Peripapillary Retinal Nerve Fiber Layer (pRNFL) Thickness – A Novel Biomarker of Neurodegeneration in Late-Infantile CLN2 Disease

Introduction

The term neuronal ceroid lipofuscinoses (NCLs) encompasses a heterogenous collection of 13 distinct neurodegenerative lysosomal storage diseases (CLN 1–14) that collectively represent the most common inherited neurodegenerative disorders in childhood.¹ These disorders are attributed to mutations in genes that code for enzymes that precipitate in the accumulation of lysosomal storage material, primarily in cells of the central nervous system (CNS), causing neuron and interneuron loss.² CLN2 disease mainly presents as a late-infantile onset NCL (LINCL) and is associated with mutations to the tripeptidyl peptidase 1 (TPP1) gene, that encodes the lysosomal soluble enzyme hydrolase TPP1. Mutations to the TPP1 gene trigger decreased activity of TPP1, resulting in the systematic accumulation of curvilinear profiles in lysosomal residual bodies. Cells of the CNS and the retina are especially vulnerable to these metabolic changes, which ultimately lead to their degeneration.³ Symptoms usually begin in early childhood, around 2–4 years of age and include decline of verbal language, loss of motor skills, dementia, and seizures.⁴ The 2–3 years following onset of symptoms are characterized by rapid deterioration of motor function, language, and swallowing, with many patients experiencing premature death in early adolescence. Ocular symptoms are a cardinal feature of late-infantile CLN2 disease, as vision loss typically manifests after 3 years of age, usually subsequently to seizures and psychomotor decline.⁵ The vision loss progresses to blindness over time. However, optical coherence tomography (OCT) imaging has enabled the detection of preliminary signs of retinal degeneration in patients as young as 2 years of age, before the display of clinical ocular symptoms. The retinopathy in late-infantile CLN2 patients has been described as a bilateral, symmetric, outer retinal degeneration that expands from the central macula to the peripheral retina. This degeneration, which is especially accelerated between 4 and 6 years of age, primarily affects the retinal pigment epithelium (RPE) and photoreceptor cells in a bull’s eye pattern (bull’s eye maculopathy).^6–8 Currently, intracerebroventricular (ICV) enzyme replacement therapy (ERT) with recombinant human TPP1 (rhTPP1) (cerliponase alfa, Brineura, BioMarin Pharmaceuticals, California, USA) is the only United States Food and Drug Administration (US FDA) and European Medicines Agency (EMA) approved treatment for CLN2 that has proven to decelerate the progression of motor and language function decline.⁹ However, the progression of retinal degeneration has only been mitigated through preclinical treatment strategies that target the retina directly. In particular, intravitreal (IVT) therapies in canine CLN2 models,^10–12 and as of recently, continuous IVT delivery of rhTPP1 in human CLN2 patients,¹³ have displayed promising results in slowing down this process. Various OCT biomarkers, among them central retinal thickness (CRT), ellipsoid zone (EZ) integrity and outer nuclear layer (ONL) thickness, have been described as indicators of late-infantile CLN2 disease progression,^3,14 and therefore may support the development and monitoring of new disease-modifying treatments.

Retinal nerve fiber layer (RNFL) thickness is an extensively studied OCT biomarker in neurodegenerative disorders. Particularly, peripapillary RNFL (pRNFL) thinning has been associated with poor disease progression in a multitude of neurodegenerative diseases including Alzheimer’s disease, multiple sclerosis and Parkinson’s disease.^15–22 As late-infantile CLN2 disease is associated with abundant outer retinal degeneration,^6,7,15 little attention has been given to the inner retinal morphology in this neurodegenerative disease. The aim of the present retrospective, observational study was to examine the presence of pRNFL degeneration in cerliponase alfa treated patients with late-infantile CLN2 disease.

As our findings were indicative of progressive pRNFL degeneration, we furthermore evaluated pRNFL changes in correlation to neurological deterioration, as assessed via the Weill Cornell Late Infantile Neuronal Ceroid Lipofuscinoses (Weill Cornell LINCL) Scale,⁵ and the Hamburg Motor and Language (HML) Scale.²³

Materials and Methods

This is a retrospective observational study based on analysis of medical history data, as well as pediatric, neurologic, and ophthalmologic examinations performed during standard of care routine consultation visits at the University Medical Center Hamburg-Eppendorf and the NCL specialty clinic at the Department of Pediatrics and at the department of Ophthalmology. The protocol of this study was approved by the local ethical committee of the Ärztekammer Hamburg, Hamburg (PV7215) and informed consent was obtained for all patients prior to enrollment. The principles of the Declaration of Helsinki were followed. The study protocol was registered at ClinicalTrials.org (registration number: NCT04613089).

