| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | COVER ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
||||||||
|
||||||||||
| ||||||||||||||||||||||||||||||||||||||||
NEWS & VIEWS |
|
Clinical trials are vital to continually improving the efficacy of interventions for cancer prevention, detection, and treatment, and a new study published in the Archives of Internal Medicine (2008;168:632–642) underscores their value.
The study, led by Benjamin Djulbegovic, MD, PhD, of the H. Lee Moffitt Cancer Center and Research Institute at the University of South Florida, Tampa, assessed success rates among National Cancer Institute-funded Phase III randomized controlled trials (RCTs) conducted by cooperative oncology groups since 1955. After pooling data from 781 randomized comparisons involving 216,451 people in 624 trials, the researchers found 25% to 50% of the new treatments tested in these trials were successful, numbers Djulbegovic says should highlight "the need to emphasize to patients that they are intimate partners in making discoveries about how to treat cancer."
"We can't find out if one treatment is better than another if patients aren't involved. Somehow more patients need to realize how important they are to the process," said Djulbegovic. Part of the answer is to do a better job of explaining how clinical trials work and to help patients better understand the principles of uncertainty involved, he says.
Djulbegovic and colleagues evaluated clinical trial success rates using a multidimensional approach, controlling for other factors typically affecting clinical trial results: publication bias, quality-assessment accuracy, and comparator use.
They analyzed data from National Cancer Institute-funded Phase III RCTs completed between 1995 and 2000 and used 3 separate outcome measures for determining whether the standard or experimental treatment was superior in a given trial. Studies that looked only for equivalence between treatments were excluded. Results of 90% of the trials included in the study had been published, and the researchers were able to obtain data for almost one-third of the unpublished studies. The research team calculated the proportion of conclusive and inconclusive trials, as well as the proportion of studies that led to a "breakthrough intervention," defined either as a therapy so beneficial it immediately becomes the new standard of treatment or reduced the death rate by 50% or more.
Of 781 randomized comparisons, the researchers were able to assess 743 for statistical significance and found 30% of the results (n = 221) to be statistically significant. Of these, 80% (n = 176) favored the experimental treatment, and 20% (n = 45) favored the standard.
However, according to Djulbegovic, statistical significance of survival and other primary outcomes only goes so far in assessing a trial's success. Focusing on the success of a trial's primary outcome makes it harder to see the value of alternative yet equally effective treatments.
"Treatment has other dimensions. It's important to supplement statistical significance with other value judgments to get the whole picture," he said. As an example, Djulbegovic noted in an interview "the introduction of lumpectomy instead of mastectomy, which resulted in dramatic improvement in quality of life of women with breast cancer."
To assess subtler issues involved in identifying successful trials, such as weighing benefits or harms of different treatments for an individual, Djulbegovic and his team evaluated the published judgments on trial outcomes as written by the original researchers. Of 766 comparisons, 41% (n = 316) of the new and promising therapies were considered superior by the trial investigators, and 59% (n = 450) of the standard treatments were found to be better.
When pooled data were compared, the new treatments on average demonstrated a 5% reduction in the death rate, and in 2% of cases, the death rate fell by more than 50%. Survival gains were highest among those cancers associated with advances in adjuvant and intensive chemotherapies, such as gastrointestinal cancer and hematologic malignant neoplasms.
The researchers also looked at whether there were differences among treatment effects during 4 different time periods and performed a time-series analysis to determine whether success rates were affected by the results of preceding trials. The time-series analysis found no significant correlation between studies at various time intervals.
"Society has received a good return on its investment in the cooperative oncology group system," the authors write. They argue that success rates could improve if the number of inconclusive trials was reduced. According to their data, 218 of the 743 randomized comparisons (29%) were inconclusive, in large part, the authors say, because the clinical trial investigators' overly optimistic predictions of the size of treatment effects led them to underestimate the number of subjects needed to achieve statistically significant differences between outcomes of the standard or experimental treatments.
| ||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | COVER ARCHIVE | SEARCH | TABLE OF CONTENTS |
