Joy Bachman was running out of options.
Ten years after a melanoma was removed from her neck, her cancer had returned. A seizure and an emergency trip to the hospital on her husband’s birthday in August 2014 had left her paralyzed on one side. Tests revealed three tumors in her brain, and cancer in her lungs, bones and adrenal glands.
After surgery to remove the tumors, doctors gave her three to six months to live. She entered a clinical trial testing a new drug combination that only made her sicker. By Christmas, she was back in the hospital, weighing a mere 89 pounds. The experimental treatment had made her septic and her abdomen swelled as if she were pregnant. Each day, hospital staff pumped 5 liters of fluid off her stomach.
Her oncologist, Dr. Brian Erickson, came into her hospital room and took her by the hand. With her family gathered around her, he told her she would likely die within five to 10 days.
But Erickson had one last thing to try. Bachman’s cancer cells had tested positive for a mutation in a gene known as BRAF that can cause cells to grow out of control.
“Your tumor is overwhelming you,” Erickson told her. “But we have a chance of starting this pill.”
The drug, dabrafenib (dah-brah-feh-nib), had only been approved by the Food and Drug Administration in the previous year. Erickson told her she’d know the drug was working if the swelling went down.
“They sent me home with hospice in my mother’s living room,” Bachman recalls. “We didn’t want the kids to see me dying.”
Four days later, her body began to drain the fluid.
“It started to work and I took of picture of me in a bikini, which I’ve never worn before in my life,” Bachman said. She sent it to Dr. Erickson with a note, “It’s working.”
Bachman’s story is not so much of a new miracle cure for cancer, but of a growing understanding of the differences between various types of cancers. Bachman survived because doctors now understand the differences between types of melanomas, and have found an effective treatment for Bachman’s subtype of melanoma.
Traditionally, cancers have been categorized according to the part of the body: breast, lung, colon, prostate, skin. Over time, researchers have learned that cancers from the same part of the body, cancers that look identical under the microscope, often have very different causes or ways they develop. That has led to the division of common cancers into smaller and smaller subgroups.
As a result, doctors can’t treat every breast cancer, every lung cancer, every melanoma the same way. The subclassification of cancers is increasingly forcing doctors to approach cancer as a rare disease, raising significant scientific and financial challenges to finding effective treatments.
To be sure, cancer is anything but rare. About 1 in 3 Americans will develop cancer in their lifetime, and cancer kills more people than any malady save heart disease.
A disease is considered rare if it affects anywhere from 2 to 15 per 100,000 people, depending on whose definition you use. Breast cancer, which affects some 155 per 100,000 women, can be divided into at least 11 subgroups. That means some subtypes affect so few patients, they would qualify as rare diseases.
“We are seeing a fragmentation of breast cancer, colorectal, lung — all the common cancers,” said Dr. John Bartlett, program director of diagnostic development for the Ontario Institute for Cancer Research in Toronto, Canada. “It’s challenging how we conceptualize all aspects of research. We’re having to think about how to do this with smaller and smaller groups of patients.”
The rare disease status among cancer subtypes is becoming so common the numbers of drugs approved under the Orphan Drug Act, which provides incentives to drug companies to develop treatments for rare diseases, has skyrocketed. The number of orphan drugs approved had only exceeded 20 in a single year twice from 1983 to 2010. Over the following seven years, however, it has never been lower than 26 in a year, including a whopping 77 in 2017. Cancer treatments have been the fastest growing part of the rare disease market, accounting for more orphan drug approvals from 2010 to 2014 than any other category of disease.
From 2013 to 2017, rare disease cancers accounted for 37 percent of all breakthrough therapies, a category of drugs where initial clinical evidence suggests a drug may offer substantial improvement over available therapies.
The subclassification of cancers is largely driven by a shift in the way cancers are diagnosed. Traditionally, pathologists looked at tumor cells under a microscope to determine if the tumor was malignant. Now, doctors are increasingly using molecular diagnostics, gene sequencing and other specialized tests to determine whether a tumor is cancerous and what type of cancer is in play.
Many predict that eventually all cancer patients will undergo genetic testing to determine what malfunctions have caused their particular form of cancer, allowing doctors to select a drug that targets that genetic mutation or the particular way it develops. The entire field of cancer is shifting away from treating large groups of patients in the same way and toward individualized treatment, an approach commonly referred to as precision medicine.
