Main

Cellular adaptation to osmotic stress is a vital process that protects cells from the effects of dehydration or edema. To maintain homeostasis cells produce osmolytes, molecules designed to create osmolality without compromising other cell functions (1). Osmolytes appear to be especially important in the brain as alterations in ion composition would affect excitability. myo-Inositol is one of the major osmolytes present in the CNS; others include sorbitol, taurine, betaine, and glycerophosphocholine.

myo-Inositol transport inside cells comprises uptake, which is selective and actively driven by SMIT (2), and efflux, which is nonselective and passively mediated by the chloride or volume-sensitive organic osmolyte-anion channel (3). Clinical studies have shown that brain cells respond to extracellular hypertonicity by increasing intracellular concentrations of myo-inositol (4). Exposure of neuronal cells (5), astrocytes (6), and glial cells (7) to enhanced extracellular osmolality in vitro was found to stimulate the expression of SMIT mRNA and hence increased intracellular myo-inositol.

Several studies in different species have shown that brain osmolality is increased by brain ischemia (8, 9). In adult rats with unilateral cerebral ischemia, SMIT expression was induced 6–72 h after the insult on both the affected and contralateral sides (10, 11). Work using the voltage-gated sodium-channel opener, veratridine, suggested that the increase in sodium influx caused by sodium-channel opening during the first stage of ischemia or injury may be the trigger for up-regulation of SMIT (12).

Brain energy metabolism after perinatal HI has been extensively investigated using phosphorus (31P) MRS and long TE proton (1H)MRS in both experimental neonatal animal models (13, 14) and term infants with NE (15, 16). These studies have supported a biphasic model of injury. In animals during HI there was an increase in brain lactate, which recovered on resuscitation. In both animal models and term infants with NE, during a secondary phase 8–12 h later there was further disruption of brain energy metabolism with a secondary increase in brain lactate. In infants with NE, the magnitude of the secondary derangement in brain energy metabolism was associated with the infant's neurodevelopmental outcome at 1 y (15–19).

Early MRI findings associated with severe perinatal HI brain injury in term infants with NE include damage to the deeper structures of the brain such as the basal ganglia and thalami and absence of the normal high SI in the posterior half of the PLIC on T1-weighted images (20).

Because myo-inositol is best measured using the technically more challenging short TE 1H MRS, in vivo data documenting acute changes in brain myo-inositol after perinatal HI have not been reported to date. The aim of this study was to investigate the early changes in myo-inositol/Cr using localized short TE 1H MRS in infants with NE and to relate these findings to the following measures of brain injury:1) basal ganglia lactate/Cr obtained using localized 1H MRS at long TEs, 2) acute changes on MRI, and 3) neurodevelopmental outcome at 1 y. We hypothesized that brain myo-inositol/Cr would be increased in the acute to subacute stages after perinatal HI as a response 1) to extracellular hypertonicity and induction of SMIT and 2) to decreased cellular ATP levels leading to decreased efflux of myo-inositol (21).

METHODS

Permission for this prospective MR study was granted by the Hammersmith Hospital Research Ethics Committee (number 93/4047), and parental consent was obtained for each study.

Subjects.

Term infants were selected for inclusion into the MR study according to at least one of the following criteria: history of fetal distress [late decelerations on cardiotocography, meconium-stained liquor, acidemia (pH < 7.1 or base deficit > 12 mM) in umbilical cord blood or arterial blood within 30 min of birth] and early abnormal postnatal neurologic signs consistent with NE.

The study population consisted of 14 infants born at gestational age (mean ± SD) of 39.6 ± 1.6 wk and having a birth weight of 3270 ± 490 g. Infants were examined using MRI and localized 1H MRS at 3.5 ± 2.1 d. Five infants had been entered into the pilot study of treatment with whole-body hypothermia for NE [Hammersmith Hospital Research Ethics Committee number 97/5133;(22)], and two were being cooled at the time of the scan. Clinical data are summarized in Table 1.

