A commentary published in Nature says that at least 90 percent of preclinical cancer research studies published in science journals are not reproducible, and therefore wrong.
C. Glenn Begley, former head of cancer research at Amgen, and Lee M. Ellis, a University of Texas cancer researcher, say that their analysis of 53 published papers describing preclinical trial results found that just six, or 11 percent, contained reproducible results.
The researchers reviewed preclinical trials - those done on animals or with cells in a lab. None of the results referenced human trials or outcomes.
The authors wrote that it is not so much fraud but a high pressure field that leads to so many non-reproducible research results being published.
The highly competitive cancer research environment encourages researchers to "publish or die" and the environment leads researchers to leave out data in studies that run counter to their conclusions, massage results, and interpret findings based more on intuition than actual science, according to the authors. The two also suggest that science journals are more willing to publish studies that will generate interest rather than those that advance science.
The authors also recommended some fixes: Researchers must do a better job of reporting negative findings; research facilities should reduce their pressure on scientists to publish or perish; and science journals should revise their policies for accepting papers for publication.
Source: Drug development: Raise standards for preclinical cancer research, Nature 483, 531-533 (29 March 2012) doi:10.1038/483531a