NORMAL CANINE BRAIN MATURATION AT MAGNETIC RESONANCE IMAGING

Introduction

The brain of altricial mammals, such as primates, dogs, cats, and rats, is incompletely developed at birth.^1–5 This is in contrast to that of precocial mammals, such as the horse, which exhibit a more mature appearance shortly after birth.^6,7 A major feature of postnatal brain maturation is continuation of neuronal myelination that began in utero. The development of the postnatal brain, and particularly the progressive myelination of the cerebral white matter as observed with magnetic resonance imaging (MRI), and correlated with histologic observations, is thoroughly described in humans, and this database is used routinely as a normal reference when evaluating abnormal neonates.^8,9 For the dog, there are few similar data, and what is available is not in a form suitable for use as a reference during evaluation of clinical cases.^4,7 The purpose of this report is to describe the qualitative changes observed with various MRI sequences during normal canine brain maturation and to identify clinically useful milestones for the transition from juvenile to mature patterns of cerebral gray and white matter relative signal intensity.

Materials and Methods

MR brain images of 17 sexually intact dogs, born naturally following normal gestation and parturition, were acquired at various times after birth. Gestation varied from 60 to 64 days, with no special consideration given to this variation when evaluating the images. The dogs used for this report originated from a breeding colony of individuals genetically affected with one of several forms of mucopolysaccharidosis (MPS): MPS-I, MPS-IIIB, or MPS-VII. These MPS variants are autosomal recessive disorders, and the individuals used for this brain maturation study underwent DNA testing before imaging to confirm they were genetically normal or were a carrier of only one allele.^10–13 The colony consists of mixed breed dogs that were of predominantly beagle, schipperke, and hound lineage with an average adult body weight of approximately 12–16 kg. At 1 week, the individuals weighed approximately 0.5 kg and at 36 weeks approximately 10 kg. All dogs were assessed as physically, behaviorally, and neurologically normal during their development.^14,15

For the MRI of dogs aged 1–8 weeks, anesthesia was induced using isoflurane^* gas masking with subsequent intubation and maintenance using isoflurane gas to effect. For dogs aged >8 weeks, butorphenol† and atropine‡ were administered intramuscularly, followed by intravenous propofol§ with subsequent intubation and maintenance using isoflurane gas.

MR brain images were acquired using a 1.5 T system and 8-channel HR Brain coil.¶ Each study consisted of nine 2D series: T2-weighted (T2W) dorsal, transverse, and sagittal images; T1-weighted (T1W) transverse and sagittal images; T2W fluid attenuation inversion recovery (FLAIR) transverse images; short tau inversion recovery (STIR) transverse images; and diffusion weighted imaging (DWI) in dorsal and transverse planes (Table 1). Apparent diffusion coefficient (ADC) maps were constructed from the DWI data using the proprietary scanner software. There were a total of 35 studies from ages 1–36 weeks postnatal (Table 2). The MRI were reviewed by a board-certified radiologist and a third-year radiology resident (E.R. and B.G.). Qualitatively, signal intensity of structures being evaluated was referenced against that of cortical gray matter intensity on the same series. In addition, subjective evaluation of structures relative to that observed in adults, based on published reports, a standard anatomic text, and from personal experience, was documented.^16–18 ADC values were measured on dorsal and transverse plane ADC maps in the subcortical white matter of the left occipital lobe using a 0.145 cm² region of interest and plotted as a function of age.

During each time point of weeks 1–12, one dog underwent euthanasia immediately following MRI, and the brain was collected immediately and prepared for histologic evaluation. Each brain was placed in 10% buffered formalin for 1 week. After fixation, each brain was placed in Bouin's solution for 24 h to increase tissue firmness. The brain was rinsed with alkaline alcohol and sectioned transversely with the intent being to place the slices in roughly the same locations as the transverse MRI. Brain sections were processed in an automatic tissue processor and embedded in paraffin. Five-micrometer-thick sections were obtained and mounted on positively charged glass slides to be stained with hematoxylin and eosin and myelin-specific luxol fast blue stains. Microscopic examination was performed by a board-certified pathologist (D.G.-T.) who was unaware of the MRI appearances. For the microscopic examination, particular attention was given to the appearance and changes in the subcortical white matter, internal capsule, and corpus callosum.

