By Aaron Krol 

March 14, 2014 | When the GeneKey cancer profiling service launched in 2008, as a division of CollabRx called CollabRx ONE, it was unclear whether the program could truly contribute to treatment decisions. The idea of personalized cancer care was already quickly gaining ground, with a growing number of therapies, especially in clinical trials, targeted to cancers with mutations in specific genes. A few companies, like Caris Life Sciences, were even beginning to look at the abundance of certain proteins in the tumor, or at the genetic sequence directly, to decide if a patient was a promising match for one of those therapies. But these approaches could rely on well-developed biological frameworks, narrowing the search to the few biomarkers that had been confidently connected to treatments. GeneKey wanted to look farther afield.

“When we ask oncologists how much of the biology of cancer is known today, I can tell you the median response is about five percent,” says James A. D. Smith, who became GeneKey’s CEO early last year. “So if you’re using the biomarker approach, you’re by definition in the five percent slice of the pie.” GeneKey wanted instead to sequence the whole exome and transcriptome of each patient’s tumor, as well as their normal tissue, to get a glimpse of the full spectrum of mutations. “Instead of generating a few hundred data points, we’re generating tens of thousands of data points for each individual patient,” says Smith.

If it sounds intuitive that more data equals greater knowledge, consider that GeneKey typically finds between one and two thousand mutations in each tumor it sequences, depending on the cancer’s progression. (Older tumors tend to accumulate more mutations.) Most of these changes prove irrelevant; some are contradictory. They can appear in any area of cellular biology, meaning GeneKey needs the guidance of outside experts to interpret molecular pathways of which the company may not have in-depth knowledge. And with information on both the DNA changes and the RNA expression levels of each gene, GeneKey is courting a huge mass of redundant data. Shouldn’t DNA mutations be predictive of RNA expression? What could be learned by flagging both?

A Telling Case 

“We didn’t know what to expect initially,” says Chief Scientist Raphael Lehrer. Lehrer is the founder of GeneKey, who began as the leader of the CollabRx ONE unit and carried the program forward as it split off from CollabRx in 2010. It wasn’t obvious, when Lehrer started taking on patients, that the combination of comprehensive DNA and RNA sequencing would yield any new insights – or whether any findings that did come of the program could add up to therapy recommendations.

Then, in 2009, GeneKey profiled a patient with a perturbation in the PIK3 pathway: a frequent den of bad actors in cancer because the genes in the pathway are involved in cell growth and proliferation. One protein in particular, mTOR, is of great interest to oncologists, because tumors that overexpress mTOR have a number of therapy options, derived from the drug rapamycin.

In this patient, unfortunately, the RNA data clearly showed that mTOR was downregulated, indicating that these therapies were inappropriate. Any standard DNA or RNA analysis of the tumor would have led to the same conclusion.

“The thing that struck us,” says Lehrer, “was we were looking upstream and downstream of mTOR more generally, and in the pathway immediately above it, we probably found ten or fifteen different disregulations, copy number changes, and one or two mutations, which all pointed to an activation of mTOR at the proteomic level. And downstream a bit, it looked as if it were active, in the sense that the proteins it was supposed to impact appeared to be impacted.” That is, the upstream DNA data seemed to hint that mTOR should have been overexpressed – and the downstream RNA data showed a lot of proteins responding as if to high mTOR levels.

GeneKey formed a novel hypothesis. The mTOR protein, as the data clearly showed, was being suppressed; but the active, phosphorylated form of mTOR was in overdrive, and that was driving the cancer. “We actually tested that out in the patient’s sample with a phosphor-proteomic assay, and found – in some sense to our surprise – that was in fact the case,” says Lehrer. The patient may respond to mTOR-targeted therapies after all.

The Large Toolbox 

These unusual cases, where GeneKey’s pathway-wide analysis reveals the exact opposite of what a targeted gene test would suggest, are exceedingly rare; Lehrer says he’s seen this scenario play out just twice. But they serve as a striking illustration of what can be missed when the standard biomarkers serve as proxies for the whole cancer. Only by looking at all the genes in the PIK3 pathway, and by comparing both DNA and RNA data, could the GeneKey team arrive at an informed therapeutic strategy.

To date, the GeneKey team has accepted somewhere between 20 and 30 patients. The service strives to be ready with a treatment option no matter what aspect of cellular biology is being disrupted. This often involves reaching out to GeneKey’s scientific advisory board, or to authors of relevant papers, to put together a personal review team for each patient. “We have a number of consultants, and then get additional consultants as needed when we find biology that does not fall in our expertise,” says Lehrer.

It also involves being prepared to deliver outside-the-box suggestions. The patients who seek out GeneKey have typically failed at least a few lines of therapy, so Lehrer says it’s less common for the team to come up with a drug approved for the patient’s own cancer type; the patients who achieve long-term remission from those drugs generally never reach GeneKey in the first place. Most commonly, the consultants will recommend a treatment from another branch of cancer, based on shared molecular targets. They also look for ongoing clinical trials, which in oncology are now almost always targeted to small genetic subpopulations; with whole exome sequences of each patient’s tumor, GeneKey can quickly find these matches.

Often, GeneKey will suggest a drug that wasn’t designed for cancer at all, as one component of a treatment strategy. Repurposing drugs is a hot topic in basic cancer research, but it remains unusual in frontline care, in part because if a genetic target is so far afield of standard practice, it almost certainly won’t be sequenced in the first place. The GeneKey team is wary of the experimental nature of any therapies with their origins outside cancer care, but, inspired by high-profile repurposed drugs like metformin and thalidomide, they will pay some consideration to any drug that hits the right pathway. “Typically,” says Lehrer, “when we’ve found a non-cancer [treatment], we’re able to link it to various mechanisms of oncogenicity. For instance, we might combine a non-cancer drug that hits more at the root with a cancer drug that hits more at the downstream effect level.”

