Measures of Brain Metabolism and Diffusion in Neonates of Mothers with Diabetes

Abstract

BACKGROUND AND PURPOSE:

Maternal diabetes may affect neonates’ long-term neurodevelopment and cognitive behavior. Brain biochemistry and white matter fiber tracts may reveal early changes of brain abnormality. The purpose of this study was to compare brain metabolites and fiber structures in infants of mothers with diabetes with those of mothers without diabetes during the early stage of neonatal neurodevelopment.

MATERIALS AND METHODS:

Pregnant women with diabetes mellitus (DM) and healthy controls were recruited prospectively. Baseline characteristics including maternal and gestational age, weight, height, body mass index (BMI), and types of diabetes were extracted from the mothers and neonates in the clinical record. Infants of mothers with diabetes and healthy controls underwent scans by using a 3T MRI scanner. T2-weighted anatomic images were acquired for voxel positioning. ¹H-MR spectroscopy from the right frontal lobe and whole-brain DTI data were acquired from the participants. Independent sample t-tests were used to compare anthropometric data and a general linear model was used to compare brain metabolites and diffusion measures in the neonates.

RESULTS:

MRI data were successfully obtained from 77 neonates (41 healthy controls and 36 infants of mothers with diabetes). Gestational diabetes was the most common diagnosis (53%), followed by type 2 (33%) and type 1 pregestational DM (14%). Weight (P = .02) and BMI (P < .01) were significantly higher in mothers with diabetes with earlier birth gestational age. Myo-inositol (mI) measurements were significantly higher (P = .02) in infants of mothers with diabetes versus controls but this was not observed in the excitatory neurotransmitter of γ-aminobutyric acid (GABA) and other high-concentration metabolites. Mean diffusivity values were significantly higher in the anterior part (P = .03) and posterior gray matter (P < .05) of the cingulate gyrus for infants of mothers with diabetes compared with controls. Axial diffusivity values were significantly higher (P = .03) in the anterior part right gray matter of the cingulate gyrus for infants of mothers with diabetes compared with controls.

CONCLUSIONS:

The increase of myo-inositol in infants of mothers with diabetes may reflect a compensatory mechanism related to altered blood glucose to preserve and promote β-cell growth to enhance the production of insulin. The increase in mean diffusivity is consistent with decreased water diffusion, which potentially reflects abnormal changes in myelination and axonal loss.

Diabetes mellitus (DM) in pregnancy may result in both short- and long-term complications for both mother and offspring. In mothers, there is increased risk of gestational hypertension, preeclampsia, and postnatal metabolic syndrome¹ while infants of mothers with diabetes are at increased risk for fetal macrosomia, poor cardiorespiratory and metabolic transition, polycythemia, and hyperbilirubinemia.^2,3 Existing literature suggests that infants born to mothers with diabetes during pregnancy may experience an increased risk of altered brain maturation. As early as 3 months of age, infants of mothers with diabetes showed lower motor and mental development when compared with those of healthy controls.⁴ At age 2 years, infants born to mothers with gestational DM showed lower expressive language skills when compared with those born to healthy mothers.⁵ A cross-sectional study reported school-aged children born to mothers with gestational DM underperformed in graphic and spatial abilities as well as manual dexterity compared with those born to healthy mothers. Working memory and the overall intellectual capacity and cognitive abilities were also found to be lower in infants of mothers with diabetes at school ages.⁶ Collectively, these studies suggest adverse impact on neurodevelopment in early communication, cognitive abilities, and spatial recognition in infants of mothers with diabetes.

For structural brain development, a small study comparing third trimester brain volumes did not detect significant differences in fetuses of mothers with gestational DM compared with type 1 DM.⁷ However, in older children (aged 7–11 years) exposed to gestational DM in utero, brain morphology and volume showed reduced radial thickness in the left hippocampus compared with a reference group of nondiabetic controls.⁸

