because it is still not known whether there is clinical validity. The discussion should be focused on how to push a technique through to utility once it has been shown to have clinical validity.

Marc Williams, who presented the Lynch syndrome case scenario (Chapter 2), agreed with the need to build an infrastructure so that some of the questions can be more efficiently answered. He noted that the infrastructure of the Intermountain system is useful in this regard. The company’s approach thus far has been very pragmatic, culling through the list of things that are potentially available and investing time and effort on those things that, from the company’s perspective, would return value for Intermountain patients.

Bruce Blumberg, who presented the genomic profiling scenario (Chapter 4), agreed that there is a problem with the current paradigm for evidence generation. Technology is advancing more rapidly than our evidentiary approach is able to keep up with. The one-disease-at-a-time, one-test-at-a-time, one-SNP-at-a-time approach to research is no longer viable.

Wicklund noted that funding for research is another important issue. Blumberg said that people tend to be more engaged in something when they have an investment in it. He recommended determining who will benefit from the evidence generation and then asking them to underwrite at least part of the cost of the research. Because they are invested in the process, they will be more likely to follow through in adopting and implementing any recommendations that come out of the studies.

HOW MUCH DATA ARE ENOUGH?

Blumberg returned to the question of “When is enough, enough?” that was raised relative to Lynch syndrome testing, and he asked it of pharmacogenomic testing for warfarin dosing. There is already enough evidence, he said, to conclude that there is no major benefit from genomic testing versus current modes of coagulation management in well-managed clinics. When the effect that is being studied is small, more evidence is needed, and larger and longer studies must be done. When will we be convinced that there is—or is not—some small benefit of genomic testing? If there is such a benefit, decisions should be made based on the needs in each individual clinical setting. For example, genomic testing for warfarin sensitivity may not add value in a well-managed coagulation clinic, but it may be helpful in other settings.

Marc Williams added that too often, not just in genetics but in medicine in general, the approach to problem solving is to assemble panels of experts and “think problems to death.” One can always construct worst-case scenarios, he said, but when ideas are implemented, it is rare for those worst-case scenarios to occur. The best-case scenario may not occur either,



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement