"Surrogate Endpoints" In Clinical Testing – Are We Being Duped?

Researchers in the field of immunotherapy haveClinical trials in cancer patients are evaluated by
increasingly held out hope for therapeutic vaccines.criteria determined by either the World Health
However, a position paper published in the EuropeanOrganization or RECIST (Response Evaluation Criteria
Journal of Cancer, May 2009,(1) questioned thein Solid Tumors) which define objective response
validity of vaccine trials and called for a criticalrates as 30% to 50% decrease in lesions with no
appraisal of how the outcomes are determined andappearance of new lesions. "Mixed responses" in
understood.which some tumors decrease while others increase
Therapeutic vaccines, unlike prophylactic vaccines, aredo not qualify as objective responses in either set of
designed to be administered after the onset ofstandard criteria. An example of deviating from the
disease so as to trigger an immune system response.standard criteria was seen in a study by Spanknebal
This emerging type of biological therapy is still mostlyet al, which reported "10 objective responses in 46
experimental, thus the vaccines are only available toevaluable patients after one course of IL-2 therapy."
patients enrolled in clinical studies.However, five of those were mixed responders
Steven Rosenburg, M.D., Surgery Chief at themeaning the true objective response rate was not
National Cancer Institute (NCI) in Bethesda, Maryland,22%, but rather, 11%.(4)
argues that optimism about the clinical application ofA classic case of misleading surrogate endpoints
therapeutic vaccines is unjustified. His research teamoccurred when studying the effect of ventricular
noted that cancer vaccine trials in 440 patients,arrhythmia after myocardial infarction. Arrhythmias
conducted at NCI Surgery Branch, had an overallare a known risk factor for sudden cardiac death and
objective response rate of only 2.6%. Much lowerthe drugs, encainide and flecainide are quite effective
than the 4.0% response rate reported in the 40in suppressing them. Consequently, almost half a
studies that involved 756 patients. The teammillion patients in the US started receiving those
concluded that the more favorable study resultsdrugs for that purpose. Researchers convinced
were based on surrogate end points rather than oncardiologists that it would be unethical to withhold
proof of anti-tumor effects.(2)these drugs from patients in the clinical study.
A surrogate endpoint is a "marker" – or measureFortunately, a controlled study of encainide and
of effect of a particular treatment that may or mayflecainide took place. Amazingly, the results showed
not correlate with a definitive endpoint that willthat while both drugs suppressed arrhythmias
validate the study. Oftentimes, the primary endpointeffectively, the actual death rate tripled! Yes, the
of a study is an undesirable outcome – a fulldrugs had a positive effect on arrhythmias, but they
manifestation of disease or even death. So thealso had an unintended and previously unrecognized
research team will zero in on "markers" that canadverse effect on overall survival.(5) Surrogate
serve as an alternative to test the benefit of aendpoints do have a place in clinical study but it
particular treatment.appears they should be used where they perform
The problem arises when the surrogate endpointbest – in preliminary screening and phase 2 trials.
"correlates" with the actual endpoint but doesn't haveDefinitive phase 3 trials should provide reliable
a guaranteed relationship. For instance, tumorevidence based on true objective clinical response.
response is frequently used as a surrogate end pointSources:
in therapeutic trials of cancer. The categories are:1. Eggermont, AMM, "Therapeutic vaccines in solid
"complete response" (tumor not visible ontumours: Can they be harmful?", Eur J Cancer. 2009
examination), "partial response" (a reduction in tumorMay 22
volume of 50% or more), and "no change of2. Eggermont, AMM, "Therapeutic vaccines in solid
progression." Unfortunately, the measure of tumortumours: Can they be harmful?", Eur J Cancer. 2009
response is NOT a determinate of overall survival.May 22
Which means that tumor response "correlates" to the3. Moertel CG. "Improving the efficiency of clinical
subject at hand but has no real effect on mortalitytrials: a medical perspective", Stat Med. 1984; 3:455-68
rates.(3)4. Rosenberg SA, Yang JC, Restifo NP. "Cancer
The fact is, early-stage clinical trials usually involveimmunotherapy: moving beyond current vaccines",
only a small number of patients, for which the resultsNat Med 2004; 10:909–15
are not always sustained in larger trials. Rarely is the5. Fleming, Thomas R, "Surrogate Endpoints And
validity of a surrogate endpoint rigorously establishedFDA's Accelerated Approval Process", Health Affairs,
even though deviation from standard accepted24, no.
criteria leads to considerable confusion.