Unusual toxicities with TNF inhibition: Heart failure and drug-induced lupus J.J. Cush John J. Cush, MD, Chief, Rheumatology ABSTRACT demyelinating disease, cytopenias, auto- and Clinical Immunology, Presbyterian Serious and unexpected adverse events, immune disorders, and heart failure. Hospital of Dallas, Clinical Professor of such as heart failure and drug-induced Other essays in this volume address the Internal Medicine, The University of Texas lupus, have been reported in patients subject of infection (10), demyelination Southwestern Medical School, Dallas, Texas, USA. receiving TNF inhibitor therapy. These (11), and lymphoma (12). This review events generally are easily recogniz- will focus on heart failure and lupus- Please address correspondence to: John J. Cush, MD, Presbyterian Hospital able, although they cannot be predicted like disease, which are rarely seen in of Dallas, 8200 Walnut Hill Lane, Dallas, nor avoided, other than by drug avoid- association with TNF inhibitor therapy Texax 75231-4496, USA. ance altogether. Many patients have and are currently without explanation. Clin Exp Rheumatol 2004; 22 (Suppl. 35): great benefit from anti-TNF therapies. S141-S147. Their intelligent use requires a firm Sources of data © Copyright CLINICAL AND EXPERIMENTAL understanding of these rare toxicities, As a consequence of drug approval, RHEUMATOLOGY 2004. so as to minimize the morbidity associ- manufacturers have a responsibility to ated with their uncommon occurrence. collect safety data actively and report Key words: TNF inhibitors, conges- these data to the appropriate agencies. tive heart failure, drug-induced lupus, Introduction For example, in the USA a manufactur- rheumatoid arthritis. The introduction of biologic agents to er must submit periodic safety reports specifically inhibit the pro-inflammato- to the FDA twice yearly (March and ry cytokine, tumor necrosis factor (TNF), September). Moreover, other parties has transformed traditional treatment are also focused on the safety of TNF paradigms for a wide range of inflam- inhibitors. These include the drug man- matory disorders. Many patients with ufacturer (each having its own global rheumatoid arthritis (RA), Crohn's coli- safety/pharmacovigilance office), regu- tis, psoriasis and a variety of spondyl- latory agencies (e.g., FDA, EMEA [Eu- oarthropathies have been afforded new- ropean Agency for the Evaluation of found disease control, especially when Medical Products]), external individu- traditional agents have failed, were only als, researchers, clinicians and patient partially effective, or were not tolerat- advocacy groups which may report ad- ed. After 6 years of use, these drugs have ditional safety data or have access to been administered to nearly 700,000 publically available drug safety reports. patients worldwide, although only a When analyzing the safety of TNF inhi- minority of patients with inflammatory bitors, data are collected from the FDA's diseases have received them. The re- Medwatch/AERS (Adverse Event Re- strained growth of this class of therapy porting System), specific freedom of is affected by high costs, limited num- information (FOI) requests for safety bers of ideal candidates, lack of clear reports, review of the medical litera- guidelines for use and a continued con- ture, and scientific abstracts and com- cern for their long-term safety (1,2). posite data derived from national and The most common toxicities seen with international patient registries (9). these agents were defined during con- Each of the manufacturers of TNF in- trolled clinical trials and drug develop- hibitors have committed itself to multi- ment (3-5). However, following FDA ple, large on-going open-label safety approval, post-marketing surveillance registries to further discern the true has disclosed a number of rare or un- incidences of rare or unusual events. Of- usual signals that underscore the need ten causality can be difficult to define, for further large, population-based stu- as many of these events could be as- dies concerning the safety of TNF in- cribed to the disorder under treatment, hibitors (6-9). These rare and unusual rather than to the treatment alone. A events include opportunistic infections, prolonged period of at least 5-10 years serious infections, possible lymphoma, is required to distinguish between two S-141 Heart failure and drug-induced lupus in TNF inhibition / J.J. Cush Table I. Adverse outcomes in CHF treated with TNF inhibitors+ (8). Study name RENAISSANCE RECOVER ATTACH TNF inhibitor Etanercept Etanercept Infliximab N (NYHA Class) 925 (NYHC 2-4) 1123 (NYHC 2-4) 150 (NYHC 3-4) Regimens Placebo, Placebo Placebo 25 mg BIW, TIW 25 mg BIW 5 mg/kg or 10 mg/kg @ weeks 0, 2 & 6 Study duration Median 12.7 mos. Median 5.7 mos 28 weeks Death rates* Placebo 14.2% Placebo 8.8% Placebo 5 mg/kg 10 mg/kg ETAN BIW 17.9% ETAN QWK 5.9% @wk 28 0 2.0% 5.9% ETAN TIW 19.8% ETAN BIW 7.2% @wk 54 8.2% 8.0% 15.7% CHF* NA NA Placebo 5 mg/kg 10 mg/kg hospitalization @wk 28 10.2% 6.0% 21.6% Concerns ? TIW dose group more problems. 