Magnetic resonance spectroscopy markers of axons and astrogliosis in relation to specific features of white matter injury in preterm infants

Abstract

Introduction

Punctate white matter lesions (pWMLs) and diffuse excessive high signal intensity (DEHSI) are commonly observed signal abnormalities on MRI scans of high-risk preterm infants near term-equivalent age. To establish whether these features are indicative abnormalities in axonal development or astroglia, we compared pWMLs and DEHSI to markers of axons and astrogliosis, derived from magnetic resonance spectroscopy (MRS).

Methods

Data from 108 preterm infants (gestational age at birth 31.0 weeks±4.3; age at scan 41.2 weeks±6.0) who underwent MR examinations under clinical indications were included in this study. Linear regression analyses were used to test the effects of pWMLs and DEHSI on N-acetyl-aspartate (NAA) and myoinositol concentrations, respectively.

Results

Across the full sample, pWMLs were associated with a reduction in NAA whereas moderate to severe DEHSI altered the normal age-dependent changes in myoinositol such that myoinositol levels were lower at younger ages with no change during the perinatal period. Subgroup analyses indicated that the above associations were driven by the subgroup of neonates with both pWMLs and moderate to severe DEHSI.

Conclusion

Overall, these findings suggest that pWMLs in conjunction with moderate/severe DEHSI may signify a population of infants at risk for long-term adverse neurodevelopmental outcome due to white matter injury and associated axonopathy. The loss of normal age-associated changes in myoinositol further suggests disrupted astroglial function and/or osmotic dysregulation.

Introduction

Neuroimaging studies have demonstrated that the majority of preterm infants have white matter abnormalities, including signal abnormalities, volume loss, cystic abnormalities, enlarged ventricles, thinning of the corpus callosum, and delayed myelination near term equivalency [1–3]. Although many of the above are now considered risk factors for later neurodevelopmental impairments, the sensitivity and specificity of these conventional imaging patterns for predicting future outcomes, particularly neurocognitive functioning, remains limited [2–6]. Moreover, even less is known about what these conventional imaging biomarkers signify at a cellular/molecular level, a critical gap in knowledge at a time when human clinical trials targeting different strategies for neuroprotection in preterm infants are emerging.

We used multimodal in vivo MR imaging to better characterize the two signal abnormalities most widely used to signify white matter injury in preterm infants in the neonatal period: punctate white matter lesions (pWMLs) and diffuse excessive high signal intensity (DEHSI). Thus far, the limited radiological-pathological data available have suggested that pWMLs, visualized as focal areas of high signal on T1-weighted MRI, indicate focal necroses with lipid-laden macrophages or possibly foci of activated microglia [6, 7], which in human neuropathologic studies has been associated with axonopathy [8, 9]. In contrast, DEHSI, visualized on T2-weighted MRI as diffuse regions of high signal, may reflect white matter gliosis, [1, 10, 11] which has been associated with arrested preoligodendrocyte maturation [12].

Importantly, although neuropathological studies demonstrate that focal necroses and diffuse white matter gliosis may frequently co-occur, there is not yet consensus as to whether they should be considered independently or combined—as in a single spectrum of white matter injury [8, 11, 13]. Similarly, there has been no consensus in neuroimaging studies as to whether pWMLs and DEHSI should be considered independently or combined to yield an overall white matter injury score [2, 14, 15], and such variability across studies may also explain the variability in the extent to which neonatal MRI is predictive of later neurodevelopmental disability [3–5, 14–16]. Previous research by our group has suggested pWMLs and DEHSI are associated with distinct patterns of metabolic alteration in the parietal white matter [17]. However, in that study, analyses pertaining to pWMLs and DEHSI were conducted in parallel and no direct statistical comparison was made between pWMLs and DEHSI.

To delineate the associations between signal abnormalities on conventional MRI and cellular/molecular aspects of white matter injury in vivo in preterm infants, we compared pWMLs and DEHSI to markers of neuronal/axonal integrity and astrogliosis derived from MR spectroscopy (MRS). N-acetyl-aspartate (NAA) is synthesized in the mitochondria of neurons and axons [18, 19] and is reliably depleted in the setting of traumatic brain injury [20], stroke [21], and other hypoxic-ischemic brain injuries [22] and, accordingly, is used as a marker of axonal integrity/axonopathy. In contrast, myoinositol, an osmolite expressed in high intracellular concentration in astroglia, is consistently elevated in the setting of multiple sclerosis and other neuroinflammatory CNS conditions [23–25] as well as in tumors characterized by a high fraction of glial cells [26], and thus is often used as a marker for astroglia.

