stopping tumor necrosis factor inhibitors in well controlled rheumatoid arthritis patients: the poet study
Marjan Ghiti Moghadam is a PhD student in the Department of Psychology, Health & Technology. Her supervisor is prof.dr. M.A.F.J. van de Laar from the Faculty of Behavioural, Management and Social sciences (BMS).
With the turn of the centurty, the treatment of patients with Rheumatoid Arthritis (RA) started to change from care to a situation where cure has become a realistic option. Basis for this change is the development and introduction of inhibitors of Tumor Necrosis Factor (TNFi) and other targets thereafter. In hind sight, the efficacy of these new class of pharmacological agents: referred to as biologicals, proofed to be better than expected, changing the perspective of many RA patients and patients suffering from other autoimmune diseases. Typically during the development and introduction of TNFi the attention was focused on efficacy and safety. Drug tapering or even more stopping, in case of favorable efficacy, was not on our minds, The paradigm that auto-immune diseases, like RA, are chronic conditions that need live long treatments, and in best cases progression could be retarded, continued to dominate the thinking of healthprofessionals and patients. We were confronted with the need to stop or discontinue TNFi, in cases where the anticipate increased risk for infections became reality (1). Guidelines to decrease the infectious risk related to the inhibition of TNF were developed and successfully implemented (biobadaser TNF en TBC) (2). Although scarcely evaluated, stopping TNFi in case of infection or increased risk for infection (elective surgery) the clinical impression was that flares occurred following stopping where restarting TNFi resulted in quick regained response (3). Studies on remission induction (4), later referred to as treating to the target of remission (t2T) (5), was a direct consequence of the impressive clincal efficacy of the introduction on biologicals ic TNFi. The high percentage of RA patients in real life studies, that were able to reach stable remission by using the therapeutic options in combination with the increasing drug costs and concern on lifelong treatments paved the way for studies on drug tapering or stopping TNFi in case of excellent clinical response (remission of low disease activity). Although scientific arguments, like immunogenicity (6) against drug tapering outweigh the arguments against immediate stopping, psychological arguments form patients as well as health professionals lead the majority of studies started with drug tapering strategies. The majority of studies are investigator initiated and for obvious reasons small and subjected to methodological limitations. The POET study, that is the basis for my thesis, was initiated by Dutch Rheumatologists and Health Policy Makers. The combined Dutch effort resulted in one of the largest studies on the subject of dose optimalization of biologicals in RA in this case by stopping. Although this study also has its limitations ic the open label design and the pragmatic randomized design, the impressive number of participating patients enabled us to address not only the primary study question but also to explore additional questions like mechanisms involved and impact for patients and society. In the first chapter of my thesis I discussed existing information on stopping and tapering of biologicals in RA as well as a theoretical framework potentially explaining underlying mechanisms.
This chapter is the primary and most important publication of the POET study, it was my privilege, on behalf of the POET team to show that stopping TNFi is possible in 48% of RA patients in sustained remission or LDA. However, this strategy results in more than threefold increase signs and symptoms of increased disease activity (flare) within 12 months as compared to patients who continued TNFi. We also showed that mean disease activity in the stop group was significantly increased throughout the follow-up period but in most patients it remained below the threshold for moderate disease activity. With regard to the fact that different flare definitions are available, we used sensitivity analyses using other definitions of flare, which showed comparable results. Our results confirm findings of several, mostly smaller studies: Preceding studies on stopping TNFi in RA demonstrated a wide range of flare risk of 18-88%. Difference in results is possibly explained by heterogeneity in RA population, definitions of flare, disease activity at inclusion, and use of concomitant csDMARDs. RCTs reporting the results of stopping specific TNFi report the following results. The US Corrona registry reported that 73.4% of RA patients experienced no flare for more than 12 months after stopping their first TNFi (7). The EMPIRE trial, PRESERVE trial, and DOSERA study reported that respectively 57.7%, 42,6% and 13% of RA patients successful stopping etanercept (8,9,10); The OPTIMA trial, HONOR and BRIGHT studies reported that respectively 56%, 58% and 18%, of RA patients remained in remission or LDA after stopping adalimumab (11,12,13). Stopping infliximab by Quinn et al, the BeSt study (after falling DMARD) and RRR reported persistent LDA of remission in 70%, 25% and 50% respectievely (14,15,16). Finaly a small study on stopping certolizumab pegol reported absence of any success (17). In our POET study, in case of flare after stopping TNFi, we suggested restarting TNFi. The observed data show that treatment success could be realized in the majority of patients (83.1% success after restarting TNFi, most of the time within 12 weeks). Moreover, there were no notable (serious) adverse events after restarting TNFi. Finely we noticed more hospitalization in the stop group than continuation group, but most of them were not related to the stopping TNFi. Reasons for hospitalization that were observed are infections, elective surgery, malignancies or fractures. We concluded that stopping TNFi in RA patients and restarting TNFi in case of flare seem to be safe. In addition to the stopping TNFi, there are multiple stategies for dose reduction also referred to as dose optimalization. For obvious reasons the variability to do so exceeds the options for direct stopping like we did in the POET study. The PRESERVA-, DOSERA- and DRESS study used different strategies to taper than STRASS- en RETRO study (9,10,18,19,20). In general these studies showed that less flare occured during dose tapering compared with stopping TNFi, but overall comparable long term effects for patients were seen. The POET study only addressed stopping TNFi. Other studies examined the effects of TNFi dose reduction. Some trials compared etanercept reduction versus stopping or continuation and concluded that reduced dosing generally resulted in an increased disease activity and or flares. Although the risk of flares during doses reduction was not to the extend of immediate stopping, the cumulative risk of al consecutive reductions is comparable with the flare risk of stopping immediately (21-24). More recently, in the DRESS study showed that disease activity-guided dose reduction of adalimumab or etanercept was not inferior to continuation with respect to flares (18), comparable to our findings.
In this chapter we search for predictors for successful discontinuation of TNFi. This chapter serves the theoretical frame work discussed in chapter one. Our analyses showed that 70% of RA patients in sustained remission or stable LDA, who used a TNFi of the antibody class, with a relative short disease duration of <10 years and a MBDA score ≤44 were most likely be able to stop their TNFi without a subsequent flare. In contrast to patients on a TNFi receptor antagonist, with a disease duration of ≥10 years and a MBDA score >44 who 80% needed to restart TNFi because of experiencing a flare. The POET observation that relative short disease duration was related to stopping without flare was in line with the the study of Kavanaugh et al., it was reported that successful stopping bDMARDs in RA patients with early disease (25). However Kuiper et al. showed that early RA patients had no lower flare rate than stablished RA after stopping bDMARDs (26). The latter might be explained by the small number of patients. In our study, RF positivity, non-smoking and normal BMI were not found to be predictors of successful discontinuation. In contrast with the data from Corrona registry and RRR study who reported these patients characteristics as a predictor for successful TNFi discontinuation (7,16). A possible explanation for the absence of a relation with RF and aCCP might be the high percentage of double positive patients in our study. Also signs and symptoms related to residual disease activity at the time of discontinuation seems to be a predictor for flare after stopping TNFi (13, 16, 27). the association with a higher MBDA score at baseline in our POET study is in line with this hypothesis. We demonstrated that the type of TNFi could be a predictor. Our analyses showed that patients using TNF antibodies (mostly adalimumab) were significantly more often able to successfully discontinue their TNFi than patients who had been using TNF receptor antagonists (mostly etanercept). TNF antibodies seems to induced prolonged disease control due to langer half-life, lysing surface TNF expressing blood cells and block TNF interaction with cell surface TNF receptors. This is in line with the recent publication of M Hashimoto et al. (28). Taken the information of the analysis presented in chapter 3 together with the theoretical framework presented in chapter 1 we found support for a role for the window of opportunity, residual disease activity and mode of TNF-inhibitor ic antibodies as compared to receptor antagonist. Obviously these post hoc analysis can not confirm the theory but are in line with the underlying hypothesis.
The multi-biomarker disease activity (MBDA) blood test measures 12 serum proteins to produce a score that has been clinically validated as a measure of disease activity in patients with RA (29-31). MBDA scores have been shown to reflect current clinical disease activity and changes in disease activity over time, including treatment responses in RA patients treated with TNFi (32). In this chapter we focused on the MBDA score as a predictor of flare after discontinuation of TNFi treatment. We used the data from patients who randomized to the stop group. In this group, the baseline serum samples were available in majority of patients to measure MBDA scores. We showed that patients with a high MBDA score at the time of TNFi discontinuation had significantly more flare in follow-up during 12 months. The data suggested that a high MBDA score, although the patient is in remission or a state of LDA, could be a sign of residual disease activity. Several previous studies have found elevated MBDA score in RA patients with remission of LDA (33-36). We reported for the first time, to the best of our knowledge, the value of the MBDA score as a predictor for flare in RA patients who discontinued TNFi treatment. Considering the three different predictor criteria for relapse (TNFi restart, medication escalation, and physician-reported flare), in our study a high MBDA score was an independent predictor of all three potential predictors but it did not remain associated when adjusting for all other potential predictors, including DAS28 score. Although the heterogenicity of the disease rheumatoid arthritis and consequently the difficulty of assessing disease activity is well known, laboratory based methods like the MBDA, up to present, are not able to overcome the limitations of composite disease activity scores. The results of the POET study do confirm that there is a relation between RA disease activity and MBDA scores but this weak relation probably does not allow the costs of the test (37).
