Tuesday 1 April 2014
Tailor-made methotrexate treatment in juvenile idiopathic arthritis: an established drug revisited
Promotor: Prof. dr A.B.J. Prakken & prof. dr N.M. Wulffraat
Defence: 1 April 2014
Methotrexate (MTX) represented the first major advancement in the treatment of rheumatic diseases. As such MTX gained the leading role in the management of juvenile idiopathic arthritis. Nevertheless, MTX is not sufficiently efficacious in all JIA patients and may lead to adverse effects such as gastrointestinal intolerance, which compromise complete disease control early on during disease course. In such cases, more aggressive therapies with biologicals are given. High efficacy of combination therapies has resulted in applying biologicals early during JIA treatment, before knowing the patients response to MTX with certainty. However, clinicians should be able to determine, before the start or early (at 3 months) during MTX therapy, which patients will be responsive to MTX and will therefore benefit from MTX monotherapy, and which patients will be partially responsive or unresponsive to MTX, thus requiring fast MTX dose escalation or addition of biologicals. To make such tailor-made treatment decisions, clinicians necessitate tools to optimise JIA treatment with MTX and in turn JIA treatment in general. This and other unmet needs of MTX have been addressed in chapter 2 and further elaborated on in the rest of this this.
In part I of this thesis, we investigate various determinants of MTX (non-)response and whether and how they could be used as tools to steer tailor-made therapeutic decisions in JIA patients. In chapter 3, we showed in 287 JIA patients that single nucleotide polymorphisms (SNPs) in genes of MTX efflux transporters ABCB1 and ABCC3 were associated with MTX response, whereas a SNP in a gene of an MTX influx transporter RFC was associated with MTX non-response during one year of treatment. These SNPs could be used as early, fast- and relatively-affordable-to-determine, objective biomarkers of MTX efficacy. However, in order to utilise SNPs in daily clinical practice to gear individualised treatment decisions, we transformed associations of SNPs with MTX efficacy into a prediction model in chapter 4. To predict which patients will be unresponsive to MTX monotherapy before MTX start, we developed a model in 183 JIA patients and validated it 104 patients. The model included erythrocyte sedimentation rate (ESR) and four SNPs in genes encoding for enzymes and transporters of the MTX metabolic pathway – MTRR, PCFT, ABCB1 and ABCC3. The prediction model classified 72% of patients correctly in the derivation cohort and 65% in the validation cohort. The model was converted into a risk score-system, ranging from 0 to 11 points, whereby each risk score carried a certain probability of being an MTX non-responder. We chose a cut-off ≥3 as the optimal score, with a 78% sensitivity and a 49% specificity, as we considered it crucial to adequately treat as many future non-responders as possible with biologicals (high sensitivity), and at the same time attempting to restrict their use as much as possible to those patients who really need them (reasonable specificity). Using the prediction model, patients who are likely to be unresponsive to MTX will receive early additional treatment with biologicals, whereas those responsive to MTX are spared costly biologicals with potentially serious adverse effects. In chapter 5, we investigated the association of MTX polyglutamates (MTX-PGs), measured at 3 months after MTX start, with disease activity (measured using the juvenile arthritis disease activity score (JADAS-27) researched in chapter 6) of 113 JIA patients followed for one year after MTX start. We determined that long-chain MTX-PG3, MTX-PG4, MTX-PG5 and their sum, measured after 3 months of MTX use, were associated with lower disease activity at 3 months and during one year of MTX treatment. The applicability of MTX-PGs as a therapeutic drug monitoring tool for steering individualized treatment decisions can be illustrated in patients who stopped MTX (n=4) and those who received additional medication (n=7), due to insufficient effect. Patients that discontinued MTX had significantly lower MTX doses and lower long-chain MTX-PG concentrations than those who continued MTX. Instead of stopping MTX, these patients may have benefited from MTX dose escalation. On the other hand, patients on additional medication at 6 months had similar MTX doses and MTX-PG concentrations at 3 months, as patients on MTX monotherapy. They remained non-responders, in spite of optimal MTX treatment (reflected by adequate polyglutamation). If timely monitored with TDM, they could have received additional medication earlier than 6 months after MTX start. TDM of MTX-PGs in JIA could guide clinicians to escalate MTX dose in patients with a low polyglutamation rate, and to offer biologicals to patients with insufficient response to MTX with adequate polyglutamation. The upcoming challenge is refinement and validation of the prediction model and MTX-PGs in steering tailor-made therapeutic decisions in a large well-defined JIA cohort followed from the start of MTX treatment, and ultimately their application in daily clinical practice.
