Objective
We sought to determine the relationship between fetal brain metabolism and microstructure expressed by brain sulcation, and corpus callosum (CC) development assessed by fetal brain magnetic resonance (MR) imaging and proton MR spectroscopy ( 1 H-MRS).
Study Design
A total of 119 fetuses, 64 that were small for gestational age (estimated fetal weight <10th centile and normal umbilical artery Doppler) and 55 controls underwent a 3T MR imaging/ 1 H-MRS exam at 37 weeks. Anatomical T2-weighted images were obtained in the 3 orthogonal planes and long echo time (TE) 1 H-MRS acquired from the frontal lobe. Head biometrics, cortical fissure depths (insula, Sylvian, parietooccipital, cingulate, and calcarine), and CC area and biometries were blindly performed by manual and semiautomated delineation using Analyze software and corrected creating ratios for biparietal diameter and frontooccipital diameter, respectively, for group comparison. Spectroscopic data were processed using LCModel software and analyzed as metabolic ratios of N-acetylaspartate (NAA) to choline (Cho), Cho to creatine (Cr), and myo-inositol (Ino) to Cho. Differences between cases and controls were assessed. To test for the association between metabolic ratios and microstructural parameters, bivariate correlation analyses were performed.
Results
Spectroscopic findings showed decreased NAA/Cho and increased Cho/Cr ratios in small fetuses. They also presented smaller head biometrics, shorter and smaller CC, and greater insular and cingulate depths. Frontal lobe NAA/Cho significantly correlated with biparietal diameter (r = 0.268; P = .021), head circumference (r = 0.259; P = .026), CC length (r = 0.265; P = .026), CC area (r = 0.317; P = .007), and the area of 6 from the 7 CC subdivisions. It did not correlate with any of the cortical sulcation parameters evaluated. None of the other metabolic ratios presented significant correlations with cortical development or CC parameters.
Conclusion
Frontal lobe NAA/Cho levels–which are considered a surrogate marker of neuronal activity–show a strong association with CC development. These results suggest that both metabolic and callosal alterations may be part of the same process of impaired brain development associated with intrauterine growth restriction.
Intrauterine growth restriction (IUGR) affects 5-10% of all pregnancies and is a leading cause of fetal morbidity and mortality. Potential sustained exposure to hypoxemia and undernutrition may have profound consequences on the developing brain in early and severe IUGR. Short- and long-term cognitive dysfunctions have been widely reported in this condition. This effect has been coined as “fetal brain programming.”
Recently, the focus of attention has shifted from early- to late-onset IUGR. This mainly corresponds to its higher prevalence and the preexisting gap of knowledge, as it was assumed to be a form of constitutional smallness. Currently, a growing body of evidence agrees that a proportion of late-onset small fetuses present signs of brain programming in association with an impaired cognitive function during infancy. Considering that the majority of small children are born near term, the impact of this form in the risk of neurodevelopmental disabilities can hardly be overestimated. This stresses the need to improve the understanding of the processes underlying brain reorganization in small fetuses and to develop biomarkers to detect brain changes at an individual level.
One of the main constrains for these purposes is that the pathophysiological basis in growth restriction is still poorly characterized. We have shown how late-onset small fetuses present significant changes in their brain metabolism and microstructure. About the first, proton magnetic resonance (MR) spectroscopy ( 1 H-MRS) is a noninvasive tool for assessing tissue metabolic profiles in vivo, which can be expressed as metabolic ratios such as N-acetylaspartate (NAA)/choline (Cho), myo-inositol (Ino)/Cho, and Cho/creatine (Cr). NAA is considered a neuronal marker also localized in oligodendrocytes. Total choline containing compounds are essential for cell membrane turnover. Total creatine (Cr) is associated with tissue bioenergetics.
Regarding brain microstructure, some of the most relevant findings in small fetuses were detected in the corpus callosum (CC) and cortical development. CC is considered a surrogate marker of white matter (WM) development and has been proposed as a sensitive indicator of normal brain development and maturation. Additionally, brain sulcation is used as a reliable estimate of gestational age and fetal cortical maturation.
The relationship between alterations in brain metabolism and microstructure has not been explored yet. This information could unveil pathophysiological changes in IUGR conditions that may be helpful in understanding the form of brain damage that these fetuses are exposed to. This information seems critical to develop targeted interventions.
The aim of this study is to determine if changes in fetal brain metabolite ratios are associated with changes in fetal brain biometrics, cortical sulcation, and CC structure.
Materials and Methods
Study cohort
This study is part of a larger prospective research program on IUGR involving fetal, and short- and long-term postnatal, follow-up. The specific protocol of this study was approved by the institutional ethics committee (institutional review board 2008/4422) and all participants gave written informed consent.
