Biologic agents, such as TNF-α inhibitors (TNFis), have revolutionized the treatment of rheumatoid arthritis (RA), but despite this advance, not all patients respond favorably. Up to 40% of RA patients do not respond to the first biologic (primary failure) or lose response over time (secondary failure). Drug survival (the time to discontinuation of a drug) is influenced by many factors including lack or loss of efficacy, adverse events (AEs), and poor adherence. Immunogenicity is defined as the ability of biopharmaceutical products (BPs) to induce an immune response, resulting in the generation of anti-drug antibodies (ADAs). ADAs are considered an important (albeit not the only) mechanism of secondary treatment failure and limited drug survival, due to effects on pharmacokinetics and bioavailability. ADAs are also implicated in treatment-related AEs, such as infusion and injection-site reactions (1). Immunogenicity testing is a mandatory, regulatory requirement for BP drug licensing, as part of the safety profile package required by both the US Food and Drug Administration (2) (FDA) and the European Medicines Agency (3) (EMA).

Therapeutic drug monitoring (TDM) and immunogenicity testing, using trough drug levels and ADAs, have the potential to improve clinical decision-making, by influencing drug selection, dose, and frequency of administration. This may allow clinicians to reduce under- and over- treatment for patients in clinical relapse or remission. There are currently no consensus guidelines recommending the use of BP drug levels and immunogenicity testing in RA, and as such, their use in clinical practice is widely variable.

TNFis (in combination with methotrexate and as monotherapy) are often selected as first-line biologic therapy in patients with RA who are refractory to non-biologic disease-modifying antirheumatic drugs (DMARDs), due to the availability of long-term data from clinical trials and extensive real world experience. Moreover, costs have recently lowered due to the advent of biosimilar TNFis. Infliximab, adalimumab, and etanercept (in bio-original and biosimilar forms) are the most frequently used TNFis, with the most available data. Here, we discuss the utility and clinical relevance of TDM and immunogenicity testing of TNFis in patients with RA, and potential future immunopharmacological strategies to personalize disease management.

Current options for managing TNFi failures in RA include cycling within class, i.e., to an alternative TNFi, or switching between class i.e., to a drug with a different mechanism of action. Published recommendations provide little guidance to determine the best strategy (47, 48). Both options are supported by data from randomized controlled trials and the real world, therefore the decision is generally empirical and based on physician discretion. This dilemma was summarized in a recent review (29). In the open-label, 52 weeks randomized Rotation or Change (ROC) trial, the treating physician selected between a second TNFi and a non-TNFi in patients with primary TNFi failure (49). The ROC trial results concluded that the reasons for improved drug survival when switching to a second TNFi was better efficacy, and with switching to a non-TNFi was reduced AEs. Further evidence from a prospective study, suggests better outcomes can be achieved using an algorithm based on trough drug levels and ADAs, compared with “empirical switching” (50).

Current treatment recommendations for RA endorse combination therapy with a biologic and DMARD (47, 48), which is consistently more effective than biologic monotherapy, possibly due to effects on immunogenicity. Methotrexate significantly increases adalimumab trough concentrations (51, 52), and in a dose- dependent manner, reduces immunogenicity (51), and improves clinical outcomes in early disease (53).

Given the limitations regarding assay diversity and data interpretation, and the lack of conclusive support for cost- effectiveness, routine use of TDM and ADA testing has not been widely adopted in British Rheumatology practice (54). There are exceptions, with local management algorithms for RA incorporating these tests (55, 56), but overall the use and interpretation of TDM and ADAs is inconsistent. By contrast, The British Society for Gastroenterology guidelines for the management of IBD includes clear, algorithmic recommendations for measurement of drug levels (±ADA) (57). In IBD, clinical decision making using drug levels and ADAs in secondary non-responders is more cost-effective when compared to empirical drug escalation (58, 59). The recent prospective, observational personalized anti-TNF therapy in Crohn's disease study (PANTS), demonstrated that low concentrations of adalimumab and infliximab at week 14 were associated with primary non-response, non-remission at week 54 and the development of ADAs (32). ADAs predicted subsequent low drug levels and concomitant immunomodulators (thiopurine or methotrexate) mitigated the risk of developing ADAs (32).

In time, readily available, accurate assays to measure drug levels and ADA titer, will hopefully arm clinicians with powerful tools to optimize the management of RA, especially in patients with secondary loss of response. A potential algorithm that could be used in future management strategies is shown in Figure 1.

