FDA Meeting March 2003: Update on the Safety of New Drugs for Rheumatoid Arthritis.
Part II: CHF, Infection and other Safety Issues.

August 2003

The Bottom Line:

• TNF inhibition is not an effective treatment for CHF, and may even lead to worse outcomes. While data from patients with RA treated with TNF inhibitors do not seem to indicate an excess of CHF exacerbations, clinicians should carefully consider the potential risks and benefits of such therapy in RA patients with CHF or risk factors for CHF.
• TB is a concern with TNF inhibitor use. PPD screening for LTBI is indicated, as is vigilance with continued use.
• A few cases of hepatotoxicity have been reported with TNF inhibitors; clinicians should be aware of this, and may consider this in their assessment algorithms for patients.

Discussion

In animal models and human disease, the inflammatory cytokine Tumor Necrosis Factor (TNF) may contribute to the pathogenesis of congestive heart failure (CHF). Elevated levels of TNF in CHF patients may contribute to cardiac cachexia, fatigue, myocyte dysfunction and changes in myocardial matrix proteins.  Based on such data it was predicted that TNF inhibition would have a salubrious effect.  However, three large clinical trials of TNF inhibitors in CHF were prematurely halted for lack of benefit or adverse outcomes (1). 

Two randomized, placebo controlled trials (RENAISSANCE and RECOVER) with 2048 patients tested the potential benefit of etanercept (ETAN), given one, two or three times weekly.  CHF patients (mean duration 4.6 - 5.6 years) were predominantly male (78%), Caucasian (84-99%), 62-65 years old, with entry ejection fractions of 22-24%. Most were NYHA Class 3 (70-72%) or Class 2 (24-27%).  Both trials were stopped for “futility”, defined as <10% benefit on morbidity or mortality endpoints (CHF hospitalization, death, improvement in NHYA class, patient global assessment). The North American RENAISSANCE study was stopped after 12.7 months with a non-significant trend towards worsening in the ETAN group (Table 1). The multinational RECOVER study was halted after 5.7 months and showed no differences between placebo and ETAN patient outcomes. It is unknown if the premature discontinuation for futility may have masked potential cardiac risks due to ETAN therapy.  The number of patients treated and duration of exposure may mollify this concern.  Review of all RA clinical trials with ETAN (N=3389) shows new onset CHF in only two placebo and two ETAN treated patients.

The infliximab (INFLX) ATTACH trial was a North American 150 patient, placebo-controlled trial involving NYHA class 3 and 4 CHF patients (1,2).  Patients received 3 infusions over 6 weeks. This study was discontinued after a week 28 interim analysis showed a dose related increase in hospitalizations and mortality (Table 1). At week 28, CHF hospitalizations (hosp) and deaths were increased in the 10 mg/kg INFLX (21.6% hosp, 5.9% deaths) compared with placebo (10.2% hospitalizations, no deaths). When patients were followed for 54 weeks, four deaths were noted in the placebo and 5 mg/kg groups, but eight deaths were seen in the 10 mg/kg treatment group. These results prompted a change in product label and a Dear Doctor letter (10/01).  In contrast, analysis of over 1,600 patients in controlled clinical trials showed new onset CHF occurred in 0.2% of INFLX and 2.1% of placebo treated patients with RA and Crohn's disease.

During the adalimumab (ADA) FDA review, the incidence of new onset CHF among RA patients treated with ADA (N=1362) or placebo (N=683) was 0.1% and 0.7%, respectively (1).  Thus, in the randomized controlled trials of all three TNF inhibitors (> 7,300 patients) a total of 18 new onset CHF reports have emerged - 7 occurring with TNF inhibitors (<0.2%) and 11 with placebo (0.7-2.1%) . 

Review of spontaneous reports from the FDA Medwatch system revealed 158 spontaneous reports of CHF (3).  FDA review of 51 of these CHF reports (30 ETAN, 21 INFLX) revealed nine CHF exacerbations and 42 new onset cases, half of the latter having no prior risk factors.  The mean age was 64 years and the median time to onset was 3.5 months (range 1day-2 yrs). Three deaths occurred.  Ten patients (4 ETAN, 6 INFLX) developed CHF under the age of 50 years.  Three had known risk factors.  The median ejection fraction was 20%.  All stopped anti-TNF therapy: three resolved, six partially improved and one patient died. 

Mechanisms underlying these adverse cardiac outcomes or predictive factors are not apparent and contrary to expectations derived from animal and human research.  Moreover, patients with milder forms of CHF are not free of risk (based on the RENNAISSANCE study) and higher doses of ETAN (TIW) or INFLX (10 mg/kg) may pose unacceptable risks for CHF patients. Further label changes and product versuss class specific language for all 3 TNF inhibitors are under consideration by the FDA.  Recommendations may include caution when using TNF inhibitors in RA patients with CHF, especially if considering higher doses (>5 mg/kg INFLX or 25 mg TIW ETAN).