A prospective cohort of 119 consecutive singleton pregnancies was included, 64 small fetuses and 55 appropriate for gestational age (AGA). All small fetuses had an estimated and confirmed birthweight below the 10th centile according to local standards and normal umbilical artery pulsatility index (<95th centile). AGA fetuses were considered those with an estimated and confirmed birthweight above the 10th centile, which were selected among normal pregnancies and followed up at our institution after accepting to participate in the study.
Cases with congenital malformations, chromosomal abnormalities, perinatal infections, chronic maternal pathology including diabetes, chronic hypertension, autoimmune and other systemic diseases, and noncephalic presentations were considered noneligible for this study.
Clinical and ultrasound data
Small fetuses were followed up in our fetal growth restriction unit from diagnosis until delivery. Gestational age was corrected from fetal crown-rump length in the first trimester. Prenatal Doppler ultrasound examinations were performed (6-2-MHz linear-curved-array transducer, Sonoline Antares; Siemens, Erlangen, Germany) within 1 week from the MR imaging (MRI) scan. All fetuses in this study had at least 1 scan within 1 week of delivery. Maternal and perinatal data were prospectively recorded in all study patients.
Fetal MR acquisition
All cases were scanned at 37 weeks of gestation in a TIM TRIO 3.0-T scanner (Siemens) without sedation following the American College of Radiology guidelines for use of medical imaging during pregnancy and lactation. A receiver radiofrequency coil with 8 phased-array elements was wrapped around the mother’s abdomen, as near as possible to the fetus. The total length of each MRI exam did not exceed 45 minutes. T2-weighted anatomical images in the 3 orthogonal planes were acquired with a single-shot fast spin-echo (half-Fourier acquisition single-shot turbo spin-echo [HASTE]) sequences, using the following parameters: 990-millisecond repetition time, 137-millisecond TE, 3.5-mm slice thickness, 260-mm isotropic field of view, voxel size 1.4×1.4×3.5 mm, in plane resolution 192 × 192, 80-degree flip angle, and 24-second acquisition time. Single voxel point-resolved spectroscopy 1 H-MRS data were acquired from a 40×20×20-mm 3 voxel positioned in the frontal lobe region according to the T2 HASTE images ( Figure 1 ), and using the following parameters: 2000-millisecond repetition time, 145-millisecond TE, 98 averages, CHESS (chemical shift selective) water suppression, and an acquisition time of 3 minutes, 24 seconds. A second anatomical T2 HASTE image was obtained after each 1 H-MRS acquisition to confirm the position of the fetal head. If the obtained spectrum was distorted by fetal movements, it was repeated.
Structural MRI were reviewed for the presence of anatomical abnormalities by an experienced neuroradiologist blinded to group status.
1 H-MRS postprocessing
MR spectra were processed using the user independent frequency domain-fitting program Linear Combination Model-Fitting (LCModel; S. Provencher Inc., http://s-provencher.com/pages/lcmodel.shtml ), version 6.1-4A. The basis set used included a total of 12 metabolites: alanine, creatine and phosphocreatine, glutamine, glutamate, phosphocholine and glycerophosphocholine, myo-inositol (Ino), lactate, N-acetylaspartate and N-acetylaspartylglutamate, and scyllo-inositol. As certain metabolites are difficult to resolve from others, analyzing them together (sum) becomes more accurate than their individual concentrations. We studied the sums of: N-acetylaspartate + N-acetylaspartylglutamate, referred to as “total NAA” (NAA); glycerophosphocholine + phosphocholine, referred to as “total Cho” (Cho); and creatine + phosphocreatine, referred to as “total Cr” (Cr). Also, since the glycine peak overlaps with myo-inositol in the same spectral region (3.55 ppm), there is a mixed contribution of both metabolites to what is processed as Ino. Results were expressed as metabolic ratios: NAA/Cho, Ino/Cho, and Cho/Cr.
A quality control was carried out to select spectra for further analysis: signal-to-noise ratio had to be >5, the coefficient of variation for the estimated concentrations (Cramer-Rao lower bounds) had to be <20% (which indicates that these metabolites could be reliably estimated ), the estimated full width at half maximum of the metabolite peaks <0.095 ppm, and main metabolic peaks had to be clearly identified visually.
Head biometries, cortical development, and CC assessment
Offline analyses of head biometries, cortical development, and CC morphometry measurements were performed using the semiautomatic Analyze 9.0 software (Biomedical Imaging Resource; Mayo Clinic, Rochester, MN). To avoid the bias of having smaller measurements in the heads of small fetuses, cortical development and CC measurements were corrected by the biparietal diameter and the frontooccipital diameter, respectively, for group comparison purposes.
