Association Between Optic Nerve Morphology by Handheld Spectral Domain Optical Coherence Tomography and Intracranial Pathology in Preterm Infants

Abstract

Purpose: This study investigates optic disc and macular morphology using handheld spectral domain-optical coherence tomography (SD-OCT) in preterm infants with and without brain pathology.

Methods: In this prospective observational study, premature newborns underwent handheld SD-OCT at 36 weeks postmenstrual age. Semiautomated quantitative OCT analysis was performed including the optic disc and macula. Routine head ultrasound radiology reports were obtained from the medical record.

Results: A total of 69 patients (35 male and 34 female) were included. The average birth weight was 982.7 ± 277.4 g and gestational age was 27.99 ± 2.57 weeks. Extremely preterm infants born at <28 weeks’ gestation had a larger cup diameter (P = 0.006), thinner nasal neural rim thickness (P = 0.010), and deeper maximal cup depth (P = 0.010) compared with preterm infants (≥28 weeks’ gestation). Brain lesions on ultrasound examination were detected in 15 patients (21.7%), with all having intraventricular hemorrhage (IVH) and 2 patients also presenting with periventricular leukomalacia. Patients with IVH had a thicker central macular retinal nerve fiber layer, ganglion cell layer, and inner plexiform layer than those without IVH (P = 0.029, P = 0.022, and P = 0.002). These differences persisted after adjusting for retinopathy of prematurity stage (P = 0.042, P = 0.034, and P = 0.007, respectively). When controlling for gestational age, the ganglion cell layer and inner plexiform layer differences remained (P = 0.011 and P = 0.032, respectively). Optic nerve parameters did not differ between infants with and without IVH.

Conclusions: IVH was associated independently with arrested foveal maturation with no discernible effect on optic disc morphology in the preterm newborn period.

Translational Relevance: The macular structural changes detected by SD-OCT in preterm newborns with brain pathology may suggest the need for closer monitoring of visual development in these infants.

Introduction

Preterm infants are at high risk of brain injury caused by hypoxic, ischemic, infectious, or inflammatory events.¹ Intraventricular hemorrhage (IVH) and periventricular leukomalacia (PVL) are common brain abnormalities that contribute to neurodevelopmental impairment.² They are also frequently associated with systemic conditions such as sepsis, necrotizing enterocolitis, and bronchopulmonary dysplasia, which also play a pivotal role in neurodevelopment.^3–5 These perinatal insults have been reported to correlate with morphological changes in the optic nerve and neuroretina.⁶ 

Both retrograde and anterograde trans-synaptic degeneration may play a role in the association between brain and ocular disease in preterm infants. Normal-sized optic discs with large cupping have been described in preterm infants with PVL.^7,8 One suggested mechanism is primary injury to the immature optic radiations leading to reduced retinal nerve fiber layer (RNFL) thickness.^7,8 This is regarded as a unique form of optic nerve hypoplasia that occurs after the scleral canals have established a normal diameter.^7,9 In addition, patients with severe IVH are known to more frequently develop optic atrophy.^10,11 Other previous studies have also demonstrated an association between thinner retinas and intracranial pathologic features along with neurodegenerative disease.^12–15 For most of these studies, retinal thickness was measured sometime after the pathological insult occurred, rather than during the newborn period. 

Handheld spectral domain-optical coherence tomography (SD-OCT) has allowed for the visualization of anatomical features of the optic nerve head and macular neuroretina in preterm infants during the newborn period.^16,17 Correlating these morphological features with intracranial abnormalities seen radiographically would enhance our understanding of intraocular structural development. The purpose of this study was to investigate optic disc and macular morphology using handheld SD-OCT in a series of preterm infants with and without brain pathology detected by head ultrasound examination. 

Materials and Methods

This prospective observational study was approved by the institutional review board at Seattle Children's Hospital and the University of Washington (Number: 15464), and the procedures used in this study adhere to the Declaration of Helsinki. Informed consent was obtained from the guardians of all participants. 

