Relation between ocular paraneoplastic syndromes and Immune Checkpoint Inhibitors (ICI): review of literature

Abstract

Purpose

To describe different ocular paraneoplastic syndromes in patients treated with Immune Checkpoint Inhibitors (ICI), its relation with different types of ICI and different types of tumors, and its implications for treatment.

Methods

A comprehensive review of the literature was performed.

Results

Patients treated with ICI can present with different ocular paraneoplastic syndromes, such as Carcinoma Associated Retinopathy (CAR), Melanoma Associated Retinopathy (MAR) and paraneoplastic Acute Exudative Polymorphous Vitelliform Maculopathy (pAEPVM). In literature, the different types of paraneoplastic retinopathy are mostly related to different types of primary tumors, with MAR and pAEPVM seen in melanoma, and CAR in carcinoma. Visual prognosis is limited in MAR and CAR.

Conclusion

Paraneoplastic disorders result from an antitumor immune response against a shared autoantigen between the tumor and ocular tissue. ICI enhance the antitumor immune response, which can lead to increased cross-reaction against ocular structures and unmasking of a predisposed paraneoplastic syndrome. Different types of primary tumors are related to different cross-reactive antibodies. Therefore, the different types of paraneoplastic syndromes are related to different types of primary tumors and are probably unrelated to the type of ICI. ICI-related paraneoplastic syndromes often lead to an ethical dilemma. Continuation of ICI treatment can lead to irreversible visual loss in MAR and CAR. In these cases overall survival must be weighed against quality of life. In pAEPVM however, the vitelliform lesions can disappear with tumor control, which may involve continuation of ICI.

Background

Immune checkpoint inhibitors are considered a recent breakthrough in the treatment of advanced cancers [1]. The immune system contains several checkpoints to prevent overactivation against healthy cells. However, tumor cells use these checkpoints to escape the immune system. In some tumors, there is an upregulation of checkpoints on the T-cell surface, including cytotoxic T-lymphocyte antigen-4 (CTLA-4) receptor, and programmed death-1 (PD-1) receptor, thereby suppressing T-cell activation against tumor cells. Blocking this inhibitory interaction enhances a specific antitumor T-cell response.

To date, various PD-1 (pembrolizumab, nivolumab), PD-ligand-1 (PD-L1; atezolizumab), and CTLA-4 inhibitors (ipilimumab) have been approved in the treatment of several malignancies, including melanoma, non-small-cell lung carcinoma, and other advanced tumors.

The development of these new drugs has improved survival rates. However, immunotherapy removes a protection against autoimmunity allowing various immune-related adverse events (IRAE), with the most common being pneumonitis, hepatitis, colitis, dermatitis, and endocrinopathies [2, 3].

Ophthalmologic IRAE are rare and have been reported in less than 1% of patients [4,5,6]. Exact rates, however, are difficult to obtain. They typically develop within weeks to months of initiating therapy and can affect various parts of the eye and orbit. Most frequently reported ophthalmic adverse events include dry eye disease and uveitis (anterior uveitis, Vogt-Koyanagi-Harada disease-like uveitis). Other reported side effects are conjunctivitis, (peripheral ulcerative) keratitis, inflammatory orbitopathy, orbital myositis, myasthenia gravis, optic neuropathy, acute macular neuroretinopathy, and paraneoplastic syndromes, such as Carcinoma Associated Retinopathy (CAR), Melanoma Associated Retinopathy (MAR) and paraneoplastic Acute Exudative Polymorphous Vitelliform Maculopathy (pAEPVM).

Ocular paraneoplastic syndromes have been well described, but the evolution after treatment with ICI remains unclear. Therefore, we conducted a literature review to systematically map the research done in this area and identify existing gaps in knowledge. We focus mainly on its pathophysiology, clinical characteristics, diagnosis, and current treatment.

Materials and methods

We performed a comprehensive literature search of the medical databases Medline (PubMed), and Embase, and Web of Science. The methodology of this literature review was written following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) statement (Additional file 1). The search strategy is given in Additional file 2. To identify potentially relevant articles, two reviewers (PC and PPS) screened all search results based on the title and abstract. Selected full-text articles were then reviewed for eligibility. To avoid missing any relevant research, one reviewer (PC) performed snowballing, by which 24 additional articles were included. Two articles were found through hand searching. Additional file 3 provides a detailed overview of inclusion and exclusion criteria.

Results

An overview of the available literature on this rare retinal manifestation is presented in Tables 1, 2 and 3.

Table 1 Cases of Melanoma Associated Retinopathy (MAR)

Table 2 Cases of Carcinoma Associated Retinopathy (CAR)

Table 3 Cases of paraneoplastic Acute Exudative Polymorphous Vitelliform Maculopathy (pAEPVM)

MAR

We found nine cases of MAR related to ICI administration: 3 patients received the combination of ipilimumab and nivolumab, 3 pembrolizumab, 1 ipilimumab, 1 nivolumab, and 1 ipilimumab + nivolumab + pembrolizumab [7,8,9,10,11,12,13,14,15, 11, 12]. The mean age was 67.75 years (range 56 - 79), and there was an equal gender ratio (1 patient not specified (NS)). All patients had known metastatic melanoma with a history of surgery in 7 out of 9 (1 with radiotherapy and dacarbazine); in 2 patients any previous treatment was not reported. The three most frequently described presenting symptoms include visual impairment, photopsia and nyctalopia. Mean best corrected visual acuity (BCVA) at presentation was 20/35 (3 NS). Time to onset varied from a few days to a maximum of 5 cycles and in 3 cases MAR was already present before the start of ICI. In 5 cases, antiretinal antibodies were found with TRPM1, aldolase and carbonic anhydrase II (CA II) as the 3 most frequent.

The antitumor efficacy of ICI was a complete response in 37.5% (3/8), a partial response in 50% (4/8), and stable disease in 1 case (1 NS). Other IRAE occurred in 6 of 9 patients. MAR was treated with corticosteroids in 7 of 9 patients (3 systemic, 2 intraocular, 1 topical and 1 subtenon), 3 patients also received an intravitreal injection of anti-vascular endothelial growth factor to treat Macular Neovascularization (MNV). One patient received intravenous immunoglobulins (IVIG) in addition to corticosteroids. In 4 of 8 cases ICI was discontinued, but in none of the cases there was a rechallenge. BCVA was reported as an ophthalmic outcome in 7 cases (worse in 3, stable in 3, and better in one eye but worse in the other eye in 1 case). Improvement was seen in an eye with MNV. Inflammation resolved under corticosteroids. The mean follow-up was 56.1 weeks (range 3 - 182).

CAR

Five CAR cases have been described: 2 with nivolumab, 1 with atezolizumab, 1 with pembrolizumab and one with the combination of nivolumab and ipilimumab [16,17,18,19,20]. There was an equal male-female ratio and the mean age was 62 years (range 52 - 75 years, 1 case age and gender NS). CAR was associated with lung carcinoma (n=2), hepatocellular carcinoma (n=1), cervical carcinoma (n=1) and endometrial carcinoma (n=1). One patient received chemotherapy concurrently and 1 lenvatinib (protein kinase inhibitor), 3 patients had already been treated before the start of ICI (1 chemotherapy, 1 chemotherapy + radiotherapy, and 1 surgery + radiotherapy + chemotherapy; 1 NS). Photopsias are the most frequently reported symptoms (n=3; 1 NS) and mean BCVA at presentation was 20/60 (1 NS, range No Light Perception-20/20). Time to onset was shortly afterwards (3 weeks, 2 cycles and “shortly thereafter”) in 3 patients, 18 months in 1 patient and was not reported in 1 case. Antiretinal antibodies were detected in 4 patients (CA II (n=2), TULP1, recoverin, GAPDH, 38 kDa, PKM2, 112 kDa, enolase and arrestin). The antitumor efficacy of ICI has not been discussed in any article. In 2 cases other IRAE occurred (arrhythmia, electrolyte imbalance, hypothyroidism, diarrhea, pericardial effusion, and memory loss).

In 3 of 5 patients CAR was treated with systemic corticosteroids, in 1 case this was in combination with rituximab. In all patients there was an improvement in presenting complaints (n=4; 1 NS). Visual acuity remained stable or improved in all cases (1 NS). In 4 patients ICI was discontinued (1 NS) and in 1 patient rechallenge together with corticosteroids and rituximab did not lead to a recurrence. The mean duration of follow-up was 9.25 months (range 3 - 24 months, 1 NS).