Measurement of trough drug levels is the most valuable test in the first instance to identify patients with low or optimal (therapeutic) circulating drug. Using ADAs for the first branching in the algorithm is probably inappropriate, as ADAs are not always clinically relevant (especially if present at low-titer) if there is sufficient circulating drug. In cases of treatment failure, supplementary knowledge of ADAs (and perhaps the titer) may be helpful in determining the etiology of suboptimal drug levels. Low drug levels without ADAs may be due to factors such poor adherence to therapy (as most biologics are self-administered injections), a higher BMI and/or faster drug metabolism, which would require different strategies compared with those for patients with detectable ADAs. To overcome this problem, optimizing the dose of biologic by reducing the interval of administration, e.g., changing adalimumab monotherapy from fortnightly to weekly [as permitted by the National Institute of Health and Care Excellence (NICE) in the U.K. (60)], or optimizing dose of concomitant immunosuppressants may recapture a response (61, 62). Emerging evidence suggests that efficacy can be re-established in ADA positive patients with secondary failure, by addition of methotrexate to infliximab treatment in IBD (63), although there is limited support for this approach in the RA literature. If these strategies are unsuccessful or not applicable, switching BP should be considered.

If ADAs are detected in the context of a low drug level, switching to a less immunogenic drug within the same class (e.g., etanercept) could be beneficial, especially if the patient has previously responded to a TNFi. Switching to a second TNFi may be successful due to differences in drug molecular structure, immunological action, immunogenicity, and pharmacokinetics, as well as different underlying disease pathogenesis (24). There is an argument however, to switch to a biologic with a different mechanism of action, as although ADAs are not cross-reactive, patients with ADAs to the first failed TNFi are more likely to seroconvert and produce ADAs with subsequent TNFis (64–67) and are thus less likely to respond to a second TNFi, especially if this is a monoclonal antibody (64, 66). Of note, ADAs to bio-originals are reactive to the corresponding biosimilars, and therefore after detection of ADA, switching a bio-original to its biosimilar version would not be recommended (68). It is plausible to suggest that a patient with ADAs, refractory to multiple biologics, may benefit from a treatment with a less or minimally immunogenic drug, e.g., a receptor fusion protein e.g., abatacept (69) or a small molecule [JAK inhibitor (70)]. In the case of non-responders with optimal drug levels, the presence/absence of ADAs is unlikely to influence subsequent management. These patients have a lower probability of response to an agent within the same class and therefore we would postulate that they are most likely to benefit from switching to a drug with a different mechanism of action (64).

Given the high cost and potential AEs associated with biologic therapies, strategies have been proposed to taper biologics (by reducing drug doses or increasing dosing intervals) in patients with sustained clinical remission, thereby reducing risks and costs overtreatment. In some studies, correlation between DAS28 (disease activity score; a composite measure of disease activity in RA) improvement and serum drug trough levels has been verified up to a threshold of drug level, above which no significant DAS28 changes occur (71). A recent study using certolizumab found that a drug level above a defined threshold was not associated with any additional clinical benefit, and therefore it may be possible in the future to use TDM to titrate treatment (15). Withdrawal of treatment in disease quiescence is an area of active research and currently there is insufficient evidence to draw meaningful conclusions about the role of TDM and immunogenicity testing. Data from ongoing, randomized controlled trials (72) using TDM or ADA to guide withdrawal strategies may inform future practice. It is reasonable to hypothesize that drug withdrawal may be possible in patients with inactive disease and undetectable drug levels or high ADA titres, as remission is probably not being maintained by treatment with the BP.

The increasing and earlier use of BPs in RA is likely to lead to a greater proportion of patients receiving these therapies. Efforts are expanding to predict, reduce and reverse BP immunogenicity to mitigate the impact on drug development, which was summarized in a recent review (73). Strategies to reduce the immunogenic potential of BPs include “de-immunizing” approaches through protein engineering e.g., rational amino acid substitutions and/or addition of epitope-masking moieties, as well as induction of peripheral tolerance (73). There are emerging concerns that immunogenicity may limit the development of newer investigational medicinal products such as the bispecific antibodies.

Despite long-standing interest and accrual of data, we are still unable to predict responses to TNFi. Prospective, longitudinal studies of BP-naïve patients may provide mechanistic information and address a critical unanswered question—why BPs are immunogenic in some patients, but tolerogenic in others. Prediction of immunogenicity may allow mitigation and management strategies to be implemented to prevent or minimize the generation of ADAs (73). Other strategies to personalize biologic selection, include pharmacogenetic testing to identify genetic factors that may predict lack of response to, or toxicities from, TNFi (74).

Further research is needed to develop standardized, clinically-validated assays for both drug and ADA testing. These tests could then be incorporated into evidence-based guidelines to optimize treatment decisions along the patient pathway: for patients with active disease about to start treatment, not responding to treatment (primary or secondary failure) or for those in remission, to permit drug tapering strategies. Taken together this may help to improve the long-term efficacy, safety profile and cost-effectiveness of BPs.

PM and JM co-wrote this manuscript and both authors approved the final version.