“That’s going to be incredibly challenging to accomplish,” said Dr. Gordon Mills, director of precision oncology at Oregon Health & Science University Knight Cancer Institute. “For a while, I think we’re going to be looking at the concept of precision oncology with smaller and smaller groups of patients that are similar to each other, and being able to develop approaches that we can demonstrate benefits to that particular group of patients with a very similar type of disease and similar types of tumors.”
That’s why the mainstays of cancer treatment remain surgery, chemotherapy and radiation. If a cancer is self-contained, surgeons will try to cut it out first. Chemotherapy offers a systemic approach, using a poison that targets fast growing cells in hopes of killing the cancer throughout the body. But chemotherapy plays the averages. Two out of three women with ovarian cancer, for example, will see a benefit from chemotherapy. But that means one out of three is exposed to a toxic treatment with no benefit. And generally doctors don’t know which one will benefit until they try it.
By subtyping cancers, researchers hope to find treatments that will work best for the specific pathway in play, killing the cancer while limiting the damage to healthy tissue and organs.
In 2010, when doctors started testing for a mutation known as EGFR in lung cancer, 15 to 25 percent of patients were able to switch from chemotherapy to a targeted therapy pill. Now, doctors can test for a half dozen more mutations, each with a targeted treatment. That has had a profound effect on patient care. Lung cancer has been one of the more deadly cancers. Without routine screenings, most lung cancers are found late, after they have spread and surgery is no longer an option. The emergence of screening programs for smokers and the development of targeted therapies has improved the outlook. The death rate for lung cancer dropped by 48 percent from 1990 to 2016 among men, and by 23 percent from 2002 to 2016 among women, with declines accelerating in recent years.
There are a number of cancers for which doctors had nothing to offer patients as recently as 10 years ago, that now have treatments that can almost be considered cures.
“The poster child for all of this is chronic myelogenous leukemia,” said Dr. Theodore Braich, an oncologist with the St. Charles Cancer Center in Bend.
CML results from a single genetic defect which produces a protein that drives the disease. Turn off that protein and you stop the disease.
“Not always, but the vast majority of cases,” Braich said. “And when you stop it, you stop it for a long, long time.”
Researchers at OHSU discovered the key to turning off that protein, bringing it to market under the brand name Gleevec in 2001.
“They thought it was going to work for six, 12 months, and then it was going to stop working,” Braich said. “That was the history of everything else.”
The five-year survival rate for CML jumped from 31 percent before Gleevec, to 59 percent after. And the scramble to find the next Gleevec was on. But that model proved hard to replicate.
Even the BRAF targeted treatments, which seem to melt away melanomas like Bachman’s, tend to work for only a year or two in most patients. Bachman has now switched to an immunotherapy drug, which ramps up the immune system to keep her cancer at bay. It’s unclear whether she’ll be able to stop taking the medication or if her immune system will need constant boosting to prevent a recurrence.
The Cancer Genome Atlas, a project launched in 2005 to catalog genetic mutations responsible for cancer, has documented some 30,000 mutations, an average of about 400 mutations per cancer. Now researchers have to figure out what all those mutations and their different combinations mean.
It’s sort of a cascade effect. Researchers must first find the differences, then understand whether those differences are important. If they are, they need to figure out if they can do something about them.
Bartlett estimates that the field has now identified about 95 percent of the changes that contribute to cancer. The ones we know about, know what they do and know how to treat, represent just the tip of the iceberg. Just below the water’s surface is the chunk of differences we’re beginning to understand but can’t affect yet. And then the bulk of the iceberg represents the differences that have been identified, but no one knows what they do, if they’re important or how to fix them.
The more mutations that researchers identify, the higher the number of possible subtypings that can emerge.
“It’s just like if you compared any two people. Are they the same height? Are they roughly the same weight? Do they have the same eye color? You’re going to split them into a certain number of groups,” Bartlett explains. “The more pieces of data you have, the more you subdivide the population.”
For decades, lung cancer was divided into four main groups. Now one of those has been further subdivided into at least 16 subtypes. Breast cancer is routinely divided into three groups. But now, one of those groups can be divided into five or six subgroups.
