In the past decade, inherited retinal diseases (IRDs) have become the most common cause of legal blindness among working age adults in several developed nations (1, 2), having overtaken diabetic retinopathy with the advent of anti-vascular endothelial growth factor (VEGF) agents and improved community screening. X-linked retinoschisis (XLRS) is a relatively uncommon IRD, with a population prevalence estimated between 1 to 200 cases per 100,000 (3–5). XLRS is characterized by separation of the neurosensory retinal layers, often during infancy or childhood (“juvenile” XLRS), resulting in schisis cavities that cause reduced central vision and relative scotomata (6, 7). Reduced structural integrity of the retina may also lead to vitreous hemorrhage, full-thickness retinal breaks, and/or retinal detachment (RD) (8). Strabismus has also been observed in children with deprivation amblyopia secondary to macular schisis (9, 10) or axial hyperopia (11).

The tendency of XLRS to affect the pediatric population and poor visual outcomes due to schisis-related RD has led to considerable interest in development of gene therapies (12–17). XLRS is caused by pathologic mutations in the RS1 gene, which encodes retinoschisin, a 224 amino acid extracellular lectin that binds retinal cell plasma membranes and enables cell-to-cell adhesion (18–21). With a coding sequence of only 675 nucleotides (22), full-length RS1 is easily packaged into conventional adeno-associated viral (AAV) vectors for gene augmentation. However, RS1 gene therapy clinical trials have thus far failed to meet their primary endpoints (14, 23), perhaps due to short time points (e.g., 9 months) of study as well as clinically significant intraocular inflammation that contributed to study discontinuation. As such, long-term functional and structural data to aid in clinical trial design as well as inform patient selection for future studies will increase the likelihood of success of any emerging gene replacement strategies.

Despite the availability of data from several large XLRS clinical cohort studies (24–30), the optimal therapeutic trial design for this disease remains unclear. More specifically, changes in visual acuity over the natural history of disease, relationships between macular schisis severity and visual function, and the relevance of RS1 genotypes to visual prognosis must be clearly understood if demonstrable efficacy of a XLRS gene- or cell-based therapy is sought. In the current work, we performed a retrospective study of the functional, structural, and genetic outcomes of a large cohort of molecularly confirmed XLRS patients from the United States, with the primary aim to demonstrate long-term structural and functional outcomes to better inform current therapeutic approaches, especially clinical trial design.

Development of demonstrably efficacious therapies for IRDs has proven challenging due to the typically slow progression of disease and limited understanding of the natural histories of different IRD entities. In addition, there are limited data on the important prognostic clinical findings of IRDs due to their rarity in clinical practice and seemingly variable presentations despite shared genotypes. Here, we have used a cohort of XLRS patients to elucidate features of disease that clinicians may use for prognosis and clinical trialists may use to improve chances of demonstrating efficacy over the typically brief course of an interventional trial.

One common finding among published XLRS patient cohorts is the slow progression of vision loss (4, 6, 24, 25, 29, 30), which has important implications for the assessment of treatment responses in a clinical trial. We found no decline in median acuity over a median follow-up of 7 years, and a mean acuity decline only equivalent to 1.57 ETDRS letters per year, which is well within the error margin of routine clinical testing. A previous systematic review of XLRS cohort studies (6) demonstrated an average annual decline of 0.22–0.5 ETDRS letters. Despite this decline, however, in those studies, as well as a more recent large cohort study from Belgium and The Netherlands (30), an initial increase in VA was observed in the first to second decade of life before a slow decline phase. This trend of visual acuity increase was not observed in our data, although we did observe progression of outer retinal atrophy over the first two decades of life, with atrophy severity correlating with worsened visual acuity. A smaller US-based cohort study (36) similarly demonstrated that outer segment thickness but not total retinal thickness correlated with visual acuity, in agreement with our current findings. The relative stability of VA through multiple years of life highlights the challenge of using VA as a primary trial endpoint, as well as the importance of patient selection for potential treatment trials.

