Diagnosis and Management of Rare Immune‐Related Adverse Events

Abstract

Oncologic treatment is being revolutionized by a burgeoning number of immune checkpoint inhibitors (ICPis). To date, seven ICPis have received Food and Drug Administration approval, targeting cytotoxic T‐lymphocyte antigen, programmed cell death, or programmed cell death ligand. Adverse events associated with checkpoint inhibition have been described in the literature. Guidelines exist for the most common of these, but as the use of ICPis becomes more common, the number of patients presenting with rare events will increase. This article reviews the diagnosis and management of rare ocular, hematological, luminal gastrointestinal, and rheumatological toxicities arising from ICPi treatment.

Key Points

• As the use of immune checkpoint inhibitors (ICPis) becomes more common, the number of rare immune‐related adverse events (irAEs) will increase. A high level of suspicion is required to identify and treat these toxicities.

• Although it can be difficult to definitively attribute rare irAEs to ICPis, a temporal and mechanistic relationship and the absence of other etiologies should make the treating physician suspicious for a rare irAE.

• Certain rare irAEs, such as celiac disease, do not require treatment with glucocorticoids. Thus, differentiating this irAE from other gastrointestinal irAEs has important implications for treatment.

Short abstract

Guidelines exist for the management of common adverse events related to immune checkpoint inhibitor therapy; however, management recommendations for less common events are lacking. This article focuses on these less well‐described toxicities, presenting case vignettes of specific toxicities and overall conclusions.

Introduction

Oncologic treatment is being revolutionized by a burgeoning number of immune checkpoint inhibitors (ICPis). To date, seven ICPis have received Food and Drug Administration approval, targeting cytotoxic T‐lymphocyte antigen (CTLA‐4), programmed cell death (PD‐1), or programmed cell death ligand (PD‐L1) 1, 2, 3, 4, 5, 6, 7. These drugs harness the adaptive immune system to recognize tumor antigens. Given this unique mechanism, autoimmune‐like toxicity can occur, affecting multiple organ systems, especially the skin, the gastrointestinal (GI) tract and liver, and the endocrine system 8, 9, 10. These common toxicities, defined as >1% (>1 in 100), have been well characterized in clinical trial adverse event summary tables. By contrast, uncommon toxicities occur at a rate of 1% to 0.1% (1 in 100 to 1 in 1,000), rare toxicities arise at a rate of 0.1% to 0.01% (1 in 1,000 to 1 in 10,000), and very rare toxicities occur at a rate of less than 0.01% (<1:10,000) 11. The number of patients who would need to be treated to see rare or very rare toxicities is too large for existing ICPi trials. However, as the use of ICPis becomes more common, the number of patients presenting with rare events will increase. National guidelines exist for the management of common ICPi toxicity, but management recommendations regarding less common events are lacking 8, 9, 10. In this article, we describe the diagnosis and management of rare ocular, hematological, luminal GI, and rheumatological toxicities (Table 1). A comprehensive review of toxicities in other specialties, including cardiology, dermatology, endocrine, gastroenterology, hematology, neurology, and renal, has been previously covered elsewhere in this series (https://theoncologist.alphamedpress.org/site/collections/iraes/index.xhtml). We have therefore focused on these less well‐described toxicities. A high index of suspicion is required for such events to facilitate timely diagnosis and treatment. Although comprehensive questioning of each organ system may not be feasible at every routine follow‐up visit, we hope to raise awareness of some of the less common toxicities so that providers can be alerted to symptoms that may warrant further workup and evaluation.

Table 1.

Common and infrequent ocular, hematologic, gastrointestinal, and rheumatic immune‐related adverse events

Toxicity	Common	Infrequent
Ocular	None	Conjunctivitis Keratoconjunctivitis sicca (dry eyes) Uveitis Vitritis
Hematologic	None	Autoimmune hemolytic anemia Immune thrombocytopenia Autoimmune neutropenia Aplastic anemia Pure red cell aplasia Hemophagocytic lymphohistiocytosis
Gastrointestinal	(Entero)colitis Hepatitis Elevated pancreatic enzymes	Gastritis Celiac disease Enteric neuropathy Pancreatic insufficiency Pancreatitis
Rheumatic	Inflammatory arthritis Polymyalgia rheumatica	Myositis Giant cell arteritis Lupus nephritis Sarcoidosis Scleroderma Other vasculitis

Ocular Toxicities

Incidence

Ocular toxicity is an unusual, but increasingly recognized, side effect of ICPi therapy. The more frequent ocular toxicities associated with ICPis are uveitis and dry eyes 12. Serious ocular and orbital toxicities of ICPis are estimated to occur in less than 1% of patients 13, the majority of whom have been described in case reports.

