NEW YORK (GenomeWeb) – At a public workshop held by the US Food and Drug Administration this week, medical professionals and patient representatives discussed their perspectives on the return of increasingly complex genetic test results.
The agency had convened the workshop, which took place at its White Oak Campus in Silver Spring, Maryland and was webcast, to gain a better understanding about the specific information patients and providers prefer to receive from genetic tests, how they would like the results to be returned, and what other information they might need to enable them to use the results effectively for making medical decisions. FDA intends to use those insights to formulate new regulation for diagnostic tests that are based on next-generation sequencing and other new technologies.
While genetic tests are not new, their complexity has grown over the past decade with the advent of technologies such as NGS that deliver not only multiple test results from a single sample but also results that are associated with uncertainty and varying levels of underlying clinical evidence.
Elizabeth Mansfield, FDA's deputy director for personalized medicine, explained that the agency does not want to restrict anyone from having their genome sequenced, but wants to keep an eye on the medical use of this information, with the goal of determining what levels of clinical evidence and contextual information patients and providers need to make medical interpretations of the results.
"How you are going to make decisions may depend a lot on how the information is transmitted, it may depend a lot on the warnings or limitations around those results and the interpretation. So we would like to hear what you would like to know," Mansfield said.
She stressed that the FDA is not planning to interfere with the practice of medicine with its proposed regulations, which some workshop participants claimed it would by dictating what information physicians report back to patients. "This is about preference for receiving results, not about the practice of medicine," Mansfield said.
The workshop was held against the backdrop of President Obama's Precision Medicine Initiative, which promises to deliver personalized treatments to patients based on their genetic makeup, health history, and other factors, and in which the FDA is playing a key regulatory role.
For example, the FDA is developing standards for the accuracy of next-generation sequencing tests for cancer and other genetic diseases, which were discussed in the context of cancer at a separate workshop last month. It is also developing criteria for public databases to support the clinical validity of these tests, and is fostering community development of open source tools to evaluate test performance through its PrecisionFDA platform, said Bruce Kuhlik, FDA's senior advisor to the commissioner.
"Through PMI, we want to promote access to innovative technologies that are accurate and provide meaningful information to patients and providers to support healthcare decisions. We want and need your input to make that a reality," Kuhlik said. That input "will help us develop guidance to achieve the bold objectives that the President has set when he launched the Precision Medicine Initiative."
"We are here today to look at some of the big questions around regulation and management of big data and new tests that will be absolutely critical to precision medicine," said Jo Handelsman, associate director for science at the White House Office of Science and Technology Policy. "We know that diagnostics are a key element for precision medicine, and FDA will play a critical role in managing the development of diagnostics, and hopefully this audience will play a role in ideas about regulation."
FDA representatives encouraged stakeholders to submit their comments on the topic to a docket by March 31.
FDA framed the workshop around three different types of genetic tests, each illustrated with a case study and accompanied by a panel discussion: genetic tests in patients who are currently well to predict their risk of future disease; genetic tests for acute disease, including germline tests for inherited conditions and somatic analyses of tumor tissue; and genetic tests for chronic diseases, which mostly covered pharmacogenomic tests to predict the response to drugs for the treatment of such diseases.
Predictive testing
One of the concerns panel participants expressed about predictive genetic tests is that doctors have neither the time nor the experience to delve into the interpretation of genetic test results, in particular if the test includes dozens, or, in the case of whole-exome or whole-genome tests, hundreds of results. In addition, they find it difficult to convey small increases in disease risk that may not be very meaningful to their patients.
Patients, on the other hand, often seem to take DNA tests more seriously than other types of medical tests and to regard the results as absolute, being surprised to find out that genetic tests can have false positive or false negative results. Patients also sometimes say at the outset that they want to receive all their genetic results, panelists said, only to discover later that they have obtained information they did not really want, which speaks to the importance of pretest genetic counseling and making patients aware of the possibility of unintended findings.
Tailoring test reports to the needs of physicians and patients, and maybe standardizing the information they contain or the language they use was another theme expressed by several panel participants. Several advocated that reports should be tiered, starting with the most important results and well-established clinical variants, and adding other results, such as variants of unknown significance, as well as more detailed information, further down. It might also be a good idea to have different versions of a report for patients, providers, and genetic counselors, they said.
Patients also often find it helpful to receive the results both in a narrative and graphic format, and to have disease risk percentages expressed in various ways. Average lifetime risks, panelists said, are generally preferable over relative risks in test reports. Reports should also answer commonly asked questions, for example whether a test is a screening test or a diagnostic test, how predictive the test is, what the results mean for family members, and what the limitations of the test are, and they should state that patients should discuss the results with their healthcare provider before making any major medical decisions. Finally, reports need to take a patient's ethnic background into account because the interpretation of variants may differ between ethnic groups. Ideally, reports should factor in a patient's clinical context, which many testing labs do not have access to today.
