Tuesday, February 26, 2013

Exome sequencing wins where traditional medical tests fail

This year's Advances in Genome Biology and Technology conference was densely packed with clinical applications of high-throughput sequencing, with the theme becoming a clear winner throughout the few days of AGBT.  Memorable talks by Christine Eng and Stephen Kingsmore last Thursday provided a high level overview regarding the state of the field.

Christine Eng, of Baylor College of Medicine, showed that clinical exome sequencing could yield a diagnosis for about 25% of patients tested.  The whole point of exome sequencing over other approaches, such as specific test for mutations in a small number of genes, is to scan for rare genetic variants that are still clinically significant.

Eng's Baylor lab is a CLIA-certified genetic testing laboratory and provides sequencing based diagnoses to physicians as clinical reports.  Exome sequencing can capture about 85% of known disease causing variants, said Eng, and raw reports from clinical samples yield about 200 thousand variations, with final reports seen by physicians showing a very small number of clinically actionable variants: generally under five.

The definition "clinically actionable" generally means that a known drug will work patients with that variant and symptoms, or that a clinical trial for that variant is underway.

Many other talks at AGBT emphasized that clinical physicians generally don't have the time or expertise to evaluate whether the hundreds of mutations that everyone are relevant to the patient.  However, Eng mentioned that a third of orders do request a more complex report, with about 25-30 actionable mutations.

The Baylor lab's throughput is about 130 exomes per month, and the test costs about $7 thousand, a number that surprised some members of the audience.  As research advances, I expect that that price tag will fall.  Eng also gave an example of a patient without a firm diagnosis until exome sequencing was performed, after enduring $25 thousand in traditional tests.

In another talk, Stephen Kingsmore discussed genetic testing of newborns, 5% of which require neonatal intensive care. In these cases, getting a diagnosis is an emergency, said Kingsmore, and doing so is especially challenging as there are about 3,700 single gene diseases, many of which present with similar clinical features in a newborn.

Kingsmore discussed STAT-Seq, Children's Mercy Hospital's pipeline for producing clinical reports from sequencing with about a 50 hour turnaround time.  He estimated the cost of testing are "about $13 thousand", but mentioned that their team is trying to get the cost down to about $5 thousand.  I suspect the price of STAT-Seq is higher than Baylor's test due to the rapidity of the former.

Where the test shines, said Kingsmore, is in determining what isn't wrong with a newborn: "Genome sequencing is very effective at ruling out diseases", he says, "you can eliminate unnecessary treatments and you can eliminate futile treatments".

Another great application he describe was in making dose adjustments; many genomic variations are associated with pharmacogenomics that are out of a normal range.  In other words, one can identify newborns that need lower or higher doses of medication for the same effect, without the guesswork and time that it would take by trial and error.

One surprising result common to both presentations was that diagnoses using clinical exome sequence were useful even when no treatment options were available.

Kingsmore, in particular, explained that firm, molecular testing based explanations have profound psychological benefits: Parent's uncertainty regarding their children can be dispelled, or they're able to prepare for intense treatments that are needed.

As with most technology, clinical sequencing is being rolled out in cases with the most need or demand first, where the impact of the results far outweigh the expensive price of the analysis.  But I don't think it will be long before similar tests are available for many other medical issues, none as critical as newborn diagnoses, but no less important to the patients that will need them or just request them.