MATERIALS AND METHODS
Data source
We used the 2003~2016 KNHIS database. The KNHIS database contains longitudinal patient data including demographics, International Classification of Diseases Tenth Revision (ICD10) diagnosis codes, procedures, prescription records (drug names, prescription and dispensing dates, days’ supply, dose, and route of administration), and type of medical utilization (outpatient, inpatient, or emergency department) of all Korean citizens. The Institutional Review Board of the Seoul National University Bundang Hospital approved the study protocol (X-1704-393-903) and waived the need for patient consent based on deidentified database. This study was conducted in accordance with the principles of the Declaration of Helsinki and Good Clinical Practice Guidelines.
Study population
We identified patients aged ≥18 years with an ICD10 code of seropositive RA (M05.x) who initiated bDMARDs (infliximab, adalimumab, golimumab, etanercept, abatacept, rituximab, or tocilizumab) (Figure 1) [9]. This RA identification algorithm has been validated to have a predictive positive value of more than 85%. Then, we applied a V-code for seropositive RA, which ensures that the patient is a beneficiary of 90% reimbursement of the drug cost from the Korean government. Requirement of seropositivity is relevant in that the use of bDMARDs among RA patients is largely limited to seropositive cases in Korea due to reimbursement criteria. Patients who had dialysis services or human immunodeficiency virus infection were excluded since their DMARD use is expected to substantially differ from that of RA population in general.
To define initiators of bDMARDs, patients were required to be free of any bDMARDs for at least 12 months prior to the first dispensing date (=index date) of a bDMARD they initiated. Among patients who satisfied the above RA identification algorithm [9], the RA diagnosis date was defined as the earliest date among DMARD dispensing dates, free of any DMARD use for ≥365 days before the RA diagnosis date.
Outcomes
We defined the following four items of optimal treatment [3,4]: 1) use of MTX as part of the initial treatment, 2) MTX dose escalation of ≥15 mg/week within 1-year post-diagnosis, 3) low dose glucocorticoid use (≤7.5 mg/day of prednisone-equivalent dose), and 4) glucocorticoid discontinuation within 6 months of treatment. Glucocorticoids were considered discontinued at 6 months if the last available dates of glucocorticoids did not exceed 180-days from RA diagnoses, namely, the first csDMARD dispensing dates, and free of further dispensing for at least 90 days.
While the international guideline such as EULAR recommendations do not recommend a step-up from MTX monotherapy to csDMARD combination before bDMARD initiation [3,4], the Health Insurance Review and Assessment Service (HIRA) of Korea require as reimbursement conditions to use MTX plus other csDMARD before initiating bDMARDs, based on previous studies showing non-inferiority of such combination therapy compared to bDMARD use [10-13]. Since the HIRA reimbursement criteria are expected to impact treatment patterns of pre-index period, we looked at csDMARD combination patterns in addition to the four proxies listed above. The csDMARDs examined regarding combination were MTX, sulfasalazine, hydroxychloroquine, leflunomide, tacrolimus, azathioprine, cyclophosphamide, cyclosporine, bucillamine, and D-penicillamine.
Patient comorbidities
We obtained information on comorbidities including cardiovascular, metabolic, pulmonary, gastrointestinal, renal, infectious, and malignancies, together with Charlson-Deyo comorbidity score (Table 1) [14].
Statistical analysis
Variables were expressed as mean±standard deviation (SD) or numbers (%). To assess longitudinal trends, we used the Cochran-Armitage test for categorical variables and general linear models for continuous variables [15,16].
To identify predictors associated with MTX dose of ≥15 mg/week within 1-year post-diagnosis and glucocorticoid-free status within 6 months of treatment, we conducted a multivariable logistic regression analysis, providing odds ratios (ORs) and 95% confidence intervals (CIs). Regression models included age at diagnosis, gender, index year, csDMARD treatments during 1-year after diagnosis, and 1-year pre-RA comorbidities. All analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA). p<0.05 was considered statistically significant.
RESULTS
Characteristics of the study population
Figure 1 shows our study population selection process. Among 10,144 bDMARD initiators, 6,986 with identifiable RA diagnosis dates were included in the study. Demographics, distribution of index years, and pre-RA comorbidities of study participants are summarized in Table 1. The mean±SD age at the index date was 56.5±10.6 years. The majority (78.2%) of subjects were females. The mean±SD interval from RA diagnosis to bDMARD initiation was 3.9±3.2 years.
