Therapeutic monitoring of TNF-alpha inhibitors in rheumatoid arthritis

5.1 A clinical expert explained that rheumatoid arthritis has a devastating effect on a person's quality of life and almost 1 in 3 people stop work within 2 years of diagnosis. The patient experts commented that active disease and flares have the biggest effect on their lives and they constantly fear their recurrence. The committee noted that there is no standard definition of flare. Further research to better define it would be beneficial but may be challenging because of the variability in disease presentation. The committee also heard that 'low disease activity' covers a wide range of disease presentations, and people with rheumatoid arthritis can continue to have pain even when their joints are not visibly swollen.

5.2 The committee heard that tumour necrosis factor (TNF)‑alpha inhibitors can be an effective treatment option for severe rheumatoid arthritis that has not responded to conventional therapy. But some people have disease that does not respond or loses response to TNF‑alpha inhibitors. Based on British Society for Rheumatology Biologics Register for Rheumatoid Arthritis (BSRBR‑RA) data, approximately 50% of people who have TNF‑alpha inhibitors stop within the first 4 years because of lack of efficacy and adverse events such as severe infections. The committee concluded that managing rheumatoid arthritis is complex. New tests to optimise the use of TNF‑alpha inhibitors could improve management of the condition and so improve outcomes and quality of life for people with rheumatoid arthritis.

5.3 The committee noted that therapeutic monitoring of TNF‑alpha inhibitors is more established in inflammatory bowel disease. It was informed that companies who make TNF‑alpha inhibitors may offer to cover the cost of the enzyme-linked immunosorbent assay (ELISA) kits if a trust switches to their drug. The committee concluded that because testing is already being done and may be provided free of charge, clinicians may potentially have access. Therefore, it was important to consider whether testing in rheumatoid arthritis is appropriate.

5.4 The committee discussed treatment options for people with rheumatoid arthritis whose treatment target (remission or low disease activity) with TNF‑alpha inhibitors had been reached. The clinical experts explained that, in the UK, most people continue their treatment at the standard dose, that is, their dose is not reduced. But the committee was also informed that an increasing number of trusts do reduce the dose of TNF‑alpha inhibitor (based on clinical judgement) for these people. These trusts would have quick access policies in place for people reducing their dose, so they can return to clinic should their disease need to be reassessed or treatment adjusted.

5.5 The committee considered the potential value of therapeutic monitoring of TNF‑alpha inhibitors in people with rheumatoid arthritis whose treatment target (remission or low disease activity) had been reached. The committee noted that compared with no dose reduction, therapeutic monitoring could help inform clinicians who could reduce their dose of TNF‑alpha inhibitor without loss of efficacy. Lower doses could mean a lower risk of side effects, such as serious infections, and lower costs for TNF‑alpha inhibitors. The committee noted that compared with dose reduction based on clinical judgement only, therapeutic monitoring may have a limited effect on the average dose of TNF‑alpha inhibitor and rates of adverse events. However, helping to better inform clinicians who can reduce their dose without loss of efficacy could lead to a lower rate of relapse and flares.

5.6 A patient expert explained that therapeutic monitoring of TNF‑alpha inhibitors can reassure people with rheumatoid arthritis who wish to consider reducing their dose. People may be uneasy about reducing their dose, fearing they may have disease recurrence or a flare as a result. The committee concluded that therapeutic monitoring of TNF‑alpha inhibitors could potentially reassure people with rheumatoid arthritis about reducing their dose more than when their dose is reduced based on clinical judgement alone.

5.7 The committee noted that therapeutic monitoring of TNF‑alpha inhibitors could improve adherence to TNF‑alpha inhibitors by helping to identify people for whom this could be a problem. The importance of adherence could then be discussed with them. However, clinical experts commented that concerns about adherence would not be the main reason for using therapeutic monitoring of TNF‑alpha inhibitors.

5.8 The committee reviewed the available evidence on the clinical effectiveness of using ELISA tests for therapeutic monitoring of TNF‑alpha inhibitors in people with rheumatoid arthritis. The committee noted that there was no evidence available for people with disease that has not responded to TNF‑alpha inhibitors or has stopped responding to TNF‑alpha inhibitors. It also noted that the evidence for people whose treatment target had been reached was limited and of poor quality. The committee was aware that no UK studies had been identified and there was no evidence for the IDKmonitor, LISA‑TRACKER and RIDASCREEN ELISA tests.

