Magnetic Resonance Spectroscopy of Brain Metabolism in Fetuses With Congenital Heart Disease

Abstract

BACKGROUND

Congenital heart disease (CHD) remains a significant risk factor for neurologic injury because altered fetal hemodynamics may be unable to support typical brain development during critical periods of growth and maturation.

OBJECTIVES

The primary objective was to assess differences in the cerebral biochemical profile between healthy fetuses and fetuses with complex CHD and to relate these with infant outcomes.

METHODS

Pregnant participants underwent fetal magnetic resonance imaging with cerebral proton magnetic resonance spectroscopy acquisitions as part of a prospective observational study. Cerebral metabolites of N-acetyl aspartate, creatine, choline, myo-inositol, scyllo-inositol, lactate, and relevant ratios were quantified using LCModel.

RESULTS

We acquired 503 proton magnetic resonance spectroscopy images (controls = 333; CHD = 170) from 333 participants (controls = 221; CHD = 112). Mean choline levels were higher in CHD compared with controls (CHD 2.47 IU [Institutional Units] ± 0.44 and Controls 2.35 IU ± 0.45; P = 0.02), whereas N-acetyl aspartate:choline ratios were lower among CHD fetuses compared with controls (CHD 1.34 ± 0.40 IU vs controls 1.44 ± 0.48 IU; P = 0.001). Cerebral lactate was detected in all cohorts but increased in fetuses with transposition of the great arteries and single-ventricle CHD (median: 1.63 [IQR: 0.56–3.27] in transposition of the great arteries and median: 1.28 [IQR: 0–2.42] in single-ventricle CHD) compared with 2-ventricle CHD (median: 0.79 [IQR: 0–1.45]). Cerebral lactate also was associated with increased odds of death before discharge (OR: 1.75; P = 0.04).

CONCLUSIONS

CHD is associated with altered cerebral metabolites in utero, particularly in the third trimester period of pregnancy, which is characterized by exponential brain growth and maturation, and is associated with survival to hospital discharge. The long-term neurodevelopmental consequences of these findings warrant further study.

CENTRAL ILLUSTRATION

Congenital Heart Disease and Altered Fetal Brain Biochemistry Fetal congenital heart disease (CHD) poses a significant risk for neurologic injury, altering cerebral blood flow and oxygen delivery that may disrupt fetal brain metabolism. Using magnetic resonance spectroscopy to measure key cerebral metabolites, we found fetuses with CHD have increased choline and decreased N-acetyl-aspartate to choline ratios (A). Fetuses with transposition of the great arteries also have increased cerebral lactate (B). These dynamic measures may serve as important, early markers of altered brain development and could signal future impairment to neurodevelopmental outcomes. NAA = N-acetylaspartate; TGA = transpositiion of the great arteries.

Congenital heart disease (CHD) is the most common birth defect, and despite substantial advancements in survival, remains a significant risk factor for neurologic injury. Impaired cortical maturation, brain growth, cerebral perfusion, and decreased cerebral tissue oxygenation in fetuses with CHD compared with healthy fetuses suggest that disturbances in brain development begin in utero for fetuses with CHD.^1–12 Quantitative magnetic resonance imaging (MRI) studies have demonstrated that fetuses with CHD have impaired global and regional brain volumes detectable in utero.^1,5,10,11 In addition, advanced noninvasive quantitative MRI measures of regional brain oxygenation and perfusion have reported significant alterations in high-risk fetuses with CHD^5,6,11 and linked hypoxia and hypoperfusion with decreased brain growth.^5,11 Furthermore, decreases in brain volumes for fetuses with CHD are associated with decreased cognitive language, motor, and adaptive function in toddlers.¹² Presumably, disruptions in cerebral blood flow and oxygenation that ensue from altered cerebral circulation in CHD may fail to meet the metabolic demands necessary for optimal neurodevelopment. However, technical challenges in the detailed, noninvasive study of fetal brain metabolism have limited the investigations of CHD on early brain development to only a handful of reports.^1,7

