Reduced Occipital Regional Volumes at Term Predict Impaired Visual Function in Early Childhood in Very Low Birth Weight Infants

Abstract

purpose. Premature infants are at increased risk of impaired visual performance related to both cortical and subcortical pathways for oculomotor control. The hypothesis for the current study was that preterm infants with impaired saccades, smooth pursuit, and binocular eye alignment at age 2 years would have smaller occipital brain volumes at term equivalent, as measured by volumetric magnetic resonance (MR) techniques, than would preterm infants without such abnormalities.

methods. Study participants consisted of 68 infants from a representative regional cohort of 100 preterm infants born between 23 and 33 weeks’ gestation. At term equivalent, all infants underwent MR imaging, and the images were coregistered, tissue segmented into five cerebral tissue subtypes, and further subdivided into eight regions for each hemisphere. At 2 years corrected, all infants completed a comprehensive orthoptic evaluation performed by a single examiner.

results. Twenty-four (35%) of the 68 infants had abnormal oculomotor control at 2 years, including abnormalities in saccadic movements (n = 7), smooth pursuit (n = 14), or strabismus (n = 9, four with esotropia and five with exotropia). When compared with preterm infants without visuomotor impairment, these infants had significantly smaller inferior occipital region brain tissue volumes bilaterally (n = 24 vs. n = 44; total tissue, mean ± SD, left, 37.9 ± 7.4 cm³ vs. 43.7 ± 7.4 cm³; mean difference [95% CI] −5.7 [−9.4 to −2.0] cm³, P = 0.003; right, 36.8 ± 7.1 cm³ vs. 41.4 ± 6.2 cm³, mean difference −4.6 [−7.9 to −1.3] cm³, P = 0.007). This difference remained significant after adjusting for intracranial volume (ICV; left, mean difference −3.5 [−6.7 to −0.2] cm³, P = 0.04; right, mean difference −2.4 [−5.2 to −0.4] cm³, P = 0.09). Within this region, the cortical gray matter volume was the most significantly reduced (left, 20.4 ± 6.2 cm³ vs. 25.4 ± 5.6 cm³, mean difference −3.1 [−5.7 to −0.5] cm³, P = 0.02; right 21.0 ± 5.4 cm³ vs. 24.9 ± 5.0 cm³, mean difference –2.2 [−4.4 to 0.0] cm³, P = 0.05, ICV adjusted). Abnormalities in saccadic eye movements accounted for the largest effect on inferior occipital regional brain volumes (left side, P = 0.02).

conclusions. Volumetric MR imaging techniques demonstrated an overall reduction in the inferior occipital regional brain volumes in preterm infants at term corrected who later exhibit impaired oculomotor function control. These findings assist in understanding the neuroanatomic correlates of later visual difficulties experienced by infants born prematurely.

Infants born prematurely are at a significantly increased risk of visual abnormalities. ¹ Some of these abnormalities, such as retinopathy of prematurity (ROP), are manifest early and are thought to result from more direct injurious factors such as reactive oxygen species. ² Other visual disturbances manifest later and may occur as a result of immaturity and the vulnerability of the preterm brain to a more global insult, resulting in alterations in the pathways and functions of the occipital cortex. ^{3 4}  

Saccades, smooth pursuit and vergence are eye movement systems that allow the image of an object to be brought onto the fovea of both eyes, for optimum image resolution, and then help maintain it there. Both saccadic and smooth pursuit conjugate eye movements rely on connectivity through the lateral geniculate nucleus, fanning out through the deep white matter to reach the primary visual cortex in the occipital lobe. The primary visual cortex (Brodmann area 17) is contained by the striate cortex at the calcarine sulcus, a deep groove on the medial surface of the occipital lobe. ⁵ The striate cortex contains neurons that respond to moving visual stimuli ⁶ and is crucial for the control of visually guided movements. ⁷ Lesions of the striate cortex induced in primates are known to impair eye movements due to the lack of visual input. ⁸ Observed deficits are greater with larger lesions, and smooth pursuit tends to be more impaired than saccades. ⁹  

More complex visual stimuli cannot be fully analyzed within the striate cortex, and further information processing is thought to be performed in the middle temporal visual area (MT or V5) and the medial superior temporal (MST) visual area. ^{7 10} Saccades triggered by a visual stimulus are dependent on these structures, as confirmed with functional magnetic resonance (MR) imaging. ¹¹ The primary visual cortex is also the first locus in the central nervous system where visual input from both eyes is combined. ¹² Maldevelopment of this region of the cortex in the primate model has been associated with unrepaired natural, infantile-onset strabismus. ^{13 14}  

Preterm infants are prone to the effects of cerebral immaturity and white matter injury (WMI). ¹⁵ Periventricular leukomalacia (PVL) comprises both focal periventricular necrosis and diffuse cerebral WMI. ¹⁶ Cystic PVL in the preterm infant has a particular predilection for the occipital region of the brain. ¹⁷ Strabismus and amblyopia are more prevalent in low birth weight preterm infants ¹ and infants who have had perinatal hypoxia. ^{18 19}  

MR imaging techniques in the preterm infant at term equivalent have demonstrated diffuse WMI injury as the commonest neuropathic condition in the preterm infant, and such a condition has been shown to be associated with reductions in total cerebral volumes. ^{15 20} However, the relationship of such global or regional reductions in cerebral volumes to later function remains unknown. 

We postulated that preterm infants with later abnormalities of the visual system that are more specific to the posterior pathways, such as saccades, smooth pursuit, and binocular eye alignment, would be characterized by smaller occipital lobe volumes as measured by volumetric MRI techniques at term. 

Patients and Methods

Study participants were members of a regional cohort of 100 very low birth weight (<1500 g) infants and/or below 33 weeks’ gestation who were consecutively admitted to a level III neonatal intensive care unit (NICU) at Christchurch Women’s Hospital (New Zealand) from July 1998 to November 2000. Infants with congenital abnormalities or whose parents did not speak English or who had moved out of the region at term were excluded. Over the recruitment period, 119 preterm infants were eligible for inclusion in the study. Of these infants, 10 died before term, 4 were missed, and 5 declined to participate. Excluding those who died, 92% of all eligible infants were recruited with informed consent. Eighty infants who had MR scans at term corrected completed full visual function testing at 2 years of age. The study was conducted in accordance with the provisions of the Declaration of Helsinki. 

MR Image Acquisition

MR imaging was performed without sedation, after each infant was fed and wrapped in a bean bag (Vac Fix; S&S X-ray Products, Brooklyn, NY). Scans were performed with a 1.5-tesla system (Signa; GE Medical Systems, Milwaukee, WI). A three-dimensional Fourier transform spoiled-gradient recalled sequence (1.5-mm coronal slices; flip angle, 45°; repetition time, 35 ms; echo time, 5 ms; field of view, 18 cm; matrix, 256 × 256) and a double echo (proton density and T₂-weighted) spin-echo sequence, 36 and 162 ms; field of view, 18 cm; matrix, 256 × 256; interleaved acquisition). 

MR Image Processing

Quantitative volumetric analysis was conducted on a computer workstation (Sun Microsystems, Mountain View, CA) by a single operator (DT). The image processing algorithms were used to reduce imaging system noise, align T₁ and T₂ images and segment the imaged volume (Fig. 1) . The segmentation method applied was a spatially varying statistical classification in which an anatomic template of a 40-week-old infant was used to modify the result of tissue classification. ²¹ The brain images were segmented into five different tissue subtypes: cortical gray matter, myelinated white matter, nonmyelinated white matter, cerebral spinal fluid (CSF) and deep nuclear gray matter. The intracranial volume was computed as the sum of these tissue subtypes. Intraobserver correlation coefficients were calculated by performing segmentations on five randomly chosen subjects five times each, at least one day apart. The intraclass correlation coefficient (95% confidence interval [CI]) was 0.84 (0.58–0.98) for cortical gray matter, 0.73 (0.38–0.96) for myelinated white matter, 0.83 (0.55–0.98) for nonmyelinated white matter, and 0.97 (0.90–1.00) for CSF. 

Comparison of regional brain volumes of different tissue classes was performed by dividing the brain into hemispheres, with each hemisphere then being further parcellated into eight regions by using a previously described methodology ^{22 23} (Fig. 2) . Each brain image was first manually registered into a standard orientation according to the Talairach atlas. ²⁴ Hemispheres were divided along the midsagittal plane, and each hemisphere was further divided into eight subregions (dorsal prefrontal, orbitofrontal, premotor, subgenual, sensorimotor, midtemporal, parieto-occipital [PO], and inferior occipital [IO] with cerebellum). The axial plane was passed through a line joining the anterior commissure (AC) to the posterior commissure (PC). The anterior limiting coronal plane was positioned at the most anterior part of the genu of the corpus callosum, the middle coronal plane at the anterior border of the AC, and the posterior through the PC (Fig. 2) . 

Parcellation allows investigation of the distribution of the different tissue subtypes throughout all eight subregions. ²⁴ The IO region contains the primary visual striate cortex area (V1, Brodmann area 17) at the calcarine sulcus (Brodmann area 18) and the inferior aspect of Brodmann area 19. The MT (V5) and part of the MST areas of the cortex are also contained in this region (Fig. 2) . The whole of the cerebellum, the dorsal brain stem and the superior and inferior colliculi are also contained in this parcel, contributing to the gray matter segment obtained from the IO region. 

The PO region contains the superior aspects of Brodmann areas 18 and 19 and also the splenium of the corpus callosum. The last contains commissural fibers that run between the left and right visual cortex. 

Parcellation was performed twice on five randomly chosen images to calculate intrarater reliability. The intraclass correlation coefficient (95% CI) was 0.95 (0.67–0.99) for the left IO parcel, 0.98 (0.88–1.00) for the right IO parcel, 0.99 (0.91–1.00) for the left PO parcel, and 0.96 (0.75–1.00) for the right PO parcel. 

MR Image Qualitative Analysis

MR images were qualitatively analyzed by a single blinded rater (TI) for WMI by evaluating and scoring the presence and severity of white matter signal abnormality, white matter volume reduction, cystic white matter abnormality, thinning of the corpus callosum and maturation of myelin. Using this score, a WMI grade of normal (grade I), mild (grade II), moderate (grade III), or severe (grade IV) was then assigned, as has been previously described. ²⁰ The SD of intraobserver differences as a mean of the average was <5%. 

Vision History and Ocular Examination

At age 2 years corrected, a complete ocular history and examination was undertaken for each infant, with particular attention to the presence of strabismus and abnormalities in saccades and smooth pursuit (Table 2) . The orthoptic and ophthalmic evaluation included assessments of best corrected visual acuity, visual fields by confrontation to assess for hemianopia, Lang stereopsis, motor fusion, ocular alignment, ocular motility, and pupil reflexes. After pupillary dilatation, media and fundus inspection and cycloplegic retinoscopy were performed. 

Orthoptic examination was performed by a single experienced tester (CG) who was “blind” to the children’s perinatal history and term MRI findings. Visual acuity was tested using Cardiff Acuity Cards, ²⁵ Kays Pictures, ²⁶ or fixation behavior. ²⁷ The Cardiff Acuity Test uses vanishing optotypes and relies on “forced preferential looking” at pictures of familiar objects. It is suitable for children aged between 12 months and 3 years, as well as older children with limited verbal communication skills. The Kay Pictures require children to be able to name simple objects. Other infants were observed for fixation behavior—in particular, whether it was central, steady, and maintained. 

Examination for strabismus included the cover-uncover and alternate cover test in primary position at near (0.33 m) and at distance (6 m). The presence or absence of nystagmus was noted with examination for any manifest or latent nystagmus. Extraocular movements were examined at near fixation in nine positions of gaze. Smooth pursuit was tested with a butterfly fixation object as a slowly moving target, at a distance of 0.33 m by direct observation. Assessment was made for each eye horizontally, nasal to temporal, and then temporal to nasal, and also vertically up and down. The result was deemed abnormal if the pursuit was not smooth, but had interrupted saccadic refixations (i.e., if “catch-up” saccadic eye movements were necessary to follow the target). 

Saccadic eye movements were tested by direct observation. Assessment was made horizontally, noting saccades in nasal to temporal then temporal to nasal directions and also vertically up and down. Two toy fixation targets were held 25 to 30 cm at a fixation distance of 0.33 m. Saccades were evoked by alternately drawing attention verbally to each of the salient targets. Saccadic eye movements were deemed abnormal if they were observed to be hypermetric (overshot the target by at least 30%), hypometric (if they fell short of the target by at least 30%), or abnormally slow. 

Visual tasks were repeated if abnormality was detected, to confirm the finding. The child’s attention and cooperation span were also noted if identified as limiting factors. 

Analysis of Results

The infant characteristics were compared using independent-samples t-tests and the Mann-Whitney statistic. The volumetric MR image analysis was performed in a blinded fashion with respect to vision-related outcomes and vice versa. The regional cerebral volumes for the abnormal visuomotor function control and the normal visuomotor function children were all normally distributed. Analysis of variance was performed for regional and total tissue volumes between the two groups adjusting for intracranial volume as a covariate. To compare the left and right regional volumes, a paired-samples t-test was used. Analysis was performed on computer (SPSS ver. 11.5; SPSS, Inc., Chicago, IL). 

Results

Of the cohort, 80 infants returned for vision assessment at 2 years. Three had died before the age of 2 years, 16 did not attend visual follow-up and one was registered blind due to ROP and did not undergo further vision testing. Of these 80 infants, 68 (85%) had post-MR image analysis at term. It was not possible to register the images of the remaining 12 for technical reasons, such as presence of motion artifacts. 

Thus, 32 male and 36 female infants were studied. The mean gestational age of the cohort of infants was 27.9 weeks, and the mean weight at birth was 1057 g. The characteristics of the infants in the visuomotor abnormality group compared with the remaining infants are shown in Table 1 . Those in the visuomotor abnormality group tended to be more premature and of a lower birth weight, but this difference did not reach statistical significance. The two groups were comparable in all other respects. Three of the 68 infants required cryotherapy for ROP. Laser ablation was not available at this site at the time of the study. There was no significant difference in WMI scores in the two groups of infants. ²⁰  

Vision Testing

For the 68 children who completed both MR imaging analysis and visual evaluation at 2 years of age, visual acuity was within normal range for age in all the children. The results of the visual function testing are summarized in Table 2 . Of the nine children with ocular misalignment, four had esotropia and five had exotropia on the cover-uncover tests. Of these nine children, four had a family history of strabismus. Of the children with exotropia, four had primary exotropia and one had exotropia after strabismus surgery for esotropia. On ocular motility testing, 14 children had abnormal smooth pursuit in both the horizontal and vertical planes. Seven children had abnormal saccades: five in both horizontal and vertical planes and two in just the horizontal plane. 

On comparison of the visual testing results between the visuomotor abnormality group (n = 24) and the remaining children (n = 44), the former had significantly greater impairment of binocular functions of stereopsis and motor fusion as would be expected (Table 2) . 

None of the children displayed any nystagmus. Optokinetic nystagmus was not tested. Eight (12%) of the children had anisocoria, but direct and consensual reflexes were normal in all the children. Fundus examination was performed in all but three children and was normal in all except one who showed peripheral retinal scarring. This patient had received cryotherapy for ROP. 

Volumetric MR Analysis

On analysis of regional cerebral volumes, children with abnormal saccades, smooth pursuit and/or strabismus had smaller IO tissue volumes compared with the rest (n = 24 vs. n = 44; total tissue, mean ± SD, left, 37.9 ± 7.4 cm³ vs. 43.7 ± 7.4 cm³, mean difference [95% CI] −5.7 [−9.4 to −2.0] cm³, P = 0.003; right, 36.8 ± 7.1 cm³ vs. 41.4 ± 6.2 cm³, mean difference −4.6 [−7.9 to −1.3] cm³, P = 0.007). This persisted on the left after adjustment for ICV (total tissue, left, mean difference −3.5 [−6.7, −0.2] cm³, P = 0.04; right, mean difference −2.4 [−5.2 to −0.4] cm³, P = 0.09; Table 3 ) and the presence of WMI (total tissue, left, mean difference −3.5 [−6.7, −0.2] cm³, P = 0.04; right, mean difference −2.4 [−5.1 to 0.4] cm³, P = 0.9). 

All the segmented tissue types were smaller in the IO region in the infants with abnormalities, but the largest reduction in tissue type occurred within cortical gray matter (left, 20.4 ± 6.2 cm³ vs. 25.4 ± 5.6 cm³, mean difference −3.1 [−5.7 −0.5] cm³; P = 0.02; right 21.0 ± 5.4 cm³ vs. 24.9 ± 5.0 cm³, mean difference −2.2 [−4.4 to 0.0] cm³, P = 0.05, ICV adjusted). The gray matter segment of the IO region contains most of the visual cortical gray matter including areas V1, MT, and MST as well as cerebellar gray matter. Within the visuomotor function impaired group, the infants with the abnormal saccades had the smallest IO volumes (left, 35.8 ± 6.2 cm³ vs. 42.3 ± 7.7 cm³, mean difference −5.9 [−10.8 to −1.0] cm³; P = 0.02, ICV adjusted). 

In the PO region, which contains the superior aspects of Brodmann areas 18 and 19 and also the splenium of the corpus callosum, there was no difference in the total tissue volumes in the visual abnormality group compared with the remaining children (left, 55.9 ± 8.3 cm³ vs. 59.4 ± 7.4 cm³, mean difference −0.1 [−2.5 to 2.4] cm³, P = 0.95; right, 55.50 ± 9.2 cm³ vs. 58.1 ± 7.4 cm³, mean difference 1.1 [−1.5 to 3.7] cm³, P = 0.43, ICV adjusted; Table 3 ). 

