Background
Autoimmune retinopathies (AIR) comprise a spectrum of relatively uncommon autoimmune retinal diseases. Although AIR have been studied for the past 40 years [
1,
2], they remain difficult to diagnose [
3] and treat. AIR include such conditions as paraneoplastic autoimmune retinopathy (pAIR), which can be further subdivided into cancer-associated retinopathy (CAR) and melanoma-associated retinopathy (MAR). In the absence of malignancy, the condition is referred to as non-paraneoplastic autoimmune retinopathy (npAIR). A commonality uniting pAIR and npAIR is that in both conditions, the integrity and function of various retinal cells, including cones, rods, and bipolar cells, are affected by antiretinal antibodies (ARAs) that are believed to arise from molecular mimicry [
4]. The cell types that are most affected in each patient, and thus the initial signs and symptoms, likely depend on which retinal proteins are targeted by the ARAs [
5‐
7]. Consequently, this causes heterogeneity in clinical presentation among patients, including central vision loss, variable changes in visual field, retinal structure, and morphology [
8]. Recently, a panel of experts proposed a list of key diagnostic criteria for AIR, among which included: the absence of an apparent cause for visual dysfunction, an abnormal ERG, and the presence of serum ARAs [
9]. Until now, there is no standard therapy or established treatment protocol, and patient outcomes following intervention are variable. However, a drug called rituximab has garnered interest as a potential treatment option.
Rituximab is a monoclonal antibody that binds to CD20, a non-glycosylated protein expressed on the surface of B lymphocytes (B-cells), inducing B-cell lysis [
10]. It was first approved by the FDA for the treatment of B-cell lymphoma, although recently it has been applied to a variety of autoimmune disorders [
11]. However, its use has not been extensively explored for immune-related retinal conditions, and only case reports and one case series have discussed rituximab administration for patients with AIR [
12‐
19]. Here, we present a case series of five patients exploring the effects of rituximab therapy for the treatment of npAIR as assessed by electrodiagnostic testing. Our aim is to provide a reference for clinicians who are seeking new options for managing this complicated disease and to demonstrate the utility of the ERG as a means of assessing response to immunosuppression in npAIR.
Methods
We performed a retrospective review of all cases of npAIR diagnosed at the Edward S Harkness Eye Institute at New York-Presbyterian Hospital (NYPH) between 2009 and 2016. Five cases were selected based on the following inclusion criteria: (1) they received at least one rituximab infusion during their disease course, and (2) they had a minimum of a six-month follow-up to assess visual function with electrodiagnostic testing, visual acuity, and multimodal imaging, as well as visual field testing when available. The diagnosis of npAIR was based on peer-reviewed diagnostic criteria [
9].
Detection of ARAs in all patients was confirmed by one of two laboratories: The Ocular Immunology Laboratory located at Oregon Health & Science University (Portland, Oregon) or The University of California at Davis Laboratory (Davis, California). Autoantibody detection was performed as previously described [
20,
21]. Briefly, serum was collected from patients and the presence of anti-retinal antibodies was determined by western blot analysis. Western blot band thickness was compared between tests and used to assess the change in response for specific antibodies over time.
Full-field electroretinograms (ffERGs) (Diagnosys LLC, Lowell, Massachusetts, USA) were recorded from both eyes with DTL electrodes according to the standards from the International Society for Clinical Electrophysiology of Vision (ISCEV) [
22] in both scotopic and photopic states. When 30 Hz flicker amplitudes were lower than 5 microvolts (μV), Burian-Allen contact lens electrodes were used to record the electric responses. The amplitudes and implicit times obtained from both eyes of each patient were compared with age-matched normal values, in which the boundary of normal limits represented two standard deviations from the mean.
The main outcome was the effect of rituximab on visual function before and after treatment and was determined based on two independent criteria: (1) stability or improvement in ERG scotopic and/or photopic response, using the last response before treatment with rituximab as the baseline (ratio = 1); and (2) improvement or stability in best corrected visual acuity (BCVA), using the last visual acuity measurement before initiation of treatment with rituximab as the baseline (ratio = 1). Ratios were calculated for ERG and BCVA by comparing the post-treatment response to baseline response (\( \frac{after\ rituximab}{just\ before\ rituximab} \)), such that values greater than 1 indicate improvements in visual function, and values less than one indicate declines in function. Secondary outcomes included subjective stability or improvement on visual field testing; retinal structure, as assessed by spectral-domain optical coherence tomography (SD-OCT) when available; and assessment of antibody titer after rituximab infusions. The following data were retrieved retrospectively for each patient: demographics (age, gender); medical history (history of other autoimmune diseases); clinical features; non-rituximab treatment data (treatment(s), response); and rituximab treatment data (dose, adverse reactions, response).