Proton Magnetic Resonance Spectroscopy (¹H MRS) is a powerful tool that allows noninvasive measurement of metabolite concentrations in the fetal brain. Such studies enable detection and quantification of dynamic biochemical metabolites necessary for optimal growth of the developing brain. We and others have previously shown the evolution of several key cerebral metabolites across gestation in healthy fetuses.^13–15 Several metabolites, including N-acetylaspartate (NAA) (a marker of neuronal density), creatine (Cr) (a marker of energy-dependent systems), and choline (Cho) (a marker of cell membrane turnover), increase with advancing gestational age (GA), indicative of normal brain maturation.^13–15 In neonates with CHD, previous studies have reported decreased NAA:Cho ratios compared with healthy controls^16,17 as well as increased levels of lactate (Lac) and Lac:Cho ratios,^16–18 consistent with disrupted early brain biochemistry in CHD. Interestingly, reductions in thalamic and cerebellar volumes in infants with CHD also have been linked with altered cerebral biochemistry,¹⁹ whereas early ratios of cerebral Cho:Cr in infants with transposition of the great arteries (TGA) were positively associated with neurodevelopment at year 1,²⁰ emphasizing the important relationship between early brain metabolism, structure, and function. However, much less is known about in utero biochemical profiles for fetuses with CHD, particularly the onset and trajectory of key cerebral metabolites during periods of rapid brain development. The main objective of this study, therefore, was to bridge this knowledge gap by assessing differences in the biochemical profile between healthy fetuses and fetuses with complex CHD using serial measurements across the latter periods of gestation and to relate these with infant outcomes.

METHODS

PARTICIPANTS.

We prospectively enrolled pregnant patients for an observational study under a protocol approved by the Institutional Review Board at the Children’s National Hospital with informed written consent obtained from all study participants. Inclusion criteria comprised pregnant patients with singleton gestations diagnosed with complex CHD on fetal echocardiography, defined as lesions with anticipated corrective or palliative open-heart surgery during the neonatal period (cases). We also recruited patients with uncomplicated pregnancies, singleton gestation, and normal prenatal screening studies (controls). Exclusion criteria for both cases and controls included congenital infection, dysmorphic features, documented chromosomal abnormality, and multiple gestations. Patients with CHD are routinely counseled regarding additional prenatal genetic testing via amniocentesis or chorionic villous sampling, and all newborns with CHD receive additional genetic counseling and recommended testing postnatally.

CLINICAL DATA.

Clinical data, including maternal age and delivery outcomes, were extracted from the medical record. GA was determined based on last menstrual period and first trimester ultrasound. Delivery mode was categorized as unscheduled cesarean delivery vs normal delivery (defined as vaginal or scheduled cesarean delivery); resuscitation at delivery included the need for respiratory or cardiac resuscitation; and death included fetal demise, and preoperative or postoperative death before discharge. Pregnant patients carrying fetuses with CHD also underwent fetal echocardiogram including Doppler sonography during the study visit. Fetuses with CHD were subsequently categorized into the following classes based on fetal echocardiogram: CHD with: 1) 2 ventricles and no aortic obstruction (2V); 2) 2 ventricles and aortic obstruction (2VAO); 3) functional single ventricle and no aortic obstruction (SV); and 4) functional single ventricle and aortic obstruction (SVAO). Doppler measures included pulsatility indices (PI) of the middle cerebral artery (MCA PI), umbilical artery (UA PI), and the cerebroplacental ratio (MCA PI divided by the UA PI), as well as combined cardiac output measures.

MRI ACQUISITIONS.

All enrolled participants were scanned on a 1.5-T MR scanner (Discovery MR450, GE Healthcare) using an 8-channel surface coil (GE Healthcare) without sedation. Anatomical images were acquired in 3 planes using 2-dimensional single-shot fast spin-echo with the following: slice thickness: 2 mm; slice spacing: 0 mm; repetition time (TR): minimum; echo time (TE): 160 ms; flip angle: 90°; orientation: superior/inferior (S/I); number of slices: 42–65; matrix: 256 × 192.

Spectra were collected from a voxel that was placed in the central brain of the fetus. The voxels were placed using each of the 3 single-shot fast spin-echo images as guides to ensure the voxel was restricted within the brain in addition to being in the center of the brain in all 3 planes. All data were acquired with TE/TR = 144/1,500 ms using Point RESolved Spectroscopy localization sequence along with Chemical Shift Selective as the water suppression sequence for acquiring water suppressed spectra. Sixteen averages of unsuppressed water spectra along with 128 averages of water suppressed spectra were acquired from each participant.

To optimize ¹H-MRS data quality, previously validated techniques were used: selection of voxel dimension according to GA (25 × 25 × 25 mm³ for fetuses with GA ≤28 and 30 × 30 × 30 cm³ for fetuses with GA >28) to ensure the voxel was confined within the fetal brain and minimize contamination from external tissues; monitoring acquisitions to detect periods with excessive fetal motion; and real-time assessments of line width during automatic shimming with repeating anatomical reference scans and repositioning voxel placement as needed.^13,14 Representative voxel location and spectrum from the fetal brain with clear peaks from NAA, Cho, Cr and myo-inositol are shown in Figure 1.