5.9 The committee was concerned that the results of the INGEBIO study may not be generalisable to the NHS. This was because of differences in the healthcare settings between Spain and the UK, and the lack of an explicit algorithm for guiding clinicians in how the test results should be interpreted and how they affect treatment. A patient expert explained that a warm climate has a favourable effect on symptoms, so there could be some differences in disease presentation between Spain and the UK. The committee noted that, because both treatment groups were enrolled at the same study sites, climate should not affect the results. Also, the committee was aware that the rate of dose reduction of TNF‑alpha inhibitors based on clinical judgement alone (standard care) and the rate of dose reduction based on clinical judgement plus therapeutic monitoring of TNF‑alpha inhibitors was similar in both study arms. The committee recalled that dose reduction based on clinical judgement is not routine practice in the NHS (see section 5.4). It concluded that INGEBIO may not be generalisable to clinical practice in the NHS.

5.10 The committee considered other limitations of INGEBIO. It noted that the study findings were presented as abstracts only, but the full study report had now been provided as commercial in confidence by Grifols. A full-text publication is being prepared. The committee was aware that the study had a non-randomised design and that baseline imbalances reported in the full study report were concerning. The external assessment group (EAG) explained that there was no adjustment for baseline imbalance in disease activity between groups. Also, there were unclear differences in clinical outcomes between the intention-to-treat analysis and the analysis that excluded 19 people who were lost to follow up. The committee noted that the study enrolled a mixed population of people with different rheumatic diseases, with only 37% of people having rheumatoid arthritis. The clinical experts explained that rheumatic diseases have different rates of immunogenicity and therapeutic ranges but algorithms to interpret test results should be similar across these diseases. The committee noted a trend towards a reduced rate of flares with therapeutic monitoring, but the difference was not statistically significant. Also, the committee noted that the rates of flares were not stratified by dose and so did not provide information as to whether doses of TNF‑alpha inhibitors can be reduced without loss of efficacy. The committee noted that without this dose-relationship information, the differences seen in INGEBIO could simply be caused by chance. The committee concluded that the clinical outcomes reported were uncertain.

5.11 The committee considered the single-centre observational study by Pascual-Salcedo et al. (2013) and the small randomised controlled trial by l'Ami et al. (2017). It discussed its doubts about the generalisability of the Spanish observational study to the NHS because of differences in healthcare setting, the lack of a control group and enrolling people with mixed rheumatic diseases. The committee noted that l'Ami et al. enrolled a small number of people, had a short follow-up time, and the median doses of TNF‑alpha inhibitor (adalimumab) in both treatment groups were not statistically significantly different at the end of the study. The committee also noted that l'Ami et al. only randomised people with high blood levels of TNF‑alpha inhibitor. It did not provide any information on treatment choices and outcomes for people with lower levels of TNF‑alpha inhibitor. The committee concluded that the 2 studies had important limitations, but they provided some support that therapeutic monitoring could help reduce doses of TNF‑alpha inhibitors without negatively affecting clinical outcomes (that is, without a subsequent increase in disease activity).

5.12 The committee considered the analytical validity of the tests. A clinical expert explained that there is no formal external quality assurance scheme for measuring levels of TNF‑alpha inhibitors and antibodies to TNF‑alpha inhibitors, but some laboratories take part in sample-exchange schemes as a form of quality assurance. Work on assuring the quality of ELISA tests for therapeutic monitoring of TNF‑alpha inhibitors is ongoing and is most advanced for infliximab, for which World Health Organization international calibration standards have been developed. These calibration standards were shown to improve the consistency of results between different laboratories. A clinical expert explained that there is variability between results generated by the different ELISA tests, especially for TNF‑alpha inhibitors other than infliximab. The committee concluded that there is still potential uncertainty in the analytical performance of the ELISAs.

5.13 The committee noted that studies on the clinical validity of measuring levels of TNF‑alpha inhibitors (that is, studies looking at correlation between test results and health states such as remission or active disease) were not included in the assessment. It concluded that considering the very limited and poor-quality direct evidence on the clinical utility of ELISA tests (that is, information showing how treatment decisions informed by ELISA test results affect outcomes for people with rheumatoid arthritis), information on the clinical validity of ELISA tests could be beneficial. However, this information would not be able to confirm their clinical utility.