Discussion
AIR is a rare autoimmune disorder characterized by the production of ARAs that target retinal proteins. Normally when the body encounters a foreign pathogen, B cells bind to a unique antigen, which causes them to mature into antibody-producing plasma cells or memory B cells, which trigger an immune response. However, in autoimmune conditions such as AIR, the B cells become reactive to self-antigens [
25] and begin to produce pathogenic ARAs. This process eventually induces retinal cell death and leads typically to a more rapid and progressive vision loss [
2] compared to hereditary retinal degenerations, which show a slow mean decrease of 10% per year on 30 Hz flicker ERG response in patients with retinitis pigmentosa, for example [
26].
In this case series, we exclusively studied patients with npAIR, which distinguishes itself from other forms of AIR by the lack of malignancy. In our cohort of npAIR patients, a high percentage (80%) were female, which is typical of autoimmune conditions [
27], and 80% had another coexisting autoimmune condition, which is also reported in patients with AIR according to previous studies [
27]. All patients had abnormal ERG responses, rapid disease progression and positive testing for ARAs. Four patients were previously taking an alternative immunosuppressant, and three initially responded well, although the beneficial effects eventually subsided. A recent study reported that in a subgroup of npAIR patients, roughly 63% responded well to immunosuppressive drugs such as cyclosporine, mycophenolate mofetil, infliximab, IVIg or steroids [
27]. Although this estimate is higher than what was observed in other studies [
6], patients’ variable and transient response to immunosuppression is not surprising given the incertitude in the pathophysiology of this disease, including the questionable pathogenicity of many subtypes of ARAs to the retina [
28,
29]. Frequent failure of treatment has encouraged the pursuit of alternative drug strategies, rituximab being one example.
Rituximab is an immunosuppressant that has been used more recently in a number of systemic conditions, including patients with IgG4-related orbital diseases [
30], myasthenia Gravis [
31], neuromyelitis optica [
32], and other ocular inflammatory or autoimmune conditions [
33]. Binding of the drug to cell receptors leads to a rapid depletion in the population of B cells for approximately 6–12 months [
34]. It is hypothesized that rituximab induces B-cell apoptosis through the activation of mitogen-activated protein kinases, natural killer cells, or the complement cascade [
11].
In this case series, we found variable responses among five npAIR patients taking rituximab. Two appeared to stabilize, while one marginally improved and two others did not respond. Although treatment regimen varied between patients, all of them reached adequately low levels of B cells, confirming the potency of the drug. There are several possible reasons why treatment outcomes were inconsistent among patients. As mentioned, four patients had previously been prescribed immunosuppressants, and while three responded initially, they eventually became resistant to treatment. We hypothesize that improvement on rituximab may have been hampered by the limited number of functional photoreceptors or connected secondary-order cells remaining in the retinas of these patients. Additionally, each patient’s genetic and immunological backgrounds may confer greater or lesser amenability of immunotherapy (including to rituximab) for the treatment of their condition, as has been suggested for other immunologically based retinal diseases [
35]. In fact, some studies explain the low sensitivity to rituximab by citing differences in B cell memory capacity for reconfiguration [
36] or the lack of specific complement regulatory proteins on the cell’s surface [
37] due to genetic differences among patients.
In addition, we did not observe a correlation between outcomes and changes in the type or level of antibodies after treatment in four out of four tested patients, which is unexpected. Indeed, in the two patients declared stable on rituximab, types (and level for one patient) of antibodies were similar at 5 and 8 months after starting rituximab. This might be because the CD20 receptor is not present on antibody-producing plasma cells, and thus, levels of immunoglobulin are not expected to decrease after rituximab infusion [
34]. This phenomenon was observed in other studies as well. Looney et al. [
38] reported improvement in patients with Lupus erythematosous following rituximab without changes in anti-double stranded DNA antibody or complement level. While Jarius et al. reported decreased antibody titers after rituximab treatment in neuromyelitis optica, the antibody always remained detectable in almost all patients [
39]. Perhaps stability following treatment without change in antibody levels may be due to decreased antigen presentation rather than antibody level, [
40] i.e. rituximab may be acting not only on B cell depletion, but on T-cell action as well [
11]. Additionally, in P1, ARAs were still measurable after 7 months on rituximab but became undetectable 14 months later. Further studies are indicated to identify the appropriate time interval for repeat testing after baseline, although our data suggest that early on, antibody levels may not serve as a suitable proxy for patients’ response to rituximab. Instead, this test may be best interpreted in conjunction with other indices of retinal function and structure (BCVA, ffERG, multi-modal imaging, etc.).