Treatment patterns and proxy fulfillments within 1-year post-diagnosis
Since the international recommendations primarily focus on initial treatment strategies after RA diagnosis [3,4], items were examined for the 1-year period after RA diagnosis or prior to bDMARD initiation, whichever came first (Table 2).
MTX was used as part of an initial treatment strategy among 54.9% of RA patients. The mean±SD initial dose of MTX was 9.2±10.1 mg/week. Within 1-year after RA diagnosis, 85.2% of patients eventually used MTX. Among them, 48.9% ever tried a dose beyond 15 mg/week. The mean±SD maximal dose of MTX was 14.1±13.1 mg/week (folate supplementation in more than 99% of MTX users). Parenteral MTX use was seen in 2.8% over this period.
Glucocorticoids were used in 75.2% of RA patients as part of an initial treatment. Only a minority of glucocorticoid users tapered off the drug, with 64.5% and 61.9% of RA patients being still on glucocorticoids at 6 months and 12 months, respectively. 79.5% of all glucocorticoid users received prednisone-equivalent doses of ≤7.5 mg/day during the 1-year post-diagnosis period. The mean daily dose of glucocorticoids was 5.5±4.2 mg.
csDMARD combination as an initial treatment strategy was seen in 47.2% of patients, with 83.0% of them using MTX. Eventually, 85.7% of RA patients used csDMARD combination. The mean±SD time from RA diagnosis to csDMARD combination was 188±479 days.
Treatment patterns and MTX dose escalation fulfillment during the 1-year pre-index period
We looked for treatment patterns including MTX dose escalation during the 1-year pre-index period before initiating bDMARDs, when RA activity would have been highly active (Table 2).
Compliant with the HIRA requirements of Korea, 90.7% of patients used MTX (folate supplementation in more than 98% of MTX users) during this period. However, only 52.6% of them ever reached a maximal dose of ≥15 mg/week. Less than 5% used parenteral MTX. We observed a high rate of glucocorticoid use (98.0%), and csDMARD combination (98.7%), and use of potent csDMARDs (leflunomide in 53.0%, and tacrolimus in 21.1%) [17,18], reflecting high disease activity before initiating bDMARDs during this period. The mean daily prednisone-equivalent dose of glucocorticoids was 5.5±3.7 mg.
Longitudinal trends in treatment patterns of csDMARDs
We looked for longitudinal changes in annual csDMARD use for 1-year post-diagnosis (Table 3), and 1-year pre-index periods (Supplementary Table 1). Highly prevalent use (82.5%~91.2%) of MTX within 1-year after RA diagnosis was observed across all calendar years. However, the proportion of MTX users with an initial or maximal dose of ≥15 mg/week had significantly decreased over the years (p for trend <0.001). Unlike such a declining trend of MTX dose, the use of tacrolimus gradually has increased (p for trend <0.001). The csDMARD combination rate within 1-year of diagnosis was consistently high (>85%) over time despite a trend of decrease (p=0.003).
During the pre-index period, MTX use (at least 85% of RA patients) and csDMARD combination (at least 97.9% of RA patients) were consistently high (Supplementary Table 1). Similar to the patterns of the 1-year post-diagnosis period, the proportion of MTX users with a maximal dose of ≥15 mg/week had significantly decreased over time (p for trend<0.001) while the use of leflunomide (p for trend=0.021) and tacrolimus has increased (p for trend<0.001).
