Introduction

Uveitis is defined as an inflammation of the iris, ciliary body, vitreous, retina and/or choroid. About 10%–15% of cases of preventable blindness in western countries are caused by uveitis or its complications.1 2 About 60 causes of uveitis have been described and can be classified into five groups, which are pure ophthalmological entities, infectious and inflammatory diseases, masquerade syndromes and drug-related uveitis.3 In western countries, approximately a quarter of uveitis cases are related to ophthalmological diseases, a quarter to systemic diseases fulfilling consensual diagnostic criteria, a quarter to presumed systemic diseases, and a quarter remain unexplained.4

The treatment for chronic non-infectious uveitis (CNIU) is mainly based on local or systemic corticosteroids or immunosuppressive therapy (such as methotrexate and mycophenolate mofetil). Recently, it has expanded to biotherapies, especially using antitumour necrosis factor (TNF). Based on the results from VISUAL15 and VISUAL26 randomised controlled trials, adalimumab (ADA) has been approved by the Food and Drug Administration and the European Medicines Agency for the treatment of non-infectious non-anterior uveitis in adult patients with insufficient response to corticosteroids.

Therapeutic drug monitoring (TDM) relies both on the determination of serum drug level and the research of antidrug antibodies in order to guide therapeutic decision. Many studies in the fields of rheumatology, dermatology and gastroenterology have shown the benefits of TDM on therapeutic efficacy and on costs.7–11 TDM helps with the evaluation of primary or secondary non-responders and with the dose adjustments for patients presenting low-drug concentrations, and can also reflect compliance. Only a few studies have evaluated the benefits of anti-TNF TDM in patients with CNIU.12–14 These studies mostly focused on ADA and showed that higher trough levels were associated with a better clinical response in uveitis patients in context of juvenile idiopathic arthritis,14 Behcet’s disease13 or sarcoidosis.13 Moreover, concomitant treatment with conventional synthetic disease-modifying antirheumatic drugs (csDMARDs) decreases the rate of antibodies against ADA (AAA)12 and positive AAA is associated with low drug level and loss of response to ADA.12

However, in these previous studies, the size of the population was small and they were mainly focused on the association of ADA trough levels and AAA with the clinical response.12–14 None of these studies has evaluated the practical application of TDM in both responders and non-responders. Two studies included only chronic anterior uveitis associated with juvenile idiopathic arthritis.12 14

The aim of the present study was to assess the relevance of TDM in CNIU treated with ADA; this was performed by evaluating the effects of adjusting the treatment regarding intraocular inflammation in both responders and non-responders.

Patients and methodsStudy design and patient recruitment

A single-centre retrospective study was conducted including adult patients with CNIU referred to the department of Internal Medicine (Hôpital de la Croix-Rousse, Hospices Civils de Lyon, Lyon, France), treated with ADA, and for whom at least one dosage of serum ADA level and an AAA serology were performed. The uveitis diagnosis was confirmed for all patients by a complete ophthalmological examination, consisting in slit lamp biomicroscopy and fundus examination. The Standardisation of uveitis nomenclature was used for anatomical classification.15 The aetiological diagnosis of CNIU was performed on the basis of internationally admitted criteria. Patients receiving ADA for any other cause than uveitis (such as spondyloarthritis) were excluded. Patients were recruited between June 2003 and July 2019. This study was registered on clinicaltrials.gov (NCT 03863782).

Anti-TNF treatment

Treatment was started in patients who presented active uveitis and who had received at least one csDMARD. A subcutaneous injection of 40 mg of ADA was performed every other week. Previous therapy (corticosteroids, methotrexate, azathioprine, mycophenolate mofetil) was continued if tolerated. Patients were evaluated clinically before starting the treatment and 4, 12 and 24 weeks after. If necessary, medical examination could be performed more often for non-responders. ADA and AAA measurements were performed in case of relapse of intraocular inflammation or persistent intraocular inflammation during the first 3 months of treatment with ADA for non-responders and after 6 months of stable uveitis before spacing the injections in responders. ADA side effects were reported directly by the patients and recorded in their medical file by the clinician.

