Immune Checkpoint Inhibitor Associated Optic Neuritis

Abstract

The three classes of immune checkpoint inhibitors can be associated with optic neuritis, which typically presents with bilateral, painless visual decline. Visual function stabilized with drug cessation and systemic steroids.

Immune checkpoint inhibitors (ICIs) are a group of systemically administered immunotherapy agents that enhance the adaptive immune response and are now widely used to treat many forms of cancer.¹ They include monoclonal antibodies that block and inhibit cytotoxic T-lymphocyte associated antigen-4 (CTLA-4, ipilimumab and tremelimumab), programmed cell death protein 1 (PD-1, nivolumab, pembrolizumab, cemiplimab) and programmed cell death ligand 1 (PD-L1, atezolizumab, avelumab, durvalumab). Generally, ICIs potentiate the T-lymphocyte response, including reactivity against tumor cells. However, unrestrained T-cells can also aberrantly target normal tissue and trigger a range of adverse inflammatory events, including ophthalmic and neurological phenomena.²

Inflammatory ophthalmic conditions occur in 1% of patients receiving ICIs and can impact almost any aspect of the ophthalmic system.^1,2 At present, there are five published single case reports of ICI-associated optic neuritis (ON).^2–6 In this retrospective case series, we report a large series of 18 eyes with ICI-associated optic neuritis in 11 patients diagnosed at our three centers between 22^nd May 2017 and 15^th January 2020. The study was compliant under Memorial Sloan Kettering Institutional Review Board and the Declaration of Helsinki. Informed consent was obtained in all patients.

All relevant data were recorded for each patient and affected eye. Demographics, clinical features, and treatment selection with outcomes are summarized in Table 1 (and examples in SFig 1 & 2 (available at www.aaojournal.org)). Clinical assessments included Snellen visual acuity (VA), presence of pupillary afferent defect (APD), color vision (Ishihara) and Humphrey visual fields (HVF). The presenting HVF defects were characterized as diffuse, central/cecocentral, arcuate/altitudinal, or normal. For magnetic resonance imaging (MRI) assessment, enhancement or T2 hyperintensity was recorded for optic nerve segments (orbital, canalicular, intracranial, combination of segments, none) in addition to the presence of demyelinating lesions.

Table 1:

Clinical characteristics of patients, their systemic disease, presenting symptoms, their eyes, their ophthalmic findings and their outcome

Patient	Age (yrs)	Gender	Drug at time of dx	No. of ICI cycles prior to ophthalmic dx	Time to follow up (mos)	Alive(Y/N)	Primary Cancer Dx	Ocular symptoms	Pain (Y/N)	Other intraocular inflammation
1 (MSK)	61	F	Ipilimumab 3mg/Kg + Nivolumbab 1mg/kg q 3weeks	3	32.53	Y	Cutaneous melanoma	Decreased vision, floaters, “smudge”	N	Anterior uveitis OU
2 (MSK)	58	F	Ipilimumab 3mg/Kg + Nivolumab 1mg/kg q 3weeks, followed by Nivolimab	52	12.10	Y	Cutaneous melanoma	Decreased vision, Floaters, halos	N	None
3 (MSK)	59	F	Nivolumab 240mg q3wks	2	7.03	N	SCLC	Decreased vision, “large” floater	N	None
4 (MSK)	71	M	Pembrolizumab 200mg monthly, pemetrexed 200mg	3	6.47	N	NSCLC	Decreased vision	N	None
5 (MSK)	54	F	Ipilimumab 1mg/Kg q6wks + Nivolumab 240mg q3wks	4	2.53	Y	NSCLC	Decreased vision	N	Choroiditis, intraretinal edema OU
6 (MSK)	68	M	Ipilimumab 3mg/Kg + Nivolumbab 1mg/kg q 3weeks	4	2.43	Y	Cutaneous melanoma	Peripheral vision loss	N	None
7 (MSK)	73	F	Atezolizumab 1200mg q3weeks	95	1.60	Y	Renal Cell Carcinoma	“Big” Floaters	N	Anterior uveitis, rare vitreous cells
8 (MEE)	69	F	Pembrolizumab 2 mg/kg q3 weeks	7	4.47	N	NSCLC	Decreased vision	N	None
9 (MEE)	65	M	Ipilimumab 3 mg/kg + Nivolumab 1 mg/kg q3 weeks	4	1.17	Y	Cutaneous melanoma	Decreased vision	N	None
10 (MEE)	63	M	Nivolumab 2 mg/kg q2 weeks	10	3.37	Y	Cutaneous melanoma	Decreased vision, redness	Y	Anterior uveitis OU
11 (WEI)	58	M	Ipilimumab 3mg/kg + Nivolumab 1mg/kg q3 weeks; Nivolumab 1mg/kg rechallenge once	4	6.00	Y	SCLC	Decreased vision	N	None

