaxy/asplenia syndrome	WMG, ANN, thromboemboli
30	1.5 mo	34 5/7	F	TAPVR	None	No
31	1.5 mo	38	F	PA	dup(16p13.11)	No
32	2 mo	37	M	AS	None	No
33	4 mo	38	F	AS	None	ANN, IPH
34	5 mo	21 1/7	F	TOF	None	No
35	5 mo	39 6/7	F	TOF	del(11p13)	No
36	5 mo	34	M	HLHS	dup(16p13.3)	SAH, EDH, IDH
37	7 mo	…	M	TGA	None	PVL, WMG, ANN, GMH, IPH
38	7 mo	36 2/7	F	TOF	None	WMG, SAH, SDH
39	7 mo	…	M	TGA	Klinefelter's syndrome	No
40	8 mo	38 2/7	F	CoA	None	WMG, ANN, IPH
41	10 mo	38 3/7	F	TOF	None	No
42	20 mo	38	M	DORV, HLHS	None	WMG, ANN
43	21 mo	…	M	DORV, TAPVR	Heterotaxy/asplenia syndrome	SAH, infarct
44	4 y	…	F	TAPVR	None	WMG, ANN, IPH
45	5 y	38 1/7	M	TGA, DORV	None	IPH, infarct
46	14 y	…	M	TGA	None	WMG, ANN, SAH
47	18 y	…	F	EA	None	ANN, SAH, IPH
48	28 y	…	F	TGA	None	WMG, ANN, infarct
49	42 y	…	M	TOF	None	Infarct
50	46 y	…	F	TOF	None	No

ANN indicates acute neuronal necrosis; AS, aortic stenosis; CoA, coarctation of the aorta; EA, Ebstein's anomaly; EDH, epidural hemorrhage; F, female; GMH, germinal matrix hemorrhage; HLHS, hypoplastic left heart syndrome; IAA, interrupted aortic arch; IPH, intraparenchymal hemorrhage; M, male; PA, pulmonic atresia; PVL, periventricular leukomalacia; SAH, subarachnoid hemorrhage; SDH, subdural hemorrhage; TA, tricuspid atresia; TAPVR, total anomalous pulmonary venous return; TAr, truncus arteriosus; TGA, transposition of the great arteries; TOF, tetralogy of Fallot; and WMG, white matter gliosis.

Table 2

Summary of Clinical Characteristics in Autopsy Cases With History of CHD Including Cardiac Diagnoses

	Total (n=50)	Intervention (n=26)*	No Intervention (n=24)*
Postnatal age at death, median (range)	18.5 d (0 d to 46 y)	209 d (3 d to 46 y)	1.0 d (0–147)
Gestational age, wk, median (range)	36.1 (20–40)	37.5 (30–40)	34.1 (20–40)
Birth weight, g, median (range)	2300 (320–3577)	2448 (1384–3194)	1750 (320–3577)
Stillborn or elective termination†	7	0	7
Male sex	25	14	11
Chromosomal abnormality	12	3	9
Cardiac diagnosis
Tetralogy of Fallot	12	6	6
Hypoplastic left heart syndrome	11	4	6
Transposition of great arteries	10	7	3
Total anomalous pulmonary venous return	8	5	3
Coarctation of aorta/hypoplastic or interrupted aortic arch	4	2	2
Tricuspid atresia	2	1	1
Severe aortic stenosis	2	2	0
Truncus arteriosus	3	1	2
Pulmonic atresia	3	0	3
Ebstein's anomaly	1	1	0

P values are by Student t test for continuous variables and Fisher's exact test for categorical variables. Correction for multiple comparisons by Bonferroni's method. CHD indicates congenital heart disease.

No statistically significant differences at the 5% level were found by intervention status after adjustment for multiple comparisons.

Adjusted P value for stillborn or elective termination was 0.056.

A summary of the cardiac diagnoses is presented in Table 2. The most common cardiac conditions identified were tetralogy of Fallot (n=12; 24%), hypoplastic left heart syndrome (n=11; 22%), transposition of the great arteries (n=10; 20%), and total anomalous pulmonary venous return (n=8; 16%). Of the 50 total cases identified, 52% (26 of 50) underwent definitive repair or palliation. Clinical data regarding whether patients underwent cardiopulmonary bypass or deep hypothermic arrest were not available for all cases. Of the 50 total cases, 30 patients (60%) had isolated CHD with no identified chromosomal abnormalities or accompanying congenital malformations. Of the remaining cases, 12 (24%) were diagnosed with concomitant chromosomal abnormalities and 8 (16%) had multiple congenital abnormalities in addition to CHD. Table 1 provides a summary of the genetic conditions encountered in the series.

Acquired Neuropathological Lesions

The overall incidence of acquired neuropathological lesions was 60% (30 of 50). Of the 50 patients in the series, 20 (40%) lacked significant acquired neuropathological lesions on postmortem examination. There were no individual lesions that were present in the majority (>50%) of cases. The most common lesions were intracranial hemorrhage, acute hippocampal injury (10 of 50; 20%), and diffuse white matter gliosis (8 of 50 [16%]; Table 3). Intracranial hemorrhage was most commonly subarachnoid (12 of 50; 24%), germinal matrix hemorrhage (10 of 50; 20%), and isolated intraparenchymal hemorrhage (9 of 50; 18%). Germinal matrix hemorrhage ranged in severity from focal, microscopic intraparenchymal hemorrhage (n=3) to grossly identifiable hemorrhage with intraventricular extension (n=5).

Table 3

Summary of Acquired Neuropathology Found in Autopsy Cases by Comorbidity and Surgical Intervention

	Total	Chromosomal Anomalies*		Multiple Anomalies*		Cardiac Surgery*		Preterm Birth*
		No	Yes	No	Yes	No	Yes	No	Yes
No. of cases	50	30	12	30	8	24	26	18	22
Cerebral cortex
Acute cortical injury	6	5	1	5	0	1	5	3	0
Large territory infarcts	6	3	0	3	3	1	5	1	1
Cerebral white matter
PVL	3	3	0	3	0	0	3	0	2
WMG	8	6	0	6	2	2	6	3	1
Hippocampus
Acute	4	3	1	3	0	1	3	3	1
Chronic	6	5	1	5	0	1	5	2	2
Basal ganglia
Acute	2	2	0	2	0	0	2	1	1
Chronic	2	2	0	2	0	0	2	0	2
Thalamus
Acute	3	3	0	3	0	0	3	1	1
Chronic	3	2	0	2	0	1	2	0	2
Cerebellum
Acute	4	3	0	3	1	0	4	3	0
Chronic	3	1	1	1	1	2	1	2	0
Midbrain
Acute	2	1	0	1	1	0	2	0	1
Chronic	3	2	0	2	1	0	3	0	1
Pons
Acute, tegmentum	4	2	1	2	0	2	2	3	1
Chronic, tegmentum	1	2	0	2	0	0	1	0	1
Acute, basis pontis	1	0	0	0	1	0	1	0	0
Chronic, basis pontis	2	0	1	0	1	0	1	1	0
Medulla
Acute	3	2	1	2	0	1	3	2	1
Chronic, olivary nuclei	2	2	0	2	0	0	2	1	1
Chronic, other	4	2	1	2	1	1	3	2	1
Spinal cord
Acute	3	2	0	2	1	2	2	0	2
Chronic	1	1	0	1	0	1	1	0	1
Intracranial hemorrhage
SAH	12	8	2	8	2	3	9	2	7
GMH	10	8	2	8	0	6	4	3	6
IPH (isolated)	9	8	1	8	0	2	7	5	1
SDH	3	2	0	2	1	0	3	0	2
IDH	3	2	1	2	0	1	2	1	2
EDH	2	0	1	0	1	0	2	0	1
Thrombi	2	1	0	1	1	1	1	1	0
Meningitis/encephalitis	3	3	0	3	0	0	3	1	0
Venous sinus thrombosis	1	0	0	0	1	0	1	0	0
Any acquired lesion† , ‡	30	21	4	21	5	10	20	11	11

P values by Fisher's exact test with Bonferroni's correction for multiple comparisons for individual lesions only. Acute and chronic injury are defined as acute neuronal necrosis and neuronal loss/gliosis, respectively. EDH indicates epidural hemorrhage; GMH, germinal matrix hemorrhage; IDH, intradural hemorrhage; IPH, intraparenchymal hemorrhage; PVL, periventricular leukomalacia; SAH, subarachnoid hemorrhage; SDH, subdural hemorrhage; and WMG, white matter gliosis.

No statistically significant differences at the 5% level were found by lesion characteristic after adjustment for multiple comparisons.

Infants without chromosomal anomalies had a higher rate of acquired lesions than infants with chromosomal anomalies (P=0.041).