Predictors associated with proxy fulfillments
The use of pre-biologic MTX ≥15 mg/week within 1-year post-diagnosis was associated with younger age at RA diagnosis (OR 1.02 per 1-year decrease, 95% CI 1.01~1.02), male gender (OR 1.27, 95% CI 1.12~1.45), combination therapy (OR 3.74, 95% CI 3.02~4.62), glucocorticoid use (OR 2.50, 95% CI 1.74~3.59), and cyclooxygenase-2 inhibitor (OR 1.27, 95% CI 1.14~1.41) during this period (Table 4). Glucocorticoid discontinuation within 6 months from treatment was associated with male gender (OR 1.34, 95% CI 1.12~1.60), maximal MTX dose ≥15 mg/week (OR 1.63, 95% 1.40~1.89), combination therapy (OR 1.34, 95% CI 1.01~1.79), and pre-RA hospitalization (OR 1.30, 95% CI 1.11~1.53) (Table 4). Among comorbidities from the 1-year post-diagnosis period, chronic viral hepatitis B or C was associated with both MTX use of ≥15 mg/week and glucocorticoid discontinuation. The Charlson-Deyo comorbidity score was not associated with these proxies.
DISCUSSION
In this population-based nationwide study, we found that approximately 13% of RA population ever used bDMARD with a mean interval of 3.9 years from RA diagnosis to bDMARD initiation. We observed that the DMARD combination and glucocorticoids were the main treatment strategy before using bDMARD in most of the RA cases. However, MTX was chosen as the 1st csDMARD in only half of newly diagnosed RA patients and MTX dose escalation (≥15 mg/week) was observed among only half of its users. The underuse of MTX as the 1st starting DMARD among RA patients is in line with prescription patterns of other countries where 40%~50% of newly RA patients initiate MTX as the 1st agent [7,19]. Glucocorticoid discontinuation was achieved in a minority (less than 10% of glucocorticoid users) of the patients, within a dose of ≤7.5 mg/day. These treatment patterns were different from the international guidelines [3,4]. Instead, we observed a high compliance to HIRA reimbursement criteria, showing MTX-based csDMARD combination before bDMARD initiation. Trend analyses showed annual decrease of initial and maximal doses of MTX with increased use of alternative drugs such as tacrolimus and leflunomide. Patient age and gender, concomitant treatments, and comorbidity status were associated with MTX dose of ≥15 mg/week within the 1-year post-diagnosis period and glucocorticoid discontinuation within 6 months.
The proportion of patients who used bDMARD in Korea (12.6%, Figure 1) was relatively low compared to that in other countries having no prescription regulations or more generous reimbursement criteria such as USA (48.6%), France (60.2%), and Japan (50.5%) [20]. The proportion however was similar to that in the UK where a similar reimbursement threshold of Disease Activity Score 28-erythrocyte sedimentation rate of ≥5.1 despite csDMARD treatment has been implemented. In addition, the interval from RA diagnosis to bDMARD initiation of 3.9 years was much longer than recommended by the guidelines in case of not achieving low disease activity or remission within 3~6 months of treatment [3,4]. Such delayed bDMARD use may lead to loss of optimal treatment window for remission induction.
Previous studies described an optimal MTX use involves MTX use as the 1st csDMARD, rapid dose escalation, and parenteral route of administration in case of insufficient response to oral MTX [6,7]. In our study, 54.7% of RA patients in Korea received MTX as an initial therapy, and only half of the users achieved the maximal dose of ≥15 mg/week within 1-year after RA diagnosis. A limited proportion of patients tried parenteral MTX. Considering persistent glucocorticoid use during this period, RA activity was likely to be high enough to require MTX dose escalation. The proportion of MTX users (<10%) who ever tried a dose of ≥20 mg/week in our study was in contrast with that (26.4%~50%) in the European or USA cohorts [7,21]. One of the reasons for this finding might be due to concerns about potential toxicity with a higher MTX dose [22-24]. It has been noted that a lower maximal dose might be more appropriate for Asians due to possible toxicities [8,25,26]. In Japan, the MTX dose is allowed up-to 16mg/week due to potential toxicities [26]. Although studies are limited on dose-dependent adverse events in Korean population, future studies are essential to generate both biological and clinical evidence for optimal MTX dose among Koreans in terms of safety.
According to 2013 EULAR recommendations, only short-term use of low dose (≤7.5 mg/day) glucocorticoids is recommended in combination with a csDMARD for up to 6 months with rapid tapering [4]. Although glucocorticoids were mostly used at a low dose, discontinuation occurred in only a minority of patients in our cohort. The persistent use of glucocorticoids might in part be due to unremitting RA activity associated with suboptimal MTX use and beliefs in benefits of glucocorticoids outweighing risks [27,28].