Measurements of serum trough ADA and AAA

To avoid interferences in the assay, blood venipuncture was performed just before drug administration, and blood samples were aliquoted. Trough ADA levels were measured from one serum aliquot using a commercially available ELISA kit (Promonitor-ADL, Progenika Biopharma SA, Spain). Results were expressed as μg/mL. The reference cut points were: <4 µg/mL (subtherapeutic levels), between 4 and 10 µg/mL (therapeutic levels) and >10 µg/mL (supratherapeutic levels). AAA were measured on another aliquot using an ELISA kit (promonitor-anti-ADL, Progenika Biopharma SA, Spain). The cut-off value for the presence of AAA was 10 arbitrary units (AU)/mL.

Uveitis outcome measurement

Uveitis clinical evaluation included visual acuity (Snellen best-corrected visual acuity) and a complete ophthalmological examination. A slit-lamp examination was used to evaluate the anterior chamber and indirect ophthalmoscopy was performed to grade vitreous haze and to evaluate posterior segment. Macula oedema was evaluated by optical coherence tomography (OCT). Vasculitis was evaluated by fluorescein and indocyanin green angiography.

Response to treatment was categorised as complete, partial or no response.13 14 Complete response was defined by the presence of grade 0 cells in both anterior and posterior segments and by the absence of any other sign of intraocular inflammation on OCT and angiography. Partial response was defined by a two-step decrease in anterior chamber cells or vitreous haze, decrease of macular oedema or vasculitis without any finding consistent with the criteria of complete response. No response was defined by a persistent intraocular inflammation without any finding consistent with the criteria of partial response. Non-responders were further categorised into primary or secondary non-responders. Primary non-responder patients were defined by the presence of persistent intraocular inflammation during the first 3 months of treatment with ADA.13 14 Secondary non-responders were defined by the presence of re-occurrence of intraocular inflammation signs after an initial partial or complete response for at least 3 months.

Data collection

The following data were collected: patient sex, age, ethnic group, follow-up duration, ADA treatment characteristics (dose, interval injection) and ophthalmological characteristics at diagnosis and after beginning ADA (anatomic type of uveitis (anterior, intermediate, posterior, pan uveitis),15 anterior chamber examination [slit lamp, granulomatous or not), laterality, presence of macular oedema, vasculitis and reason for starting ADA treatment (severe uveitis, side effects or failure of previous treatment)).

Since the primary objective was to assess TDM-guided optimisation in CNIU patients, were also analysed: ophthalmological characteristics after the start of ADA treatment, date of TDM, and therapeutic adjustment made after TDM (injection intervals, dose and change in treatment).

Statistical analysis

Data were expressed as median (IQR) for continuous variables, mean (SD) for normally distributed continuous variables and frequency (percentage) for qualitative variables. Fisher’s exact test and χ2 tests were used for comparisons involving qualitative variables, and the Wilcoxon rank-sum test was used for comparisons involving continuous variables.

All tests were two tailed, and a p< 0.05 was considered significant. All statistical analyses were performed using R statistical software (R Foundation for Statistical Computing, Vienna, Austria).

ResultsPatient characteristics

Among the 98 adults with CNIU who were treated with ADA, 44 patients underwent TDM (online supplemental figure 1). Their median (IQR) age at diagnosis was 35 (27–47) years, the M/F sex ratio was 0.69, 33/44 (75%) patients were non-Caucasians, 9/44 (20%) were North African, 1/44 (2.5%) was Asian and 1/44 (2.5%) originated from sub-Saharan Africa (table 1).

Table 1

General characteristics, ophthalmological findings and aetiological distribution in 44 patients

The anatomical distribution of uveitis was: anterior uveitis 7/44 (15.9%), intermediate uveitis 7/44 (15.9%), posterior uveitis 13/44 (29.6%) and panuveitis 17/44 (38.8%). A total of 42/44 (95%) patients had bilateral uveitis, 15/44 (34%) uveitis were granulomatous. Retinal vasculitis was present in 19/44 (43%) patients (table 1).