Infants who had cardiac interventions had a higher rate of acquired lesions than infants who did not have cardiac surgery (P=0.020).

Hypoxic‐ischemic lesions were identified in 42% (21 of 50) of cases. The most common sites of injury were: (1) hippocampus (10 of 50; 20%); (2) cerebral white matter/diffuse white matter gliosis (8 of 50; 16%); and (3) cerebellum (6 of 50; 12%). Large territorial infarcts were observed in 12% of cases (8 of 50). PVL was rare and observed in only 3 of 50 cases. Hypoxic ischemic lesions within the basal ganglia were also rare (2 of 50). Within the brainstem, the incidence of acute or chronic injury was <10% in all locations, including the basis pontis and inferior olives. Histological sections of spinal cord were submitted for only 24% (12 of 50) of cases. Evidence of hypoxic‐ischemic lesions were present in 3 cases, including acute necrosis of the cervical (n=2), lumbar (n=1), and sacral (n=1) spinal cord and atrophic lateral corticospinal tracts (n=1).

Large territory ischemic or hemorrhagic infarctions were identified in 12% (6 of 50) of cases. Five of these 6 lesions were ischemic, with 1 hemorrhagic infarction of the frontal lobe. Four of the 6 were judged to be acute based on neuropathological examination. Of the 2 subacute or chronic cases, 1 was in the lentiform nucleus and the other involved the middle and posterior cerebral artery territories. Two of the 6 cases showed intravascular thrombi, fibrinous in 1 case, and likely thromboembolic in the other.

There was evidence of acute infectious processes, including meningitis and encephalitis, in 3 cases. One case demonstrated evidence of acute brainstem encephalitis characterized by microglial nodules and neuronophagia. An additional case demonstrated extensive parenchymal necrosis in the right frontal and parietal cortices associated with presence of acute inflammation and fungal organisms consistent with an acute fungal meningoencephalitis. One case showed evidence of focal terminal acute meningitis characterized by mild acute inflammation in the leptomeninges overlying the superior frontal gyrus.

Of the 50 total cases identified, 18% (9 of 50) died at a postnatal age >1 year. Most of these long‐term survivors demonstrated acquired neuropathological lesions (89%; 8 of 9). A variety of neuropathological lesions were identified, including intracranial hemorrhage (n=5), large territory infarctions (n=4), acute status spongiosus (n=1), brainstem encephalitis (n=1), acute fungal meningoencephalitis (n=1), and venous sinus thrombosis (n=1).

Analysis of Acquired Neuropathology in Patients With Chromosomal Abnormalities

The cohort was classified into 3 groups: group 1: isolated CHD (n=30); group 2: CHD with an associated chromosomal abnormality (n=12); and group 3: CHD with multiple congenital anomalies, but normal chromosomes (n=8). Incidence of acquired neuropathological lesions in groups 1 to 3 was 70% (21 of 30), 33% (4 of 12), and 63% (5 of 8), respectively. There was a lower incidence of acquired neuropathological lesions in patients with chromosomal abnormalities compared with patients with isolated CHD (4 of 12 versus 21 of 30; P=0.041; Table 3). Of the stillborn patients included within the cohort (n=7), 57% had chromosomal abnormalities (4 of 7), 29% had isolated CHD (2 of 7), and 14% had multiple congenital anomalies (1 of 7). All stillborn patients lacked acquired neuropathological lesions (0 of 7; 0%).

Analysis of Acquired Neuropathology in Cases With Cardiac Procedures

Of the 50 total cases identified, 52% (26 of 50) underwent repair or palliative procedures. Acquired neuropathological lesions were identified in 77% (20 of 26) of those undergoing cardiac procedures compared with 42% (10 of 24) in those patients not undergoing cardiac procedures (P=0.020). Of those undergoing cardiac procedures, the most common lesions were: (1) subarachnoid hemorrhage (9 of 26, 35%); (2) hippocampal injury (acute or chronic; 8 of 26 [31%]); (3) isolated intraparenchymal hemorrhage (7 of 26; 27%); and (4) diffuse white matter gliosis (6 of 26; 23%). PVL was only observed in 3 of the 26 cases. There was a higher gestational age (37.5 versus 34.1 weeks), and higher postnatal age at death (209 days versus 1 day), in patients undergoing cardiac procedures compared with those who did not. A summary of the acquired neuropathological lesions in cases by presence or absence of cardiac intervention is presented in Table 3. For 3 patients, the intervention was carried out entirely percutaneously (right ventricular outflow tract stent placement, a pulmonary artery balloon dilatation, and 1 case with multiple balloon valvuloplasties and atrial septostomy). Of these 3 patients, 1 had cerebellar hemorrhage, 2 showed acute hypoxic ischemic injury, and 1 had no acquired neuropathological lesions. None had periventricular leukomalacia or white matter gliosis.