pAEPVM

The search strategy yielded nine cases of pAEPVM related to ICI (ipilimumab n=4, nivolumab n=3, pembrolizumab n=2) [21,22,23,24,25,26,27,28]. The mean age was 62.8 years (range 46 - 78). 5 of 9 patients were male. The primary tumor in all cases was a melanoma, mainly mucocutaneous. In 2 patients the tumor had already been treated surgically, 1 had a history of surgery and radiotherapy, and 1 of surgery, chemotherapy and nivolumab. The patient described by Sandhu et al. was previously treated with a B-type Raf proto-oncogene (BRAF) inhibitor, Mitogen-activated protein kinase kinase (MEK) inhibitor and ipilimumab. Pembrolizumab was given concomitantly with 2 BRAF inhibitors, dabrafenib and vemurafenib. 1 patient received ipilimumab after nivolumab. In 3 patients, ICI was the first-line treatment (1 patient NS). Mild loss of vision is the most frequently described symptom, reported in 6 out of 9 patients. BCVA at presentation was 20/25 (range 20/100 - 20/20). Time to onset averaged 10.25 weeks (range 3-28). The antitumor efficacy of ICI was discussed in 5 of the 9 cases and varied widely: progression (n=2), partial remission (n=1), reduction after rechallenge (n=1), and no recurrence (n=1). Other IRAE occurred in 33% (immune-related thyroiditis, sarcoid-like syndrome, elevated liver transaminases, and pneumonitis). The ICI was stopped in 5 patients and in the case of Sandhu et al.. Vemurafenib (BRAF inhibitor) was stopped. In 3 cases no additional treatment was started, and 5 patients received corticosteroids (systemic (n=2), intraocular, topical and in 1 case together with chemotherapy). In the case of Lincoff et al., pAEPVM was already present before the start of ipilimumab. After surgery and initiation of this ICI, a slow improvement in symptoms occurred. Only in 1 of 5 patients (Kemels et al.) a rechallenge occurred together with surgical resection of the primary tumor, after which a significant reduction of the subretinal fluid (SRF) was noted. In most cases there was a resolution of the SRF (n=6), and the subretinal deposits (n=3) persisted. The mean duration of follow-up was 17.4 weeks (range 3-40).

Discussion

We describe the findings of ocular paraneoplastic syndromes with checkpoint inhibitors. A comparison between the three paraneoplastic syndromes is presented in Table 4.

Table 4 Characteristics of paraneoplastic syndromes

Interestingly, these paraneoplastic syndromes are mainly seen in specific primary tumors. For example, MAR is exclusively described in melanomas; CAR mainly in patients with small cell lung carcinoma, but is associated with a variety of cancers. pAEPVM has also been documented in several melanoma and carcinoma cases, but is often related to mucosal melanoma. However, ICI were initially only indicated in metastatic melanoma, which may skew these results.

MAR, CAR and pAEPVM are rare retinopathies that can occur without or after initiation of ICIs. Given that only case reports exist for now, the exact incidence of these paraneoplastic syndromes whether or not in association with ICIs is currently unknown. Since ICIs can induce an increased anti-tumor response, a potential cross-reaction may result in exacerbation or induction of a predisposed paraneoplastic phenomenon.

The exact underlying pathophysiology is not yet fully understood, but molecular mimicry is the globally accepted mechanism. Presumably, the increased anti-tumor response induced by ICI leads to an increased cross-reaction of antibodies against non-tumor antigens; namely against the Retinal Pigment Epithelium (RPE) in pAEPVM, against bipolar cells in MAR, and against photoreceptors in CAR [15, 22, 23, 29,30,31,32].

Antiretinal autoantibodies give rise to bilateral retinal damage and visual disturbances, which are much more pronounced in CAR and MAR compared to pAEPVM [26, 30, 32, 33]. In CAR, cone dysfunction results in a decrease in visual acuity, impaired color vision, and central scotomas. A dysfunction of the rods is more likely to lead to prolonged dark adaptation, nyctalopia and (mid)peripheral visual field defects/scotomas.

The time to onset varies between 2 weeks and 18 months.

In pAEPVM, the antibodies probably directed against RPE, disrupt their pump and transport function. It is believed to be an immune response against bestrophin [34, 35]. The clinical picture therefore resembles autosomal recessive bestrophinopathy with the only difference that the latter has a shallow anterior chamber. Subretinal fluid and subretinal accumulation of yellowish material occurs at the posterior pole [21, 22, 25]. These vitelliform lesions are typically hyperautofluorescent indicating lipofuscin deposition in the RPE cells [36]. Optical coherence tomography (OCT) shows zones of subretinal fluid and deposits of hyperreflective material. Fluorescein angiography (FA) reveals blockage at the vitelliform lesion without retinal or optic nerve leakage.

In CAR and MAR fundoscopic findings are initially rather subtle with sometimes retinal vessel attenuation, and presence of intraocular inflammation; evolving into retinal pigment epithelial mottling, retinal atrophy, and optic disc pallor [30, 33, 37]. OCT shows loss of the outer retinal layers with foveal sparing. A (para)central scotoma can be visualized on the visual field. Findings on fundus autofluorescence (FAF) and FA are rather variable and not pathognomonic. In CAR, FA sometimes shows retinal vasculitis. Hyperautofluorescence around a hypoautofluorescent zone reflects the actively affected photoreceptors in CAR.

Full-field electroretinography (ERG) provides an objective evaluation of retinal function and is therefore an important diagnostic test [33]. In CAR, depending on the degree of damage to the rods and/or cones, a reduction of the a-wave and consequently b-wave is seen, most pronounced in photopic and/or scotopic conditions. In MAR, ERG reflects impaired ON-bipolar cell function which typically manifests as an electronegative ERG. This pattern is also seen in the complete type of congenital stationary night blindness (cCSNB) [38].