WMI made no independent impact on the findings when factored in as a covariate. In addition, the proportions of infants in both groups who had some impairment of the corpus callosum (including the splenium) on qualitative analysis was not significantly different (Table 1) . 

Of note, in all infants, the left IO parcel was larger than the right (41.6 vs. 39.8 cm³, mean difference 1.8 cm³ [1.0–2.6]; t = 4.47, P < 0.001, paired-samples t-test) as was the left PO parcel when compared with the right (58.2 vs. 57.2 cm³, mean difference 1.0 cm³ [0.2–1.8]; t = 2.49, P = 0.01, paired-samples t-test). 

There was no significant difference in the volumes of the other six regions of each hemisphere between the impaired oculomotor function group and the remainder of infants on both sides. 

Discussion

In our study, preterm infants with impairment in visuomotor control demonstrated smaller IO regional cerebral volumes. This finding is consistent with the known anatomic role of this region in the integration of visual information for the eye movement systems. This finding persisted after adjusting for total intracranial volume and WMI. Of the segmented tissue subtypes, cortical gray matter volumes were most significantly reduced in the infants with impaired visual function. Of the three related neuro-ophthalmic functions we studied, only abnormal saccades showed a strong association with a reduction in occipital volumes. 

There are clear limitations in our study, which we attempted to address in our methods. These include that the clinical testing of smooth pursuit and saccades has the potential for bias because of its subjective nature and is very challenging to undertake accurately, given the relatively short concentration span of a 2-year-old. To address this, the tests were performed by an experienced pediatric tester who repeated measurements as necessary. Saccades, for example, were only deemed abnormal if they fell repeatedly short or overshot the target by at least 30% of the distance between the two targets. Other recognized limitations in this study include the MR methodologies with the subjective nature of the qualitative evaluation of WMI and the imprecision of the quantitative MR analysis methods. The parcellation techniques are limited in their anatomic localization with the commissure and may vary in relation to the functional regions between individuals. The quantitative MR methods we have used to obtain volumetric data are identical with those used by Peterson et al. ^{22 23} who compared parcellation volumes in children born prematurely with those in term control subjects. Our use of a single experienced analyst in the quantitative and parcellation techniques was intended to reduce variability in the techniques, with robust intraobserver results. Preliminary work from our group suggests that such “in vivo” MR volumes appear to reflect changes in volumes, ²⁸ but further validation of these methods along with clearer anatomic-functional delineation is required. 

However, despite these limitations, our data support in vivo in the preterm infant that impaired visuomotor control is related to occipital region volumes by corrected term. This is neuroanatomically consistent with the role of the posterior neural tracks to the primary visual cortex in these ocular functions. ^{7 8 9 12} Our data would also support that alterations in structural development in this region that have occurred by 40 weeks’ corrected gestation continue to affect oculomotor control 2 years later. We postulate that we were unable to demonstrate differences in volume in other regions between the two groups of children because the visual functions are more diffusely distributed across these regions, and our techniques would not have the sensitivity to detect them. 

Saccades are described as pathologic when they over- or undershoot or cannot be induced voluntarily. Similarly smooth pursuit is abnormal if there is failure of initiation or cessation or if the velocity of pursuit does not match the target velocity. Strabismus is misalignment of the eyes. Saccadic eye movements, smooth pursuit, and vergence are all systems of oculomotor functions associated with visual cues with distinct anatomic substrates and physiological organization. ⁵ Such eye movements are a result of interplay between visual sensory input which localizes to the primary visual cortex through the optic radiations, and oculomotor movements that have much of their basis subcortically. Precise saccades and smooth pursuit are essential for normal visual development, and conversely normal visual function is essential for purposeful oculomotor control. 

The primary visual cortex and the prestriate cortex are important in providing visual information for both saccades and smooth pursuit, with activation also seen in the occipital and parietal cortex, ²⁹ but there are complex neural pathways that facilitate eye movements onward from the primary visual cortex, including connectivity to the frontal and temporal eye fields. 

For saccadic eye movements, these eye fields provide input to the brain stem saccade generator. ³⁰ Excitatory and inhibitory burst neurons then control the initiation and termination of saccades. Voluntary or learned saccades are controlled by the frontal eye field and are modulated via the cerebellar vermis or fastigial nucleus, whereas reflex saccades rely on connectivity via the parietal cortex and the superior colliculus. ³¹ Both, the frontal and parietal eye fields need cortical visual information for purposeful saccades. ^{30 31} Descriptions involving saccadic dysfunction have included faulty initiation, faulty accuracy or velocity, and involuntary saccadic intrusions onto a steady eye position or movement. 

