News Comments -- George Church

Gene tests raise murky medical issues (Boston Globe, 30-Dec-2013, Callum Borchers)

"sequencing holds little value for most people at this time." --Robert Green
Comment: Car airbags and vaccines hold little value to "most people". How do you know if you fall into the category of "most people" in advance?

"mythology that sequencing can tell us absolutely everything. The truth is, it tells us a lot of things that are incomplete or ambiguous." --Robert Green
Comment: Blood pressure and throat exams also do not tell us "absolutely everything". Diagnostics normally focus on what can be done rather than what cannot. Why not for genetics?

"There isn't really a compelling case for doing it." --Mike Eisen
Comment: Given that 3% (or 10%, see of babies have (highly penetrant) recessive diseases avoidable by pre-martial/pre-conception counseling and given that the (reimburseable) cost of treating each is $50K to $200K per dose for life (or $1.6 M per one-time cure like Glybera) plus costs to the family, e.g. working-hours and stress-related issues. At $2000 per genome, you only need to save $70K per affected child to be cost-effective, averaged across the population.

--George Church 1-Jan-2014

Note that the $2000 cost above has been available for bulk orders for 3 years and is now close to $1000 (see: Illumina HiSeq X Ten, Forbes view)

The current CLIA approved whole genome services for individuals (not bulk) include:

1) since Nov-2012 : Illumina Trugenome   (cost $5000 for technical sequence data).

2) since Aug-2013: Partner's Healthcare   (cost $9000)

3) Most of the other CLIA approved labs only offer partial genomes. Please let me know if you know of others.

One other option (research not CLIA) since Jul-2009:   (no cost to the individual, but requires open sharing of the data).

--George Church, updated 13-Feb-2014

Harvard geneticist George Church on genome testing: "I wouldn't wait. I didn't wait." (Boston Globe, 13-Dec-2013, Callum Borchers)

Improving Genome Understanding. (Nature, 9-Oct-2013, George Church)

Comment: Reasons that wealthy, tech-savvy citizens avoid learning about their own genome include:

1) Many focus on what personal genome knowledge can't do rather than what it can do. We bought cell phones before they were smart phones, before they had cameras, etc. But we won't buy a genome when "all" it can detect is Tay Sachs, HCM and a few hundred other highly predictive and medically actionable alleles.

2) Many assume that since most genomes don't produce actionable results, then there is no reason to study our personal genomes. Does the fact that most cars never deploy their airbags mean that we should not have airbags? When we ask what did one person (e.g. George Church) learn from one genome, we fall into this trap.

3) Many seem to think that it is hard for a medical geneticist (and/or software) to report only the actionable results, and hence many people will panic and request unnecessary medical therapies. We have the tools already to focus on actionable genes, just as we have the tools to focus on actionable blood chemistry and actionable imaging data. Each medical test has potentially distracting or scary incidentalomas, but zero is not the correct place to set the bar for any of those diagnostics.

4) Confusion over ways to reduce costs. Some feel that common SNPs, exomes, disease-specific panels, ancestry, or family risks should be prerequisites (or alternatives) for whole genome sequencing (WGS). But those have serious blind spots (eg detecting inversions and translocations which can be de novo null alleles). Those 5 can be used as post-filters rather than pre-requisites for WGS. WGS has a current cost of $1000 and $5000 price. Exome prices are $1000 to $5000. DTC SNP test price= $99. And doing nothing is $0. Should our epitaphs say that we saved $5000, but lost 10 to 70 QALY (quality-adjusted life years)?

5) Many dismiss highly predictive DNA variants as "rare" or non-actionable. Most are actionable via pre-conception genetic counseling. Others actionable even at late onset via preventative surgery, drugs etc. This is not about small probabilistic effects or pharmacogenetics or generic lifestyle advice, but very high impact actions. Yes; these are individually rare, but collectively common (in the 5 to 9% range depending on population). Many so-called common diseases (like cancer at 10-15% of deaths) are not more common since they represent lumping of many different diseases. We all have N=1 histories and we need to make diagnoses (and research) based on large, precision medicine datasets, including environmental and WGS.

Example of a genome variant report from PGP1

--George Church 13-Dec-2013 (corrections, additions, other sugestions are welcome)