Head biometries were assessed by head circumference, and biparietal and frontooccipital diameters as described elsewhere.
Cortical development was assessed by measuring the following fissure depths bilaterally: Sylvian, insular, and parietoccipital in axial slices, and calcarine and cingulate in coronal slices as reported elsewhere.
CC length was measured. Also, CC morphometry was assessed by measuring the total area and the area of a subdivision in 7 portions: rostrum, genu, rostral body, anterior midbody, posterior midbody, isthmus, and splenium as described by Witelson et al CC assessment was performed following a previously reported methodology ( Figure 2 ).
Statistical analysis
Student t test for independent samples and Pearson χ 2 or Fisher exact tests were used to compare quantitative and qualitative data, respectively. To detect differences in the studied metabolic ratios between cases and controls, Student t test for independent samples and a multivariate analysis of covariance, adjusting by maternal smoking, gestational age at MRI, and maternal body mass index, were conducted. Correlations between the frontal lobe metabolic information and the cortical development and CC assessment were obtained by partial correlations. Results were considered to be significant at a P value < .05. All statistical calculations were done using statistical software (SPSS, version 17.0; IBM Corp, Armonk, NY).
Results
Baseline features and clinical outcomes in the study groups
A total of 119 fetuses, 64 late-onset small fetuses and 55 controls, were evaluated. Ultrasound and MRI data were obtained from all patients included in the study and no differences were found in the gestational age at which they were performed. A higher rate of smokers was found in the late-onset small fetuses when compared to AGA. Also, small fetuses presented a significantly higher rate of labor induction and emergency cesarean delivery than AGAs ( Table 1 ). No signs of intracranial pathology were found in any of the fetal MRI. Considering the whole cohort, 9 cases were discarded for microstructural assessment due to insufficient image quality. At the same time, 1 H-MRS data from 77 (64.7%) cases were selected for further analysis, after applying the quality-control criteria discussed earlier. There were no statistically significant differences between cases with interpretable and uninterpretable spectra in terms of maternal and fetal features.
| Characteristic | Small fetuses (n = 64) | AGA fetuses (n = 55) | P value a |
|---|---|---|---|
| Maternal age, y | 31.6 ± 5.8 | 32.9 ± 4.2 | .18 |
| BMI, kg/m 2 | 22.3 ± 3.5 | 23.2 ± 3.9 | .17 |
| Primiparity, % | 79.7 | 63.6 | .05 |
| Non-Caucasian ethnicity, % | 25 | 27.3 | .77 |
| Smoker, % | 26.6 | 3.7 | < .01 |
| GA at US, wk | 37.6 ± 1.1 | 37.3 ± 0.84 | .10 |
| GA at MRI, wk | 37.7 ± 3.3 | 38.4 ± 6.4 | .16 |
| GA at birth, wk | 38.5 ± 1.0 | 40.0 ± 1.1 | < .001 |
| Induction, % | 81.3 | 28.3 | < .001 |
| Cesarean delivery, % | 35.9 | 15.1 | .01 |
| Birthweight, g | 2374.5 ± 259.75 | 3411.2 ± 281.6 | < .001 |
| Weight centile | 3.2 ± 2.5 | 53.9 ± 24.5 | < .001 |
| Male/female | 35/29 | 27/28 | .54 |
| Neonatal head circumference, cm | 32.74 ± 1.55 | 34.69 ± 2.28 | < .001 |
| Neonatal head circumference centile | 18.21 ± 19.31 | 53.96 ± 29.21 | < .001 |
| Neonatal length, cm | 45.71 ± 1.67 | 50.0 ± 1.90 | < .001 |
| Neonatal length centile | 8.9 ± 9.04 | 55.50 ± 27.90 | < .001 |
| Neonatal acidosis, b % | 10.5 | 2.3 | .10 |
| Apgar score <7 at 5 min, % | 4.7 | 0 | .10 |
| Emergency cesarean delivery, % | 29.7 | 7.5 | < .01 |
| NICU stay length, d | 0.06 ± 0.5 | 0 | .35 |
a Student t test for independent samples, Pearson χ 2 , or Fisher exact test as appropriate
Head biometries and microstructural differences between groups
Late-onset small fetuses showed smaller head biometries expressed by significantly reduced biparietal diameters, and smaller frontooccipital diameters and head circumferences. They also presented significantly smaller CC with smaller callosal subdivisions except for the rostral and splenial subdivisions that did not reach statistical significance. This group also showed greater insular and cingulate sulci depths ( Table 2 ).
