Fingerprint Cancer

Where a Cancer Lives May not be as Important as What it’s Made of

C3Fall2013_Final-4Imagine you live in Buffalo or El Paso or Tallahassee. Would you still eat Thai food, listen to the Beatles and watch Harrison Ford movies? Of course you would. Sure, if you live in San Francisco you might be a little more likely to experiment with yoga and if you live in Manhattan you might be a little more likely to be a Yankees fan, but no matter where you live, you would remain you.

Researchers at the University of Colorado Cancer Center are learning the same is true of cancer. For example, it turns out that two lung cancers may be as different as a skinned knee and a runny nose—if they’re made by different genetic mutations, you can’t treat them with the same targeted drug. And far-flung cancers like breast and bladder may be stopped cold by the same treatment, as long as their genetic structures are similar.

Cancer used to be defined by where it lives in the body. Increasingly it’s being defined by the fingerprint of the genes that create it. It’s not just where it lives, but also what it’s made of that counts. And CU Cancer Center is at the leading edge of this paradigm shift in the way we classify and treat the disease.

Which Keys Fit Which Locks?

C3Fall2013_Final-5Let’s start with a Colorado hometown story and a good example of how replacing a cancer’s site-specific label with the code of its genes can lead to new, life-changing treatments.

“Think about treating cancer as putting a key in a lock,” says Robert C. Doebele, MD, PhD, associate professor of medical oncology at CU School of Medicine. “A cancer-caus­ing genetic mutation is like a lock. Then for each lock, there’s the specific key of a drug. You put the right key in the right lock and you’ve unlocked or solved a type of cancer.”

So what keys and what locks?

“People go about it from both directions—they have a drug and wonder what cancer it could treat, or they have a cancer and wonder what drug will treat it. In this case we decided to look at the locks first—what previously unknown genetic mutations cause lung cancer?” Doebele says.

It sounds like a big question and it is: There are about 40,000 genes in the human body. That’s 40,000 bits of stuff that make things (to use scientific terms). Unless you have an identical twin, chances are darn good that no one else on earth has the same stuff as you. And everything from sunshine to overcooked steaks to living and breath­ing has the potential to change the stuff and things you were born with so that by the time you’re reading this article, not all your cells look the same.

In other words, the human genome is a massively confusing soup through which not even a psychic or a Basset Hound can sniff. That is, until you pare it down from 40,000 genes to the top suspects. See, there are 180 or so genes that have been identified as very common causes of cancer.

For example, one of these genes is BRCA1. When in good working order, BRCA1 repairs damaged DNA in breast tissue, or if a cell’s DNA is beyond repair, BRCA1 helps to mark that cell for destruction. But when BRCA1 is itself damaged, it can’t repair or destroy damaged cells and so these damaged cells with their suspect DNA stick around and cause breast cancer.

There are roughly 180 genes like BRCA1 that, when malfunctioning, have been shown to cause cancer—and so rather than searching through 40,000 genes, Doebele limited his search to this worrisome shortlist.

A Thyroid Cancer Gene In Lung Cancer Tumors

But why look for locks that you already know exist? Well, many of these genetic locks have only been paired with cancers at the site where they were discovered. For example, BRCA1 is known as a breast cancer gene. But recently, it’s been implicated in ovarian and even bladder cancer. Likewise, genes that control cell growth in the presence of androgens have long been known to drive prostate cancer, and researchers at CU Cancer Center have shown that these same genetic signatures can contribute to bladder and breast cancers. And mutations in the genes EGFR, ALK and ROS1 have been shown to drive lung cancer—and now it turns out these mutations also drive subsets of colon cancer.

So there’s precedent for the genetic causes of cancer jumping the sites where they were discovered. Instead, one mutation may drive many different “kinds” of cancer at many different sites.

Doebele wondered if any known genetic causes might jump into lung cancer. And so he collected 36 samples from tumors that tested negative for any known genetic driver of lung cancer and sent them to the lab for gene sequencing. If not a known driver, maybe there was an unknown driver? And if there’s an unknown driver, maybe it was one of these 180 suspicious candidates?

Sure enough, samples started coming back positive for mutations in the gene NTRK1. If Doebele’s samples had been from thyroid tumors there would’ve been no surprise: When accidentally mashed together with another nearby gene, NTRK1 signals cells to constantly grow and divide, leading to the cancer known as papillary thyroid carcinoma. But Doebele’s samples weren’t from the thyroid; they were from lung tumors.

In other words, the human genome is a massively confusing soup through which not even a psychic or a Basset Hound can sniff. That is, until you pare it down from 40,000 genes to the top suspects. See, there are 180 or so genes that have been identified as very common causes of cancer.