Table 1 Clinical details of birth-asphyxiated infants
Full size table

In the five infants (patients 7, 9, 12, 13, and 14) who were selected to receive the hypothermia treatment, hypothermia was initiated within 6 h of birth and maintained using a commercial air-cooling system (Polar Air, Augustine Medical, Eden Prairie, MN, U.S.A.), which induced hypothermia by blowing cool air through a translucent, perforated paper blanket placed over the infant. The air temperature was adjusted to maintain the rectal temperature between 33 and 34°C for 48 h. Using an extension to the air-cooling system this treatment could be continued while the infant was being studied in the magnet (patients 13 and 14). After 48 h of hypothermia the infant was rewarmed at 0.5°C/h. Patients 7, 9, and 12 were studied after 48 h of hypothermia when they were normothermic.

The MRI and MRS examinations were performed sequentially in natural sleep or after sedation with oral chloral hydrate (50–100 mg/kg) in infants who were self-ventilating and i.v. morphine infusion in those requiring mechanical ventilation. For immobilization and noise damping, the infant's head was partly surrounded by a bag filled with polystyrene balls from which the air was evacuated. Neonatal intensive care, including mechanical ventilation, i.v. fluids, and inotropic drug administration, was continued as appropriate during the scans, and all infants were accompanied by one pediatrician at all times. Heart rate, oxygen saturation, and axillary and rectal temperatures in those infants being cooled were monitored throughout the procedure. Mechanical ventilation was provided if necessary by an MR-compatible neonatal ventilator (babyPAC Neonatal Ventilator, pneuPAC Limited, Luton Beds, England).

Magnetic resonance imaging.

Cerebral MR data were obtained using a Marconi Medical prototype 1.5-T MR system (Marconi Medical Systems, Cleveland, OH, U.S.A.) and a double-tuned pediatric birdcage coil. The study protocol consisted of multislice T1-weighted conventional spin echo (CSE 500/15) and dual proton-density, T2-weighted fast spin echo (FSE TR 4200 TE 15/210) transverse images and sagittal T1-weighted conventional spin echo images. Infants were divided into two groups by analysis of the MR images by a researcher blinded to the MRS results. The appearance of the SI of the PLIC was assessed on both the T1- and T2-weighted images and graded as normal or abnormal. In some cases the SI from the PLIC was equivocal because of a delay in the evolution of the loss of signal. In these situations the basal ganglia were assessed for the presence of other areas of abnormal SI, and if present the MR image was classified as abnormal.

Magnetic resonance spectroscopy.

A voxel (2 cm × 2 cm × 2 cm) within the basal ganglia was defined, and 1H MR spectra localized to this voxel were obtained using the PRESS localization sequence (23), with computer-optimized slice-selective radio-frequency pulses, using a repetition time of 2 s, 128 data collections, and TE values of 40 and 270 ms. Water suppression was obtained with three initial chemical shift-selective saturation pulses. The voxel was placed in the basal ganglia on the left side in all cases and not specifically from an area including any lesion.

The 1H MR data were filtered using a Gaussian filter (line broadening, 1.5 Hz), Fourier transformed, and manually phased. The singlet peaks were routinely assigned at both long and short TEs to Cho (3.22 ppm), Cr (3.02 ppm), and NAA (2.02 ppm). The methyl group of lactate was a doublet centered at 1.34 ppm at TE 270 ms. At TE 40 ms the lactate peak overlapped with lipid and macromolecule resonances, and no attempt was made to separate the contributions. The MR spectrum of myo-inositol (3.56 ppm) was particularly complex because the six protons have similar but slightly different chemical shifts. The coupling pattern at 1.5 T was not first order; however, many authors have described how myo-inositol can be characterized at TE 40 ms by a singletlike peak at 3.56 ppm within the spectral resolution achieved in a clinical scanner (24, 25). The resonance at 3.56 ppm may also contain smaller contributions from inositol monophosphate, inositol polyphosphates, and glycine. In principle, definitive assignment to myo-inositol could be confirmed by quantifying the other myo-inositol resonance at 4.1 ppm. This peak, however, could not be reliably fitted because of its proximity to the residual water signal, and therefore it was not possible to precisely determine contributions other than myo-inositol. A complex multiplet could also be assigned to glutamine/glutamate (2.14–2.47 ppm).

Peak area ratios were calculated by fitting the peaks to a combined Lorentzian/Gaussian line shape. The starting point of the line-fitting algorithm was, first, to specify the expected resonances and then to limit the line width so that overlapping peaks were not obscured by a spurious fit to a broad resonance. For the spectra acquired with TE 40 ms, the maximum peak width of the fitted line shapes was limited to 0.1 ppm. The peak area ratio of myo-inositol/Cr was measured from the spectrum acquired using TE 40 ms, and lactate/Cr and NAA/Cr were measured from the spectrum acquired with TE 270 ms.

