Two roads diverged in a wood, and I –
I took the one less traveled by,
And that has made all the difference.
Robert Frost, "The Road Not Taken"
I worry that, as clinical researchers, we are increasingly choosing the roads more traveled when it comes to defining trial outcomes – selecting the ones that make little difference to our patients.
I recently attended a meeting where results were presented from clinical trials in myelodysplastic syndromes (MDS). For the purposes of this essay, it could have been a conference focused on leukemias, lymphomas, plasma cell dyscrasias, or hematopoietic cell transplantation populations. But, in this case it was MDS.
You probably recognize the scene: darkened auditorium, international audience, wide range of sartorial choices, variable percentages of participants focused on either their smartphones or the speaker, no open seats available near the aisles.
The first presentation of the session provided an overview of exciting clinical trials that are exploring new drugs or drug combinations. The future seemed bright, the preclinical bases for the studies sound, the potential for making a difference in our patients' lives huge. It set the stage for the results from the clinical trials to follow.
The next presentation, a study of a rationally developed drug for patients whose MDS expressed a specific molecular profile, showed great results: an overall response rate (ORR) more than double what we have seen with other drugs on the market.
But how was "response" defined? There were some of the standards we all recognize: complete remission (CR), partial remission (PR), and even hematologic improvement, which is specific to MDS. But lumped into the category was the term "marrow CR," abbreviated CRm, which is defined as having bone marrow (BM) with <5 percent myeloblasts (and a decrease by >50% from pretreatment blast percentages, conventionally having started with ≥5% blasts in the BM), irrespective of whether a person's blood counts have normalized.
As a reminder, this can also be achieved with high-dose, cytotoxic chemotherapy such as cyclophosphamide or with a massive radiation exposure. And it has no association with improved survival.
The next presentation of a different clinical trial showed a similarly high ORR, including the same CRm endpoint lumped in with other definitions of "response." This time, though, the investigators also included stable disease (SD) as a response.
Now, wait a second. I thought SD means the hematologic malignancy doesn't get much better or much worse, whereas response means it gets better by some metric. So that would be like saying the Cleveland Browns were Super Bowl contenders last year because they went 1-15, whereas the previous year they didn't even come close, going 3-13.
That's right, fellow Steelers fans – I went there.
Other clinical trials in myeloid malignancies have piled on, incorporating CRm into ORR calculations, or the suspicious term "morphologic leukemia-free state," abbreviated MLFS. MLFS has essentially the same definition as CRm, but with one additional letter in the abbreviation.
Regarding its value, see the above discussion on cyclophosphamide and massive radiation exposure.
Perhaps even more disturbing, trials reported at a number of conferences are not including endpoints like response duration, often choosing instead to display waterfall plots of time on treatment. These show colorful bars spreading across a slide, each representing an individual patient and the length of time that patient received a drug. These can be both visually and therapeutically misleading – our eyes focus on the long bars and can't process a true average –and do not reflect clinically meaningful benefits.
Some have argued that we should be more accepting of endpoints such as better response rates, as even small incremental benefits can add up, over time, to more substantive improvements in outcomes.
Others have made the case that an ORR that includes CR and PR alone may be good enough, as these might be considered interim markers reasonably likely to translate to a clinically meaningful benefit such as overall survival (OS; adopting the type of definition the U.S. Food and Drug Administration might use for accelerated approval).
Yet, a number of recent randomized trials, particularly those in predominantly older adults with myeloid malignancies, have shown a doubling of CR rates in the experimental study arms, compared with the control arms, with no impact (or, arguably, even a negative impact) on OS.
So, rather than representing true advances, focusing on ORR – and, worse yet, diluting the definition of response with clinically meaningless abbreviations – only fools us into thinking that fairly active drugs or drug combinations are truly advancing our therapeutic armamentarium. This wastes time in drug development, tricks us into off-label drug use, and, most importantly, exposes our patients to treatments with the false hope of progress.
Oh, I kept the first for another day!
Yet knowing how way leads on to way,
I doubted if I should ever come back.
Let's keep the more traveled road for another day and move away from conducting clinical trials with endpoints that maximize numbers we can report to our colleagues to designing studies that focus on what's meaningful to our patients.
Mikkael A. Sekeres, MD, MS
Editor-in-Chief