We first tested the hypothesis that the presence of pWMLs would be associated with a decrease in NAA, reflecting axonopathy, independent of DEHSI. In contrast, we hypothesized that increasing DEHSI severity would be associated with an elevation in myoinositol, reflecting astrogliosis, independent of pWMLs. Additionally, because pWMLs and DEHSI were considered distinct entities exerting independent effects on axons and astroglia, we also tested whether different subgroups defined by pWMLs and DEHSI (i.e., pWMLs without DEHSI, pWMLs with DEHSI) were associated with different effects on NAA and myoinositol, suggesting that pWMLs and DEHSI may be recombined to define distinct patterns of white matter injury (WMI).

Materials and methods

Case selection

Data from 108 preterm neonates (mean gestational age at birth 31.0 weeks±4.3; range 23–36 weeks; mean postconceptional age at scan 41.2 weeks±6.0; range 25.7–60.7 weeks) were included in this study. Details regarding case selection are available in [17]. Briefly, all preterm infants who underwent clinically indicated MRIs were screened prospectively as part of ongoing longitudinal studies of neurodevelopment in neonates with prematurity. All available cases were included in this study provided: (1) the imaging study had been completed on an infant born before 37 gestational weeks of age; (2) the infant was not older than 60 weeks postconceptional age (PCA; calculated as the interval between the mother’s last menstrual period and birth plus postnatal age) at the time of the MRI; (3) there was no evidence of cerebral abnormality other than pWMLs or DEHSI (i.e., large vessel acute or chronic infarction, parenchymal hemorrhage, infection, tumor, or cerebral malformation); and (4) there was no clinical or laboratory evidence of liver failure, hyperbilirubinemia (requiring exchange transfusion), or underlying inborn error of metabolism. Prior results from this cohort have been published in [17].

MR data acquisition

MRI studies were acquired under clinical indications (most often to assess brain injury following preterm birth) on a GE 1.5 T (Signa LX, GE Healthcare, Milwaukee, WI) MR System using a customized neonatal transmit-receive head coil. Some studies were conducted using an MR compatible incubator; however, most were conducted with the neonate wrapped in a blanket and secured with appropriate physiological monitoring equipment. Per clinical protocol, most infants were sedated with choral hydrate. Ear protection was achieved using foam ear plugs in conjunction with MiniMuffs (Natus Medical Inc., San Carlos, CA). Conventional imaging studies were acquired with the MRS studies and included a 3D coronal SPGR sequence (echo time (TE)=6 ms; repetition time (TR)=25 ms, FOV=18 cm; matrix=256×160; slice thickness 1.5 mm, spacing 0 mm) or axial and sagittal T1-weighted FLAIR sequences (TE=7.4, TR=2100; TI=750; FOV=20 cm; Matrix=256×160), axial T2-weighted FSE sequence (TE=85 ms, TR=5000 ms, FOV=20 cm, matrix=320×160 or 256×128), and a diffusion-weighted sequence (TE=80; TR=10000; FOV=22 cm; Matrix=128×128; slice thickness=4.5 mm, spacing 0 mm).

¹H spectra were acquired from a single voxel (approximately 3 cm³) placed in the parietal white matter dorsolateral to the trigone of the lateral ventricle in the left hemisphere using a point resolved spectroscopy (PRESS) sequence with a short TE of 35 milliseconds (ms), a TR of 1.5 s, 128 signal averages, and a total acquisition time for each spectrum of approximately 5 min, including scanner adjustments. The parietal white matter location was selected because (1) the parietal white matter is known to be a region of vulnerability in preterm infants and (2) developing axons from numerous thalamocortical and corticocortical association pathways traverse that region [27, 28].

Determination of pWMLs and DEHSI based on conventional MR images

Conventional MRI scans (T1-, T2-, and diffusion-weighted sequences) for all studies were independently reviewed by two investigators (AP and JLW) and scored for the presence of both pWMLs (defined as punctate T1-hyperintense lesions in the periventricular white matter and corona radiata; Fig. 1) and DEHSI (defined as high signal on T2-weighted MR images in the cerebral white matter) and scored on a four-point scale: 0/within normal limits, 1/mildly increased, 2/moderately increased, and 3/severely increased (Fig. 2) (see also Online Resource).

Fig. 1.

Punctate white matter lesions. Each case was reviewed by two investigators and scored for the presence of punctate white matter lesions, visualized as focal lesions with high signal intensity in the periventricular white matter (arrows)

Fig. 2.