Patient	Laterality of Optic Neuritis	APD involved eye? (Y/N)	Optic nerve exam at dx	MRI type (sequences): findings	Initial visual field pattern (mean deviation dB)	Ophthalmic Treatment	Optic nerve exam at after treatment	Posttx visual field pattern (mean deviation dB)
1 (MSK)	Bilateral	N	Trace nasal edema OU	MRI orbits (T1, T2, fatsat): central T2 hyperintense changes of bilateral intraorbital optic nerves continuous to prechiasm, enhancement of sheath complexes	Cecocentral detect OD (NR)	PO Pred 80mg w/ taper over 1 mos, topical prednisolone, timolol/dorzolamide	Trace nasal pallor	Cecocentral detect OD (NR)
2 (MSK)	Bilateral	N	Trace diffuse edema OU	MRI brain and orbits (T1, T2, fatsat, DWI): no abnormality	Cecocentral defect OU (−6.82 OD, −6.73 OS)	IV DXM 3g, IVIg x 1, ritux x 3, PLEX x 5, PO pred 80mg taper over 2 mos	Trace diffuse pallor OU	Cecocentral defect OU (−5.64 OD, −6.44 OS)
3 (MSK)	Unilateral (left)	Y	Full disc OS, with hemorrhage, progressed to 2+ disc edema	MRI orbits at time of 20/30 vision (T1, T2, fatsat, DWI, FLAIR): no abnormality	Diffuse OS (−24.81 OS)	PO pred 60mg	Pallor of superior left disc	NA
4 (MSK)	Bilateral	Y	Diffuse disc pallor and edema	MRI orbits (T1, T2, fatsat, DWI): Papilledema, bilateral enhancement intraorbital optic nerve sheaths	Diffuse OU (−30.4 OD, −24.86 OS)	IV methylpred 1g x 5 days, PO pred 80mg w/ taper over 2 mos	Disc pallor with resolved edema, thinning OU	Diffuse OU (−29.25 OD, − 24.75 OS)
5 (MSK)	Bilateral	Y	Disc edema OU	MRI brain and orbits (T1, T2, fatsat, DWI): mild enlargement of left greater than right distal intraorbital optic nerve, no enhancement	Inf arcuate OU (−4.2 OD, −9.48 OS)	PO pred 80mg w/ taper over 2 mos	Normal	No defect (0.95 OD, − 0.35 OS)
6 (MSK)	Unilateral (right)	Y	Disc sectoral edema OD (SN)	MRI brain (T1, T2, fatsat DWI, FLAIR): no abnormality	Inf altitudinal OD, Inf arcuate OS (−15.8 OD, −5.85 OS)	IV methylpred x 3days, PO pred 1 mg/kg/day w/ taper over 2 mos	SN disc pallor, corresponding NFL OCT thinning	Inf arcuate OD, Inf arcuate OS (−6.83 OD, 0.03 OS)
7 (MSK)	Bilateral	N	2+ optic nerve edema OU	None	No defect OU (0.23 OD, 0.44 OS)	PO pred 80mg w/ taper over 2 mos	Normal	No defect OU (0.69 OD, 0.463 OS)
8 (MEE)	Unilateral (right)	Y	Optic nerve edema OD	MRI orbits (T1, T2, fatsat, DWI, FLAIR): no abnormality	Inf arcuate defect OD (−12.37)	IV methylpred x 48 hrs, PO DXM 4 mg	Optic nerve pallor OD	NA
9 (MEE)	Unilateral (left)	Y	Optic nerve edema OS superior>inferior, flame hemorrhages	MRI orbits (T1, T2, fatsat, DWI, FLAIR): no abnormality	Cecocentral OS (−2.84)	None	Slightly less edema, resolved flame hemorrhages	NA
10 (MEE)	Bilateral	N	Optic nerve edema OD, and then edema 2 weeks later OS	MRI orbits (T1, T2, fatsat, DWI, FLAIR): no abnormality	Inferior arcuate OD, superior altitudinal OS (NA)	IV methylpred, PO pred 60 mg w/ taper. topical difluprednate, timolol/ dorzolamide, brimonidine	Optic nerve pallor OU	Inferior arcuate OD, superior altitudinal OS (6.09 OD, −9.52 OS)
11 (WEI)	Bilateral	Y	Pink, C/D asymmetry OU (0.5 OD; 0.7 OS)	MRI orbits (T1, T2, fatsat, DWI, FLAIR): Left greater than right optic nerve enhancement: Prechiasmal OU, canalicular OS	Diffuse OU (−22.71 OU)	IV methylpred 1g x 5 days and 5 PLEX, PO pred 50mg w/ taper over 6 mo	Pink OD, 4+ pallor OS	NA