FIGURE 1. Magnetic Resonance Imaging and Spectroscopy in Fetus With Congenital Heart Disease.

Magnetic resonance imaging of fetus with congenital heart disease, with representative voxel placement for proton magnetic resonance spectroscopy (¹H-MRS) measure centered in the fetal brain, along with the corresponding ¹H-MRS output using LCModel. Representative ¹H-MRS spectrum also denotes the resonance frequency of key cerebral metabolites, including myo-inositol, choline, creatine, N-acetylaspartate (NAA), and the inverted lactate doublet.

SPECTRAL ANALYSES.

We performed frequency and phase corrections using programs written in MAT-LAB before quantification in “LCModel” analysis software, in part to mitigate effects of motion (Supplemental Figure 1).^14,21 Metabolite concentrations were estimated in LCModel using water spectrum as an internal reference and the concentrations were reported in “institutional units.” LCModel basis set included the following 15 metabolites: alanine, aspartate, Cr, glutamine, glutamate, glycerophosphocholine, glutathione, Lac, myo-inositol (Ins), NAA, N-acetylaspartylglutamate, phosphocholine, phosphocreatine, scyllo-Inositol (sI), and taurine. Macromolecules and lipid spectra available in LCModel were used to account for the macromolecule baseline in the fitting routine. Data with Cramer Rao Lower Bound >20% for total Cho: glycerophosphocholine + phosphocholine (Cho) were excluded from further analysis; for all other metabolites, the exclusion criteria included Cramer Rao Lower Bound >100% to avoid biasing the results to higher metabolite concentrations.^22,23 All resulting data underwent visual inspection by a single investigator, blinded to case status, as a final step as part of quality control.

STATISTICAL ANALYSES.

Statistical methods used linear mixed effects (LME) modeling framework to analyze the conditional mean of metabolites while accounting for the clustered data structure of multiple scans per subject (fetus). However, median mixed effects models were used to analyze Lac levels, given non-normal distribution.

Group differences in mean metabolite concentrations between healthy fetuses and fetuses with CHD in mean or median metabolite concentrations were evaluated by including a group term in LME models while controlling for GA at the time of the assessment. Then the comparisons between healthy fetuses and fetuses with CHD were further stratified to determine differences in cerebral metabolites in the second and third trimesters (GA ≤28 weeks and GA >28 weeks).

Fetuses with CHD were further analyzed to compare the mean or median metabolite levels among 3 subgroups: 2V disease (class 1 and 2, excluding TGA), SV (class 3 and 4), and TGA, with pairwise multiple comparisons. We also explored the relationships between each metabolite and individual Doppler sonography measures in separate models. All analyses accounted for GA at MRI.

As part of the study, subjects had repeated measurement of metabolites at different GA. To analyze the impact of specific metabolites on pregnancy and infant outcome, we considered the latest metabolite measurement for each subject. We used logistic regression to evaluate the relationship between fetal metabolite measurements and binary outcomes, such as death, resuscitation at delivery, and unplanned cesarian section, accounting for GA at MRI.

Hypothesis tests were 2-tailed; P value of 0.05 or lower was considered significant. P values and 95% CIs presented have not been adjusted for multiplicity, and, therefore, inferences drawn from these statistics may not be reproducible. All analyses were conducted using software R version 4.05 (R Foundation for Statistical Computing) and LME models were implemented using the lme4 package version 1.1–26.²⁴

RESULTS

BASELINE CHARACTERISTICS OF THE COHORT.

Of the 333 unique pregnant participants, 221 had healthy fetuses and 112 had fetuses diagnosed with CHD. Participants were scanned during the second and third trimester of pregnancy with GA ranging from 18 to 39 weeks. Approximately one-half of the cohort underwent 2 fetal MRI studies at least 4 weeks apart; 113 (51.1%) controls and 58 (51.7%) CHD cases underwent 2 fetal MRI scans. No participant had structural or destructive injury noted on qualitative evaluation of fetal brain MRI. Twenty-eight neonates with CHD died before discharge (Supplementa1 Table 1). Additional characteristics of the study cohort are shown in Table 1 and Supplemental Figure 2. Of the 112 participants with CHD, 20 (18%) underwent additional prenatal genetic testing via amniocentesis or chorionic villous sampling and 76 (68%) underwent additional testing postnatally (Supplemental Table 2).

TABLE 1.