Subjects

The inclusion criteria were (1) premature newborns in the neonatal intensive care unit undergoing retinopathy of prematurity (ROP) screening examinations (standard criteria include gestational age [GA] <30 weeks, birth weight <1500 g, or infants with birth weight between 1500–2000 g with an unstable clinical course), (2) adequate quality optic nerve or macular handheld SD-OCT images for analysis, and (3) brain ultrasound examination performed at least one time. Patients were excluded if they had an additional chromosomal anomaly or other congenital malformation unrelated to prematurity. Fundus examination using indirect ophthalmoscopy was performed as a part of standard clinical practice to determine the presence and stage of ROP, level of plus disease, clinical follow-up schedule, and treatment. Baseline demographic data and other details were obtained from the medical record, including GA, postmenstrual age, sex, race, ethnicity, birth weight, delivery method, maternal history, and systemic disease (bronchopulmonary dysplasia or chronic lung disease, necrotizing enterocolitis or other inflammation of the intestine, neonatal sepsis, and other inflammatory or infectious disease). 

Handheld SD-OCT Imaging

Handheld SD-OCT (Envisu C2300, Leica Microsystems, Wetzlar, Germany) was performed at the time of each clinical ROP screening examination. The pupils were dilated with an ophthalmic solution containing 1% phenylephrine hydrochloride and 0.2% cyclopentolate (Alcon, Inc, Fort Worth, TX). Patients were in the supine position without anesthesia during imaging. Both fovea and optic nerve images were taken with volume scans. For this study, OCT images obtained at a single time point per infant closest to 36 weeks GA (range, 34–38 weeks GA) were used for analysis. If the acquired optic nerve image was tilted beyond 30°, it was excluded from the analysis. Images were also excluded in cases of significant macular edema distorting the foveal contour or inadequate image quality for image analysis, as determined by a trained grader (H.J.). 

OCT Analysis

Quantitative OCT analysis was conducted in a semiautomated manner using a computer algorithm (Python, Wilmington, DE). Two trained graders (H.J. and C.P.) performed the analysis. Interobserver reliability was determined between graders, and data measured by the more senior grader (H.J.) were used in the final analysis. 

Figure 1A illustrates the semiautomated optic disc analysis. The OCT B-scan involving the deepest point of the optic nerve cup was selected for analysis by the grader. The edge of the retinal pigment epithelium and the deepest point of the optic disc were determined manually by the grader. The software then calculated the disc diameter, defined as the distance between the edges of the retinal pigment epithelium. Cup diameter was measured as the length of the line parallel to the disc diameter 150 µm anteriorly, between the two points intersecting the internal limiting membrane. Nasal and temporal rims were calculated as the distance between the disc and cup edges. Maximal cup depth was measured as a perpendicular line between the cup diameter and the deepest point of the cup. Average RNFL thickness was measured from 1200 to 1600 µm on both sides from the center of the cup. 

Figure 1.

 

Semiautomated measurements of parameters in SD-OCT at the optic nerve (A) included disc diameter, cup diameter, maximal cup depth, nasal and temporal neural rim thickness, nasal and temporal neural height, and nasal and temporal peripapillary RNFL thickness. At the macula (B), measurements included RNFL, GCL, IPL, and whole retinal thickness in the central, nasal and temporal areas.

Figure 1B demonstrates the macular analysis. The grader selected the B-scan at the deepest point of the foveal pit. The software calculated RNFL, ganglion cell layer (GCL), inner plexiform layer (IPL), and whole retinal (internal limiting membrane to Bruch's membrane) thickness. Measurements were obtained centrally (average thickness within 250 µm nasally and temporally from center), nasally (average thickness 500–2000 µm nasally from center), and temporally (average thickness 500–2000 µm temporally from center). All segmentation was reviewed individually by the graders. If errors in segmentation were identified in key areas, manual adjustments in the segmentation were performed. 

Brain Imaging

Brain ultrasound examination was usually obtained for clinical purposes within the first 2 weeks after birth and again after 36 weeks postmenstrual age, with interval ultrasound examinations occurring if IVH was detected. The brain ultrasound examination was interpreted by experienced radiologists as a part of standard clinical care. The radiology reports were reviewed for the following findings: IVH, PVL, infectious signs (including calcification), and hydrocephalus. An abnormality at any time point was classified as an abnormal lesion. 

