In the past decade, much new knowledge about the molecular underpinnings of cancer has accumulated, and the array of molecular aberrations in each individual tumor can be assessed through genomic sequencing and other tests. The rationale for and feasibility of developing molecularly targeted therapies have never been stronger, and there are hundreds of candidate drugs in the development pipeline,” said Adrian Senderowicz, MD, Chief Medical Officer, IGNYTA, Inc, and Planning Committee Chair of the National Cancer Policy Forum’s workshop, “Policy Issues in the Development and Adoption of Molecularly Targeted Therapies for Cancer,” held recently in Washington, DC.
That said, many challenges remain in the development and appropriate implementation of these new therapies, Dr. Senderowicz added.
Too Much of a Good Thing?
Bruce Johnson, MD, Chief Clinical Research Officer, Dana-Farber Cancer Institute, Boston, traced the beginnings of these issues to April 2004, when three three different laboratories from the Dana-Farber Cancer Institute, Massachusetts General Hospital, and Memorial Sloan Kettering Cancer Center discovered a link between dramatic responses to treatment with the epidermal growth factor receptor (EGFR) inhibitors (gefitinib and erlotinib) and mutations of EGFR in lung cancer patients. Investigators at the Boston hospital believed that it was important to routinely test lung cancers for the mutations of EGFR so they could be initially treated with gefitinib or erlotinib rather than chemotherapy. Therefore, the hospitals sent tumor slides to the Laboratory of Molecular Medicine at Harvard for EGFR mutation testing in a CLIA environment.
One year later (2005), Genzyme Corporation announced commercial availability of an EGFR mutation test to identify patients likely to respond to targeted therapies to treat non–small cell lung cancer. Without the testing in unselected patients, only about 10% of them responded to gefitinib or erlotinib. The patients with sensitizing mutations of EGFR responded more than 50% of the time.
Between then and now, hundreds of compounds have been or are being tested. They belong to two broad groups: therapeutic monoclonal antibodies, which target specific antigens found on the cell surface, and small molecules that penetrate the cell membrane to interact with targets inside a cell.
According to My Cancer Genome (www.mycancergenome.org), a personalized cancer medicine resource managed by the Vanderbilt-Ingram Cancer Center, Nashville, there are now 74 targets for kinase inhibitors, with 324 compounds in development; 65 targets for therapeutic antibodies, with 106 compounds in development; 12 targets for immunotherapies, with 39 compounds in development; and 34 “other” targets, with 79 compounds in development. In addition, 54 agents have been approved by the U.S. Food and Drug Administration (FDA) to date.
Mycancergenome.org provides current information on the mutations that cause cancer and their related therapeutic implications, including available clinical trials. It is chock full of data—so much, in fact, that clinicians can be overwhelmed, not knowing what to do with them, and some just stay away, said Mia Levy, MD, PhD, Ingram Assistant Professor of Cancer Research and Director of Cancer Clinical Informatics, Vanderbilt University.
Testing the Molecules
Before designing therapies, the molecules themselves must be tested. Dr. Levy described the four main types of gene alterations:
- Single-nucleotide variants, also known as point mutations, result from a base substitution at one nucleotide, producing a change in the amino acid sequence or premature truncation of an encoded protein.
- Small duplications of consecutive nucleotides, insertions, or deletions involving one or more nucleotides, or more complex mutations involving simultaneous deletions and insertions of one or more bases. They may result in addition or subtraction of amino acids in a protein or cause its premature truncation.
- Exon or gene copy number changes, including large duplications or deletions of entire exons affecting protein function or changes in the entire gene.
- Structural variants or large structural anomalies of genetic material, including translocations or inversions that result from breakpoints between multiple chromosomes or within a single chromosome, often resulting in fusion genes and associated fusion proteins.
Mutations can cluster in “hotspots” where tumors from different patients harbor the same recurrent mutation. Some hotspots are frequent, others rare. For example, the BRAF V600E mutation occurs in 40% of all melanomas, whereas BRAF L597S occurs in less than 1% of all melanomas.
“Tumor-specific genetic alterations can also be detected in the bloodstream. This presents a great opportunity for cancer diagnostics and for detecting the mutational makeup of tumors without doing a tissue biopsy. Such a ‘liquid biopsy’ is, however, like finding a needle in a haystack,” said Matthias Holdhoff, MD, Assistant Professor of Oncology, Johns Hopkins University School of Medicine, Baltimore.
In the blood of cancer patients, Dr. Holdhoff explained, the fraction of DNA fragments that harbor tumor-specific mutations is very small compared to the abundant amounts of normal DNA fragments from healthy cells. This problem is tackled by using highly sensitive polymerase chain reaction–based assays that essentially count DNA fragments in blood one by one. For example, these techniques can detect even 5 fragments of tumor-derived DNA in a pool of 10,000 normal DNA fragments from healthy cells.
The following tests are in current use:
- Allele-specific polymerase chain reaction detects a specific single-nucleotide variant. It can find mutant DNA if present in 1% to 5% of tumors tested, but it cannot detect other mutations that may be present.
- Sanger dideoxynucleotide sequencing can find unknown mutations, including single-nucleotide variants, small duplications, insertions, deletions, and indels. It can detect gene fusions if RNA from the fusion transcript is first extracted from the specimen. It requires mutant DNA to be present at a frequency of 20% to 25%.
- Pyrosequencing can detect unknown mutations in a small targeted region. It can be done quickly and can detect mutant DNA at 5%, but the types are limited.
- Mass spectrometry and single-base extension assay both detect targeted single-nucleotide variants in mutant DNA at 5% to 10% and in more than one gene.
- Multiplex ligation-dependent probe amplification finds exon and gene copy number, and depending on the experimental design, can also detect single-nucleotide variants. It requires mutant DNA to be present at 20% to 40% and works best on fresh frozen tissue.