Inter-eye disease symmetry is a cardinal feature of most IRDs, and this feature is essential for the conduct of interventional trials that utilize untreated fellow eyes as an untreated control. In XLRS, prior studies have suggested good inter-eye correlation using CST as the outcome measure: for example, in one study of 120 patients, an r-squared value of 0.83 was observed, compared to an r-squared value of 0.47 for VA (37). However, our data show greater inter-eye variability than previously observed. Using the same outcome measures, we obtained r-squared values of 0.146 and 0.029 for inter-eye CST and VA, respectively. Only CST was statistically significant, but this exhibited greater deviation from linearity compared to the previous study, probably due to the smaller sample size used in the current study. It follows that interventional clinical trials involving only tens of patients, such as those mentioned previously (14, 23), would be highly susceptible to inter-eye variability confounding the results if a fellow eye control study design is used. Although a fellow eye comparator design is still likely optimal because it minimizes the risks to the untreated fellow eye from the experimental therapy, a large patient number and long follow-up would be required with such a design to demonstrate clinical benefit.

Due to the challenges with VA as a clinical trial endpoint, including accurate quantitative measurement of intra-eye variations in CST, ORA, or related structural measurements will be valuable to consider for trial design. For ORA, however, consideration needs to be given to the extent of ORA at the commencement of the trial, as it seems unlikely that restoration of normal RS1 function would reverse pre-existing photoreceptor loss. Moreover, our observation that ORA continued to develop in a patient despite resolution of visible foveoschisis on OCT suggests that it may not be possible to halt progression of ORA once it begins. Therefore, to maximize structural preservation, XLRS trials should prioritize enrollment of patients with intact outer retinal structure at baseline, which based on our findings would require patients predominantly under 20 years of age. In addition to structural outcomes, it is likely that additional functional outcomes will be essential to demonstrate therapeutic efficacy in XLRS treatment trials. The abnormal responses of XLRS patients on microperimetry, static perimetry, electroretinography, and pupillometry are well established and have been proposed for use as trial endpoints (38–41). In a recently reported AAV2-mediated RS1 gene augmentation trial, clinical response at 12 months was demonstrated only for two endpoints: static perimetry, with two responders of 25 participants, and reduction in cyst cavity volume on SD-OCT, with one responder of 23 participants (23). Despite these limited responses, it is likely that further optimization of both structural and functional endpoints will enable demonstration of therapeutic efficacy in future trials.

Additionally, we acknowledge that there may be additional mechanisms of visual acuity loss in XLRS patients that were not detected in the present study. We did not observe the development of frank optic atrophy in our cohort, but it is conceivable that chronic inner retinal schisis of the macula may lead to damage of the nerve fiber layer and progressive loss of visual acuity. Glaucoma has been found in association with macular schisis in several reports (42, 43), although we are unaware of a study that has investigated this systematically in XLRS patients via OCT of the nerve fiber layer. It may be useful to explore this possibility in currently available XLRS patient cohorts as an additional prognostic factor for vision loss.

Contemporary interventions for XLRS remain limited to management of macular schisis with topical or systemic CAIs (44–46) (VA and CST improve for 2 years then wane), and vitreoretinal surgery in cases of schisis-related RD. (47, 48) We observed a small but significant reduction in CST at 3–6 months following commencement of a topical CAI, consistent with previous reports of CAI anatomical efficacy (46). This was not accompanied by a significant improvement in VA, which is consistent with our finding that CST did not correlate significantly with VA, although we acknowledge that a long-term follow-up may be required to observe the benefits of this intervention. Previous work by Ambrosio and colleagues (46) showed that CAIs offer a transient benefit to VA and CST over 2 years before their effect declines, and we have speculated that the degree of response might be related in part to integrity of deep capillary plexus vasculature (34). We speculate that long-term study of CAI use in XLRS patient may demonstrate efficacy in the form of slowed ORA secondary to CST reduction, although this effect is likely to be modest. Future longitudinal studies using OCT angiography (which was not included in this study) might be informative.

RD is a relatively common complication of XLRS, reported to occur in 0–23% of cases (24, 49). Similarly, we observed RD in 7.7% of eyes in our series, all but one of which underwent vitrectomy with stable retinal findings post-operatively. Perhaps unsurprisingly, peripheral retinoschisis was seen in all eyes with RD, and this was observed more often than in eyes without RD. No eyes in our series were managed with vitrectomy in the absence of RD, although previous work from Japan has suggested than this may decrease the risk of RD in XLRS cases with severe peripheral retinoschisis (50). The same group suggested that foveoschisis may be related to abnormal vitreo-retinal adherence (50), although current literature suggests that retinoschisin is primarily involved in intra-retinal, intercellular adhesion via plasma membrane-bound Na/K-ATPase (20, 51–53). Clinically, our data suggest that XLRS patients with peripheral schisis be counselled about their heightened risk for RD compared to patients without peripheral involvement, but at present the weight of evidence does not support the use of prophylactic vitrectomy in XLRS patients.

In our cohort, a strong association between null RS1 genotypes and moderate or worse visual impairment at the final visit was demonstrated, which was independent of presenting structural features or the age at symptom onset. Potential phenotype–genotype correlations have been suggested previously for XLRS. An electrophysiological study by Vincent and colleagues from the United Kingdom (26) demonstrated that nonsense, splice-site, and frame-shift mutations in RS1 were associated with electronegative ERGs, delayed flicker responses, and abnormal pattern ERGs. Missense mutations in the same study demonstrated more variable ERG responses, consistent with variation in residual RS1 function. In contrast, several groups were unable to demonstrate associations between XLRS disease severity and genotype, the first being a UK study of 86 XLRS patients (25). That study examined visual acuity and structural changes among patients with truncating and missense RS1 mutations but did not find significant differences between the groups, although supporting data were not shown in the final article. A recent large cohort study (30) of 340 XLRS patients from The Netherlands and Belgium did not identify any clear phenotype–genotype correlations linking the severity of RS1 mutations to the clinical severity of disease, although again supporting data were not shown in the article. A Spanish cohort study of XLRS found that severe genotypes were not correlated with RD, VH, or strabismus, while VA was not included as an outcome measure (54). Another paper, based on study of a Chinese cohort of XLRS patients, found that poor visual outcomes tended to associate with null RS1 genotypes (55). We used the cutoff of moderate visual impairment (Snellen 20/63 or logMAR 0.5) or worse as a clinically relevant outcome measure with which to establish a phenotype–genotype correlation because this represents the median visual acuity of XRLS patients in both this and prior cohort studies (7, 30), enabling separation of patients into two relatively evenly-sized groups of visual acuity. Patients with null mutations had 7.8-fold the odds of at least moderate visual impairment at last follow-up, regardless of their initial age at symptom onset. Our additional finding that more extensive ORA occurred more often among null RS1 genotype eyes also hints at an explanation for this correlation. It will be interesting to determine if a null genotype has similarly strong predictive power for functional outcomes when applied to other regional cohorts.

To date there have been two phase 1/2 gene therapy trials for XLRS, both involving intravitreal AAV-mediated RS1 gene augmentation (14, 23). Despite these pioneering efforts, neither trial demonstrated clinically meaningful improvements to retinal function as measured by VA, perimetry, OCT, or electrophysiology. Moreover, both trials had cases of intraocular inflammation in the form of anterior and/or posterior uveitis, though this was generally manageable with systemic steroids. In light of data presented in the current work and elsewhere (6, 9, 25, 30, 37), we suggest that the natural history of XLRS may preclude observation of a clinically meaningful outcome in the brief 12–24 months observation window of a conventional clinical trial. The low average rate of visual decline (<2 ETDRS letters/year), presumably high incidence of amblyopia among XLRS patients, inter-eye asymmetry, and minimal change in macular thickness over long observation periods, should prompt a re-thinking of how an interventional XLRS treatment trial should be designed. Using contemporary structural and functional outcome measures, we suggest that a more prolonged trial period (likely greater than 5 years) will be necessary to observe stabilization of the disease unless patients are enrolled during their visual acuity decline phase, in which case a briefer trial may be possible. A more optimally designed trial would involve young patients with null genotypes, intact outer retinal structure, and normal or near-normal acuity at trial commencement. This may increase the chance of trial success, compared to the phenotypically and genotypically heterogeneous patient groups that are typically used for IRD treatment trials.

The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author.

The studies involving human participants were reviewed and approved by University of Iowa Institutional Review Board. Written informed consent to participate in this study was provided by the participants’ legal guardian/next of kin.

BF and IH conceived the study. BF drafted the manuscript and performed the data analysis. JR, ES, SR, HB, ADr, ADu, JA, and ES critically revised the manuscript for intellectual content. All authors contributed to the article and approved the submitted version.

This work was supported by Iowa Institute for Vision Research, University of Iowa, Iowa City, Iowa, National Institutes of Health P30-EY025580, Research to Prevent Blindness Unrestricted Grant to the University of Iowa, and Chakraborty Foundation Grant (Drack) and Professorship (Dumitrescu).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.