Recent statistical analyses of data from a Phase II trial ofImmunex Corp.'s soluble tumor necrosis factor receptor (TNFr)have confirmed that the drug harmed patients with sepsis whoreceived medium and high doses of the drug.

A biostatistician retained by Immunex concluded that thenegative results could not be fully explained by baselineimbalances between study groups, and therefore represented astatistically significant relationship between administration ofthe drug and death.

The new finding raises disturbing questions about furtherstudies using similar drugs and has cast a shadow over thetherapeutic strategy of blocking a patient's natural immuneresponse during sepsis. The unexpected and unfortunateresults also serve as a sobering reminder of the risks inherentin testing experimental drugs. Last year fialuridine (FIAU), anagent to treat hepatitis B, killed five patients in clinical trials.

Immunex of Seattle announced the results of the TNFr trial lastOctober without providing details. The company alsoannounced at the time that it was discontinuing furtherdevelopment of TNFr for sepsis, although it will test the drug inother immune-related diseases.

Janet Wittes of the Washington, D.C.-based statistical consultingfirm Statistics Collaborative performed the analytical work onthe TNFr data. Her analyses and the detailed TNFr results werereported last week at the Third International Congress onImmune Consequences of Trauma, Shock and Sepsis in Munich,Germany. Jerald Sadoff, director of the division ofcommunicable diseases and immunology at the Walter ReedArmy Institute of Research in Washington, D.C., presented thedata.

The Immunex TNFr trial enrolled 141 patients in four differenttreatment groups to study the safety and efficacy of TNFr atdifferent doses. Mortality rates for the double-blinded trialwere as follows:

No. patients who died/Total no. patients in group (mortalityrate)Placebo: 10/33 (30%)Low Dose TNFr: 9/30 (30%)Medium Dose TNFr: 14/29 (48%)High Dose TNFr: 26/49 (53%)Statistical significance of dose-related trend toward mortality:p=0.016.

Possible Explanations

Immunex's associate medical director of clinical research, JanAgosti, on Tuesday warned that a statistically significantfinding may not always be clinically significant. Agosti said thestudy was too small to show statistically significant imbalancesbetween groups but that clinical differences were significant.She called the statistician's verdict "an important observation,"but said it must be interpreted with clinical knowledge.

Steven Opal, associate professor of medicine at BrownUniversity in Providence, R.I., and one of the principalinvestigators of the Immunex Phase II TNFr trial, agreed thatthe small size of the Immunex TNFr trial makes it difficult toplace complete faith in the statistical findings. "There couldhave been a randomization imbalance not apparent to thestatistician but true nonetheless," he said. "However, one is leftwith the strong suspicion that there was a peculiar toxicitywith this therapy in some patients."

According to Walter Reed's Sadoff, patients in the medium- andhigh-dose treatment groups had a higher number of organfailures (such as lung injury) due to sepsis than patients in theplacebo group. Since organ failure is one measure of diseaseseverity, such an imbalance might explain why so many moreof the treated patients died. He also said that in the study,patients with Gram-positive infections suffered highermortality rates than those with Gram-negative infections.(Gram-negative and Gram-positive are terms for the two broadclasses of bacteria). Since there were more Gram-positiveinfections in the treatment groups than the placebo group, suchan important imbalance might have affected outcome.

But Sadoff conceded that his observations were at odds withthe statistical conclusion that imbalances between the groupsalone did not adequately explain the harmful effect of TNFr. Hesaid he had not yet had an opportunity to discuss the issuewith Wittes.

Anti-Cytokine Approach Questioned

Some experts have questioned the strategy of altering apatient's natural and complex immune response to infection.

Cytokines, such as TNF and interleukin-1 (IL-1), are proteinsthat activate and are activated by immune system cells such asneutrophils and macrophages. They are part of the complexlanguage that allows cells to alert one another and coordinatean attack on foreign matter that enters the body. Whenbacteria invade a patient's blood, the theory goes, cytokines arereleased in excess and produce tissue injury in the host.

While cytokines appear capable of causing serious harm (thusthe strategy of inhibiting them appears logical), there is alsoevidence to suggest that they are a crucial part of the body'sdefense against infection.

The problem is, no one is certain when in the sepsis cascadecytokines change from being beneficial to harmful. Blocking apatient's TNF or IL-1 (or the receptors that appear to regulateTNF and IL-1) at the wrong time could impair a patient'sability to fight off infection. Likewise, inhibiting a cytokine atthe right time might theoretically save the patient from severeorgan damage or death.

"There is a potentially important message here for anti-cytokine therapies and for anti-TNF therapies in particular,"said Brown University's Opal. "Some have argued that a certainamount of TNF is necessary for a patient to deal with infection.If you accept that, then this data has disturbing and disquietingimplications for the entire anti-cytokine field."

Ethical Questions

The TNFr data may raise some thorny ethical questions forHoffmann-La Roche Inc. of Basel, Switzerland, which hasalready begun enrolling patients in an international,multicenter Phase I/II clinical trial to test its soluble TNFreceptor -- with a proposed mechanism of action similar toImmunex's drug -- in patients with sepsis. Genentech Inc. ofSouth San Francisco, Calif., will supply the drug, known asTNFr-IgG, for the clinical trials.

A spokeswoman for Hoffmann-La Roche said its TNF receptorhas different biological properties than Immunex's drug and istherefore not expected to run into the same problems.However, some researchers, including Agosti, refutedpurported differences between the two drugs, saying there areno clear scientific data to prove claimed disparities betweenthe two drugs' affinity for TNF.

-- Lisa Piercey Washington Editor

(c) 1997 American Health Consultants. All rights reserved.