For example, one of these genes is BRCA1. When in good working order, BRCA1 repairs damaged DNA in breast tissue, or if a cell’s DNA is beyond repair, BRCA1 helps to mark that cell for destruction. But when BRCA1 is itself damaged, it can’t repair or destroy damaged cells and so these damaged cells with their suspect DNA stick around and cause breast cancer.

There are roughly 180 genes like BRCA1 that, when malfunctioning, have been shown to cause cancer—and so rather than searching through 40,000 genes, Doebele limited his search to this worrisome shortlist.

Testing Tumors

C3Fall2013_Final-6But was it the random chance of an unlucky couple of lung cancer cells that happened to hold this mutation—or was the pattern prevalent enough in the overall tumor to make it worth targeting?

Doebele didn’t know. Yet.

He took the idea back to the lab of Leila Garcia, PhD, director of the Cytogenetics Shared Resource at CU Cancer Center and a top, international expert on designing tests that could show how many cells inside a tumor hold a specific genetic mutation. Garcia looks at cancer with the technique known as fluorescence in-situ hybridization (FISH). She designed a FISH test to look for NTRK1 fusions. Basically, it made cells with this genetic anomaly light up. And cells within Doebele’s lung cancer patient samples lit up.

“So I walked up the street to Arrary BioPharm, a biotech firm in Boulder and it turns out they had a couple compounds that target this gene just sitting on the shelf,” Doebele says. In other words, now with a clear picture of the lock, it turned out there was already a key and it was sitting there in a lab not a half-hour drive from Doebele’s office.

With a lock, a test to know where the lock lives, and now a key, Doebele is laying the groundwork for possible clinical trials that would use one of Array’s compounds to target the gene NTRK1 in the subset of lung cancers seen malevolently active on Garcia’s FISH test.

That’s how collaboration gets done.

Fingerprints, Specs, and Drug Development

C3Fall2013_Final-6.2This is cutting edge. It’s right here. But CU Cancer Center certainly isn’t the only institution where researchers are picking apart cancer based on mutations instead of sites where it lives.

The National Cancer Institute is prioritizing a program called Strategic Partnering to Evaluate Cancer Signatures (SPECS) to help cancer centers and other research institutions pool their expertise to discover new mutations, define their importance and design tests to see which tumors hold these mutations. Basically, this groundwork lays mutations at the feet of drug companies who have the capital and expertise needed to bring drugs to market that target these mutations.

And it’s a major revolution for targeted cancer therapies. Imagine this: you find a mutation that’s present in a tiny subset of one cancer type. Do you think it’s worth many, many millions of dollars for a drug company to bring a drug to market that they can eventually hope to sell to this small slice of the cancer population? Don’t count on it.

But if a specific mutation isn’t just driving the growth of, say, 2 percent of lung cancers, but also 2 percent of colorectal and ovarian and head & neck and prostate cancer, all of a sudden you have a larger market and a powerful incentive for drug companies to develop the treatment. By finding mutations across cancer types, SPECS may not only give drug companies targets, but may demonstrate the market potential that can help the idea of a drug become reality.

With that in mind, national and international researchers from the SPECS group recently met at CU Cancer Center to take stock and explore next steps. Among this group of elite researchers is Fred R. Hirsch, MD, PhD, professor of medical oncology and pathology at the CU School of Medicine. Hirsch leads the only SPECS group focused on lung cancer. And basically the challenge is this: While genetic ways to understand and treat the most common form of non-small cell lung cancer have exploded in recent years, its cousin, squamous cell lung cancer, has lagged far behind. Treatment options are basic: surgery, chemotherapy and radiation.

“The SPECS program has been very successful,” Hirsch says. “The FDA recently approved a new, genetic signature in micro-RNA that can help refine the prognosis for breast cancer, discovered by a SPECS group. And the leukemia group is also far ahead. The lung group is the youngest in the SPECS family but we’re honored to have received a sizeable grant a year ago to bring the understanding of squamous cell lung cancer in line with these other diseases.”

The SPECS lung cancer group that Hirsch heads includes researchers from eight institutions in the U.S. and Canada, and is at the stage of genetically profiling around 1,000 squamous lung cancer samples. Then they will pair this profiling with patient outcomes. What genetic signatures predict strong outcomes and which signatures predict poor outcomes? If Hirsch can discover which genes are “bad” in squamous cell lung cancer, he can immediately offer genetic models that can help patients and doctors plan around a more accurate prognosis. And these same “bad” genes may provide drug companies with targets.

A Web of Genes

C3Fall2013_Final-7With Doebele, Garcia and Hirsch, these are stories of what happens when one gene goes bad: it turns on or off or makes something it shouldn’t and the result is cancer.

But what happens when more than one gene is to blame?