dx = diagnosis, ICI = immune checkpoint inhibitor, SCLC = small cell lung cancer, NSCLC = nonsmall cell lung cancer, MSK = Memorial Sloan Kettering, MEE = Mass Eye & Ear, WEI = Wilmer Eye Institute, APD = afferent pupillary defect, dx = diagnosis, MRI = magnetic resonance imaging, HVF = Hymphrey visual field, OU = oculus uterque, OD = oculus dextrus, OS = oculus sinister, Inf = inferior, sup = superior, SN = superonasal, MSK = Memorial Sloan Kettering, MEE = Mass Eye & Ear, WEI = Wilmer Eye Institute, pred = prednisone, DXM = dexamethasone, ritux = rituximab, PLEX = plasmaexchange, Posttx = post-traetment, mos = months, NR = not recorded, T1 = T1-weighted, T2 = T2-weighted, fatsat = fat-saturation, DWI = diffusion-weighted imaging, FLAIR = fluid-attenuated inversion recovery, NA = not available

Our cohort confirms, as in the five published cases,^2–6 that all three classes of ICIs (anti-CTLA4, anti-PD-1, and anti-PD-L1) have the potential to induce ON, suggesting their common mechanism of immune activation is the principle etiology of the disease. Our cohort reveals a unique feature regarding the timing of this disease. Patients received a median of four drug cycles prior to ICI-associated ON, however, timing of onset was variable and ranged from 2 to 95 cycles. The four patients who developed “late” optic neuritis after the 4-cycle median, were all maintained on prolonged checkpoint inhibition monotherapy. This suggests that the more toxic combination therapy may impart a short-term risk for optic neuritis, but that protracted use of a single ICI is associated with a delayed risk.

ICI-associated ON has clinical features that are not entirely consistent with the “classic triad” (unilateral decreased vision, dyschromatopsia 92% and pain >90%) of optic neuritis secondary to other etiologies. All patients in our cohort were symptomatic, with the vast majority (90%) complaining of painless decreased vision and/or “floaters” (despite no vitreous opacities). 64% of patients had bilateral optic nerve involvement. 89% of eyes had at least one objective measurement of optic nerve dysfunction besides decreased VA, including APD and HVF abnormality (15 eyes, 83%); but dyschromatopsia in only 67% of eyes (Table 2, available at www.aaojournal.org). The unique features of bilateral, painless decline of vision in the context of intact color vision, are unlike “classical” optic neuritis.

Ancillary testing can be particularly helpful in supporting the diagnosis, particularly in cases of subtle clinical findings. The optic nerve was variably abnormal in all eyes (Table 1). However, in four eyes with subtle findings (trace pallor and VA > 20/30), HVF testing and/or MRI were helpful in establishing diagnosis. Lumbar puncture did not reveal malignant cells or infectious etiology in any of the four patients tested; and serum paraneoplastic and autoimmune testing was negative/inconclusive in three patients tested. ON in the context of ICI would raise the suspicion for drug-associated disease. Metastatic optic nerve infiltration would be the next likely diagnosis, but this is unlikely with the primary malignancies in this cohort.

At the time of ON diagnosis, the ICI was discontinued in all patients. All but one patient received steroids, given orally (4 pts, 36%) or intravenously with oral taper (6 pts, 55%). Two patients received rituximab, intravenous immunoglobulin (IVIG), and/or plasmapheresis. In 16 eyes with worse than 20/20 or 20/25 baseline vision, the VA improved in 12, was stable in 2, and worsened in 2. The mean improvement in vision from presentation to follow-up was 2 Snellen lines.