For smooth pursuit, eye movement information is sent to the pontine nuclei and to the cerebellar cortex via the posterior parietal cortex and the MST areas and then onto the motor neurons of the three oculomotor cranial nerves. ⁶ As for saccades, the frontal and supplementary eye field are responsible for the voluntary control of predictable smooth pursuit. ^{32 33} In our data we were not able to detect any alterations in brain volumes in the frontal or temporal regions in relation to poor oculomotor function. This limitation may relate to a lack of sensitivity of our techniques with smaller volumetric reductions and may be better addressed in future studies by a combination of diffusion tensor techniques with investigation for these specific fiber tracts alongside volumetric evaluation. 

The IO brain volume parcel contains the primary, secondary, and tertiary visual cortices and adjacent structures including the MT and part of MST areas related to visual function. Other structures that are important in visuomotor function and lie within this parcel include the dorsal brain stem, superior colliculi, and cerebellum. A delineation of the primary or other areas of visual cortex was beyond the scope of this study. Brain stem-related structures are clearly important in the execution of eye movements, but defining more specific structural differences in the brain stem was also beyond the scope of the study. The relatively small volumes of the brain stem nuclei would be unlikely to contribute to the regional volumes studied. 

With its role in the generation of saccades, smooth pursuit, and the vestibulo-ocular reflex, the cerebellum is important in normal oculomotor function. In the present study, the cerebellum was included in the IO parcel. There is some evidence that the cerebellar volumes calculated using MR methods in preterm infants compared with term control subjects may be reduced in later childhood ^{22 34} and particularly in preterm infants with major neuropathic conditions. Reduction in cerebellar volumes in such preterm infants has also been correlated with impaired neurodevelopmental outcomes. ³⁴ Using similar techniques with manual outlining of the cerebellum in the preterm infant, we have previously documented the cerebellar hemisphere volume to be approximately 11 cm³, or contributing approximately 25% of the size of the IO parcel. ³⁵ Thus, in the present study, a reduction in cerebellar volume may well have contributed in part to the differences in IO regional volumes, but this is unlikely to provide the complete explanation for our findings. The presence of structural cerebellar changes in the newborn period remains to be confirmed, ³⁶ and its association with impaired oculomotor function is not independently confirmed in our analysis. 

In the present study, the proportion of infants with WMI on qualitative review of MR images at term was similar in both groups of infants. With our relatively small numbers, WMI or impaired corpus callosum development did not have a significant impact on the specific oculomotor functions we tested. There is more general evidence, however, for defects in vision, visual perception, and coordination in relationship to WMI in ex-premature infants, and in patients who have had cerebral irradiation and cytotoxicity. ^{3 37 38 39} In the present work, we attempted to study a representative cohort of preterm infants consecutively admitted to a newborn intensive care whereas other groups have reported preterm infants selected with specific and more severe cerebral injury in relation to visual outcomes. ^{3 4 37} There are few data relating MR imaging to oculomotor function in preterm infants. ^{4 19} However, children with occipital, cortical, and subcortical injuries have been shown to have a higher prevalence of exotropia and esotropia. ⁴⁰  

Our data suggest that the major cerebral tissue type responsible for the regional reduction in brain volume within the IO region is gray matter, of both cerebral and cerebellar origin. Decreases in cerebral gray matter volume may be related to alterations in gyral and sulcal development as well as the very complex integrity of the cortical layers. The neuropathic basis for the reduction in cortical gray matter volumes is unknown, although some evidence exists for injury in both neuronal and axonal elements in the severe cystic form of periventricular leukomalacia. ⁴¹ As only one of our infants displayed this severe cystic form of PVL we hypothesize that it is more likely that a sublethal alteration in neuroaxonal connectivity resulted in a deafferentation and secondary degeneration of the neuronal elements of the cortical gray matter. Whether such reductions in gray matter volume are secondary or primary events is unclear and is an important area for study in understanding the nature of both regional and global impact of prematurity. ^{42 43}  

A variety of visual abnormalities have been described in ex-preterm infants. ¹ There is preliminary evidence of abnormalities in eye movement in children with coordination difficulties and in children who have been premature, when compared with control subjects. ⁴⁴ Our study group had a high prevalence (35%) of abnormalities in smooth pursuit, saccadic eye movements, and/or vergence. Normal eye movements and coordination may be closely related in the functional coupling between perception and action. ⁴⁵ There is emerging evidence that abnormal eye movements may affect coordination and fine motor function ⁴⁶ and will contribute to later reading difficulties. ^{47 48}  

