Neutralizing Tumor Necrosis Factor Activity Leads to Remission in PatientsWith Refractory Noninfectious Posterior Uveitis FREE

Noninfectious posterior segment intraocular inflammation (PSII) encompassesa group of immune-mediated conditions that are variably characterized by relapsing-remittinginflammation of the choroid, retina, ciliary body, and vitreous.¹ Traditionallyclassified as intermediate, posterior, and panuveitis, these conditions predominantlyaffect young and middle-aged people and are a significant cause of visualimpairment.² The mainstay of therapy for PSIIis corticosteroids, with the addition of immunosuppressive agents such ascyclosporine, azathioprine, tacrolimus, and mycophenolate mofetil if required.^{3 - 4} With this approach, ocular inflammationcan be controlled in most patients, but many remain refractory to therapyor develop a toxic reaction to the drug, indicating the need to develop moreeffective and safer therapies.

Posterior segment intraocular inflammation is a presumed antigen-specificCD4⁺ T-lymphocyte–mediated autoimmune disease characterizedby macrophage- and T-lymphocyte–mediated damage to the eye.⁵ The experimental model of PSII, experimental autoimmuneuveoretinitis (EAU), has provided abundant evidence that tumor necrosis factor α(TNF-α) plays a pivotal role in the pathogenesis of inflammation andtissue destruction in PSII,^{6 - 8} andthis is supported by findings of elevated levels of TNF-α in the ocularfluid and serum of patients with uveitis.^{9 - 11} Indeed,a TNF-α blockade with the p55 TNF receptor fusion protein (TNFr-Ig)leads to profound suppression of retinal inflammation and structural damagein patients with EAU, modulating T-lymphocyte and macrophage function withinthe retina.^{12 - 14}

Therefore, following the successes of TNFr-Ig therapy in patients withEAU, the established benefit of anti-TNF therapies in patients with otherautoimmune diseases, such as rheumatoid arthritis and Crohn disease,^{15 - 18} andtheir recent successful use in patients with ocular Behçet disease,¹⁹ we report the use of TNFr-Ig in a phase 1/2 studyin patients with PSII. The objectives of this study were to evaluate the efficacyand tolerability of short-term therapy with TNFr-Ig in patients with refractorynoninfectious PSII and to estimate its impact on health-related quality oflife (HQOL) and vision-related quality of life (VR-QOL).

Patients were recruited to this open-label, nonrandomized, pilot studyof TNFr-Ig for PSII from 2 regional referral centers for uveitis in the UnitedKingdom, the Bristol Eye Hospital, Bristol, and the Aberdeen Royal Infirmary,Aberdeen. The study was approved by the ethics committees of each center,and informed consent was obtained from all patients. The inclusion criteriawere chronic noninfectious sight-threatening PSII and treatment failure withprednisolone (>10-mg/d maintenance dosage) and at least 1 immunosuppressiveagent, most commonly cyclosporine, because of either refractory disease ordrug intolerance. Reasons for exclusion from the study were pregnancy, diabetesmellitus, renal disease, concurrent infection, and recent live vaccinations.

The TNFr-Ig molecule is a chimeric molecule comprising the extracellulardomain of the human p55 TNF receptor fused to the hinge, CH₂ andCH₃, domains of the human IgG1 heavy chain; and was manufacturedby the Therapeutic Antibody Centre, University of Oxford. This construct issimilar to the one described by Peppel et al.²⁰ TheTNFr-Ig molecule inhibits human TNF-α function in vitro and is a potentantagonist of TNF-α in patients with EAU.^{12 - 13} Anintravenous infusion of TNFr-Ig, 50 mg, in isotonic sodium chloride solutionwas administered over 4 hours. A second infusion of TNFr-Ig, 100 mg, was givento patients who responded to the first infusion but developed a sight-threateningrelapse of intraocular inflammation within 3 months. Preexisting immunosuppressionwas maintained or reduced while the response to TNFr-Ig therapy was beingevaluated. Antibody responses to TNFr-Ig were measured in patients' serumsamples before each treatment, at 2 weeks, and at 1, 2, 3, and 5 months aftertreatment by a sandwich enzyme-linked immunosorbent assay, as described elsewhere,^{21 - 22} with some modifications to improvereliability.

