Drug-induced Inflammation in Patients on TNFα Inhibitors

Keywords::

• drug-induced uveitis
• sarcoidosis
• granulomatous
• inflixamab
• etanercept
• adalimumab

Local and systemic corticosteroids constitute first-line therapy for most patients with noninfectious uveitis,^1,2 whereas noncorticosteroid immunosuppressive agents are typically used in patients requiring more sustained immunosuppression.^3–9 Traditional corticosteroid-sparing agents include antimetabolites, such as methotrexate, azathioprine, and mycophenolate mofetil; leukocyte-signaling inhibitors, such as cyclosporine and tacrolimus; and, infrequently, the alkylating agents cyclophosphamide and chlorambucil, which are generally reserved for the most severe vision- or life-threatening conditions.^9,10 The choice of any particular immunosuppressive agent can be complex and must be tailored to the needs of the individual patient.^7,9

The introduction of tumor necrosis factor alpha (TNFα) inhibitors to treat systemic autoimmune disease in the late 1990s revolutionized the management of systemic inflammatory diseases, including moderate to severe rheumatoid arthritis, inflammatory bowel disease, juvenile idiopathic arthritis, inflammatory bowel disease, and the various spondyloarthropathies. These agents have also gained increasing acceptance as off-label therapies for patients with refractory uveitis.^11–15 Five TNFα antagonists are currently available for clinical use: infliximab (Remicade), a chimeric murine-human full length monoclonal antibody given by intravenous infusion every 4–8 weeks; etanercept (Enbrel), an antibody Fc-TNF receptor fusion protein injected subcutaneously twice weekly; adalimumab (Humira) and golimumab (Simponi), fully humanized full-length monoclonal antibodies given as subcutaneous injections every 2 and 4 weeks, respectively; and certolizumab (Cimzia), a fully humanized Fab antibody fragment linked to polyethylene glycol to extend drug half-life, which is administered as a subcutaneous injection monthly. Infliximab, etanercept, and adalimumab have each been used to treat a wide range of uveitic disorders,^11–15 with particularly promising results noted in patients with moderate to severe Behçet disease^16,17 and juvenile idiopathic arthritis.¹⁸ Experience with golimumab¹⁹ and certolizumab²⁰ to treat ocular inflammatory disease is limited, but encouraging.

Despite the documented therapeutic benefits of TNFα antagonists, their use is not without risk.^21–23 Safety data collected from both randomized clinical trials and postmarketing surveillance studies have highlighted the occurrence of infusion or injection site reactions, which occur in approximately 20% of patients, but are typically manageable. Less common, although potentially more severe, complications include reactivation of tuberculosis; severe fungal infections, cytopenias, or, rarely, aplastic anemia; increased risk of malignancy; worsening of preexisting heart failure; and exacerbation or development of demyelinating disease.²³ Unexpected inflammatory effects of TNFα inhibitors have also been noted and include exacerbation or initiation of drug-induced autoimmune diseases, such as psoriasis, systemic lupus erythematosus, vasculitis, and uveitis—particularly with etanercept.²⁴

Two articles in this issue of Ocular Immunology & Inflammation highlight the paradoxical production of uveitis in patients receiving anti-TNFα therapy. The first, by Fonollosa and colleagues,²⁵ describes a 30-year-old woman with psoriatic arthritis well controlled with etanercept for 4 years. The patient then developed acute fever, fatigue, headache, decreased vision, and floaters. Ocular examination revealed 20/20 vision in each eye with mild bilateral granulomatous anterior uveitis and intermediate uveitis accompanied by snowball vitreous opacities and segmental periphlebitis. Investigations ultimately revealed an elevated serum angiotensin-converting enzyme (ACE) level together with hilar and mediastinal adenopathies. Mediastinal lymph node biopsy disclosed noncaseating granulomas consistent with sarcoidosis. The patient’s uveitis resolved following discontinuation of etanercept and her psoriasis was ultimately controlled with systemic prednisone, methotrexate, and ustekinumab, an antibody-based interleukin 12/23 inhibitor. Other possible causes of uveitis were excluded with ancillary testing. A review of the literature performed by the authors identified 6 additional cases of sarcoid uveitis linked to anti-TNFα therapy, 5 of whom were on etanercept and 1 of whom was on infliximab. The delay in onset of uveitis ranged from 3 weeks to nearly 6 years. The types of uveitis varied, and included anterior, posterior, and panuveitis. Two patients had periphlebitis and 1 had a focal chorioretinitis. The uveitis resolved in 5 of the affected patients once the TNFα inhibitor was replaced by a non-TNFα targeting immunosuppressive agent, and in 1 patient with the addition of systemic corticosteroids and the ultimate replacement of etanercept by adalimumab.

The second report, by Seve and associates,²⁶ describes a 61-year-old woman with rheumatoid arthritis, on adalimumab for 3 years, who then developed bilateral panuveitis with granulomatous keratic precipitates, periphlebitis, and peripheral multifocal choroiditis. Laboratory investigations revealed an elevated serum ACE level, but were otherwise unremarkable. Biopsy of a skin nodule on the patient’s face that appeared at the same time as the onset of uveitis revealed noncaseating granuloma formation. The skin involvement resolved completely and the uveitis improved following withdrawal of the adalimumab and administration of systemic corticosteroids.

Daïen and colleagues have estimated the prevalence of paradoxical sarcoid-associated inflammation in patients receiving TNFα inhibitors to be approximately 1 in 2800.²⁷ A review by Toussirot and Pertuiset in 2010 identified 28 such cases, including 16 associated with etanercept, 8 with infliximab, and 4 with adalimumab.²⁸ Time from initiation of anti-TNFα therapy to diagnosis of sarcoid-associated inflammation varied from 1 month to 5 years. Pulmonary involvement was most common (86%), followed by cutaneous (29%) and ocular (14%). Seve and associates²⁶ and others^29,30 have aptly pointed out that the occurrence of such responses in the setting of multiple TNFα antagonists supports the notion of drug class-related idiosyncratic dysregulation of immune function in selected patients.

The mechanism(s) whereby TNFα inhibitors produce paradoxical inflammatory effects are unclear. One hypothesis that has garnered considerable attention involves the inverse and interdependent relationship between TNFα and interferon levels (both α and γ), and the role of increased interferon in the development of autoimmune diseases,³¹ including both psoriasis³² and sarcoidosis.^33,34 Similar changes in the cytokine balance in response to TNFα inhibition have been suggested to elicit auto-antibody and immune complex formation, both of which may have broadly inflammatory effects.³⁵ Others have suggested that TNFα inhibition may directly or indirectly increase the rate of infection by organisms implicated in noncaseating granuloma formation, such as Propionibacterium acnes and nontuberculous Mycobacterium species.^27,36 It has been hypothesized further that differential and/or less complete inhibition of circulating and tissue-based TNFα by etanercept might promote a regional cytokine imbalance, thus explaining the apparent increased relative risk of granuloma formation in patients taking etanercept as compared to antibody-based TNFα antagonists. This higher risk of paradoxical granuloma formation in patients on etanercept is consistent with the observation that etanercept is generally ineffective at treating severe systemic granulomatous disorders, such as Wegener granulomatosis and progressive pulmonary sarcoidosis.27 Genetic and environmental factors may also contribute to the propensity for paradoxical granuloma formation,37 since both interferon polymorphisms38 and variations in human leukocyte antigen (HLA) expression39 have been independently associated with the development of interferon-induced sarcoidosis.

Taken together, the evidence is now clear that TNFα inhibitors should be added to the growing list of agents known to cause or promote intraocular inflammation (Table 1).^40–47 While the clinical characteristics of the uveitis induced by TNFα antagonists vary from patient to patient, suggestive clinical findings appear to include the presence of granulomatous features and the development of periphlebitis—features often observed in ocular sarcoidosis. As with other agents known to promote intraocular inflammation, TNFα-associated uveitis typically resolves with full recovery of vision following withdrawal of the inciting agent.

TABLE 1  Drugs associated with uveitis.^40–47

Systemic

TNFα antagonists^a

Rifabutin^a

Biphosphonates^a

Sulfonamides^a

Oral contraceptives^a

Diethylcarbamazine^a

Fluoroquinolones^a

Topiramate^b

Trifluoperazine^b

Quinidine^b

Ibuprofen^b

Reserpine^b

Topical

Prostaglandin analogues^a

Metipranolol^a

Corticosteroids^a

Cholinomimetics^a

Antibiotics^a

Betaxolol^b

Intraocular

Antibiotics^a

Cidofovir^a

Urokinase^a

Plasmin/microplasmin^a

Antibodies (ranibizumab, bevacizumab, etc.)^a

Triamcinolone acetonide^a

Vaccines

Bacille Calmette-Guérin (BCG)^a,c

Influenza^a

Hepatitis B^a

Measles, mumps, and rubella (MMR)^b

Diphtheria, tetanus, and pertussis (DPT)^b

Varicella^b

Smallpox^b

Note. A comprehensive list of drugs reported to be associated with uveitis may be found at www.eyedrugregistry.com.⁴²

^aStrong evidence.

^bAnecdotal evidence.

^cUveitis also associated with skin placement of purified protein derivative (PPD).

ACKNOWLEDGEMENT

Supported in part by the Pacific Vision Foundation (ETC) and the San Francisco Retina Foundation (ETC). Dr. Suhler receives institutional support from Research to Prevent Blindness and the Department of Veterans’ Affairs. We thank Ramana S. Moorthy, MD, FACS for thoughtfully reading and commenting on an early version of this editorial.

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.