Neurodevelopmental outcome.

Neurodevelopmental outcome was assessed at 1 y of age using Griffith's DQ (26) and a detailed neurologic examination quantified with an OS (27). This score included items for tone, posture, passive and elicited motility and tone, interaction, reflexes, and vision and hearing responses. Infants were classified as normal if they had no abnormal neurologic signs (that is, a normal OS) and achieved a Griffith's DQ of >85. All other infants were regarded as having an abnormal neurodevelopmental outcome. Infants who died in this study did so because of neurologic problems and were scored with a DQ of <20.

Statistical considerations.

Data distributions were inspected and were transformed to normality when appropriate. ANOVA was used to compare the mean myo-inositol/Cr, lactate/Cr, NAA/Cr, and Cho/Cr in the group with normal MRI and outcome with the spectroscopy findings in the group with abnormal MRI and outcome. Regression analysis was used to compare spectroscopy findings with the Griffith's DQ. Results are presented as mean ± SD.

RESULTS

The MRI and MRS data in all infants were acquired at 3.5 ± 2.1 d after birth. MRI grading and MRS results for each infant are illustrated in Table 2. Representative MR images and 1H MR spectra are illustrated in Figures 1 and 2.

Table 2 MRI and MRS results with outcome data
Full size table
Figure 1
figure 1

Infant born at 41 wk gestational age by emergency cesarean section with Apgar scores of 1 at 1 min and 5 at 5 min (patient 4). The infant was examined by MRI and MRS on d 4. This infant had normal MRI and a normal neurodevelopmental outcome at 1 y of age. A, T1-weighted (conventional spin echo 500/15) transverse MR image illustrating bilateral high SI from myelin in the posterior half of the PLIC (arrow). The position of the 8-cm3 voxel from which the 1H MRS data were obtained is shown. B, T2-weighted (fast spin echo 4200/15/210) transverse MR image demonstrating bilateral regions of low SI representing myelin in the PLIC (arrow). C, 1H MR spectrum acquired from the voxel shown in A and B using PRESS localization sequence, TE 40 ms. myo-Inositol (mI)/Cr was 0.85. D, localized 1H MR spectrum acquired using TE 270 ms. Lactate/Cr was 0.44.

Full size image
Figure 2
figure 2

Infant born at 39 wk gestational age by assisted breech delivery with Apgar scores of 0 at 1 min and 0 at 5 min (patient 13). This infant was enrolled at <6 h of age into the pilot study of treatment with whole-body hypothermia for NE. MRI and MRS studies were undertaken on d 1 while the patient was being cooled. This infant had abnormal MRI and died on d 2 of age. A, T1-weighted (conventional spin echo 500/15) transverse MR image illustrating loss of the SI from the PLIC (arrow). There is evidence of brain swelling. There are some small abnormal SIs in the basal ganglia and thalami (arrows). The position of the 8-cm3 voxel from which the 1H MRS data were obtained is shown. B, T2-weighted (fast spin echo 4200/15/210) transverse MR image. There is no obvious low signal attributable to myelin in the PLIC but an exaggeration of the low SI in the lateral thalamus (arrow). C, 1H MR spectrum acquired from the voxel shown in A and B using PRESS localization sequence, TE 40 ms. Relative mI levels were increased, as illustrated by the mI/Cr metabolite ratio of 1.41. D, localized 1H MR spectrum acquired using TE 270 ms, illustrating markedly elevated lactate/Cr of 3.61.

Full size image

The serum sodium concentrations and osmolality are given in Table 2. In patient 11 the increase in osmolality was transient.

Six infants had normal MRI and eight infants had abnormal MRI of the basal ganglia and PLIC. There was no evidence of long-standing injury on any of the MR scans. The gestational age at birth of those infants with normal MRI was 40.0 ± 1.7 wk and in the group with abnormal MRI was 39.4 ± 1.6 wk (p > 0.05). The birth weight of those infants with normal MRI was 3160 ± 680 g and in the group with an abnormal MRI was 3360 ± 320 g (p > 0.05). The age at time of scan in the group with infants with normal MRI was 4.5 ± 2.2 d and in the group with abnormal MRI was 2.7 ± 1.8 d (p > 0.05).