A systemic and ophthalmic examination was undertaken immediately beforeTNFr-Ig therapy, at 2 and 4 weeks after therapy, and then at 4- to 6-weekintervals according to clinical activity and response to treatment. The baselineassessment included a medical history, a general physical and ophthalmic examination,a blood pressure reading, a urinalysis, a chest x-ray film, and blood tests,including a complete blood cell count, a creatinine level test, liver functiontests, a glucose level test, a urate level test, a C-reactive protein profile,and a lipid profile. Patients were examined at each follow-up visit for adverseeffects and clinical response to TNFr-Ig therapy. The international UveitisScoring System was used to assess clinical disease activity, and lens opacitieswere graded using the Lens Opacities Classification System III.^{23 - 24}

Three self-administered questionnaires were used to assess HQOL, VR-QOL,and adverse effects experienced before TNFr-Ig therapy and after 1, 3, and6 months. Health-related quality of life was evaluated using the validatedUnited Kingdom standard version of the 36-Item Short-Form Health Survey,²⁵ which consists of 36 items grouped into 8 subscalesto measure health, including physical functioning, social functioning, rolelimitations because of physical problems, role limitations because of emotionalproblems, mental health, energy/vitality, bodily pain, and general healthperception. The 36-Item Short-Form Health Survey subscale scores range from0% to 100%, with higher scores indicating better health. Vision-related qualityof life was measured using the vision core module 1 (VCM1), a 10-item questionnairethat provides a subjective measure of concern regarding vision, with scoresranging from 0.0 (best score) to 5.0 (worst score), with 50 intervals.²⁶ Finally, patients completed an adverse effect questionnairethat contained a comprehensive list of well-recognized adverse effects toimmunosuppressive agents. This questionnaire, which addressed the overalleffect of adverse effects on quality of life (asking "how much have theseproblems interfered with your quality of life?"), was scored from 0 (no adverseeffects) to 5 (extreme adverse effects).

The primary outcome measure was best-corrected logMAR visual acuitymeasured at 4 m with the Early Treatment Diabetic Retinopathy Study chart,scored for individual letters. The chart was illuminated with an illuminationunit (Lighthouse Chart Illumination Unit; Lighthouse International, New York,NY). Secondary outcome measures were binocular indirect ophthalmoscopy (BIO)score, cystoid macular edema, Lens Opacities Classification System III gradingof cataract, adverse effects, VR-QOL, and HQOL. True changes in visual acuityand BIO score following TNFr-Ig therapy were defined as an improvement invisual acuity of at least 2 lines (a decrease in the logMAR score of at least0.2) or a decrease in the BIO score to 0 in either eye within 1 month of treatment.To evaluate the duration of response to TNFr-Ig, the end of the period ofresponse was defined by a decrease in visual acuity of at least 2 lines oran increase in BIO score of at least 1 in either eye.

Statistical analysis was performed using the Wilcoxon signed rank test.A software program (Prism, version 3.02; GraphPad, San Diego, Calif) was usedfor all statistical calculations, and significance was attributed when P<.05.

Seventeen patients were enrolled in the study between June 1, 2001,and May 28, 2002. All patients who met the study enrollment criteria duringthis period agreed to participate. Their mean age was 43 years (range, 25-59years), and 11 patients were women. The mean duration of uveitis at enrollmentwas 6 years (range, 0.5-12 years), and the mean follow-up after TNFr-Ig therapywas 8 months (range, 4-16 months), excluding 1 patient who was unavailablefor follow-up after 1 month. Uveitis was bilateral in 14 patients and unilateralin 3 patients. Systemic diagnoses included tubulointerstitial nephritis anduveitis syndrome in 1 patient and Behçet disease in 2 patients. Clinicallyapparent nonocular manifestations of systemic disease were absent in all patients.Prednisolone plus 1 other immunosuppressive agent failed in 3 patients beforethey received anti-TNF therapy, and at least triple therapy failed in theremainder of patients. Twelve patients received a single infusion of TNFr-Ig,4 received 2 infusions, and 1 received a third infusion. Table 1 describes the clinical characteristics, systemic therapybefore and after TNFr-Ig, and the outcome of treatment for each patient.