¹H-MR spectroscopy has been well described in healthy and high-risk fetuses and neonates.⁹ However, there are no existing neonate MR spectroscopy studies in infants of mothers with diabetes. Only 1 fetal MR spectroscopy study reported that the Cho/Cr ratio was marginally lower in the fetal brain of mothers with DM between 34- and 38-weeks’ gestational age (GA).¹⁰ Results from this study were limited by the measurements of only 2 metabolites while other high concentration metabolites were not reported. Results from the fetal MR spectroscopy were in line with a study of children with poorly controlled DM that showed ratios of NAA/Cr and Cho/Cr were significantly lower in the pons and lower NAA/Cr in the left posterior parietal white matter.¹¹

Very few studies have examined the role of disrupted white matter organization in infants of mothers with diabetes. Available evidence points to microstructural white matter abnormalities characterized by the reduction of fractional anisotropy (FA) values in the splenium of the corpus callosum, the posterior limb of the internal capsule, and the thalamus in infants of mothers with diabetes compared with controls.¹² Decreased FA values were associated with poorer neurocognitive performance in the infants of mothers with diabetes group, suggesting altered homogeneity of fiber orientation in these 3 brain regions. Observations were in line with infants of mothers with high maternal fasting glucose levels as lower FA values were found in the right amygdala.¹³

The metabolic exposures of hyperglycemia and impaired placental function in utero may disturb the biochemical environment of the brain¹⁴ and result in microstructural changes of developing white matter fiber tracts.^10,12 However, studies of early brain biochemistry and fiber tract integrity in this high-risk group are lacking. Therefore, the objective of this study was to compare in vivo brain metabolites and fiber tracking in infants of mothers with diabetes with those of healthy mothers without diabetes.

MATERIALS AND METHODS

Participants

Details reported for this study fulfill items in the Strengthening the Reporting of Observational Studies in Epidemiology checklist as shown in the Supplemental Data. Healthy term neonates and infants of mothers with diabetes were recruited prospectively in an exploratory observational study of brain development between 2019–2022 under a protocol approved by the local institutional review board. Informed written consents were obtained from the parents of each participant. Inclusion criteria comprised neonates of mothers with pregestational (including type 1 and type 2) DM or gestational DM. Exclusion criteria included any contraindications of MRI, documented chromosomal or genetic abnormality, congenital abnormality, central nervous system abnormality, and multiple gestations.

Clinical Data

Clinical data for neonates including sex, GA, postmenstrual age (PMA) at the time of MRI, birth weight, scan weight (weight at the time of MRI), blood glucose levels, and history of complications were extracted from the medical record. Clinical data for mothers including maternal age, weight (second trimester), height, body mass index (BMI), type of diabetes, and history of medications also were extracted.

MR Spectroscopy Acquisition

All neonates underwent MRI scans on a 3T scanner (Discovery MR750, GE Healthcare) by using an 8-channel head coil without sedation during natural sleep. Anatomic images were acquired by using a T2-weighted 3D fast spin-echo sequence (section thickness = 1 mm, spacing = 0 mm, TR = 2500 ms, TE = 64.7–89.9 ms, flip angle = 90°, number of slices = 120, matrix = 160 × 160). Unedited MR spectra were acquired by using Point Resolved Spectroscopy (PRESS) (TR/TE = 1500/35 ms, 128 transients) and 2 transients of water reference were collected. GABA-edited spectra were acquired by using Mescher-Garwood Point Resolved Spectroscopy (MEGA-PRESS)¹⁵ (TR/TE = 1800/68 ms, 256 transients). Sixteen-millisecond editing pulses were placed at 1.9 ppm during ON and 7.5 ppm during OFF acquisitions. Data were collected from a voxel that was positioned in the white matter of the right frontal lobe (20 × 15 × 15 mm³) as shown in Fig 1. Full scan parameters are reported in the Supplemental Data.

FIG 1.

A representative voxel location in the predominantly white matter right frontal lobe (20 × 15 × 15 mm³) of an infant of a mother with diabetes (male, GA 39.4 weeks).

DTI Acquisition

Diffusion data were acquired by using a single-shell acquisition with a single-shot spin-echo EPI sequence (TE = 8.7 ms, TR = 8 seconds, 200 mm FOV, 128 slices, 1.6 × 1.6 mm in-plane resolution, and 3-mm slice thickness). The diffusion scheme included b-values of 2500 s/mm², 64 diffusion directions, and 4 nondiffusion weighted volumes.

