SEARCH

SEARCH BY CITATION

To the Editor:

It is axiomatic that an effective and safe therapy should be administered as early in the course of an illness as possible. In a recent editorial in Arthritis & Rheumatism (A&R), Pincus et al state that a delay of ≥3 months before institution of aggressive therapy for rheumatoid arthritis (RA) “appears undesirable if not unacceptable” (1). In the same editorial, the authors state that “many people who meet the American College of Rheumatology (ACR; formerly, the American Rheumatism Association) 1987 revised criteria for RA for <6 months have a self-limited process rather than a progressive disease and do not require urgent intervention.”

The message of urgency in the editorial by Pincus et al derives in part from the report by Möttönen et al (2) in the same issue of A&R. This report concerned the effect of a delay of a few months from onset of symptoms to institution of therapy in a study of a single drug (sulfasalazine) versus combined sulfasalazine, methotrexate, hydroxychloroquine, and prednisolone therapy. Pincus et al acknowledge Möttönen et al's finding that delay in starting single-drug therapy impairs the subsequent capacity to control symptoms of RA, but they fail to note that, in the same study, a similar delay when using more than a single drug resulted in no decreased efficacy. Arguments have been advanced that there may be a window of opportunity in which to start disease-modifying antirheumatic drug (DMARD) therapy (3, 4), and that “some patients may respond to intervention in a fundamentally different way than they would if therapy were delayed” (5).

A consequence of these messages of urgency to diagnose RA rapidly and to institute aggressive therapy equally rapidly is that they may be misapplied in the community, especially by nonrheumatologists. Since RA remains a disease of very variable course and, as stated by Visser et al (6), “since treatment with disease-modifying antirheumatic drugs (DMARDs) is only justified when the risk:benefit or cost-effectiveness ratios are favorable,” it is valuable to review the various observations regarding the accuracy of early diagnosis of RA as well as our ability to predict which patients' disease will follow a progressive erosive course, and to consider whether our current medications will show effectiveness and safety in the community setting as well as in the controlled environment of the clinical trial.

Among individuals referred to the Norfolk Arthritis Register (NOAR), a community-based inception cohort, Harrison et al studied 358 patients with very early inflammatory polyarthritis (IP). Fifty-two percent of these patients were classified as having RA (7). Using a definition of remission as “no arthritis on examination in a patient who has not taken second-line drugs or steroids in the preceding 3 months,” these investigators contrasted their experience with findings from studies that were “hospital or clinic based populations thus hampered by referral selection bias.” In their study, 25% of the whole cohort had disease remission at 2 years. Nineteen percent of the cohort initially diagnosed as having RA had natural disease remission at 2 years, as did 42% of those initially diagnosed as having undifferentiated IP (UIP); further, “patients who had been treated with second line drugs or steroids at any time during the 2 years were less likely to be in remission.” An urgency approach to such a community-based cohort would probably result in unnecessary exposure of a significant population of patients to potentially toxic medications. In contrast to RA patients in most clinical trials of new drugs for RA, only one-third of Harrison et al's community-based patients were rheumatoid factor (RF) positive.

The 2002 Update of the ACR Guidelines for the Management of Rheumatoid Arthritis states that “the majority of patients with newly diagnosed RA should be started on disease-modifying antirheumatic drug (DMARD) therapy within 3 months of diagnosis” (8). The ACR 1987 revised criteria for RA are, in fact, classification criteria and not diagnostic criteria and “were not useful for predicting RA in patients with arthritis onset within the previous year” (9).

Visser et al (6), using a logistic regression analysis of the baseline variables of 524 consecutive, newly referred patients with early arthritis, used the diagnosis at the 2-year followup visit to define three disease groups: self-limiting arthritis, persistent nonerosive arthritis, and persistent erosive disease. The initial diagnosis of early inflammatory arthritis led to 60% with self-limiting arthritis, 16% with persistent nonerosive arthritis, and 24% with persistent erosive arthritis at the 2-year followup visit. Of those classified as having RA (30% of the total), 10% had self-limiting disease, 22% had persistent nonerosive disease, and 68% had persistent erosive disease (calculated from Table 2 of Visser et al [6]). According to Visser et al, “patients who had no signs of arthritis at followup but who were taking DMARDs were classified as having persistent arthritis.” This community-based population provides a view of the larger population of persons presenting with IP and differs from that of those qualifying for inclusion in randomized clinical trials that assess the effects of delays in starting therapy. It is noteworthy that almost one-third of the RA patients had a disease course over 2 years characterized as self limiting or persistent nonerosive.

The magnitude of our uncertainty regarding accurate diagnosis of early RA is clearly presented in the review of reports from early arthritis clinics by Berthelot et al (10), in which “there was a striking lack of agreement among researchers about the nosology and/or taxonomy of many cases of mild arthritis” and “some cases first classified as undifferentiated arthritis (UA) subsequently prove to be RA whereas others too quickly diagnosed as RA resolve on follow-up.” Visser et al (6) provide diagnostic criteria that allow distinction of persistent nonerosive disease from persistent erosive disease. Such distinctions would justify allowing sufficient time to decide whether to use some of the current, potentially toxic therapies. In his editorial regarding the study by Visser et al, Scott references and compares prognostic factors identified in cohort studies and states that “expensive, disease-modifying therapies, especially immunotherapy, should only be given to patients at a high risk of severe disease, and the current ACR criteria do not help in identifying such cases” (11).

It seems clear that, if early treatment is to be defined as time since diagnosis of RA, a significant period of time may need to elapse from first symptom to certainty of diagnosis, and prognostic criteria will need to be applied to distinguish the subtypes of disease course to make appropriate therapeutic decisions. This is not to imply that a great deal should not be done for the patient during this decision-making period by using antiinflammatory agents, including low-dose glucocorticoids which, administered for short duration, have acceptable side effects and have been reported to have a beneficial effect on the course of radiographic measurements of joint destruction (12).

The observation that powerful antiinflammatory agents, such as etanercept and infliximab, could influence the rate of radiographically observed joint damage clearly accelerated the impulse to early therapy (i.e., prior to first observed damage). Studies based at secondary care centers and using the ACR classification criteria, which may include erosions as a baseline requirement, are clearly affected by selection bias. Bukhari et al (13) reported on the time to first occurrence of erosions using a community-based inception cohort in which patients with IP, but without baseline erosions, were followed up for 2 years. Of the 95% of entry subjects who met ACR criteria for RA by the fifth visit, 44% developed erosive disease. An interesting subgroup of patients developed their first erosions after 2 years. Overall, it was estimated that 33–83% of such patients would develop erosions, usually within the first 2 years. This leaves a significant percentage of patients presenting with RA who do not develop early erosive disease, which is the target of aggressive combination therapy and, especially, of the anti–tumor necrosis factor (anti-TNF) biologic agents.

Brennan et al (14) addressed the issue of identifying those at high risk of developing erosive disease by developing an algorithm based on the NOAR cohort subgroup of patients with new-onset IP. By dividing the patient population and then testing the derived algorithm on the second, or validation, group, these investigators found that the erosion status of 79% of patients was predicted correctly using just three variables: RF status, swelling of ≥2 large joints, and disease duration of >90 days. This predictive trio should allow reduced urgency for a significant segment of patients with early RA. Bukhari et al (15) subsequently reported that, using multivariate regression techniques and after adjusting for baseline variables, only RF positivity was a useful predictor of disease progression; further, a dose–response relationship between RF titer and worsening damage could be found that was independent of treatment.

It would therefore, appear that, for patients presenting early with IP, clinical and laboratory diagnostic criteria are readily available that will permit likelihood statements regarding a nonerosive course of disease. Such patients should be treated as nonurgent (or at least less urgent) cases, and the minimum necessary amount of nonsteroidal antiinflammatory drugs and/or DMARDs should be used together with physical modalities and careful monitoring. Application of prognostic indicators of future erosive disease should further guide sequential therapy.

Along with the understandable desire to prevent irreversible bone and joint lesions is the belief that our new and powerful antiinflammatory drugs will provide safe and effective therapy. The gold standard for measuring the efficacy of individual drugs, as well as combinations of these drugs, remains the randomized, double-blind, clinical trial. Although differences in radiographic progression have assumed great importance when comparing drug(s) with placebo, the effects on disease activity as measured by the ACR 20% improvement criteria (ACR20) (16) and, to a lesser extent, the ACR50 and ACR70 (17) have dominated the clinical trial efficacy outcome literature.

Suarez-Almazor (18) has provided an enlightening analysis of four of the most widely used DMARDs, expressing efficacy as the absolute benefit attributable to the drug (calculated as drug ACR20 result − placebo ACR20 result), yielding 20% absolute benefit for methotrexate, 26% for leflunomide, 48% for etanercept, and 16% for anakinra. It is, of course, hoped that the degree of absolute benefit in the clinical trial will be observed when the drugs are used in the community under conditions far removed from those imposed in the randomized controlled efficacy trial. Drug effectiveness (i.e., under community conditions) can be influenced by access, adherence, diagnostic accuracy, and other factors. Suarez-Almazor calculates the potential community effectiveness as 8% for methotrexate, 11% for leflunomide, 19% for etanercept, and 6% for anakinra. This means, for example, that when treated with etanercept in the community, only 19 of 100 patients with RA might be expected to achieve an ACR20 level of benefit attributable to the drug. Although these estimations lack both rigor and confirmation, it is vital to consider that conditions and results recorded in formal randomized controlled trials are not to be expected when medications are used in the community.

Finally, results of postmarketing surveillance of some of our most recent drugs, leflunomide, etanercept, and infliximab, must give us pause. Reports regarding liver toxicity (19, 20) and other adverse side effects (21) associated with leflunomide have led to a petition to the Food and Drug Administration (FDA) for withdrawal of this drug. Documentation contained in this petition, including FDA adverse event reports, is available online at http://www.citizen.org/documents/1614.pdf. The anti-TNF biologics are rapidly accumulating warnings regarding their use in individuals susceptible to infections (22) (RA itself having a long history of increasing the likelihood of infection), and these drugs have also been reported to be associated with lupus-like phenomena (23) and demyelinating disorders. Since none of these agents are believed to affect the etiology of RA, and all commonly lead to resurgence of the inflammatory disease process upon their withdrawal, it must be considered that to be used successfully, they may have to be used indefinitely, with all that this implies regarding adverse side effects and costs.

RA remains a disease of very variable presentation and progression. The subset of patients usually studied in controlled clinical trials of new drugs commonly has clinical and laboratory characteristics associated with a bone-destructive and disabling disease course. Treatment of such patients with early and “aggressive” therapy would appear warranted. When applied to community-based persons with UIP, this sense of urgency may be inappropriate, given the problems described in the diagnosis of early RA and given the existence of recognizable clinical subsets of patients whose disease will probably follow a self-limited, nonerosive course. Since all of our current DMARDs treat elements of disease pathogenesis and not disease etiology, we may anticipate their use for prolonged periods of time. Postmarketing surveillance is essential so that the risk:benefit ratio in the community can be monitored closely.

REFERENCES

  1. Top of page

Wallace V. Epstein MD, MACR*, * University of California, San Francisco.