TNF Inhibitors vs Standard Therapy in Rheumatoid Arthritis
In patients with rheumatoid arthritis, our strategy of treatment with synthetic disease modifying drugs achieved non-inferior outcomes to the NICE approved strategy of treatment with tumour necrosis factor inhibitor over 12 months. It also cost substantially less. Our primary outcome—the score on the health assessment questionnaire—fell in both groups, indicating reduced disability at the trial endpoint. The difference between groups favoured the disease modifying drug strategy, with 95% confidence intervals within the prespecified boundary of non-inferiority. Both strategies improved quality of life similarly and resulted in minimal erosive progression. The tumour necrosis factor inhibitor strategy cost an additional average £5545 (€7570, $8586) per patient. Its main benefit was rapid falls in disease activity and more early remissions. As synthetic disease modifying drugs are "slow acting" agents, this difference is not surprising. Patients who did not respond to six months of treatment with disease modifying drugs and who switched to tumour necrosis factor inhibitors were not disadvantaged.
Over 12 months, health and social care costs were £5545 less in patients who were treated with the combined drug strategy. These reduced costs were associated with reduced scores on the health assessment questionnaire, which have immediate economic benefits. Our results support using the combined drug strategy for 12 months. Longer term observational and detailed limitations economic modelling studies are needed to show whether this approach has any enduring benefits.
Tumour necrosis factor inhibitors have many advantages. Their innovative nature, rapid effects, and ease of use attract patients and clinicians. Their defined mechanism of action contrasts with the uncertainties about the molecular effects of drugs like methotrexate. Combinations of lower cost synthetic disease modifying drugs, however, are relevant for advanced health economies and might be even more relevant in health economies where biologics are unaffordable.
The patients in our trial had diverse ethnicities and deprivation levels and were seen in routine practice settings in geographically dispersed centres in England. We focused on patient centred outcomes, which are crucial for people with arthritis. Levels of disease activity replicated levels in national and international registers before and during treatment.
Our trial had several limitations. Firstly, many eligible patients did not participate, mainly because they did not consent. Although non-consenting patients could have responded differently to participants, participation levels replicated other English grant funded trials in rheumatoid arthritis and recent tumour necrosis factor inhibitors trials. Non-participation is likely to reflect patients' concerns about all intensive treatments and our cautiously worded patient information sheet.
Secondly, disease modifying drugs regimens varied because standardisation was impossible in this context. Thirdly, few trials in rheumatoid arthritis use the health assessment questionnaire as their primary outcome measure. It was relevant in our trial because it measures changes important to patients, is sensitive to improvements with all tested treatments, and helps NICE to determine treatment benefits. Scores on the health assessment questionnaire change less in those with late rheumatoid arthritis, but both groups had clinically important improvements, and the durations of disease in our patients were comparable with those in pivotal biologic trials. Although minimally important differences in scores might be less than 0.22 in routine practice, this lower difference would not change our conclusions.
Fourthly, our trial was un-blinded because blinding is impractical when many different treatments are used. The primary outcome, however, was self completed by patients and not directly influenced by clinicians. Fifthly, our trial lasted only 12 months because when it was designed longer delays in assessing biologic treatments raised ethical concerns.
Finally, the effects of withdrawal and switching treatments need careful consideration. Only a few patients (16/205, 8%) were lost to follow-up and required the imputation of missing data. As the complete case analysis gave similar findings to the intention to treat analysis, it seems unlikely that withdrawals influenced our conclusions. Comparable numbers of patients withdrew from both trial arms, suggesting withdrawals were unlikely to be influenced by the treatment strategies. Overall withdrawal rates were similar to those in comparable rheumatoid arthritis strategy trials involving biologics and disease modifying drugs. The frequencies with which patients discontinued tumour necrosis factor inhibitors were comparable with UK national registry data of routine practice results. The evidence suggests withdrawals in our trial were unlikely to have influenced our main conclusions.
In early rheumatoid arthritis three head-to-head trials have shown that combinations of disease modifying drugs achieve similar benefits to tumour necrosis factor inhibitors. RACAT, the only other trial in established active rheumatoid arthritis, showed that triple therapy (methotrexate, sulfasalazine, and hydroxychloroquine) achieved similar benefits to etanercept plus methotrexate in incomplete methotrexate responders treated for 12 months. Our trial and RACAT had comparable results, despite differences in patients and trial design. They highlight the role of head-to-head trials against effective conventional comparators to evaluate expensive treatments. Although toxicity remains a potential concern with combined disease modifying drugs because single trials like ours cannot fully assess risks, such combinations have been used for many years without evidence of major toxicity.
The current trail questions the status quo of use of biologics in patients with active rheumatoid arthritis in whom methotrexate does not achieve the required response. In these patients the clinical and economic evidence from our trial supports preferentially starting combinations of synthetic disease modifying drugs. We also believe that modernising biologic treatment regimens for rheumatoid arthritis requires more strategy trials. Increasing efficacy is important, and optimising cost effectiveness is crucial. Options include giving biologics to patients with rheumatoid arthritis who are likely to respond to specific agents, discontinuing biologics when patients do not fully respond, and tapering treatment when patients achieve sustained remissions.
Discussion
Interpretation
In patients with rheumatoid arthritis, our strategy of treatment with synthetic disease modifying drugs achieved non-inferior outcomes to the NICE approved strategy of treatment with tumour necrosis factor inhibitor over 12 months. It also cost substantially less. Our primary outcome—the score on the health assessment questionnaire—fell in both groups, indicating reduced disability at the trial endpoint. The difference between groups favoured the disease modifying drug strategy, with 95% confidence intervals within the prespecified boundary of non-inferiority. Both strategies improved quality of life similarly and resulted in minimal erosive progression. The tumour necrosis factor inhibitor strategy cost an additional average £5545 (€7570, $8586) per patient. Its main benefit was rapid falls in disease activity and more early remissions. As synthetic disease modifying drugs are "slow acting" agents, this difference is not surprising. Patients who did not respond to six months of treatment with disease modifying drugs and who switched to tumour necrosis factor inhibitors were not disadvantaged.
Over 12 months, health and social care costs were £5545 less in patients who were treated with the combined drug strategy. These reduced costs were associated with reduced scores on the health assessment questionnaire, which have immediate economic benefits. Our results support using the combined drug strategy for 12 months. Longer term observational and detailed limitations economic modelling studies are needed to show whether this approach has any enduring benefits.
Tumour necrosis factor inhibitors have many advantages. Their innovative nature, rapid effects, and ease of use attract patients and clinicians. Their defined mechanism of action contrasts with the uncertainties about the molecular effects of drugs like methotrexate. Combinations of lower cost synthetic disease modifying drugs, however, are relevant for advanced health economies and might be even more relevant in health economies where biologics are unaffordable.
Generalisability
The patients in our trial had diverse ethnicities and deprivation levels and were seen in routine practice settings in geographically dispersed centres in England. We focused on patient centred outcomes, which are crucial for people with arthritis. Levels of disease activity replicated levels in national and international registers before and during treatment.
Limitations
Our trial had several limitations. Firstly, many eligible patients did not participate, mainly because they did not consent. Although non-consenting patients could have responded differently to participants, participation levels replicated other English grant funded trials in rheumatoid arthritis and recent tumour necrosis factor inhibitors trials. Non-participation is likely to reflect patients' concerns about all intensive treatments and our cautiously worded patient information sheet.
Secondly, disease modifying drugs regimens varied because standardisation was impossible in this context. Thirdly, few trials in rheumatoid arthritis use the health assessment questionnaire as their primary outcome measure. It was relevant in our trial because it measures changes important to patients, is sensitive to improvements with all tested treatments, and helps NICE to determine treatment benefits. Scores on the health assessment questionnaire change less in those with late rheumatoid arthritis, but both groups had clinically important improvements, and the durations of disease in our patients were comparable with those in pivotal biologic trials. Although minimally important differences in scores might be less than 0.22 in routine practice, this lower difference would not change our conclusions.
Fourthly, our trial was un-blinded because blinding is impractical when many different treatments are used. The primary outcome, however, was self completed by patients and not directly influenced by clinicians. Fifthly, our trial lasted only 12 months because when it was designed longer delays in assessing biologic treatments raised ethical concerns.
Finally, the effects of withdrawal and switching treatments need careful consideration. Only a few patients (16/205, 8%) were lost to follow-up and required the imputation of missing data. As the complete case analysis gave similar findings to the intention to treat analysis, it seems unlikely that withdrawals influenced our conclusions. Comparable numbers of patients withdrew from both trial arms, suggesting withdrawals were unlikely to be influenced by the treatment strategies. Overall withdrawal rates were similar to those in comparable rheumatoid arthritis strategy trials involving biologics and disease modifying drugs. The frequencies with which patients discontinued tumour necrosis factor inhibitors were comparable with UK national registry data of routine practice results. The evidence suggests withdrawals in our trial were unlikely to have influenced our main conclusions.
Overall Evidence
In early rheumatoid arthritis three head-to-head trials have shown that combinations of disease modifying drugs achieve similar benefits to tumour necrosis factor inhibitors. RACAT, the only other trial in established active rheumatoid arthritis, showed that triple therapy (methotrexate, sulfasalazine, and hydroxychloroquine) achieved similar benefits to etanercept plus methotrexate in incomplete methotrexate responders treated for 12 months. Our trial and RACAT had comparable results, despite differences in patients and trial design. They highlight the role of head-to-head trials against effective conventional comparators to evaluate expensive treatments. Although toxicity remains a potential concern with combined disease modifying drugs because single trials like ours cannot fully assess risks, such combinations have been used for many years without evidence of major toxicity.
The current trail questions the status quo of use of biologics in patients with active rheumatoid arthritis in whom methotrexate does not achieve the required response. In these patients the clinical and economic evidence from our trial supports preferentially starting combinations of synthetic disease modifying drugs. We also believe that modernising biologic treatment regimens for rheumatoid arthritis requires more strategy trials. Increasing efficacy is important, and optimising cost effectiveness is crucial. Options include giving biologics to patients with rheumatoid arthritis who are likely to respond to specific agents, discontinuing biologics when patients do not fully respond, and tapering treatment when patients achieve sustained remissions.
SHARE
