This study showed that over a 2-year observation period, significant thinning of the pRNFL and mGCIPL was observed throughout the course of the disease, but that these changes were significantly more pronounced early in the disease course. The attenuation of atrophy with longer disease duration is suggestive of a plateau effect. Our findings argue in favor of significant progressive neuronal and axonal injury occurring during the earliest stages of MS, providing support for early intervention to prevent or forestall this injury.
Establishing the timing of neuroaxonal loss in MS is crucial as we explore methods for assessing potential therapeutics meant to treat progressive disease and prevent this decline. It has been presumed in much of the literature that neuroaxonal loss is the substrate underlying volume loss on brain MRI. However, routinely used sequences on brain MRI lack specificity in this regard, making the retina an attractive additional site to monitor. Use of atrophy as a biomarker in progressive MS trials is dependent on the assumption that continued atrophy accumulates in the progressive phase of disease. This study helps to establish that the rate of atrophy in progressive disease may be lower and helps inform timing and sample size calculations for trials using OCT as an outcome.
In the present study, a longitudinal investigation of three retinal layers was performed, the pRNFL, mGCIPL and mINL. Firstly, our data showed significant thinning of the pRNFL (−1.1 µm), over 10 times more than observed in healthy reference subjects (−0.1 µm). This is in line with previously reported findings of longitudinal changes in pRNFL thickness obtained using time domain OCT [16, 18, 31]. In more recent studies using spectral-domain OCT, Narayanan and colleagues reported an annualized decrease in pRNFL of −1.49 μm for MSNON eyes, and −1.27 µm for MSON eyes, but they did not assess for the impact of disease duration or assess a cohort with a broad representation of disease durations . Saidha and colleagues, however, reported a considerably smaller yearly decrease of −0.36 µm (MSON and MNON eyes together) .
Second, regarding the mGCIPL, we reported significant thinning over a 2-year observation period (−1.1 µm). This decrease in mGCIPL thickness, more than twice as much as observed in the reference population, was similar for MSON and MSNON eyes. In a recent longitudinal studies, Narayanan and colleagues reported comparable findings, with a yearly mGCIPL decrease of −0.53 µm in MSON and −0.49 in MSNON eyes , whereas Saidha et al. reported an annual change of −0.34 µm (MSON and MSNON together) .
Third, with respect to the mINL, the present study showed no consistent changes in patients with MS across the cohort, independent of disease duration or disease type. To our knowledge, no other longitudinal studies have investigated changes in mINL thickness. Interestingly, in human retinal postmortem samples, extensive loss of neurons in the mINL has been demonstrated after about 20 years of disease duration . This discrepancy between in vivo and in vitro data would argue in favor of mINL thinning as being a dynamic process, instead of a constant process presenting a steady decrease during the disease course. This hypothesis is further strengthened by the reported increase in mINL thickness associated with MME. The occurrence of MME in MS , is characterized by the presence of cystic areas of hyporeflectivity in the mINL of the retina and it was shown that INL thickness changes were dynamic and were associated with increased disability . Since none of the subjects experienced MME during the study, the present data do not allow for interpretation of the direct effect of MME on changes in mINL thickness, but we do not rule out the possibility of dynamic changes at a time scale not measured within the present study.
Taken together, available longitudinal literature consistently demonstrates progressive loss of pRNFL and mGCIPL in patients with MS. It was, however, previously unclear whether this inner retinal layer damage is a continuous process (as is often assumed with linear modeling), or if the underlying process is dependent on the stage of the disease. Previous studies have shown that retinal axonal injury was already present early in the disease course of MS [20–23]. Although the present data confirm this finding, we have also conclusively shown that the rate of pRNFL and mGCIPL thinning is dependent on disease duration and is greatest in early phases of the disease. Importantly, previous data on annual change of inner retinal layer thickness are inconsistent. Given the findings of the present study, we believe that annual change is clearly influenced by disease duration. Therefore, the inconsistent findings reported in literature are probably partly caused by unaccounted differences in disease duration between study cohorts.
Furthermore, whereas pRNFL and mGCIPL thinning were both present in eyes of patients with relapsing MS, the eyes of SPMS patients only showed substantial thinning of the mGCIPL. This finding is in concurrence with a previous longitudinal study by Henderson and colleagues, who reported no significant decrease in pRNFL thickness after an average follow-up of over 18 months in a cohort of progressive MS patients . This finding suggests that although both neuronal and axonal injuries occur in the relapsing phase of the disease, only subtle neuronal damage continues in the progressive phase. Notwithstanding, this finding should be interpreted with caution, as the effect was [although statistically significant (p = 0.015)], only modestly stronger than observed in healthy references.
The present data support the large body of evidence on the effects of previous episode of MSON on pRNFL and mGCIPL thickness. Consistent with many other studies [10, 36–38], we demonstrated that the eyes of patients who experienced previous episodes of MSON showed significantly more damage of both inner layers at baseline, compared to unaffected (MSNON) eyes. The longitudinal data of the present study, however, extend to these findings, as we demonstrated that although the MSON eyes showed significantly more injury at baseline, the rate of thinning over time was similar for ON affected and unaffected eyes. Consequently, while prior episodes of MSON act as a major confounder in cross-sectional studies, its effect seems considerably less important in longitudinal investigations.
A major strength of the present study was the longitudinal design. Although cross-sectional studies have provided very useful and important data on retinal layer injury in MS, the longitudinal design of the present study, together with the very broadly representative cohort of patients, allowed for interpretation of the dynamics of retinal changes in different stages of MS. Although the 2-year observation period of the present study showed significant changes, the absolute changes were relatively small. Therefore, our current studies are focused on longer observation periods (3–5 years) with repeated OCT and clinical assessments. Our study may have several limitations. First, the control group was relatively small. Although this group was fairly homogeneous and similar to the patient population, a larger sample would increase statistical power. Similarly, although patients with a disease duration across the entire arc of disease were included, those with a progressive disease course (SPMS and PPMS) may have been underpowered, as the majority of patients had relapsing remitting MS. Second, extensive low-contrast visual acuity and color vision testing was not included in the present study. Both measures would provide important information on visual functioning and should be included in future studies .
In summary, this study showed that over a 2-year observation period, significant thinning of the pRNFL and mGCIPL, but not the mINL, was observed in patients with MS throughout the course of disease. These changes in pRNFL and mGCIPL were more pronounced early in the disease course. The attenuation of atrophy with longer disease duration suggests that the earliest phases of disease provide the optimal time to prevent permanent neuroaxonal injury.