MR Spectroscopy Analysis

Spectral data were modeled by using Osprey-v2.5¹⁶ within Matlab R2023b. Preprocessing steps included residual water removal, eddy-current correction and robust spectral registration. The basis set applied for quantification was generated by using real vendor waveforms and sequence timings.¹⁷ Details of the metabolite basis functions and Gaussian basis functions for macromolecules and lipids were listed previously.^18,19

Metabolite measurements were reported in ratios to total creatine (tCr) and water-referenced concentration levels. GABA plus macromolecules (GABA+) was modeled by using the GABA-edited difference spectrum and other metabolites of interest were modeled by using the PRESS spectrum. Data quality metrics including SNR and full width at half maximum were calculated. Spectral data with metabolite SNRs and linewidths that were smaller than 6 and larger than 0.1 ppm, respectively, were excluded before statistical analyses, presumably due to bad shimming, poor water suppression, and motion artifacts.

DTI Analysis

The diffusion-weighted images were denoised and corrected for motion and eddy current distortions. Diffusion data were fitted by using DTIFit in FSL (http://fsl.fmrib.ox.ac.uk/fsl/fsl-4.1.9/fdt/fdt_dtifit.html) to compute FA, mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD). These metrics were then registered to the Edinburgh Neonatal Atlas–ENA50 by using ANTs (http://stnava.github.io/ANTs/).²⁰ After registration, means and standard deviations of each diffusion measure were computed in each of the 87 dHCP parcellations as shown in Fig 2.

FIG 2.

A representative neonatal atlas parcellation that contains 87 regional brain structures by using the automated dHCP structural pipeline. The full list of parcellation regions is listed in the Supplemental Data.

Statistical Analysis

All statistical analyses were performed by using SPSS Version 29 (IBM). Independent sample t-tests were used to compare anthropometric data. General linear models with Bonferroni corrections were used to compare brain metabolite levels and diffusion measures. Sex and PMA at the time of MRI were controlled as covariates for comparisons. Data points less or greater than 3 standard deviations were interpreted as outliers and excluded. P values < .05 were considered statistically significant. Of note, given the exploratory nature of this work, the level of significance was determined by the uncorrected a for each individual brain region.²¹

RESULTS

MRI scans were successfully performed in 41 controls and 36 infants of mothers with diabetes. For MR spectroscopy, 9 PRESS and 9 MEGA-PRESS data sets were excluded from controls and 4 PRESS and 2 MEGA-PRESS data sets were excluded from cases. For DTI, 13 and 8 data sets were excluded from controls and cases, respectively, due to motion artifacts or incomplete acquisition of diffusion data.

Clinical Demographics

Body weight at second trimester (P = .02) and BMI (P < .01) in mothers with DM were significantly higher compared with control mothers, as indicated in Table 1. Gestational diabetes (53%) was the most common type of DM followed by type 2 (33%) and type 1 (14%) DM. More than one-half (56%) of the mothers with DM were receiving insulin medication. For neonatal comparisons, birth GA was significantly earlier (P < .01) in infants of mothers with diabetes (37.9 ± 1.7 weeks) compared with controls (39.0 ± 1.8 weeks). Weight of infants of mothers with diabetes at the time of MRI was trending higher (P = .08) and blood glucose levels were borderline lower (P = .05) compared with healthy controls.

Table 1:

Baseline characteristics of mothers with (cases) and without (controls) diabetes and their neonates

	Controls	Cases	P Value
Number (n)	41	36	—
PRESS (n)	32	32
MEGA-PRESS (n)	32	34
DTI (n)	28	28
Mothersa
Age (years)	36.0 ± 5.7	34.2 ± 6.9	.23
Weight (kg)	78.3 ± 10.4	90.7 ± 27.5	.02b
Height (cm)	163.9 ± 6.2	166.3 ± 5.5	.15
BMI (kg/m2)	29.3 ± 4.3	34.6 ± 8.2	<.01b
Type of diabetes
Type 1	—	5 (14%)	—
Type 2		12 (33%)
Gestational diabetes		19 (53%)
Insulin medications		20 (56%)
Hypertension	3 (7%)	8 (22%)	—
Hypothyroidism	3 (7%)	9 (25%)
Neonates
Sex, male (n)	22 (54%)	16 (44%)	.36
Birth GA (weeks)	39.0 ± 1.8	37.9 ± 1.7	<.01b
PMA (weeks)	42.5 ± 2.6	42.3 ± 2.7	.80
Birth weight (g)	3260.4 ± 513.3	3240.2 ± 663.4	.89
Scan weight (g)	3938.8 ± 1042.3	4328.3 ± 714.2	.08
Glucose levels (mg/dL)c	64.4 ± 16.7	49.5 ± 19.0	.05

