Evidence for drugs and biomarkers in oncology guidelines (GLs): A survey of the National Comprehensive Cancer Network (NCCN) colon cancer panel.

Abstract

e16578 Background: GLs influence practitioners, payers, and policy about the adoption of new cancer drugs and biomarkers. The NCCN GLs, referenced by Centers for Medicare & Medicaid Services, are the most frequently-updated oncology GLs with methodology based upon evidence review and expert consensus. The sources of evidence and weight of expert opinion in developing these GLs have not been described. Methods: We developed a multiple choice and ranking survey to examine how new drugs and biomarkers reach an NCCN panel for discussion, which sources of evidence are preferred, and the timeliness of GLs for new drugs and biomarkers according to source of evidence. Anonymous survey links were e-mailed to members of the NCCN Colon Cancer Panel with permission from the NCCN Board of Directors and Colon Cancer Panel Chair. Analysis was descriptive. Results: 69% (18 of 26) panelists responded including 56% medical oncologists, 17% surgeons, and 17% radiation oncologists. 44% were members for > 6 yrs, 22% for ≤ 2 yrs. Recent publications or presentations and new FDA approvals were the most common factors leading to discussion of a drug or biomarker at a panel meeting. The preferred supporting evidence for GLs was randomized, controlled clinical trial (RCT) data for 80% of respondents; meta-analyses (MA), non-RCT data, and expert opinion were preferred by 7% each. RCT data, systematic review (SR), and MA were reported by 87%, 67%, and 60% of respondents, respectively, as highly likely to lead to timely inclusion of a new drug or biomarker in GLs. GLs about new drugs were perceived as timely by 80%, compared to 53% for GLs about new biomarkers. Conclusions: RCT data was strongly preferred as the source of evidence for GLs about drugs and biomarkers. In contrast to GLs from other organizations which rely upon SR and MA, this preference for RCT data may be due to the dynamic, real-time development of the NCCN GLs. GLs about new biomarkers were perceived as less timely and more likely to require revision than GLs about new drugs. The findings from this pilot survey may inform future studies to improve the timeliness of GLs for rapidly-evolving technologies such as biomarkers for which RCT data may not be available.