Statistical Analysis

Intergrader reliability for OCT measurements was assessed using the intraclass correlation coefficient. Patients were divided into two groups based on the presence or absence of brain lesions detected by ultrasound examination, and further stratified by GA as extremely preterm (defined as birth at a GA of <28 weeks) and nonextremely preterm (birth at a GA of ≥28 weeks). 

Characteristics and OCT measurements were compared between the two groups based on brain lesions, with and without adjusting for GA and ROP stage as potential confounders. OCT measurements were also compared between extremely preterm and nonextremely preterm groups. Correlations between the demographics and the OCT parameters were also analyzed. All statistics used a generalized linear mixed model to accommodate for multiple measurements per infant. Given the exploratory nature of the study and its relatively small sample size, adjustments for multiple comparisons were not performed. Continuous variables are presented as mean ± standard deviation. A P Value of ≤0.05 was considered statistically significant. 

Results

Participant Demographics

After excluding 33 patients lacking OCT imaging during the required time period, 6 patients for inadequate quality to perform analysis, and 7 patients for significantly distorting macular edema, a total of 69 patients (35 male and 34 female) were included in the study. Race/ethnicity included 39 (56.5%) White, 13 (18.8%) Hispanic, 4 (5.8%) Asian, 2 (2.9%) American Indian, 2 (2.9%) Black, 1 (1.4%) Pacific Islander, and 8 (11.6%) other. Average birth weight was 982.7 ± 277.4 g and GA was 27.99 ± 2.57 weeks. Bronchopulmonary dysplasia or chronic lung disease were diagnosed in 31 (46.3%), necrotizing enterocolitis in 6 (9%), neonatal sepsis in 13 (19.4%), and other inflammatory or infectious disease in 8 (11.6%) patients, including urinary tract infection (n = 3), gastritis (n = 1), thyroiditis (n = 1), meningitis (n = 1), cellulitis (n = 1), and acquired cytomegalovirus infection (n = 1) (Table 1). 

Table 1.

 

Demographics and Other Characteristics of Patients With and Without Brain Lesions on Head Ultrasound Examination

 

Forty-three eyes (31.6%) were diagnosed with ROP, including 17 (12.5%) eyes with stage 1 disease, 15 (11.0%) eyes with stage 2 disease, and 11 (8.1%) eyes with stage 3 disease. Nondistorting macular edema was detected in 8 (5.8%) eyes (Table 2). 

Table 2.

 

Ophthalmic Features of Patients With and Without Brain Lesions on Head Ultrasound Examination

 

Abnormal brain lesions were detected in 15 patients (21.7%), with IVH in all 15 cases. PVL was diagnosed in two patients (13.3%) in addition to IVH. Comparing patients with IVH with those without, there were several characteristics that differed between the two groups; among patients with IVH, birth weight (P = 0.034) and GA (P = 0.002) were lower than in patients without IVH. Vaginal delivery was a more frequent delivery method (P = 0.009), and bronchopulmonary dysplasia or chronic lung disease were more frequent comorbidities (P < 0.001), and ROP (P = 0.049) was more common among those with IVH compared with those without IVH (Tables 1 and 2). 

Interobserver Reliability

Interobserver reliability was assessed regarding all OCT semiautomated measurements between the two independent graders (H.J. and C.P.). All intraclass correlation coefficient results were in the range of excellent except for nasal macular whole retinal thickness (intraclass correlation coefficient = 0.419, 95% confidence interval, 0.146−0.605) (Table 3). 

Table 3.