Maternal-Fetal Cohort Characteristics

	Controls (n = 221)	CHD (n = 112)	P Value
MRI studies, n	333	172
Maternal age, y	30.57 ± 6.88	32.03 ± 5.78	0.04
Gestational age at MRI, wk – all scans	31.52 ± 5.14	31.92 ± 4.17	0.35
Gestational age at MRI, wk – second trimester	24.13 ± 2.55	25.07 ± 1.32	0.11
Gestational age at MRI, wk – third trimester	33.49 ± 3.69	32.97 ± 3.27	0.16
Male fetus	112 (50.9)	66 (59.5)	0.18
Gestational age at delivery, wka	39.38 ± 1.45	38.50 ± 1.42	<0.01
Birth weight, gb	3,302 ± 461	3,093 ± 581	<0.01
Birth weight, z scoreb	−0.17 ± 0.84	−0.34 ± 1.14	0.14
Apgar score, 1 minc	8.13 ± 1.23	7.33 ± 1.59	<0.01
Apgar score, 5 minc	8.82 ± 0.89	8.27 ± 1.04	<0.01
Cardiorespiratory resuscitationd	10 (4.5)	57 (50.9)	<0.01
Death before dischargee	0 (0.0)	28 (25.0)	<0.01
CHD Diagnostic Class
Class 1: 2 ventricles		54 (48.2)
Class 2: 2 ventricles with arch obstruction		11 (9.8)
Class 3: single ventricle		21 (18.8)
Class 4: single ventricle with arch obstruction		26 (23.2)

Values are n, mean ± SD, or n (%).

^aData available for 314 (94%) subjects.

^bData available for 303 (91%) subjects.

^cData available for 276 (83%) subjects.

^dData available for 329(99%) subjects.

^eData available for 295 (89%) subjects.

CHD = congenital heart disease; MRI = magnetic resonance imaging.

FETAL BRAIN BIOCHEMISTRY IN FETUSES WITH CHD VS HEALTHY FETUSES.

Detectable levels of NAA, Cho, and Cr were quantified in all studies. For the remaining metabolites, Ins was quantified in 503 of 505 acquisitions (99%), sI (scyllo-inositol) in 496 (98%) of ¹H MRS acquisitions, and Lac was quantified in 372 (74% overall and 74.9% of controls vs 71.3% of CHD cases; P = 0.40). Comparison of brain metabolite differences between the 2 cohorts showed significantly higher Cho and lower NAA:Ch ratio in CHD fetuses compared with the healthy fetuses from low-risk pregnancies. We did not observe significant differences in any other metabolites between the 2 cohorts. Table 2 shows least square means estimates from LME models of key metabolites, controlling for GA. We also compared metabolite profiles in the second and third trimesters of pregnancy. We did not observe any significant difference in metabolite levels between the 2 cohorts during the second trimester. However, our results showed significantly higher levels of Cho (P = 0.008) and lower levels of NAA:Cho (P < 0.001) in fetuses with CHD compared with healthy fetuses during the third trimester (Table 3). We also did not observe any significant difference in metabolite levels relative to maternal age, however there was an inverse relationship with Cr and maternal age in fetuses with CHD (Supplemental Table 3).

TABLE 2.

Cerebral Metabolites in Fetuses With and Without CHD

	Controls (n = 333)a	CHD (n = 170)a	P Valueb
Myo-inositol, IU	13.7 ± 3.95	13.3 ± 4.14	0.37
Lactate, IU	1.24 (0.06–2.08)	1.25 (0–2.39)	0.16
Scyllo-inositol, IU	0.37 ± 0.13	0.38 ± 0.14	0.47
Cho, IUc	2.35 ± 0.45	2.47 ± 0.44	0.02
NAA, IUc	3.37 ± 1.27	3.34 ± 1.21	0.25
Cr, IUc	2.85 ± 0.76	2.87 ± 0.89	0.76
NAA/Cho, IUc	1.44 ± 0.48	1.34 ± 0.40	<0.01

Values are mean ± SD or median (IQR).

^aNumber of ¹H-MRS scans.

^bP values obtained from linear (or median) mixed models to account for GA and multilevel data structure; no corrections for multiple testing were applied.

^cTotal metabolites (Cho = glycerophosphocholine + phosphocholine; Cr = creatine + phosphocreatine; NAA = N-acetylaspartate + N-acetylaspartylglutamate).

GA = gestational age; IU = institutional units; NAA = N-acetylaspartate; other abbreviations as in Table 1.

TABLE 3.