| Characteristic | Small fetuses (n = 57) | AGA fetuses (n = 53) | P value a | P value b |
|---|---|---|---|---|
| BPD, mm | 89.28 ± 3.30 | 94.61 ± 3.5 | < .01 | < .01 |
| FOD, mm | 109.22 ± 4.99 | 115.49 ± 4.28 | < .01 | < .01 |
| HC, mm | 313.31 ± 11.28 | 331.55 ± 10.64 | < .01 | < .01 |
| Cephalic index c | 81.85 ± 3.71 | 81.95 ± 3.08 | .87 | .93 |
| Corpus callosum | ||||
| Length/FOD | 0.363 ± 0.02 | 0.358 ± 0.02 | .17 | .18 |
| Total area/FOD | 1.05 ± 0.17 | 1.165 ± 0.23 | < .01 | .03 |
| Rostrum area/FOD | 0.07 ± 0.04 | 0.081 ± 0.05 | .18 | .64 |
| Genu area/FOD | 0.121 ± 0.04 | 0.145 ± 0.06 | .01 | .02 |
| Rostral body area/FOD | 0.235 ± 0.04 | 0.266 ± 0.05 | < .01 | < .01 |
| Anterior mid body area/FOD | 0.128 ± 0.03 | 0.139 ± 0.03 | .03 | .23 |
| Posterior mid body area/FOD | 0.123 ± 0.02 | 0.135 ± 0.03 | .02 | .12 |
| Isthmus area/FOD | 0.105 ± 0.02 | 0.118 ± 0.03 | < .01 | .01 |
| Splenium area/FOD | 0.271 ± 0.06 | 0.280 ± 0.07 | .48 | .85 |
| Cortical development | ||||
| Left insular depth/BPD | 0.299 ± 0.05 | 0.267 ± 0.03 | < .01 | < .01 |
| Right insular depth/BPD | 0.38 ± 0.06 | 0.326 ± 0.07 | < .01 | < .01 |
| Left Sylvian fissure/BPD | 0.155 ± 0.04 | 0.15 ± 0.02 | .6 | .38 |
| Right Sylvian fissure/BPD | 0.151 ± 0.04 | 0.153 ± 0.04 | .85 | .69 |
| Left parietoccipital fissure/BPD | 0.139 ± 0.02 | 0.132 ± 0.03 | .34 | .09 |
| Right parietoccipital fissure/BPD | 0.138 ± 0.03 | 0.135 ± 0.33 | .64 | .38 |
| Left cingular depth/BPD | 0.099 ± 0.01 | 0.088 ± 0.01 | < .01 | .01 |
| Right cingular depth/BPD | 0.098 ± 0.01 | 0.088 ± 0.01 | .02 | .02 |
| Left calcarine fissure/BPD | 0.176 ± 0.03 | 0.182 ± 0.04 | .46 | .86 |
| Right calcarine fissure/BPD | 0.178 ± 0.04 | 0.189 ± 0.03 | .23 | .41 |
a Student t test for independent samples comparing small vs AGA fetuses
b Multivariate analysis of covariance statistical analysis adjusting by maternal smoking, gestational age at magnetic resonance imaging, and maternal body mass index
Metabolic differences between groups
When metabolic ratios were compared between the clinical groups, significantly reduced NAA/Cho and increased Cho/Cr ratios were present in late-onset small fetuses when compared to AGAs. Only decreased NAA/Cho remained significant after adjusting for potential confounding factors ( Table 3 ).
| Variable | Small fetuses (n = 43) | AGA fetuses (n = 34) | P value a | P value b |
|---|---|---|---|---|
| NAA/Cho | 1.203 ± 0.241 | 1.371 ± 0.278 | .007 | .015 |
| Ino/Cho | 2.855 ± 0.497 | 2.804 ± 0.460 | .673 | .797 |
| Cho/Cr | 0.717± 0.081 | 0.669 ± 0.117 | .044 | .731 |
a Student t test for independent samples
b Multivariate analysis of covariance statistical analysis was used to compare measurements adjusting for maternal smoking, gestational age at magnetic resonance imaging, and maternal body mass index.
Correlations between brain metabolic ratios and microstructural features
Considering both cases and controls, significant correlations were found between NAA/Cho ratios and head biometries. Also, NAA/Cho showed a significant correlation with CC length and area ( Figures 3 and 4 ), and areas from all callosal subdivisions except for the rostrum. No significant correlations were found between NAA/Cho and any of the cortical sulcation parameters, neither between Cho/Cr or Ino/Cho and any of the head biometries, callosal measurements, or sulcation parameters ( Table 4 ).