Case Vignette

Mr. A is a 63‐year‐old white man with metastatic melanoma who developed floaters in his right eye after cycle 3 of nivolumab. His best corrected visual acuity measured 20/25. There was no anterior uveitis. A dilated fundoscopic examination revealed mild vitritis, and cycle 4 of nivolumab was delayed. Mr. A was treated with topical steroids (prednisolone acetate 1%) four times daily in his right eye for 3 weeks, followed by gradual taper. The symptomatic floaters and vitritis resolved and have not recurred after 15 cycles of nivolumab (Fig. 1).

Figure 1.

Optical coherence tomography in Mr. A. Vitreous cells were visible on optical coherence tomography (A) and resolved after topical steroids (B).

Other Rare ICPi‐Related Ocular Conditions

Less frequently encountered ocular side effects include inflammatory orbitopathy (myositis, dacryoadenitis), meibomian gland dysfunction, scleritis, peripheral ulcerative keratitis, vitritis, uveal effusion (Fig. 2), retinal vascular occlusion, choroidal neovascularization, immune retinopathy, myasthenia gravis, cranial nerve palsy, and optic neuritis (Table 2) 14, 15, 16, 17, 18, 19, 20.

Figure 2.

Examples of a normal fundus photograph and a rare ocular irAE. (A): Normal‐color fundus photograph. (B): Uveal effusion in a 59‐year‐old white man with metastatic head and neck squamous cell carcinoma treated with pembrolizumab.

Table 2.

Ocular toxicities associated with immune checkpoint inhibitor treatment

Condition	Clinical signs and symptoms
Inflammatory orbitopathy	Proptosis, limitation of eye movement, chemosis Pain
Meibomian gland dysfunction	Eyelid or conjunctival redness, blocked Meibomian gland orifices Pain, foreign body sensation, or irritation
Scleritis	Redness, violaceous hue Deep or aching pain
Peripheral ulcerative keratitis	Crescent‐shaped ulceration at margin of corneal limbus Pain, redness, tearing, decreased vision
Vitritis	Vitreous cells and haze Floaters
Uveal effusion	Serous choroidal detachment Decreased vision
Retinal vascular occlusion	Intraretinal hemorrhages, vascular tortuosity, and congestion Painless, acute decreased vision
Choroidal neovascularization	Retinal neovascular membrane with hemorrhage, lipid, or fluid Decreased vision, metamorphopsia (distortion of central vision)
Immune retinopathy	Mild retinal pigment epithelium changes, may be subtle Photopsias (flickering lights), decreased vision, visual field loss
Myasthenia gravis	Ptosis, limitation of eye movement Diplopia
Cranial nerve palsy	Limitation of eye movement, facial paralysis Diplopia, decreased vision
Optic neuritis	Optic nerve edema, may be normal in appearance Acute decreased vision, pain with eye movement

Grading and Diagnostic Workup

As immunotherapies are being combined with other agents that are known to cause ocular toxicity, oncologists should consider taking a brief an ocular history at baseline and with each cycle, including presence of vision changes (blurred vision, floaters, visual field defects), pain, and redness or irritation. New ocular symptoms should prompt referral to an ophthalmologist. Examination includes assessment of visual acuity, intraocular pressure, ocular motility, visual fields, inspection of the ocular adnexa (eyelids and lashes, orbit), anterior segment (conjunctiva, sclera, anterior chamber, iris, lens), and fundoscopic examination (vitreous, optic nerve, vasculature, retina, choroid). Ancillary imaging can facilitate diagnostic workup and documentation of findings. Studies such as slit lamp and color fundus photography, optical coherence tomography, angiography, ultrasonography, visual fields, neurophysiologic testing (electroretinography, visual evoked potential) and neuroimaging (computed tomography, magnetic resonance imaging [MRI]) can be crucial, depending on the extent of involvement detected.