Note:—Data were compared by using independent sample t-tests for mothers.

^aMothers’ weight, height, and BMI were measured atthe second trimester.

^bIndicates significant differences between healthy controls and cases.

^cGlucose levels were available in only 35 cases and 8 controls.

Brain Biochemistry

Representative mean spectra for PRESS and GABA-edited difference spectrum are shown in Fig 3. Linear combination modeling showed successful fittings for tNAA, tCr, tCho, Glu, and Gln (collectively as Glx) and ml in PRESS, and GABA+ in the GABA-difference spectrum in MEGA-PRESS. Metabolite measurement of mI was significantly higher (P = .02) in the infants of mothers with diabetes group compared with controls regardless of the reporting method. There was a trend for greater concentrations of GABA+ in the infants of mothers with diabetes group as compared with the controls, although this did not reach statistical significance (P = .13), as indicated in Table 2. Measurements for other high-concentration metabolites including tNAA, tCho, tCr, and Glx were similar between the 2 groups. The average SNRs and linewidths were 16.3 and 7.8 Hz, respectively, indicating moderate signal and good shimming for the obtained data.

FIG 3.

Mean spectra and modeling of the in vivo data from neonates of mothers with diabetes. The linear combination modeling of high-concentration metabolites including tNAA, tCr, tCh, Glx, and ml are clearly visible in (A) the PRESS spectrum and GABA+ in (B) the MEGA-PRESS GABA-edited difference spectrum. Macromolecules and lipids were modeled for metabolite concentration measurements.

Table 2:

Metabolite concentration measurements acquired from the right frontal lobe in neonates of mothers with (cases) and those without (controls) diabetes

	Controls	Cases	P Value
Ratio of tCr
tNAA/tCr	1.09 ± 0.16	1.06 ± 0.18	.56
tCho/tCr	0.46 ± 0.08	0.45 ± 0.07	.56
Glu/tCr	1.14 ± 0.31	1.07 ± 0.41	.38
Gln/tCr	0.36 ± 0.30	0.46 ± 0.40	.35
Glx/tCr	1.51 ± 0.39	1.54 ± 0.53	.97
ml/tCr	1.52 ± 0.32	1.73 ± 0.24	.02a
GABA+/tCr	0.68 ± 0.27	0.83 ± 0.55	.21
Water-referenced
tNAA	6.26 ± 1.55	6.16 ± 1.40	.96
tCho	2.58 ± 0.48	2.55 ± 0.31	.91
tCr	5.81 ± 1.64	5.84 ± 1.04	.89
Glu	6.50 ± 1.99	6.41 ± 2.95	.81
Gln	2.13 ± 1.99	2.65 ± 2.12	.40
Glx	8.62 ± 2.64	9.05 ± 3.49	.70
ml	8.63 ± 1.98	9.99 ± 1.63	.02a
GABA+	0.54 ± 0.20	0.69 ± 0.45	.13

Note:—Measurements are reported in ratio of tCr and water-referenced concentration (institutional unit [IU]). General linear model with Bonferroni corrections was used to compare measurements between the 2 groups, and sex and PMA were controlled as covariates. GABA+ was measured from the MEGA-PRESS difference spectrum and other metabolites of interest were measured from the short-TE PRESS.

^aIndicates significant differences between healthy controls and cases.

