The results from less than half of early-stage cancer studies could be reproduced in a years-long project, the Scientist reports.
The journal eLife, the Center for Open Science, and Science Exchange launched their Reproducibility Project: Cancer Biology effort in 2014 with the goal of replicating the key findings from 50 high-profile preclinical cancer studies published between 2010 and 2012. Due to budgetary and other constraints over the years, the team had to pare back its goals and now reports in eLife that it attempted to replicate 50 experiments from 23 papers.
Through this exercise, the group led by Timothy Errington at the Center for Open Science found that most studies do not provide enough information for other scientists to replicate their work and that the original authors were "extremely" or "very helpful" for 41 percent of the experiments, but they were "not at all helpful" or did not respond for 32 percent of experiments. In a meta-analysis presented in a separate paper, Errington and his colleagues further report that they had a 46 percent success rate in replicating the earlier studies' findings.
"The truth is we fool ourselves. Most of what we claim is novel or significant is no such thing," Vinay Prasad from the University of California, San Francisco, who was not involved in the analyses, tells the Associated Press.
Michael Lauer, deputy director of extramural research at the National Institutes of Health, adds at the AP that he was disappointed in the number of scientists who were not helpful or didn't respond to requests for help and says that the agency will work to improve data sharing among researchers it funds.