NanoString Nails Breast Cancer Study, Challenging Genomic Health
Seattle-based NanoString Technologies has made a big bet that the future of the company depends on turning its genetic analysis research tool into a diagnostic workhorse. Today, it presented some hard data that suggests it is on its way.
The company reported today that its instrument, called nCounter, was able to predict whether or not women with early-stage breast cancer were likely to have a recurrence in the future, by looking at an array of 50 genes known as the PAM50 signature. The information went beyond the usual 21-gene analysis done by Redwood City, CA-based Genomic Health (NASDAQ: GHDX). While some women get clear information today from Genomic Health about when they are at “low” or “high” risk of recurrence, doctors and patients often are confused about how aggressive they should be with chemotherapies when the results put patients into the “intermediate” risk group. NanoString’s tool could prove valuable because it classified fewer patients in the “intermediate” risk category.
The findings for the NanoString study were presented today in front of several thousand people at a plenary session of the San Antonio Breast Cancer Symposium.
“This is one of the single most important events in NanoString’s history, along with the launch of our first commercial system,” says Brad Gray, NanoString’s CEO, when reached by phone at the conference. “It really validates we can develop diagnostics on the nCounter, and that our first product is likely to be a significant one.”
NanoString, a spinoff from the Institute for Systems Biology in Seattle, introduced the first commercial version of its tool in the summer of 2008, for research purposes only. The technology provides scientists with a digital readout on the extent to which genes are dialed on or off in a sample—what’s known as gene expression analysis. The tool has gained popularity the past couple years with academic customers, especially those at The Broad Institute of MIT and Harvard, because of its ability to help spot patterns in complex diseases like cancer where 50 or 100 genes might be perturbed instead of just one.
But the research market has its limits, and NanoString has been thinking big about the diagnostic potential for the nCounter. In October, the company introduced a second-generation product that’s supposed to have 50 percent higher bandwidth (known as throughput in the genetics business); more flexible software for analyzing the data from the instrument; and hardware that is manufactured in line with more rigorous, consistent diagnostic industry standards. Last month, the company raised $20 million in venture capital from a syndicate that includes GE Healthcare and former Genzyme CEO Henri Termeer, to help pursue its diagnostic opportunity.
NanoString’s vision is to challenge Genomic Health directly, with a different kind of business model, Gray says. While Genomic Health runs its sophisticated 21-gene analysis at a centralized lab that doctors ship samples to, NanoString envisions selling instruments to labs around the world so technicians can perform the diagnostic test themselves on site.
Here’s how the study worked: NanoString obtained 1,017 tumor samples taken from women in England who enrolled in the landmark study known as ATAC, and got acceptable results from 1,007 of them. The women had early-stage breast cancer, and were treated a couple different ways. They were followed up for 10 years, and records exist on how many of them ended up having recurrences or not.
NanoString analyzed the tumor samples with its 50-gene signature in about six weeks at its Seattle labs, in September and October, Gray says. Based on what the machine said, NanoString classified women as either low, intermediate, or high risk, just like how Genomic Health validated its Oncotype DX breast cancer test, Gray says. Once the data was gathered, researchers then compared the NanoString predictions for each individual with what records showed actually happened.
The NanoString and Genomic Health tests were almost the same when it came to classifying women as “low” risk, as the NanoString machine put 428 women in that camp, and Genomic Health’s test classified 434 women that way, Gray says. But the big difference was in the “intermediate” and “high” risk groups. NanoString’s test put 192 women in the “intermediate” gray zone, while 243 women were put in that group by Genomic Health, Gray says.
Mitch Dowsett, a breast cancer researcher at The Royal Marsden Hospital and The Institute of Cancer Research in London, and the lead investigator working with NanoString, said in a statement: “In the 1,007 samples for which results from both tests are available, the NanoString PAM50 ROR score provided significantly more prognostic information than the Oncotype DX RS score and assigned fewer patients to the intermediate risk group than Oncotype DX,” test from Genomic Health.
The data wasn’t made available in advance to conference organizers, as it usually is, and it wasn’t summarized in abstracts beforehand, so people were clearly processing the information for the first time as they heard it live this afternoon in San Antonio. In an unusual move, conference organizers agreed to showcase the study results before the data even existed, because of their strong interest in the design of the study, and the information it might yield, Gray says.
“What we saw was that PAM50 is a very accurate predictor of whether women will or will not recur,” Gray says.
Steve Shak, Genomic Health’s chief medical officer, offered the following statement tonight on the new data from NanoString: “While this is interesting early, prognostic data, it needs to be replicated to demonstrate predictive value and in order to be applied to clinical practice. Oncotype DX is the only test that has clinical evidence validating its ability to predict the likelihood of chemotherapy benefit as well recurrence in early stage breast cancer.”
NanoString designed this study so that it could stand on its own as part of an application for regulatory approval to start selling its test in certain parts of the world, such as the European Union, Gray says. More data from another trial will likely be needed before NanoString attempts to seek U.S. approval from the FDA, he says. But because these trials are retrospective—meaning they depend on getting samples and records on patients that enrolled in past studies—they can be done more quickly than so-called “prospective” studies which would require recruiting and enrolling more than a thousand new patients and then following them up for years to see if they have recurrences.
Gray declined to offer a timeline for when the next study can get started and finished, and he didn’t say when the company wants to turn in its application to the FDA. But he did note that NanoString’s first meeting with the principal investigator of today’s study was only one year ago at the same San Antonio conference, and it only took the company six weeks to go from processing its first tumor sample to generating the data on the accuracy of its predictions.
“We’d like to go a little further down the path before committing to timeline,” Gray says. But he added, “we’re working as diligently as we possibly can on our second act.”