Brain MRI findings and their association with visual impairment in young adolescents born very preterm

This study is part of the longitudinal multidisciplinary study of visuomotor capacity in very preterm infants (LOVIS) project that is a population-based prospective follow-up study of 113 very preterm infants born <32 weeks of gestational age, in 2004–2007 in Uppsala county, Sweden [19]. Forty-seven children born very preterm and 22 full-term controls (born >37 weeks of gestational age, in 2005–2009) who underwent both brain MRI and concurrent ophthalmological assessment at 12-year follow-up were included. The mean age for MRI assessment among very preterm born children was 12.8 years, median 12.6 years, range 11.7 to 14.6 years, and among controls 13.0, 13.0, and 12.2 to 13.8, respectively. A flowchart describing the very preterm study population is shown in Figure 1. Parental written consent was obtained after written and oral information. The LOVIS study had ethical approval from the regional research ethics committee in Uppsala (Dnr 03/665 and 2016/400).

One experienced neuroradiologist (JW) and one resident (AGK) examined the scans independently of each other’s results and blinded to perinatal characteristics including gestational age. The images were assessed quantitatively and qualitatively. WMI was noted and categorized into punctate high signal lesions, confluent high signal intensities, or cystic changes. Signs of previous hemorrhage and infarcts were also assessed. In case of differing diagnosis (12 cases), the scan was re-evaluated and discussed until consensus. For quantitative measurements, the mean was computed for the two investigators results, and agreement was assessed with intraclass correlation coefficient and Bland–Altman plot. Quantitative measurements were used to assess ventricle dilatation, thinning of the corpus callosum, and white matter loss as well as the thickness of the chiasma and the optic nerves. Ventricle size was evaluated with Evans index; the ratio of maximum width of the frontal horns and maximal internal diameter of skull at the same slice employed in axial MRI (Figure 2a). To assess posterior ventricle dilatation, we introduced the “posterior ventricle index” as a new measurement for evaluation of occipital white matter reduction. Similarly to Evans index, the maximum width of the posterior parts of the lateral ventricles (at the trigonum) was related to the maximal internal diameter of skull (Figure 2b). The size of the corpus callosum was assessed by measurement of the surface area in mm² on a mid-sagittal slice. The posterior part of the corpus callosum was assessed by measuring the surface area of the posterior half when divided into two parts (Figure 2c). The mean of the bilaterally smallest white matter part next to the trigone of the lateral ventricles was calculated in mm in order to assess the loss of white matter (Figure 2d). The thickness of the chiasma and the mean of the bilaterally optic nerve thickness were calculated in an axial T1-weighted slice (Figure 2e).

All children born very preterm were screened for ROP during the neonatal period. ROP was defined according to the international classification of ROP [11], and the criteria for treatment followed the ETROP definition [20]. Mild ROP was defined as stages 1 and 2, and severe ROP was defined as stages 3–5. Ophthalmological follow-up had previously been performed at 2.5 years of corrected age [21] and at 6.5 years of chronological age [22]. In the present study, a new ophthalmological examination was performed at 12 years, all by the same orthoptist. The children used their habitual glasses at the examination. Visual acuity (Sloan letters 3 m test) was assessed, and the best-corrected binocular distance acuity (with habitual glasses) was used for calculations. Contrast sensitivity was measured binocularly at 3 m with the Lea Hyvärinen 2.5% contrast sensitivity test. The cut-offs for further analyses regarding visual acuity and contrast sensitivity and MRI data were defined as visual acuity ≤1.0 and contrast sensitivity <0.5. Manifest strabismus was assessed with the help of a cover/un-covertest at near (33 cm) and at distance (5 m). Stereo acuity was examined at near with the TNO test. A stereopsis of more than 60 s of arc was regarded as subnormal. Manifest nystagmus was recorded. Full-term controls underwent the same ophthalmological assessment.

