Drs. Phillips and Husni contributed equally to this study.
Experience with etanercept in an academic medical center: Are infection rates increased?†
Version of Record online: 7 FEB 2002
Copyright © 2002 by the American College of Rheumatology
Arthritis Care & Research
Volume 47, Issue 1, pages 17–21, February 2002
How to Cite
Phillips, K., Husni, M. E., Karlson, E. W. and Coblyn, J. S. (2002), Experience with etanercept in an academic medical center: Are infection rates increased?. Arthritis & Rheumatism, 47: 17–21. doi: 10.1002/art1.10243
- Issue online: 7 FEB 2002
- Version of Record online: 7 FEB 2002
- Manuscript Accepted: 19 JUL 2001
- Manuscript Received: 1 JUN 2001
- NIH. Grant Numbers: AR-07530-17, K08-AR-02074-1
- Arthritis Foundation Investigator Award
- Infection rates;
- Tumor necrosis factor alpha
There is little established information regarding the safety of antitumor necrosis factor therapies used outside the setting of clinical trials. This study evaluated the long-term safety and tolerability of open-label use of etanercept when used to treat patients with a variety of systemic rheumatic diseases. Reduction of concomitant corticosteroid and disease-modifying antirheumatic drug was also assessed.
Retrospective medical record review of 180 patients who were started on etanercept between December 1998 and April 2000 at an academic medical center.
Most patients (81%) remained on therapy for longer than 6 months, and a significant number (43%) of patients for longer than 12 months. Etanercept was prescribed for rheumatoid arthritis (RA) in 144 patients and for diseases other than RA, including ankylosing spondylitis, psoriatic arthritis, and polymyositis, in 36 patients. Fifty-six percent of patients taking corticosteroids were able to reduce their dose and 51% of patients were able to taper their methotrexate dosages. Forty-three patients (26%) discontinued etanercept. Reasons for discontinuing therapy included serious adverse events (2.9%), of which infection was most common. These included a psoas abscess secondary to Mycobacterium avium-intracellulare, septic wrist, bacteremia, and septic total hip replacement. Two deaths associated with infection were seen.
The majority of the studied patients tolerated etanercept for longer than 6 months. Many of these patients were able to subsequently taper or even discontinue corticosteroid and methotrexate therapy. Serious infections occurred in this patient population. Our results underscore the value of long-term observation under the conditions of clinical practice beyond controlled clinical trials.
Tumor necrosis factor α (TNFα) plays an important role in the pathophysiology of rheumatic diseases. It is produced by macrophages, lymphocytes, neutrophils, and keratinocytes and is a critical mediator of the immune response. TNFα exerts its effect by binding to specific cell surface receptors. Competitive inhibition of TNFα by the receptor extracellular domain renders TNFα biologically inactive (1). Soluble TNFα receptor agents belong to a class of medications known as biologic response modifiers. Etanercept is a TNFα receptor:IgG1 fusion protein that inhibits TNFα function. It was the first biologic modifier to be approved by the US Food and Drug Administration (FDA) for use in rheumatoid arthritis (RA) (2, 3).
Concern regarding the relationship between TNFα inhibition and vulnerability to infection stems from the known activities of TNFα. TNFα participates in the activation of T cells during antigen presentation, and has been shown to stimulate B cells (4). Shortly after its approval by the FDA, 6 deaths from infection were reported among approximately 25,000 patients. The package insert was changed to reflect this potential risk in patients with active infections, including chronic or localized infections. Postmarketing surveillance reveals that more than 85,000 patients have been treated with etanercept; little is known outside of clinical trials about the experience and safety of this novel therapy.
We examined the tolerability and safety of etanercept in an academic medical practice. Retrospective chart review was performed for each patient for the duration of etanercept therapy. Male and female patients between the ages of 18 and 80 were eligible for inclusion in the study. The association between adverse events and underlying diagnosis or concomitant medications was explored. Treatment with disease-modifying antirheumatic drugs (DMARDs) was assessed, and corticosteroid use was examined. Adverse events, including infections, were identified in our patients before and during etanercept therapy.
PATIENTS AND METHODS
Patients. A total of 180 patients were enrolled during the study period. These patients were not previously enrolled in an etanercept or any other clinical trial. Following FDA approval for etanercept in November 1998, injection-teaching sessions were held for all patients initiating etanercept at the Robert B. Brigham Arthritis Center at Brigham and Women's Hospital. Patients for this study were identified by their enrollment in etanercept teaching sessions. The setting is an urban academic medical center that serves as a tertiary referral center for systemic rheumatic disease. Prior approval for the study was obtained from the institutional review board of the hospital.
A systematic retrospective chart review was conducted on all 180 patients identified through the etanercept teaching sessions for the duration of the study period, from December 1998 to April 2000. Patients in the study received 25 mg of etanercept twice weekly. Demographics were obtained for each of the 180 patients. Of the initial cohort, 168 were followed for the entire period of the study. Twelve patients were unable to be followed while on etanercept. Six of the twelve patients moved away from the site of the center, 4 did not attend followup visits, and two did not begin etanercept although enrolled in the teaching session (both patients were also subsequently lost to followup in the Arthritis Center).
Each patient served as his or her own control for purposes of analyses. A chart review was performed for each patient in the study for the same time period as etanercept treatment duration. This was done for each patient beginning just prior to start of etanercept. As an example, a patient on etanercept for 6 months duration would have his or her chart reviewed for the 6 months prior to etanercept initiation (preetanercept period) and for the 6 months while on etanercept (postetanercept). Adverse events were ascertained in the same manner during the preetanercept period as in the postetanercept period. All records were reviewed using a predefined data collection sheet that included demographics, duration of etanercept use, history of joint replacement, previous DMARD use, concomitant corticosteroid and methotrexate use, minor and serious adverse events, and off-label use. Medically important or serious adverse events were defined as those requiring intravenous antibiotics or hospitalization.
Statistical analysis. Statistical significance was determined using chi-square analysis and Fisher's exact probability test for comparison of rates of adverse events between the patients prior to etanercept use and during etanercept treatment. Statistical significance for all comparisons was set at P < 0.05. Analysis was performed using SISA-Binomial by SISA software (Uitenbroek, Germany, 1997; available at http://home.clara.net/sisa Accessed Jan. 8, 2001).
Demographics. Demographics of our study population are listed in Table 1. Of the initial cohort, 93.3% (168) were included in the final calculations. The 12 excluded patients were lost to followup during the time period of the study. Thirty-five men (19%) and 145 women (81%) with a mean age of 52.8 ± 15 years were evaluated initially. The majority of patients had a diagnosis of RA (119 of 180; 66%) or juvenile rheumatoid arthritis (25 of 180; 14%). Of the other 36 patients, 17 were diagnosed with psoriatic arthritis and 4 with ankylosing spondylitis or spondylitic variant. One patient treated with etanercept had evidence of dermatomyositis, and 1 was thought to have adult-onset Still's disease. The remaining patients carried a diagnosis of undifferentiated inflammatory arthritis. Forty-two patients of the initial cohort had evidence of severe disease with one or more joint replacements.
|Age, years||52.8 ± 15.6|
|Rheumatoid arthritis||119 (66%)|
|Juvenile rheumatoid arthritis||25 (14%)|
|Psoriatic arthritis||17 (9%)|
|Ankylosing spondylitis||4 (2%)|
|Number with joint replacements||42 (23%)|
|Current use of|
Outcomes. The median duration of etanercept use was 10 months with a range of 1–19 months. The mean duration of therapy was 9.87 months. Our study cohort had 27% (46) taking etanercept for less than 6 months, 29% (48) taking etanercept for 6–12 months, and 44% (74) taking etanercept for longer than 12 months. Of the patients followed for the entire study, 73% (123) were taking etanercept for at least 6 months. Patients (n = 43) discontinued etanercept for a variety of reasons. Only 1 patient discontinued the medication following an injection site reaction. Three patients did not wish to take injections and discontinued for this reason, and 2 discontinued because their insurance would not cover the medication. Two patients discontinued the medication prior to elective surgery. Seven patients discontinued the medication due to pregnancy. Eleven patients discontinued due to lack of efficacy. Four patients discontinued the medication due to serious infections (Table 3). Three patients discontinued the medication after developing minor infections (i.e., upper respiratory infection) and 1 patient discontinued the medication after developing bronchiolitis obliterans obstructing pneumonia. A patient with a diagnosis of dermatomyositis discontinued the medication due to rising creatine kinase level. Four discontinued following a rash, 3 discontinued due to nausea, and 1 discontinued due to migraine.
Corticosteroid and DMARD use. Prior to treatment with etanercept, 112 of 168 patients (67%) were taking prednisone daily. Of this subset, 63 (56%) were able to taper or discontinue prednisone (Figure 1). Most of the study patients (101) were taking methotrexate at the time of etanercept initiation. More than half (n = 52) were able to decrease or discontinue methotrexate (Figure 1). Other DMARDs used concomitantly with etanercept included plaquenil, azathioprine, leflunomide, and cyclophosphamide (8 patients were taking one of these DMARDs in addition to etanercept, see Table 1). Six patients were taking sulfasalazine with etanercept. Many of these medications were subsequently tapered or discontinued.
Baseline adverse events. As previously mentioned, patients served as their own controls because we reviewed their charts before and after etanercept treatment. Thus, if a patient was treated with etanercept for 6 months, the 6 months prior to etanercept use was reviewed as a control. When reviewing this time period (control period) to obtain a baseline infection rate, 32 of 168 patients (19%) experienced minor side effects (i.e., upper respiratory infection, urinary tract infection, cellulitis) prior to beginning etanercept. Three control patients developed serious infections during the evaluation period, all requiring intravenous antibiotics. One was diagnosed with pneumonia, one with acute cholecystitis, and another with a deep-tissue foot infection.
Minor adverse events. Of the total number of patients followed (n = 168), 91 (54%) experienced a minor or severe adverse event (Table 2). The rate of minor adverse events was statistically greater than the rate experienced by patients prior to etanercept therapy (P ≤ 0.0001). Six patients discontinued therapy as a result of minor adverse events. The most common minor complications were injection site reaction or mild infection (upper respiratory infection or sinusitis). Six patients reported injection site reactions. Sixteen patients reported upper respiratory infection, cough, or sinusitis symptoms while taking etanercept. Five patients were evaluated for chest pain during treatment with etanercept. Rare patients experienced infections requiring oral antibiotics, such as cellulitis or bronchitis. Fourteen patients developed a skin rash that was self-limited, and 5 patients reported symptoms of depression.
|Adverse events||Baseline adverse events (%)||Treatment adverse events (%)||P|
|Minor||32 (19%)||86 (51%)||<0.0001|
|Serious||3 (1.8%)||5 (2.9%)||0.22|
Serious adverse events. Five patients developed serious infection while taking etanercept (Table 3). One patient developed acute cholecystitis, had an uneventful recovery after laparoscopic cholecystectomy, and ultimately continued therapy. The remaining 4 patients discontinued therapy upon diagnosis of the infection. These included a patient with a septic wrist caused by Staphylococcus aureus and a patient with an arthroplastic hip infection caused by MRSA (methicillin-resistant Staphylococcus aureus).
|Patient number||Age (years)||Duration of etanercept therapy (months)||Diagnosis||Comorbidities||Prednisone DMARDs||Serious adverse event|
|35||62||10||RA||IDDM||Methotrexate||Staphylococcus aureus sepsis with septic wrist|
|149||69||8||Psoriatic arthritis||—||Acute cholecystitis|
|163||76||12||RA||Myelodysplasia||Prednisone||Deceased; bacteremia following colonoscopy|
|170||41||14||Psoriatic arthritis||—||Methotrexate||MRSA, septic arthroplastic hip|
|179||75||12||RA||Asthma||Prednisone||Deceased; MAI, psoas abscess, intestinal perforation|
Statistical analysis reveals no significant difference in frequency of serious infections between the control group (patients used as their own control) and the study cohort (P = 0.22). However, 2 deaths related to infection were seen in patients taking etanercept. Subgroup analysis was also performed to assess the number of infections in patients who may have more severe disease (identified by joint replacement) or diagnoses other than RA. Patients who may have additional immunosuppression from corticosteroid or DMARD use were also evaluated for baseline infections and infections while taking etanercept. Subgroup analysis revealed no significant difference between controls and patients with diagnoses other than RA (P = 0.25), history of joint replacement (P = 0.49), or concomitant prednisone (P = 0.28) or methotrexate use (P = 0.31).
Given recent warnings regarding demyelinating disease and TNF inhibition, it is worthwhile to note that no patient was diagnosed with multiple sclerosis or another demyelinating disease during the study. In addition, there were no new diagnoses of malignancy or aplastic anemia in our patient population.
Two patients in our population died as a result of complications from sepsis associated with etanercept therapy. Patient 163 was a 76-year-old woman with a history of myelodysplasia, hypothyroidism, spinal stenosis, and RA. She had been taking etanercept for longer than 12 months in addition to her usual medications, which consisted of atenolol, furosemide, levothyroxine, rofecoxib, and prednisone (previously). Following colonoscopy with polyp removal, she developed bacteremia and died. Patient 178 was a 79-year-old woman with a medical history of chronic obstructive pulmonary disease, esophageal dysmotility, seasonal allergies, osteoporosis with multiple fractures, and RA for 19 years. Concomitant medications at the time of presentation included amitriptylline, calcitonin nasal spray, hydroxychloroquine, loratadine, nabumetone, oxycodone, and prednisone. Treatment with etanercept occurred for 12 months prior to presentation with a psoas abscess. Following surgical drainage, cultures grew Mycobacterium avium-intracellulare. Two months after her initial presentation, the abscess recurred. Postoperative course was complicated by a colonic perforation that was treated with a subtotal colectomy and ileostomy formation. She developed congestive heart failure, renal failure, and wound dehiscence. Wound cultures grew enterococci and Citrobacter. She continued to experience problems with oxygenation and worsening metabolic acidosis. At the request of the family, further aggressive measures were withheld, and she died on hospital day 45.
Comparison with other cohorts. We compared the infection-associated mortality rate seen in our patient population with the rates from previously published RA studies. It has long been recognized that increased mortality rates are seen in patients with RA compared with healthy controls. Striking increases from infectious causes of death are seen in RA patients (observed to expected ratio of 6.2 for patients followed for 35 years) (5). Other studies have estimated the mortality for RA patients from infectious causes to be 4.5% (45 of 1,000 patients) (3 of 67 patients), 2.4% (5 of 208 patients), or 2% (4 of 199 patients) (6, 7). Differences in the values obtained in different studies have been attributed to various causes, such as immunosuppression or time of presentation. We found no statistically significant difference comparing these populations with our study population (2 of 168). However, mortality rates may rise with duration of disease, and studies of longer duration may demonstrate higher mortality rates associated with infection in patients taking etanercept.
Long-term observation of patients in a clinical setting is an important adjunct to safety data generated in controlled clinical trials. A clinical practice setting differs from the clinical trial setting with regard to comorbidities, underlying diagnosis of the patient, and concurrent use of other medications. These differences may lead to the occurrence of complications that are not demonstrated in clinical trials. This has been illustrated with 3 recently developed medications. Following FDA approval, fenfluramine-phentermine, terfenadine, and cisapride were all removed from the market when postmarketing surveillance showed increased rates of morbidity and mortality thought to be associated with their use.
The role of TNFα in the immune response is very complex. Although some studies suggest it may limit the extent and duration of inflammation, a role for TNFα in susceptibility to infection is clearly suggested by data from animal models of TNFα inhibition and blockade (8). Studies in mice have shown that animals treated with soluble TNFα receptors are more likely than untreated animals to succumb to infection with Listeria monocytogenes (9–11). A role for TNFα in production and maintenance of granulomas has been proposed previously (8). Indeed, recent studies demonstrate TNF p55 receptor (TNFRp55) gene-deficient mice infected with Mycobacterium avium have increased mortality compared with immunocompetent mice. Newly formed granulomas are not maintained in the gene-deficient mice, and all die from infection early in the course of disease. In contrast, immunocompetent mice remain alive, though chronically infected, for many months after the initial infection (12).
Infection is a major cause of death in patients with systemic rheumatic disease. Both the underlying disease and recommended therapy may predispose to infection. Despite a high number of serious infections in our treated group, comparison with the control group (n = 168) revealed no significant difference. Subgroup analysis likewise revealed no obvious trends in susceptibility to infection. Significantly more patients reported minor adverse events following initiation of etanercept, including bronchitis, rash, and upper respiratory infections. Although this may represent reporting bias following initiation of a novel therapy, the effect of mild immunosuppression cannot be excluded. The 12 patients lost to followup, if added to our results, may reveal that the number of infections was underassessed. Further studies with larger numbers of patients followed for a greater length of time may reveal an increased risk of serious infection in patients taking etanercept.
Two deaths were related to infection while taking etanercept therapy. Neither patient had symptoms or physical exam findings to indicate active or chronic infection before starting etanercept. However, each had a secondary risk factor for potential immunosuppression (diabetes mellitus and myelodysplasia). These findings raise the issue of asymptomatic infections. More rigorous monitoring may be needed for patients prior to and during etanercept therapy, such as tuberculin (Mantoux) testing, chest radiographs, or serial complete blood counts. Use of entanercept in patients with comorbidities such as diabetes or myelodysplasia certainly requires caution. Patients with signs of active infection should discontinue etanercept and undergo aggressive treatment.
The results of this study demonstrate that a majority of patients were able to taper either methotrexate or corticosteroids while taking etanercept. Higher doses of corticosteroids are a well-known risk factor for adverse events. In patients requiring methotrexate or prednisone therapy, the incidence of side effects may be decreased if the dosage used is the lowest effective dosage. Patients responding to etanercept may reduce their risk of concomitant adverse events secondary to other drug therapies by decreasing corticosteroid or methotrexate use.
Current therapies for systemic rheumatic disease exhibit cumulative toxicity that limits their usefulness. Biologic response modifiers hold great promise in that they target specific cytokines involved in the inflammatory process. Solubilized TNFα receptors have been well tolerated in clinical trials. Indeed, the majority of our patients were able to tolerate etanercept for longer than 6 months. Such well-tolerated therapies are attractive to clinicians and will likely be increasingly used off-label for all rheumatic diseases. Etanercept is currently being prescribed for systemic inflammatory diseases other than RA. Characterization of such off-label use may be relevant when cataloguing adverse events in a clinical population, as risk:benefit ratios may be altered. The results of this study support further long-term observation under the conditions of a general rheumatology practice as a supplement to controlled clinical trials.
The authors wish to express their appreciation to Dr. Michael Weinblatt for his advice and recommendations in the preparation of this manuscript.
- 4Immune and inflammatory responses in TNF alpha-deficient mice: a critical requirement for TNF alpha in the formation of primary B cell follicles, follicular dendritic cell networks and germinal centers, and in the maturation of the humoral immune response. J Exp Med 1996; 184: 1397–1411., , , .