Seasoning to Perfection: How to Optimize Anti-TNF Therapy

KEYWORDS:

• Adverse effects
• anti-TNF
• immunosuppression
• pharmacokinetics
• uveitis

Every physician who cares for patients with non-infectious uveitis or other immune-mediated diseases seeks to recommend therapy by finding the optimal balance between likelihood of therapeutic benefit versus chance of risk, i.e. an adverse effect. Throughout most of the world, caregivers who advise patients with uveitis have few choices of medications. If oral, topical or regional corticosteroids fail along with the synthesized anti-metabolites such as methotrexate or mycophenolate, the choice usually comes down to a biologic. And the biologic of choice is often adalimumab, based on pivotal, placebo-controlled, randomized trials.^1,2 If in certain situations adalimumab is recommended to treat uveitis, what is the best way to maximize its benefit to risk ratio? We liken this to how a great chef might season a slow cooking meal, stirring, tasting, and optimizing the spices. In this issue of OII, McKay, and colleagues report on one potential reason why adalimumab therapy could fail: the body could generate anti-drug antibodies (ADAbs) that neutralize the monoclonal antibody.³

Many factors could contribute to a medication failing to achieve its therapeutic goal. First, the pathogenesis of the disease might be independent of the mechanism targeted by the medication. Acute retinal necrosis should respond to appropriate anti-viral therapy, but birdshot chorioretinopathy would not be expected to improve with antiviral therapy. Since non-infectious uveitis is a heterogeneous collection of diseases, the possibility that TNF is not contributing to the disease should always be borne in mind. Second, non-adherence, which could be driven by multiple factors including cost, is a possible explanation. Third, genetic factors can affect the metabolism of many medications.⁴ Fourth, we are increasingly appreciating that the microbiome affects the pharmacodynamics of many drugs.⁵ Surprisingly, this role for the microbiome even includes an impact on the efficacy of monoclonal antibodies.⁶ One group is analyzing whether functional brain magnetic resonance imaging could predict the response to the TNF inhibitor, certolizumab, in patients with rheumatoid arthritis.⁷ Others have suggested that machine learning could be used to optimize the benefit from TNF inhibitors.⁸ Clearly, biomarkers would be valuable in these pursuits. Identifying informative biomarkers has been challenging for rheumatoid arthritis, a disease, which is both more common and more uniform than uveitis.

Finally, and most relevant to the study by McKay and colleagues, a monoclonal antibody could prove ineffective due to the development of neutralizing ADAbs. Monoclonal antibodies generally have a name that ends as “umab,” designating fully humanized, or “imab,” designating chimeric, usually partially mouse. An “imab” like infliximab is more likely to induce neutralizing ADAbs than a “umab” like adalimumab.⁹ Rheumatologists are the physicians who are arguably the most likely to prescribe a biologic. Some studies suggest that therapy can be optimized by monitoring for the development of ADAbs, determining trough drug levels and/or measuring the bioactivity of the monoclonal antibody,¹⁰ while other reports concluded that such monitoring rarely adds to therapeutic decision-making.^11,12 A related question is whether the addition of an anti-metabolite-like methotrexate to anti-TNF therapy could reduce the likelihood that neutralizing antibodies will develop. Such a strategy is plausible but must be weighed against the additional risk, cost, and inconvenience of a second immunosuppressive medication. Many rheumatologists will at least consider this strategy if the underlying disease like rheumatoid arthritis is likely to respond to an antimetabolite and avoid this strategy if the underlying disease like ankylosing spondylitis is not responsive to anti-metabolite therapy. For example, the addition of adalimumab to methotrexate in the SYCAMORE trial for uveitis associated with juvenile idiopathic arthritis (JIA) was a rational strategy because methotrexate itself is useful in controlling the uveitis associated with JIA.¹³

ADAbs are associated with lower trough levels of anti-TNF drugs, higher risk of hypersensitivity reactions and lower rates of remission.⁹ Measuring ADAbs and/or trough drug levels is a logical and attractive approach to individualize drug dosing precisely. However, thus far in clinical trials, therapeutic drug monitoring of TNF-inhibitors does not improve clinical outcomes.^12,14 There are likely multiple reasons for this, including differences in underlying disease, variability in assays measuring ADAbs and variability in ADAbs themselves. Most assays measure binding of the antibody to the drug without measuring the in vivo neutralizing effect of the ADAb, and thus may detect ADAbs of little clinical significance.¹⁵ It is not surprising that patients may have ADAbs and be in remission.¹² Changing therapy purely based on therapeutic drug monitoring may not improve clinical outcomes.¹⁵ Given these limitations, one could argue that monitoring the patient’s disease is more important than attempting by a laboratory test to predict treatment failure, particularly in an organ lending itself well to monitoring, such as the joint, skin, or eye.

In the open-label extension of the Visual I and Visual II trials, patients and physicians were allowed to use clinical judgment that included the use of topical, regional, or oral corticosteroids to maximize therapeutic benefit.¹⁶ In this paradigm, roughly 80% to 90% of subjects were graded as a therapeutic success. McKay and colleagues have explored an important potential cause of a disappointing therapeutic result, neutralizing antibodies. We can be confident that future care for patients with uveitis will continue to improve. From our perspective as rheumatologists, ADAbs have not yet proven to help guide clinical decisions. However, as we improve our ability to understand and detect clinically relevant ADAbs, measuring these antibodies may eventually become a routine part of patient care. A physician treating patients with uveitis will be able to use a strategy long employed by the greatest of chefs, precise titration of the right combination of ingredients.

Declaration of Interest

JTR consults for AbbVie, Gilead, Roche, Novartis, Revolo, Roivant, Neoleukin, Santen, Horizon, UCB. He receives royalties from UpToDate. He receives research support from Horizon and Pfizer. He serves on the data monitoring board for Celgene-Bristol Myers. MAF consults for Revolo.

Additional information

Funding

This work was supported by NIH Grants EY026572 and KL2TR002370, and the OHSU Department of Medicine Wheels Up Program. JTR receives financial support from the Grandmaison Fund for Autoimmunity Research, the William and Mary Bauman Foundation, and the Stan and Madelle Rosenfeld Family Trust.