Apart from the biomedical perspective of the disease the perspective of the patient is by intention reflecting the consequences of the disease for the health status of the patient. Therefore, in the POET study we addressed in addition to the outcomes in the biomedical perspective the impact for the patient as wells. We demonstrated that stopping TNFi had a significant but short-term negative impact on patient-reported outcomes (PROs) such as physical and mental health status, health utility, pain, disability and fatigue. There are only a few studies in which the effect of stopping TNFi on burden of disease were analyzed. The randomized controlled trial PRESERVE study showed after 12 months a better score on general health, pain, disability, health utility, sleep, and fatigue in patients who continued etanercept plus methotrexate than patients who stopped etanercept and used methotrexate only (9). In addition, the observational RRR study showed that after one-year the mean HAQ-DI scores were significantly lower in patients who remained in remission after stopping infliximab versus those that flared (16). This result was not supported by OPTIMA – and ADMIRE-studies (11,38). Also the result of the open-label nonrandomized HONOR study did not show a significant difference HAQ-DI scores after one year adalimumab discontinuation between patients that sustain in remission and patients who flared (12). In a very small observational study of Brocq et al, in 21 patients, they did not found significant differences in one-year HAQ-DI scores between stopping of TNFi (39). In general the negative biomedical effects of stopping TNFi in RA patients seem to be reflected in unfavorable Patient Reported Outcomes. However in line with the handling with care design of the different studies the negative outcomes are small and of short duration not impacting the long-term health of patients trying to stop or tapper theirexpensive treatments.
Since drug cost was one of the important arguments for the discussion on stopping biologicals as a cost analysis of stopping versus continuing TNFi in case of good disease control ( remission or LDA, as defined by DA28) was almost mandatory. We demonstrated that stopping TNFi in this group of RA patients can save cost but comes with a small loss of QALY. The mean saved cost was €368,269 per QALY lost. We showed that, except for the first follow-up point after baseline, the mean health utility did not significantly differ over time between patients in the stop and continuation groups. We suggest that stopping TNFi in well controlled RA patients comes with considerable costs saved and temporarily limited loss of quality of life. In the, also Dutch, tapering study, the authors showed that guided down titration of TNFi could save €390,493 for each QALY lost (40) remarkably compared with discontinuation of TNFi in our stop study. In another cost analysis based on the STRASS study, the authors reported that step-down strategy results in saving 53,417 euros per QALY lost (41). In the PRESERVE study, in which dose reduction was studied in the case of remission, Kobelt et al. reported the cost per QALY gained for the half-dose strategy versus stopping etanercept varies between €14,000 and €29,000, depending on the time frame. The author suggested that a dose reduction is the most advantageous strategy in patients with moderate RA, although cost advantage of stopping TNFi was not separately reported (42). Some studies only reported dug cost saved: Van der Maas et al. reported a mean cost reduction of 3474 euros per patient due to infliximab tapering (43). A comparable result has been shown by Murphy et al. in a cohort of 79 patients with rheumatoid arthritis, psoriatic arthritis, and ankylosing spondylitis. They reported a mean saving of 3800 euros per patient per year after dose reduction of etanercept or adalimumab (44).
In this chapter we did a post-hoc analyses of data to identify the most reliable measurement instruments for disease exacerbation. We showed that the composite disease activity indices, physician and patient global assessment of disease activity and pain could be used for measuring disease exacerbation and flares in RA. Previous studies have reported pain as most important treatment priority for patients (45). In other studies the physical function was found to be one of the best performing indicators of treatment benefits (46-51). So, the poor performance of physical function in our analyses contrasts with previous studies. In our analyses the CDAI, SDAI and DAS28 performed equivalently as measures of disease exacerbation. Similar results were observed for these three scores in two previous studies (52,53). Acute-phase reactants in our study seems to give a little additional information on disease activity exacerbation. We suggested that fatigue, emotional wellbeing or participation to a composite score may not contribute much to its reliability to predict flare. Previous studies also found limited responsiveness for participation and emotional wellbeing, But the value of fatigue reported different (54-59).
Because of the potentially risks but especially because of increasing costs of longterm TNFi treatment, it was considered important to investigate whether established RA patients with sustained well controlled disease activity, defined as DAS28 remission or low disease activity can safely and effectively stop their therapy with TNFi. The primary study result is that within 12 months after stopping TNFi compared to those who continued TNFi experience a threefold increased risk of signs and symptoms of increased disease activity (flare). Of course stopping antirheumatic therapy in this case TNFi can only be considered in patients in whom the treatment goal is reached. It does not need to be advocated that In case treatment goals are not reached treatment escalation is indicated. We do realize that an intense discussion on the definition of remission is ongoing. Our result support the need of a definition of remission without residual disease activity, although operationalization in a global applicable definition remains a challenge beyond the goals of this thesis. The definition of increased disease activity (flare) differs between studies, we did do sensitivity analysis using different definition of flare all coming to the same conclusion. Taking into account that RA is a live long disease discontinuation of successful antirheumatic therapy remains a risk. Successful stopping of TNFi seems to be possible for a subgroup of RA patients but comes with an increased risk of flare that in the POET study and other studies was observed to be manageable. It is important to find out which group of patients have more risk of flare after discontinuation. We showed that discontinuation of TNFi does not have long-term consequences for the patients-related outcomes. Finely, Our analysis shows that stopping TNFis in RA patients with stable controlled disease is cost effective but results in a small QALY loss. In conclusion, discontinuation of TNFi in well controlled RA patients is a realistic option Obviously, the decision to try to stop or taper TNFi must be a shared decision between rheumatologist and the patient.