In part II, we focus on various facets of the most common MTX-related adverse effects – MTX intolerance – in JIA and rheumatoid/psoriatic arthritis in adults (RA and PsA). In chapter 7, we determined the prevalence of MTX intolerance, which included gastrointestinal symptoms (abdominal pain, nausea and vomiting) occurring after, before (anticipatory) and when thinking of (associative) of MTX intake or injection (the latter two being the conditioned responses to physical symptoms), as well behavioural symptoms such as restlessness and crying when taking MTX. In a cohort of 297 JIA patients, we showed that the prevalence of MTX intolerance, determined using a newly validated Methotrexate Intolerance Severity Score (MISS), reached a high 50.5%. The prevalence of MTX intolerance was 23% higher in patients on parenteral (67.5%) than on oral MTX (44.5%). In chapter 8, we demonstrated that MTX intolerance occurred in RA and PsA patients as well, although to a much lower extent, namely in 11% of 291 patients, suggesting a weaker classic conditioning response in adults than in children taking MTX. Similarly to JIA, however, MTX intolerance prevalence was higher in patients on parenteral (20.8%) than on oral MTX (6.2%). Nevertheless, in order to establish whether parenteral MTX, intrinsically, carries a higher risk of gastrointestinal complaints, patients using parenteral MTX from the treatment start should be compared to those using oral MTX from the start. We conclude that arthritis patients on MTX should be monitored with the MISS, as it allows early detection of MTX intolerance, which could create a window of opportunity for timely treatment of MTX intolerance. In chapter 8, we compared three treatment strategies of MTX intolerance in a randomised controlled trial, specifically: the effect of oral MTX and behavioural therapy (n=15) or parenteral MTX (n=17) compared with the standard of care treatment consisting of oral MTX with an antiemetic (n=16), on MTX intolerance in JIA. Contrary to our hypothesis, behavioural therapy did not target MTX intolerance more successfully than the other two strategies. Unexpectedly, all three treatment strategies had beneficial effects on MTX intolerance. Furthermore, MTX intolerance scores declined in the first week of enrolment, strongly suggesting that participation in the trial and concomitant positive expectations and internal motivation, rather than given treatments, mediated the observed beneficial effects in all three groups. Nevertheless, the beneficial effect continued during the entire follow-up, which could be contributed to the direct effect of treatment strategies as well. We conclude that treatment strategies for MTX intolerance should be tailored individually, based on motivations and expectations of both the patients and their parents. Finally, in chapter 9, we developed and internally validated a prediction model for MTX intolerance (occurring at 6 or 12 months after MTX start) in a cohort of 152 JIA patients. The model consisted of routine clinical variables: JIA subtype, JADAS-27, parent/patient assessment of pain, antinuclear antibody, alanine transaminase, thrombocyte count, creatinine and an interaction term between creatinine and JIA subtype. The model classified 77.5% of patients correctly, and 66.7% after internal validation by bootstrapping, and was transformed into a risk score, ranging from 0 to 17 points. We chose a cut-off ≥6 as the optimal score, at which 82% of intolerant patients were classified correctly (high sensitivity), while maintaining correct classification of 56.1% of tolerant patients (moderate specificity). Using the prediction model, patients at risk of developing MTX intolerance would be identified at MTX start. Furthermore, these patients could be frequently monitored (using the MISS) and their physical gastrointestinal symptoms identified early and treated timely, thus preventing the development of a classical conditioning response and hence the development of MTX intolerance.
In part III, chapter 10, we focus on quantitative and qualitative effects of MTX treatment on the key players of autoimmune inflammation in JIA – the regulatory (Treg) and effector (Teff) T cells – at MTX start and at 3 and 6 months upon MTX start. We showed that Treg phenotype and suppressive function on Teff proliferation and cytokine production were not affected by MTX treatment. Contrary to our hypothesis that MTX would have an anti-proliferative effect on Teff proliferation, we determined that neither in vitro exposure to low nanomolar concentrations of MTX, corresponding to low-dose MTX treatment, nor the low-dose MTX treatment itself inhibited proliferation of T cells ex vivo and upon T cell receptor stimulation. Instead, Teff proliferation at 6 months was enhanced in both responders and non-responders, suggesting that T cell proliferation could be directly affected by MTX, rather than by clinical improvement. We demonstrated that increased concentrations of IFNγ, but not of pro-inflammatory IL-6 and TNFα, in plasma of JIA patients after 6 months, which paralleled enhanced proliferation of T cells. Increased plasma levels of IFNγ, in concert with increased T cell proliferation indicate that MTX does not attenuate but rather enhances the effector T cell function in JIA patients during MTX treatment. This immunological data is contrary to the common belief that low-dose MTX treatment in rheumatic diseases has immunosuppressive properties.
The data described in this thesis shows that MTX treatment can and should be optimised and tailored to individual patients by using tools such as prediction models and/or therapeutic drug monitoring to determine which patients will be responsive to MTX only, and which patients will need more aggressive treatment with biologicals. Also, MTX treatment should be optimised by predicting and treating MTX intolerance, which occurs in over one half of JIA patients. In order to optimise JIA treatment in general, optimising treatment with MTX, first, is warranted. Future studies should focus on validation of the abovementioned tools for steering tailor-made therapeutic decisions and on their refinement with novel genetic and immunological biomarkers in large international JIA cohorts, followed by their application in daily clinical practice.