Omics tests need validation

Tools

Biomarkers are being used every day to target development of new drugs and to diagnose and guide treatment of disease, and they have the potential to be ever more powerful as research into genomics, proteomics and all the other omics continues. However, as a new report from the Institute of Medicine recommends, there need to be solid evaluation and validation processes in place before any of these tests make it to the clinic.

These concerns and recommendations arise from recent findings that some personalized medicine tests that have been based on omics research have not proved valid. The issues arose from claims from researchers at Duke University about genomic tests predicting efficacy of cancer chemotherapies in certain patient groups. There were problems with data and methodologies, and because these were not spotted, or not acted upon, papers were published that had to be retracted, patients were enrolled into inappropriate clinical trials, and companies were launched prematurely. These tests have now been acknowledged as invalid.

To avoid this happening again, which could potentially put patients at risk and place doctors in a vulnerable position, the report recommends that investigators should ensure that their data and computational processes go through independent scrutiny, and journal editors should require researchers to disclose their data and codes at the time of a paper's submission. As part of the process, researchers should discuss the tests with the regulatory authorities at the beginning of validation studies and not change the tests during clinical trials without letting the authorities know.

The key recommendation is for integrity and transparency at every stage of the process and reporting and investigating any conflicts and issues, from beginning to end. Even journal editors (and newsletter editors) can play a role here, spotting and reporting errors.

"We hope that this report will help all members of the investigative team understand the entire pathway of translating omics discoveries into clinical tests and recognize and avoid the potential pitfalls at each stage," said committee chair Gilbert Omenn, professor of internal medicine, human genetics, and public health, and director, Center for Computational Medicine and Bioinformatics, University of Michigan, Ann Arbor. "We believe that past problems, such as the Duke case, could have been prevented had a clearly defined process been available and been utilized. Scientific and clinical progress in omics test development will be accelerated if these recommendations are broadly adopted."

Want to read more about personalized medicine? In the ebook Companion diagnostics: The future of medicine, we've looked at the development of targeted drugs and companion diagnostics, including the successful and not-so-successful projects, and what issues (and opportunities) there will be for regulators, biopharma companies and payers moving into the future. -- Suzanne Elvidge (email)