Finally, it is worthy of note that the left IO parcel was found to be larger than its counterpart on the right in all infants. Cerebral asymmetry has been well documented, ⁴⁹ and pathology studies have demonstrated that the left occipital lobe is wider than the right ⁵⁰ and the left inferior parietal lobule is larger than the right. ⁵¹  

In conclusion, our study demonstrated abnormalities in specific oculomotor systems in this cohort of preterm born infants at 2 years and evokes the hypothesis of a relationship with occipital region cerebral volumes, as measured by MR techniques. Abnormal visual functions in relation to immaturity and cerebral abnormality may contribute to the overall picture of neurosensory problems encountered by children who were born prematurely. 

 Supported by the Neurologic Foundation of New Zealand, Health Research Council of New Zealand, Lotteries Grants Boards of New Zealand, the Canterbury Medical Research Foundation, a research grant from the Whitaker Foundation; and National Institute of Mental Health Grant R21 MH67054 and National Institutes of Health Grant P41 RR13218.

 Submitted for publication June 26, 2005; revised December 13, 2005, and March 6, 2006; accepted May 22, 2006.

 Disclosure: D.K. Shah, None; C. Guinane, None; P. August, None; N.C. Austin, None; L.J. Woodward, None; D.K. Thompson, None; S.K. Warfield, None; R. Clemett, None; T.E. Inder, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

T₁-weighted (left) and T₂-weighted (middle) MR-images of a preterm infant at term, corrected, segmented into different tissue subtypes (right), by using a 40-week gestation template. This infant was born at 26 weeks’ gestation (birth weight, 886 g) and had grade-2 WMI.

Figure 2.

 

(a) Parcellation divides each hemisphere of the brain into eight subregions and allows the volumes of particular regions such as PO and IO to be studied. (b) A composite left lateral T₁-weighted parasagittal MR image of an infant at term showing the locations of regions of importance for eye movements, including the primary visual cortex (V1), the middle temporal visual area (V5), and the MST visual area in relation to the IO and OP regions. (c) Midsagittal T₁-weighted image of an infant at term showing the IO region with the striate cortex in the calcarine sulcus (bottom arrow), the cerebellum and the dorsal brain stem. The PO region contains the superior aspects of Brodmann areas 18 and 19 and also the splenium of the corpus callosum (top arrow).

Table 2.

 

Vision Function Test Results for All Patients

Examination	Results	Impaired (n = 24)			Remainder (n = 44)			P
Visual acuity		Left	Right	Left	Right
Test method (all)	6/12	2	2	9	9	0.56 left
	6/9.5	16	15	23	23	0.41 right
34 Cardiff cards	6/7.5	1	1	0	0
21 Kay’s pictures	6/6	0	1	4	4
13 Fixation behavior	CSM*	5	4	8	8
	UN, †	0	1	0	0
Visual field	No major abnormality	23	23	42	42	0.94 left
	Unable to test	1	1	2	2	0.94 right
Ocular alignment	Orthophoria	15			43			<0.001
	Esotropia	4			0
	Esophoria	0			1
	Exotropia	5			0
Lang stereopsis	Present	9			42			<0.001
	Absent	14			1
	Unable to test	1			1
15^ BO motor fusion	Present	11			39			<0.001
	Absent	13			2
	Unable to test	0			3
Nystagmus	None	23			44			0.18
	Unable to test	1			0
Smooth pursuit	Normal	10			44			<0.001
	Abnormal	14 (vertical and horizontal planes)			0
Saccades	Normal	17			44			<0.001
	Abnormal	7 (5 vertical and horizontal; 2 only horizontal)			0

*  Fixation behavior: central, steady and maintained.

†  Fixation behavior: unsteady and not maintained.

Table 1.

 

Patient Characteristics of Infants with Visuomotor Impairment Compared with Those of the Remaining Infants