Differences in continuous background characteristics between study children and those who withdrew were studied using the independent samples t-test or Mann–Whitney U-test. For the categorical background characteristics, Pearson’s Chi-square test or Fisher’s exact test was used. Differences in continuous brain MRI variables between children born very preterm and full-term controls were evaluated using the independent samples t-test (Table 2). Differences in continuous ophthalmological variables between children born very preterm and controls were studied with Mann–Whitney U-test, while Pearson’s Chi-square was used for categorical variables. Associations between perinatal risk factors and brain MRI findings as well as brain MRI findings and ophthalmological outcomes were studied using generalized linear models. No child with treatment-requiring ROP was included, and therefore, no adjustment for ROP was necessary in the statistical analysis [10]. Residuals were checked to justify the analysis. Possible multicollinearity was checked; correlation coefficient ≥0.8 and/or tolerance value <0.1 and/or Phi’ and Cramers ≥0.8 was considered a sign of multicollinearity. Intraclass correlation coefficient was calculated to assess interrater agreement. The statistical analysis was performed using an SPSS version 27.0 and 28.0 (IBM SPSS Statistics, IBM Corporation, NY, USA). A two-tailed p-value of <0.05 was considered statistically significant.

Perinatal characteristics of the included 47 children (19 girls) born very preterm are shown in Table 1. Of the participating children, none was treated for ROP, 12 had IVH grade 1, none had IVH grades 2–4, and 2 children had PVL as diagnosed on neonatal ultrasound. Apart from treated ROP, no other differences regarding perinatal characteristics between children who participated in the present study (n=47) and those who withdrew (n=66) were found (Table 1). A total of 22 full-term controls (11 girls) were included.

A total of 15 (32%) children born very preterm had focal WMI: 5 (11%) had small punctate high signal lesions, and 10 (21%) had confluent high signal intensities in the white matter. No child had cystic lesions. Among full-term controls, 1 (5%) child had punctate high signal lesions. None of the focal white matter lesions were found in the optic radiation but were located in the frontal and parietal white matter regions. No control children had confluent high signal intensities or cystic lesions. There were no signs of previous hemorrhage or infarcts in either the preterm children or in full-term controls.

The mean values of Evans index, posterior ventricle index, peritrigonal white matter, corpus callosum area, and posterior corpus callosum area are presented in Table 2. Evans index (0.27 vs 0.25, p<0.001) and posterior ventricle index (0.47 vs 0.45, p=0.02) were larger in children born very preterm than in controls. We calculated lower estimates for posterior corpus callosum area in the preterm born children, but this difference did not reach statistical significance (p=0.08). Higher gestational age was associated with larger corpus callosum area (β=10.7, 95% CI 0.59–20.8) but not with other brain measurements in children born very preterm. The optic nerves were significantly decreased among the study children compared to the controls (4.49 vs. 4.12, p<0.001). No significant difference was found for the chiasma. The interrater agreement, estimated with the intraclass correlation coefficient, was 0.77, 0.54, 0.80, and 0.71 for the Evan’s index, posterior ventricle index, corpus callosum, and posterior corpus callosum, respectively. The intraclass correlation coefficient for peritrigonal white matter thickness was 0.14 and 0.20 for the right and left sides, respectively. For the chiasma, the intraclass correlation coefficient was 0.47 and for the mean of the optic nerve 0.77.

The results of the ophthalmological examinations are presented in Table 3. There was no difference in visual acuity (p=0.3) and contrast sensitivity (p=0.3) between children born very preterm and full-term controls. Manifest strabismus was found in 6 of the preterm children but in none of the full-term controls. Further, stereo acuity was normal (≤60) in 36 (77%) and subnormal (≥120) in 11 (23.4%) children, out of which 7 (14.9%) with no stereopsis at all, while all full-term controls had normal stereo acuity (p<0.001). There were no children with manifest nystagmus.

In children born very preterm, increased posterior ventricle index increased risk for visual acuity ≤1.0 (p=0.03) and contrast sensitivity <0.5 (p=0.005). Decreased peritrigonal white matter thickness associated with impaired visual acuity (p=0.04). Decreased thickness of the chiasma increased risk for subnormal stereo acuity (>60). No associations with strabismus were found. Regarding the other MRI measurements, Evans index, corpus callosum area, and posterior corpus callosum area, no significant associations with the ophthalmological outcomes were found (Table 4).