Based on the Common Terminology Criteria for Adverse Events (CTCAE; version 5.0, 2017) 21, grade 1 toxicity is mild (may be asymptomatic but clinically apparent based upon examination findings) and frequently does not require intervention. Grade 2 events are moderate and symptomatic, limit instrumental activities of daily living (ADLs), and are associated with visual acuity of 20/40 or better (or 3 lines or fewer decreased from baseline). Grade 3 toxicity leads to a marked decrease in vision (worse than 20/40, or more than 3 lines decreased from baseline, but better than 20/200), limitation of self‐care ADLs, severe pain, and visual field defects and may require intravitreal injection of steroids or surgery. Grade 4 events are characterized by blindness (i.e., vision of 20/200 or worse).

Management

Symptoms, such as dry eyes, can be treated with topical over the counter artificial tear (saline) lubricants. Mild inflammatory conditions can be managed with topical steroids (prednisolone acetate 1%, one drop, four to six times daily, depending on severity). Moderate to severe inflammatory toxicity may require periocular (sub‐Tenon's) or intravitreal injection of steroids, or systemic glucocorticoids (1 mg/kg/day). The decision to discontinue ICPi therapy is complex and should be made after a thorough discussion among the oncologist, the ophthalmologist, and the patient. Mild to moderate toxicity can be managed medically, whereas severe side effects may require surgical intervention and/or cessation of therapy 22.

Conclusion

It is important to establish an ophthalmic baseline before initiation of ICPi therapy and to inquire about symptoms at follow‐up visits 23. Although severe ocular toxicities are rare, early referral to an ophthalmologist can be vision‐saving.

Immune‐Mediated Thrombocytopenia

Incidence

ICPi‐associated immune thrombocytopenia (ICPi‐ITP) is an isolated acquired thrombocytopenia, temporally related to treatment with ICPis and thought to be due to auto‐antibody formation, immune‐mediated platelet destruction, and inappropriately low platelet production. The exact incidence of ICPi‐ITP is unknown. In one retrospective series of 2,360 patients with melanoma receiving ICPis, only 11 patients developed thrombocytopenia (0.5%), with the platelet drop occurring on average 70 days after ICPi initiation (range, 12–173 days) 24. The differential diagnosis for ICPi‐ITP is broad; it includes myelophthisic processes, consumptive coagulopathy, microangiopathic disease, platelet sequestration, drug‐induced thrombocytopenia, infection, and pseudo‐thrombocytopenia.

Case Vignette

Mr. B is a 77‐year‐old man with metastatic melanoma treated with pembrolizumab. A complete blood cell (CBC) count prior to cycle 1 revealed a white blood cell (WBC) count of 11 × 10³/mcL, a hemoglobin of 12.3 g/dL, and a platelet count of 120 × 10³/mcL. Routine bloodwork performed 18 days after starting pembrolizumab revealed a WBC count of 10 × 10³/mcL, hemoglobin of 11.4 g/dL, and a platelet count of 31 × 10³/mcL. The patient denied mucosal bleeding, increased bruising, or petechiae. A direct antiglobulin test was negative, and hemolysis labs (lactate dehydrogenase, haptoglobin, reticulocyte count, and indirect bilirubin) were not consistent with red blood cell (RBC) destruction. Review of his peripheral blood film revealed hypochromic red blood cells, mature myeloid forms, giant platelets without platelet clumping, and the absence of schistocytes, dacrocytes, or nucleated RBCs. Prednisone 60 mg orally once per day was started for a presumed diagnosis of ICPi‐ITP.

Grading and Diagnostic Workup

CTCAE defines grades 1, 2, 3, and 4 severity of ICP‐ITP as having a platelet count of 75 × 10³/mcL to less than the lower limit of normal, 50–75 × 10³/mcL, 25–50 × 10³/mcL, and < 25 × 10³/mcL, respectively 21. However, it is critical to take the pre‐ICPi platelet count into consideration, as many patients treated with ICPis have myriad reasons to have low platelets at baseline. The diagnostic workup includes a CBC and evaluation of the peripheral blood film in order to rule out the presence of schistocytes, which would suggest a microangiopathic process. If CBC derangements aside from thrombocytopenia exist, other causes of thrombocytopenia must be considered. Viral studies (hepatitis B, hepatitis C, and human immunodeficiency virus) are recommended if there is concern for exposure. Bone marrow biopsy is not necessary if the history, physical examination, and blood film are supportive of a diagnosis of ICPi‐ITP.

