Discussion

Searching the Gray Literature (for this horizon scan): LexisNexis, CHI vs. Others

For our project, the meaning of gray literature evolved into referring to "any literature which is not peer reviewed." However, given the broad definitions for key terms such as "gray literature," "genetic test," "biomarker," and "in development," one of the early challenges of this project was to identify gray literature which was appropriate and applicable to our particular task order. To this end, we found little precedent or guidance for systematic searching of the gray literature for the purpose of a horizon scan in genetic testing for cancer.

In the NLM's Etext on Health Technology Assessment Information resources, finding the gray literature is demonstrated through the HTA example, which starts with the HTA Database.⁵ We began our search with the NLM report and continued to add gray literature resources to this list as our project evolved. We carefully documented our methods in searching the gray literature for this project because we believe that our report may serve as a novel yet useful example of how to perform a gray literature search for a horizon scan project. After rigorous exploration of numerous databases and resources, we were able to identify a handful of gray literature sources that could be loosely categorized into three different categories of utility for our project: high utility, low to moderate utility, and not applicable.

For the purposes of our horizon scan, we found that the LexisNexis and CHI databases to have the greatest utility for finding cancer genetic tests of interest to our project. One final caveat, our particular method for searching the gray literature is not meant to be definitive or apply universally to all projects. Instead, our report is meant to provide one example of how the gray literature can be searched for the purposes of a horizon scan on diagnostic genetic tests in clinical development for cancer.

Genetic tests for cancer

Part I: Current genetic tests

In both parts of this project, we started our search with the same approach of reviewing the scientific and gray literature while also attending scientific conferences and interviewing expert opinions. In order to accomplish the first goal of identifying genetic tests currently available for cancer care, we found the commercial literature and Web sites for the largest diagnostic test companies to be the most useful for addressing the first part of this project. We found 62 genetic tests for 9 different cancers. Given the broad definition for genetic test that we used, it should not be a surprise that we found such a wide range of genetic tests available, from basic protein biomarkers like AFP, to the new multi-gene assay, Oncotype Dx, for predicting disease recurrence in breast cancer. The compilation of one-page summaries provides more detailed information on each genetic test currently available in the database. In addition, each summary contains a brief synopsis of different search strategies and the amount of literature that may be indicative of the amount of time and effort that would be needed to focus on potential projects such as a systematic review of the clinical validity and utility studies for an individual genetic test.

Part II: Genetic tests in development

One of the challenges that we encountered during this second part of the project was trying to determine what genetic test was appropriate for being "in clinical development" for cancer. Earlier, we discussed the five different phases of development that a genetic test or biomarker must achieve in order to eventually gain clinical acceptance (Figure 1). The evaluation and testing of a diagnostic test can be a long and time consuming process; therefore, a horizon scan for genetic tests in development may only be interested in tests that are emerging from later phases of clinical development (phases 2–5) in order to identify tests with the greater likelihood of having more immediate clinical impact.

The next challenge we faced was to then find the most efficient resources and databases that would identify genetic tests in development relevant to our project. As we gained experience with the scientific literature, gray literature, and professional meetings and interviews, we discovered that MEDLINE searches were useful for finding information of pre-clinical exploratory biomarker and genetic tests. However, we also found that MEDLINE and other databases that search the scientific literature were not as efficient in identifying genetic tests in the later phases of development. Instead, the gray literature, and in particular LexisNexis and CHI, were the most useful in identifying genetic tests in development and with more immediate commercial potential. The results for Part II of this project can be found in Database III, which is a compilation of genetic tests identified through various gray literature resources and expert interviews.

Finally, we found 104 genetic tests in development through our systematic search of the gray literature and other sources. We discovered that the LexisNexis and CHI resources had the highest utility for identifying genetic tests in development that may have more immediate commercial impact. However, we are aware that despite all our efforts, these lists are not necessarily comprehensive and that it is possible that a handful of promising genetic tests for cancer exist but do not appear in our database. Perhaps, this is to be expected since we are entering a time of unprecedented growth for the medical diagnostic industry as physicians, patients, and society are just beginning to benefit from the fruits of labor stemming from the completion of the Human Genome Project in 2003.

Return to Contents
Proceed to Next Section