'Next Generation' sequencing, or NGS, was the big theme of the ESHG meeting this year.
We're doing some of that in my lab, though so far in a small way. We chose a machine that most closely duplicated the sorts of thing we did already, so we could make the transition as smoothly as possible. What's working pretty well now is rapid analysis of two genes. Next we'll be moving to a panel of about 35 genes, 25 covering all the different cancer risks we deal with routinely, plus a group of research genes we think will fall into the moderate risk category, or maybe the high risk but rare category.
This is more or less what many of our colleagues are doing, and I was hoping to see results and discussions of the difficulties of this strategy at the ESHG. There was some, but only from the french groups that I communicate with regularly anyway. Nothing new.
The NGS presentations most all took a different tack: that of looking at the entire genome, or entire protein-coding sequence (the « exome »).
It's interesting to do that when the list of genes on your panel changes regularly, or when you're really not sure what to put on it in the first place.
But at least with my panel, I have a reasonable idea of what to expect from each of the genes on it. I've got something to say to the patient about what I find there.
With the whole exome, you're going to find stuff all over the place, and then you've got to decide somehow what's relevant.
I saw a lot of talks on a lot of different disorders. They'd start with 40,000 variants, then filter those this way and that, and filter them again, and come up with the golden nugget that explains their case. Well, at least there's a decent hypothesis about why this variant explains that case, and not any of the other variants they tossed out.
I didn't see any talks where they filtered and filtered and came up with nothing. Guess those studies don't get presented.
One of the major issues with these exome studies is what to do the the off-target findings. We used to say 'incidental', but that word seems inappropriate when you've deliberately looked for stuff. Even my gene panels can come up with off-target findings – if I have a breast cancer case, and find a mutation in the polyposis gene, I wasn't looking at polyposis, it just happens to be on the panel. But my patient wasn't prepared to hear about colon cancer risk; that wasn't the point of the test: what does the clinician say about it? That case is fairly clear, at least in my lab. We give any results that are pertinent for cancer predisposition, and we do now broaden the pre-test consultation to discuss the possible results more generally. There might also be results that indicate one is a carrier of a gene for a serious rare disease, and if it happens that your partner also carries that gene, you could have some very sick kids. It gets complicated.
But it's an even bigger problem when you've gone looking for an explanation for, say, a developmental disorder, and you learn something about cardiac risk or diabetes. On top of the patient's result, most of these studies looking to diagnose childhood syndromes also test both parents (which really helps weed out the irrelevant variants), and now you'll find things in the parents that the kid didn't happen to inherit, but you know they're there and future kids or the parents themselves are at risk.
To face this, people are developing different levels of informed consent. There's already a general consensus not to give results that have no clear consequences. That would be a long list and wouldn't mean much at all. Really pertinent results can get lost in all the noise. Then the patient can consent to be told all results, or just results with significant health risk that you can do something about, or very narrowly just results directly related to the pathology for which he was referred. In Europe there's careful preservation of the right not to know, while in the US, the tendency is toward the obligation to know.
I could ramble on, but the coffee is ready.