5.14 The committee considered the choice of model structure. It recalled the uncertainties associated with INGEBIO (see sections 5.9 and 5.10), which provided the main clinical inputs for the model. Because of this, the committee agreed with the choice of a simple modelling approach. It concluded that the model results were of limited value because of a lack of robust clinical data.

5.15 At the first meeting, the committee discussed the primary analyses based on the Ucar et al. and Arango et al. conference abstracts (see section 5.16). It noted that the costs of managing health states appeared to be high. At the second meeting, the committee considered the additional analyses based on the INGEBIO full study report and updated health state costs. The committee concluded that because of the error in the health state costs, the results of the primary analyses were inaccurate. It noted that none of the analyses adjusted for the baseline imbalances between the 2 study arms in INGEBIO. Therefore, the costs and quality-adjusted life years (QALYs) estimated from the model are likely to be flawed. The committee also agreed that the degree of uncertainty in the current clinical evidence was too high to use the model for decision making.

5.16 At the first meeting, the committee discussed differences between the 2 sources of clinical data from INGEBIO for the 2 primary analyses and noted both were conference abstracts. It was aware that the Ucar et al. intention-to-treat analysis reported time in remission, whereas Arango et al. excluded people lost to follow up and reported time in remission or low disease activity pooled together. As a result, health states were defined differently in the 2 primary analyses: remission compared with active disease (low to high disease activity) in the first analysis, and remission or low disease activity compared with active disease (moderate to high disease activity) in the second analysis. The committee agreed that in the EAG's model based on INGEBIO, the time spent in each health state was a key driver of the cost-effectiveness results. The committee also noted that if the comparator in the model was no dose reduction it would be likely that the amount of drug would be a key driver of the cost-effectiveness results, and not the time spent in each health state.

5.17 The committee noted that the rates of adalimumab dose reduction in INGEBIO were similar in the 2 treatment groups. As a result, the model did not provide information on whether therapeutic monitoring could offer savings to the NHS on the costs of adalimumab compared with the current practice of no dose reduction (see sections 5.4 and 5.5). The committee also noted that the similar rates of dose reduction in both groups explained why the results were not sensitive to changes in the price of adalimumab, even when discounts of up to 80% were considered. The committee agreed that in the NHS, rates of dose reduction and biosimilar prices are expected to affect the cost effectiveness of therapeutic monitoring of TNF‑alpha inhibitors. The committee was aware that in Gavan's cost-effectiveness modelling, based on the BSRBR‑RA data, therapeutic monitoring of TNF‑alpha inhibitors was generally cost effective compared with no dose reduction. But it was unlikely to be cost effective relative to dose reduction based on clinical judgement. The committee concluded that the EAG model may not be representative of NHS practice, in which dose reduction of TNF‑alpha inhibitors is not routinely done.

5.18 The committee acknowledged that the rates of flares in INGEBIO were not stratified by dose and so the relationship between adalimumab dose and the rate of flares was not captured in the model. It concluded therefore, that the model may not accurately reflect the experience of people with rheumatoid arthritis in the NHS whose dose of TNF‑alpha inhibitors is reduced.

5.19 The committee noted that the cost of a phlebotomy appointment appeared to be high but clinical experts explained that it likely represents the true cost of an outpatient phlebotomy appointment. They commented that although people with rheumatoid arthritis taking TNF‑alpha inhibitors (especially those also taking methotrexate) have frequent monitoring, an additional phlebotomy appointment may be needed to measure trough drug levels. This additional appointment would not be needed if drug levels of TNF‑alpha inhibitors could be measured at any time in the administration cycle. This was explored in sensitivity analyses.

5.20 The committee discussed the limitations of the economic model. It considered that although the clinical studies for therapeutic monitoring of TNF‑alpha inhibitors show promising results, the degree of uncertainty in the clinical evidence was too high for it to be able to use the incremental cost-effectiveness ratios (ICERs) for decision making. It concluded that the scope of any further changes to the modelling assumptions would be limited without more robust clinical data. The committee noted other evidence gaps such as: