Top 10 Reasons Against Routine Genomic Profiling of Tumors in Patient Care
When they did comprehensive screening for gene mutations at MD Anderson, in a huge number of patients, they found an actionable target in 31%, but, of these 31%, only 10% responded to the targeted therapy. Overall, only 2.4% of all the patients receiving genomic profiling had a response. This is absolutely horrible. I think that genomic profiling is in general a scam, yet virtually everyone is doing it. I think that certain types of targeted genomic profiling is probably worthwhile -- e.g. EGFRtermtermterm mutations in lung cancer, but phenotype analysis can test for the same drugs using cell culture as a platform.
San Francisco, CA—Neal J. Meropol, MD, of the University Hospitals Seidman Cancer Center, and Case Western Reserve University, Cleveland, OH, has long advocated against unnecessary treatment and testing of patients with cancer. At the 2015 Gastrointestinal Cancers Symposium, Dr Meropol outlined his reasons why clinicians should not bend to pressure to routinely test all tumors. “Here are my top 10 reasons why I believe we are not ready for routine molecular profiling of tumors,” he told the audience, listing the following reasons.
1. Assay Platform Limitations
Variability in the sensitivity and specificity of the many platforms being promoted raises questions regarding analytic validation, said Dr Meropol, asking questions that raise concern. How are the genes selected for these panels? Does the platform look at the transcriptome or just the genome? Are we looking at epigenomic changes that may be relevant in selecting treatments? What is the turnaround time for the results? And what is the cost of these assays?
2. Tumors Are Heterogenous and Complex
Mutations in a tumor may be different between sites. “Although they may look the same under a microscope, colon cancers and gastric cancers are extremely complex, and extremely heterogenous in terms of their molecular profiles,” Dr Meropol said. Nor will finding a single driver mutation guarantee that a single drug intervention will be effective, because of “cross talk” between pathways occurring downstream of a key mutation. “If we’re going to use a tumor biopsy for selecting treatment for an individual patient,” he said. “This biopsy should be done proximate to the time that we’re going to intervene with a new therapy.”
3. We Don’t Know the Drivers
According to Dr Meropol, definitions used in the MATCH trial for identifying drivers to explore in a clinical trial are not yet good enough for routine care. “The reasons not to try it are that these are costly interventions, they may not work, they have side effectsterm, and they provide false hope. This should not be our routine approach with patients,” he advised.
4. We Don’t Have the Evidence that Links Drugs to These Drivers
The best level evidence is an FDA-approved dyad, but Dr Meropol stressed that this level of evidence is missing in nearly all drug cases. “Even if an agent meets a clinical end point, and there’s evidence of target inhibition, and there’s plausible evidence of a predictive or selection assay or analyte,” he said, “until it’s been proved in a prospective clinical trial, the evidence may be viewed as weak in terms of routine clinical practice.” Preclinical evidence is an even poorer prognosticator for patient outcomes. Dr Meropol iterated the need for incentivizing the development of biomarkers worldwide.
5. Investigational Drugs Are Not Widely Available
There are limited clinical trial sites; patients have to seek studies and travel for them. “Not everybody lives in close proximity to a research center that has access to multiple clinical trials and new agents,” Dr Meropol said, “and getting compassionate access to a new drug in development is a logistically complicated process. It’s time-consuming and rather opaque.”
6. It Isn’t Practical to Screen Many to (Maybe) Help a Few
The evidence is simply not there at this time, he said, highlighting a phase 1 study conducted at M.D. Anderson Cancer Center that intended to show the benefit of identifying clinical trials that might be appropriate for patients’ tumors. Of the 1283 patients assessed, 31% had at least 1 mutation, and among those who could be matched for treatment, approximately 10% had an antitumor response. But the overall response rate based on matching was only 2.4%.
7. There Is No Mechanism to Pay for Drugs for Off-Label Use
Off-label use of expensive targeted agents is increasingly scrutinized by payers, and costs are falling on patients. “Recommending cancer drugs with high copays may not be ethical without strong evidence that it’s going to help that individual patient,” he said.
8. Drug Approval Based on Tumor Type, Not on Genotype
The current (and old) paradigm is histology-based and requires large prospective phase 3 trials to provide the level of evidence that leads to a FDA and worldwide drug approval. But an emerging paradigm for drug approval is genome-based, using small studies looking for big effect to make decisions, “but we’re simply not there yet,” he said.
9. No Statistical Approach to Interpreting a Series of Anecdotes
Dr Meropol asked: How much evidence is needed to conclude that a particular mutation should receive a particular therapy? How many patients are needed with outcome data across how many tissues of origin? And when looking at a major response, how can we distinguish between a fluke and an outcome that is real? The answers to these questions remain unknown.
10. Unintended Consequences for Patients
Given the potential for confusion over interpretation, it is important to know who is interpreting the data and making the recommendations. “We don’t want to give our patients false hope,” said Dr Meropol. “We don’t want to subject them to the risks of needless biopsies, and we don’t want to subject them to the financial burden of therapies and procedures that are not destined to help them.”
Source: Association for Value-Based Cancer Care (February 2015, Vol 6, No 1 - Personalized Medicine)