Reports of studies of cancer prognostic factors are largely suboptimal, a new study concludes.
Researchers believe that certain characteristics of a tumor, such as the expression of a particular gene, may help doctors predict a patient's prognosis--how they respond to a certain chemotherapy drug or how likely they are to survive their cancer, for example. However, most proposed prognostic markers have not been validated and have not changed clinical practice.
Panayiotis A. Kyzas, M.D., Ph.D., of the University of Ioannina School of Medicine, Ioannina, Greece, and colleagues decided to examine the quality of reports of studies of cancer prognostic factors and see whether certain aspects of report quality were associated with particular conclusions about a tumor marker. Among 331 studies identified from 20 meta-analyses of cancer prognostic markers, the researchers didn't find any association between study quality and reported outcome. However, they did find several shortcomings: An important statistical test called a power calculation was rarely reported, most studies were too small to reveal modest effects of a particular tumor marker, many studies did not state that the analysis was blinded, and only a small percentage of studies were prospective.
"Our analysis indicates that the reporting of important quality in formation in cancer prognostic studies is often suboptimal," the authors write. "Given the potential clinical importance of this prognostic information, efforts to improve the design and reporting of these studies are warranted."
Note: The Journal of the National Cancer Institute is published by Oxford University Press and is not affiliated with the National Cancer Institute. Attribution to the Journal of the National Cancer Institute is requested in all news coverage. Visit the Journal online at http://jncicancerspectrum.oxfordjournals.org/.
Cite This Page: