THE POTENTIAL AND THE PROBLEMS

According to Sean Tunis, director of the Center for Medical Technology Policy, which was a cohost of the workshop, molecular diagnostics could have a transformational impact on medicine. Though molecular diagnostics currently apply to only about 2 percent of the population, that number could eventually rise to 60 percent (Ferreira-Gonzalez et al., 2008). PricewaterhouseCoopers (2009) has estimated that the diagnostic and therapeutic segment of the personalized medicine market will be $42 billion by 2015, with a 10 percent annual growth rate.

The current reality belies this vast potential, however. As Teutsch et al. (2009) have written, “Of most concern, the number and quality of studies [of genetic tests] are limited. Test applications are being proposed and marketed based on descriptive evidence and pathophysiologic reasoning, often lacking well-designed clinical trials or observational studies to establish validity and utility but advocated by industry and patient interest groups” (p. 3). This is a “serious indictment,” said Tunis, and contrasts strongly with the expectations for the field.

The Centers for Medicare & Medicaid Services' (CMS's) guidelines for the evaluation of diagnostics tests center on two questions. First, is the evidence adequate to determine whether the test provides accurate diagnostic information? Second, if the test changes accuracy, is the evidence adequate to determine how the changed accuracy affects health outcomes? The clear message is that diagnostic accuracy by itself is not enough, said Tunis. The important factor is whether test results lead to changes in practice that can be linked to improved health outcomes.

Establishing this linkage to improved patient outcomes requires clear, predictable, and consistent standards of evidence by which diagnostic technologies will be judged, stated Tunis. These standards, in turn, will dictate the infrastructure and partnerships that are needed, and they will be essential for investors and entrepreneurs to judge accurately the risk and potential returns on investment.

In the past, coverage decisions have not necessarily been based on clear evidentiary standards, Tunis said. In some cases, coverage decisions have been dictated by legal challenges. Private payers tend to follow the lead of CMS in making their coverage decisions. For the field to move forward, Tunis said, a decision-making process among the payers needs to be more clearly tied to a clear and shared understanding of clinical utility.

FROM TUMOR MARKERS TO BIOMARKERS

Cancer diagnostics have progressed from an era of tumor markers to biomarkers, said McCormack. Tumor markers were helpful in making decisions, but they were also problematic. Many were based on serum tumor markers that were validated using sample banks that were not well pedigreed or well stored and were drawn from readily available patients. The evidence generated for such markers “met the standard of the day,” said McCormack, “but left a lot to be desired.”

Today, far more biomarkers are available. From 1960 to 1989, fewer than 50,000 publications in the Library of Medicine mentioned biomarkers. In just the first decade of the 21st century, more than 250,000 did. During that period, medicine has moved from classifying cancers based on organ or tissue to classifying cancers based on pathway. Laboratory medicine also has evolved. An explosion of technology, especially at the genomic level, has resulted in procedures and results that require skilled specialists to acquire and analyze data. Yet the answers provided by those data “are not overly obvious,” said McCormack, and the data are being interpreted by a generation of practitioners and researchers who were not necessarily schooled to fully understand that information. “It is apparent to everyone that our ability to discover potentially beneficial markers far exceeds our ability to translate them for patient use,” he added.

A CHANGING LANDSCAPE

Highlighting how the need for clinical utility information has recently altered the development of molecular diagnostics, McCormack drew on his experience in having been involved in several major diagnostic development projects. Of four tests—the development of the prostate specific antigen (PSA) test in 1993, high-throughput hepatitis testing in 1997, testing for tumor cells in the blood in 2004, and a two-gene pathology lab test to detect cancer cells in women undergoing resection for primary breast cancer in 2007—only in the last case did regulatory approval require a demonstration of clinical utility, which was done in a postmarket context. In launching these tests, McCormack was repeatedly asked three questions by providers. First, has the test been validated? Second, is the test covered? And, third, how will this test change the way I practice medicine? “It was this last question that, of course, stumped me the most, because I did not have the clinical utility … they were looking for.”

It was not until this most recent test that the clinical utility question drove a change in the development process, said McCormack. Veridex partnered with the Southwest Oncology Group to study the clinical utility of basing therapy on counts of circulating tumor cells 3 weeks after the start of therapy. The study opened in the fourth quarter of 2006, closed to enrollment in March 2012, and still has 2 years before outcomes can be determined. Nonetheless, noted McCormack, “it shouldn't take 6 years to [demonstrate the utility of a test]. We need to be creative to deliver the information that people want.”

INHERENT TENSIONS

Tunis emphasized the inherent tension between the level of certainty about risks and benefits and innovation and early access to new technologies. The optimal balance to maximize long-term public health is not easy to determine and varies by stakeholder interest and perspective. Thus, said Tunis, “there is a critical need for stakeholders to come together and develop some common, shared understanding of what constitutes adequate evidence of clinical utility and a process for doing that.”

Clear, consistent, and predictable evidentiary expectations are essential, said Tunis. “Until we translate that dialogue into specific methodological recommendations for designing studies of clinical utility, we can't really move the field forward in a meaningful and predictable way.”

These methodological recommendations are as much a product of a social consensus as a scientific consensus, Tunis remarked. They reflect a collective social judgment about what is optimal, and they cannot be determined by specialists in a single field working in isolation. In balancing public policy objectives that compete with each other, the stakeholders need to arrive at a consensus that can be translated into scientific and technical terms.

Footnotes

1

Andrew Freedman, chief of the Epidemiology and Genomics Research Program's Clinical and Translational Epidemiology Branch at the National Cancer Institute, cited these four questions during his workshop presentation, which is summarized in Chapter 4.