Important Points Made by the Individual Speakers
- Molecular characterization of drugs has far preceded the molecular characterization of cancer.
- As cancer drugs target smaller subsets of people, new drug development paradigms are needed that require smaller trials and demonstrate substantial effects.
- Regulatory science needs to advance to be able to approve effective drugs with limited amounts of information.
- Despite the value that genetic testing can produce, a relatively small number of patients with cancer receive such tests.
In the third scenario discussed at the workshop, the patient has developed cancer 5 years following her deep vein thrombosis:
The individual is seen at age 50 with cough, dyspnea, and chest discomfort. Evaluation reveals a lung mass; bronchoscopy and biopsy reveal a non-small-cell lung cancer. Her tumor is found to have variations that allow the use of targeted therapy, and with treatment the patient goes into remission.
The clinician who discussed the case, Kenneth Offit, chief of clinical genetics service in the Department of Human Genetics, Memorial Sloan-Kettering Cancer Center, changed the case slightly to make it more realistic. He posited that the woman had advanced non-small-cell lung cancer, because if it were a local lung cancer she would simply be referred for surgery. He added that the patient experienced a relapse after going into remission and that the case was followed by further testing and treatment.
Standard chemotherapy for advanced non-small-cell lung cancer, as that treatment existed about a decade ago, produced only a median survival increase of between 7 and 8 months, Offit observed (Kelly et al., 2001; Schiller et al., 2002). Even when tyrosine kinase inhibitors began to be used in 2003, prior to FDA approval of epidermal growth factor receptor (EGFR) mutation testing, they produced only modest survival advantages (Kris et al., 2003). Once testing for EGFR gene mutations began in 2004, response rates reached the remarkable rate of 75 to 82 percent, said Offit, with progression-free survivals of 8 months to a year or more (Pao et al., 2004).
Nevertheless, most cases are not success stories. And resistance to tyrosine kinase inhibitors commonly develops (Balak et al., 2006; Bean et al., 2008), which requires treatment with another targeted drug. Randomized controlled trials for EGFR testing have been carried out and point to a progression-free survival on the order of 3 to 4 months (Mok et al., 2009). “It is a striking advance, but we want to keep all of this in perspective,” Offit said.
Current practice at Sloan-Kettering is to test lung cancers with targeted testing. Driver mutations for between half and two-thirds of lung cancers have been identified and can be tested, and these mutations point to targeted therapies that have already been approved by the FDA or are in the pipeline for approval. Still, Offit noted, access to testing can be a problem, depending on geographical residence.
Offit briefly described several studies that have found costs per QALY of around $100,000 for targeted cancer interventions (Atherly and Camidge, 2012; Bradbury et al., 2010). These costs per QALY are higher than for BRCA1 testing (Plevritis et al., 2006) and testing for Lynch syndrome (Ladabaum et al., 2011).
In discussing the second model of genomic information delivery in which targeted data plus other actionable information is provided, Offit described the Integrated Mutation Profiling of Actionable Cancer Targets screening model, which captures information on 230 cancer genes, though Sloan-Kettering is currently running this as a research arm only. Two major issues with this type of screening are variants of unknown significance and
the required consents. These variants can impact pharmacogenomics as well as noncancer disease risk in future generations.
Offit noted that they spend 2.5 hours to do the counseling for next generation sequencing panels and wondered what could realistically be done when potentially receiving information from 20,000 genes. And he noted that companies who provide these services use consent forms that are impractical for patients. “Your patient is supposed to say, ‘I am in the mood today to look at my mitochondrial genes, but let’s leave out the PI3 kinase pathway, okay?’ And they will write that down on the commercial consent form. Rather unrealistic.”
As an alternative, he pointed to a consent form developed at the University of Michigan that asks whether patients want to receive results “that do not have a direct impact on [the] care of my current cancer.” Patients can choose to receive results “that may have significance for biological family members” or results “that are not related to your cancer but may have potential medical impact for you.”
Finally, Offit briefly mentioned the delivery of complete information from whole genome sequencing, including the incidentalome. In cancer therapy, whole genome analysis may reveal germline changes that impact therapeutic activity and that have implications for familial risk as well. Full disclosure of this information without medical practice standards, he said, will likely lead to clinical and economic inefficiencies—if not to chaos. “We have a lot of cheap tests in medicine. But just because it is cheap doesn’t mean that you take it outside of the medical model,” he concluded.
