- Top of page
The results of randomized controlled trials (RCTs) conducted in carefully controlled clinical research environments often do not reflect the actual practice settings in which the medications are subsequently prescribed (1). For example, the narrow eligibility criteria employed to enroll patients in RCTs tend to exclude elderly patients with multiple concurrent illnesses, who are at higher risk for experiencing adverse events. RCTs may exaggerate treatment benefits by including more skillful physicians and participants with a greater likelihood for improving. Important but infrequent adverse events may not be recognized for years after the medication has been extensively used in the community (2). Postmarketing studies on newly approved drugs thus have a vital role in detecting toxicities not identified in RCTs and evaluating drug efficacy in nonresearch populations and environments.
The Food and Drug Administration (FDA) approved leflunomide (LEF), a disease-modifying antirheumatic drug (DMARD), for the treatment of rheumatoid arthritis (RA) in October 1998. LEF inhibits lymphocyte proliferation and hence the clonal expansion of T cells in RA by blocking dihydro-orotate dehydrogenase, an enzyme critical for de novo pyrimidine synthesis (3). Several multinational RCTs have demonstrated that LEF is a safe and effective DMARD equivalent to methotrexate and sulfasalazine for treating the signs and symptoms of RA and retarding disease progression as measured by radiography (4–10). Although longer-term tolerability and effectiveness have been reported in some postmarketing studies (11, 12), high discontinuation rates in daily practice have also been noted (13–15).
In this article, we describe the use of LEF in clinical practice, including initial dosage, coadministration with other DMARDs, duration of administration, and reasons for discontinuation, in a large national cohort of American veterans during the 33 months immediately following the drug's introduction. The Department of Veterans Affairs (VA) administers the largest integrated health care system in the United States (16). Its unique national databases permit a large number of patients for whom LEF has been prescribed to be easily identified and their medication use characterized.
- Top of page
This cohort was predominantly male (92%), older (47% were over 65 years of age), and white (63% were white, 7% African American, and race was undocumented in 26%) (Table 1). The mean annual income was $23,000 (SD $48,000, median $14,000).
Table 1. Sociodemographic characteristics of 3,325 leflunomide recipients
|Characteristic||Leflunomide recipients, % (n)|
| Female||8.0 (267)|
| Male||91.3 (3,038)|
| Missing||0.6 (20)|
|Age, years|| |
| <34||1.6 (53)|
| 35–44||6.0 (192)|
| 45–54||20.7 (689)|
| 55–64||24.5 (813)|
| 65–74||31.6 (1,051)|
| >74||15.3 (507)|
| White||62.5 (2,080)|
| African American||6.9 (230)|
| Other||4.7 (140)|
| Missing||26.6 (875)|
|Annual income in 1998–2001|| |
| <$20,000||61.4 (2,043)|
| $20,000–$40,000||28.2 (938)|
| $40,001–$60,000||4.2 (140)|
| >$60,000||5.2 (174)|
| Missing||0.9 (30)|
Use of LEF was widely adopted in the VA system shortly after FDA approval, as indicated by the observation that prescriptions were being written by providers at 110 VAMCs within 9 months following the drug's introduction. The number of patients newly started on LEF rapidly increased following FDA approval from 75 patients in the 4th quarter of 1998 to 325 patients in the 4th quarter of 1999. As noted in Table 2, the number of patients receiving new prescriptions remained relatively constant, in the range of 230 to 425. In the second quarter of 2001, nearly 2,000 patients received at least 1 LEF prescription.