DEHSI. Using a modified version of the scale by Maalouf and colleagues [1], DEHSI was scored qualitatively on a four-point scale based on the extent of signal abnormality within the white matter: no signal abnormality (a), high signal restricted to the periventricular region only, classified as mild (b), high signal in the periventricular regions extending into the centrum semiovale, classified as moderate (c), and high signal extending from the periventricular white matter into the intragyral white matter (d). The corresponding ¹H spectra acquired from a ~3 cm voxel in the white matter dorsolateral to the trigone of the left lateral ventricle are displayed below. The peaks corresponding to NAA and myoinositol are highlighted in black, while other prominent peaks included in LCModel to generate a better model fit are highlighted in gray. Note that (d) also had pWMLs, which are depicted in Fig. 1. NAA=N-acetyl-aspartate, mI=myoinositol, Glu=glutamate, Gln=glutamine, Cho=choline, Cr=creatine, Lac=lactate

Metabolites analyzed and data processing

We focused our analyses on two key metabolites: NAA and myoinositol. Absolute concentration for each metabolite was quantitated from the MRS spectra using LCModel software (Stephen Provencher Inc., Oakville, Ontario, Canada, LCModel Version 6.1–4 F). In accordance with prior publications [17, 29, 30], metabolite concentrations were corrected for the varying fractions of cerebrospinal fluid and tissue water content in the parietal white matter region of interest. For absolute quantitation, the signal from unsuppressed water was used as internal concentration reference. MR spectra of low quality were removed by limiting the sample empirically to spectra with a linewidth (measure of field homogeneity) of <5 Hz and signal-to-noise ratio (SNR) ≥5. Cramer Rao bounds were typically less than 15 % (as calculated by LCModel).

Statistical analyses

Statistical analyses were carried out in Interactive Data Language (IDL; Exelis Corporation) and visualized in IDL and in SPSS (V.20, IBM Corporation). Standard linear regression analyses were performed to test the relation of NAA and myoinositol concentrations to the presence of pWMLs and the severity of DEHSI. In those analyses, the following variables were simultaneously used as regressors: age, pWML main effect, DEHSI main effect, pWML × age interaction, DEHSI × age interaction, and pWML × DEHSI interaction.

For the final analyses comparing each of the three white matter injury subgroups (pWMLs only, moderate/severe DEHSI only, and combined pWMLs/mod-to-severe DEHSI) to the “controls” (infants without either white matter injury feature), we repeated the above regression analyses, pairwise (i.e., including infants from one white matter injury subgroup and the controls). No mathematical correction was made for multiple comparisons, and all comparisons have been reported.

Results

Punctate white matter lesions and DEHSI in relation to NAA, an axonal marker

NAA concentration is known to increase rapidly during the perinatal period [29, 31, 32], and we observed a similar pattern in this sample (t [101]=9.74, p<0.001). Controlling for age, linear regression analysis demonstrated that the presence of pWMLs was associated with a significant decrease in NAA concentration (p=0.015) independent of DEHSI. There were no associations between DEHSI and NAA concentration or significant interactions (Table 1 and Fig. 1 in Online Resource).

Punctate white matter lesions, DEHSI and myoinositol, a marker of astrogliosis

Like NAA, myoinositol showed marked age-related changes (t [101]=−3.95, p<0.001). Linear regression analysis demonstrated a significant DEHSI-by-age interaction (p=0.008), but no significant associations or age-associated interactions among pWMLs and myoinositol (Table 1 in Online Resource).

To further understand the DEHSI-by-age interaction in association with myoinositol concentration, we median-split the infants with regard to DEHSI. Next, we compared the slopes of the age-associated changes in myoinositol concentrations in the cases with no/mild DEHSI (N=70) and in the cases with moderate/severe (N=38) DEHSI (Fig. 2 in Online Resource). Post hoc analyses indicated that at younger ages (i.e., 37 weeks postconceptional age), myoinositol concentration was lower in the cases with moderate to severe DEHSI (t [101]=−2.62, p=0.01). Moreover, whereas the infants with no/mild DEHSI demonstrated an age-associated decline in myoinositol concentration (r=−0.61 p<.001), there was no such age-associated decline in myoinositol concentration for the infants with moderate to severe DEHSI (r=−0.24, p>0.1).