Cancer doctors now routinely test breast tumor biopsies for those subtypes, knowing that if a woman has certain markers, there are treatments that are more likely to work. How far down the road of testing they’ll go to determine what mutation is driving the cancer depends on the doctor and the woman’s situation.
Doctors can run a panel of tests that identify the most relevant mutations. Those tests were once prohibitively expensive for most patients, but the cost has come down and insurance companies are increasingly covering the panel tests.
“The reality is the panels don’t help as much as the companies tell you,” Braich said. “The likelihood that you’re going to find a driver mutation that truly is targetable is small.”
Those tests are much more likely to point patients to drugs on the market that have a limited benefit or to potential clinical trials.
“There the barrier is often that’s it’s 1,000 miles away,” Braich said.
Often, community oncologists can get the same medications being used in clinical trials from the manufacturer and allow patients with no good alternatives to try those drugs. That patient’s experience, however, isn’t captured as part of any study and therefore doesn’t help move the field forward.
There are plenty of examples of patients who have found an effective treatment through panel testing after exhausting all standard options, but those examples are more the exception than the rule.
In fact, oncologists are still debating whether routine panel testing of all patients make sense.
Is it better and more cost effective to treat 100 cancer patients with chemotherapy, or to do the genetic testing in hopes of finding the 5 to 10 percent that will have a known mutation with a matched treatment?
The likelihood of finding a treatable mutation is low. A recent study by researchers at OHSU found that only about 9 percent of patients with metastatic cancer will be eligible for a targeted treatment based on genetic testing, and just 5 percent will actually benefit.
And even for those who do benefit, the gains are often short-lived. For most of the common tumors — breast, lung or colon — a detailed genetic analysis usually finds a number of aberrant pathways. Targeting just one of those pathways might slow the tumor down temporarily, but rarely provides a lengthy remission. Tumors usually evolve a resistance mechanism to most of the targeted therapies.
“The targeted therapies, outside of CML, buy you typically a month to maybe a couple of years,” said Erickson, the oncologist with Summit Medical Group — Bend Memorial Clinic. “They usually won’t buy you more than a piece of time.”
That could allow some patients to go from targeted therapy to targeted therapy, each extending their lives for months to years in succession, perhaps buying them time until the next discovery.
Understanding the way these various subgroups of cancers arise give researchers a clue about how to stop them. But the subclassification of cancers is wreaking havoc on the traditional ways of doing research. With smaller subgroups of cancer, each group contains fewer potential test subjects.
“It’s making it nearly impossible to do what we used to do which is open up a single arm trial where you have a homogenous patient population and you’re testing just one intervention,” said Dr. Charles Blanke, an oncologist who chairs a global cancer clinical trials network based at OHSU. “You can’t find enough patients.”
Even when clinical trial networks collaborate to find patients with less common cancers to enroll in clinical trials, it can be challenging to find enough participants. Such networks are now working to improve screening methods, hoping to find a greater percentage of patients with a given mutation and to enroll more of them into a study.
“It’s about getting more patients into the top of the funnel, if you will, to try to shake out an increasing number at the bottom of the funnel,” Blanke said.
Having smaller subgroups of cancer also means that patients are less likely to qualify for any given study.
“If we have some subcategory of lung cancer that will affect 5 percent of patients, we are going to have 95 out of 100 patients we talk to about a trial who have no option,” Blanke said. “That’s totally unfair to them.”
To solve these challenges, cancer researchers are turning to basket trials, which enroll patients in a broad category of cancer and then assign them to specific arms of the trial based on their genetic markers.
Dr. Otis Brawley, the former chief medical officer for the American Cancer Society and now a professor at Johns Hopkins University, explains that if researchers want to test a drug for a particular marker in lung cancer, they need to find at least 40 to 50 patients with that marker, most of whom do very well on the drug. Researchers need to set up a system so that when patients present with lung cancer at a research site, they can be tested to see if their cancer carries that particular marker. They need to have doctors willing to enroll patients in that clinical trial and oversee their care during the study. Each institution running a clinical trial must get the approval of its trial design by an internal committee.
But even a larger academic medical center may only see 300 patients with lung cancer in a given year. If a particular marker affects only 2 to 3 percent of lung cancers, that’s only 6 to 9 patients per center that could enroll. And to enroll in the clinical trial, those patients need to have few other serious medical conditions that could skew results. That means doctors at different hospitals must collaborate to find enough patients to garner meaningful results.