Historically, it was believed that inflammatory insults to the optic nerve typically impacted the papillomacular bundle, and therefore would be associated with a central scotoma.⁷ However, in the Optic Neuritis Treatment Trial (ONTT), the most common presenting HVF pattern was diffuse in 48%.⁷ If diffuse field loss by 24-degrees Humphrey is considered equivalent to a central deficit by Goldmann perimetry, then the majority of patients in the ONTT did display a central deficit.⁷ By contrast, this cohort had a relatively even distribution of visual field patterns including diffuse (28%), arcuate/altitudinal (39%), central/cecocentral (22%), and no defect (11%). This suggests that either ICI-associated inflammation may not favor the papillomacular bundle or that inflammation leading to ischemia may play a role in the pathogenesis. However, it is difficult to draw conclusions from HVF alone, because ascribing certain causes of HVF deficits to specific locations of nerve fiber bundle damage is not always reliable.⁷

The MRI imaging in our cohort demonstrated a propensity for ICI-associated ON to spare the retrobulbar and proximal nerve (Table 1). The reason for this is unclear, but may suggest differences in the level of ICI-associated immune activation along the length of the nerve. Furthermore, these MRI findings do not provide a consistent explanation for the painless nature of this optic neuritis. An intraorbital inflamed nerve sheath would be in close proximity to mechanical disruption by extraocular muscle movement, and more prone to inducing pain. One patient had this feature (patient 4), but he reported no pain. Strikingly, pain was reported in only a single patient with papillitis (patient 10), but this patient had concomitant anterior uveitis. The relative lack of pain with ICI-associated ON is a distinct feature, but without a clear explanation.

In conclusion, the potential consequences of ICI-associated inflammation on the ocular structures are extensive and can be profound, including ON. The presentation of ICI-associated ON appears to have a unique presentation compared to classical ON. Specifically, it tends to be bilateral with painless visual decline, and more often with intact color vision. It is advisable for patients on ICI to report any visual changes and be referred for an ophthalmic assessment. It is important to recognize these adverse effects typically occur within four cycles of ICI therapy, but have the potential of late occurrences, primarily in the context of prolonged monotherapy. Ancillary testing, including visual fields and MRI, can be useful diagnostic tools particularly in instances of subtle clinical findings. Visual field patterns can be variable and not restricted to central defects. In this cohort, most eyes had MRI demonstrating sparing of abnormalities in the retrobulbar and proximal nerve. Fortunately, as demonstrated in this case series, patients with ICI-induced ON who receive systemic steroids have the potential to retain or improve their vision and field of view.

Supplementary Material

1

SFigure 1: Representative case 5. At presentation diffuse optic nerve edema was noted in both eyes (right eye shown) as demonstrated on fundus photography with nerve fiber layer sheen along superior and inferior arcades, and peripheral retinal pigment epithelium (RPE) changes (A), and concomitant peripapillary choroiditis and subretinal fluid as demonstrated by hyper autofluorescence (B). Ultrasound demonstrated optic nerve enlargement and thickened choroid (C). Following cessation of ipilimumab and nivolumab and 1.9 months following systemic steroids, the optic nerve edema improved (D), along with resolution of the choroiditis (E). Ultrasound demonstrated corresponding optic nerve and choroidal improvement (F).

2

SFigure 2: Representative case 6. Magnetic resonance imaging at presentation demonstrating mild enhancement surrounding the intraorbital optic nerve sheaths and mild bulging of optic discs at optic nerve insertion on T1-weighted post-gadolinium axial image (A), mild enhancement again noted on T1-weighted coronal image (B). Sectoral superior optic disc edema of the right eye was noted on presentation as demonstrated on fundus photography (C) with corresponding inferior visual field defect (D). Enhanced depth Optical Coherence Tomography (ED-OCT) (Spectralis; Heidelberg Engineering, Heidelberg, Germany) confirms corresponding superior relative nerve fiber layer thickening seen on thickness map and quadrant thickness (E). Sectoral optic disc edema resolved following cessation of ipilimumab and nivolumab and 3.6 months following systemic steroids (F). Inferior visual field defect showed corresponding marked improvement (G), and ED-OCT demonstrated sectoral optic atrophy (H).