- Fluorescence in situ hybridization detects gene copy number changes and targeted structural variants but not in solid tumors.
- Next-generation sequencing: custom panels, hybridization capture, and whole-exome sequencing detect substitutions, duplications, insertions, deletions, indels, exon and gene copy number changes, and select translocations—for a total of 1,144 exomes sequenced.
- Next-generation sequencing: whole-genome sequencing detects substitutions, duplications, insertions, deletions, indels, gene copy number changes, and chromosome inversions and translocations. It is the most comprehensive of all, but it requires more tumor tissue, sophisticated bioinformatics, and large computational demands for data storage.
Next-generation sequencing is one of the most powerful assay methods to date, said P. Mickey Williams, MD, Director, Molecular Characterization Laboratory, Frederick National Laboratories for Cancer Research. “It is popular in all aspects of cancer research and clinical management,” he noted. For example, the National Cancer Institute’s MATCH study is a multiarm basket study in which each arm includes multiple tissues. The goal is first to identify mutations, amplifications, and gene fusions to determine eligibility and then to assign patients to a relevant agent regimen using a rules-based approach. “This type of study requires screening large numbers of tumors, and it needs many targeted treatments,” he explained.
Companion Diagnostics
Targeted therapies are a good thing—if they work. But they don’t always, and for this reason there are companion diagnostics. A companion diagnostic is a diagnostic test whose results are essential for the safe and effective use of a particular drug or biologic agent. These devices can be used to identify patients who are most likely to benefit from a therapeutic product, identify patients who are likely to be at increased risk of serious side effects, and monitor response to treatment.
If a diagnostic test is inaccurate, said E. David Litwack, PhD, a member of the FDA’s Division of in Vitro Diagnostics and Radiological Health, then the treatment decision based on it may not be correct.
Ideally both the companion diagnostic and drug should be developed together, preferably before the drug enters clinical trials. In fact, the FDA may require a companion diagnostic if a new drug is designed to treat patients with a specific molecular target. One of the Agency’s recommendations, said Dr. Litwack, is for pharmaceutical and diagnostic companies to partner early in the process.
Dr. Senderowicz added a note about benefit and risk. “A false-positive test would cause a patient to receive unneeded treatment, along with its potential risks, without benefit. A false-negative means a patient would not receive needed treatment.”
Use of companion diagnostics began in 1998 with a companion that detects excessive levels or extra copies of HER2 in breast tumors. Since then, manufacturers have increasingly accepted the fact that diagnostic tests can greatly increase clinical success. For instance, there are now two companions for cetuximab (Erbitux), one for deferasirox (Exjade), one for afatinib (Gilotrif), one for imatinib (Gleevec), and seven for trastuzumab (Herceptin).
Using and Paying for All This
High technology is always exhilarating, but is molecular targeting of practical clinical use, and if so, who’s going to pay for it?
David Eberhard, MD, PhD, Director, Pre-Clinical Genomic Pathology, Lineberger Comprehensive Cancer Center, University of North Carolina, Chapel Hill, posed four critical questions that are frequently addressed with laboratory assays and diagnostic tests: (1) Diagnostic: Does the patient have a condition, and if so, what is it and what caused it? (2) Prognostic: How is the patient going to do? (3) Predictive: How will the patient respond to an intervention? (4) Pharmacodynamic: Is the intervention having an effect?
The usefulness of laboratory results for answering these questions requires analytical validation—that is, how well an assay will measure a molecular event: its range, accuracy, precision, bias, and reproducibility. It also requires clinical validation—that is, its strength of association with the condition of interest: What can it be used for, and does it have actual value in health care? And the key question, said Dr. Eberhard: “Does it offer more than what we have now?”
As to who will pay for the technology, that remains to be seen, said
Donna Messner, Vice President and Senior Research Director, Center for Medical Technology Policy, a Baltimore company that looks at quality and relevance of clinical research.
“Molecular diagnostic tests have the potential to transform oncology practice, but integration of biomarkers into practice has been inefficient so far because evidence of clinical utility is inadequate, studies of clinical validity have been incomplete or flawed, there is no shared evidentiary framework, and clear and predictable methodologic standards are lacking.”
She added that there has been explosive growth in the number, complexity, and costs of tests in the past 20 years. For example, genetic testing is now available for more than 2,000 conditions, and more than 125,000 variants have been discovered in thousands of genes. “But the consequences of wrong information and/or decisions can be disastrous.”
What’s more, payers are demanding ever-closer scrutiny as tests proliferate. Coverage policies vary widely among payers because they serve different patient populations, they have differing financial and organizational models, they may not have reviewed the same evidence for each test, and corporate cultures differ.
For instance, some companies think that a test is medically necessary if it has a direct effect on clinical care. Others believe that the disease in question must be preventable or treatable, and still others require that the test results in a change in the intensity of surveillance and/or treatment. Some companies want to see improvement in outcomes when a test is used, and almost all will ask whether another, extant clinical tool is available for the same purpose.
Other reasons that insurers balk at covering genomic tests include too few published studies demonstrating clinical utility, the often-minor role of genetics in complex diseases, and the availability of alternative effective screening methods.
Even without the coverage issues, there are still clinical problems in precision medicine, said Lillian Siu, MD, Professor, University of Toronto, Princess Margaret Cancer Centre Drug Development Program. “It is difficult to keep up with a fast-growing body of knowledge, ensure that patients understand test results, find treatments (either approved drugs or appropriate clinical trials) to match the mutations found, and realize value generated through these agents.” ■
Disclosure: Dr. Levy is a consultant and a scientific advisor for Personalis and a consultant for GenomOncology. Drs. Litwack, Eberhard, Holdhoff, and Siu reported no potential conflicts of interest.