Results

At each postnatal age, the different individuals exhibited similar signal characteristics on all sequences. The signal intensity of the subcortical white matter was characterized by a signal reversal on all sequences. Initially, the relative intensity of white vs. gray matter was reversed from that observed in the adult. This was followed by a transition phase where the white and gray matter were isointense to each other. Finally, there was a transition to the white and gray matter relative signal intensity characteristic of that observed in the adult, with progressive refinement to the appearance of a mature brain. On FLAIR images, the white matter exhibited more signal transitions than in T1W or T2W images, similar to that in human brain but with earlier milestones of transition¹⁹; this is described in detail below. On all sequences, signal intensity changed from caudal to rostral at the brainstem, and from central to peripheral at the internal capsule and cerebrum (Table 3).

The corpus callosum, as a distinct feature, varies in its temporal appearance on various sequences as described below. However, on all sequences, the complete corpus callosum from genu through body and splenium was not distinctly identifiable at 3 weeks or earlier but was clearly distinguishable at 8 weeks and later, and had an adult appearance at 16 weeks.

On T1W, T2W, and STIR images, the cerebellum had a mature appearance by approximately 6 weeks, while the cerebrum had a predominantly adult appearance at 16 weeks.

In FLAIR images, there was a delay in the appearance of brain maturation compared with other sequences, as described below. There was continued volume expansion and mild refinement of subcortical white matter arborization and border distinction continuing to 36 weeks. A separate observation was that the normal neurohypophyseal focal T1-hyperintensity was recognizable at all ages.^20–22

T2W white matter development was evaluated in dorsal (Fig. 1), sagittal (Fig. 2), and transverse (Fig. 3) images. Dorsal images of Fig. 1 are at the level of the dorsal internal capsule and are representative of the subcortical white matter changes throughout the rostral and caudal portions of the telencephalon. The sagittal images of Fig. 2 are at midline and were thought to be the most useful for assessment of the cerebellum and corpus callosum. The transverse images of Fig. 3 are at the level of the interthalamic adhesion; this location was chosen because it depicted the relevant gray and white matter anatomy at the mid-telencephalon and because it was the most consistently acquired image location for all individuals at all ages.

At 1 week, the brainstem was isointense while the subcortical white matter was T2-hyperintense (Fig. 2). From 2 to 6 weeks, the brainstem became more T2-hypointense (Fig. 2). Within the brainstem, the caudal colliculi, the trapezoid bodies, lateral lemnisci, and cerebellar peduncle were hypointense to the remainder of the brainstem (Fig. 4). These areas became isointense to the remainder of the brainstem by 6 weeks, when the brainstem exhibited a mature appearance. During the same time, there was rapid development of the cerebellar arbor vitae, with an adult appearance to the cerebellum at 6 weeks (Fig. 2). The internal capsule was mildly T2-hypointense at 2 and 3 weeks and was distinguishable as a T2-hypointense feature on dorsal and transverse images at 4 weeks of age (1, 3). On T2W image identification of the internal capsule at weeks 2 through 4, there was no consistent bias toward which part, rostral or caudal crus, was identified earliest. The corpus callosum was poorly and inconsistently identified at 4 weeks but was clearly evident as a thin and sometimes incomplete T2-hypointense feature on mid-sagittal images beginning at 6 weeks. The complete corpus callosum, including the genu, body, and splenium, was not consistently seen until 8 weeks.

The cerebral white matter was uniformly T2-hyperintense during the first 4 weeks but its signal decreased visibly during this time (1-3). At 6 weeks, the subcortical white matter was isointense or variably mildly hypo- and hyperintense, with all individuals having a mild heterogeneous appearance to the cerebral cortex (Fig. 3). At 8 weeks and later, the subcortical white matter was T2-hypointense, with its signal gradually decreasing, until 16 weeks when the white matter appearance was similar to that of an adult (1, 3). A mild appreciable decrease in relative intensity and increase in relative volume of the white matter continued to 36 weeks (1, 3).

