REGULATORY ISSUES

Greely described three major regulatory issues that will be important as the technology progresses through these three developmental stages. First, he said, sequencing machines will need to become Class III medical devices. They will need to be regulated by the FDA and will require proof of safety and efficacy. Before the instruments can be used for clinical purposes, their developers will need to demonstrate their sensitivity and their specificity, and the machines will need to be assessed for their ability to phase chromosomes and sequence areas of high GC content.

Second, the laboratories that do the sequencing will need to meet stricter proficiency standards. These facilities will need to be certified under CLIA and should additionally be certified by CAP. The CLIA certification process will need to establish clear proficiency standards before the sequencing of patient genomes is allowed for clinical purposes. Laboratories will need to demonstrate the kinds of quality control systems they have in place, the kinds of procedures they have for repeating unclear or unexpected results, and how they will make sure that important positive results are not false-positives. Meeting these standards will require “effective regulatory attention to clinical laboratories,” Greely said.

The most difficult regulatory issues will involve the interpretation of genomic data. When whole-genome analysis is used in the clinic, interpretation will need to be largely automated because, Greely said, “with 6.8 billion base pairs in a diploid genome, you are not going to get a human brain looking at all of those.” Software packages will sort findings into different categories for action, and where information belongs will depend partly on the medical needs and characteristics of the patient. These software packages will presumably allow for sequence validation and updates, Greely said, and as such will be considered medical devices that should also be regulated by the FDA for safety and efficacy.

ETHICAL ISSUES: THE DOCTOR–PATIENT RELATIONSHIP

Greely discussed four ethical issues, all of which revolve around the doctor–patient relationship. First, genomic information should be disseminated to a patient within the confines of medical care. Already many companies offer results directly to consumers, and many consumers argue for their right to access such information outside of a medical relationship. Greely argued, however, that as is the case with a patient obtaining a tissue biopsy or imaging scan, sequencing data should not be acquirable without the assistance of a physician. Many health care providers may know relatively little about genetics, Greely said, but “they know something about tests and they know something about patients.” They can listen to patients, tell them when they do not need to take a result too seriously, and determine when a patient is confused or worried about something. They have skills and information that an e-mail from a direct-to-consumer company will not provide. Health information is “powerful and important and likely to be misinterpreted by people to their own harm,” Greely said.

The second aspect of the doctor–patient relationship involves which information a doctor needs to provide to a patient. The answer is fairly straightforward, Greely said. Under normal circumstances, the health professional should provide the patient with any information that is clinically important but not necessarily with information that is not clinically significant. For example, a patient might have a whole-genome sequence to test for BRCA1 or BRCA2 mutations with negative results. However, if the test revealed mutations related to Lynch syndrome, which confers a very high probability of getting colorectal cancer during a lifetime, a physician should have to reveal this information to a patient. “You are in a clinical relationship with the physician,” Greely said. “If the physician orders an X-ray on you because he is worried about your liver and sees a big growth on your kidney, he is not going to say your liver is fine and not mention anything about the kidney.” Genomic information can be just as clinically significant as information from a medical imaging test, and physicians are responsible for the information that genetic testing reveals. On the other hand, physicians do not need to tell patients about things that are not clinically significant, Greely said, although he admitted that this is a difficult phrase to define clearly. Still, health care providers do not need to convey information about variants of unknown significance that might cause undue anxiety.

The third issue—consent—also involves the information that doctors need to convey to their patients. Physicians need to let patients know that they will tell them about clinically significant results from genomic testing, and they need to explain that they will not tell patients about results that are not clinically significant unless specifically requested. Physicians should ask their patients if they want to know more, in which case access to additional information can be provided. “Having them consent to the fact that you are not going to tell them about absolutely everything that is in those 6.8 billion base pairs is, I think, both ethically a good idea and maybe legally a good idea,” Greely said. “The goal of the medical profession shouldn't be to eliminate possibilities that it will get sued but to do a good job in taking care of its patients, which in turn will limit the number of times it loses lawsuits.”

The fourth category is the most challenging, Greely said: How can health care providers tell patients about all of the genetic results that are clinically significant? Based on early experiences with whole-genome analysis, an average patient might have roughly 100 clinically significant genetic variants, with 5 to 10 things that are quite important and 90 to 100 that deserve mention. Even if a genetic counselor spent just 3 minutes talking about each thing, discussing 100 results would require 5 hours. “Who is going to give that counseling? Who is going to listen to that counseling? And most important, who is going to pay for that counseling? We are going to have a real challenge trying to educate patients about what this means.”

Research projects are needed to look at how best to convey information to patients about their genetics, Greely said. The solution will probably involve presentations using various media combined with some face-to-face discussions, especially for the most important results. “You don't want them just reading it on the Web,” he said. And doctors should be careful that patients understand the implications of the information, he said: “‘Yes, your BRCA1 and BRCA2 genes are normal, but no, that doesn't mean you should stop getting mammograms,’ which is the kind of mistake I am afraid patients will likely make that could be fatal.”

PRIVACY ISSUES

Greely also discussed five privacy-related issues. First, where should information be stored? Options mentioned at the workshop include re-sequencing as needed, giving it to the patient, inserting it into the EMR, or putting it in the cloud. “My own sense right now is the right answer is, yes, all of the above,” Greely said. On one end of the spectrum, the information is always available for re-sequencing since it cannot be removed from the nucleus. On the other end, if a patient changes doctors or health plans, as most Americans do, having the information in the cloud will enable the information to be retrieved by a new health care system.

Second, a germline sequence will not change over time, but interpretations will change continually as research progresses and as a patient goes through life. Standards will be needed for how often data need to be re-queried with software interpretation tools and how to pass new information on to doctors, said Greely.

The third category involves children. Today, children are generally not tested for health conditions that are not important during their minority. But if all children are tested neonatally, how will those data be treated? Should the data be stored in a lockbox and given to them when they are 18? Should the data be given to their parents or pediatricians? “I don't know the answer to this one,” Greely said. “It is a tricky one, but it is one where we are going to have to come up with an answer.”

Fourth, the huge amount of genomic information that will soon be stored in EMRs and in the cloud could be enormously valuable for the understanding of disease. How much pressure should society exert on individuals to encourage them to let those data be used for research? Perhaps we should consider that patients could be compensated in some manner when they allow their information to be used for research purposes. This is another issue that will be difficult to resolve, Greely said.

Finally, technological approaches cannot completely resolve confidentiality problems. “The good news of genetic data is it will be protected as much as other health data,” Greely said. “The bad news is it will be protected as much as other health data.” There will always be some security risk involved with storing the data since it, like any other data, can be compromised illegally or through poor oversight by the user, such as losing a password or a laptop. Even if data are anonymized, people can be re-identifed if phenotype and genotype data are combined. “Confidentiality cannot be guaranteed, and we need to be honest with people about that,” Greely said. “We need to tell them there is some privacy risk.”

OTHER ISSUES

Greely concluded by briefly mentioning several other issues that will loom large in an era of genomic medicine. Parents will more easily be able to make decisions about the genetic traits of their children. The information will be available for governments to construct universal forensic databases. People will make invalid and discriminatory connections between genetic data and personal or group attributes.

As the cost of sequencing continues to plummet, genome sequencing will become ubiquitous, Greely said. “We can manage the transition in ways that make it less painful, or we can manage it or fail to manage it in ways that make it more painful,” he said. “I applaud the Institute [of Medicine] for this workshop, because it is really important for us to try to figure out how to use this in ways that maximize its benefits and try to minimize its harms.”