Nine patients (53%) achieved a successful response to TNFr-Ig therapyaccording to the primary outcome measure, a 2-line improvement in visual acuityin 1 or both eyes within 1 month of treatment. In 8 (57%) of 14 patients withvitreous haze before treatment, represented by a BIO score of 1 or more, theBIO score improved to 0 in 1 or both eyes within 1 month of TNFr-Ig therapy;and macular edema resolved in 5 (56%) of 9 affected patients. Twelve (71%)of the patients achieved complete cessation of intraocular inflammation followingTNFr-Ig therapy. The change in visual acuity and BIO score for worse and bettereyes between baseline and 1 month following treatment is shown in Table 2. Figure 1 shows the improvement in visual acuity and BIO score forall uveitic eyes for 22 of 23 treatment periods during the 3 months followingtreatment (1 patient was unavailable for follow-up after 1 month). The medianduration of response to TNFr-Ig was 3 months (range, 0.5-9 months), and 2patients were still in remission at final follow-up at 9 months without anyfurther increase in immunosuppression (Figure2). A reduction in concomitant immunosuppression was possible for11 patients (65%) following TNFr-Ig therapy (Table 1). Progression of cataract during the follow-up period occurredin 4 patients by 1 grade using the Lens Opacities Classification System III.

Three patients developed infusion-related adverse effects. Patient 7complained of fatigue, headache, dizziness, and nausea during her first infusion,which got progressively worse during her second and third infusions, althoughnot severe enough to warrant abandoning the treatment. Patient 12 complainedof nausea and fatigue during the infusion, and patient 16 developed a mildhypersensitivity reaction characterized by a urticarial rash that resolvedwithout treatment. Adverse effects reported in the days and weeks after TNFr-Igtherapy included mild headache in 2 patients, lymphocytopenia lasting lessthan 3 months in 2 patients concomitantly receiving tacrolimus or cyclosporine,nausea and vomiting 3 days after the infusion in 1 patient, and a severe itchapproximately 1 month after the infusion in 3 patients.

The median (interquartile range) VCM1 score decreased from 2.2 (1.6-3.0)at baseline to 1.7 (0.6-2.5), 1.7 (0.6-2.2), and 1.6 (0.3-2.4) at 1, 3, and6 months, respectively, following TNFr-Ig therapy, indicating that a significantand sustained improvement in VR-QOL occurred (P =.009, .02, and .03 for baseline vs 1-, 3-, and 6-month scores, respectively). Table 3 displays the 36-Item Short-FormHealth Survey subscale scores, the VCM1 scores, and adverse effect scoresat baseline and longitudinally during follow-up. The mental health and bodilypain subscale scores increased significantly following TNFr-Ig therapy (P = .04 and P = .006, respectively).A significant improvement in the self-assessed adverse effect score occurredfollowing TNFr-Ig therapy, from a median (interquartile range) of 3.0 (2.0-3.0)at baseline to 2.5 (0.8-3.0) after 1 month (P = .04).

Pretreatment and posttreatment samples from patients 11 and 13, bothof whom responded well to TNFr-Ig, gave strong responses to the anti–TNFr-Igassay. Among the other 15 patients, 13 did not have a response significantlyabove the pretreatment level. Patient 8 had transient weak responses on day14 after the first treatment and on days 13 and 26 after the second treatment.Patient 16 had a weak response on day 14, but no response in other samples,including after the second treatment.