Design and Participants

Our study comprised a total of 20 patients with genetically and enzymatically confirmed diagnosis of CLN2 disease. Patient’s detailed demographic and genotype data are illustrated in Table 1. All patients were treated with 300 mg of cerliponase alfa during the course of the study, administered through ICV infusions once every two weeks. The ages at which cerliponase alfa treatments were started are provided in Table 1.

Table 1 Patients’ Detailed Demographic, Genotypic and Clinical Information

Study Procedures

Patients underwent OCT imaging during general anesthesia using a Spectral Domain OCT (Heidelberg Engineering OCT2 Flex, Heidelberg Engineering Inc., Heidelberg, Germany). PRNFL layer segmentation values and CRT calculation were manually confirmed with the help of the provided software (Heidelberg Eye Explorer 1.10.12.0). CRT, global (average) pRNFL (G-pRNFL), and the pRNFL values in each of the six sectors around the optic nerve head [temporal superior pRNFL (TS-pRNFL), temporal pRNFL (T-pRNFL), temporal inferior pRNFL (TI-pRNFL), nasal superior pRNFL (NS-pRNFL), nasal pRNFL (N-pRNFL) and nasal inferior pRNFL (NI-pRNFL)] were similar for both eyes of each patient and considering the disease’s symmetrical progression of retinal degeneration,⁷ they were averaged. Normative CRT and pRNFL thickness values in healthy children were obtained from data presented by Turk et al.²⁴ They assessed normative CRT and pRNFL values in a cohort of 107 healthy children using a Spectral Domain OCT device manufactured by the same manufacturer as the OCT2 Flex module utilized in our study (Heidelberg Engineering, Heidelberg, Germany). Medical history data were obtained from the DEM-CHILD database (ClinicalTrials.gov identifier NCT04613089). Each patient’s neurological disease progression was quantitatively assessed by a pediatric neurologist from the NCL-specialty clinic utilizing the disease specific Weill Cornell LINCL Scale, and HML Scale.^5,25

Statistical Analysis

Statistical analyses were performed using SPSS ver. 27.0 (IBM, Armonk, NY, USA). A two-sided t-test was applied to analyze differences between the observed cohort and the control group. Each statistical variable was tested for normality with a Shapiro–Wilk test. In accordance with the results, non-parametric correlation analysis (two-sided Spearman rank correlation) was performed.

Results

Participants

A total of 40 eyes of 20 patients were examined at least once and were included in this observational study. Our cohort included 12 (60%) male patients and 8 (40%) female patients. Mean age at examination was 6.90 years (range 2–11.08 years, SD 2.36) for all patients. All patients started receiving cerliponase alfa treatment prior to the presented examinations at a mean age of 4.40 years (range 0.58–6.75 years, SD 1.36).

Fourteen patients underwent serial examinations, whereas 6 patients were only examined once, resulting in a total of 84 OCT scans, 42 Weill Cornell LINCL Scale, and 42 hML Scale scores. Follow-up examinations were performed at a mean time of 1.38 years (range 0.42–2.17 years, SD 0.50). The patient’s data sets are displayed in detail in Tables 1 and 2.

Table 2 Patients’ Age, CRT and RNFL Values

Global RNFL Assessment

Mean G-pRNFL thickness was 77.02 μm (range: 43–103, SD 13.61), displaying a significant decrease (p < 0.0001) compared to the normative values described by Turk et al²⁴ (Table 3). Thirty-nine out of the 42 measured G-pRNFL thickness values (92.86%) were beneath the normative range. The youngest patient with measurable G-pRNFL loss was 2 years of age. G-pRNFL thickness values demonstrated a very strong positive correlation with the disease-specific clinical rating scales,²⁶ exhibiting a Spearman correlation coefficient (r_s) of 0.849 (p < 0.01) for the Weill Cornell LINCL Scale, and 0.833 (p < 0.01) for the HML Scale. When correlated with age at examination, G-pRNFL displayed a significant negative correlation (r_s = - 0.557, p < 0.01).