NAA and myoinositol in association specific WMI subtypes defined by pWMLs and DEHSI alone and in combination

In the above analyses, we controlled for the possible effects of each injury type on the NAA and myoinositol concentration, statistically. It is possible that the effects of pWMLs and DEHSI on NAA and myoinositol are categorically different in infants with a single WMI feature compared to infants with combined features. Thus, we ran a final analysis across subgroups where we selected out neonates with a single white matter injury feature, that is a group of neonates with pWMLs without DEHSI (i.e., pWMLs and DEHSI = no/mild, n=11) and a group of neonates with DEHSI without pWMLs (i.e., DEHSI = moderate or severe; no pWMLs, n=19). We then defined a combined group consisting of neonates with both injury patterns (pWMLs and DEHSI = moderate or severe, n=19). Finally, we designated a “control” group consisting of neonates without either injury feature (i.e., DEHSI = no only; no pWMLs, n=22).

Controlling for age, group level comparisons among the infants with each white matter injury pattern alone (groups 1 and 2 above) and the infants without white matter injury (“controls”) did not reveal significant differences regarding NAA or myoinositol; however, parallel to above, there was a trend for NAA to be lower in the neonates with pWMLs alone (t [29]=−1.95, n.s.) (Fig. 3, see also Table 2 in Online Resource). Furthermore, for the combined pWML/DEHSI group, there was both a statistically significant decrease in NAA (t [37]=−2.70, p=0.01) and WMI-by-age interaction for myoinositol (t [37]=2.67, p=0.011) (Fig. 3, see also Table 2 in Online Resource). As above, the interaction demonstrated an age-dependent decrease in myoinositol in the “control” infants (r=−0.78 p<0.001) but no change in myoinositol concentration across ages in the infants with the combined pWML/DEHSI injury pattern (r=0.14, p>0.5).

Fig. 3.

NAA and myoinositol concentrations among white matter injury subgroups versus infants without apparent white matter injury. In the graphs above, “control” infants (pWML=0; DEHSI=0) are represented as blue circles and infants with the specified WMI pattern are represented as green triangles. NAA and myoinositol concentrations were not different from controls for cases classified as pWML alone or DEHSI alone. In contrast, there was a main effect of white matter injury on NAA concentration (p=0.01) and age-related interaction in relation to myoinositol (p=0.011) such that cases with moderate to severe DEHSI did not evidence an age-associated decline in myoinositol. Post hoc analyses indicated that the slope of myoinositol as a function of age was not significantly different from zero for the combined group (r=0.14, p>0.5) whereas there was a statistically significant negative slope for the “control” group (r=−0.78, p<0.001)

Discussion

The major finding in this study was a strong association between punctate white matter lesions and diminished NAA in the parietal white matter. Additionally, moderate to severe DEHSI eliminated the age-associated decline in myoinositol concentration during the perinatal period. However, somewhat unexpectedly, the above associations were essentially only present in the neonates who displayed the combined injury pattern (i.e., pWMLs plus moderate to severe DEHSI). Infants with moderate to severe DEHSI, but without pWMLs were not different from control infants with regard to NAA or myoinositol. Similarly, although there was a trend toward diminished NAA in the infants with pWMLs alone, the difference in NAA (or myoinositol) only reached significance among infants with the combined injury pattern.

Following advances in neonatal intensive care, in the past decades, there has been a pronounced shift in the pattern of white matter injury as defined by neuroimaging or histopathological examination [12, 13, 33]. Large, cystic lesions (i.e., cystic periventricular leukomalacia) are now rarely observed in vivo or at autopsy [12, 13]. However, non-cavitary lesions (e.g., punctate T1-hyperintense lesions on MRI, non-cavitary focal necroses and microcysts on histopathologic exam and diffuse astrogliosis) have remained prevalent while other features such as diffuse signal abnormalities (e.g., DEHSI) have gained attention [1, 2, 8, 12–14]. Overall, this shift in the white matter injury features observed in vivo and at autopsy has prompted the need to reevaluate white matter injury in the modern era. Moreover, in an era where infant mortality rates are low, even among some of the very extreme preterm infants, and when molecular therapies targeting different cellular/molecular mechanisms of white matter injury are emerging, there is a pressing need to develop in vivo biomarkers, which can be used to both identify and monitor infants who might benefit specific therapies for neuroprotection or neurorehabilitation.

In line with this, infants with pWMLs, particularly those who also had moderate to severe DEHSI, had markedly diminished NAA, a marker for axonopathy. These results, which are in line with results from diffusion tensor imaging studies [34], suggest pWMLs together with moderate to severe DEHSI may be an appropriate biomarker by which to select infants who would benefit from a therapy targeting axonal growth/development (e.g., certain growth factors).