Cerebral Metabolites in Fetuses With and Without CHD in Second and Third Trimesters

	Second Trimester (GA 13–28 wk)			Third Trimester (GA 29–40 wk)
	Controls (n = 97)a	CHD (n = 30)a	P Valueb	Controls (n = 237)a	CHD (n = 140)a	P Valueb
Myo-inositol, IU	14.19 ± 4.92	13.87 ± 4.67	0.70	13.43 ± 3.47	13.15 ± 4.03	0.50
Lactate, IU	1.30 (0.69–2.50)	1.25 (0.09–2.88)	0.29	1.14 (0.00–1.94)	1.25 (0.00–2.09)	0.18
Scyllo-inositol, IU	0.31 ± 0.13	0.33 ± 0.14	0.82	0.39 ± 0.13	0.39 ± 0.14	0.46
Cho, IUc	2.29 ± 0.51	2.38 ± 0.54	0.53	2.38 ± 0.43	2.49 ± 0.42	0.01
NAA, IUc	2.26 ± 0.80	2.43 ± 1.00	0.65	3.82 ± 1.14	3.54 ± 1.05	0.16
Cr, IUc	2.21 ± 0.63	2.24 ± 1.18	0.83	3.10 ± 0.66	3.00 ± 076	0.91
NAA/Cho, IUc	1.00 ± 0.29	1.04 ± 0.33	0.94	1.62 ± 0.42	1.40 ± 0.39	<0.01

Values are mean ± SD or median (IQR).

^aNumber of ¹H-MRS scans.

^bP values obtained from linear (or median) mixed models to account for GA and multilevel data structure; no corrections for multiple testing were applied.

^cTotal metabolites (Cho = glycerophosphocholine + phosphocholine; NAA = N-acetylaspartate + N-acetylaspartylglutamate; Cr = creatine + phosphocreatine).

Abbreviations as in Tables 1 and 2.

RELATIONSHIP BETWEEN CHD DIAGNOSTIC GROUPS AND FETAL BRAIN BIOCHEMISTRY.

We compared the metabolic profiles in fetuses within distinct CHD diagnostic groups, based on cardiac physiology. In this analysis, we compared fetuses with 2V (Class 1 and 2, excluding TGA), SV (Class 3 and 4), and TGA, given that despite having 2 ventricles, TGA physiology represents a severe form of CHD with significant hypoxemia. Fetuses with TGA had significantly higher Lac followed by fetuses with SV physiology. Fetuses with 2V physiology (excluding TGA) had the lowest levels of Lac among the 3 groups. Pairwise comparison of metabolites among these groups showed significantly different Lac levels between: 1) 2V (excluding TGA) and SV fetuses (P = 0.014); and 2) TGA and fetuses with SV physiology (P = 0.014) (Table 4, Supplemental Table 4). We did not detect any significant differences in metabolite levels based on the presence or absence of AO in fetuses with CHD.

TABLE 4.

Cerebral Metabolites Between CHD Diagnosis Groups

	Adjusted Meansa				Pairwise Comparisonb
	CHD Class 1 and 2 (n = 58)c	CHD Class 3 and 4 (n = 77)c	TGA (n = 33)c	P Valued	Class 1–2 vs Class 3–4	Class 1–2 vs TGA	Class 3–4 vs TGA
Myo-inositol, IU	12.68 ± 0.57	13.41 ± 0.50	13.70 ± 0.77	0.49	0.74 (0.50)	1.02 (0.50)	0.29 (0.76)
Lactate, IU	0.64 ± 0.18	1.35 ± 0.18	1.76 ± 0.40	<0.01	0.71 (0.01)	1.12 (0.01)	−0.42 (0.34)
Scyllo-inositol, IU	0.36 ± 0.02	0.38 ± 0.02	0.41 ± 0.02	0.31	0.02 (0.40)	0.05 (0.40)	0.03 (0.40)
Cho, IUe	2.39 ± 0.06	2.47 ± 0.05	2.58 ± 0.08	0.17	0.08 (0.33)	0.19 (0.19)	0.11 (0.33)
NAA, IUe	3.20 ± 0.13	3.34 ± 0.11	3.56 ± 0.17	0.23	0.14 (0.40)	0.36 (0.26)	0.22 (0.40)
Cr, IUe	2.80 ± 0.11	2.94 ± 0.09	2.79 ± 0.14	0.52	0.14 (0.57)	−0.01 (0.96)	−0.15 (0.57)
NAA/Cho, IUe	1.33 ± 0.04	1.36 ± 0.04	1.36 ± 0.06	0.86	0.03 (0.97)	0.03 (0.97)	0.00 (0.97)

Values are mean ± SD or median (IQR).

^aData are mixed models, accounting for GA.

^bPost hoc pairwise comparisons of adjusted mean difference (P value), corrected for multiple tests using Benjamini-Hochberg method.

^cNumber of ¹H-MRS scans.

^dOverall P value to assess differences across CHD groups obtained from linear (or median) mixed models to account for GA and multilevel data structure.