DTI Findings

A representative DTI of diffusion metrics including FA, MD, AD, and RD is shown in Fig 4. Significantly higher MD (P = .03) and AD (P = .03) values were reported in the anterior part of the gray matter of the right cingulate gyrus in infants of mothers with diabetes compared with the controls. Significantly higher MD (P < .05) values were reported in the posterior part of the gray matter of the right cingulate gyrus in infants of mothers with diabetes compared with the controls. Illustrations for comparisons of AD and MD values in the cingulate gyrus where significant differences were reported between controls and infants of mothers with diabetes are shown in boxplots (Figs. 5 and 6). Full boxplots with all 87 parcellations are reported in the Supplemental Data for AD and for MD. Full numerical values are presented in the Supplemental Data for FA and MA and for AD and RD. Anatomic illustrations of brain parcellations with significant differences that are registered to the ENA50 atlas, highlighting significantly different ROIs as shown in Fig 7.

FIG 4.

A representative DTI of diffusion metrics for (A) FA, (B) MD, (C) AD, and (D) RD of an infant of a mother with diabetes (female, GA 38.7 weeks).

FIG 5.

A boxplot of AD values for the cingulate gyrus region between healthy controls (blue) and infants of mothers with diabetes (red). A significant higher AD value was reported in the anterior part of the right gray matter of the cingulate gyrus in infants of mothers with diabetes as shown with an asterisk.

FIG 6.

A boxplot of MD values for the cingulate gyrus region between healthy controls (blue) and infants of mothers with diabetes (red). Significant higher MD values were reported in the anterior and posterior part of the right gray matter of the cingulate gyrus in infants of mothers with diabetes as shown with asterisks.

FIG 7.

Diffusion measures were registered on an infant of a mother with diabetes (male, GA 39.4 weeks) for visualization. In the anterior part of the gray matter of the right cingulate gyrus (red), mean diffusivity and axial diffusivity values were significantly higher in infants of mothers with diabetes compared with healthy controls. In the posterior part of the gray matter of the right cingulate (blue), mean diffusivity and radial diffusivity were significantly higher in Infants of mothers with diabetes compared with healthy controls.

DISCUSSION

Infants born to mothers with diabetes in pregnancy are at risk for neurobehavioral concerns later in life. Despite the potential for altered intrauterine exposures influencing early brain development, research studies of brain development in infants of mothers with diabetes are very limited. Our study reveals significant differences in levels of myo-inositol and in the altered gray matter diffusion of the right cingulate gyrus.

Our overall cohort of infants was predominantly born at term. Infants of mothers with diabetes were born at a younger GA (37.9 weeks) compared with controls (39.0 weeks). These data are in line with a population-based study suggesting an earlier birth GA in infants of mothers with diabetes, especially in the setting of additional maternal comorbidities.²² Therefore, although there is a difference in GA at birth between groups, PMAs at MRI scan were comparable, which may have a more significant impact on metabolite levels.²³ Interestingly, neonatal body weight at the time of MRI was marginally but noticeably higher by 10% in infants of mothers with diabetes compared with the controls, despite similar birth weight and corrected age at the time of MRI. This findings are in keeping with a retrospective longitudinal analysis indicating an accelerated weight gain in infants of mothers with diabetes during the early stage of development compared with healthy controls.²⁴

An important finding of this work was the increase of mI in infants of mothers with diabetes compared with controls. mI is one of the most abundant in vivo metabolites located mainly in glial cells of the human brain. It functions as an osmolyte and is a key precursor of membrane phospho-inositides and phospholipids²⁵ that plays an important role in glucose regulation due to its characteristic of insulin-sensitizing effect that improves glucose homeostasis. Recent studies suggested that mI supplementation reduced insulin resistance after pregnancy and incidence rate of gestational DM, in particular for overweight and obese pregnant women.^26,27 A meta-analysis involving more than 400 participants showed supporting evidence that mI supplementation significantly reduced the incident rate of gestational DM and preterm delivery in pregnant women compared with controls.²⁸ Other benefits including the decrease of blood pressure, total and low-density lipoprotein cholesterol, and serum triglycerides and increase of high-density lipoprotein in pregnant women were reported.²⁸