The best way to predict the future is to understand the past and where we are now, said Stephen Eck, vice president and global head of medical oncology for Astellas Pharma Global Development. In the past, the pharmaceutical industry was not very interested in oncology. Drug discovery was highly empiric, drawing extensively from natural products and alkylating agents. Companies could sell only a few doses to any one person because of cumulative toxicity and low incidence. “There were better things to do with your capital if you were a drug developer,” Eck said.
This situation began to change as companies recognized that a drug used for a few adults in one cancer market could also be used in other cancer markets, said Eck. Drugs became more tolerable and could be used longer, which increased sales. Premium pricing made small single indications attractive. Scientific advances made rational drug discovery faster and cheaper. For all these reasons, treatments for small subsets of cancer became more economically attractive.
Today, a variety of drugs are used to treat non-small-cell lung cancer.
Most of these drugs are used without any genetic testing, and the market is highly segmented, with people making decisions using data of varying quality. The best situation would be for the market to segment along scientific lines so that patients received what they needed, not along the lines of what people thought on average would be a good idea, Eck said.
Molecular characterization of drugs has far preceded the molecular characterization of cancer, Eck observed. Today, drugs are highly purified and have well-known structures tied to their pharmacology. “We can understand the drug itself at an exquisitely detailed level,” said Eck. Diseases such as lung cancer are much less well understood, but molecular diagnostics are starting to reveal their secrets. In the future, said Eck, “somebody will be diagnosed with lung cancer. They will get a biopsy. It will be molecularly characterized. And the therapy will be chosen based on the unique attributes of that patient’s tumor.” These therapies are not curative, and as tumors mutate and metastasize, new therapies will be needed. But cancers can be periodically reassessed, leading to selection of the next targeted therapy.
For this vision to become economically feasible, several things are needed, said Eck. Faster and cheaper technologies for molecular characterization need to be developed. New drug development paradigms are needed that require smaller trials and demonstrate substantial effects. Smaller trials will require improvements in pharmacovigilance, because clinical and biological information will need to be collected once a drug is put out into the marketplace to substantiate trial results. Regulatory science needs advances to be able to approve effective drugs with limited amounts of information. In return, Eck said, drugs need to be used effectively in oncology, especially with so many being used off label for nonapproved indications with little to no evidence to support that use. To pay for schools, food, and other basic services, we must end the steady increases in the cost of health care, said Eck, and drug development could help control costs if it could bring drugs to market for less money.
In 2004, at the age of 44, Richard Heimler was returning home from a business trip when he began experiencing chest discomfort and shortness of breath. A chest X-ray taken in the emergency room revealed a 3-millimeter spot on his right lung.
A positron emission tomography scan and biopsy confirmed that he had a malignant non-small-cell lung cancer, even though he was not a smoker and had no family history of lung cancer. “At that point, after getting that diagnosis, I was numb and naïve,” Heimler said. “I was glad at the time I didn’t know that 60 percent of lung cancer patients die in the first
year and that 90 percent die in the first 5 years. I also did not know that 200,000 people are diagnosed with lung cancer every year, and 180,000 people die from lung cancer every year” (ACS, 2012).
Heimler was fortunate that his mother was a genetic counselor. Serving as his advocate, she found him the best doctors, hospitals, and treatments in New York. “I recommend that everyone has an advocate, or someone to do the research, absorb the information, or at times fight for you. Newly diagnosed patients and their caregivers should know their patient advocacy organizations. They are a great resource and support,” Heimler said.
He initially had a pneumonectomy—the removal of his right lung—followed by chemotherapy. He then had a recurrence of the cancer, this time a brain tumor, which was removed surgically. A later brain tumor was killed by gamma knife radiation. He then had a tumor below his rib cage, which was removed surgically, followed by chemotherapy. “I know this sounds like a horror story, but I am still standing,” he said. Then, 3 years ago, spots appeared on his left lung. “That was our worst fear, because that’s the only lung I have. So at that point, if it’s a baseball game, I was in the bottom of the seventh inning, top of the eighth.”