The aetiological distribution was: idiopathic (15/44, 34%), juvenile idiopathic-arthritis (5/44, 11%), Behcet’s disease (4/44, 9%), sarcoidosis (4/44, 9%), birdshot retinochoroidopathy (3/44, 7%), Vogt-Koyanagi-Harada’s disease (3/44, 7%), pars planitis (3/44, 7%) and for 1 case sympathetic ophthalmia, Blau syndrome, Crohn’s disease, Cogan’s syndrome, idiopathic multifocal choroiditis, punctate inner choroidopathy, and serpiginous choroiditis (table 1).

Laboratory and TDM

ADA was started because of severe uveitis 3/44 (6.8%), side effects from previous treatment 14/44 (31.8%) or failure of previous therapy 44/44 (100%). Previous and concomitant systemic treatments were corticosteroids 44/44 (100%), azathioprine 11/44 (25%), methotrexate 10/44 (22.7%), ciclosporine 4/44 (9%), mycophenolate mofetil 8/44 (18%), leflunomide 1/44 (2.3%), hydroxychloroquine 3/44 (6.8%) and/or infliximab 1/44 (2.3%). Side effects of ADA were encountered in 5/44 (11.4%) patients, they were abdominal pain 2/44 (4.5%), headache 1/44 (2.3%), neutropenia 1/44 (2.3%) and drug-induced lupus 1/44 (2.3%), abdominal pain and drug-induced lupus led to the discontinuation of ADA and the start of methotrexate and mycophenolate mofetil in two responders who had low ADA levels.

All of the 44 patients completed at least 3 months of treatment with ADA without relevant adverse event. After 3 months, 35/44 (79.5%) patients displayed a favourable clinical response, among these 21/35 (60%) patients achieved a complete response and 14/35 (40%) achieved a partial response. The remaining 9/44 (20.5%) patients were considered as primary non-responders, while 10/44 (22.7%) patients were considered as secondary non-responders after a mean (range) follow-up of 22 (4–39) months.

TDM results

For each of the 44 patients, 1–4 TDM were performed, and a total of 79 dosages were performed. The median (IQR) trough ADA was 5.7 (0.75–10) µg/mL. TDM was performed in responders before decreasing the frequency of injections (n=48 dosages), in primary non-responders (n=11 dosages) and in secondary non-responders (n=20 dosages). A total of 33/79 (41.8%) dosages found ADA level above the ADA efficacy threshold, and 46/79 (58.2%) dosages were below the threshold or undetectable. AAA were detectable in 6/44 (13.6%) patients, for whom the titres ranged from 21.2 to 316.3 AU/mL.

The median (IQR) ADA levels of responders (7 (2.5–11) µg/mL) was significantly higher than for non-responders (1.8 (0.35–5.7) µg/mL; p=0.0004). There was no statistically significant difference in ADA trough levels between primary non-responders and secondary non-responders patients. In AAA-positive patients, the ADA trough levels were <0.3 µg/mL, and were significantly lower than the median (IQR) ADA trough levels of patients without AAA (4.9 (2–10) µg/mL; p=0.0001). ADA trough levels did not differ significantly neither between complete and partial responders (p=0.22), nor between patients with concomitant immunosuppressors (median (IQR)=5.7 (0.54–8.9) µg/mL) and patients without immunosuppressor (median (IQR)=5.7 (2.3–11) µg/mL; p=0.46).

AAA-positive patients

Among the six AAA-positive patients, one was treated by a combination therapy including methotrexate. Also, five patients were non-responders and four of them discontinued ADA in light of the TDM results (data were missing for one patients). The treatment was switched to golimumab (for three patients) or methotrexate (for one patient). One patient had a stable uveitis despite high AAA titre (26.4 AU/mL) and low ADA level (<0.3 µg/mL), and uveitis remained stable after ADA discontinuation.

TDM-guided therapeutic adjustments and clinical outcome

TDM results led to a change in treatment for 10/11 (91%) of primary non-responders: an increase in the frequency of injections 3/11 (27.2%), an increase in the drug dose 1/11 (9%) or a switch to another treatment 6/11 (54%) (golimumab, infliximab or interferon) (one missing data). Among secondary non-responders, an increase in the frequency of injections was performed for 9/20 (45%) patients and a switch to another treatment for 4/20 (20%) patients (golimumab, tocilizumab or interferon; figure 1).