MR spectroscopy studies using adult cohorts with DM showed that mI measurements were increased by approximately 8% and 20% in the frontal cortex and frontal white matter, respectively, compared with the participants without diabetes.^29,30 The levels of increased mI in adults with DM were in line with the results from this study as mI was increased by 13%–14% in infants of mothers with diabetes. The underlying mechanism for the elevation of mI in infants of mothers with diabetes is uncertain but investigations in adult cohorts suggest that it may be the result of increased cerebral osmolality in patients with diabetes.³¹ A previous study reported that mI gradually decreased with gestational age in the early stage of infancy.³² In this study, the higher mI measured in infants of mothers with diabetes compared with healthy controls could be an indication of delayed neurodevelopment in infants of mothers with diabetes. It is similarly important to note that though the infants of mothers with diabetes cohort presented here was born at a slightly younger gestational age than controls, which may have influenced mI measures, PMA at MRI was accounted for in the analysis, so that early birth GA may not fully explain the noted differences. Additional studies in both term and preterm cohorts are needed to validate these findings. Because mI plays a crucial role by acting as a secondary messenger molecule involved in insulin signaling that impacts glucose uptake, the increase of mI in infants of mothers with diabetes may help with cell osmotic to balance glucose levels.

It is important to note that while this work found significant differences in mI, there are other brain metabolites that could be influenced by prenatal exposure to maternal diabetes. Scyllo-inositol is another form of inositol that plays an important role in insulin signaling in addition to glycerophosphorylcholine and other choline-containing compounds such as phosphocholine, that may impact glucose intolerance and insulin resistance.³³ Existing literature also reported that betaine may modulate GABA homeostasis and the GABAergic system that plays an important role in regulating insulin and blood glucose levels.^34,35 Measurements of these additional metabolites would broaden the overall understanding of the brain biochemistry in infants of mothers with diabetes but due to their low concentration and the narrow spectral resolution in clinical field strength, accurate measurements are challenging.

In this study, blood glucose levels were noticeably lower (P = .05) in infants of mothers with diabetes likely due to residual effects of high prenatal glucose exposure and subsequently high insulin levels. Technological advances in MR spectroscopy now allow for the investigation of regional brain glucose metabolism, though these require Chemical Exchange Saturation Transfer (CEST) MRI, high-field J-difference editing sequence, or¹³C MRS^36–38 and similar studies in infants and children are limited. Future work is warranted to bring similar measures to the investigation of glucose metabolism in high-risk populations, such as infants of mothers with diabetes, to advance our understanding of energy metabolism during the perinatal period.

For DTI, significantly higher MD and AD were reported in the anterior part of the gray matter of the right cingulate gyrus and significantly higher MD in the posterior part of the gray matter of the right cingulate gyrus in infants of mothers with diabetes compared with healthy controls. There is very little literature regarding microstructural development in infants of mothers with diabetes, as there is a single study in fetuses of mothers with diabetes by using ADC.¹⁰ While no significant difference was reported in ADC in the regions of interest in fetuses of mothers with diabetes compared with those of healthy mothers,¹⁰ this difference may be due to the nature of ADC, which measures the overall rate of water diffusion, whereas MD reflects the average diffusion across multiple directions that consider the directional nature of water diffusion. Moreover, the technical challenges in acquiring fetal DTI may limit the direct comparison to neonatal measures, as well as the rapid changes in both white and gray matter maturation during this period.

More specifically, MD indicates the average diffusion of water molecules within brain tissues and fibers. Tissue damage or disruption of fiber structures allows higher freedom of movement of water molecules in all directions, leading to the increase of MD values.³⁹ In this study, the increase of MD values was reported in the gray matter of the right cingulate gyrus, which was in line with the literature as MD values were reported to be higher in multiple brain regions, including the right cingulum bundle in adult patients with type 2 DM.⁴⁰ Available research also reported regional gray matter loss in patients with type 2 DM in different regions, including the cingulate gyrus.^41,42 Our findings suggest that the integrity of gray matter in the cingulate gyrus in infants of mothers with diabetes may have been impacted by the exposure of mothers with diabetes during pregnancy. While the etiology is unclear, long-term exposure to high blood sugar levels may result in regional inflammation and edema.⁴³ Of note, the cingulate gyrus is one of the brain regions that develops early during the fetal period compared with other cortical areas,⁴⁴ and may be more susceptible to long-standing metabolic changes, as with long-term exposure to elevated glucose.