10 mg/kg @ higher risk NYHA Class II are NOT at lower risk +All studies were prematurely halted for futility or poor outcomes; *at the end of study. large populations (>10,000) of treated creased circulating levels of TNF in pa- had entry ejection fractions of 22-24%. and untreated patients, and these com- tients with congestive heart failure Most were NYHA Class 3 (70-72%), parisons are confounded by the fact (CHF), especially in those with NYHA with some being NYHA Class 2 (24- that patients who are treated with bio- class III and IV disease that is associat- 27%) These studies were discontinued logic agents are likely to have a more ed with severity and higher mortality after 12 and 5.7 months respectively severe clinical status. Finally, when risk. Transgenic mice which overex- for "futility"; defined as less than 10% post-marketing analyses yielded unex- press TNF have dilated cardiomyopa- benefit in morbidity or mortality end- pected or serious safety signals, the FDA thy and premature death. These data points (CHF hospitalizations, death has twice initiated public proceedings led many investigators to an a priori rates, improvement in NHYA class, to examine the safety of TNF inhibitors assumption that TNF inhibition would patient global assessment). in clinical practice. These proceedings improve cardiac function and survival While no benefit was seen, several pa- have been published elsewhere as part in patients with CHF (11). tients did experience adverse cardiac of the American College of Rheumatol- Clinical trials with both etanercept and outcomes, even those with NYHA class ogy (ACR) Hotline series (6-9). infliximab were initiated in patients 2 disease. The multi-national RECOV- with heart failure who did not have ER trial did not show benefit or wors- TNF and heart failure arthritis or inflammatory diseases (14- ening using conventional dosing of 25 TNF has been presumed to play a cru- 16). After 6-12 months of study these mg weekly or twice weekly. The North cial role in the pathogenesis of heart trials were halted prematurely due to American RENNAISSANCE trial show- failure and cardiac cachexia (2, 13). the absence of effect and/or poorer out- ed a non-signifcant trend toward more Systemic effects of TNF, IL-1 and oth- comes (8, 9). Reasons for the absence flares of CHF and hospitalization in er proinflammatory cytokines are well of anticipated benefit with either drug, those treated with higher doses of etan- known causes of anorexia, cardiac ca- or for the paradoxical outcomes seen in ercept, 25 mg thrice weekly. It is not chexia, endotoxic shock, etc. These cy- patients on higher doses of etanercept known if the premature cessation of tokines may also have untoward vascu- or infliximab, are not understood. these trials may have masked further lar effects by promoting the expression Table I summarizes the clinical trials cardiac risks associated with etanercept of adhesion molecules and alterations that were reviewed at a March 2003 therapy. in blood flow (13). TNF has been shown FDA conference concerning the safety Infliximab was also studied in a pilot, to have negative inotropic effects on of TNF inhibitors (8,9). Etanercept was phase II trial of 150 RA patients (8, 9, the myocardium and may further con- studied in two randomized controlled 16). This North American, placebo- tribute to left ventricular dysfunction trials (RENNAISSANCE and RECO- controlled trial involved NYHA class 3 and cardiomyopathy. TNF also causes VER trials) involving 2,048 patients and 4 CHF patients. Patients were ran- myocyte dysfunction and death, and with NYHA class 2-4 heart failure (14, domized to receive 3 infusions of pla- may cause myocardial fibrosis and al- 15). These patients had a history of cebo or 5mg/kg or 10mg/kg infliximab teration of myocardial matrix proteins CHF for a mean of 4.6 to 5.6 years, over 6 weeks and were then followed. to contribute further to the myocardial were predominantly male (78%), Cau- This study was discontinued after an in- failure. Several studies have shown in- casian (84-99%), 62-65 yrs old, and terim analysis at week 28 (TableI) indi- S-142 Heart failure and drug-induced lupus in TNF inhibition / J.J. Cush cated that CHF hospitalizations and adalimumab treated patients (N=1362) identified incident cases of heart failure deaths were higher in the 10 mg/kg in- and 0.5% in placebo-treated patients (N in 13,171 RA patients receiving TNF fliximab treated group (21.6% hospital- = 683). Therefore, in the randomized inhibitors and other DMARDs (17). In izations, 5.9% deaths), compared with controlled trials of all 3 TNF inhibitors their study, heart failure was more the placebo treated group (10.2% hospi- involving more than 7,300 patients, a common in RA patients (3.9%) than in talizations, no deaths).