Recently, at our cancer center's weekly leukemia meeting, things got ugly.
Usually, we use the meeting to discuss the progress of our inpatients; patients recently discharged from the hospital; new outpatient consults; and any programmatic or administrative issues that arise, to keep everyone in our group (docs, nurses, physician assistants, social workers, and research personnel) “in the loop” about clinical and nonclinical developments. It's a great venue to discuss challenging patient management issues, new clinical trials, and to introduce any recent publications that may influence our practice.
Last week, we invited the head of molecular pathology to the meeting so he could review some of the tests we were already sending, and so he could ask us what seemed like a simple question: “What other molecular tests do you guys want to send on all your patients?”
My word, you'd think he had just asked us to debate who should win American Idol. The clinical folks wanted to stick with the basics, the pathologists wanted to prioritize what was justified by volume and test complexity, while the translational researchers started arguing that we should send absolutely everything they could think of—not just routine cytogenetics and fluorescence in situ hybridization for core binding factor abnormalities, BCR-ABL, MLL, or lesions common in myelodysplastic syndromes, but also tests for NPM1, FLT3 (both internal tandem duplications and tyrosine kinase domain lesions), CKIT for patients with core binding factor abnormalities, CEBPA, DNMT3A, WT1, RAS, p53, TET2, JAK2, IDH1, IDH2—there were bifocals and pocket protectors flying everywhere!
How do you make sense of the wealth of molecular information coming out almost weekly further refining prognosis in leukemia (and confirming what we've all know for years—that every cancer subtype really represents hundreds of different diagnoses, each driven by discrete clonal abnormalities), and use that information practically to treat your patients?
I did what any confused program leader would do—I emailed a bunch of other leukemia program leaders to see what they do, to develop some kind of consensus approach to testing leukemia patients. I learned that we all send approximately the same tests; we tend to be late adaptors; and we all expressed some hesitancy in determining what was the right time to incorporate new molecular tests into our decision-making.
Why the controversy?
There are a couple of big issues. Most centers with dedicated leukemia programs also have large clinical and translational research arms, many of which are doing incredible work at defining the biologic underpinnings of these diseases, and the clinical implications of that biology. It's much easier for these researchers to collect information about tissue abnormalities prospectively, with a patient sitting in front of them, rather than having to rely on consented, donated, and stored specimens that must be retrieved for testing. It's also cheaper for researchers if results from testing are considered “standard of care,” and thus paid for by insurers, or under clinical budgets.
Now, naturally this all has to be done ethically—so leukemia docs have to agree that a test is so convincingly a game changer that we should all be using it to make clinical decisions. The problem you run into is that many hospital admissions for a leukemia diagnosis and therapy are capitated. In other words, a hospital is paid a set amount for a patient's entire admission, whether that patient receives inexpensive, generic treatment and undergoes minimal testing, or the latest, boutique chemotherapy complemented by multiple, complicated, expensive molecular tests. The more tests we order, the more likely hospitals are to lose money on our patients, and refuse to allow their admission.
The clinical challenge with collecting that great molecular information in real time is that we will feel compelled to use it to make treatment decisions. For example, the Washington University group recently reported that 22% of 281 acute myeloid leukemia patients had DNMT3A mutations, and that those with mutations, which were more common among patients with intermediate risk disease, had a worse survival than those without mutations. So, is this impressive enough data that, today, if I detect this abnormality in one of my own leukemia patients, I should recommend a bone marrow transplant in first complete remission? What if my patient otherwise has good clinical prognostic indicators, like a younger age, low presenting white blood cell count, and no comorbidities? Has this lesion (or other molecular abnormalities) been studied in the setting of enough other well-recognized prognostic indicators to make it to the prime time of clinical practice?
There's an inherent conflict in how we approach the application of molecular abnormalities to our patients in that, often, the information about lesions is collected retrospectively, makes a big splash, and then gathers a groundswell of support to be incorporated into practice before it is validated prospectively. Techniques as sophisticated as whole genome profiling are wildly exciting, and ultimately will enable us to fundamentally understand these heterogeneous diseases. When I talk to other leukemia docs, though, I get the sense that we all recognize this conflict, and as a result try to be prudent in applying this new, exciting information to the patient sitting in front of us.
No pathologists were harmed in the writing of this editorial.