The results show that TNF blockade with TNFr-Ig is effective for thetreatment of refractory noninfectious sight-threatening PSII. Within 1 monthof anti-TNF therapy, 53% of the patients achieved an improvement in visualacuity of at least 2 lines, the primary outcome measure. Of the patients withvitreous haze, 57% experienced an improvement in BIO score to 0, and cystoidmacular edema resolved in 56% of the patients when present. Therapy with TNFr-Iginduced complete remission in 71% of the patients. The efficacy of TNFr-Igin patients with PSII was demonstrated not only by the improvement in clinicalmarkers of intraocular inflammation but also by the improvement in HQOL andVR-QOL scores. Furthermore, the clinical benefit of repeat infusions was demonstratedin all 5 patients who received a second infusion.

The response to TNFr-Ig was maintained for a median of 3 months, andall but 3 patients relapsed during the follow-up period, indicating that whileshort-term control of ocular inflammation occurred, long-term remission throughthe induction of immune tolerance was generally not achieved. All but 1 patientcontinued to receive adjuvant immunosuppression following TNFr-Ig therapy,although at a lower dose in most patients. As in rheumatoid arthritis andCrohn disease, where the response to the anti-TNF monoclonal antibody infliximabis finite, repeat infusions of TNFr-Ig, and possibly adjuvant immunosuppression,would be required to maintain remission for the long-term. Nussenblatt andcolleagues²⁷ demonstrated that long-term therapywith an anti–interleukin 2 receptor monoclonal antibody (daclizumab)controlled PSII without the need for other immunosuppressive therapies. However,targeting cytokines in this way is only likely to provide short-term reliefof inflammatory symptoms, rather than long-term disease modulation.²⁸ This contrasts with using the T-lymphocyte–depletingmonoclonal antibody alemtuzumab (Campath-1H),²⁹ whichleads to long-term remission of ocular inflammatory disease after a 5-daycourse, possibly by inducing profound lymphocytopenia that on recovery allowsthe immune response to be "reset," as confirmed in animal models.³⁰

Therapy with TNFr-Ig was well tolerated, and apart from 3 mild infusion-relatedreactions and temporary lymphocytopenia in 2 patients, no serious adverseeffects occurred. A significant reduction in the self-reported adverse effectscore following TNFr-Ig therapy, probably as a result of a reduction in concomitantimmunosuppression, is further evidence that the treatment was largely uncomplicated.The clinical efficacy of TNFr-Ig in this cohort of patients is strengthenedby the improvement in VCM1 scores following treatment, which paralleled theimprovement in the clinical markers of ocular inflammation, visual acuity,and BIO score. Because the VCM1 is a validated instrument for measuring subjectiveconcern regarding vision and has been used to measure VR-QOL in patients withuveitis,^{26 ,31} the results confirmthat TNFr-Ig led to an objective and subjective improvement in ocular inflammation.Substantial improvements in the mental health and bodily pain subscales ofthe 36-Item Short-Form Health Survey also occurred, and these functional improvementslend further support to the efficacy of TNFr-Ig therapy.

The 3 infusion reactions that occurred were mild and short-lived. Twoof these occurred during a repeat infusion following an uneventful first infusion,highlighting the possible immunogenicity of TNFr-Ig and the risk of increasinghypersensitivity reactions on repeat treatment. However, none of these infusionreactions were serious enough to warrant abandoning the treatment. Interestingly,mild injection site reactions and infusion reactions with 2 commercially availableanti-TNF agents, etanercept and infliximab, are also common.^{32 - 33} Experiencewith infliximab in patients with rheumatoid arthritis and Crohn disease suggeststhat serious adverse effects with TNF blockade are rare, but recently, reactivationof tuberculosis has been highlighted as a rare but potentially fatal complicationof treatment.³⁴

None of the patients in this study made high levels of specific antibodiesin response to TNFr-Ig. Two patients had high responses in the antibody assaybefore treatment, indicating that they may have had natural antibodies thatrecognized TNFr-Ig. A weak and transient antibody response above the pretreatmentlevels was seen in 1 patient after each of 2 infusions, and a weak response,of doubtful significance, was seen in 1 other patient.