Figure 1

TDM and therapeutic adjustment in non-responders. AAA, antibodies against; ADA, adalimumab; TDM, therapeutic drug monitoring.

Among non-responders patients for whom the ADA level was above the efficacy threshold, TDM led to a switch to another treatment in 4/9 (44%) cases and an increase in the frequency of injections in 2/9 (22%) cases. For 3/9 (33%) cases, there was no modification in the ADA treatment, but local (2/3 cases) or systemic (1/3 case) corticosteroids were increased. In all patients for whom ADA was discontinued and switched to another one, partial or complete response was observed.

Among all non-responders (primary and secondary), TDM led to an increase in the frequency of injections in 13/31 (41.9%) cases, an increase in the drug dose in 1/31 (3.2%) cases, and a switch to another treatment in 10/31 (32%) of cases (infliximab, golimumab, interferon, tocilizumab or methotrexate; figure 1) (one missing data). There was no modification in the ADA treatment for 7/31 (22%) of cases; local corticosteroids were increased for 6/7 (85%) cases (including 3/6 intravitreal steroids) and systemic corticosteroids were increased in 1/6 (15%) case. In case of a therapeutic adjustment in primary and secondary non-responders patients, the switch from ADA treatment to another treatment led to a partial or complete response in 9/10 (90%) cases, while an increase in the frequency of injections led to a partial or complete response in 11/13 (84%) cases.

Among responders, 31/48 (64.5%) dosages were above the efficacy threshold (4 µg/mL) and 17/48 (35.4%) were below. Among responders for whom the ADA level was above the efficacy threshold, the frequency of injections was decreased for 15/31 (48.4%) cases. Among them, 3/15 (20%) patients underwent a uveitis relapse (figure 2). None of the 16/31 (51.6%) patients for whom the ADA treatment was unchanged experienced a relapse. TDM led to change in treatment in 12/17 (70.5%) responders for whom the ADA level was below the threshold: increase in the frequency of injections 8/12 (66.6%), switch to another treatment because of side effects 2/12 (16.6%) or discontinuation of ADA treatment 2/12 (16.6 %).

Figure 2

TDM and therapeutic adjustment in responders. ADA, adalimumab; TDM, therapeutic drug monitoring.

Discussion

In fields such as gastroenterology, dermatology and rheumatology, emerging evidence is supporting the use of TDM for the optimisation of biological efficacy, safety and cost-effectiveness ADA treatment.7 8 However, to date, little evidence is available regarding the utility of biological TDM in CNIU.

The present study is the first retrospective study to report the clinical use of a TDM-based strategy in adults with CNIU receiving ADA. TDM revealed the presence of AAA about a fifth of non-responder patients, associated with undetectable ADA drug levels, underscoring immunogenicity as a major cause of loss of response in uveitis patients receiving biotherapies. This association between AAA positivity, low ADA drug level and loss of treatment response has been previously demonstrated in uveitis patients.12 Also, the prevalence AAA-positive patients observed herein was similar to a published study.13

As expected, responders had significantly higher ADA levels than non-responders, suggesting that ADA trough levels are associated with clinical response, as it was showed in previous studies in ophthalmology13 14 as well as in other fields (rheumatology,8 16 dermatology7 or gastroenterology.9

In these other fields, TDM is already recommended in routine clinical practice (or at least for non-responder patients).7 9 As shown for patients with rheumatoid arthritis17 or inflammatory bowel diseases,18 TDM should be performed for non-responder patients before concluding to failure of TNF inhibitor treatment. The use of TDM for ADA treatment has been associated with a substantial reduction in costs of medication in context of RA10 and inflammatory bowel disease.17