Table 3 Descriptive Statistics of RNFL Thickness and CRT Values in Children with Late Infantile CLN2 Disease and in Healthy Children

Evaluation of RNFL Sectors

A detailed evaluation of pRNFL sectors was performed in order to identify the different structural changes in each sector. The descriptive statistics for these values, and for the normative values for each sector²⁴ are illustrated in Table 3. Because clinically relevant pRNFL alterations are primarily observed in the temporal sectors of the optic nerve head, further assessment and correlation analysis will be focused on these sectors.

Mean TS-pRNFL and TI-pRNFL thickness values were 94.12 μm (range: 42–129, SD: 22.07) and 99.36 μm (range: 55–157, SD: 24.42) respectively, exhibiting a significant reduction (p < 0.0001) from the normative values,²⁴ with 38 TS-pRNFL values (90.48%) and 37 TI-pRNFL values (88.10%) falling below the normative range. Both TS-pRNFL and TI-pRNFL displayed moderate, significant positive correlations towards the clinical scores. Spearman correlation coefficient of TS-pRNFL was 0.788 (p < 0.01) for the Weill Cornell LINCL Scale and 0.739 (p < 0.01) for the HML Scale, whereas TI-pRNFL presented r_s values of 0.709 (p < 0.01) for the Weill Cornell LINCL Scale and 0.653 (p < 0.01) for the HML Scale. In the temporal sector, T-pRNFL thickness revealed a mean of 70.71 μm (range: 46–100, SD: 12.19). T-pRNFL values did not show any significant difference (p < 0.0566) when compared to healthy children,²⁴ as 12 of the 42 T-pRNFL values (28.57%) were below the normal range. Furthermore, T-pRNFL displayed fair, positive correlations towards the Weil Cornell LINCL Scale (r_s = 0.389, p < 0.05) and the HML Scale (r_s = 0.417, p < 0.01).

In addition to these assessments, we analyzed the correlation between pRNFL thickness values in the above sectors and the age at examination. Our data indicated that TS-pRNFL (r_s = - 0.553, p < 0.01) and TI-pRNFL (r_s = - 0.504, p < 0.01) exhibited significant, negative correlations towards the age at examination. T-pRNFL did not show any significant correlation to age (r_s = - 0.259, p > 0.05). The correlation coefficient values for all pRNFL sectors in regard to age at examination, the Weill Cornell LINCL Scale and the HML Scale are showcased in Table 4.

Table 4 Spearman Correlation Coefficient (Rs) Values Between RNFL Thickness and CRT Values, Age at Examination, Weill Cornell LINCL Scale Scores, and HML Scale Scores

Central Retinal Thickness Analysis

Our patient cohort reported a mean CRT of 167.14 (range: 88–270, SD: 58.35), demonstrating a significant decrease (p < 0.0001) compared to healthy children, as 34 of the 42 CRT measurements (80.95%) were lower than the normative range²⁴ (Table 3). The earliest age at which CRT loss was documented in our cohort was 2 years. When studying the correlation between CRT and the disease-specific clinical Scales, we discovered that CRT correlated fairly with the Weill Cornell LINCL Scale (r_s = 0.578, p < 0.01) and the HML Scale (r_s = 0.577, p < 0.01) while also demonstrating a fair negative correlation with age at examination (r_s = - 0.549, p < 0.01) (Table 4).

Discussion

Late-infantile neuronal ceroid lipofuscinosis type 2 (CLN2) is a progressive neurodegenerative disorder manifesting with seizures, motor deficits, language impairment and vision loss.⁴ Outer retinal degeneration has been a well-documented feature of this disease.^6–8 However, few studies have suggested inner retinal involvement.³ In the herein study, we identified the presence of progressive pRNFL degeneration in treatment experienced patients with genetically confirmed diagnosis of late-infantile CLN2 disease.

Natural history studies on the disease’s ocular manifestations have described a bilateral and symmetrical,⁷ progressive retinal degeneration that commences from the macula and expands to the peripheral retina in a bull’s eye pattern.⁶ This degenerative process, which is especially amplified during the pivotal period of 48 to 72 months⁷ is initially identifiable in the parafoveal area, at the layer of the photoreceptors, the RPE,⁶ and the ONL layer, which precedes EZ loss.³ These parafoveal changes progress further to the fovea, and subsequently to the outer macula,³ before ultimately advancing centrifugally to the peripheral retina.