At the same time, neither imaging pattern alone or in combination portended an increase in myoinositol, our marker for astrogliosis. While this does not preclude an underlying gliosis in the setting of either injury pattern, the finding that the normal age-associated changes in myoinositol concentration were altered in the setting of moderate to severe DEHSI (in particular, among neonates with comorbid pWMLs) does suggest other interpretations. First, although the biological basis for the decline in myoinositol concentration during the perinatal period is not well-understood, the fact that myoinositol is localized to the astroglial compartment suggests that, at minimum, the age-associated decline can be interpreted as a maturation change in astroglia. Thus, one straightforward interpretation for the loss of age-associated changes in myoinositol concentration in the setting of pWMLs coupled with moderate to severe is that at a cellular level, the injury signified by the combined pattern suggests disrupted astroglial maturation.

At the same time, the finding that myoinositol was lower in the infants with combined pWMLs/DEHSI at younger ages points to additional concerns regarding osmotic stress. In addition to being a potential indicator of astroglial concentration, myoinositol is a well-established osmolite and data from cell culture and human studies indicate that intracellular myoinositol concentrations decrease in the setting of high extracellular water content and concomitant osmotic stress [35, 36]. In line with this, previous studies have also demonstrated increased diffusivity in association with DEHSI [37] as well as prolonged T2 relaxation [38]. Thus, although additional research is needed, it is possible that targeting the underlying cause of the osmotic stress may be a further strategy for neuroprotection in this population. Toward this goal, future studies may also want to consider whether DEHSI is associated with specific prematurity complications (e.g., pleural effusions, ascites), recurrent use of certain drugs or even underlying genetic vulnerabilities.

Axonopathy

An ongoing controversy in the human neuropathology literature is the extent to which focal necroses in the periventricular white matter are associated with a diffuse axonopathy [8, 12]. Staining for fractin (caspase-cleaved actin, an apoptotic marker), Haynes and colleagues were able to demonstrate widespread axonopathy in regions surrounding and distant from an acute or organizing focal necrosis [8]. In contrast, Buser and colleagues did not observe axonopathy distal to the microscopic cysts in their contemporary cohort of autopsy cases [12], which may be attributable to methodological differences—in particular, a bias toward chronic lesions in the study of Buser et al.—or to the shift in white matter injury in the modern era. In this cohort, which is from a similar era as the contemporary cohort in [12], we observed a decrease in NAA in association with pWMLs, particularly in those cases with comorbid moderate to severe DEHSI. Notably, at the imaging resolution (~3 cm cubic volume) of our MRS voxel, it is impossible to localize a decrease in NAA to the region defined by a pWML or even cluster of pWMLs. Moreover, in this study, the parietal white matter voxel was placed in the same standard anatomical location in each infant regardless of the presence or location of pWMLs. Thus, even when present by chance in an MRS voxel (~3 cm³), the size of the pWMLs (~1 mm in diameter) precluded them from accounting for the majority of the tissue volume. Accordingly, the decrease in NAA observed in association with the pWMLs most likely reflects a widespread characteristic of the parietal white matter tissue and not merely the markedly small tissue volume accounted for the pWMLs themselves.

NAA is synthesized in the mitochondria of neurons/axons, diffuses down the axoplasma, and is degraded by mature oligodendrocytes by aspartoacylase providing acetyl groups for myelin lipid synthesis [18, 19]. Accordingly, NAA concentration in a given tissue volume can be affected by the number of mature neurons/axons, the density of mitochondria within the neurons/axons, oxidative metabolism, and the number/metabolic capacity of mature oligodendrocytes in the surrounding tissue. Additionally, there is note of limited data from cell culture that immature oligodendrocytes may have some capacity for synthesizing NAA [39]. Thus, while not necessarily indicative of axonal loss per se, the reduced NAA observed in association with pWMLs does at least suggest that there has been a reduction in number or function of their mitochondria, consistent with axonal damage, and could be primary or secondary to diffuse neuronal loss in the thalamus or cortex. Alternatively, the decreased NAA may reflect an immature state of axonal development; however, it should be noted that despite increasing with age, NAA concentration remained lower throughout the age range included suggesting that there has been an overall loss in number or in the functional capacity of axons in the parietal white matter.