Management

Patients with grade 1 ICPi‐ITP do not require an intervention and can continue receiving ICPis with close clinical monitoring 24. The American Society of Clinical Oncology‐National Comprehensive Cancer Network guidelines recommend discontinuation of ICPis if the platelet count falls below 75 × 10³/mcL (grade 2 toxicity); we are in agreement with this recommendation, and we also suggest temporary cessation of ICPis if the platelet count falls more than 50% from baseline 9. Treatment for ICPi‐ITP should be initiated when platelets are <50 × 10³/mcL, a threshold higher than for primary immune thrombocytopenia (ITP), as patients with cancer may be at higher risk for bleeding (depending on tumor location, need for invasive procedures, or other patient‐specific factors). Resumption of ICPis should be deferred until the platelet count is higher than 75–100 × 10³/mcL; however, rechallenge with ICPis may be possible 25.

Recommendations for treatment of ICPi‐ITP are derived from case reports and expert opinion. Options mirror those used in idiopathic ITP, and include immunosuppressive medications, thrombopoietin receptor (TPO) agonists, antifibrinolytics, and platelet transfusions 26. First‐line treatment consists of high‐dose glucocorticoids and/or immune globulin, with TPO agonists and the anti‐CD20 monoclonal antibody, rituximab, used as second‐line agents for refractory thrombocytopenia 24. The TPO agonists may be a particularly attractive option for treatment of ICPi‐ITP given their relatively rapid onset of action 27, the ability to titrate weekly dosing throughout ICPi therapy, and the concern that glucocorticoids may abrogate the antitumor effect of ICPis. However, data regarding the use of TPO agonists in ICPi‐ITP are limited 25. Proposed treatment guidelines are outlined in Table 3.

Table 3.

Treatment options for immune checkpoint inhibitor‐associated immune thrombocytopenia

Treatment category	Details
First‐line treatment
Glucocorticoids	Prednisone: 1 mg/kg/day (dose range, 0.5–2 mg/kg per day) orally for 2–4 weeks tapered over 4–6 weeks to the lowest effective dose. High‐dose dexamethasone: 40 mg per day orally as a single daily dose for 4 consecutive days. Treatment may be repeated 2 more times in 2‐week intervals. Methylprednisolone: 1,000 mg/day IV for 3 days for severe bleed.
Immune globulin	1,000 mg/kg/day given once per day for 2 days (total dose 2,000 mg/kg), or 400 mg/kg/day given for 5 days. Used if rapid response is needed and/or contraindication to glucocorticoids.
Second‐line treatment
TPO agonists	Romiplostim:a Start at 1–3 mcg/kg/week. Eltrombopag: Start at 50 mg per day with maximum dose of 75 mg per day. Patients of East Asian ancestry or those with liver impairment should start at 25 mg per day.
Rituximab	375 mg/m2 intravenously once per week for 4 weeks.
Adjuncts: for acute bleeding or platelet count <10 × 103 mcL
Platelet transfusion	Check platelet count 30 minutes after transfusion to determine response.
Antifibrinolytics	Aminocaproic acid: 4 g IV bolus followed by 1 g/hour for a maximum dose of 24 g in 24 hours. Tranexamic acid: 1.3 g orally three times per day for 5 days or longer depending on clinical scenario.

Abbreviations: IV, intravenous; TPO, thrombopoietin receptor.

^aWe prefer romiplostim over eltrombopag given ease of weekly titration.

Conclusion

Although rare, hematologic toxicities of ICPis are becoming well recognized with the widespread use of immunotherapy 28. ICPi‐ITP mimics primary ITP in presentation and response to therapy. Rigorous laboratory correlates are lacking; however, we speculate that the pathophysiology of ICPi‐ITP may be due to immune dysregulation, loss of self‐tolerance, and decreased platelet production 29. Although glucocorticoids are an effective first‐line treatment, TPO agonists are a promising tool that warrant further study.

Rare Luminal GI Toxicity

Incidence

The precise incidence of ICPi‐related celiac disease is presently unknown 30, 31. In a well‐defined cohort of 377 patients with melanoma at the Massachusetts General Hospital, we identified two cases of incident celiac disease (0.5%). We have seen an additional seven cases in our cancer center outside of this defined cohort. We have associated ipilimumab, pembrolizumab, and nivolumab with celiac disease, but the sample size is too small to allow an estimation of relative risks among the specific regulatory pathways.

Case Vignette

Ms. C is a 72‐year‐old woman with stage IIIB pulmonary adenocarcinoma who presented to GI clinic with new‐onset, nonbloody, water diarrhea after five cycles of pembrolizumab. She was having three to five bowel movements a day with occasional nocturnal bowel movements (CTCAE grade 2 diarrhea). Most of her bowel movements were postprandial, coming an hour after eating, but she had not noticed any association with specific foods, and she had no nausea or vomiting.

Serological testing for celiac disease returned positive with IgA 158 mg/dL and tissue transglutaminase (TTG)‐IgA 37 U/mL (range, 0–15 u/mL). Clostridium difficile testing was negative by polymerase chain reaction. Esophagogastroduodenoscopy and colonoscopy were then performed. Her small intestinal and colonic mucosa were grossly normal (Fig. 3), with biopsies showing increased intraepithelial lymphocytes in the duodenum consistent with celiac disease. Colonic biopsies were normal, excluding concurrent PD‐1 blockade‐induced colitis. Ms. C was started on a gluten‐free diet with resolution of her symptoms, without the use of glucocorticoids, and she resumed pembrolizumab therapy without incident.

Figure 3.

Normal endoscopic findings in Ms. C.

Other Rare ICPi‐Related GI Conditions 32

Hemorrhagic gastritis has been reported with PD‐1 blockade in the absence of other luminal inflammation 33. In addition, disorders of GI motility have been reported as rare manifestations of neurologic toxicity from checkpoint blockade 34.

Grading and Diagnostic Workup

Celiac disease does not fit well into the standard CTCAE severity grading criteria 21. Diarrhea grade gives some information about the impact on the patient's quality of life but has little value for either treatment recommendations or long‐term health risks. We advocate considering all immunotherapy‐induced celiac disease as a grade 2 toxicity with the mainstay treatment being a strict gluten‐free diet without the need for glucocorticoids.

The workup for immunotherapy‐associated diarrhea should include serologic testing for celiac disease. TTG‐IgA paired with a total serum IgA shows the highest sensitivity and specificity for the disease 35. Confirmatory endoscopic biopsies are recommended, and we routinely pair these with colonic biopsies, as some of our cases of incident celiac disease have also had colonic inflammation.

It is unknown whether symptomatic celiac disease that appears during ICPi treatment represents activation of latent, undiagnosed celiac disease or if the disease is truly de novo. We recommend considering celiac disease as associated with immunotherapy when the patient becomes symptomatic after initiation of immunotherapy and he or she has no known prior history of the disease.

Management

ICPi‐associated celiac disease is a non‐life‐threatening toxicity that can be managed through a strict gluten‐free diet. Most patients will not require glucocorticoids or any other form of immune suppression, although a minority of these individuals will have other types of luminal inflammation, such as enterocolitis. Importantly, in patients with celiac disease and enterocolitis, diarrhea will not resolve without instituting gluten restriction. In addition, if ICPi‐associated celiac disease is misdiagnosed, it can lead to unnecessary treatment with steroids or premature discontinuation of ICPi. For isolated celiac disease associated with checkpoint blockade, we do not recommend steroids or recommend any changes to the immunotherapy regimen.

Conclusion

At our institution, we assess TTG‐IgA and IgA on all ICPi‐treated patients with diarrhea and confirm serological diagnosis with endoscopic biopsies when patients are otherwise clinically stable. We hypothesize that these patients likely have a latent form of celiac disease that was actively under immune suppression by PD‐1/PD‐L1 or CTLA‐4 prior to the onset of therapy and that treatment allowed this controlled immune response to become clinically significant.

Rheumatic Toxicity

Incidence

The incidence of ICPi‐related vasculitis, specifically giant cell arteritis (GCA), is difficult to calculate given its rarity. GCA has been reported after the administration of pembrolizumab and ipilimumab 36, 37. Polymyalgia rheumatica (PMR) is a more common immune‐related adverse event (irAE), although the true incidence is also challenging to ascertain, as PMR is often grouped with other types of inflammatory arthritis. A recent prospective study estimated the incidence of PMR among patients receiving ICPis at a single institution to be approximately 2% 38.