Table 2. Initiation and discontinuation of leflunomide as a function of calendar quarter
|Calendar quarter||Leflunomide prescriptions written, n (%)||Patients on whom leflunomide treatment was||Patients who received leflunomide during quarter, n|
|Initiated* n (%)||Discontinued† n (%)|
|4th 1998||161 (0.6)||75 (2.3)||2 (0.14)||75|
|1st 1999||791 (3.0)||231 (7.0)||56 (4.0)||304|
|2nd 1999||1,108 (4.2)||275 (8.3)||69 (5.0)||490|
|3rd 1999||1,917 (7.3)||325 (9.8)||116 (8.3)||752|
|4th 1999||2,352 (9.0)||322 (9.7)||122 (8.8)||937|
|1st 2000||2,826 (10.8)||379 (11.4)||167 (12.0)||1,175|
|2nd 2000||2,257 (8.6)||243 (7.3)||158 (11.4)||1,168|
|3rd 2000||3,385 (13.0)||424 (12.8)||243 (17.5)||1,476|
|4th 2000||2,827 (10.8)||324 (9.7)||173 (12.4)||1,476|
|1st 2001||4,099 (15.7)||381 (11.4)||285 (20.5)||1,792|
|2nd 2001||4,423 (16.9)||346 (10.4)||NA||1,934|
Because of LEF's prolonged half-life and the desirability of rapidly attaining a steady state blood level, the FDA approved an oral loading dose of 100 mg daily for 3 days, followed by 10–20 mg daily. In clinical practice, significant variations in the loading patterns were observed because only 61% received the recommended loading sequence, 4% received 100-mg tablets daily for 2 days, 3% received some other loading pattern, and 31% did not receive any loading dose (Table 3).
Table 3. Leflunomide loading patterns
|Loading pattern||% (n)|
|100 mg for 3 days||61.2 (2,037)|
|100 mg for 2 days||4.4 (147)|
|100 mg once weekly for 3 weeks||0.5 (18)|
|No load; begin 10 mg/day||5.4 (180)|
|No load; begin 20 mg/day||25.8 (859)|
|Other loading pattern||2.5 (84)|
LEF was prescribed without any other DMARD in 33% of patients, with methotrexate in 24%, hydroxychloroquine in 11%, hydroxychloroquine + methotrexate in 10%, or with other combinations in 22% (Table 4). LEF was discontinued in 42% (1,391 of 3,325) of patients. The median time to LEF discontinuation, estimated by survival analysis, was 17.6 months (95% confidence interval [95% CI] 15.7–18.7; Figure 1). Of the patients who discontinued LEF, 570 (41%) did so within 3 months and 876 (63%) within 6 months (Table 5). Logistic regression results indicated increased odds for LEF discontinuation with 3-day loading (odds ratio [OR] = 2.0, 95% CI 1.76–2.4), age <44 years (OR = 1.6, 95% CI 1.2–2.1), age >75 years (OR = 1.8, 95% CI 1.4–2.2), and annual income <$60,000 (OR = 1.8, 95% CI 1.3–2.5). Reduced odds for discontinuation were associated with loading patterns other than 100 mg for 3 days (OR ranging from 0.2 to 0.6) and annual incomes >$60,000 (OR = 0.6, 95% CI 0.4–0.8). Results were identical when the analysis was repeated with Cox proportional hazard modeling. At the time of discontinuation, LEF was being administered alone in 34%, in combination with methotrexate in 17%, with etanercept in 10%, with hydroxychloroquine in 9%, or with other combinations in 30% (Table 6).
Table 4. DMARD combinations at initiation of LEF among 3,325 patients*
|DMARD combination||% (n)|
|LEF only||32.6 (1,086)|
|LEF + MTX||23.4 (780)|
|LEF + MTX + SSZ||1.6 (55)|
|LEF + MTX + HCQ||10.3 (345)|
|LEF + MTX + HCQ + SSZ||1.4 (49)|
|LEF + HCQ||10.8 (361)|
|LEF + HCQ + SSZ||1.5 (51)|
|LEF + SSZ||3.0 (101)|
|LEF + etan.||2.7 (90)|
|LEF + AZA||2.4 (83)|
Figure 1. Discontinuation of leflunomide: Kaplan-Meir survival analysis. SDF = survivor density function (proportion with no event at a given time).
Download figure to PowerPoint
Table 5. Duration of leflunomide therapy among 1,391 discontinuers
|Months to discontinuation||% (n)|
Table 6. DMARD combinations at discontinuation among 1,391 patients who discontinued LEF*
|DMARD combination||% (n)|
|LEF only||34.3 (477)|
|LEF + MTX||17.4 (242)|
|LEF + MTX + etan.||5.5 (77)|
|LEF + MTX + HCQ||5.1 (71)|
|LEF + MTX + HCQ + etan.||1.9 (27)|
|LEF + HCQ||9.4 (130)|
|LEF + HCQ + etan.||2.7 (38)|
|LEF + SSZ||1.8 (25)|
|LEF + etan.||10.0 (139)|
|LEF + AZA||1.6 (22)|
The medical records of 291 patients who discontinued LEF at 7 VAMCs were reviewed. Of these discontinuers, most (80%) were being treated for RA, the only approved indication; 8% received LEF for psoriatic arthritis, and 12% for other indications, e.g., discoid lupus, lupus, polymyositis, polymyalgia rheumatica, and giant cell arteritis. The most commonly cited reasons for discontinuation were lack of efficacy (30%), gastrointestinal symptoms including diarrhea (29%), and noncompliance and lost to followup (14%) (Table 7). Of the discontinuers, 19 (6.5%) were eventually restarted on LEF, often at a lower dosage.