As a proinflammatory cytokine in experimental models, TNF-α activatesT lymphocytes and macrophages and up-regulates endothelial adhesion moleculesand other proinflammatory cytokines, playing a central part in the inductionand maintenance of inflammation in autoimmune reactions.³⁵ Consequently,following many clinical trials, etanercept, a p75 TNF receptor fusion protein,and infliximab, a chimeric monoclonal antibody to TNF-α, are routinelyused in severe cases of rheumatoid arthritis and Crohn disease.^{15 - 16 ,18 ,36} Severalrecent case series have highlighted the potential benefit of TNF-α blockadein ocular inflammatory disease; Sfikakis et al¹⁹ describedthe rapid resolution of panuveitis in 5 patients with Behçet diseasewho were treated with infliximab, and in another prospective study,³⁷ etanercept led to improvement in 10 of 16 eyes withjuvenile chronic uveitis. In a retrospective series of 16 patients with uveitisor scleritis treated with infliximab or etanercept, improvement in ocularinflammation occurred in both patients receiving infliximab but only in 4of 14 patients treated with etanercept. Infliximab was also highly effectivein patients with severe refractory scleritis, leading to remission in all4 patients treated.³⁸ In a randomized controlledtrial³⁹ of etanercept for uveitis, no significantdifference in efficacy was found for etanercept over placebo in preventingrelapses of ocular inflammation in patients with inactive uveitis whose methotrexatetherapy was tapered. Differences in drug construct (p55 vs p75 TNF receptor),the use of concomitant immunosuppression, and the intravenous rather thansubcutaneous route of delivery in the present study may explain the greaterefficacy of TNFr-Ig in our series. Furthermore, the study of Foster et al³⁹ evaluated the ability of etanercept to maintain,rather than induce, remission in patients with active PSII, as in the presentstudy.

In patients with EAU, it has been shown that during TNFr-Ig therapythere is presumed inhibition of retinal T-lymphocyte–derived interferon γand T-lymphocyte activation and, more recently, inhibition of macrophage activation.^{13 - 14} Soluble TNF is regarded as a ligandfor p55 and is required for the generation of a full phenotype of inflammatorylesions in experimental models.^{40 - 41} Thisimplies that the principal action of TNF receptor 1 and, therefore, TNFr-Igis against soluble TNF. In the present study, we could show no significantsuppression of TNF production, although in our cohort it was uniformly lowin unstimulated T-lymphocyte cultures (data not shown).

In summary, we have demonstrated that TNF-α blockade with TNFr-Igwas effective in rescuing patients with sight-threatening noninfectious PSIIresistant to conventional immunotherapy. Therapy with TNFr-Ig not only ledto the improvement of clinical markers of ocular inflammation but also toimproved VR-QOL and HQOL. In addition to its use as a rescue therapy in thisseries, TNFr-Ig may have a role as a maintenance immunotherapy for patientswho are refractory to conventional agents, but repeat infusions would be requiredto maintain long-term remission. The results of this pilot study support theneed to assess anti-TNF therapies in longer-term treatment regimens for PSIIin the context of randomized clinical trials.

Corresponding author: Andrew D. Dick, FRCP, FRCOphth, MD, Divisionof Ophthalmology, University of Bristol, Bristol Eye Hospital, Lower MaudlinStreet, Bristol BS1 2LX, England (e-mail: A.Dick@bristol.ac.uk).

Submitted for publication June 16, 2003; final revision received November14, 2003; accepted November 17, 2003.

This study was supported by Fujisawa Ltd, London, England (Mr Murphy);by The Grampian University Hospital Trust for Research and Development, Aberdeen,Scotland (Dr Greiner); in part by Millenium Pharmaceuticals Inc, Boston, Mass(Drs Rebello and Hale); and in part by The National Eye Research Center, Bristol,England.

We thank the team at the Therapeutic Antibody Centre, University ofOxford, Oxford, England, for assistance with the manufacture of TNFr-Ig.