Neuroimaging biomarkers of outcome

Considerable research has focused on developing a biomarker that could predict neurodevelopmental outcome for neonates born preterm, providing not only accurate prognosis for clinicians and families but also affording the opportunity to initiate rehabilitation before critical periods have passed and neurodevelopmental deficits have emerged. In the past decade, there have been many noteworthy studies demonstrating, in general, that the more white matter injury features observed in the neonatal period, the worse the long-term neurodevelopmental prognosis [2, 3, 15]. However, when considering specific features alone (e.g., pWMLs, DEHSI), the results have been inconsistent [5, 14, 16, 40]. The findings from this study suggest that specific features alone may not be enough to signify abnormal development, and that rather the combination of features (e.g., pWMLs and DEHSI) may be a better predictor for adverse developmental outcome.

Prior studies of brain metabolism in preterm neonates with white matter injury

To date, there have been remarkably few published studies examining the association between white matter injury and brain metabolism in preterm neonates. Interestingly, an early study by Robertson and colleagues, which compared a small group of preterm infants with destructive white matter lesions (i.e. cystic PVL and hemorrhagic parenchymal infarction) to preterm infants without such lesions failed to demonstrate differences in the ratio of NAA/creatine in association with white matter damage; however, the ratio of myoinositol/creatine was higher in the white matter injury group [41]. On the other hand, a recent study did demonstrate decreased NAA/creatine in preterm neonates with white matter injury—characterized predominantly as punctate white matter lesions—imaged in the preterm period; however, analyses based on white matter injury subtypes were not conducted [32]. Additionally, NAA concentration has been shown to be diminished in neonates with hypoxic-ischemic encephalopathy and in fetuses and neonates with congenital heart disease, and in those studies, predictive of long-term outcome [22, 42, 43].

Limitations

A biomarker specific to astrogliosis in neonates remains elusive. In this study, as in many previous studies, myoinositol, an osmolite present in high intracellular concentration in astroglia was used as a potential surrogate indicator of astrogliosis. However, near term equivalency, myoinositol decreases dramatically as a function of age, coincident with the transient proliferation of microglia in the white matter [44]. Thus, myoinositol is already a challenging marker to use for astrogliosis in a term-equivalent neonate due to the rapid underlying developmental changes during this period. Moreover, as demonstrated here, the role of myoinositol as an osmolite further confounds its potential role as an astroglial marker. Further work is needed to develop a more specific marker for astrogliosis in preterm white matter injury.

Additional limitations in this study include the evolving nature of the signal abnormalities characteristic of pWMLs and DEHSI and the timing of our clinically indicated MRIs. As recently demonstrated in a large, prospective cohort of preterm infants, pWMLs are most readily apparent on neuroimaging studies completed soon after birth, but may still be apparent at term-equivalency or later [45]. In contrast, DEHSI, as defined in most neuroimaging studies, is a feature apparent at term equivalency, representing the sequelae of prior injury and/or ongoing gliosis with impaired oligodendrocyte maturation. Indeed, both pWMLs and DEHSI have been associated with gliosis (the former with microglial activation as noted in the introduction), and it is for this reason that we investigated the relations between both lesions and myoinositol concentrations in the present study. However, given the difference in the timing of the lesions on MRI, a more ideal study design would have been to conduct both early and term-equivalent MRIs and then investigate associations between pWMLs and DEHSI (classified retrospectively from the term-equivalent exam) and MRS biomarkers at both time points. Relatedly, there has been variability in the manner in which pWMLs and DEHSI may be defined in neuroimaging studies, and indeed, other studies have distinguished isolated pWML lesions from clusters or conglomerates [45], as well as whether the “posterior periventricular cross-roads” were visible in DEHSI [5, 40]. Future studies should consider whether further WMI subtypes can be distinguished metabolically. Finally, our inclusion of both a broad range of gestational ages (i.e., 23–36 weeks) and postconceptional ages at MRI (i.e., 25–60 weeks) is a further limitation of this study and may have confounded our ability to relate metabolic markers of axonopathy and astrogliosis to the observed neuroimaging patterns.

Conclusion

Among preterm neonates imaged in infancy, punctate white matter lesions, particularly in conjunction with moderate to severe DEHSI, is associated with diminished NAA diffusely in the white matter. Additionally, the presence of pWMLs in combination with DEHSI altered the normal developmental decline in myoinositol concentration during this period and further suggested the disruption of astroglial function and/or osmotic dysregulation. Overall, pWMLs in conjunction with moderate to severe DEHSI may be the most potent conventional imaging biomarker signifying a population of infants at high risk for long-term adverse neurodevelopmental outcome due to white matter injury and associated axonopathy.