Tuberculosis. The association between TNF inhibitor use, tuberculosis (TB) and opportunistic infections was previously reviewed by the FDA Arthritis Advisory Committee (see Hotline Sept. 2001) (4).  During the March 4, 2003 FDA meeting these adverse events were updated, and additional information has since become available (5). Animal models demonstrating biologic plausibility, along with pharmacovigilance data on prevalence and temporal association, support the association of TB with TNF inhibitor therapy in RA patients.

Although primary infections have been seen, most of the TB cases appear to be reactivation of latent TB infection (LTBI). Screening for LTBI with PPD according to ATS/CDC guidelines is advised for patients being considered for treatment, and seems to have been effective in reducing the occurrence of TB in treated patients. There appear to be differential risks of TB with the currently available inhibitors, although the specific mechanisms underlying this difference remain to be delineated. Most cases of TB are still from outside the U.S. even though most use of the agents is in the U.S., highlighting the effect of local factors. Extrapulmonary disease is much higher among TB cases with all three TNF inhibitors than would be expected in the general population with TB (18%). The potential for false negative screening tests and the possibility of acquiring primary TB while on therapy underscore the need for continuous vigilance on the part of clinicians for the possibility of TB in patients treated with these agents.

Hepatotoxicity with TNF inhibitors. In the ATTRACT trial, mild to moderate LFT elevations (<3 x upper limit of normal [ULN]) were observed  more often in the INFLIX (37%) than placebo (29%) groups (all groups received background MTX).  In a Crohn's disease (CD) study (ACCENT I) all patients were maintained on background immunomodulatory therapies (e.g., 6MP, azathioprine, mycophenolate, or  MTX) and were then randomized to receive repeated doses of placebo, 5 mg/kg or 10 mg/kg of INFLIX.  Hepatic enzyme elevations were more common in the 5 or 10 mg INFLIX (42%) group compared with placebo (36%) treated CD patients. Moderate LFTs elevations (>2 but <3 times the ULN) were seen in 24 INFLX (5mg or 10 mg/kg) patients, but only in eight placebo patients.  None went on to develop liver impairment. In the ERA study, although most LFT elevations were seen in the MTX group, four patients in the ETAN 10 mg group (none in the 25 mg group) developed mild to moderate LFT elevations.  Lastly, in the ADA clinical trials, nine ADA treated patients (<4%) developed > 2 fold elevations of AST or ALT.  Four of these remained elevated during the trial. Many of these were confounded by background MTX or DMARD use and the frequency was similar to placebo controls. One patient with a history of fatty liver died while receiving ADA, but was never noted to have an increased AST or ALT in the trial. These data suggested that mild to moderate elevations of hepatic enzymes may occur with TNF inhibitor use.   

FDA post-marketing surveillance has disclosed 134 spontaneous reports of liver failure associated with TNF inhibitor use.  In their detailed review of 50 well-documented cases (31 INFLIX, 19 ETAN) they found confounding diagnoses or hepatoxin exposure in 43 cases (13 sepsis, eight tuberculosis or INH use, five alcohol related, three viral hepatitis, two GVHD, six hepatotoxic drugs, eight other).  However, in seven cases (five INFLIX, two ETAN) no other cause could be identified suggesting that TNF inhibitor use may have lead to hepatic failure.  It was noted that the chance occurrence of liver failure in the general population is estimated to be one per million population.  Clinicians should be aware of these rare events and report similar findings to the FDA at www.medwatch.com.

Arthur Kavanaugh, MD, John J. Cush, MD, Eric Matteson, MD, Hotline Editors - August 2003

Drs. Cush and Kavanaugh have received research grants from and consulted for Abbott, Amgen and Centocor, Inc. Dr. Matteson has received research grants from and consulted for Amgen and Centocor, Inc.

References

1. FDA Center For Drug Evaluation and Research Arthritis Advisory Committee Safety Update TNFα Blocking Agents (Transcripts)
2. Chung ES, et al.   Circulation. 107:3133-40, 2003
3. Kwon HJ, et al.  Ann Intern Med. 138:807-11, 2003.
4. Cush JJ. Matteson EL. Hotline September 2001 - FDA Advisory Committee Reviews Safety of TNF Inhibitors
5. Leff J.  Amgen personal communication.

Hotline is provided by the ACR Communications and Marketing Committee as a service to members. This Hotline reflects the views of the authors, and does not represent a position statement of the College.

© 2003 American College of Rheumatology

Previous | Index | Next