Pharma’s Cutting Edge

NEJM — Detection of Mutations in EGFR in Circulating Lung-Cancer Cells (Sub Req)

Look to history for a sense of the import of today’s medical advances. 

Most authorities recognize TR Ashworth, writing in the Australian Medical Jorunal, as making the first discovery of cells in the blood similar to those in patients’ solid tumors (post-mortem).  That was in 1869.  I couldn’t find the original article online, but the citation is: [Ashworth TR (1869) A case of cancer in which cells similar to those in the tumours were seen in the blood after death. Aus Med J 14: 146–149, as referenced in this minireview.]

Surely, shortly after word of this discovery spread, scientists began pondering ways of exploiting this knowledge to diagnose and track the progression of cancer.  But a huge technological gulf separated the idea of using circulating tumor cells for the benefit of cancer patients and its application in the clinic.  As you might imagine, the number of circulating tumor cells (CTCs) relative to the total number of circulating blood cells is relatively small, necessitating reliable tumor cell-enrichment and detection technologies to make diagnostic/theranostic use of CTCs.

Fast forward some 135 years from Ashworth’s initial discovery to the routine use of techniques to exploit it.  In January 2004, Veridex, a J&J company, received US FDA approval to market CellSearch, a device for magnetically labeling antibody-selected epithelial cells in a blood sample, for use in patients with metastatic breast cancer as an aid to monitoring response to treatment.  The theory behind the device and accompanying analysis of the BRCA CTCs is that the number of CTCs correlates (inversely) with response to treatment.  To my knowledge, this is the only such separation/analysis system available in the U.S. for non-experimental monitoring of response to therapy via CTCs.  It has now been approved for marketing for metastatic colon and prostate cancer in addition to BRCA.

Unfortunately for Veridex and for patients, the system has not been shown to affect outcomes in metastatic breast cancer, where it has been most thoroughly studied, leading ASCO to make the following practice recommendation:

The measurement of circulating tumor cells (CTCs) should not be used to make the diagnosis of breast cancer or to influence any treatment decisions in patients with breast cancer. Similarly, the use of the recently US Food and Drug Administration–cleared test for CTCs (CellSearch Assay; Veridex, Warren, NJ) in patients with metastatic breast cancer cannot be recommended until additional validation confirms the clinical value of this test.

Read the rest of this entry »

In Milestone, FDA Pushes Genetic Tests Tied to Drug - WSJ.com

Kudos to Anna Wilde Matthews, from the Wall Street Journal (Subscr. Req.), who has done a superb job documenting some of the practical difficulties of implementing genetic predictors of therapeutic response in clinical practice.  Here’s a snippet from her piece:

Some specialists say testing hasn’t been proved to reduce the risks of the drug. They fear patients who don’t get the tests and run into trouble will sue doctors, citing the drug’s label. While Medicare covers the tests, which generally cost between $300 and $500, major insurers such as Aetna Inc., WellPoint Inc. and Cigna Corp. don’t. The insurers say they need more evidence about the benefits.  “It would be irresponsible and potentially harmful to suggest that testing be used, or even mentioned, in the label,” said University of Washington professor Ann Wittkowsky in an interview before the FDA’s decision. “It is fascinating science, but it is not yet ready for prime time.” Larry Lesko, director of the clinical pharmacology office at the FDA, says the agency has “substantial” evidence to support the new label and hopes it will improve safety by informing doctors.

Dr. Lesko’s hope for a benefit of the genetic predictors is commendable; it shows he’s open to supporting new ways of improving the safety of potentially dangerous drugs.  It’s also irrelevant to prescribers and insurers who need evidence of actual benefit, not just the hope of benefit, to guide their practices.  Before FDA supports the widespread use of such theranostics, it must require theranostics vendors, or in some cases drug manufacturers, to perform the clinical studies needed to support their use.

In the absence of such studies, doctors prescribing warfarin should not deviate from their current empirical dosing practices using measurement of INR to gauge response.  The one exception to this recommendation is a high suspicion of genetic predisposition to atypical warfarin pharmacokinetics based on a suggestive family or personal history.  In such cases, the prior probability of a positive test might warrant its use prior to dosing.