Case Vignette

Ms. D is a 76‐year‐old woman with metastatic rectal adenocarcinoma, previously treated with neoadjuvant 5‐fluorouracil and radiation, resection, and eight cycles of adjuvant FOLFOX, followed by FOLFIRI (Leucovorin Calcium [folinic acid], Fluorouracil, Irinotecan Hydrochloride) and bevacizumab with progressive disease, who was started on ipilimumab and nivolumab as part of a clinical trial. Five days after her second cycle of combination ICPi, she developed right‐sided temporal headaches, blurry vision, shoulder and hip pain, and stiffness. She had a history of PMR and GCA that had been diagnosed 20 years prior and had been treated with prednisone at that time with a single recurrence 3 years after the initial diagnosis. At the time of her current presentation she was started on prednisone 60 mg/day given concern for GCA, with rapid improvement of her symptoms. Erythrocyte sedimentation rate (ESR) was not checked at the time of initial diagnosis, but a week after starting prednisone her ESR was 35 mm/hour, and she underwent a right sided temporal artery (TA) biopsy, which was negative for arteritis. She was evaluated by rheumatology, neurology, and ophthalmology and underwent a brain and face MRI to evaluate for metastases, cavernous sinus inflammation, or orbital inflammation. The MRI did not reveal a mass lesion, infarction, or abnormalities of the optic pathway or visualized cranial nerves, but there was evidence of enhancement of the frontal and parietal branches of the left superficial temporal artery on the T1 postcontrast fat‐suppressed images (Fig. 4) 39, 40, 41. She declined a repeat bilateral temporal artery biopsy, temporal artery ultrasound, or large‐vessel imaging.

Figure 4.

Enhancement of the frontal and parietal branches of the left superficial temporal artery (arrows) on the T1 postcontrast fat‐suppressed magnetic resonance images of Ms. D.

This case highlights the challenges in making a diagnosis of GCA in patients receiving ICPis. Although clinical uncertainty regarding the exact diagnosis remains in this case, her symptoms were most likely related to the ICPis and could represent GCA and PMR, although this could not be confirmed given negative biopsy. Her prednisone was tapered over the course of 3 months. For usual GCA unrelated to ICPis, treatment with prednisone is usually a prolonged taper over the course of a year. An expedited taper was chosen in this case because of the diagnostic uncertainty and risk of adverse effects with prolonged high‐dose prednisone. Immunotherapy was not restarted.

Other Rare ICPi‐Related Rheumatic Conditions

In a recent systematic review of all rheumatic adverse events caused by ICPis, vasculitis was reported in two clinical trials, and reports were otherwise limited to case reports 42. Another review identified 53 cases of vasculitis associated with the use of an ICPi, of which 20 were confirmed cases 43. Small‐, medium‐, and large‐vessel vasculitides affecting numerous organs including the skin, large vessels, and peripheral nerves have been reported 44, 45, 46. In addition to PMR, other forms of inflammatory arthritis have been described including small joint polyarthritis mimicking rheumatoid arthritis and large joint oligo‐articular arthritis mimicking the seronegative spondyloarthropathies 38, 42. These irAEs are far more common than vasculitis, occurring at rates of 1%–43% 2, 42, 47, 48, 49. The wide range of reported incidence is likely due in part to the heterogeneity of how musculoskeletal irAEs are reported in clinical trials; a recent prospective study estimated the incidence of musculoskeletal irAEs that were referred to rheumatology at a single institution to be 6.6%, which likely provides a more accurate estimate of the incidence of this irAE 38. Other rare rheumatologic irAEs that have been reported include sicca symptoms mimicking Sjogren's syndrome, myositis, lupus nephritis, sarcoid, and scleroderma 42, 50, 51, 52, 53, 54, 55.