At that point, Heimler’s oncologist tested his tissue to identify any tumor variations associated with his cancer. He tested positive for the ALK gene, which is only present in 3 to 5 percent of patients with lung cancer (Kris et al., 2003; Riess and Wakelee, 2012), and his oncologist told him about a new clinical trial with Pfizer for patients with this tumor variation. Drug companies and diagnostic testing companies need “to educate our doctors about all the new targeted treatments,” Heimler said. “Doctors must recommend to all their cancer patients to have diagnostic testing for all available targeted treatments. It is also important to implore our doctors to archive enough of our tissue to get an accurate sample for diagnosis.”
Because the tumors on his lung were small, the risks were minimal for Heimler to enter the stage 2 clinical trial for 6 months. He began taking the drug crizotinib (Xalkori)—three 250-milligram pills in the morning and another three pills in the evening—and has experienced minimal side effects.
A few months after beginning the trial, his tumors began to shrink and then cavitate, and no new activity occurred. The week of the workshop was his 2-year anniversary on crizotinib. He reported, “My doctors cannot see any tumors, and I feel great and have a good quality of life. So I am very, very thankful.”
Heimler has found the clinical trial to be economically advantageous. His treatments are free for the rest of his life as long as he stays in the trial, as are his computed tomography scans, the magnetic resonance imaging scans of his brain, and other tests. If he leaves the trial, however, his drug costs $6,000 every 3 weeks, of which Medicare covers only about $4,000.
He does not qualify for any financial assistance for his medication, but his parents are helping to ensure that economics is not a factor in his care. He has two private disability plans, plus Social Security disability. “I will never be rich, and I will never be able to save any money, but I have sufficient income to live comfortably and raise my two children.”
Lung cancer may not be curable, but it is treatable and livable, Heimler said. “Lung cancer is a relentless, unforgiving, and nondiscriminating disease. But thanks to advances in personalized medicine and diagnostics, it does not have to be a death sentence.” On the basis of his experiences, he has become an advocate devoted to raising awareness for lung cancer. Lung cancer is the number one cause of cancer deaths in the world. It kills more people than any disease but heart disease, and it kills more people than breast, colon, kidney, skin, and prostate cancer combined (ACS, 2012). People with lung cancer are often subject to a “blame the victim” mentality, because other people think they brought the disease on themselves by smoking. This perception has an effect on research dollars, Heimler observed. According to the Lung Cancer Alliance, federal research funding per cancer death is $26,398 for breast cancers, $13,419 for prostate cancer, $6,849 for colon cancer, and just $1,442 for lung cancer.1
“I hope that in the years ahead we can change the perception of lung cancer in the general public, the medical profession, the government, and the media in order to generate a fair and proportionate share of research funding for this deadly disease. We need to raise lung cancer to a national public health priority,” said Heimler.
“I am one face of lung cancer,” he continued, “who is benefiting from diagnostic companies and drug companies working together to identify and produce targeted therapies to make a difference for lung cancer. And even though my targeted therapy might only be effective for 3 to 5 percent of lung cancer patients, there are still 11,000 people each year who can benefit from this drug alone. It is time to take advantage of advances in personalized medicine by getting the right medicines to the right people. This new world of science is giving me hope that maybe my children and I will have more time to create new memories in the years ahead.”
The panel of economists began by discussing an observation made by Heimler and several other workshop participants—that genetic testing still occurs for a relatively small number of cancer patients. As Ramsey observed,
1 According to estimates by the NIH, fiscal year 2011 funding for lung cancer was $221 million, breast cancer was $715 million, colorectal cancer was $313 million, and prostate cancer was $284 million (http://report.nih.gov/categorical_spending.aspx [accessed March 19, 2013]).
genomic testing can provide enhanced value if patients are directed to drugs where they have a better response rate and fewer side effects. But the delivery and reimbursement systems are structured in a way that acts to restrict rather than enhance testing, which may be a hindrance to rapidly evolving personalized medicine.
Billings pointed out that costs are involved with moving samples among sites and conducting different tests even as a disease continues to progress. One clear benefit of a rapid and comprehensive mutational test is that it would produce comprehensive knowledge quickly.