Clinicians also called for trusted sources of information on the Internet for patients to better understand their results, in particular since patients frequently consult "Dr. Google" before seeing their physicians about results they received ahead of time, and often confront doctors with inaccurate information. Doctors, on the other hand, would appreciate having a clinical support system in place that allows them to call an expert if information on the test report seems confusing to them, according to one panelist.
One of the challenges of predictive genetic tests is that the interpretation of variants may change over time, necessitating dynamic test reports, which might be difficult to implement in electronic health records, which are designed to be stable and not easily changed, panelists pointed out.
Tumor and other disease tests
A second panel discussion centered on the use of genetic testing in the context of acute diseases, in particular for patients with a suspected hereditary condition and for the analysis of tumors in cancer patients in order to find a suitable targeted therapy.
Like in the previous discussion, several panelists said a tiered approach would be a good way to present test results, starting with variants that have strong evidence or FDA-approved treatments attached to them, continuing with variants with less compelling evidence or with treatments that are in clinical trials, and finishing with variants of unknown significance. Panel participants stressed that this approach would allow doctors and patients to choose the information they want to consider and act upon, for example whether they would want to enroll in a clinical trial based on evidence gained only in cell lines. Ideally, test reports would be customized to patients and doctors' wishes —for example to obtain only actionable results or results with high clinical certainty — but that is mostly not feasible today.
Even non-actionable results would be important to include in the report, panelists argued, because if those results provide a molecular diagnosis for an inherited disorder, that in itself is often important for families. In addition, pathogenic mutations can help with research into genetic diseases and their phenotypes, which could lead to new treatments in the future.
The interpretation of variants of unknown significance will necessarily differ between patients with a strong or weak family history of a disease, such as cancer. VUS should go into the electronic health record, according to one panelist, so patients can be identified and re-contacted if the status of their VUS shifts, which might make them eligible for clinical trials, for example.
When it comes to tumor testing, patients and physicians also need to be made aware of the possibility of finding germline mutations that can have profound effects on family members and may require more in-depth discussions with a genetic counselor, panelists pointed out. However, many tumor panels today are not validated for germline results and may or may not report them. Patients should also be informed ahead of time whether insurance will cover their test, and how the test results might affect their ability to obtain life, disability, or long-term care insurance.
According to one panelist, in the absence of standards today, tumor test reports should also contain information about the analytical validity of the results, given that the performance of technologies such as NGS differs between variant types, such as single nucleotides, copy number variants, and indels. They should also contain information about tumor heterogeneity and the test's detection limit. Based on that information, physicians and patients can then decide how to use the results.
While various professional organizations are working on guidelines and standards for test reports, commercial test providers want to differentiate their own reports from those of others, leading to inherent tension, the panelist pointed out, though there should be a requirement for certain types of information to be included.
PGx for chronic diseases
A third workshop panel focused on genetic tests for chronic diseases, illustrated by a hypothetical case study of a young patient with a history of depression and schizophrenia that is moderately well controlled with drugs and therapy who takes a genetic test to help him find "the perfect drug" for his condition.
Pharmacogenomic results, like the one provided by this test, are often desired by participants in research studies, according to one participant, as they tend to be the most actionable results.
One panelist, herself suffering from chronic diseases, said she would find it critical to have access to genetic information that can help her to choose the best drug or mitigate its side effects, especially for psychiatric disorders, where drugs are often chosen by a process of trial and error that can be lengthy. However, another panelist who is a clinician cautioned that if the genetic evidence is not strong, it would be difficult to decide how to act on the information.
According to another panelist, a study of drugs commonly used in psychiatric disorders that came out last year concluded that with the exception of a few tests for relatively rare gene variants, not many genetic tests exist today that can help doctors make treatment decisions or predict adverse events.
Another panelist said she worries about tests promising to deliver the "perfect treatment" because pharmacogenomic tests, like the one in the case study, provide probabilistic information and no perfect answers. One participant, in fact, mentioned that she received a direct-to-consumer genetic test from 23andMe and uploaded her data to different platforms for analysis, receiving different results that conflicted with her personal and clinical experience. Nevertheless, she said she would not want the FDA to be too prescriptive about what types of information to include in a test.
A pharmacogenomic test like the one in the case study could open a dialog between a patient and doctor to discuss new treatment options, panelists said, even though the test result may not be the only consideration.