^eTotal metabolites (Cho = glycerophosphocholine + phosphocholine; NAA = N-acetylaspartate + N-acetylaspartylglutamate; Cr = creatine + phosphocreatine).

NAA:Cho = N-acetylaspartate to choline ratio; TGA = transposition of the great arteries; other abbreviations as in Tables 1 and 2.

RELATIONSHIP BETWEEN CEREBRAL BIOCHEMICAL PROFILES AND DOPPLER SONOGRAPHY IN FETUSES WITH CHD.

The MCA PI from fetal Doppler studies had a significant negative correlation with cerebral Ins levels (standardized coefficient [β] = −0.20; P < 0.01). The UA PI trended with cerebral Lac (β = −0.10; P = 0.07), NAA (β = −1.84; P = 0.07), and NAA:Ch (β = −0.11; P = 0.05). We did not detect any significant association between fetal brain metabolites and the cerebroplacental ratio (CPR) or fetal cardiac output measures (Table 5).

TABLE 5.

Cerebral Metabolites and Doppler Sonography Measures for Fetuses With Congenital Heart Disease

	MCA PI		UA PI		CPR
	Standardized Coefficient (95% CI)	P Value	Standardized Coefficient (95% CI)	P Value	Standardized Coefficient (95% CI)	P Value
Myo-inositol	−0.20 (−0.35 to −0.06)	0.01	−0.09 (−0.25 to 0.07)	0.25	−0.07 (−0.20 to 0.06)	0.27
Lactatea	−0.08 (−0.22 to 0.07)	0.30	−0.10 (−0.22 to 0.01)	0.07	0.02 (−0.11 to 0.15)	0.80
Scyllo-inositol	0.04 (−0.09 to 0.18)	0.55	−0.06 (−0.2 to 0.08)	0.39	0.02 (−0.09 to 0.14)	0.70
Choa	−0.05 (−0.19 to 0.08)	0.43	−0.07 (−0.21 to 0.08)	0.35	−0.06 (−0.18 to 0.05)	0.26
NAAa	−0.09 (−0.22 to 0.04)	0.17	−0.05 (−0.19 to 0.09)	0.44	−0.07 (−0.18 to 0.04)	0.20
Cra	−0.07 (−0.17 to 0.04)	0.22	−0.10 (−0.22 to 0.01)	0.07	−0.03 (−0.12 to 0.06)	0.46
NAA/Choa	−0.09 (−0.19 to 0.02)	0.11	−0.11 (−0.22 to 0.00)	0.05	−0.01 (−0.10 to 0.08)	0.84

Data account for GA and multilevel data structure; no corrections for multiple testing were applied.

^aTotal metabolites (Cho = glycerophosphocholine + phosphocholine; NAA = N-acetylaspartate + N-acetylaspartylglutamate; Cr = creatine + phosphocreatine).

CPR = cerebral-placental ratio; MCA = middle cerebral artery; PI = pulsatility index; UA = umbilical artery; other abbreviations as in Tables 1, 2, and 4.

RELATIONSHIP BETWEEN FETAL BIOCHEMICAL PROFILES, PREGNANCY, AND INFANT OUTCOMES.

Fetuses with detectable cerebral Lac were 1.75 times more likely to die before discharge compared with those without (OR: 1.75; P = 0.04). Prenatal cerebral Lac also trended with an increased likelihood of cardiorespiratory resuscitation at delivery (OR: 1.41; P = 0.06). Other key metabolites were not otherwise associated with delivery mode, resuscitation at delivery, or survival (Table 6).

TABLE 6.

Fetal Cerebral Metabolites and Birth Outcomes

	Delivery Mode		Resuscitation at Delivery		Death
	OR	P Valuea	OR	P Valuea	OR	P Valuea
Myo-inositol	0.99	0.94	0.97	0.49	0.92	0.24
Lactate	0.65	0.34	1.41	0.06	1.75	0.04
Scyllo-inositol	0.13	0.51	0.32	0.45	42.98	0.11
Chob	0.37	0.77	0.51	0.59	0.57	0.78
NAAb	4.14	0.56	1.39	0.68	2.02	0.62
Crb	1.33	0.51	1.04	0.90	0.37	0.09
NAA/Chob	0.04	0.60	0.13	0.33	0.07	0.46

Data adjusted for GA at MRI, indicating relative change in odds for birth outcomes per unit change in metabolite concentration (IU) and corresponding P value.

^aNo corrections for multiple testing were applied.

^bTotal metabolites (Cho = glycerophosphocholine + phosphocholine; NAA = N-acetylaspartate + N-acetylaspartylglutamate; Cr = creatine + phosphocreatine).