Grading and Diagnostic Workup

Patients receiving ICPis who develop new‐onset severe headaches, particularly with features concerning for GCA such as jaw claudication, visual disturbances, or symptoms of PMR, should be evaluated urgently, ideally by a multidisciplinary team including oncology, rheumatology, neurology, and ophthalmology. In a review of patients referred for a TA biopsy with concern for standard GCA (as opposed to ICPi‐related GCA), the presence of jaw claudication and diplopia increased the likelihood ratio of the patient having GCA, and thus these features should be given particular attention 56. The differential diagnosis of headaches in a patient receiving immunotherapy is broad, and consideration should be given to metastatic disease, infection, neurological irAEs, or other rare irAEs including cavernous sinus inflammation or orbital inflammation. ESR and C‐reactive protein should be checked and can help in the diagnostic workup but are nonspecific and must be taken in the context of the entire clinical picture. A positive temporal artery biopsy remains the gold standard for diagnosis of GCA and should thus be performed promptly in the appropriate clinical setting. Although diagnostic yield of a TA biopsy does diminish with time, positive biopsy results can still be seen up to 4 weeks after the initiation of glucocorticoids, and generally a biopsy within 2 weeks of starting glucocorticoids is acceptable 57, 58, 59. If clinical suspicion is high, treatment should not be delayed while waiting for the biopsy to be obtained. The presence of confusion, focal neurologic deficits, prominent pulmonary symptoms, metabolic abnormalities including rapidly rising serum creatinine, digital ulcerations, or other new rashes should raise concern about an alternative diagnosis such as infection, other types of vasculitis, or alternative irAEs. The current CTCAE grading for vasculitis is inclusive of all forms of vasculitis and does not offer specific guidelines based on the type of vasculitis 21. Biopsy‐proven GCA should be considered grade 3 or higher based on the current CTCAE guidelines. Although GCA is rarely life threatening, it can be vision threatening, and thus prompt recognition and treatment are required.

Management

For patients in whom clinical suspicion for GCA is high, treatment with high‐dose glucocorticoids (60 mg prednisone if there are no visual symptoms; consider 1,000 mg intravenous methylprednisolone if there are vision changes that are consistent with GCA) should not be delayed while awaiting diagnostic studies. In patients with classic GCA unrelated to ICPis, the use of tocilizumab improves glucocorticoid‐free remission 60. However, there is not enough evidence or experience to know whether tocilizumab may also play a role in the treatment of patients with ICPi‐GCA. Given the rarity of this complication, there are no formal recommendations regarding what to do with the ICPi; at a minimum the ICPi should be delayed.

Conclusion

ICPi‐associated GCA has been reported, but it remains a rare irAE. As highlighted here, the diagnosis can be challenging, and consideration must be given to other causes of headache or vision changes in patients receiving ICPis. In the right clinical setting, when symptoms are suggestive of GCA, a temporal artery biopsy should be obtained. Multidisciplinary evaluation by rheumatology, neurology, and ophthalmology is helpful. Mechanistically, low or absent expression of PD‐L1 by dendritic cells residing in the vessel wall of vasculitic vessels likely allows PD‐1‐expressing T cells to go unchecked, leading to the release of effector cytokines and the destruction and inflammation of vessel walls 61, 62. Given the important role of the immune checkpoint inhibitors in the pathophysiology of large‐vessel vasculitis, it is likely that this irAE will become increasingly common as the ICPis are used more frequently.

Overall Conclusion

Our case examples of rare or uncommon irAEs highlight the need for a high index of suspicion when treating patients with ICPis. It is critical to review a symptom history as well as updated laboratory testing at each visit before administration of ICPis. The low incidence of these events makes it difficult to distinguish these episodes from random events. However, our cases have a mechanistic relationship and temporal association with the ICPis, making it more likely that they are associated with ICPi therapy. Although there may be differences in the incidence of these rare events among the available ICPis, particularly between PD‐1/PD‐L1 pathway inhibitors and anti‐CTLA‐4, all of the events described in this article are presently too rare for us to speculate on differences in incidence. Additional research is needed in this field to define diagnostic biomarkers for immune‐related toxicities associated with checkpoint inhibitors and to develop comprehensive treatment guidelines.

Author Contributions

Conception/design: Sara R. Schoenfeld, Mary E. Aronow, Rebecca Karp Leaf, Michael Dougan, Kerry L. Reynolds

Manuscript writing: Sara R. Schoenfeld, Mary E. Aronow, Rebecca Karp Leaf, Michael Dougan, Kerry L. Reynolds

Final approval of manuscript: Sara R. Schoenfeld, Mary E. Aronow, Rebecca Karp Leaf, Michael Dougan, Kerry L. Reynolds

Disclosures

Michael Dougan: Novartis (RF), Genentech (CA). The other authors indicated no financial relationships.

(C/A) Consulting/advisory relationship; (RF) Research funding; (E) Employment; (ET) Expert testimony; (H) Honoraria received; (OI) Ownership interests; (IP) Intellectual property rights/inventor/patent holder; (SAB) Scientific advisory board