Monday, May 10, 2010

Personal Genomes in Clinical Care. Quake paper Falls Short!

With all due respect to the scientists involved in analyzing Stephen Quake's genome in clinical context.

You did a major league $h!tty job.

No offense.

I can only assume this based on what you reported in the lancet paper.

Start by asking yourself.

"Is Stephen healthier because of what that genome and clinical assessment added to his care?"

I am speaking precisely on this topic at the Consumer Genomics Conference on June 3rd at 830 AM. So I will hold off on all my arguments....But,

The Paper
even says

"We noted that most of the sequence information is difficult to interpret, and discussed error rates"

Ummm, ok. Nice counseling session.

"patients with whole genome sequence data need information about more diseases with a wide clinical range"

Perhaps that person could actually be a physician, maybe a generalist?

"For this we offered extended access to clinical geneticists, genetic counsellors and clinical lab directors"

Nice! Joubert's is not Gilbert's is not Plavix. Thanks for stopping by.

I did appreciate that your paper calculated pretest probabilities. Unfortunately these were based on a pedigree which had no ethnicity and incomplete clinical data.

1. No Glycohemoglobin to evaluate for diabetes risk or maybe even diagnose it
2. No Iron Studies to evaluate for Hemochromatosis, yet you state genes may set him up for it.
3. No documentation of a physical exam including DRE for prostate hypertrophy/cancer or PSA
4. No dietary history? No Smoking history? No social history?

Shall I go on?

You show increased risk for Diabetes post test as well as prostate cancer, obesity, CAD, MI, Asthma, NHL, RA (no ESR/CRP/CCP?)

You projected an increased risk for 7 and decreased for 8. Yet no Assessment of MCI etc in Alzhemiers disease? My god, you did a stress test in an asymptomatic patient who exercises daily.

"Although the methods we used are nascent, the results provide proof of principle that clinically meaningful information can be derived about disease and response to drugs in patients with whole genome sequence data"

Translated: We made up a system and used novel DNA results to hypothesize about disease risk using research fellows, computer programs an excellent cardiologist (Not a GP) and an Echo machine.......But we skimped on the physical exam, use of primary care doctors, complete blood counts and other clinically useful testing and procedures.

I admire your efforts, but

A. You have missed the boat in using not all the tools at hand
B. By being Genome-centric, we miss the clinical picture.

"Although no methods exist for statistical integration of such conditionally dependent risks, interpretation in the context of the causal circuit diagram allows assessment of the combined effect of environmental and genetic risk for EVERY individual"

Translation: Nothing exists statistically to evaluate disease interaction and how it may increase risks of interlinked disease.

Ask yourself, "What have we done to make Stephen Quake healthier from this test?". Other than hype the use of a genome clinically?

This paper was all genome and NO CLINICAL ASSESSMENT!

The Sherpa Says: The only thing of note that is important here is the CYP2C19 data.......
I have seen abnormal CGH data in a child with severe developmental delay come directly from a high functioning mother who was a power litigator. The genome scan as it stands now is noise. It also requires a full team a month to intepret. Clearly not ready for medical prevention or prognostication, sorry.


Red Herring said...

Well, at least he got a modicum of that most precious clinical resource - time with a well-credentialled physician. Then again, how long did he spend in the waiting room. As you know, if FT/WT > 1.0 (the ratio of face time to wait time greater than zero), the old lady from Dubuque will be happy.

Anonymous said...

If you look at the supplement that describes the amount of work and curation that went into this paper, it represents a laudable improvement in interpretation of genetic variation compared to what DTC companies currently provide.

While I think personal genome sequencing and interpretation is still premature for clinical use, the paper made use of novel methods for interpretation of complex disease risks. I don't think it is a big deal that the paper failed to describe the results of doing a DRE on Steve Quake.

BTW, you seem awfully critical for someone who has never published a single paper in this area of research yet blogs like he is an expert.

Steve Murphy MD said...

Clinically, Yes I am an expert. Just because I haven't published a paper that represents nothing more than a description of how they curate a genome and infer clinical risk does not make them any more fantastic than anyone else.

I think you should really spend some time doing medicine before you pop off with angry attacks without any justification.

Have you read my editorial?

Nature Biotechnology "In Need of a Reality Check?"

Maybe you should pubmed me first?

Oh and probably the fact that I sit on 2 research boards solely dedicated to this area of work means nothing as well?

Great curation of a genome, horrible clinical assessment. That's my take

Anonymous said...

No offense Steve, but most of your rants on this blog qualify as you "popping off with angry attacks" that are often way over the top and rude.

And I have spent time doing medicine (I'm an MD) and I have looked you up on Pubmed - not impressed (hence my previous comment).

Steve Murphy MD said...


Would love to know who you are so we can see how fantastic your CV is.

Either way, I stand by my comment.
Great way to clinically analyze a genome. Horrible way to clinically analyze a patient.

If you were part of the paper, great. But next time do the right medical tests to augment your genome.