All infants with normal MRI (n = 6) had a normal outcome at 1 y; none of these infants had been cooled. All those infants with abnormal MRI (n = 8) had an abnormal outcome at 1 y; five of these infants had been cooled.

For all infants myo-inositol/Cr and lactate/Cr were significantly correlated (p < 0.01;Fig. 3). This relationship remained significant when the two infants who were cooled at the time of the study were excluded (p < 0.05). For all infants myo-inositol/Cr was not related to NAA/Cr or Cho/Cr.

Figure 3
figure 3

myo-Inositol/Cr versus lactate/Cr and the regression line for all infants demonstrating a significant correlation between the two ratios (p < 0.01). myo-Inositol/Cr and lactate/Cr are both plotted on logarithmic scales to enable illustration of the spread of results.

Full size image

The myo-inositol/Cr in the infants with normal MRI and outcome was 0.75 ± 0.14 and in the group with abnormal MRI and outcome, 1.09 ± 0.26 (p = 0.05;Fig. 4A). There was a relation between the Griffith's developmental scale at 1 y of age and the myo-inositol/Cr ratio in the basal ganglia in the first week (p < 0.01;Fig. 4B).

Figure 4
figure 4

A, myo-inositol/Cr ratios plotted according to normal or abnormal neurodevelopmental outcome. There was a significant difference between groups (p < 0.01). B, myo-inositol/Cr in the basal ganglia in the first week plotted against the Griffith's developmental scales at 1 y, demonstrating a linear relationship (p < 0.01). Those infants with a DQ of 20 died.

Full size image

The lactate/Cr in the infants with normal MRI and outcome was 0.34 ± 0.08 and in the group with abnormal MRI and outcome, 1.05 ± 1.08 (p < 0.01;Fig. 5). There was a relation between the Griffith's developmental scale at 1 y of age and the lactate/Cr ratio in the basal ganglia in the first week (p < 0.01).

Figure 5
figure 5

Lactate/Cr ratios plotted according to normal or abnormal neurodevelopmental outcome. Lactate/Cr is plotted on a logarithmic scale to enable illustration of the spread of results. There was a significant difference between groups (p < 0.01).

Full size image

The NAA/Cr in the infants with normal MRI and outcome was 1.56 ± 0.27 and in the group with abnormal MRI and outcome, 1.30 ± 0.16 (p = 0.06;Fig. 6). There was no association between the Griffith's developmental scale at 1 y of age and the NAA/Cr ratio in the basal ganglia in the first week.

Figure 6
figure 6

NAA/Cr ratios plotted according to normal or abnormal neurodevelopmental outcome. There was a trend toward a lower NAA/Cr in the abnormal group, but this did not reach significance (p = 0.6).

Full size image

The Cho/Cr in the patients with normal MRI and outcome was 2.18 ± 0.32 and in the group with abnormal MRI and outcome, 1.97 ± 0.33 (p = 0.3). There was no association between the Griffith's developmental scale at 1 y of age and the Cho/Cr ratio in the basal ganglia in the first week.

DISCUSSION

This study found that myo-inositol/Cr was increased in the basal ganglia in the first week after birth in infants with NE who have sustained a severe HI brain injury as evidenced by MRI, basal ganglia lactate/Cr ratios, and neurodevelopmental outcome at 1 y. myo-Inositol/Cr was correlated to lactate/Cr.

Term infants up to 1 wk (mean age, 3.5 ± 2.1 d) of age only were studied for the following reasons. First, it was important to examine the acute brain changes because in vitro work has demonstrated up-regulation of SMIT up to 48 h after an insult (9). Second, as brain myo-inositol concentrations are known to decrease with increasing postnatal age (28), confounding effects because of age were limited. Third, metabolite changes occurring with secondary energy failure were likely to have developed by 1–3 d and this was therefore an appropriate time to acquire data. Unfortunately, at this early stage after perinatal HI injury, MRI changes are still evolving unless they are of antenatal onset. Loss of the normal SI of the PLIC may take 1–2 d to be evident on conventional T1- and T2-weighted images after HI; for example, before d 3 the presence or absence of the PLIC has a 91% positive predictive value for outcome whereas after d 3 the predictive value is 100%(20). For this reason we also took into account global SI changes in the basal ganglia to categorize the MR images into normal and abnormal groups.