“That’s a huge logistical burden,” Brawley said.
Large-scale clinical trials that enroll 600 people, randomizing half to the study medication, and half to traditional therapy, to find a difference in 15 to 20 people out of the 600 aren’t going to happen anymore, he said.
“The new clinical trial in cancer is going to be 30 to 40 people aimed a specific marker,” he said. “And we’re going to need clear evidence that the drug is beneficial.”
Such studies are designed to provide a quick way to determine whether certain drugs hold promise. But some experts fear the difficulty in conducting high quality clinical trials with smaller groups of patients might lead to a lowering of the standards of proof. Traditionally, drugs have been evaluated using randomized controlled trials, in which patients are randomly assigned to receive either the study drug or some established therapy. The two groups are then compared to see which did better. Particularly for rarer conditions, many trials have struggled to enroll enough participants to provide meaningful results.
Dr. Vinay Prasad, an OHSU hematologist-oncologist who has been a vocal advocate for evidence-based medicine, worries that researchers will start using the rare disease status of certain cancer subtypes to get out of having to do high quality studies.
“The subtyping has given them a new way to get what they want,” he said. “If you’re somebody who doesn’t want to do trials for your drug and you just want to get it on the market based on some surrogate endpoint in a group of 50 people, now you have something you can use as sort of a branding for that message.”
He points to the example of vorinostat, a drug manufactured by Merck for the treatment of t-cell lymphoma. The drug was approved in 2006, based on a phase II study. Phase II studies generally consider whether patients benefit from a treatment and what potential side effects might be. Phase III studies compare it against existing treatments to determine whether it’s an improvement.
“To my knowledge, vorinostat has never shown a survival benefit in any single randomized trial in all those years,” Prasad said. “Why? Because it came to the market for a rare disease.”
Randomized controlled trials can be done, he said, even with the rarest diseases. Adrenal cortical cancer, known as ACC, affects only 0.7 per 100,000 people. Yet, researchers have already published one randomized controlled trial and a second one is due to be published this year.
“What happened was many doctors across the world who see ACC said, ‘let’s work together and let’s do this so we can settle the question,” Prasad said.
When the FDA recently looked at drug approvals for rare diseases, the agency found it didn’t matter how rare the conditions were. Whether they affect 1 of out 100,000 or 6 out of 100,000, about a third of drugs that come to market are based on randomized controlled trials.
“I know that many people say you quickly get to the point where we run out of patients,” Prasad said. “I think a lot of those types of concerns are overstated.”
Prasad said cancer researchers need a more organized system for getting patients into clinical trials. Currently only about 3 percent of cancer patients participate in research studies.
“If we could get that to 20 percent, you’d answer these questions very quickly,” he said.
That could be accomplished with better use of technology to connect patients to studies. Electronic health records, he said, could be programmed to tell doctors about open clinical trials anytime they enter a cancer diagnosis.
There may be a desire to rush treatments to patients with cancers for which no good options exist today, but that can also put those patients at greater risk. Not only would researchers not know whether those drugs really work, there is a potential to make things worse.
Bartlett said some treatments aimed at shutting down a certain pathway wind up accelerating another pathway, leading to faster cancer progression.
“It’s not that we’re not going as fast as we can,” he said. “It’s that we are ensuring we’re doing what we’re doing with a high level of clinical integrity, so we don’t end up inadvertently harming people.”
The advantage to smaller study groups is that researchers can be much more nimble. The OHSU Knight Cancer Center in Portland has a study underway mostly with breast cancer patients, testing a new drug combination. The study is designed to carefully track how the medications are impacting tumor growth, and the study design allows patients to be transitioned to another experimental combination of drugs if the first one stops working.
The research team tests the cancer for as many biomarkers as they can, gets experts together to discuss what drugs or combinations of drugs are mostly like to work, and then tries that for six to 12 weeks. They monitor the patients closely and if the tumor responds, they can take another biopsy to see what changes occurred within the tumor. If the tumor isn’t responding, they can shift the patient to another treatment.
“I hate to say it, but it’s using the patient as their own experiment and changing therapies based on all the molecular biomarkers that we’re looking at,” said Dr. Chris Corless, executive director of the OHSU Knight Diagnostic Laboratories.
Beth Shia, 66, of Lake Oswego had a double mastectomy with reconstruction after being diagnosed with breast cancer in 2011. When tests showed it was a rare, hard-to-treat form of breast cancer, she followed up with chemotherapy. In 2013, she found another lump in her reconstructed breast, she underwent surgery and chemotherapy once again. Last year, she found another lump near her belly button. The cancer had recurred and had spread. There was cancer in her brain, in her lymph nodes and in her lungs. She didn’t need doctors to tell her the prognosis.
“I knew what metastatic breast cancer meant,” she said.
She underwent a targeted radiation treatment of the cancer in her brain, and her doctor recommended she enter the OHSU study, becoming the study’s third participant. The researchers analyzed the genetic data from her tumor samples and recommended two medications: one a targeted therapy that has shown benefits in treating her particular form of breast cancer and another that marshals the immune system to fight her cancer.
She started the treatments in July and by September, imaging tests showed no sign of cancer in her brain, and 70 percent to 80 percent reductions in the tumors in her lungs and lymph nodes.
Shia was ecstatic to hear the news.
“I had to tell my family so many bad things over the summer to be able to tell them some good news, they cried with happiness,” she said. “It’s going to go down as one of the highlights of my life.”
In January, the next round of testing provided even better news. A brain MRI and the multiple CT scans showed no evidence of any active disease.
“Truthfully, I don’t believe this will make it go away. This will just allow me to live many years with this, continuing some treatment,” Shia said. “I’m hopeful if this particular protocol or combination of drugs doesn’t work, then there will be something else I could try.”
The study is extremely resource-intensive with researchers tracking how cancer cells evolve in response to treatment, and using that information to tailor a drug combination for the patient. In the first 16 months of the trial, doctors enrolled 38 patients, getting biopsies from 28, finding targetable mutations in 22 and treating five.
“It took us years to set it up, and it’s expensive,” Corless said. “But ultimately it’s like anything else in society: If you scale it up, sort of industrialize it so to speak, you can reduce the cost and make it available to everybody.”
Cancer drugs are already among the most expensive therapies available today, and the increased number of cancer subtypes could drive costs even higher.
Treatment with Gleevec costs $120,000 per year, and although most public and private health plans cover the treatment, it adds significantly to health care spending in the U.S.
As more targeted treatments are developed for smaller subgroups of cancers, many wonder whether the health system will be able to foot the bill.
Some experts argue that by better targeting therapies to genetic markers, the health system is going to get more bang for the buck.
Spending $5,000 to sequence a cancer’s genetic signature could help patients avoid a $100,000 treatment that doesn’t help them.
But more subtypes also creates greater competition for research dollars, a phenomenon Brawley refers to as “disease olympics.”
When he was the medical director of the American Cancer Society, he frequently got petitions from patient advocacy groups pressing for more research funding for their cancer subtype.
“They proudly explain that they are 300 patients and family members and they want more research in ALK-positive lung cancer,” Brawley said.
“Screw RAS-positive lung cancer, screw BRAF lung cancer, screw HER2/Neu lung cancer. We want more ALK positive research, and the hell with breast cancer.”
The process is becoming increasingly competitive and political, slicing up the pie of research dollars into smaller and smaller pieces. The National Cancer Institute, Brawley said, currently funds only 10 to 12 percent of the grants submitted to it that it says merit funding.
“Those great scientific ideas aren’t being funded,” he said.
Brawley believes that precision medicine is an incredibly important development within the cancer field, but it does not signal the end of the disease.
“It is some important treatments that can help people live longer and will cure some people,” he said. “And I just used that four-letter word, but it’s not going to be the answer to all cancer.”
Still the trend is adding treatments at a significant pace, and that has doctors like Erickson excited about what they might be able to offer patients.
“Twenty years ago, if you came in with (chronic myelogenous leukemia) I would look at you and make kind of a sad face and say, ‘We’re going to do our best,’ but we don’t really have much to offer and our life expectancy was three years,” he said.
Now he has patients in their second decade of life free of cancer. And the pace of discovery seems to be accelerating.
“The first 10 years of my career it was just a couple of new drugs a year. Now we’re getting a handful of new drugs each year,” Erickson said. “It makes it so much easier to go to work.” •