Currently, ICV ERT with cerliponase alfa is the only US FDA and EMA approved treatment that slows the progression of neurological decline in late-infantile CLN2 disease.⁹ However, studies on treatment experienced patients have illustrated the ineffectiveness of ICV infusions in mitigating retinal impairment in the context of this disease. In a previous investigation, our team highlighted that despite undergoing ICV ERT treatment, patients still exhibited a continuous decline in CRT, with a marked increase in the deterioration rate between 56 and 80 months of age.⁸ Recently, Huang et al reaffirmed through OCT analyses that treatment experienced patients display the same pattern of outer retinal deterioration, which is observed in natural history studies.^3,14 Introducing a novel OCT-based clinical staging system, the authors illustrated parafoveal changes and foveal sparing in stage 2, followed by foveal involvement in stage 3, and eventually global macular degeneration, in stages 4 and 5.¹⁴ In the herein study, we were able to further reinforce these findings by revealing a significant decrease in the CRT values of our treatment experienced cohort. Moreover, we found a significant inverse association between CRT and the age at examination, confirming the results of Kovacs et al,⁷ and of our previous investigation.⁸

Although the investigation of inner retinal alterations in late-infantile CLN2 disease has comparably received less attention, electroretinography (ERG) studies on treatment naïve LINCL patients²⁷ and CLN2 canine models²⁸ provided results supporting inner retinal involvement in this neurodegenerative disease. Additionally, Katz et al²⁹ identified inner nuclear layer (INL) degeneration and accumulation of fluorescent lysosomal storage material in canine retinal ganglion cells, and in a later OCT natural history study, Kovacs et al suggested that RNFL may be affected in advanced stages of the disease as well.³ Nevertheless, conflicting evidence from preclinical²⁸ and clinical studies alike^7,14 have led to a lack of consensus within the scientific community regarding RNFL loss in late-infantile CLN2 disease.

In our present analysis of 84 OCT scans of 20 late-infantile CLN2 patients we identified that pRNFL thickness values were significantly decreased when compared to healthy children. Furthermore, we discovered that despite treatment, retinal ganglion cell axons followed a significant inverse association in correlation to age (Figure 1). The presence of pRNFL thinning was observable in patients as young as 2 years of age, before the onset of the rapid retinal decline period and preceding the presentation of motor and language deficits. These findings challenge the conclusions drawn by Kovacs et, who described the absence of pRNFL thinning during the early stages of outer retinal degeneration, suggesting that pRNFL loss manifests in advanced stages of the disease, after the age of 5 years.³ Our outcomes of progressive pRNFL degeneration in late-infantile CLN2 are consistent with the documented nerve fiber layer loss, exhibited in other neurodegenerative metabolic diseases. Characteristically, pRNFL thinning has been recognized in patients with type 1 Gaucher disease (GD1), Adrenoleukodystrophy (ALD) and Wilson’s Disease (WD) alike.^30–32

Figure 1 Global RNFL values correlated with age at examination, the Weill Cornell Clinical Scale scores and the Hamburg Motor Language Scale scores.

Subsequently, we explored the potential relationship between pRNFL degeneration and clinical deterioration in regard to patient’s motor and language functions, as assessed via the disease specific Weill Cornell LINCL and HML clinical rating Scales. Our analysis concluded that the severity of pRNFL degeneration correlates very strongly with the decline in neurological function, as patients with the lowest pRNFL thickness values displayed the worst clinical symptoms. Even though CRT values also correlated significantly with the clinical scores, their correlation coefficients were inferior to those established by pRNFL (Table 4).