The impact of mild hypothermia on the rate of evolution of MR changes and the severity of the brain injury are important considerations. MRS changes have been studied in experimental animal models; mild hypothermia started after severe transient HI in the newborn piglet reduced the delayed elevation in lactate peak area ratios (29) and the delayed cerebral energy failure measured by phosphocreatine to inorganic phosphate ratio (30). Although there are no MRS data in the newborn infant during or after cooling, the secondary rise in brain lactate/Cr is likely to have been ameliorated by cooling. Five of the eight infants with abnormal outcome were cooled; none of the six infants with a normal outcome were cooled. However, there was still a significant difference in lactate/Cr between outcome groups. The effect of cooling on brain myo-inositol concentrations or the activity of SMIT is unknown. Clinically relevant effects of hypothermia include mild hypokalemia (31), suppression of antidiuretic hormone release causing a diuresis, and a reduction in renal blood flow and GFR (32). In our pilot study of mild hypothermia, although cooled infants had a mild metabolic acidosis, a slightly raised blood lactate, and a relative hypokalemia, there was no significant change in plasma osmolality (22). Although it is not impossible that hypothermia may lead to an increase in myo-inositol concentrations, this is unlikely and hypothermia could not account for the increased myo-inositol concentrations seen in the infants who were cooled. Thus although the effect of cooling warrants further study it is unlikely that our results can be explained by this alone.

Metabolite ratios were used in this study as it was difficult to acquire a sufficient amount of data during the examination in these unstable patients to achieve reliable absolute quantification. Furthermore, the use of water as an internal reference may be problematic in this group of patients because both the water concentration and relaxation parameters may be changing in areas of gross pathology. We used Cr as the metabolite of reference for lactate, myo-inositol, NAA, and Cho because Cr concentrations have been reported to be relatively stable after an HI insult (33). The reported regional variation in metabolite concentrations (34) was avoided as data were specifically collected from the basal ganglia (20).

At short TEs, broad resonances have been reported to be present at 0.9, 1.3, and 2.05 ppm; these have been attributed to macromolecules and lipids (35). These compounds have been shown to be increased after HI events (36) and will overlap with the lactate resonance at 1.34 ppm at TE 40 ms; however, it was outside the scope of this study to define the contribution from lipids and macromolecules. At TE 270 ms, signals from these broad compounds are absent, allowing accurate assignment of the resonance at 1.34 ppm to lactate.

We assigned the peak at 3.56 ppm to myo-inositol (37). The comparison of the spectra obtained at TE 270 ms with those obtained at TE 40 ms are consistent with this assignment; however, we cannot rule out some contribution from glycine. This deserves further study, for example, using even shorter TEs or higher magnetic fields.

myo-Inositol is one of several osmolytes found in mammalian cells. There are three main groups: polyalcohols, such as sorbitol and myo-inositol; methylamines, such as glycerophosphocholine and betaine; and amino acids, such as glycine, glutamine, glutamate, aspartate, and taurine (38). myo-Inositol is involved in several cellular processes in the CNS. It is perhaps best known for intracellular signal transduction (39); as a precursor of phospholipids and phosphoinositides myo-inositol is also involved in cell membrane structure, adhesion, growth, vesicular trafficking, and survival. Most intracellular myo-inositol, however, is thought to function as a modulator of cell volume during persistent osmotic stress (40); the myo-inositol peak visible by MR represents this free myo-inositol, as myo-inositol involved in signal transduction is mostly bound and present in very small concentrations (41).

myo-Inositol and its transporter are thought to play an important role in normal fetal brain development; SMIT mRNA is diffusely and abundantly expressed in the fetal rat brain with a down-regulation at birth (42). Reasons for this elevated brain myo-inositol in fetuses are unknown. Clinical MRS studies have demonstrated that myo-inositol concentrations decrease after birth in the normal term infant brain and reach adult levels within the first few months of life (28).

