Need for circumspection in prescribing tumor necrosis factor inhibitors and other biologic response modifiers
Article first published online: 7 FEB 2002
Copyright © 2002 by the American College of Rheumatology
Arthritis Care & Research
Volume 47, Issue 1, pages 1–4, February 2002
How to Cite
Matteson, E. L. (2002), Need for circumspection in prescribing tumor necrosis factor inhibitors and other biologic response modifiers. Arthritis & Rheumatism, 47: 1–4. doi: 10.1002/art1.10244
- Issue published online: 7 FEB 2002
- Article first published online: 7 FEB 2002
- Manuscript Received: 7 AUG 2001
- Manuscript Accepted: 7 AUG 2001
The article by Phillips et al in this issue of Arthritis Care & Research entitled “Experience with etanercept in an academic medical center: are infection rates increased?” is the first long-term experience with a biologic response modifier (BRM) reported in a meaningfully large group of patients treated in general practice outside the context of a clinical trial (1). Patients were treated for a variety of rheumatic diseases; 80% of the 180 patients had rheumatoid arthritis (RA) or juvenile RA, and the remaining 20% had one of several other conditions, principally spondylarthropathies but also adult Still's disease, dermatomyositis, and other less well-characterized diseases. Most if not all of the patients with RA appear to have had longstanding or severe disease, and 23% of the initial 180 patients had undergone at least one joint replacement.
Using the pre-etanercept clinical course of the same patients prior to initiation of etanercept therapy as the control for the period of etanercept therapy, the authors found no overall increase in the risk for serious infection after initiation of etanercept. In the pre-etanercept period, 19% of the evaluated 168 patients had what were interpreted as minor infections (cellulitis, upper respiratory infections, urinary tract infections, etc.), and 3 patients had serious infections (requiring intravenous antibiotic therapy), including pneumonia, cholecystitis, and deep soft-tissue infection. There were no deaths.
During the etanercept period, 16 patients reported minor infections, including cellulitis or bronchitis, although the number of patients actually requiring antibiotic therapy is not provided. Five serious infections occurred and etanercept was discontinued in 4 of these patients. Two of these patients subsequently died. One was a 76-year-old woman with longstanding RA and myelodysplasia who was taking only etanercept and rofecoxib for treatment of arthritis at the time of her death, which was due to sepsis following colonic polypectomy. The other was a 79-year-old woman taking etanercept, prednisone, hydroxychloroquine, and nabumetone for longstanding RA, who died of multimicrobial infection following initial diagnosis and treatment of a Mycobacterium avium-intracellulare psoas abscess.
There was a clear increase in the number of minor infections and other complications (54% of patients, including injection site reactions and rash, but also depression) compared with the pre-etanercept data. These complications necessitated discontinuation of therapy in 6 patients. Certainly some of the increase in side effects may reflect greater vigilance on the part of both patients and physicians in the period of etanercept therapy because of the concern about possible side effects, especially infection.
Although the risk of serious infection did not reach statistical significance, the small number of patients and relatively short duration of followup on etanercept (median of 10 months) precludes a final judgment about the risk of infection for patients taking etanercept. These same factors limit the conclusions of subgroup analyses regarding the possible contributing role of comorbidities to these adverse events. Factors identified included previous joint infection, concomitant medications such as methotrexate, and type of disease (RA, spondylarthropathy, and others). The authors do not address whether any other disease-modifying antirheumatic drug (DMARD) therapy was initiated after etanercept, which may have altered the risk for significant adverse events, although the number of such cases is likely to have been low.
Etanercept use appeared effective for most patients, and indeed about one-half who were taking methotrexate or prednisone were able to reduce the dosages of these medications or discontinue their use altogether. Viewed from another perspective, although etanercept appeared effective for most patients, there were many patients who discontinued therapy during the observation period. Reasons for discontinuation are as varied as for any other medication used to treat RA (adverse events, cost, other health concerns, lack of efficacy, etc.). Whatever the reason, over one-fourth (approximately 27%) had discontinued therapy by 6 months, and only 44% of patients in this study had been taking etanercept for more than 12 months. The number of patients for whom 12 months or more of observation was available who were still taking etanercept at 12 months is not provided, but is estimated to be between 50% and 70%, a figure similar to that of methotrexate in long-term studies.
Judgments about the use of tumor necrosis factor (TNF) inhibitors and BRMs that soon are to be marketed should be made in the context of a sound understanding of the nature and course of RA and the other diseases for which they will increasingly be used. RA is a serious disease that rarely remits. Disability occurs in more than 80% of patients, and mortality may be doubled, especially in patients with systemic disease (2–4).
The goal of RA treatment is to eliminate synovitis and disease activity; and when this is not possible, to control synovitis and disease activity to the fullest extent possible, thereby reducing disease-related symptoms, morbidity, and mortality (5). A true cure of the disease is not on the horizon, but improved outcomes have been realized in terms of reduced morbidity and mortality (6, 7). Initiation of effective DMARD therapies early in the disease course, and the emergence of methotrexate as a gold standard therapy against which most other therapies are measured, are major contributors to this improvement (8–10).
In contrast to standard DMARDs, the mechanism of action of BRMs, such as the TNFα inhibitors etanercept and infliximab, is better understood. TNF has a pivotal role in the immune system, and is a critical mediator of local inflammation. Among the many properties important in general inflammation and in the inflammation of RA, TNF causes activation of the vascular endothelium, recruitment of inflammatory cells, and activation of T cells, B cells, and platelets.
It is not known whether TNFα inhibition further increases the already increased background risk of infection in RA (5). In the setting of infection, the effects of TNFα inhibition may be detrimental, so that infections do not remain localized but become systemic, leading to syndromes of septic shock, disseminated intravascular coagulation, and death; these agents may also mask signs of serious infection (11). By the latter part of 2000, the Food and Drug Administration (FDA) had received reports of 2,444 patients who had been prescribed the TNFα inhibitor etanercept and 928 patients who had been prescribed the TNFα inhibitor infliximab (12). Of these, 93 patients receiving etanercept and 27 receiving infliximab had died. Despite knowledge about the mechanism of action of these agents, there is no practical knowledge of how much or how long TNFα inhibition is necessary to achieve therapeutic benefits and minimize such side effects as serious infection, but also pancytopenia, demyelinating disease, and other adverse events (13).
A problem in translating results of clinical trials to clinical practice is that long-term use of a given DMARD by individual patients is uncommon, due to primary and secondary treatment failure, and often due to drug toxicity. A weakness of all estimates of DMARD utility is that there are few studies of these agents extending longer than 1 year, with the notable exception of methotrexate (14). After 5 years of treatment with methotrexate, about 50% of patients are still on therapy, whereas fewer than 20% of patients remain on other DMARDs. Even so, side effects of methotrexate therapy may occur in as many as 70% of patients, although they usually do not lead to discontinuation of the drug (9). An additional problem in translating the results of a randomized clinical trial evident from the article by Phillips et al is that patients enrolled in clinical trials of etanercept are not representative of disease in the community, or of the way the drug is used in clinical practice. Patients in the study by Phillips et al had a variety of diseases, each possibly yielding different implications for measures of safety, especially infection risk.
DMARD-related side effects were common in the entanercept study by Phillips et al. The therapy dropout and adverse-event rates suggest that the long-term time taking etanercept alone or in combination with other DMARDs as used in general rheumatologic practice is, at the very least, not likely to exceed that of methotrexate.
In the next few years, methotrexate, used alone or in combination with other DMARDs, will continue to be the preferred DMARD for patients with established or severe, newly diagnosed disease. It has a well-recognized long-term efficacy and toxicity profile, low cost, and the lowest long-term drug discontinuation rate of any currently available DMARD (14). As single therapy, it has been shown to be superior or equivalent to other standard DMARDs, providing approximately 60–70% objective improvement, including slowing of radiographic progression (8, 15). Although patients taking etanercept respond more quickly in terms of clinical and radiographic parameters, the 1-year clinical response of disease to etanercept in a clinical trial was no different than that of methotrexate (16). Ultimately, only long-term studies of functional outcome, morbidity, mortality, and cost of BRMs in routine clinical practice will provide direction for their use.
Although most patients in the study by Phillips et al likely had at least some third-party payer coverage for the cost of etanercept, there were at least 2 patients who discontinued the drug because of cost. This fact highlights the issue of cost as a very real problem for the individual patient with rheumatic disease, and demands analysis from a broader perspective. Cost of older and newer DMARD therapies, especially BRMs, weigh strongly in the equation about their utility. Current direct costs of TNFα inhibitors are high, often more than $15,000 a year. Costs of other BRMs, including emerging TNF inhibitors and agents such as the interleukin-1 receptor antagonist, will likely also be in this price range.
A thorough understanding of cost considerations should weigh direct and indirect costs and years of life lost due to the disease and treatment complications. A simulation analysis of the excess cost of RA compared to nonarthritic controls during the first 25 years after a diagnosis of RA, and adjusted for differential survival among patients with RA, indicated that the median lifetime incremental costs of RA in 1995 US dollars range approximately from $61,000 to $122,000 (higher for younger individuals) (17). A single intervention (such as autologous bone marrow transplantation) costing up to $60,000 may be cost-saving if it eliminates the downstream incremental costs of RA, especially if it can be shown to be effective at the outset of disease (17). The cost efficiencies of expensive BRM therapies requiring continued administration over many years of treatment are less clear. In early aggressive and persistently active disease, newer agents will challenge the role of methotrexate, if the use of these drugs can establish remission and limit joint damage with a favorable long-term safety and cost profile.
Until more long-term safety and efficacy data are available, marketed BRMs such as the TNFα inhibitor etanercept should be prescribed only upon the recommendation of physicians with specialty training in the diseases for which they are used.
Perhaps with the exception of very rare or very unusual infections, proof of association between a drug and an infection observed in a patient with RA is difficult to assert, and ultimately must be proven in a controlled study. Until more evidence about infection risk is available, it would seem prudent for physicians prescribing BRMs, such as TNFα inhibitors, for RA to follow the guidelines suggested by O'Dell (13):
- 1Patients with new-onset RA should initially be treated with conventional DMARDs.
- 2Physicians, especially primary care physicians, need to be aware of the possibility of severe infections in patients taking these agents, and of the need to stop TNF inhibition at the first sign of infection of any kind. With the role that TNF plays in keeping infections localized, time is critical. These agents should be restarted only when the infection has resolved completely.
- 3Patients should be aware of these risks, should stop taking BRMs promptly if any infection (including skin infection) or fever occurs, and contact their physician for further guidance.
When therapy with one of these TNFα inhibitors is contemplated, screening for past and current serious infections and other adverse events should be performed, and may include testing for tuberculosis by purified protein derivative and/or chest radiograph, serial blood counts, and serial clinical examination. If active infection is suspected, the drug should be discontinued and the patient appropriately treated. Recommendations for other BRMs will likely be similar, and dictated by their specific side effect profiles.
Many patients with RA will have acceptable disease control with conventional DMARDs. The most effective DMARD or combination of DMARDs appropriate for the degree of disease should be used. DMARDs such as hydroxychloroquine and sulfasalazine are often used for patients with very limited disease activity, although most patients will require more intensive therapy. Patients with active disease should have an initial trial of the best therapy available. In general, this means use of methotrexate and/or a similar agent, such as leflunomide. Patients not tolerating, or responding poorly to these agents may respond to combination therapy with methotrexate, sulfasalazine, and hydroxychloroquine. Combination therapies with BRMs, currently the TNFα inhibitors etanercept and infliximab, should be considered for use in patients who have uncontrolled disease activity (5, 18, 19).
Off-label usage, for example in idiopathic inflammatory myopathies and spondylarthropathies, may provide important therapeutic benefits for patients with these diseases. Use of such BRMs as etanercept for off-label indications should be supported by clinical trials and systematic postmarketing safety and efficacy data collected from practices where these agents are used for these indications. To improve the knowledge base about BRMs as they are used in daily clinical practice, physicians prescribing these agents should carefully record any suspected drug-related adverse events. Ideally, a system such as that currently being considered by the American College of Rheumatology should be devised for independent reporting of these and other disease- and treatment-related outcomes. In the meantime, physicians should be aware of and use the FDA Medwatch mechanism for reporting suspected serious adverse events, including infections, malignancies, new autoimmune diseases, and others.
- 8Historical overview of the treatment of rheumatoid arthritis with an emphasis on methotrexate. J Rheumatol 1996; 23: 34–7..
- 12Postlicensure reports of infection during use of etanercept and infliximab [abstract]. Arthritis Rheum 2000; ( 43 suppl 9): 2857., , , .