 

Interobserver Reliability of the OCT Optic Disc and Macular Parameters Between the Two Graders

 

OCT Measurements, Demographics, and Other Comorbidities

A lower GA corresponded with thicker central macular RNFL (P = 0.009) and central macular IPL (P < 0.001). Lower birth weight correlated with a thinner temporal macular RNFL (P = 0.045) and both nasal (P = 0.046) and temporal (P = 0.017) macular whole retinal thicknesses. Bronchopulmonary dysplasia and chronic lung disease were associated with increased central macular thickness of all individual layers as well as total thickness (RNFL, P = 0.007; GCL, P = 0.011; IPL, P = 0.020; whole retina, P = 0.026). All other comparisons between OCT measurements, demographics, and other comorbidities were not significant. When comparing extremely preterm and nonextremely preterm infants, cup diameter was larger in extremely preterm infants (1206.69 ± 268.23 vs. 924.93 ± 365.86 µm; P = 0.006), whereas disc diameters were not different. Accordingly, the cup-to-disc ratio was greater in extremely preterm infants (0.66 ± 0.15 vs. 0.51 ± 0.18; P = 0.003). Nasal neural rim thickness was thinner in extremely preterm infants (366.12 ± 200.82 vs. 569.30 ± 292.78 µm; P = 0.010) and temporal neural rim thicknesses were not different. Maximal cup depths were also different; extremely preterm infants demonstrated deeper maximal cup depth compared with nonextremely preterm infants (418.69 ± 135.15 vs. 300.43 ± 151.31 µm; P = 0.010) (Table 4). 

Table 4.

 

Optic Nerve and Macular Measurements Using Handheld SD-OCT in Preterm Infants According to GA

 

OCT Measurements and IVH

All optic disc parameters were not different between the two groups of patients with and without ultrasound brain lesions (IVH). (Table 5) In contrast, among macular OCT parameters, central RNFL (8.86 ± 3.96 vs. 6.79 ± 3.43 µm), GCL (10.14 ± 5.77 vs. 7.23 ± 4.45 µm), and IPL (31.00 ± 10.50 vs. 24.44 ± 6.85 µm) thicknesses were significantly greater in patients with IVH compared with those without IVH (P = 0.029, P = 0.022, and P = 0.002, respectively) (Table 5, Fig. 2). After controlling for ROP stage, these differences remained significant (P = 0.042, P = 0.034, and P = 0.007, respectively), whereas controlling for GA alone retained significance only for the GCL and IPL thicknesses (P = 0.011 and P = 0.032, respectively). However, when controlling both GA and ROP stage simultaneously, none of these variables maintained statistical significance (P = 0.160, P = 0.094, and P = 0.064). Of note, the two confounders—GA and ROP stage—were significantly correlated (P = 0.003). 

Table 5.

 

Optic Nerve and Macular Measurements Using Handheld OCT in Preterm Infants According to the Presence or Absence of Brain Lesion on Head Ultrasound Examination

 

Figure 2.

 

Macular retinal thickness by handheld SD-OCT in preterm infants according to presence of brain lesion by head ultrasound examination (IVH in all cases with two cases of PVL). Central thicknesses were greater for those with brain lesions compared with those without for (A) RNFL (P = 0.029), (B) GCL (P = 0.022), and (C) IPL (P = 0.002), whereas nasal and temporal layer thicknesses did not differ between groups. (D) Whole retinal thickness (central, nasal, and temporal) did not significantly differ between groups. Significant differences are marked as *. These differences remained significant even after adjusting for GA for GCL (P = 0.011) and IPL (P = 0.032).

Discussion

SD-OCT measurements differed according to the presence of IVH in premature infants; central macular RNFL, GCL, and IPL thicknesses were greater in patients with IVH than in patients without IVH. Parafoveal retinal thicknesses and optic disc parameters did not differ between the groups. OCT findings associated with IVH most closely corresponded with measures of foveal immaturity, rather than markers of optic nerve damage. 

GA is recognized as a significant factor influencing optic nerve parameters, which is aligned with our study. In our study, we found that extremely preterm infants had larger cup diameters and cup-to-disc ratios compared with nonextremely preterm infants, although disc diameters were not different. Preterm infants previously were found to have a larger vertical cup diameter and vertical cup-to-disc ratio compared with term infants, whereas cup depth and vertical disc diameter were not significantly different.¹⁷ PVL may play a role; PVL has been associated with abnormally large optic cups with a thin neuroretinal rim and normal-sized optic discs, with extreme prematurity as a risk factor.⁹ However, only two subjects in our study were diagnosed with PVL based on head ultrasound examination. Nonetheless, ultrasound examination may have limited sensitivity in the early diagnosis of PVL,^18,19 and it is unclear whether some of our patients had a missed diagnosis of PVL. 