In pAEPVM, a normal ERG is seen.

In addition to its diagnostic value, ERG can also be considered as an indicator of treatment response.

Antibody testing, detected by Western blot, enzyme-linked immunosorbent assay, or immunohistochemical methods, is another interesting diagnostic tool [33]. Numerous antiretinal antibodies have been characterized in CAR and MAR [39, 40]. The most commonly described antiretinal antibodies include recoverin, a 23 kDa calcium binding protein found on photoreceptors; α -enolase, a 46 kDa ubiquitous glycolytic enzyme; arrestin (48 kDa), CA II (30 kDa), and transient receptor potential cation channel subfamily M member 1 (TRPM1) expressed on retinal ON bipolar cells. TRPM1 mediates its depolarization in response to light, which is reflected in the b-wave on ERG. Mutations in the TRPM1 gene have also been documented in CSNB [41, 42]. In autoimmune retinopathy the seropositivity for known antiretinal antibodies at presentation is only 50 - 65% [43,44,45]. In addition, antiretinal antibodies can also be found in control patients. The absence of antiretinal antibodies therefore does not exclude the diagnosis.

Given the progressive visual impairment especially in CAR and MAR, rapid diagnosis and early treatment initiation is crucial. However, the treatment of ocular paraneoplastic syndromes can be challenging. Many treatment options have been described in literature, but globally there are two strategies [15]. On the one hand, reduction of autoimmunity can be achieved through immunosuppression or immunomodulation. On the other hand, tumor cytoreduction, obtained by surgery, chemotherapy or immunotherapy, can lead to decreased tumor antigen production and thus decreased cross-reaction [46].

With better tumor control by resection of the primary tumor or good effect of ICI, the tumor load can be reduced or disappear, resulting in a reduced T-cell and secondary decreased B-cell response with consequently less cross-reaction [30]. Hence, sometimes, improvement can occur after using ICI as described in some articles [7].

Strikingly, paraneoplastic syndromes might be associated with a favorable tumor response in metastatic melanoma [47].

On the other hand, it is sometimes difficult to wait for the beneficial effect, because damage can occur fairly quickly, especially with CAR and MAR. This damage is irreversible, even after tumor control. In those cases, it may be indicated to stop the ICI and still try corticosteroids and/or other immunosuppressive/immunomodulatory therapy. Since there is no pronounced decrease in vision with pAEPVM, a wait-and-see approach can be considered [22].

Suppression of autoimmunity can be achieved through multiple mechanisms, such as corticosteroids, rituximab, IVIG, and plasmapheresis; however, there is conflicting evidence in literature, with varying degrees of success [30, 46].

Tapering dose systemic corticosteroids are also sometimes administered. However, the potential negative impact of this drug on tumor response should be taken into account when used before or in conjunction with ICI [48]. Therefore, this decision is always made in consultation with an oncologist.

Ideally, the treatment provides good tumor control, resulting in less cross-reactivity. Furthermore, the Ig(immunoglobulin)-mediated side effects should be tackled, without compromising tumor response.

Novel immunotherapeutic drugs, such as efgartigimod or rozanolixizumab aim at reducing pathogenic autoantibodies by inhibiting the neonatal Fc receptor (FcRn) for binding immunoglobulin G (IgG) [49]. These drugs have a high affinity for FcRn and compete with IgG to bind this receptor. Since FcRn protects IgGs against lysosomal degradation and thereby prolongs their half-life, these drugs can reduce circulating IgG antibodies. These new drugs target pathological IgG and thus may act specifically on humoral immunity while not affecting cellular T cell immunity which is important for tumor control. This may show promise in paraneoplastic exacerbations after ICI.

Based on the pathophysiology, pAEPVM is known to be reversible, which explains its relatively favorable visual prognosis. This is in contrast to CAR and MAR where the damage at the level of the photoreceptors or bipolar cells is irreversible, resulting in a poor visual prognosis. This is in line with the included case reports in which a fairly good visual outcome is described for pAEPVM, in contrast to CAR and MAR.

Conclusion

Immune checkpoint inhibitors can induce an exacerbation of paraneoplastic syndromes via an increased antitumor response and thus cross-reaction against ocular structures, among others. The type of paraneoplastic syndrome varies by tumor. The diagnosis is mainly clinical, in which electroretinography and determination of serum antiretinal autoantibodies offer a diagnostic added value, especially for CAR and MAR. The treatment remains controversial where good tumor control is desired with consequent reduction of cross-reactivity, combined with suppression of immunoglobulin-associated side effects.