Hacker News Reader

Comment by JunkDNA | Hacker News Reader

I'm on one of the NIH grants to look at the impact of clinical genomic sequencing on patients, which includes a bioethics perspective. The hardest thing is definitely that our ability to measure greatly outpaces our ability provide meaningful interpretation. You get into all these discussions about what constitutes a medically significant finding. It's very hard to quantify when you start to grapple with it, and slippery slopes abound. Happy to answer specific questions on this in more detail if anyone wants.

View on HN

A comment I left on a LessWrong post in July might help people understand why it is such a problem:

The genome is the ultimate spaghetti code. I would not be at all surprised if some genes code for the direct opposite characteristics in the presence of other genes. It is going to take more than just running relatively simple correlation studies to untangle the functions of most genes. We are going to have to understand what proteins the genes code for and how the proteins work.

Edited for clarity, eliminated a couple of ambiguous pronouns.

You're getting at the heart of the matter as far as the complexity, but just to put an even finer point on the it:

A single gene sequence can code for many different alternate versions of the same protein. These different variations can have pretty wildly different biochemical properties. Furthermore, the protein products of a given gene varies by environment, ethnicity, organ, tissue, cell type, and stage of life (development, infancy, puberty, etc...).

If that weren't daunting enough, many proteins have a secondary function of regulating their own production and the production of other (somehow related) proteins. This means that there are these highly complex interdependent feedback loops governing the activity and concentration of proteins.

The closest programming analogy might be spaghetti of LISP-like macros that are themselves spaghetti code that writes LISP-like macros.

But all that covers the basic research question of "how is this gene connected to human disease?". Even when you get past that, it's not so easy to know what results to supply back to patients. Do you tell the parents of an 8 year old who has had her genome sequenced that their daughter has a very high predisposition of getting cancer in her 40's? The daughter is too young to consent to hear that info. Now you're forcing the parents to make the choice for her. Not only that, even though the child is the one in for the test, the parents now (possibly without fully thinking it through) are recipients of information about themselves. All that because they wanted a genetic test to see why she's shorter than average.

While it's undoubtedly unpleasant to hear that you're more likely to get cancer than most people, that information could save your life. I'm not sure that I see a moral quandary here.

It's about what people choose to know. There are many people who don't want to know about things they can't always do something about. Many people would rather live a carefree life than know they can get a particular disease. This is a point of view I have trouble wrapping my head around personally, but it can't just be discounted. My example is also a little more clear cut than most. There are many cases where it's not so definitive. Some of the risk calculations result in increases over the baseline for the population of 10%. What does that even mean? It's all pretty poorly understood. Hence there is a real desire to only report back the really strong data. But that unfortunately results in eliminating useful, but highly speculative data.

In the off chance that it doesn't save their life, theres a great concern that it would significantly decrease her quality of life preceding a diagnosis. not just psychologically--theres insurance to worry about here too.

Baseline chance of cancer is what, half? Finding out you're likely to get a specific kind should barely affect quality of life. It'll just tell you where to scan.

What kind of maths/statistics are you using?

You echo the sense that I've gotten after a cursory look into the field.

What do you think are they key barriers limiting clinical demand for genomic data?

Well there's a couple of things at work. For one, the key barrier to "clinical demand" is that clinicians (other than highly specialized ones) are not really prepared for dealing with genetic results. Most primary care doctors haven't thought about genetics since med school. I would also say that many of them question the utility of such data in clinical care. So I would describe the demand side as "tepid" except for the cases of specialities where it makes a big impact.

The above cultural issue is compounded by the fact that genetic testing is the only kind of test that transcends time. Your DNA sequence determined today could (and likely will) mean something different in the future as we learn more about the genome. But medicine historically has been very transactional: your doctor orders a test, the result comes back, you discuss results, make some lifestyle or treatment decisions and move on. With genetic tests, you could be "normal" today and tomorrow a paper lands in the New England Journal of Medicine that demonstrates with high certainty you're going to get early onset dementia or something. Well that's a problem in the current system (at least in the US). Whose job is it to go back and analyze people's genomes and update them with new info like this? Who do you bill for that? How do you notify someone? Some people don't want to know such things, can they opt out?

Targeted genetic testing has been used for years to identify specific diseases. But the tech to do huge swaths of the genome affordably is only a few years old. So even the diagnostic tech is still a bit "beta". That's the final piece limiting adoption. We're all collectively figuring out how to do this and how it fits in with existing regulatory requirements. All that "figuring stuff out" takes time.

Thanks for the response - really interesting stuff. I'd love to take a look anything that you/your collaborators have published in this space if anything's out yet.

23 and Me sells subscriptions to future information about your genes.

Reply on news.ycombinator.com