Douglas Moeller, MD
Nobody disputes the need for careful, judicious diagnostic testing. But as the number of individual gene tests continues to grow, clinicians are struggling to keep up. On the one hand, using diagnostic markers like KRAS in colon cancer or hormone receptor status in breast cancer to select the most appropriate chemotherapy regimen is essential. It’s the standard of care. On the other hand, the role of many other gene markers is still not fully defined.
One way to keep up with the “what’s what” in molecular diagnostics is to maintain a framework for classifying the tests that a practice uses regularly. Here’s an example of such a framework.
Screening: An ideal screening test is simple to do, sensitive (low false negative rate), specific (low false positive rate), and inexpensive. The goal is to test 100 percent of a targeted population for affected members. The principles of screening are straightforward; putting it into practice, less so. For example, the debates for or against PSA testing for prostate cancer and mammography modalities for breast cancer have not convincingly been resolved.
Diagnostic: An analyte that clearly distinguishes a favorable characteristic for treatment group #1 with a not-so-favorable response that moves a patient into treatment group #2 becomes important when the cost and/or toxicity of treatment group #1 is significant. The emergence of KRAS for colon cancer, and the continuing debate surrounding it, highlight this category.
Predictive Risk: BRCA testing to confirm high-risk status for women with strong family history of breast cancer highlights the challenge of a focused screening test that is moderately expensive, and challenging to apply to the entire population. By narrowing the tested cohort to high-risk individuals without sufficient family history may be missed (orphans, only children in small families).
Prognostic Risk: Similar to predictive risk in a screening population, this category is distinguished by starting with a known problem and attempting to further characterize markers for response. OncotypeDX for breast cancer is probably the best example of a test that uses multiple markers to assess the response to treatment. Twenty-one genes are assayed and ranked via statistical algorithms to create a recurrence score between zero and 100. Higher numbers suggest the need for more aggressive treatment.
Companion Diagnostics: Metabolic markers have emerged in recent years as critical variables to assess in advance of the actual treatment process. Genetic variation in the cytochrome p450 pathways can determine faster or slower drug clearance rates, leading to dosage or treatment interval adjustments in individuals. Although this is the heart of personalized medicine, the term precision medicine is frequently being used in this setting.
Panel Testing: There are several types of testing that assess multiple markers in the same test. Two major subcategories must be distinguished. The first is Multi-Analyte Algorithm Assays (MAAA, CPT Manual, Appendix O), a term chosen by the AMA/CPT Editorial Panel to replace In Vitro Diagnostic Multivariate Index Assay (IVDMIA).
Oncotype DX testing, mentioned above, can also be classified in this category. Oncotype DX Breast, for example, assays 21 specific genes and, via statistical profiling, assigns a composite risk score based on phenotype outcomes in a reference population. Steadily increasing clinical familiarity with this approach is affirming the usefulness of this prognostic approach, but individual outcomes will still vary from predicted outcomes.
The second category, multiple analyte testing (against dozens, hundreds, and even thousands of genes) is now available as Next Generation Sequencing (NGS) platforms automate the laboratory component that examines exomes (eg, 21,000 human protein genes) and full genomes. Interpreting the results of new mutations and unexpected, untargeted variations in adjacent genes remains a significant challenge for clinicians moving forward in this field.
The pace of change and innovation in molecular diagnostics testing won’t be slowing anytime soon. The pressure is on providers to make the best clinical decisions in a rapidly changing environment where few tests have sufficient clinical evidence to determine their clinical utility, and where many are lab-developed tests (LDTs) here, protocols and processes can differ significantly. Job one: Classify the tests your practice uses regularly, and keep that framework well maintained.