The next-gen sequencing feeding frenzy has a new customer in mind: clinical laboratories. Given the rapid advances this kind of sequencing has enabled in basic research of disease, it's a savvy move — but one that may be premature for a clientele that interacts directly with physicians.
Rong Mao, an assistant professor at the University of Utah and co-medical director of molecular genetics at ARUP Laboratories, has embraced the new sequencing platforms and is trying to figure out how to make them work in a clinical setting.
The promise, Mao says, is tremendous. Take cystic fibrosis as an example: originally, testing for this disease was performed using a single mutation; today, a panel of 32 mutations is used to improve accuracy. However, even that readout only offers about 90 percent sensitivity in a Caucasian patient and about 70 percent in an African-American, Mao says. As far as physicians are concerned, she adds, those aren't numbers you can report to a patient. Sequencing the entire cystic fibrosis gene boosts sensitivity to about 97 percent, and sequencing in conjunction with deletion and duplication data brings it to about 99 percent, Mao says. That's where physicians finally feel comfortable presenting results to a patient.
Clearly, using sequencers is a no-brainer. But with the number of sequencing reactions needed — "for clinical diagnostics, we need every exon sequenced from both directions for quality control purposes," Mao says — Sanger technology is far too expensive. (For cystic fibrosis, she says, there are 27 amplicons to sequence, with a total of 54 sequencing reactions for quality controls.)
While next-gen sequencing is the obvious solution, it still doesn't bring the price down to where it will have to be for frequent use in a clinical lab. Mao's team has begun to experiment with pooling samples as one possible solution. "This way you can significantly reduce the cost," says Mao, whose lab does not currently have a next-gen sequencer. She does have access to and performs studies on both the Roche 454 and the Illumina platforms, and hopes to buy an instrument this year or next. "We're still in the stage of shopping the different platforms and [seeing] which one would be the best to fit our application," she adds.
Cost reductions will only go so far in getting the technology implemented in clinical labs. "We need some quality control parameters to be able to monitor for this sequencing assay," Mao says. Her team is working on this problem right now, which involves defining an appropriate cutoff for heterozygous calls and figuring out how much depth is required to confidently call a base using short-read platforms. Sample prep is another challenge. Long-range PCR has the double whammy of needing a large number of reactions and being fairly expensive, while genome selection arrays are still very new to the market. "The capture array still has a lot of problems — for example, the uniformity of capture," Mao says. "You will see peaks and valleys from the next-generation sequencing results because the probe [does] not always perform equally." For a disease like neurofibromatosis, where pseudogenes have as much as 90 percent sequence conformity, these arrays have proven unable to consistently capture the functional gene and not the pseudogene, Mao says.
The technology will also have to be feasible — technically and in pricing — for diseases more complex than cystic fibrosis, which is linked to a single gene. Congenital hearing loss, which Mao says affects one in 500 people and may be included in routine newborn screens, has 19 associated genes.