Abbreviations as in Table 2.

DISCUSSION

In this prospective observational study using serial fetal ¹H MRS assessments, we demonstrated higher levels of prenatal Cho and lower NAA:Cho ratios in the CHD cohort compared with controls. These differences were not observed in the second trimester of pregnancy, but emerged in the third trimester, a period of exponential brain development and growth. Our study also showed increased levels of cerebral Lac in fetuses with TGA and those with SV compared with those with 2V physiology. The presence of elevated fetal cerebral Lac also was significantly associated with death before discharge. We further demonstrate a negative correlation between MCA PI and cerebral Ins in fetuses with CHD, linking cerebral hemodynamics with metabolic profiles (Central Illustration).

CEREBRAL METABOLITES AS MARKERS OF CELLULAR DEVELOPMENT.

Cho, a marker of cell membrane turnover, has been noted to be elevated in early neurodevelopment, then it decreases during overlapping periods of brain myelination and maturation.^25,26 In this work, elevated levels of cerebral Cho are consistent with delays in cerebral development that have been described in both fetuses and neonates with complex CHD.²⁷ Conversely, NAA increases with advancing GA but decreases with brain injury, so that the NAA:Cho ratio can detect deviations of both neuronal development and myelination.¹ Previous studies of newborns with CHD also have demonstrated decreased NAA:Cho ratios compared with healthy controls.^16,17,28 Similar patterns have been described late in pregnancy for fetuses with CHD, specifically between 36 and 38 weeks of gestation.⁷ We have previously shown that cerebral NAA:Cho ratios increase during the latter one-half of gestation,^1,14 and that rate of increase was significantly lower in a smaller cohort of 36 fetuses with CHD.¹ In this larger cohort of subjects, we now demonstrate that these key metabolic differences emerge specifically in the third trimester. This coincides with a period of exponential growth of the brain that reflects the highly dynamic and metabolically active periods of cortical organization and myelination. The rapidly developing fetal brain is known to be vulnerable to decreased substrate delivery.²⁹

CEREBRAL METABOLITES AS MARKERS OF OXYGENATION AND METABOLISM.

One of the hallmarks of complex CHD is the impact of chronic hypoxemia on early brain development.^{2,5,6,28,30,31} This prolonged period of hypoxemia begins in utero and triggers a number of compensatory mechanisms, including changes in cerebral autoregulation, as well as metabolic changes at a cellular level.^30,32–34 However, the chronicity of hypoxemia associated with CHD, unlike acute hypoxia-ischemia, may overwhelm these compensatory mechanisms, leading to a characteristic pattern of brain injury and maldevelopment.^28,31 Lac, a marker of anaerobic metabolism, is notably elevated after acute hypoxia-ischemia in both newborns and adults.^35,36 In the fetus, elevated cerebral Lac also has been reported in several high-risk conditions, including CHD.^1,37 However, cerebral Lac also can be detected in the healthy fetus.¹⁴ In this work, cerebral Lac was detected in healthy fetuses as well as fetuses with CHD, however, our findings show higher levels in fetuses with TGA. We also demonstrated a significant association between elevated levels of fetal cerebral Lac and death before discharge. These data suggest that complex CHD lesions, such as TGA, may have a greater reliance on anaerobic pathways to support metabolism compared with 2V CHD and healthy controls but also may reflect diminished capacity to autoregulate cerebral perfusion and oxygen delivery when Lac levels increase. The optimal cutoffs of cerebral Lac that may distinguish normal from abnormal physiology, however, remain to be determined, and may vary based on the specific fetal CHD diagnosis.

CEREBRAL METABOLITES AND BLOOD FLOW.

We describe an inverse relationship between cerebral Ins and MCA PI values. Ins, well described in brain spectroscopy, is considered a glial marker because it is predominantly found in astrocytes.³⁷ Increased levels of Ins are thought to occur with proliferation of glial cells, as well as inflammatory conditions, including chronic hypoxemia and white matter disease.^37–39 Decreases in the MCA PI are thought to result from periods of low-oxygen delivery, to increase cerebral blood flow, and are independently associated with adverse pregnancy outcomes.^32,33,40 In CHD, Doppler images of the MCA have shown decreased MCA PI values in fetuses with obstructive lesions, particularly hypoplastic left heart syndrome, as well as fetuses with TGA.^32,33 Thus, we hypothesize that the relationship between increasing Ins and decreasing MCA PI values may reflect inflammatory changes associated with chronic hypoxia. However, it is important to note that alterations in the MCA PI, and the reactivity of the MCA, are highly variable based on specific CHD lesions, highlighting the limitations of Doppler studies alone to reliably assess cerebral metabolism.^32,41

CEREBRAL METABOLITES AND NEURODEVELOPMENTAL OUTCOMES.

