Methotrexate in psoriasis under real‐world conditions: long‐term efficacy and tolerability

Dear Editor, Methotrexate is a mainstay in psoriasis treatment. Yet the evidence base for this treatment is sparse. Recent evidence has accumulated with methotrexate serving as a comparator treatment for other biologics.1 2 3 4 5 However, clinical studies do not directly reflect long‐term drug performance in routine care.6 Capturing treatment outcomes under real‐world conditions is notoriously difficult due to selection bias, reporting bias and incomplete outcome documentation. Here we summarize the results of an observational study that minimizes these limitations due to a unique set of factors, as follows. Firstly, all drug prescribing, lab reports and inpatient admissions are electronically captured for a population of approximately 420 000 in Perthshire, Tayside and North‐East Fife, Scotland. Secondly, all patients requiring systemic psoriasis treatment are managed, and electronically captured, by a single department. Thirdly, the population is marked by low mobility, and treatment outside the National Health Service is minimal. Thus we were able to retrieve a near‐complete record of all treatment episodes during an observational period exceeding 10 years, encompassing 915 patient‐treatment years (333 patients, 401 treatment episodes, median duration 33 months). The patients' clinical characteristics are given in Table 1. Additional methods and results, and Tables S1–6 and Figures S1–4 containing further information, are provided as Supporting Information. Clinical characteristics of the Tayside real‐world methotrexate (MTX) treatment cohort The data shown summarize the clinical profile accessible through electronic data mining as detailed in the methods. aFive of the 333 patients were aged < 18 years. bTopicals include agents containing steroids, calcipotriol, coal tar, dithranol, retinoids and salicylic acid, but not emollients. cDefined as continuous systemic medical treatment for > 6 months with a British National Formulary code drug for antidepressant therapy prior to starting MTX. dTwenty‐nine of the patients with diabetes had type 2 diabetes, one patient had type 1 diabetes and one patient had impaired glucose tolerance. Clinical characteristics of the Tayside real‐world methotrexate (MTX) treatment cohort The data shown summarize the clinical profile accessible through electronic data mining as detailed in the methods. aFive of the 333 patients were aged < 18 years. bTopicals include agents containing steroids, calcipotriol, coal tar, dithranol, retinoids and salicylic acid, but not emollients. cDefined as continuous systemic medical treatment for > 6 months with a British National Formulary code drug for antidepressant therapy prior to starting MTX. dTwenty‐nine of the patients with diabetes had type 2 diabetes, one patient had type 1 diabetes and one patient had impaired glucose tolerance. As treatment outcomes (e.g. Psoriasis Area and Severity Index and Physician's Global Assessment scores) are not generally documented in routine care, we initially asked whether treatment duration as such could serve as a proxy for efficacy, assuming that treatment is unlikely to continue in the absence of any benefit. In order to identify candidate time points to serve as a threshold marking ‘success’, we first analysed the treatment duration in all patients who had experienced failure due to either inefficacy or adverse effects. As shown in Figure 1a, the majority of failures occurred during the first year of treatment. Once beyond this point, patients are likely to remain on treatment long term. We therefore sought to validate this cut‐off point as an indicator of ‘success’ using independent outcomes. Indeed, patients remaining on methotrexate beyond 12 months displayed significantly reduced need for treatment with all classes of topical agents, phototherapy and psoralen–ultraviolet A. Furthermore, after 12 months there was less inpatient treatment compared with ‘failures’ (Fig. 2). These data suggest that a treatment duration of > 12 months is a reasonable surrogate efficacy outcome marking ‘success’. (a) Rate of treatment discontinuation in patients with psoriasis treated with methotrexate under real‐world conditions plotted for up to ten years. (b) Rate of treatment discontinuation in patients with psoriasis treated with methotrexate under real‐world conditions plotted at higher temporal resolution for up the initial three years. The data shown represent patients discontinued due to either lack of efficacy (dashed line), or adverse effects (AEs: dotted line), or both cumulated (solid line). Data from patients discontinued for reasons unrelated to AEs or efficacy (e.g. family planning) were excluded from the analysis. The data shown represent individual treatment episodes in n=364 patients at baseline. The requirement for psoriasis treatments in patients achieving 1 year of methotrexate duration vs. patients discontinued prior to 1 year of treatment. (a) The percentage of patients who received a topical treatment, as indicated in the figure, in the 12‐month interval before baseline and the 12‐month interval beginning 1 year after baseline. Dark shaded columns, patients treated > 1 year; light shaded columns, patients discontinued at < 1 year. (b) The percentage of patients – labelled as in (a) – who received an increased or decreased number of scripts for each of the topical treatments listed in the figure in the 12‐month interval before baseline or the 12‐month interval starting 1 year after baseline. (c) Use of phototherapy treatments in patients achieving 1 year of methotrexate treatment vs. patients discontinued during the first year of treatment. UVB, ultraviolet B; PUVA, psoralen–ultraviolet A. (d) The percentage of patients requiring inpatient treatment for psoriasis in the 12‐month interval before baseline vs. the 12‐month interval starting 1 year after methotrexate treatment initiation. **P < 0·001. A higher‐resolution temporal analysis of efficacy furthermore revealed that early (up to 9 months’ treatment) and secondary failure due to lack of efficacy (Fig.1b, arrowhead) occur at an almost equal frequency of approximately 6·0%. Beyond 24 months, lack of efficacy is rare (Fig.1b). We next compared the treatment‐limiting adverse events (AEs) occurring under real‐world conditions with those published in clinical trials, the latter being available only for short‐term treatment duration (Table 2). The overall spectrum of limiting AEs is comparable between clinical