Table 7. Reasons for discontinuation of leflunomide among 291 patients*
|Reason for discontinuation||% (n)|
|Lack of efficacy||30 (87)|
|All gastrointestinal||29 (83)|
|Noncompliance or lost for followup||14 (41)|
|Other reasons||10 (28)|
|Liver function test result elevation||5 (15)|
|Weight loss||2 (6)|
Data extracted in April 2002 from the central VA database (PTF) indicated that 103 of the 3,325 LEF recipients had died, including 78 patients who were in the group of 1,391 discontinuers. The median interval between the last LEF prescription and the date of death was 3.6 months (mean 8 months, SD 8.6 months); for 31 patients (44%) the date of death occurred at least 6 months after their last LEF prescription. Seven of the 78 discontinuers who died were patients at 1 of the 7 medical centers that collaborated in the present study, and their medical records were reviewed. An additional 17 deaths, not reported in the central VA database, were noted on chart review at participating medical centers. Among all patients (291) whose charts were reviewed, 24 were dead but no deaths were attributed to LEF, although it was being taken at the time of death in 11 patients.
- Top of page
We report the use of LEF in a national cohort of predominantly elderly, low-income, male veterans. Only 61% received the recommended loading dose of 100 mg daily for 3 days. LEF was initially prescribed without any other DMARD in one-third of patients, with methotrexate alone or with methotrexate and other DMARDs in 37% of patients, and with DMARDs other than methotrexate in the remainder. Forty-two percent of patients discontinued LEF, 41% within 3 months and 63% within 6 months. Patients were more likely to discontinue LEF if they had initially received a 3-day 100-mg loading dose, were younger than 44 years or older than 75 years, and reported an annual family income of less than $60,000. Abnormal liver function test results were cited as a reason for discontinuation in only 5% of these patients. No deaths were attributed to LEF. Our data therefore support the conclusion that LEF is a relatively safe drug in clinical practice, and that a loading dose of less than 100 mg daily for 3 days is better tolerated.
The in vivo activity of LEF is attributed to its active metabolite A-771726 (M1), which has a minimum half-life of 15 days. In the absence of a loading dose, it has been estimated that attainment of steady-state plasma concentrations would require nearly 2 months. The 25 mg/day regimen was found to be most effective, but subsequently a dose of 20 mg/day was determined to maximize the probability of clinical success based on a population pharmacokinetic model analysis (17). In the pivotal multinational clinical trials (4–7), a 100-mg loading dose for 3 days followed by 20 mg/day was used to ensure rapid attainment of steady-state levels. Only in the open label (18) and randomized (19) studies to evaluate the combination of LEF and methotrexate, was 100 mg loading for 2 days and an optional daily dose of 10 mg used.
Anecdotal evidence suggests that rheumatologists omit loading doses or prescribe alternate loading patterns to avoid gastrointestinal toxicity. In our cohort, besides some common regimens (Table 3), 3% of patients received unconventional loading regimens, such as 60 mg for 3 days and 40 mg daily for a week. Patients were less likely to discontinue if they received loading patterns other than 100 mg for 3 days. Although it is possible that the alternative loading regimens may compromise clinical efficacy, such regimens seem to be tolerated better and help ensure continuation of therapy. Our study did not examine whether the higher discontinuation rate with the 3-day loading dose was caused by a higher frequency of side effects, but such an association between 3-day loading and occurrence of early adverse events has been reported (20).
Combination DMARD therapy is frequently used by rheumatologists (21). Although there have been no RCTs to evaluate combination of LEF with DMARDs other than methotrexate, such combinations are used in practice. In our cohort, 34% of patients were started on LEF without another DMARD, and 49% were prescribed LEF with some combination of methotrexate and/or hydroxychloroquine. Among the 1,391 patients who discontinued LEF, 34% were taking only LEF at the time of discontinuation, and 52% were taking some combination of methotrexate and/or hydroxychloroquine.
The most frequently reported adverse events related to LEF include diarrhea, nausea, rash, and alopecia. Liver toxicity, most commonly mild transaminitis, was also reported in early trials (4, 5, 22). The combination of methotrexate and LEF has been associated with particularly high levels of liver toxicity, with up to 63% of patients in other studies developing transaminitis or more severe liver damage (18, 23). The European Agency for the Evaluation of Medicinal Products reported 296 cases of hepatic adverse events during 104,000 person years of LEF therapy, of whom 232 patients developed liver enzyme abnormalities, 2 patients developed cirrhosis, 15 patients developed liver failure, and 15 patients died (10 deaths possibly related to LEF) (24). In our cohort, presumed LEF-induced liver enzyme elevation was the reason for drug discontinuation in only 15 patients (15 of 291 or 5% of reviewed medical records of discontinuers), and no deaths were attributed to LEF. The explanation for the relatively low prevalence of liver damage we observed is speculative, but may include differences in patient clinical characteristics and the retrospective design of this study.
The number of withdrawals and the reasons for discontinuation in the 3 major clinical trials on LEF are summarized in Table 8. The overall discontinuation rates were about 20–30% in these trials. In a large postmarketing study on 4,250 RA patients receiving LEF between 1998 and 2000, Wolfe et al (11) reported that 25% of the patients had discontinued the medication at 1.1 years. This cohort was predominantly white (90%) and female (80%) with a median annual income of $35,000. The best predictor of discontinuing therapy was the occurrence of side effects, although prior DMARD use and concomitant use of prednisone or methotrexate also increased discontinuation. In contrast, in a community-based practice in the US (14), 52% of patients discontinued LEF within a year, 35% because of inefficacy, and 17% because of adverse events. Similarly, in a cohort of 99 RA patients from a community practice in the Netherlands, LEF was discontinued in 61 patients (62%) after a median of 12 weeks (range 1–64 weeks) with 55% of them discontinuing due to side effects (13). In a Swedish postmarketing surveillance study, only 22% of patients were still receiving LEF at 20 months, and most withdrawals were due to side effects that occurred in the few weeks following initiation (15). Our study is more in agreement with these studies than the RCTs or the postmarketing study of Wolfe et al (11).
Table 8. Withdrawal rates reported in the pivotal leflunomide randomized clinical trials
|Name of the trial||US301||MN301||MN302||MN302|
|Total number of patients randomized to leflunomide||182||133||501||292|
|Discontinuation rate, %||47||28||30||20|
|Reasons for discontinuation, %*|| || || || |
| Adverse events||22||14||19||8|
| Lack of efficacy||17||7||7||6|
| Protocol violation||0||NA†||NA†||NA†|
| Lost to followup||0.5||NA†||NA†||NA†|
The primary strengths of this study are that virtually every individual who received LEF from the VA was identified and included in the data analyses. In addition, to our knowledge, this is the first study of LEF use that includes a large number of older males from predominantly lower income groups.
The major limitations of this study are that it is retrospective, and (because of logistical difficulties associated with obtaining HSS approval from a large number of VAMCs) medical record review was performed only on patients who discontinued LEF at 7 academically affiliated facilities. In addition, race was undocumented in 26% of patients, which diminishes the validity of this variable in the model developed to predict discontinuation; the cohort was diagnostically heterogeneous; it was not possible to examine the association between disease duration or severity and outcome; the definition of inefficacy was based on individual physician interpretation rather than standardized definitions; drug toxicity was not defined in a consistent manner across medical centers; it was not possible to assess the relationship between comorbid medical illnesses or non-DMARD medication use and risk for drug toxicity; the discontinuation rate likely represents the highest estimate because review of medical records revealed that 6.5% of the patients who met criteria for discontinuers subsequently resumed the drug; and the definitions applied to identify overlapping DMARD use at the initiation and discontinuation of LEF probably do not precisely establish intended combination treatment regimens in all patients.
In this large, predominantly male, low-income national cohort with significant numbers of elderly patients, LEF was used with relative safety. The VA's national pharmacy and clinical databases provide an excellent resource for postmarketing evaluation of medication use and toxicity, particularly for an older male population that typically does not participate in clinical trials.