The application of ¹H MRS to measure cerebral metabolism in several high-risk populations has revealed significant relationships between early biochemical profiles and later neurodevelopment.^42–44 Early reports suggest ¹H MRS may provide promising biomarkers that predict neurodevelopmental outcomes^7,20; in 1 study of 10 infants with TGA, parietal white matter measures of Cho and Cr were significantly higher compared with healthy newborns, remained elevated at year 1, and were associated with psychomotor and cognitive delays.²⁰ Another study of 17 infants with CHD and 46 healthy controls noted that late third trimester measures of frontal NAA:Ch ratios were positively correlated with neurodevelopmental scores at 4 to 6 months of age.⁷ These data reinforce the notion that early neurodevelopment is characterized by critical windows of cellular maturation that form the foundation of later neurodevelopmental potential and highlight specific windows of vulnerability in the evolving brain.

STUDY LIMITATIONS.

As with many studies of the fetal brain, motion-related artifacts remain a significant challenge. Faster acquisition times and postprocessing methods can alleviate some but not all these challenges. It is also known that absolute values of brain metabolites vary based on the GA and brain region studied.⁴⁵ In the fetus, voxel size relative to the fetal brain inherently includes varying proportions of gray matter, white matter, and cerebrospinal fluid. Further advances in the technical acquisition and quantification of fetal brain metabolism are warranted for tissue-specific analyses. Studies of CHD often are confounded by the variability in the types of structural malformations and subsequent impact on hemodynamics.⁴⁶ Differences in systemic oxygenation, cardiac output, pulsatility, and cerebral blood flow converge to result in significant variations in cerebral perfusion, oxygenation, and metabolism, based on individual lesions. Although our cohort of fetuses with CHD enabled us to stratify fetuses into functional diagnostic groups (SV vs 2V physiology vs TGA), the deviations in the metabolic profiles reported might not be representative of individual lesions within CHD. Despite the large cohort of CHD fetuses described, the heterogeneity of the population studied was not adequately powered to evaluate individual CHD lesions. Last, the functional significance of these findings on long-term neurodevelopment are uncertain; these studies are ongoing.

CONCLUSIONS

In this study, we describe the emergence of altered NAA and Cho trajectories in fetuses with CHD during the third trimester. These findings underscore the fact that the third trimester is a period characterized by exponential brain maturation, suggesting that fetuses with CHD may be unable to adequately meet the increasing metabolic demands of the developing brain. NAA and Cho, key metabolites representative of the necessary scaffolding that support later neurodevelopment, are disrupted so that without sufficient cellular development, functional connections may not develop optimally. Elevations of cerebral Lac in fetuses with TGA further suggest that cerebral metabolism is determined by both cerebral oxygenation and perfusion, whereas elevated Ins may reflect inflammatory changes resulting from chronic hypoxemia. The hemodynamic changes that result from CHD, along with disrupted oxygenation and nutrient supply, may hinder typical neurodevelopment, and begin during gestation. Additional studies are warranted to identify the best predictor of both short-term survival as well as long-term morbidity. Advancing neurodevelopmental outcomes for survivors of CHD will require a better understanding of the disruptions in brain development beginning in the intrauterine period. Fetal ¹H MRS can provide the necessary insights to identify optimal periods and assess the efficacy of future intrauterine interventions.

PERSPECTIVES.

COMPETENCY IN MEDICAL KNOWLEDGE:

Altered hemodynamics disrupt early brain development in fetuses with complex CHD, compromising long-term neurodevelopment.

TRANSLATIONAL OUTLOOK:

Although measures of fetal brain Lac are associated with decreased survival, longer-term follow-up is needed to more completely understand the impact of disrupted cerebral metabolism on long-term neurodevelopment.

ABBREVIATIONS AND ACRONYMS

¹H MRS

proton magnetic resonance spectroscopy

2V(AO)

2-ventricle physiology (with aortic obstruction)

CHD

congenital heart disease

Cho

choline

Cr

creatine

GA

gestational age

Ins

myo-inositol

Lac

lactate

LME

linear mixed effects

MCA

middle cerebral artery

MRI

magnetic resonance imaging

NAA

N-acetylaspartate

PI

pulsatility indices

sI

scyllo-Inositol

SV(AO)

single-ventricle physiology (with aortic obstruction)

TE

echo time

TGA

transposition of the great arteries

TR

repetition time

UA

umbilical artery