Richard Smith: Feeding back to individuals the findings of research on their genes

Richard SmithAround the world research groups are sequencing the genome of tens of thousands of people, and a crucial question is what to do about feeding back to individuals findings that may be “clinically significant.” My immediate reaction was “Of course you should feed the findings back,” but after chairing a session on this topic at a conference in Oxford I realised that this was a complex question with uncertain answers—as, I must confess, are most questions when you examine them closely.

 The question arises with the UK10K study, which is sequencing the genomes of 10 000 people in the UK, some of them patients but many of them not.
The researchers know that they will find genes linked to disease. Some will be genes already known to be linked to disease, some will be genes that are discovered in patients with known diseases, and some will be new findings. New genes linked to disease will be uncovered not only by the primary researchers sequencing the genes but also by secondary researchers conducting meta-analyses on shared data from the study and other studies.

These findings might be either “pertinent” in that they are linked to the disease being studied in a group of patients or “incidental.” It’s the incidental findings that present the biggest difficulty, said Matt Hurles, who is a researcher but also cochair of the Ethics Advisory Group set up to advise the UK10K study. The problem is further complicated because there will be both false positives and “sample swaps,” where samples from different patients are mixed up, something that is inevitable in such a large study.

Ideally all those in the study consent to having their whole genome sequenced, the deposition of their data in a database, and a system of feeding back findings of individual clinical significance. The UK10K study comprises many other studies, and some people consented some time ago. It may be that the consent was adequate for what is now proposed in the UK10K study, or it may not be. It is for a combination of the principal investigators of the individual studies, the Ethics Advisory Group, and the local research ethics committees to decide whether existing consent is adequate. If not, the people must be “reconsented” (apologies to grammatical purists, but this is how those doing the work speak).

Somebody in the audience pointed out that one off consent is usually inadequate for most things. Informed consent is a regulatory concept with the emphasis on disclosure not understanding. It’s impossible for consent to sequencing of your whole genome to be fully informed because there are many “unknown unknowns.”  Sharon Terry, chief executive officer of the Genetic Alliance, used the example of saying “I do” when getting married—you have no idea of what lies ahead. For that reason informed consent might be better conceived as a continuing conversation rather than a single event.

The Ethics Advisory Group says that principal investigators must have “robust management processes for handling potential clinically significant findings” but doesn’t specify what they are. Again it’s for the principal investigators and local research ethics committees to decide, meaning inevitably that there will be variation. The advisory group does says that there should be “equality of consideration,” a nice concept that could, I’m sure, give rise to months of debate in a court. There are not clear precedents for the feeding back of incidental findings, and the law isn’t clear either.

If findings are going to be fed back, says the Ethics Advisory Group, then the individual must have consented to feedback (and must have had the chance to decline), the findings must be validated, and there must be mechanisms for clinical management.

But what is “clinically significant” and who decides? The advisory group says that a finding is “clinically valid” if “variants of the same type in the same gene have been unambiguously identified as being pathogenic in a peer reviewed journal.” My cynical reaction was that peer review is a pile of horseshit (as I’ve said many times), precious little in journals is “unambiguous,” and genes are not pathogenic or not pathogenic but rather carry a risk of a particular size (often small). The decision should be made by those “whose professional duty it is to make such judgements.”

Other questions are who is going to feedback and how? The onus is on the principal investigator, but many will not be clinicians. If somebody is to be told they have a risk of a rare disease then they would probably want to be informed by somebody with a deep understanding of the genetics and of what might be done clinically—probably two different people and almost certainly not the person’s GP. Nor are people likely to want to be informed by a letter or email.

Again there will be variation in what happens, and, perhaps unsurprisingly, those principal investigators who have decided what they will do will not feedback incidental findings.
But studies show that most people do want to be informed about such risks, although there is a minority who do not. Furthermore, people might want to be informed of some risks and not others. Sharon Terry described a system on the internet, PrivateAccess, that allows individuals to set their own levels in a nuanced way.

The first question to Matt Hurles was “If I’m in your study and want to have the raw data of my genome sequencing can I?” The answer was “No. You can go to a commercial company and get your genome sequenced.” The reason was the project’s research data are “not of a clinical diagnostic standard.” The question and the answer produced a forest of hands, and a Norwegian leapt up and said that under Norwegian (and he thought European law) you would have to give the data. Hurles said that English law was not clear. The Norwegian later suggested that there might be a low bar for giving data, emphasising the potential problems, and a high bar for systematic feedback of clinically significant findings. It’s the interpretation of the meaning of the data rather than the data themselves that worries people.

There were “ordinary people” and “patients” at the meeting, and they are on the Ethics Advisory Group of UK10K, but I was left with the impression that this important and difficult debate was being conducted primarily by “insiders” and that it would be good to broaden it. What I wonder in a horrible moment would the Daily Mail make of it?

Richard Smith was the editor of the BMJ until 2004.