Analysis of Acquired Neuropathology in Premature Infants

Of the 40 patients whose gestational age was known, 55% (22 of 40) were premature and 45% (18 of 40) were born at term. The most common lesions in premature patients were subarachnoid hemorrhage (7 of 22; 32%) and germinal matrix hemorrhage (6 of 22; 27%). There was no difference in the proportion of patients with acquired neuropathology in premature patients compared with patients born at term (11 of 22 [50%] versus 11 of 18 [61%]; P=0.537). Acquired neuropathological lesions occurring in premature cases and term cases are summarized in Table 3.

Analysis by Lesion Severity

Our cohort included aortic stenosis (n=2), coarctation of the aorta (n=2), interrupted aortic arch (n=1), and truncus arteriosus (n=4). Given that these lesions often present without cyanosis, we separated them for additional analysis. Of these 9 cases, 3 showed no acquired neuropathological lesions, 3 showed hemorrhages (germinal matrix or subarachnoid) only, 1 acute neuronal necrosis only, and the remaining 2 showed multiple acquired neuropathological lesions. Excluding these lesions from analysis, 17 of the 41 remaining cases (41%) show no acquired neuropathological lesions and 6 of 41 showed periventricular leukomalacia (14%). Excluding these 9 cases from analysis did not affect the results (not shown).

Discussion

Because of dramatic advancements in the medical and surgical management of infants born with congenital cardiac malformations, many of these patients are surviving into adulthood. Long‐term cognitive deficits have thus become the major source of morbidity in this patient population, leading to an increasing emphasis on improving neurological outcomes. In our study of inpatients with CHD coming to autopsy at two major pediatric cardiac referral centers, we found a high incidence of acquired neuropathological lesions, most commonly subarachnoid and germinal matrix hemorrhages and neuronal injury in the hippocampus. The hemorrhagic lesions may be attributable to the need for anticoagulation in many of these patients, although we were unable to address this question with the available data. Hippocampal injury is a common neuropathological correlate of severe hypoxic‐ischemic brain injury. Interestingly, however, the incidence of periventricular leukomalacia, traditionally seen as a hallmark of CHD neuropathology, was rare, observed in only 6% of patients overall and 12% of patients undergoing cardiac interventions.

The most significant difference between our cohort and historical controls is the decreased incidence of periventricular leukomalacia. This lesion, previously a neuropathological hallmark of CHD, is surprisingly rare in our cohort. The overall incidence of PVL in the current study was 6%, whereas earlier reports of the incidence of PVL from existing neuropathological studies of patients with CHD ranged from 11% to 61%.9, 10, 11 The highest reported rate of PVL was in Kinney et al,9, who only included patients undergoing intervention, and reported a 61% rate of PVL, compared with 12% in patients undergoing intervention in our cohort. The effect is smaller when comparing our entire cohort to previous studies including all patients, which report rates between 11% and 25%.10, 11, 14 This suggests that much of the improvement is attributable to technical improvements in operative and perioperative management of patients surviving longer and/or undergoing surgical or percutaneous interventions, with a residual rate of injury attributable to in utero and perinatal hypoxic‐ischemia injury. This is consistent with recent studies showing a significant burden of neurological injury in utero and preoperatively.15, 16, 17, 18, 19 Rates of hippocampal and olivary injury and white matter gliosis were also lower in our study compared with historical controls, but given that these lesions require microscopic sampling and are more subjective, these findings should be interpreted with caution. Rates of subarachnoid hemorrhage in the current cohort are consistent with those previously reported.10, 11 Table 4 summarizes the incidence of acquired neuropathological lesions among historical controls. This comparison is limited by differences in sampling between studies and the fact that some of the lesions observed in our cohort were not included in previous studies.