Previous studies have concluded that ADA trough levels below therapeutic range in non-responders should raise an alert about patient treatment adherence or immunisation against ADA.7 9 19 The present study showed that increasing the frequency of ADA injections to once per week is an effective adjustment in non-immunised non-responders for whom the ADA drug concentration was low, similar conclusion have been formulated elsewhere.9 18 As only one patient benefited from an increased dose (80 mg every 2 weeks), no recommendation can be made. Addition of low-dose csDMARDs could also be considered as it was showed that methotrexate had additive effect to TNF-inhibitor20 and could increase TNF-inhibitor levels,20–22 however, this strategy was not assessed in our study. In these patients, close monitoring should be performed, another TDM and a clinical examination 1 month after could be recommended.18 In immunised non-responders, ADA could be replaced by another TNF-inhibitor.7 9 10 22

As it was showed in our study, in non-responders with ADA trough levels within the therapeutic range, a switch out of the therapeutic class is recommended, including interferon-alpha or antibody directed against IL-6 receptor, such as tocilizumab.10 23 Indeed, in patients who do not respond to ADA despite adequate TNF-inhibitor levels, this therapeutic class is probably not suitable for their disease.9 10 22

In responders with high ADA levels (>10 µg/mL), decreasing the frequency of injections should be considered, as this overtreatment is not cost-effective; however, as it was shown herein, a relapse of uveitis can occur and a close follow-up is required. In immunised responders with ADA levels below the efficacy threshold, a close monitoring is recommended. If no relapse of uveitis occurs, ADA can be discontinued, as suggested elsewhere.7 10 Therapeutic adjustments in non-immunised responders with ADA levels below the efficacy range have been assessed in a few studies.7 10 22 24 In the present study, when the frequency of injections was increased, uveitis remained stable in these patients, as well as in case of ADA discontinuation. As a low ADA level is not expected to have any beneficial clinical effects, ADA discontinuation has been considered in another study.10 Nevertheless, in the context of inflammatory bowel disease, proactive TDM is emerging as an important tool for optimising anti-TNF therapy. Studies have suggested that the intensification of ADA treatment to achieve levels within the therapeutic range even in responders is associated with better disease control.22 24 25 In responders, subtherapeutic ADA trough levels have been shown to increase the risk of developing AAA and secondary loss of response.22 ADA intensification or the addition of a csDMARD in non-immunised responders with ADA below the therapeutic threshold may prevent AAA formation and loss of response.22

Overall, ADA discontinuation should be considered for patients with a complete response and without risk factors for disease relapse. ADA intensification or adding methotrexate should be discussed for patients with severe uveitis or with risk factors for disease relapse.

Based on the clinical approach recommended in the fields of dermatology,7 rheumatology10 and gastroenterology22 and on the results of our study, we propose two algorithms for practical application of TDM in CNIU patients non-responding (figure 3) and responding to ADA (figure 4).

Figure 3

Proposed algorithm for therapeutic drug monitoring of ADA in non-responders with CNIU. AAA, antibody against ADA; ADA, adalimumab; CNIU, chronic non-infectious uveitis; TNF, tumour necrosis factor.

Figure 4

Proposed algorithm for therapeutic drug monitoring of ADA in responders with CNIU. High-risk features: risk factors of disease relapse or severe consequences in event of relapse. AAA, antibody against ADA; ADA, adalimumab; CNIU, chronic non-infectious uveitis.

The present study has several limitations. The small size of the population and the heterogeneity of the included uveitis patients limits the generalisation of its results. Moreover, the retrospective design may have led to some bias in data collection. Further randomised controlled studies including larger numbers of patients are warranted.

Conclusion

TDM of ADA treatment proved relevant to provide CNIU patients with a personalised and optimised treatment course (in terms of frequency and type of drug). In case of therapeutic adjustment in non-responders patients, an improvement was observed in a very large majority of patients. In case of decreased frequency of drug administration in responders with trough ADA levels above the efficacy threshold, no relapse was observed in a large majority of cases. Further prospective studies with larger numbers of patients are needed to confirm the results obtained herein, especially regarding the medicoeconomic aspects of TDM.

Data availability statement

All data relevant to the study are included in the article or uploaded as online supplemental information.

Ethics statementsPatient consent for publication

Not required.

Ethics approval

This study received approval from the local ethics committee in February 2019 (No 19–31).