Veenstra commented that somatic genomic testing is quite different from the other scenarios examined at the workshop. In the case of cancer, the testing clearly has the potential to provide benefits to patients by avoiding the costs and adverse effects of ineffective treatments. Also, at least for some forms of testing, the cost of the test is much less than the cost of the therapy. “From a value perspective, why wouldn’t you do [genetic testing]? What is hard to get my head around is that only a minority of patients are being offered this testing.”
Heimler added that the first thing he asks people who have been diagnosed with lung cancer is whether they have been tested for the ALK gene. Mostly they say no because they are smokers and their doctors have told them that they probably do not have the gene. “But I think it is absurd for a doctor not to test every single patient, especially with a disease like lung cancer where this might be our only hope of extending our lives.” Veenstra noted that clinicians are not used to treating patients in which just a small percentage of people respond to a specific medication. “It is a different paradigm,” he said.
Eck responded that medicine is an inherently conservative enterprise that is slow to change. For example, the data in favor of lumpectomy versus radical mastectomy far outpaced its uptake in the community, but eventually it was adopted. One participant added that this type of slow adoption is observed with every type of new technology, not just genomics.
A participant observed that the increasingly complex landscape of cancer genomes poses an immense informatics challenge to be able to aggregate, analyze, interpret, store, and share these data. Many cancer centers have strong data systems, but they tend to be siloed, sometimes even within centers. Data from these systems need to be exchanged across institutions, across cities, and across the globe. In addition, the challenges in sharing data are cultural, financial, and technical, and all three of these areas must be addressed.
Timothy Ley, the Lewis T. and Rosalind B. Apple Chair in Oncol-
ogy at Washington University, suggested that the research community is committed to sharing data on cancer genomes and somatic mutations. He mentioned that The Cancer Genome Atlas is helping to share data widely. Another participant pointed out that molecular diagnostics companies and the FDA both have significant roles to play as well. Evidence on the use of genomic tests is hard to generate without data being shared, and government agencies can help facilitate this sharing, the individual said.
Eck pointed to some of the difficulties in sharing data, including intellectual property provisions, restrictions in informed consent forms, and societal norms. Nonetheless, industry is becoming more transparent as it recognizes the advantages of increased access to information. And some companies have begun putting information immediately into the public domain because patenting can be costly and difficult, and releasing information prevents other companies from patenting it. “It is hard to protect data. It is expensive. It is sometimes just easier to give it away,” Eck said.
One participant proposed an economic study that would compare targeted treatment based on EGFR testing with standard approaches to therapy as a way of demonstrating the value of testing. He also observed that cancer is undergoing a paradigm shift away from a histological definition of cancer toward a molecular characterization, which makes it difficult to compare new approaches with the current standard of care.
Offit also observed that society cannot afford to pay for very expensive cancer drugs that produce, on average, only a few extra months of life. Even for people whose expenses are covered because they are in clinical trials, those costs have to be covered somehow. Grosse agreed, noting that the common threshold of paying $100,000 per QALY is not affordable in the long run. “It works for rare diseases or for short-term treatment, but it can’t go to scale.”
Ramsey said that targeted therapy is probably not going to lower the cost of cancer care. Producing better results does not mean low drug costs. In fact, the drugs that elicit better responses tend to be more expensive. For drugs such as bevacizumab and crizotinib, with the current price structure and gains in survival, “it is not particularly good value.”
In response to a question about why new drugs tend to be priced at about $100,000 per life year saved, Ramsey said that drug pricing is extraordinarily complex, “and I don’t think there is always science involved.” A drug may be developed for one cancer and produce a given survival gain, but when it is used for another cancer and produces a much smaller survival gain, the price is the same, even though the value of the drug differs from cancer to cancer.
Katrina Armstrong, professor of medicine at the University of Pennsylvania School of Medicine, noted that the way billing and reimbursement are done in hospitals can discourage testing. These financial systems should be
set up so that tests are done at the time of maximum value. A participant noted that CMS may accept a proposal to move toward a billing system that relies on a test-specific code, which could ease this problem. The interpretation of a test result would be part of the physician’s fee schedule, however, which may further influence incentives.