Remember the confusing soup of the human genome? Well, it’s about to get even more confusing. That’s because genes interact in complex and sometimes unpredictable ways. Maybe a good gene acts bad in the presence of a third gene? Or a twist in a gene affects another far downstream? Or many genes combine to create a catastrophic effect?

As wonderful as it is to notice and treat the direct one-to-one effects of a gene that causes cancer, sometimes it’s not that easy. Sometimes a cancer can’t be defined by one gene that creates it and instead is a complex fingerprint of many genes, all acting together in a spider-web of interactions in which the shaking of one thread creates shaking somewhere else and eventually the whole web wobbles.

 Dan Theodorescu, MD, PhD, director of CU Cancer Center, developed a model that squints at these webs, taking into account its genetic ups and downs to create a concise picture of exactly how the web wobbles—and recommends drugs and treatment strate­gies that will stop its shiver. It’s called CO-eXpression EXtrapolatioN or COXEN. At its heart it’s a marriage of math and genetics.

With Doebele, Hirsch and Garcia, we’ve been looking at many samples of the same kind of cancer to see what they all share—what mutation creates them and so what mutation, turned off, could cure them? Theodorescu’s COXEN model does something a bit similar, but across many different cancer types. See, the National Cancer Institute maintains 60 cancer cell lines—60 dishes in which specific “flavors” of cancer cells grow—lung or bladder or breast, etc.

Theodorescu looked not at one slice of the NCI-60 but across all these cells. The NCI-60 have now been hit with so many experimental drugs that we have a darn good idea of how they react to just about anything that’s been made. We even know how cancer patients with these types of tumors fare with certain drugs and treatments.

But what makes cells and patients respond to drugs? Theodorescu hypothesized that it was their genetic expression—their fingerprint—and he asked an important question: Across these 60 different types of cancer, was there anything in the genes that could predict how these tumors respond to treatments? In other words, is there a genetic signature that can predict how all cancers—regardless of where they live or even their specific one-gene mutations—respond to certain kinds of drugs?

The answer is a resounding yes. Theodorescu showed that cancers across the NCI-60, no matter if they were from breast or ovary, respond in certain ways to certain drugs. And then he showed that he could extend these results even past the limits of these 60 cell lines—even in bladder cancer (Theodorescu’s specialty), which was not represented in the NCI-60, cells with genetic signatures he’d seen responded in expected ways to drugs.

A cancer’s genetic fingerprint predicts its response to treatment.

The COXEN principle has identified genetic markers that predict how a tumor will respond to therapy, and the use of these markers is now making its way up the cancer treatment food chain: Thomas Flaig, MD, medical director of CU Cancer Center’s Cancer Clinical Trials Office and assistant professor of medicine at the CU School of Medicine, has NCI approval for a national clinical trial of COXEN in bladder cancer patients.

“In bladder cancer, you commonly treat the cancer with chemotherapy for a couple months and then follow with surgery. But sometimes tumors don’t respond to drugs and you’ve wasted a couple months. We hope this trial will show whether COXEN can predict based on a tumor’s genetic signature which preoperative chemotherapy is best and also show which tumors won’t respond to drugs at all, so that we can go straight to surgery and save time,” Flaig says.

C3Fall2013_Final-8For obvious reasons, you can’t just release an algorithm into the wilds of clinical care and let it start recommending treatment options. If this COXEN trial shows that it knows ahead of time how treatments will turn out and what treatments might be better, Flaig and Theodorescu are optimistic that in the next clinical trial COXEN will be allowed to say not only, “I told you so,” but also, “Here’s a better way.”

“In essence we hope COXEN could guide treatment selection, matching each patient to the best drug based on the tumor’s genetic characteristics. This truly personalized approach could revolutionize cancer care,” Theodorescu says.

Effectively, what researchers at CU Cancer Center and elsewhere are asking is, “Are we grouping cancers the right way?” says Doebele. “Might these things that we describe as lung, breast or colorectal cancer be better described not by their site but by their genetic characteristics? I don’t want to discard decades of clinical work – but while viewing cancers at their site may help recommend the right surgery, chemotherapy and radiation, it does nothing to recommend new, personalized or precision treatments.”

How should we best understand cancer in order to treat it? By the top-down approach of its site or the bottom-up approach of the genetics that built it? By moving from the first to the second, CU Cancer Center is a leader in the way future cancer care will be delivered.

“The cure for cancer is in your genes,” says Theodorescu.

 

FacebookTwitterGoogle+Share

About the author: Garth Sundem

In addition to writing for the University of Colorado Cancer Center, Garth is the author of the books The Geeks' Guide to World Domination, Brain Candy, and Geek Logik. Contact him at garth.sundem [at] ucdenver.edu.

Comments are closed.