Thursday, November 8, 2012

Why genetically profile a glioma?

A couple of weeks ago I met with representatives of Castle Biosciences regarding their proprietary gene expression profile assay for glioblastomas, called DecisionDx-GBM, as well as their multi-methylation test for grade II and III gliomas, called DecisionDX-G-CIMP. During the meeting, I brought up a crucial issue raised by a Neuropathology Blog reader. I'll quote part of the reader's comment: "I doubt anyone wants to give a patient a life expectancy prediction based on the results of this test, since individuals may fall at any point on a survival curve. So I would not be eager to recommend this test until there are alternative treatments for those in poor prognosis groups." The representative's response to this concern was that prognostication is an important part of the decision-making process that a neuro-oncologist, in collaboration with the patient, takes into account in recommending a course of treatment. For example, when should Avastin be introduced into the regimen? If a patient has a tumor with a genetic signature that has a longer median survival, perhaps the neuro-oncologist would be more likely to hold back on the introduction of Avastin, keeping it in his armamentarium for later in the disease course. I would add that if I personally had the misfortune of being diagnosed with a glioblastoma, I would like to have as much prognostic information as possible just to help me adjust psychologically to dealing with the the disease. I realize that one individual can fall at any point on a survival curve. But put yourself in the patient's shoes for a moment. Wouldn't you want this test done, as well as IDH1 mutation, MGMT methylation status, 1p/19q testing, and whatever else that might be available to get as much information as possible about this disease has invaded you? Since, as Castle Biosciences states, insurance will cover this test, I see both practical and intangible benefits to doing it. I would be curious to hear what other neuropathologists have to say about this issue in general and about the Castle Biosciences tests in particular. I should note that the Castle tests were developed by the widely-respected neuropathologists Dr. Kenneth Aldape at MD Anderson. I am seriously entertaining the idea of routinely using the Castle Biosciences tests at my institution and would be interested in the advice of the neuropathology community as a whole regarding this issue. Please post to comments.

5 comments:

henry brown said...

Although you might think that more prognostic information is better, this may not be the case. More information may only confuse the issue -- the more tests you run, the more likely you will get discrepant results. Also, to be very cynical, you should look at a recent NewEngJMed article Oct 25 2012 volvol 367 #17 pp 1616-1625. Briefly, patients with bad cancers are going to die, chemo only prolongs their life, and oncologists commit sins of omission in not educating patients about their dismal chances (i.e., patients think that chemo and radiotherapy will 'cure' them). So I am not a big fan of expensive tests that do not lead directly to a possible targetted therapy. Nevertheless, I do IDH1 testing, and I expect to start MGMT pretty soon!

Brian E. Moore, MD, MEd said...

I'll take a look at that NEJM article, Henry. Thanks! Now, given what you said, why do you do IDH1 testing?

Craig said...

Given the strong (though not perfect 1:1) link between G-CIMP and IDH mutations, I'm uncertain as to whether testing for G-CIMP adds significant value to IDH screening.

In general there's a growing disconnect between genetic testing on tumors and an increasing emphasis on cost containment in healthcare. For example, ASCO's now recommending upfront HER2 and ALK FISH testing for breast and lung cancers, respectively, even though we have fairly reasonable algorithms that target the subset of cases with a higher likelihood of having amplification or rearrangement. We could save a ton of $$ by being more parsimonious with such tests.

I'm as big a fan of personalized medicine and molecular diagnostics as anyone, but it still has to be cost-effective.

Craig said...

Given the strong (though not perfect 1:1) link between G-CIMP and IDH mutations, I'm uncertain as to whether testing for G-CIMP adds significant value to IDH screening.

In general there's a growing disconnect between genetic testing on tumors and an increasing emphasis on cost containment in healthcare. For example, ASCO's now recommending upfront HER2 and ALK FISH testing for breast and lung cancers, respectively, even though we have fairly reasonable algorithms that target the subset of cases with a higher likelihood of having amplification or rearrangement. We could save a ton of $$ by being more parsimonious with such tests.

I'm as big a fan of personalized medicine and molecular diagnostics as anyone, but it still has to be cost-effective.

Brian E. Moore, MD, MEd said...

Craig's comment begs the question I posed to Henry: When done for prognostic rather than diagnostic purposes, why even do IDH1 immunohistochemistry if the result yields no implications for treatment?

Dr. Diamandis develops network to help pathologists interface with AI computational scientists

A neuropathology colleague in Toronto (Dr. Phedias Diamandis) is developing some amazing AI-based tools for pathology and academia. He hel...