The Challenges With Ensuring the Validity and Utility of Diagnostic Tests

Evidence-Based OncologyApril 2015
Volume 21
Issue SP6

Precision medicine has touched every aspect of healthcare today, and—as is evident from President Obama’s State of the Union speech for 2015—is front of mind with the federal government, which has plans to increase basic research funding for the National Institutes of Health and boost the regulatory role of the FDA. However, while diagnostic tests play a critical role in physician decision making— particularly in oncology—by generating data to improve treatment outcomes, questions remain regarding the clinical utility of some of these tests, as well as the validity of laboratory-developed tests (LDTs). While next-generation sequencing is on the horizon for the FDA’s regulatory purview, assessing the validity and utility of existing FDA-approved tests and LDTs, and deciding who pays for the tests, are important issues that remain unaddressed.

The American Journal of Managed Care convened an expert panel of 2 payer representatives and 2 practicing oncologists that I moderated, to discuss these and other issues with diagnostic testing that are plaguing the healthcare world. Unique perspectives on the topic were contributed by Francisco J. Esteva, MD, PhD, professor of medicine, director of breast medical oncology, and associate director of clinical investigation, Laura and Isaac Perlmutter Cancer Center, NYU Langone Medical Center; Daniel F. Hayes, MD, clinical director, Breast Oncology Program, Stuart B. Padnos professor in breast cancer research, University of Michigan Comprehensive Cancer Center; John L. Fox, MD, MHA, associate vice president, medical affairs, Priority Health; and Bryan Loy, MD, MBA, vice president oncology, laboratory and personalized medicine, Humana.

The discussion started with the panelists being asked to scrutinize the “value” of a diagnostic test beyond its analytical performance: “Between outcomes for a particular patient population, improved healthcare quality, the cost of a test, and how complex the procedure is for conducting a particular test, what are the different parameters that clinicians and payers consider?”

For a long time, biomarkers included in these diagnostic tests had not been validated, answered Esteva, resulting in smaller retrospective—not prospective—trials being conducted to evaluate the tests. Referring to biomarkers identified in breast cancer, he added that the estrogen receptor (ER), the progesterone receptor, and HER2—the standard of care—had not been validated via a prospective trial. While he believes that despite the lack of prospective data, there is value of some of the new diagnostic tests used in breast cancer, such as OncotypeDx, Mammaprint, and Prosigna, he sees a problem with the lack of clinical or outcomes data of some of the genomic/next-generation sequencing tests now in use.

According to Hayes, we critically need to take 3 actions: modify the regulatory environment, discuss the analytical validity of a diagnostic test, and identify who really decides the utility of LDTs. Unlike drug manufacturers, whose products face rigorous evaluation through clinical trials, manufacturers of diagnostic tests have the option of gaining FDA approval through a Premarket Approval process, or 510(k) clearance, or an FDA-independent process that requires approval through the CMS-initiated Clinical and Laboratory Improvement Amendments (CLIAs). LDTs, developed by individual laboratories, can escape FDA evaluation as long as they have CLIA approval, which “really means that the laboratory adheres to good laboratory practices, but has nothing to do with the value of that test to take care of an individual patient,” he emphasized.

The “clinical utility” of a diagnostic test needs attention, said Hayes, because it indicates whether analyzing for the particular biomarker on the diagnostic test improves patient outcomes. However, he pointed out that the FDA approval process does not require the manufacturer to establish clinical utility— analytical and clinical validity suffice. So the FDA-approved tests, in his opinion, may not be valuable in terms of patient care, and even if they are, there are no studies to prove so. While acknowledging that some of the LDTs in use are quite good, he pointed to the fact that this has led to a “buyer beware” market. However, he added, there is a need to clarify who should be the most vigilant—patient, doctor, third-party payers, or practice guidelines committees like the American Society of Clinical Oncology (ASCO) or the National Comprehensive Cancer Network (NCCN).

Fox said that he distinguishes predictive tests from prognostic tests. In his opinion, while the clinical utility of predictive tests is on the right track, he identifies problems with the clinical utility of prognostic tests. Even though a good association between the prognostic test and an outcome may be established, he said, informed decisions are lacking.

Loy agreed with Fox that a perfectly good test yields no value for payers if used inappropriately. Pointing to deficiencies in the process—between the test being ordered by an oncologist and the results being produced—he said that as a payer, “The whole system of care is very important to us.” While reiterating that CLIA approval is contingent only on good laboratory practices, Loy added that in addition to the test performance, the results and how they are presented to the clinicians for interpretation are drawing payers’ attention, along with the incremental utility of a test to ensure coverage.

Fox alluded to the FDA announcement that LDTs, especially those used for high-risk prediction and to guide treatment, will need pre-market review by the FDA.1 He worries that this will restrict the development of the tests and may not necessarily solve the problem. Hayes argued that several laboratories are developing these indigenous diagnostic tests without demonstrating any analytical validity or clinical validity, which would never be allowed if they were developing a therapeutic regimen. Applauding the FDA decision, Hayes said it will greatly improve the yield. He believes that rather than suppress innovation, a well-developed test that bears the promise of getting the right drug to the right patient may in fact garner the support of third-party payers.

Citing a workshop on the topic that he helped develop, Hayes pointed to 2 possible ways of generating valuable evidence with these tests: 1) conduct a prospective clinical trial; or 2) conduct a retrospective trial with archived tumor specimens from previous clinical trials, but only after drafting a complete protocol that would include the laboratory, statistical, and analytical plan—a “prospective retrospective study.”