ERG is a relatively objective test that not only serves as an important tool in diagnosing AIR, but also enables assessment of the severity of retinal dysfunction. Monitoring patients’ response to treatment is challenging, as many visual tests such as BCVA and visual field exams are subjective and vary based on the patient’s affect, learning curve for complex tests, and cooperation [
41‐
43]. For example, Mizener et al. observed that ERG was more sensitive than visual fields in assessing progression of three patients with npAIR [
2]. Despite some variability between sessions, ERG testing in this study also proved to be an effective tool for monitoring patients over time, and changes in ERG outcomes correlated closely with patient symptoms [
44], thus highlighting the utility of this test.
ERG data on npAIR patients is not as abundant as it is for patients with CAR [
45‐
47] and MAR [
48]. In general, while some patients present with greater rod than cone dysfunction initially, for a small minority, cones are affected first [
8]. In this case series, four out of five patients presented with moderate to severe rod-cone dysfunction on ffERG, while only one showed cone-rod dysfunction initially. An electronegative appearance on the maximal response was detected in P1 and P3, suggesting that the inner retinal layers were most affected. The variable effect of different ARAs in each patient may explain this disturbance pattern wherein the inner retinal layers are targeted, which has been demonstrated frequently in MAR but is less commonly reported in CAR [
49] and npAIR [
2]. With progression of the disease, the ERG recording eventually becomes extinguished in most patients. Our findings demonstrate that ERG is an effective strategy for monitoring npAIR patients over time in an objective manner that facilitates clinical decision making by supplementing findings from retinal imaging, BCVA, and visual field tests.
Thus far, two case reports have been published studying the response of patients with npAIR to rituximab, and in each [
14,
16], patients were reported to benefit from the drug, with overall improvements in retinal function. Additionally, one case series studied six npAIR patients receiving rituximab and/or combination therapy [
19]. They found that following mono- or combinatorial therapy, 66.7% of eyes had stable visual acuity, 50% showed stability on visual field testing, and 33.3% showed stability or improvement on ERG. They also found that at least one pathogenically proven ARA band resolved after treatment. Overall, they concluded that stability or improvement on two or more tests in 83.5% of patients could be considered a successful treatment. Contrastingly, we observed much more variability among patients in our cohort and concluded that only 60% were stabilized or improved following treatment.
There are several notable differences in study design that may account for the differences observed in our findings. The primary difference lies in the standardization of the dosage of rituximab, which was administered at 375 mg/m
2 every week for 8 weeks, then 375 mg/m
2 monthly [
50,
51]. Contrastingly, patients in our study were treated with rituximab in a customized fashion based on symptoms and specialist preferences. A secondary difference is their combinatorial approach, wherein rituximab was co-administered with oral cyclophosphamide or bortezomib in 4 out of 6 of their patients, while ours were treated with rituximab exclusively. However, there are many points of consensus between our studies: visual acuity was on average stabilized in both cohorts, adverse events occurred in a minority of patients, and we both found unpredictable results in ARA outcomes, making their interpretation challenging. Additionally, our findings build on Foster et al.’s by supplying OCT and ARA titer testing results, although repeated testing among larger cohorts is still greatly needed.
Some limitations of our study should be acknowledged. While one of our patients was under 10 years of age and one older than sixty, the typical npAIR diagnosis is made between the ages of 20 and 25 years of age. Additionally, the regiment of rituximab was not standardized across patients. Some patients followed the rituximab protocol that was developed for B-cell lymphoma, while some were prescribed a regiment that was originally designed for rheumatoid arthritis patients, and yet others used non-standard protocols. Additional experience with other patients with autoimmune disorders and retinal atrophies may guide dosing of rituximab in the future. Additionally, the time point of assessment and antibody testing following rituximab infusions were also different for each patient, and the optimal follow-up time after infusion cannot be determined from these data. Testing blood before and at varying intervals after each rituximab infusion is one strategy that could determine the ideal timing for ARA testing in future studies, which in turn may enable better delineation of the drug’s effects on ARAs. Finally, ratio analysis for ERG was used in this series in order to obtain an efficient comparison strategy of responses at different points for each patient, although actual voltage numbers could also have been used.