	Impaired (n = 24) Mean, SD (Range)			Remainder (n = 44) Mean, SD (Range)			P
Sex	12 male, 12 female			20 male, 24 female			0.72
Multiple births	17 singletons, 7 twins			25 singletons, 19 twins			0.26
Gestational age at birth (wk)	27.3, 2.4 (23–33)			28.2, 2.3 (23–32)			0.10
Birth weight (g)	969, 287 (440–1600)			1116, 315 (460–1790)			0.06
Days on CPAP	15.5, 19.0 (0–58)			15.3, 18.7 (0–72)			0.92
Number requiring CPAP	21 (88%)			39 (89%)
Days ventilated	11.3, 19.7 (0–78)			6.4, 14.8 (0–67)			0.25
Number ventilated	20 (83%)			28 (64%)
Days on O2	51.1, 55.6 (0–214)			32.0, 40.5 (0–135)			0.11
Infants needing home O2	5 (21%)			7 (16%)
Cryotherapy for ROP	1 infant			2 infants
Highest stage ROP	Left eye	Right eye	Left eye	Right eye	0.19 left eye 0.18 right eye
0	12 (50%)	12 (50%)	30 (68%)	30 (68%)
1	7 (29%)	7 (29%)	10 (23%)	10 (23%)
2	3 (12%)	2 (8%)	2 (5%)	2 (5%)
3	1 (4%)	2 (8%)	2 (5%)	2 (5%)
WMI grade							0.81 WMI
1	6 (25%)			8 (18%)
2	13 (54%)			28 (64%)
3	4 (17%)			7 (16%)
4	1 (4%)			1 (2%)
Corpus callosum (CC)	4 (17%) normal			12 (27%) normal			0.29 CC
	13 (54%) isolated thinning			26 (59%) isolated thinning
	6 (25%) marked thinning			6 (14%) marked thinning

Table 3.

 

Regional Cerebral Volumes Compared between the Two Groups of Infants by Analysis of Variance after Adjusting for ICV

	Mean Volume ± SD (mL)			Mean Difference	95% CI	P
Infants with abnormal saccades, smooth pursuit, and strabismus compared with the remaining preterm infants
Region/tissue	Cases n = 24	Remainder n = 44
Inferior occipital region, left
Total tissue	37.9 ± 7.4	43.7 ± 7.4	−3.5	−6.7 to −0.2	0.04
Cortical gray matter	20.4 ± 6.2	25.4 ± 5.6	−3.1	−5.7 to −0.5	0.02
Myelinated white matter	1.4 ± 0.9	1.5 ± 1.3	0.1	−0.5 to 0.7	0.73
Unmyelinated white matter	15.8 ± 5.0	16.6 ± 16.6	−0.5	−3.0 to 2.1	0.72
Cerebrospinal fluid	1.5 ± 0.9	1.6 ± 1.1	0.1	−0.4 to 0.6	0.70
Inferior occipital region, right
Total tissue	36.8 ± 7.1	41.4 ± 6.2	−2.4	−5.2 to 0.4	0.09
Cortical gray matter	21.0 ± 5.4	24.9 ± 5.0	−2.2	−4.4 to 0.0	0.05
Unmyelinated white matter	14.0 ± 5.1	14.6 ± 5.0	−0.3	−2.9 to 2.4	0.84
Myelinated white matter	1.5 ± 1.0	1.6 ± 1.6	0.1	−0.7 to 0.8	0.88
Cerebrospinal fluid	1.4 ± 0.9	1.6 ± 1.3	0.0	−0.6 to 0.6	0.92
Parieto-occipital region
Total tissue (left)	55.9 ± 8.3	59.4 ± 7.2	−0.1	−2.5 to 2.4	0.95
Total tissue (right)	55.5 ± 9.2	58.1 ± 7.4	1.1	−1.5 to 3.7	0.43
Infants with abnormal saccades compared with the remaining preterm infants
Inferior occipital region (left)	Cases n = 7	Remainder n = 61
Total tissue	35.8 ± 6.2	42.3 ± 7.7	−5.9	−10.8 to −1.0	0.02
Cortical gray matter	18.2 ± 6.7	24.2 ± 5.9	−5.6	−9.4 to −1.7	0.005
Infants with abnormal smooth pursuit compared with the remaining preterm infants
Inferior occipital region (left)
Region/tissue	Cases n = 14	Remainder n = 54
Total tissue	39.7 ± 7.5	42.1 ± 7.8	−0.5	−4.3 to 3.4	0.81
Cortical gray matter	20.5 ± 7.4	24.4 ± 5.7	−2.4	−5.4 to 0.7	0.13
Infants with strabismus compared with the remaining preterm infants
Inferior occipital region (left)
Region/tissue	Cases n = 9	Remainder n = 59
Total tissue	34.6 ± 6.9	42.7 ± 7.4	−4.1	−8.8 to 0.7	0.09
Cortical gray matter	19.3 ± 7.3	24.3 ± 5.9	−1.6	−5.4 to 2.3	0.43

The authors thank the study families for their participation; Carole Spencer, Lisa Borkus, and the Canterbury Radiology Group for assistance with MRI and follow-up data; and Jeffrey Neil for his input.