Since GeneKey goes where the data tells it to, it sometimes arrives at scenarios that are not ideal – for instance, clinical trials or drug combinations with less evidence behind them. But in working with patients with the most difficult cases, the company must sometimes balance the strength of evidence arising from the patient’s data against what’s in the literature. That’s why GeneKey stresses its role as a consultant service to oncologists, and always takes a back seat in actually making treatment decisions.

“What we’re doing is discussing with the oncologist what we found in the patient, and what the evidence is for it,” says Lehrer. “So for instance, if there’s a non-cancer drug whose use in cancer has only been studied in animals, then we can tell them that. And they can make educated decisions based on their own assessment of the patient, of the evidence, and of the risk-benefit of the drug.” In fact, while patients do sometimes seek out GeneKey on their own initiative – until 2013, word of mouth was the only way the company reached patients – Lehrer considers the oncologist’s interest to be just as important if GeneKey’s approach is to make a difference. “We don’t accept a patient into the program until after we talk to the oncologist,” he says, “because the end consumer of our analysis is the oncologist.”

Niche Market or Industry Model? 

In at least one sense, GeneKey has exceeded its founders’ expectations. Asked how often the company is able to recommend a new therapeutic strategy, Lehrer says, “It’s much higher than we would have thought. It’s in the range of 95%... And I’d say in about 85% of those cases, it includes something that the oncologist has not yet considered – or often, they say they would not have considered.” A new treatment is far from a sure bet in cancer, but GeneKey’s suggestions have often slowed or reversed the disease’s progression. With only a few dozen cases under its belt, the company is already improving and extending patients’ lives.

On the other hand, as GeneKey moves from proof-of-concept to commercial stage, the future of its exhaustive style of analysis is still unknown. Since GeneKey opened as CollabRx ONE in 2008, biomarker-based cancer diagnostics have flourished. Caris Life Sciences continues to grow, and has been joined by competitors like MolecularHealth and current investor darling Foundation Medicine, whose service sequences a record 200 cancer-associated genes.

Foundation Medicine’s success relies in large part on balancing its desire to cast a wide net against a more conservative impulse: it only tests genes connected to well-validated treatment options. This makes it easy to convince payors that the company’s tests are cost-effective diagnostics, and a large number of insurance providers – including, crucially, Medicare – are willing to cover a Foundation Medicine test. Restricting the number of genes sequenced also keeps the list price under $6,000 per test. These factors helped Foundation Medicine to sell around 3,000 tests in 2013.

GeneKey, by contrast, performs two whole exomes on every patient – one for the tumor, and one for the normal tissue – amounting to around 25,000 genes, twice over, before considering the transcriptome. The price tag of the service is in the tens of thousands of dollars, and so far, insurance companies haven’t offered reimbursement. Add in a much greater time commitment for each patient, focused on exploring brand-new oncological pathways, and GeneKey faces obstacles to scaling up to meet its biomarker-based counterparts – although Lehrer points out that the company is now raising funding to do exactly that.

Smith, the CEO, hopes this state of affairs can change in the future. “We’ve had some preliminary discussions with at least one of the large national health insurance carriers, who are looking at molecular profiling generally so that they can advise their plan members,” he says. “They clearly got it. They understand how it can save money for the system, because frankly, at the cost of new, highly-targeted therapeutics, avoiding using the wrong one just once will certainly pay for a GeneKey profile.” For now, however, the service is an out-of-pocket expense, which severely limits the number of patients it can reach.

On the other hand, GeneKey offers by far the more comprehensive service. By definition, it can characterize any patient Foundation Medicine can, plus many more. And the cases that GeneKey has the unique power to solve are the same cases that tend to advance the state of knowledge around cancer. “The output [of our analysis] looks a bit like a journal article on a specific patient,” says Lehrer, revealing not only which genes are mutated, but how they interact with related genes and ultimately change the balance of proteins in the cell.

In the larger picture, GeneKey’s greatest influence on cancer care may come not from the patients it works with directly, but from similar patients who benefit from the company’s deep research. GeneKey hasn’t yet published case studies – “we’re in the process of writing our first manuscripts,” says Lehrer – but it does patent novel indications, making them publicly visible.

“The first example of such a patent was issued last spring,” he adds, “for the use of multiple tyrosine kinase inhibitors in BRAF-mutant metastatic melanoma patients who’ve been resistant to BRAF inhibitors.” That’s another interesting case of GeneKey’s method picking up an unexpected twist on known biology. A melanoma patient with unusual BRAF expression levels was not responding to the usual combination of BRAF and MEK inhibitors, so GeneKey looked upstream in the pathway. Their profile traced the change to a mutation to a tyrosine kinase inhibitor, and recommended sunitinib, a drug normally used in renal cancer. “We had a call with the oncologist,” says Lehrer. “The patient was put on the drug the next day, and there was a symptom response within 48 hours – the abdomen shrinking to normal size, the patient able to eat where they couldn’t before and have normal bowel movements.”

These are inherently n-of-one cases, and to give rise to new pharmaceuticals, the basic findings of GeneKey would have to travel through the decades-long drug discovery and clinical trials process. But as long as the company looks at the spectrum of FDA-approved drugs, its success stories will open up new options for physicians and patients who aren’t served by the existing boundaries of knowledge.

At the same time, the cost of genetic sequencing continues to fall, while the number of genes that are known cancer biomarkers grows. In the next five or ten years, the price difference between GeneKey’s service, and the biomarker panels of companies like Foundation Medicine, could start to look less forbidding – and the two strategies could even start to bleed together. As physicians and payors adapt to personalized medicine, today’s exhaustive approach may yet become tomorrow’s standard of care.