• Report

Technology Assessment on Genetic Testing or Molecular Pathology Testing of Cancers with Unknown Primary Site to Determine Origin [Internet]

Citation

Meleth, S., Whitehead, N., Swinson Evans, T., & Lux, L. (2013). Technology Assessment on Genetic Testing or Molecular Pathology Testing of Cancers with Unknown Primary Site to Determine Origin [Internet]. (AHRQ Technology Assessment Report; No. Project ID: CANU5011). Rockville, MD: Agency of Healthcare Research and Quality (AHRQ).

Abstract

Results: We reviewed cytogenetic analysis and three genomic TOO tests (CancerTypeID, miRview, and PathworkDx) for analytical validity, clinical validity, and clinical utility. The published evidence in each of these areas is variable. Some data on analytic performance were available for all of the genomic TOO tests, but the evidence was sufficient to confirm validity only for the PathworkDX test. We could not compare analytic validity across tests because different data were reported for each test. We found sufficient evidence to assess the validity of the statistical algorithms for CancerTypeID and miRview. We were unable to assess the validity of the statistical algorithm for the PathworkDx TOO. Each test has three or more publications that report on the accuracy of the tests in identifying the TOO of known tumor sites. The accuracy rates across all of the studies for each of the three tests are fairly consistent. The meta-analytic summary of accuracy for the three tests with 95% CI is as follows: CancerTypeID—85 percent (83% to 86%); miRview mets—85 percent (83% to 87%), and PathworkDx—88 percent (86% to 89%). The accuracy of the tests in CUP cases is not easily determined, because actual TOO is not identified in most cases. The evidence that the TOO tests contributed to the diagnosis of CUP was moderate. Low evidence supported the clincal usefullness of the TOO tests in making diagnosis and treatment decisions. Low evidence also supported the length of survival amnog CUP patients who received the test. The evidence was insufficient to answer other key questions on the effect of the tests on treatment or outcomes. Conclusions: The clinical accuracy of all the three tests is similar, ranging from 85 percent to 88 percent. The evidence that the tests contribute to identifying a TOO is moderate. We do not have sufficient evidence to assess the effect of the tests on treatment decision and outcomes.