T1W white matter progression was assessed in sagittal and transverse images (5, 6). The brainstem exhibited a mildly heterogeneous isointensity at 1 week that progressed to a mature uniform mild T1-hyperintensity by week 4 (Fig. 5). The cerebellum also exhibited a mature appearance by week 4 (Fig. 5). At 2 weeks, the internal capsule was inconsistently identified as a faint T1-hyperintensity, at 3 weeks it remained faint but was consistently observed, and by 4 weeks it was a clearly distinguished T1-hyperintense structure (Fig. 6). On T1W image identification of the internal capsule at weeks 2–4, there was no consistent bias toward which part, rostral or caudal crus, was identified earliest. The corpus callosum was poorly and inconsistently observed at 2 and 3 weeks but at 4 weeks it was clearly identifiable on mid-sagittal images, although thin and sometimes incomplete. At 6 weeks, the corpus callosum was similar in appearance to its appearance in the adult (Fig. 5).

As in T2W images, the subcortical white matter underwent three phases in T1W images, but its relative intensity is opposite from that observed on the T2W images, and with an earlier onset of, and more rapid progression through, the transition to its adult appearance. Subcortical white matter T1-hypointensity was observed during the first 2 weeks (Fig. 6). The transition to a more intense region occurred between 3 and 4 weeks (Fig. 6). At week 3, where there was a 4-day postnatal age variation among the individuals imaged, the white matter was variably mildly T1-hypointense to isointense, with the older two individuals (23 days) having uniformly T1-isointense white matter, and the younger two individuals (20 and 21 days) having predominantly mildly T1-hypointense white matter, with the exception of T1-isointensity at the rostral pole of the cerebrum. At 4 weeks, all individuals had faint but consistent white matter T1-hyperintensity at the rostral pole of the cerebrum with the remainder of the subcortical white matter isointense. At 6 weeks, the subcortical white matter was characterized by thin branching hyperintense streaks and up to 16 weeks there was a progressive extension and expansion of the T1-hyperintensity throughout the entire volume of subcortical white matter to an adult appearance (Fig. 6). There were only mild changes observed between 16 and 36 weeks, with the original marked T1-hyperintensity becoming more subdued and diffuse with age (Fig. 6).

White matter progression in STIR images was assessed in the transverse plane (Fig. 7). The intensity changes in white matter on STIR images were similar to that seen on T2W images. From 6 weeks on, the myelinated white matter was characterized by a pronounced improvement in the conspicuity of the corpus callosum, the internal capsule, and subcortical white matter on STIR images (Fig. 7) compared with T2W images (Fig. 3).

White matter progression on FLAIR images was assessed in the transverse plane (Fig. 8). On FLAIR images, the brainstem remained isointense through 4 weeks and at 6 weeks exhibited a heterogeneous mild hypointensity that progressed to a mature appearance at 12 weeks. The internal capsule was poorly distinguished during the first 6 weeks; however, from 8 weeks onward, it was a conspicuous hypointense structure (Fig. 8). The corpus callosum was first seen at 8 weeks and was distinguishable in its entirety at 12 weeks and onward. Distinct from T1W, T2W, and STIR images, the subcortical white matter exhibited five phases on FLAIR images (Fig. 8). There were two juvenile subphases with two subsequent transition phases and one maturing phase, which was delayed relative to that observed on the other sequences. This is similar to that described in humans.²³ At weeks 1 and 2, the cerebral white matter was distinctly hypointense. At 3 weeks, there was a shift toward mild hyperintensity. At 4 and 6 weeks, the subcortical white matter was distinctly hyperintense. At 8 weeks, there was a second transition to generalized white matter hypointensity, with mild variation between individuals. Finally, after 12 weeks, the subcortical white matter remained hypointense, as in the adult. Through 36 weeks, there was continued progression with arborization and expansion of the hypointense regions. This maturation was slower than in the T2W, T1W, and STIR images (Fig. 8).

The subcortical white matter exhibited a progressive decrease in diffusivity with age. This was appreciated as a progression on the ADC maps (Fig. 9A) similar to that of the T2W images (Fig. 3), with the measured ADC decreasing throughout the period evaluated. The ADC decreased most rapidly during the first 4 weeks and then gradually after that (Fig. 9B).