While agreeing with Hayes, Loy said that such efforts are already being considered and that increased governance by the FDA will help payers improve their understanding of the clinical utility of a test by ensuring accountability. Citing the examples of HER2 and ER that Esteva referred to, he added that we are still learning about these tests because much needed information was not available earlier. Hayes added that the CancerLinQ program being developed by ASCO, which is expected to be launched later this year in partnership with the software company SAP,2,3 would be a big help. “I think one of the hopes of ASCO is that CancerLinQ will at least provide relatively consistent guidelines across the country,” he said, adding that while it will not replace clinical judgment, it will provide data on how tests and drugs work in the real world.


Acknowledging that the NCCN and ASCO greatly influence the use of pharmaceutical drugs and biologicals, Esteva pointed out that the lack of evidence with diagnostic tests might result in initial guidelines being developed by the organizations based on expert opinions alone. However, he thinks the NCCN, ASCO, and the European Society for Medical Oncology will lead the guideline-development front for test developers. While NCCN guidelines direct a clinical pathway approach, ASCO, Hayes said, looks at evidence-based personalization of the treatment; the 2 together work well toward standardizing clinical practice, he added.

Fox said that from a payer’s viewpoint, there is value in paying for genetic testing that can help personalize treatment and direct appropriate use of drugs like Tarceva and Herceptin, avoiding false-positive treatment. As a payer, he thinks that these tests are underutilized, as patients who should be tested are not and may not receive the most appropriate therapy. Hayes responded that he was happy to hear of the change in approach by the payer community, citing an experience he had a few years earlier when presenting a case for the OncotypeDx test that was still under development. Hayes related how when he presented data on the value of the 21-gene panel test to Blue Cross and Blue Shield—it could help identify ER-positive, node- negative patients who would not bene t from chemotherapy—the company representative said that while such an expensive test might be ordered by oncologists, they would continue chemotherapy in spite of its results. “If we put our money where our mouth is, if we’re going to use these tests, if we’re going to order these tests, we need to use the tests to drive care, but that means we’ve got to be sure that the tests are accurate and reliable.”

Esteva thinks that while the NCCN and ASCO provide valuable guidelines, the organizations need to keep up their pace in including tests that have been FDA approved or have updated validity or utility data, to avoid barriers to the use of the approved tests. Fox agreed, adding that Priority Health is ready to pay for drugs and tests included in the NCCN guidelines, but if a valuable test is not included in the guideline, it poses a challenge.


Loy stressed the importance of the role of genetic counselors in the process, indicating that Humana has them on board. He said that sometimes those who order the tests may not be sufficiently informed and may order the wrong test or misinterpret the results, which can affect treatment decisions. He also highlighted the importance of involving patients in the conversation and keeping them informed.

In Esteva’s opinion, while genetic counselors can prove valuable in discussions on determining a patient’s risk for developing cancer, he thinks they may have limited input with treatment decisions. Hayes agreed, and added that getting the different providers taking care of the patient to work as a team can eliminate discrepancies in who orders the tests and what treatment ensues.

In Loy’s opinion, too, the end user— the medical oncologist who drives treatment— should be the decision maker regarding which diagnostic test should be ordered, but added that he’s concerned with the lack of consideration for medical history prior to ordering tests for [breast cancer markers] BRCA1 and BRCA2. “It feels like we still have a long way to go in terms of creating some accountability and a measurement and education system to be able to get folks ordering (tests) thoughtfully and getting the results—getting the most value out of the dollars that will be spent on the tests, both in screening but to a lesser extent in the predictive and the prognostic arena.” Reemphasizing the need for educated decisions in diagnostic testing, Fox added that Priority requires genetic counselors be involved before patients are tested and also to discuss results after.

According to Hayes, diagnostic tests have immense value in deciding who not to treat. It is known, he said, that ER-negative patients do not benefit from anti-estrogen therapy. “We’re willing to withhold a pretty tolerable drug that, if it works, has enormous benet, but we’re still willing to withhold it from that group of patients.” On the flip side, he said that ordering panels of tests to screen for the risk of susceptibility to various cancers is uncalled for. Fox and Hayes then discussed how this process could be streamlined so patients are appropriately tested at the point of care and whether ASCO could be involved in the process, which brought the discussion back to CancerLinQ and how it could strengthen clinical decision support.


The discussion ended with the participants reiterating the importance of multi-stakeholder involvement in the process, additional regulation and better guidelines, accountability, and systems that could integrate clinical information with molecular data—all of which would contribute to achieving improved clinical decisions at lower costs.

To hear the complete discussion, visit


1. FDA takes steps to help ensure the reliability of certain diagnostic tests [press release]. Silver Spring, MD: FDA Newsroom; July 31, 2014.

2. Caffrey M. ASCO announces partner for groundbreaking CancerLinQ platform. Am J Manag Care. 2015;21(SP3):SP74.

3. Cavalio J. ASCO announces progress in the development of CancerLinq, with the first version due in late 2015. The ASCO Post website. Published January 21, 2015. Accessed March 3, 2015.

Related Videos
Screenshot of Sancy Leachman, MD, PhD, smiling
Screenshot of Sancy Leachman, MD, PhD
Sashi Naidu, MD, Carolina Blood and Cancer Care Associates
Dr. Lalan Wilfong
Dr. Ryan Huey
David A. Eagle, MD, New York Cancer & Blood Specialists
Screenshot of Sancy Leachman, MD, PhD, smiling
Manmeet Ahluwalia, MD, MBA.
Nicolas Ferreyros, BA, managing director, The Community Oncology Alliance
Susan Sabo-Wagner, MSN, RN, OCN, Oncology Consultants
Related Content
© 2023 MJH Life Sciences
All rights reserved.