Compared to the French cohort showing 10%~25% of csDMARD combination during the first year after RA diagnosis [7], more than 85% of csDMARD combination during the same period in our study was striking. Besides, almost all patients (>95%) in our study ever used csDMARD combination before bDMARD initiation probably due to HIRA requirements. It is still debatable whether to use csDMARD combination following failure of csDMARD monotherapy. Unlike 2008/2012 ACR guidelines that still offered an option of combining csDMARDs following MTX monotherapy failure [2], 2010/2013 EULAR recommendations did not support a step-up from MTX monotherapy to csDMARD combination in patients with poor prognostic factors (e.g., seropositivity as in our patients) [3,4]. However, comparable efficacy of csDMARD combination compared to bDMARD use has been claimed for patients with early arthritis and for MTX refractory patients in previous studies [10-13]. Therefore, csDMARD combination could be a viable option before bDMARD use among patients who cannot afford high costs of bDMARDs [29].
The longitudinal trend analysis showed that the average MTX dose has continually decreased over time. But use of leflunomide or tacrolimus has gradually increased particularly before initiating bDMARDs. Both leflunomide and tacrolimus have shown comparable efficacy to MTX in RA patients [17,18,30,31]. Since the HIRA reinforces csDMARD combination before initiating bDMARDs, physicians might prefer using lower dose MTX plus other potent csDMARDs rather than optimizing MTX dose in an attempt to obtain early achievement of HIRA requirements or to avoid toxicities at higher doses.
The real-world treatment patterns in our study could be explained by multiple reasons. It is inevitable that HIRA criteria altered treatment patterns during the pre-index period. However, the initial treatment strategy may be minimally affected by the HIRA requirements. Thus, a low proportion of MTX use as the 1st DMARD, lack of MTX dose escalation, or continuous glucocorticoid use could be associated with other reasons than HIRA requirements, such as preference to potent non-MTX drugs [17,18], concerns for toxicities of high dose MTX [22-26], and beliefs in benefits of glucocorticoids [27,28].
In our multivariable regression analysis, significant predictors of higher MTX dose and glucocorticoid discontinuation included demographics (younger age and/or male), concomitant treatments, and comorbidity status. Younger age was associated with MTX use of ≥15 mg/week probably due to less concerns for toxicities. Concomitant treatments (combination therapy, glucocorticoids use, and cox-2 inhibitors) associated with higher MTX dose also reflect high RA activity of the given patients. On the other hand, glucocorticoid discontinuation was associated with aggressive treatment strategies (combination therapy and MTX use of ≥15 mg/week). This is consistent with our previous findings [32]. Thus, there is a possibility that bDMARD was unnecessarily used in a subgroup of patients whose disease activity might have been controlled with elevated doses of MTX. We found significant associations of higher dose of MTX and glucocorticoid withdrawal with previous hospitalizations and chronic viral hepatitis, but not comorbidities. A higher rate of chronic hepatitis B or C might be explained by more vigilant surveillance following higher MTX dose or glucocorticoid treatment because high dose users or glucocorticoid users were likely to visit the clinic more frequently due to high disease activity.
Several strengths are worth to mention in this study. First, few previous studies have evaluated treatment patterns of csDMARDs before bDMARD initiation, particularly in relation to early (≤1-year from diagnosis) treatment strategies. Second, analyses were relevantly comprehensive, comparing real-world treatment patterns and evaluating annual trends. Third, this nationally representative database can ensure generalizability of the study findings to those countries where they require csDMARD failure to qualify for reimbursement of bDMARD therapy [33]. Fourth, we could also determine the effect not only of clinical treatment guidelines but also of national health regulations on prevalent treatment patterns.
Our study has some limitations. First, using claims database, we did not have information on RA severity/activity and specific reasons for suboptimal MTX use, e.g., toxicities. Second, only bDMARD initiators were included in this study. Therefore, we did not compare them with never- bDMARD users who may have had different treatment patterns. Nevertheless, our findings are still relevant in terms of active RA patients because bDMARD users show in general high RA activity and need more rigorous treatment. Third, we used the 2004~2016 database, and the pre-biologic treatment patterns would be different at present. However, our data on longitudinal treatment patterns provide useful information to infer trajectories on current treatment patterns.