The degree of risk experienced by cancer patients in Phase I trials appears to have been reduced since 1991, according to a new study.
Phase I studies are designed to identify a therapeutic dose for a new compound and represent the first use of a drug in humans. Thus, participation carries some risk for patients and only a limited potential for benefit.
To get an idea of how large the risk actually is, a group of Massachusetts researchers designed a retrospective analysis of Phase I trials that tested novel compounds and that were presented at the ASCO Annual Meetings of 1991 to 2002.
The investigators found that the death rate due to drug-related toxicities was 0.54% in the 213 studies analyzed, which included data on 6,474 cancer patients.
Furthermore, when the team divided the trials into three groups based on the year the study was presented at the meetings—1991 to 1994, 1995 to 1998, and 1999 to 2002—they found that there was a significant drop in both the overall death rate on study and the rate of treatment-related deaths between the earliest and latest time periods: The treatment-related death rate was 1.1% in the studies reported between 1991 and 1994 and 0.6% for those between 1999 and 2002.
In the analysis, published in the Journal of the American Medical Association (2004;292:2130-2140), the overall death rates declined during each of the three periods as well, from 3% in the first period to 2.2% in the second period, and then to 1.0% in the most recent. Of the 137 deaths from any cause, 35 (26%) were reported by study investigators as being treatment-related.
The team was surprised, however, to find a simultaneous drop in the overall response rate, said the lead author, Thomas G. Roberts, Jr., MD, MSocSci, an instructor of medicine at Harvard Medical School and attending oncologist at Massachusetts General Hospital Cancer Center.
The overall response rate was 3.8% during the period of 12 complete years. Yet, the response rate was 6.2% in the studies reported between 1991 and 1994, 2.6% between 1995 and 1998, and 2.5% from 1999 to 2002.
Changing Face of Phase I Trials
Dr. Roberts and his colleagues hypothesized that the decline in death rates during Phase I studies is the result of multiple factors, including an increased oversight by regulatory agencies and internal review boards after a couple of highly publicized unfortunate deaths of patients on study.
Additionally, there have been dramatic improvements in supportive care during the 12-year period covered by the analysis, with widespread use of new antiemetics and hematologic growth factors.
Moreover, the team noted, an increasing proportion of the drugs entering the clinic are biologically targeted therapies, as opposed to cytotoxic agents. By definition, these drugs should have a better safety profile, since they are designed to disrupt the function of one or a few proteins, rather than general cellular processes, like DNA replication.
Safety with biologics may also be improved because researchers aren't necessarily aiming for the maximally tolerated dose of a drug, but rather want to identify a dose that blocks the target's function but doesn't have broad effects on the cell.
“The idea used to be to get through Phase I, set a dose, and move on to Phase II and III,” explained Howard Burris, III, MD, Director of Drug Development at the Sarah Cannon Cancer Research Center, who was not one of the authors.
“But now a few hundred patients at the Phase I level may be able to tell you what you need to know, if you really do the biopsies that tell you whether you've inhibited your target.”
These trials, if done right, are better for the patients too, he continued. “There is nothing worse than taking Phase II patients, who are generally minimally treated, and exposing them to the wrong dose, be it too high or too low, and causing toxicity or not getting activity.”
However, incorporation of such molecular data is lagging behind in trial design, according to Dr. Burris and a number of other experts.
A recent report in the Journal of the National Cancer Institute by researchers in the National Cancer Institute of Canada Clinical Trials Group at Queen's University in Kingston, Ontario, found that out of 60 Phase I studies involving targeted therapies for the treatment of solid tumors, 36 relied on toxicity to set the dose for Phase II studies, and only two trials used information on the biological response of the targeted protein or a surrogate marker as the primary measure for dose determination.
Response Rate Drop Reflects Changing Population, Drugs
Although the decline in deaths on study is good news for both physicians and their patients considering participation in Phase I trials, the concomitant drop in response rates may be worrisome.
However, Dr. Roberts and his colleagues suggest that the reasons for the decline may not be cause for concern when examined more closely.
There has been a move in the past decade to rely more on outside evaluation for response rates, engaging otherwise uninvolved radiologists to review investigator-reported responses. That shift in methodology might account for some of the decline in overall response rates reported, since outside reviewers tend to come up with lower numbers than the physician investigators in a trial do.
Also, responses are traditionally measured as a decrease in tumor size, but many biologically targeted agents are cytostatic rather than cytotoxic. Thus, they may not induce tumor shrinkage even though the patient is deriving benefit from the therapy.
“We need to do a better job of evaluating these agents rather than just determining the typical response by CT scans,” said Dr. Roberts. “Now that we've shown that these trials are safe, we have to concentrate on gleaning more information from the trials and increasing the chance of patient benefit.”
“All evidence suggests that traditional chemotherapies have a monotonic dose response, so the old way of doing things [using a maximally tolerated dose to set Phase II dosage] was probably okay,” said Ian F. Tannock, MD, PhD, of the Department of Medical Oncology and Hematology at Princess Margaret Hospital and the University of Toronto, coauthor with Eric X. Chen MD, PhD, of an editorial that accompanied Dr. Roberts' study.
“But with biological agents, response may not be a good endpoint. They may not have a monotonic dose response, and you don't know where you are on the dose-response curve unless you do a biopsy test, preferentially in the tumor itself.”
Without this information, Dr. Tannock continued, one cannot determine if a failure to see a response is due to a real drug failure or failure to get on the right part of the activity curve.
The decline may also reflect the downside to an overall improvement in the cancer field. Over the last several decades more therapies have become available to patients. That means, however, that patients entering a Phase I trial are likely to be more heavily pretreated and farther along in their disease course than they would have been 12 years ago.
The exposure to a wide variety of drugs prior to enrollment in a trial may mean that their tumors are resistant to new therapies, which would result in lower response rates overall.
Dr. Roberts and his coauthors also note that the drop in response rates might be partially an artifact of their methodology. Because hematologic cancers have a variety of response criteria, the researchers chose to exclude these studies from their retrospective analysis.
Yet there have been a variety of very successful drugs developed for blood cancers during the 12-year period studied, including imatinib. Thus, the researchers might be reporting an artificially low response rate, compared with all Phase I trials involving novel agents.
To focus the evaluation on the safety of trials testing novel drug, Dr. Roberts and his colleagues did not include combination therapies, drugs that had been previously approved by the FDA for any indication, or studies that involved radiation.
Approximately half of the trials used biological agents, rather than cytotoxics. However, as Drs. Tannock and Chen note in their editorial, the researchers did not discuss what proportion of trials in each four-year period involved cytotoxic or biologic agents, which might have clarified some of the changes in both safety and response rate.