Notably, it became evident during our study that the temporal sector was the only sub – sector that did not display a significant nerve fiber loss compared to healthy controls. Furthermore, T-pRNFL thickness values did not demonstrate a significant correlation towards the age at examination, in contrast to the strong negative linear relationship displayed between TS-pRNFL and TI-pRNFL and age (Figure 2). These differences became even more profound when we examined the relationship between pRNFL loss in each sector and the patient’s neurological status, as pRNFL thickness values in the temporal superior and temporal inferior sectors exhibited greater correlations towards the Weil Cornell LINCL and HML Scales, compared to those in the temporal sector. These findings suggest that in late-infantile CLN2 disease pRNFL loss occurs predominantly in the superotemporal and inferotemporal region, whereas the temporal nerve fibers are relatively preserved throughout the disease (Figure 3). This regional pattern of retinal nerve fiber loss is similar to the sectoral neurodegeneration trend described in glaucoma, where pRNFL thinning is primarily observed in the temporal superior and temporal inferior sectors and the temporal sector is generally spared until the later stages of disease.^33,34 The notion behind this specific pattern of sectoral degeneration, which has been identified in other neurodegenerative diseases as well, including Alzheimer’s disease,³⁵ is still unclear.³⁶ It has been suggested that neurodegeneration in the superior and inferior quadrants may be more profound, because of the larger number of neurons that are naturally located in these regions, compared to the temporal retina.³⁷ Furthermore, the difference between the size and the axons length of the bigger parasol cells (M – cells) that are primarily located superiorly and inferiorly, and the smaller midget cells (P – cells) which are mainly found in the temporal sector may affect cellular susceptibility to amyloid deposition, metabolic requirements, and myelin sheath turnover.^35,36 These considerations, together with other potential factors including differences in vascularization and glial cell support could impact the degenerative process, culminating in the distinctive pattern of neurodegeneration described above.³⁵

Figure 2 Temporal, Temporal Superior and Temporal Inferior pRNFL values correlated with age at examination.

Figure 3 Heidelberg SD-OCT RNFL analysis of a 9.5-year-old patient with late-infantile CLN2 disease. PRNFL is significantly decreased in the global, temporal superior and temporal inferior pRNFL sectors.

Based on our findings, we support the notion that progressive pRNFL loss in late-infantile CLN2 disease is primarily the result of CNS driven neurodegeneration and not secondary to outer retinal degeneration. The detection of pathologic pRNFL thickness values in patients as young as 2 years of age, combined with the presence of pRNFL deterioration in patients with preserved CRT values suggest that RNFL degeneration may occur concurrently with, or even prior to perifoveal outer retinal alterations.

In addition, an analysis of the pRNFL sectors highlighted that the temporal superior and temporal inferior sectors were the first, as well as the most frequently affected regions by the RNFL degeneration, in contrast to the temporal nerve fibers which comparably appeared unimpaired. These observations challenge the assumption that pRNFL loss in late-infantile CLN2 disease occurs as a secondary effect to macular degeneration, as if that would be the case, then neurodegeneration would primarily affect the papillomacular bundle and manifest as temporal pRNFL loss.^35,38 Retinal ganglion cell pathology appears to occur independently of outer retina pathology.

The strong interrelation between CNS and pRNFL deterioration, in combination with our findings may suggest that pRNFL loss could be a direct result of the neurodegeneration in late-infantile CLN2 disease and not a secondary feature to retinal degeneration. The potential of OCT assessed pRNFL thickness could add value as a potential biomarker for monitoring and predicting neurodegenerative progression and evaluating response to treatment. The effectiveness of RNFL as a surrogate for disease progression has been extensively studied in several metabolic, and non-metabolic neurodegenerative diseases including Niemann – Pick type C disease, Wilson’s Disease, Huntington Disease and more.^{15,22,39–41} The substantial body of research on the potential of RNFL as a biomarker is plausible considering its fast and noninvasive nature, its repeatability, and its reproducibility.^42,43

Based on the outcomes of the herein study, we believe that OCT analysis of pRNFL thickness is a reliable, noninvasive tool that can contribute significantly to the tracking of disease progression in treatment experienced, and inexperienced patients, as well as in the development of new disease modifying treatment approaches.

The herein investigation represents one of the largest studies on the ocular manifestations of late-infantile CLN2 disease, considering the rarity of the disease and the low compliance rate of patients, on account of the occurring CNS degeneration. However, a larger sample size and additional prospective longitudinal examinations would still be required in order to reinforce the findings of our study. In light of these limitations, we strongly encourage further investigations on RNFL alterations in late-infantile CLN2 patients, as well as further studies on the correlation of RNFL changes and CNS degeneration in the framework of this disease. In view of the detected degeneration of various retinal layers in CLN2 disease, ophthalmic examination including OCT imaging should be part of follow-up visits in CLN2 patients.