Conversely, measurements of FA primarily reflect white matter integrity in the developing brain. While there are few studies of FA in offspring of mothers with DM, a study in children (3–10 years old) reported no significant changes between exposed infants and controls,⁴⁵ similar to our results.

There are a number of limitations. Infants of mothers with diabetes and healthy controls were well matched for PMA at the MRI scan but GA at birth was significantly younger in infants of mothers with diabetes due to the higher risk of prematurely born infants of mothers with diabetes compared with infants of healthy mothers.^3,22 The underlying impact from the younger GA at birth to brain development of infants of mothers with diabetes is little studied, though PMA at the time of scan should better reflect the full fetal-neonatal neurodevelopmental period that is equivalent to both cohorts. Data acquisitions were limited in the white-matter–rich right frontal lobe that is less developed during the early neonatal period compared with other regions. Additional data acquired from more mature regions with different predominant tissue types such as the thalamus and cerebellum may allow for broader investigations of metabolite changes in brain regions with more impact from the direct exposure of a diabetes environment in infants of mothers with diabetes. For DTI, the high b-value (2500 s/mm²) applied for data acquisition is susceptible to motion resulting in low SNR and may be limited in sensitivity to microstructure, particularly in neonatal white matter. A multishell acquisition with moderate b-values (eg, b = 700–1500 s/mm²) can balance SNR and sensitivity to diffusion properties, enabling more accurate microstructural modeling. Future work should explore this lower range of b-values to provide optimal data for neonatal assessments. Last, this study is performed by using a relatively small cohort of infants of mothers with diabetes and findings are close to the threshold of the significance level leading to a higher potential of bias and false-positives. Moreover, this work did not incorporate multiple corrections despite the multiple regions assessed to enhance the detection of meaningful brain regions affected, as there are arguments on how and in what conditions α corrections should be made²¹ and different methods of correction generate a new threshold to determine the significance and interpretation of the results.⁴⁶ The consistent findings in the cingulate gyrus in both AD and MD values suggest the findings reflect true physiologic differences. Nonetheless, additional longitudinal studies with a larger sample size would increase the power and reliability of the findings.

CONCLUSIONS

In summary, results from our neonatal study are in keeping with biochemical and fiber structural alterations seen in older children and adults with long-standing diabetes. These data suggest that diabetes in pregnancy is associated with early-life disturbances in neonatal brain development in infants of mothers with diabetes. We posit that these postnatal brain changes may be linked to a high blood sugar environment during critical stages of fetal brain microstructural and biochemical development. Future studies to relate changes of MR spectroscopy and DTI metrics in infants of mothers with diabetes relative to glucose control during pregnancy may provide important mechanistic insights to changes described in this work. Long-term follow-up is needed and currently underway to determine the functional impact of these neonate brain findings on later neurodevelopmental outcomes.

SUMMARY.

PREVIOUS LITERATURE:

A previous study reported that brain metabolites in fetuses of mothers with diabetes are lower than those of healthy mothers. The impact of maternal diabetes on the neurodevelopment of neonates is unknown.

KEY FINDINGS:

Increased myo-inositol (mI) levels are reported for the first time in infants of mothers with diabetes compared with those of mothers without diabetes. In addition, diffusion measures suggest noticeable changes in the gray matter of the right cingulate gyrus in infants of mothers with diabetes.

KNOWLEDGE ADVANCEMENT:

This study measures the impact of maternal diabetes mellitus on the brain microstructure and metabolism of offspring.

ABBREVIATIONS:

AD

axial diffusivity

BMI

body mass index

CEST

chemical exchange saturation transfer

DM

diabetes mellitus

ENA50

Edinburgh Neonatal Atlas 50

FA

fractional anisotropy

GA

gestational age

Gln

glutamine

Glu

glutamate

IDM

infant of diabetic mother

MEGA-PRESS

Mescher-Garwood Point Resolved Spectroscopy

MD

mean diffusivity

MI

myo-inositol

PCr

phosphocreatine

PMA

postmenstrual age

PRESS

Point Resolved Spectroscopy

RD

radial diffusivity

tCho

total choline

tCr

total creatine

tNAA

total N-acetylaspartate