Data on genetic predictors:

This interview with a Wyeth scientist is interesting to me mostly because I’m a believer that the future of cancer diagnostics, including theranostics, lies primarily in peripheral bood sample analysis. It’s sounds like such an obvious point, doesn’t it? After all, most routine diagnostics today are done using a peripheral blood sample, so why shouldn’t future diagnostics rely on peripheral blood as well? But it’s not at all obvious. Despite efforts to develop blood-based cancer diagnostics (think CA125, alpha-fetoprotein, HCG, etc), most of the “sexy” research in cancer diagnostics until recently in this field has been in tumor- or tissue-based (e.g. lymph node) diagnostics. Think growing a patient’s melanoma or lymphoma cells in vitro to look for cell-surface markers, for instance. The idea of using peripheral blood, and specifically, components of peripheral blood, such as peripheral blood mononuclear cells (PBMCs), that represent systemic responses to tumors and metastases is actually rather radical. I intend to keep a very close eye on this emerging field and will report significant advances.

The linked NYT article briefly discusses one of the more salient economic points faced by sponsors and academics interested in developing predictive tests of drug efficacy with established drug classes. It’s also a point that I’ve seen very little attention paid to in the lay media. I’m taking about the issue of class effects. Predictors of efficacy when the clinical read out (i.e. time to definitive efficacy determintation) is protracted have the potential to be useful, even when the drug class is associated with low toxicity and low cost. This utility is heightened when the disease being treated is serious. CHF fits this profile tightly. Beta blockers are cheap and relatively non-toxic and CHF is serious with a protracted time between therapy initiation and definitive efficacy. But they’re an established class, with many members. It’s likely that a predictive test will exhibit specificity spill-over among class members, as the article implies. The only way to to test for such spillover is a very large clinical trial that no single sponsor will want (or be able) to conduct. Even a consortium of sponsors would not be able to study every beta blocker, and only those beta blockers that will still be under patent protection after the trial is over would be considered worthy of study by the industry.

Governments have not even begun to come to grips with the implications of this situation. In the U.S., absent a single point of responsibility for public health economic policy, agencies such as CMS, AHRQ, NIH and FDA should ideally collaborate to determine whether and how such studies should be conducted and funded. The calculus will boil down to likelihood of finding predictive markers, the cost of developing the markers (including the mega study), the predictive value of the markers and the cost to use them routinely, and the potential cost savings to the healthcare system (particularly the publicly funded portion of that system) resulting from deployment of such tests. My sense is that we are unlikely to see such mega studies in the forseeable future, as the potential risk-adjusted, discounted cost savings will be found to be marginal relative to the costs of deployment in nearly all cases. As a result, when such tests are introduced after being studied for one member of a class, researchers will undoubtedly study them retrospectively (i.e. apply the tests to case series and prospective clinical studies done for other reasons) and then apply the tests clinically (usually off-label and without insurance coverage) based on these post hoc inferences. It’s far from ideal scientifically or policy-wise, but it’s going to be the reality for many established drug classes. It’s the main reason why governments should continue to encourage predictive tests of efficacy as new members of recent and new classes are developed.

I’ve been doing quite a bit of thinking about our microbiome (the collection of microorganisms that inhabit our bodies) of late. I’m thinking that we (as an industry) have grossly underappreciated its importance to intra- and inter-individual variability in disease phenotype and response to therapies, especially oral drug therapies. Pharmaceutical scientists whose work involves studying such variability and designing strategies to exploit and/or overcome it should immerse themselves in this rapidly growing field. The blog post I’ve linked to provides a good initial reference list. The work of Jeremy K. Nicholson at the Imperial College in London and Jeffrey I. Gordon of the Washington University in St. Louis is particularly important and provocative. Let’s add microbiomics to our pharmaceutical lexicons and give the 100 trillion bugs that inhabit our guts the attention and respect they warrant.