Histologically, there was a progression of cerebral myelination from 1 to 12 weeks, consistent with other data.¹ The internal capsule and corpus callosum, as well as the trapezoid bodies, medial and lateral lemnisci, and cerebellar peduncles, demonstrated the earliest signs of myelination, evident at 1 week. A major feature in the corpus callosum and internal capsule at 1 week was the presence of interstitial water with minimal numbers of haphazardly organized mononuclear round cells (oligodendrocyte-like cells) (Fig. 10A), while at 2 weeks, these regions begin to have higher organization and formation of rows or parallel cords of cells surrounding parallel myelinated fibers (Fig. 10B). At week 4, the main change was the marked reduction in interstitial fluid and the beginning of intercrossing organization of myelinated fibers (Fig. 10C). Little difference existed from weeks 4–6, with the exception of an apparent reduction in the concentration of cells (Fig. 10D), which is most likely produced by the thickening of the myelinated axons due to an increase in myelin deposition. A dense appearance of the myelinated fibers with reduction in the size and numbers of oligodendrocyte-like cells was observed at week 8 (Fig. 10E), with little change at week 12 (Fig. 10F).

The progression of myelination in the subcortical white matter was slower than in other areas such as the trapezoid bodies, internal capsule, or corpus callosum, exhibiting a large accumulation of interstitial fluid at 1 week of age (Fig. 11A), with minimal reduction at week 2 (Fig. 11B). Less fluid was observed at 4 weeks (Fig. 11C), and there was better organization of the oligodendrocyte-like cells. At week 6 (Fig. 11D), moderate myelination was present, while interstitial fluid was still identified. By 8 weeks (Fig. 11E), the myelinated fibers had a higher organization and the amount of fluid was minimal, giving a more dense appearance to the myelinated tissue. At 12 weeks (Fig. 11F), the subcortical white matter appeared denser and intercrossing fibers were present. At 1 week, minimal myelination was observed in most of the subcortical white matter. Even at this early stage, however, a higher concentration of myelinated fibers was observed in the frontal lobes than in the parietal and occipital lobes. The mildly higher level of myelination in the frontal lobe than the occipital lobe remained evident at 12 weeks.

For comparative purposes, transverse images from two regions acquired using the sequences employed in this study are shown in Fig. 12.

Discussion

The changing appearance of the maturing brain on the various MRI sequences can be attributed to several combined processes, with their subsequent complex effects on T1- and T2-relaxation time and proton density.^1,24–26 There was a rapid decrease in the relatively high extracellular water content of the neonate brain. In addition, there was development of axons and dendrites and proliferation and differentiation of glial cells, particularly in the subcortical white matter. Concurrent with these changes was the progressive development and extension of myelin from oligodendrocytes. Myelin contains a relatively high proportion of lipid; however, this lipid is in a different form from that of adipose tissue and this affects its T1 and T2 relaxation. Myelin is a sheath consisting of multiple layers composed of glycolipids, phospholipids, proteolipids, cholesterol, and proteins. Some of these are hydrophobic while others interact strongly with water. The process of myelination consists of both production of an initial sheath, with composition similar to an oligodendrocyte cell membrane, and a transition to the mature compact structure. During this maturation process, the chemical composition and the structure of the sheath undergo changes to the final compact form. There is intra-axonal and extracellular water outside the sheath as well as water molecules within the layers of the sheath. The strong interaction of the intramembranous water with several of the component molecules is thought to result in a pronounced T1-shortening effect on this component of the water. At the longer echo times of T2W signal generation, the reduced signal intensity observed with myelination is attributed to the effect of the decreasing axonal and extracellular water content being stronger than that of increased lipids.²⁴

The process of myelin formation and maturation continues into early adulthood following the dramatic rapid early changes observed in the juvenile, as late as into the second or even third decade of life in humans.^7,8,27 It is not established clearly how long similar changes take in the dog but, based on our observations, they likely continue to and beyond 36 weeks, the last imaging time in our study. This is supported by a previous observation that the lipid percentage of dry matter in developing canine brain gradually increased up to 1 year.¹