Table 4

Comparison of the Incidence of Acquired Neuropathologic Lesions With Historical Controls

Cohort	N	Patients	Years	PVL	WMG	ANN	SAH	GMH	None
Current study	26	Cardiac surgery only	2000–2017	12%	34%	46%	34%	25%	23%
Kinney et al, 20059	38	Cardiac surgery only	1985–1993	61%	79%	68%	NR	NR	0%
Current study	50	All patients	2000–2017	6%	24%	32%	24%	20%	40%
Mito et al, 199111	296	All patients	1981–1989	11%	NR	21%	15%	2%	NR
Glauser et al, 199010	40	All patients	1980–1985	25%	NR	28%	38%	NR	43%
Bozoky et al, 198414	45	All patients	1980–1984	18%	NR	29%	13%	NR	7%

ANN indicates acute neuronal necrosis; GMH, germinal matrix hemorrhage; NR, not reported; PVL, periventricular leukomalacia; SAH, subarachnoid hemorrhage; and WMG, white matter gliosis.

In the present cohort, there was a greater proportion of patients with acquired neuropathological lesions in cases undergoing repair or palliation compared with those who did not (77% versus 42%, respectively). Patients with chromosomal abnormalities were also less likely to have acquired neuropathological lesions than those without. Both the no‐intervention and chromosomal abnormality groups contained a significant number of stillborn patients. In our cohort, none of the stillborn patients had acquired neuropathology, which likely explains, at least in part, the lower incidence of lesions in the chromosomal abnormality and no‐intervention groups.

The present study is limited by the clinical information reported within the autopsy reports. Because of changes in medical record systems over the past 20 years at both institutions, it was not feasible to conduct a systematic review of the entire medical record, including operative notes and imaging studies. Although it was possible to reliably determine whether the patient had undergone intervention, the intraoperative methods used, including cardiopulmonary bypass and deep hypothermic circulatory arrest, were not reliably reported. Our cohort also includes a heterogenous mixture of various cardiac diagnoses and procedures.

The present cohort is also subject to unavoidable variations in autopsy practice. Because of the limitations in medical record review noted above, we were not able to calculate the autopsy rate in the course of our study, and similar autopsy rates are generally not reported in historical studies. However, one of the authors (M.M.H.) is familiar with the autopsy services at the 2 institutions in the current study (University of Iowa and Children's Hospital of Philadelphia) and has previously published work using the autopsy archives at Boston Children's Hospital, which were used for the Kinney et al study.9 As major academic centers, the clinical staff at all 3 institutions has, and continues to have, a strong commitment to obtaining autopsy consent for deaths attributed to CHD, making significant changes in autopsy rate in this population specifically unlikely. In addition, the most likely change would be a decrease in autopsy rates with an increasing emphasis on the sickest patients and those with the most severe lesions. This would artefactually increase the rate of neuropathological lesions, biasing our study toward the null result.

Furthermore, incidence rates of neuropathological lesions requiring microscopic diagnosis may be under‐represented because of sampling. In a significant proportion of cases, microscopic examination was not performed in specific brain regions that were deemed grossly normal. We also did not have access to cases dying outside the hospital, but this would tend to artefactually increase the rate of lesions, given that patients dying in the hospital are likely the most severely ill.

Given that the purpose of this study is to examine the pathology of CHD, it is, by definition, not fully representative of the CHD population as a whole, many of whom survive to adulthood. It is, however, representative of critically ill patients dying of complications of cyanotic CHD. Based on review of the inclusion criteria, demographics, and cardiac diagnoses of the previous studies noted in Table 4, it is representative of this population, enabling tentative conclusions about temporal trends in the pathology of CHD. It should be noted that Kinney et al only included patients undergoing intervention.9 To account for this, as noted above and in Table 4, we compared results in that publication with the subset of our patients undergoing intervention.

Our results suggest that the classic pathological lesions of CHD, particularly PVL, are becoming increasingly rare, presumably attributable to improvements in cardiac surgery, interventional cardiology, and critical care. We and others have identified similar trends in premature infants, where rates of germinal matrix hemorrhage and PVL have decreased dramatically over the past decades.20, 21 Although they speak to rapid technological improvements, these trends suggest that the burden of neurological disability in patients suffering from early‐life hypoxic‐ischemic injuries is increasingly attributable to more‐subtle lesions of neuronal connectivity and function, which are more difficult to detect on imaging or by histology and therefore pose a greater diagnostic and therapeutic challenge. Further progress in this area will require more‐detailed human and animal model studies to identify neuropathological substrates and therapeutic strategies for these subtler, but nevertheless life‐changing, injuries.

Sources of Funding

This work was funded, in part, by grants from the National Institutes of Health (UL1TR002537 and K23NS109284) and the Williams‐Cannon Foundation, all to Hefti.

Disclosures

None.