trials and routine care. Of note, ‘fatigue’ is reported only under real‐world conditions, among both short‐term and delayed‐onset AEs (Table 2). Although this lacks a precise definition, the overall incidence of this AE appears too high to discount it altogether, being a limiting factor in 1–2% of patients (Table S1). Possibly, the lack of a standardized definition complicates capture of this AE in clinical trials. Treatment‐limiting adverse events in methotrexate (MTX) treatment in clinical trial vs. real‐world conditions The data shown are the absolute numbers and incidences of each adverse event for all treatment‐limiting adverse events documented in both cohorts. The complete dataset for limiting adverse events observed in only one cohort is listed in Table S1 (see Supporting Information). LFT, liver function test. aData derived from a recent meta‐analysis of published clinical trial data on MTX in psoriasis (West et al., in press). b’Short term’ indicates treatment duration < 12 months. The precise median observational period documented in published clinical trials is specified in Table S1. cCases of ‘infection’ in the short‐term and clinical trial data are exclusively pneumonia. Treatment‐limiting adverse events in methotrexate (MTX) treatment in clinical trial vs. real‐world conditions The data shown are the absolute numbers and incidences of each adverse event for all treatment‐limiting adverse events documented in both cohorts. The complete dataset for limiting adverse events observed in only one cohort is listed in Table S1 (see Supporting Information). LFT, liver function test. aData derived from a recent meta‐analysis of published clinical trial data on MTX in psoriasis (West et al., in press). b’Short term’ indicates treatment duration < 12 months. The precise median observational period documented in published clinical trials is specified in Table S1. cCases of ‘infection’ in the short‐term and clinical trial data are exclusively pneumonia. The relative importance of the top two limiting AEs, hepatotoxicity and nausea/vomiting, increases over time (Fig. S2). However, an analysis of the cumulative incidence of all AEs over the entire treatment duration also shows that the top AEs (nausea/vomiting, leucopenia, liver function test aberrations) are in fact nonlimiting in most patients (Table S1). Of note, we also observed that nausea/vomiting as a limiting factor could be overcome by switching to subcutaneous injection in half of the affected patients (see the Results in the Supporting Information). This suggests that discarding methotrexate treatment due to this side‐effect occurring after oral dosing may be premature. In fact, subcutaneous treatment allowed stable dosing in nine of 27 patients who had previously failed oral methotrexate (Table S6, Fig. S4), underscoring the importance of exploring this approach before discarding methotrexate as a treatment option. We next analysed treatment outcomes stratified according to starting dose (Table S2). Efficacy was very similar between the different starting doses administered. The data show that a uniform starting dose of 10 mg once weekly would not be expected to affect performance adversely, but would allow for standardization of protocols. We then asked whether the degree of disease control is affected by the dose administered at steady state. We found no significant difference in the percentage of patients achieving a reduction in topical treatment prescriptions between any of the dose ranges administered. Likewise, no dose range was over‐ or under‐represented at steady state at any point in time (Fig. S4). Thus, no single dose range is associated with a higher level of long‐term disease control, suggesting that – in the absence of limiting AEs – efficacy may occur at any dose up to 25 mg in any given patient. Perhaps the overall most striking finding is that, at least in this cohort, lack of efficacy as a limiting factor occurs much less frequently than commonly assumed, in only approximately 10% of patients. The most frequent reason to stop methotrexate treatment is not lack of efficacy but limiting AEs. The present study has some important limitations. Firstly, it lacks a placebo control. Moreover, it is impossible directly to relate the reported outcomes to Psoriasis Area and Severity Index or Physician's Global Assessment scores. In that regard, perhaps the biggest obstacle in real‐world studies is the lack of systematically recorded outcomes, including both AE reporting and, critically, efficacy as such. To address the latter, we here show that ‘stable treatment on methotrexate > 12 months’ is a proxy for ‘success’, as validated by a series of independent objective outcomes (topical treatments, ultraviolet treatment, inpatient treatment). This quantity should be accessible in most departments, thus providing an effective indirect measure of treatment performance. In the present cohort, the outcomes consistently demonstrate that approximately two‐thirds of patients with psoriasis benefit from methotrexate. The clinical profile of our cohort and its main comorbidities (hypertension, diabetes, depression) mirrors those reported for the U.K.‐wide sample in the British Association of Dermatologists’ Biologic Interventions Register,7 thus rendering our data informative for advising patients on the probabilities of treatment efficacy, as well as the specific AEs. In addition, the outcomes reported here provide an initial comparator for local prospective audit of methotrexate performance to ensure that this inexpensive treatment is not underused. Funding sources: no external funding. Conflicts of interest: none declared. Table S1. Cumulative long‐term incidence of the adverse‐effect spectrum in methotrexate treatment. Table S2. Primary treatment outcome stratified by range of starting dose. Table S3. Steady‐state dose range observed in patients on long‐term stable treatment. Table S4. Initial dose of methotrexate by route of administration. Table S5. All treatment‐limiting adverse events in Tayside vs. published cohorts. Table S6. Treatment outcomes for subcutaneous methotrexate administration in patients previously having failed oral methotrexate. Fig S1. Rate of treatment discontinuation due to lack of efficacy or side‐effects. Fig S2. Spectrum of treatment‐limiting adverse events for methotrexate in psoriasis. Fig S3. Treatment outcome broken down by dose range on stable treatment. Fig S4. Documented reasons for switching from oral to either subcutaneous or intramuscular dosing of methotrexate.