Based on our results, the MRI features of normal canine brain maturation parallels that in humans, but proceeds at an accelerated rate. The brainstem and cerebellum mature first, followed by the cerebrum, in a central to peripheral pattern. The changes observed on T1W images preceded those on T2W and STIR images, with changes seen in FLAIR images being even more delayed. All MRI changes lagged behind histologic changes. At the brainstem, the auditory system regions of the trapezoid bodies, cerebellar peduncles, lateral lemnisci, and caudal colliculi were clearly identified areas of early myelination with histology, and this correlated as distinct areas of early T2W hypointensity in all individuals at 2 weeks of age attributed to the associated decrease in relative water content in these regions. This is consistent with findings of early myelinating structures in humans as well that previously described in the dog.^1,28,29

Three observations made with MRI and histology in developing human brain were not observed in developing canine brain. First, the posterior limb of the internal capsule (caudal crus) in humans consistently has early myelination that precedes myelination in the anterior limb (rostral crus).^30–32 In the dog, we observed simultaneous or inconsistent myelination throughout the internal capsule. A previously described central origin of myelination in the dog was not observed in our dogs.¹ Second, in humans and other primates, myelination occurs in the subcortical white matter of the occipital lobes before myelination in the frontal lobes.^3,4,25,33 Based on our observations, the opposite occurs in the dog with myelination of the rostral pole of the cerebrum preceding that of the caudal pole. This is supported by the appearance of the T1W images at 3 and 4 weeks. Our findings are supported by others.¹ Third, there are conflicting reports in humans whether the corpus callosum develops sequentially starting with the splenium, the genu, or body, and in what direction it proceeds.^32,34,35 We observed the corpus callosum to develop simultaneously throughout its length, without a consistent nidus or directionality.

The imaging parameters we used are also commonly used for the evaluation of the adult canine brain, with the exception of the STIR images, which are not a sequence used commonly for the assessment of canine brain disease. STIR images are, however, often used in human pediatric MRI protocols.³⁶ Inversion recovery sequences with short inversion times providing more T2W (STIR) and medium inversion times providing more T1W (T1 FLAIR) are useful for increasing the contrast between the cerebral cortex and subcortical white matter, which is undergoing normal myelination or is fully myelinated.^37,38 For STIR images, the specific parameters needed for myelin attenuation are generally determined for individual systems.^39,40 We did not conduct a trial to optimize contrast in STIR images, rather we used those based on prior experience in adult dogs.

For routine evaluation of the immature canine brain, our results suggest that, as with the adult, standard transverse and sagittal T2W and T1W images are adequate. The added benefit of FLAIR images is debatable for human infants, depending on their age.^19,36,41 The added value of FLAIR images is questionable in dogs aged <3–4 months based on our observations in the normal dog. This is because from 3 weeks until 12 weeks of age, the subcortical white matter is mildly hyperintense to gray matter on FLAIR sequences and thus the conspicuity of most lesions that result in increased T2W hyperintensity at the periventricular regions or white matter would not be improved as they are in the adult. STIR images may provide useful information, or increased diagnostic confidence available from T2W images alone, for leukodystrophies, demyelinating diseases, and cortical dysplasia in patients with intractable seizures. Some of these conditions have been described in the dog, while others are currently recognized only in humans.^42–45 This supposition will require confirmation with future studies involving pathologic conditions. Unfortunately, T1 FLAIR images were not included in this study. They may allow earlier and more conspicuous evidence of myelination than T1W images. The diffusivity within the subcortical white matter of the neonate canine proved to decrease during maturation, as expected with the cellular changes occurring in this region. DWI has proven useful in the early evaluation of animal model and human neonates suffering from hypoxic ischemic encephalopathy and is a matter of current research; however, the clinical utility of this as applied to the dog remains to be determined.^46–49

In conclusion, the normal temporal changes of the developing canine brain that are apparent in MRI are orderly and predictable, correlate with histologic observations of myelination and decreasing water content, and are generally consistent with those observed in humans, but proceed at an accelerated pace. These changes reflect the combined effects of decreasing water content and increasing myelin formation and maturation. Understanding the normal progression of changes observed in MRI during the development of the immature brain should reduce the risk of misinterpreting normal changes as disease and failing to recognize altered development.