Journal of the National Cancer Institute Advance Access originally published online on September 25, 2007
JNCI Journal of the National Cancer Institute 2007 99(19):1430-1432; doi:10.1093/jnci/djm181
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
© Oxford University Press 2007.
NEWS |
Unusual Comparison Between Trials Uncovers Key Genetic Polymorphisms
A standard rule of thumb in clinical trial design and analysis is that researchers should not compare results from two different trials, which can lead to spurious conclusions. Yet researchers in the Southwest Oncology Group (SWOG) have designed trials to do just that.
Using an unorthodox method, they have uncovered genetic differences that may partially explain why Japanese lung cancer patients have better outcomes in clinical trials than those of American patients. The team is now using similar trial designs to look for genetic differences that might account for the increased responsiveness of Japanese patients to drugs that block epidermal growth factor receptor (EGFR) activity.
Instead of enrolling Japanese and American patients in one trial, an impractical approach given the regulatory hurdles, the researchers designed three different but simultaneously run trials, each with an identical control arm, referred to as a "common arm." They then compared patient outcomes among the control arms in the three independent trials. "The common arm approach is a good strategy to allow for comparisons between trials and between populations," said David Gandara, M.D., director of the thoracic oncology program and associate director for clinical research at the University of California–Davis Cancer Center in Sacramento, who led the unusual new study.
|
But how reliable would the results be, given that they were derived by intertrial comparisons? Stephen George, Ph.D., chief of the division of biostatistics at Duke University and director of the statistical center for the Cancer and Leukemia Group B cooperative group, said it's not ideal but not a bad approach either. "It is about the best you can do under the circumstances," he said. "They are likely to be able to get some reliable answers out of it."
An International Conundrum
The trial came about because of the persistent difference between Japanese and American lung cancer patients. "In study after study, the Japanese patients do better, but they have more side effects," Gandara said. Regardless of the regimen being tested, Japanese lung cancer trials show a higher response rate, better survival, and more toxic effects than trials based in the United States or Europe. "We hypothesized this is due to tumor biology and to pharmacogenomic differences in the host between Japanese people and typical [white people]," Gandara said. But no data were available to directly test the idea. The best way to look for such genetic variation would be to enroll patients from both locations into one trial and look for an association between clinical response and single-nucleotide polymorphisms in the patients’ DNA or mutations in the tumor. The problem, however, is that the Japanese research groups are not currently eligible to participate in trials funded by the National Cancer Institute, making a joint trial difficult. (Some international research groups, such as the European Organization for Research and Treatment of Cancer, have completed the regulatory paperwork required for such joint projects, but the Japanese collaborative groups have not.)
"Since we couldn't do a trial with them, what I developed was the concept of a common arm," Gandara said. The approach has not been used before, as far as he and his collaborators know. At the time of the original discussions in 1999, SWOG was about to run a phase III lung cancer trial in which the control arm was paclitaxel–carboplatin. Gandara and his Japanese colleagues agreed that two Japanese phase III trials would use the same control arm. Therefore, the control arm in all three trials had identical eligibility, staging, and treatment requirements by prospective design.
The clinical results from those three arms followed the expected trend, with Japanese patients deriving greater benefit. One-year survival in the SWOG control arm was 37%, compared with 51% in the control arm of the Four-Arm Cooperative Study and 57% in the Japan Multinational Trials Organization control arm. Response rates were similar in the common arms, but more of the Japanese patients developed grade 4 neutropenia: 69% compared with 26% in the SWOG control arm.
At the American Society for Clinical Oncology annual meeting in June, Gandara reported that the trials had revealed some genetic polymorphisms that were statistically significantly associated with certain clinical outcomes—and that they were more common in the Japanese patients than the U.S. patients. The team obtained DNA samples from 78 patients in the SWOG arm and an equal number in the Japanese Multinational Trials Organization arm. They tested seven genetic polymorphisms that are known to alter either paclitaxel metabolism or DNA repair after exposure to platinum compounds, such as carboplatin. Of those, five were unevenly distributed in the two populations. In particular, two variants were associated with better response or progression-free survival.
"It is a relatively small study, but still it is very provocative," Gandara said. "We don’t look at it as definitive, but it shows that this is an important issue that needs to be studied further. No one has done this before" because the comparison requires such an unprecedented design.
Just how to view results from Japanese lung cancer trials has been a question for researchers and clinicians in the United States, according to Giuseppe Giaccone, M.D., Ph.D., chief of the medical oncology branch in the Center for Cancer Research at the National Cancer Institute. Now instead of ignoring them, researchers can see, because of Gandara's unusual approach, that there could in fact be biological differences that might account for some of the improvements in outcome typically seen. "Now we have to think differently and say ... the results that they have are real. It is not because they are doing the studies in a funny way. So we start understanding that there are differences in the way drugs are handled, beyond the differences in health systems and ways that people are treated," he said.
Gandara, however, points out that the results show that the differences are really due to differences in allele frequencies between the two populations. The alleles themselves predict response, regardless of whether the person carrying them is Japanese or American. "Genetics trumps race," he emphasized.
Examining a New Design
"Our common arm trials are just meant as a simple first step toward collaborations with the Japanese and to try to understand the similarities and differences in outcome and toxicity based on the same treatment," said John Crowley, Ph.D., CEO of Cancer Research and Biostatistics and director of the SWOG Statistical Center in Seattle, who oversaw the trial design and data analysis.
Although the common arms were part of independent studies, Crowley and Gandara tried to minimize variables wherever possible, including analyzing all the data in the SWOG statistics center.
They did a pretty good job of it, according to George, who was not involved in the trial. He noted that the SWOG group tried to compensate ahead of time for many of the issues that become pitfalls when people try to compare outcomes between trials after the fact. The trials were done at the same time, which minimizes the likelihood that supportive care, for example, would be different. They agreed upon definitions, eligibility, and staging. And they included the hard endpoint of 1-year mortality, which gets around the variability associated with assessing progression-free survival. "There is still a bit of an issue: Even with common eligibility and treatment criteria, you could still ask if the patients are really the same," he said. "Even if you took the same list [of eligibility criteria] and applied it in two countries, you might get different patients on the trial because of how people are referred in or because of practice differences. That is a known issue in clinical trials in general."
Despite the value of the common arm approach in this specific situation, George cautions that it is not likely to be appropriate in many other situations—and that the standard rule to avoid comparisons still sticks in almost all cases.
However, Gandara still wants to try the common arm approach in one more situation: looking for genetic differences that might help explain why Japanese lung cancer patients respond better to drugs that inhibit EGFR, such as gefitinib and erlotinib. Researchers already know that Japanese lung cancer patients are more likely to be never smokers, female, and have adenocarcinoma, all of which are associated with an increased likelihood of having a mutation in EGFR that makes it particularly likely to respond to such drugs. But Gandara wonders whether there are underlying genetic differences between the populations that lead to those characteristics: Are there, for example, EGFR alleles that are more likely to mutate and lead to cancer and then respond to anti-EGFR therapy? Are those alleles more common in the Japanese population?
In hopes of finding out, his team plans to meet with Japanese researchers this fall to hammer out the details for several single-arm phase II trials to be jointly run in the U.S. and Japan. If the trials go as currently planned, lung cancer patients will receive standard single-agent erlotinib therapy. The researchers will harvest tumor tissue and germline DNA and look once again for correlates of response. And they’ll do it by comparing between single-arm phase II trials run in two different locations but with identical trial designs.
"I think one should conclude that this is a positive way of doing things," Giaccone said. "It would be good if they had the same thing within the same study. Sometimes it is just impossible."
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||