Based on the results of this study, IVH may be an independent risk factor for arrested foveal maturation. Central macular inner retinal thicknesses were thicker in patients with IVH in our study, and these differences remained significant for GCL and IPL after controlling for GA. Parafoveal thicknesses showed no differences, suggesting that these changes are not related to optic nerve disease. During foveal maturation, the inner retinal layer thicknesses typically decrease during the period between 31 and 42 weeks, driven by inner retinal migration.^20,21 GA, ROP severity, and ROP treatment are known to influence this process.^20,22,23 IVH is associated with systemic angiogenic changes thought to stem from an imbalance of angiogenic and antiangiogenic factors.^24–26 We hypothesize that microvascular development in the fovea may also be influenced by such changes in the presence of IVH, contributing to foveal immaturity. Lavric et al.²⁷ identified a diminished foveal avascular zone in preterm infants using OCT angiography, associated with IVH and foveal immaturity (as measured by foveal depth). 

ROP severity could also potentially influence retinal layer thicknesses. Because IVH is a significant risk factor for ROP severity, it may exacerbate the impact of ROP on retinal development.²⁷ Nevertheless, even after adjusting for ROP stage as a confounding factor, the difference between the two groups remained significant. 

We did not observe any differences in optic disc morphology associated with IVH. Previous research demonstrated that premature infants with posthemorrhagic hydrocephalus exhibited shallower optic cups.¹⁷ However, none of the subjects in our study had hydrocephalus. Both the timing and severity of the intracranial insult may influence whether or not we see changes in optic disc morphology. PVL primarily occurs during the antenatal and perinatal periods, whereas IVH predominantly occurs in the early postnatal period, particularly in extremely preterm infants. This difference may explain the lack of association between IVH and optic disc morphology, despite the association between prematurity and optic disc morphology, where PVL may play a role. Moreover, optic disc changes may take time to develop after an intracranial insult, and our early imaging study during infancy may not capture later development of optic atrophy resulting from IVH. 

Our study has limitations. First, as an observational study, we cannot exclude additional comorbidities that coincide with IVH with the potential to influence optic nerve and macular morphology. IVH is inherently more common in lower GA infants, requiring careful adjustment for confounders. We controlled GA and ROP stage, but factors like ROP zone may also influence retinal thickness. However, adjusting for multiple correlated variables poses challenges, including potential multicollinearity. Future studies with larger cohorts could better control for these factors and incorporate ROP zone classification to further clarify its impact. Second, we did not analyze axial length or refractive error, which may affect the OCT measurements. Third, as an exploratory study limited by a small sample size, we elected not to perform statistical adjustments for multiple comparisons. However, the positive results of this study were consistent with previous research and remained significant after adjusting for GA and ROP stage. Therefore, these findings are credible and less likely to be attributable to chance alone. Fourth, manual segmentation was typically needed when image quality was suboptimal owing to a lack of infant sedation. Subjectivity was, therefore, introduced, which may influence results. 

Our study has notable strengths. We performed OCT measurements much earlier in the postnatal period than most prior studies, thereby minimizing confounding effects from later phases of neurodevelopment and disease. By examining the OCT morphology of both the optic nerve and macula, we were able to provide a more comprehensive exploration of the impact of IVH on neuroretinal development. 

In conclusion, IVH occurring in preterm infants independently arrested foveal maturation with no discernible effect on optic disc morphology in the newborn period. More careful monitoring of visual development in preterm infants with a history of IVH may be warranted. 

Acknowledgments

Supported by Knights Templar Eye Foundation Career Starter Grant, Latham Vision Research Innovation Award, Alcon Research Institute Young Investigator Award, Violet Sees, and unrestricted grants from Research to Prevent Blindness and National Institutes of Health (EY00130) to the University of Washington Department of Ophthalmology. 

Disclosure: H. Jeon, None; J. Bunk, None; C. Park, None; K.M. Perez, None; M. Dighe, None; L. Ding, None; L.E. Grant, None; A. Shariff, None; T.B. Gillette, None; K. Tarczy-Hornoch, None; M.T. Cabrera, None