The role of myo-inositol in osmoregulation in the CNS has been addressed in clinical studies. Studies using short TE 1H MRS have demonstrated an increase in brain myo-inositol in chronic hypernatremic states (4) and a decrease in myo-inositol in hyponatremic states (43). In all our infants except one the serum sodium and osmolality were within the normal range. In this infant the serum sodium and osmolality were mildly raised (148 mM and 315 mOsm/kg, respectively); part of the increase in basal ganglia myo-inositol is therefore might be attributable to this but probably minimally so. Excluding this patient from the analysis did not change the significant difference between normal and abnormal groups.

myo-Inositol/Cr was increased in those infants with abnormal MRI of the basal ganglia and in those infants with an abnormal neurodevelopmental outcome at 1 y. There was a correlation between myo-inositol/Cr and lactate/Cr ratios. A possible explanation for this acute change in myo-inositol/Cr may be induction of SMIT (10, 11), one of the many gene expressions to be influenced by an acute insult. Activation of SMIT may be in response to the influx of cations occurring during membrane depolarization at the start of the insult. This has been demonstrated with neuronal cells exposed to veratridine (causing cellular uptake of sodium); these cells developed an up-regulation of SMIT mRNA levels (12). Veratridine used with blockade of myo-inositol uptake demonstrated increased cytotoxicity in neuronal cultures, suggesting a possible protective role of myo-inositol. Accumulation of osmolytes is a slow process taking hours or days in contrast to the regulatory cell volume increase that occurs with ions (44). As there are no studies on long-term changes in brain myo-inositol/Cr after HI, it is unclear whether the increase in myo-inositol during the early hours and days after HI persists.

The relative reduction in cellular ATP levels in the injured brain may also lead to increased brain myo-inositol/Cr as efflux of myo-inositol from the cell requires the nonhydrolytic association of ATP with the volume-sensitive organic osmolyte-anion channel (21). Thus, decreased cellular ATP availability may be associated with decreased myo-inositol release, exacerbating neuronal swelling.

Other disorders associated with increased brain myo-inositol levels include Alzheimer's disease (45), preterm white matter damage at a corrected age of term (46), and Down syndrome (47). In the latter condition, up-regulation of the SMIT gene, which has been mapped to the long arm of chromosome 21 (48), was demonstrated in cultured fibroblasts (49), possibly explaining the increased brain myo-inositol concentrations seen in these patients.

In vitro work has suggested that myo-inositol is a marker for glia (50); clinical studies in adult white matter disease caused by HIV-1 cognitive motor complex attribute elevated white matter myo-inositol to increased glial activity (51). Gliosis as a cause for increased myo-inositol in these infants with NE is unlikely as they were studied so acutely after the insult; furthermore, recent studies have demonstrated that myo-inositol is not a specific marker of glia (52).

In this study, brain lactate/Cr was increased in infants with an abnormal neurodevelopmental outcome; these results confirm previous studies investigating changes in brain lactate/Cr within the first week of life and subsequent neurodevelopmental outcome (16, 34, 53). In addition, this study demonstrates an association between early increases in lactate/Cr and myo-inositol/Cr and a possible relationship between these biochemical changes and MRI abnormalities.

We demonstrated that brain NAA/Cr measured at a mean of 3.5 d was not significantly related to myo-inositol/Cr nor was it reduced in infants with an abnormal neurodevelopmental outcome. There was not a relationship between NAA/Cr and MRI abnormalities. NAA/Cr in the normal outcome group demonstrated a wider spread than NAA/Cr in the abnormal group. Previous studies in infants with NE at <1 wk of age (median, 1.8 d) concur with our findings (34); studies conducted in infants with NE at ≥1 wk of age have demonstrated a decrease in NAA/Cr (19) and NAA/Cho (53) ratios in infants with a poor neurodevelopmental outcome. This delay in reduction in NAA/Cr or NAA/Cho may indicate that neuronal loss occurred during a prolonged time or that a reduced NAA concentration may be masked by increased metabolite T2 values during secondary energy failure (54).

Brain Cho/Cr measured at a mean of 3.5 d was not significantly related to myo-inositol/Cr nor was it reduced in infants with an abnormal neurodevelopmental outcome. There was not a relationship between Cho/Cr and MRI abnormalities. Previous studies have suggested a decrease in Cho after perinatal HI (33); however, this did not reach significance in our study.

In summary, myo-inositol has several important roles in the CNS, one of which is cell volume regulation during persistent osmotic stress. myo-Inositol may also play a part in the response to HI injury. Our preliminary data suggest that at a mean age of 3.5 d, brain myo-inositol/Cr was increased in the basal ganglia of infants with NE with severe brain injury as evidenced by increased lactate/Cr, abnormal MRI, and an abnormal neurodevelopmental outcome at 1 y.