“All great truths begin as blasphemies.”
—George Bernard Shaw
(Annajanksa, 1918)
Until about 15 years ago, persons with acute myelogenous leukemia (AML) were considered candidates to receive a blood cell or bone marrow allotransplant in first remission only if they had had an HLA-identical sibling donor. However, advances using alternative donors such as an HLA-matched unrelated person, HLA haplotype–matched relative, or HLA-matched umbilical cord blood cells from an unrelated donor have changed the transplant landscape.
Currently more than half of all allotransplants for persons with AML in first remission use a donor other than an HLA-identical sibling. Many see this as a giant leap forward, perhaps akin to Mao’s Cultural Revolution. In China, where government policy means most people with AML do not have a sibling, more than 1,500 HLA haplotype–matched transplants are done annually, many for persons with AML in first remission. That is more procedures of this type than are done in the rest of the world combined.
All of this is quite exciting, but one really important question remains unanswered: should anyone with AML, even those with a perfect donor, receive a transplant in first remission? We recently discussed this issue,1 raising the blasphemous thought that transplants in first-remission AML may be inappropriate, even in persons with high-risk AML. All of what follows, of course, hinges on whether we can agree on the definition of “high-risk AML.”
Four Prerequisites
Our opinion is based on analyses of data from clinical trials as well as on theoretical considerations. Four conditions are needed to justify transplants for AML in first remission:
- (1) We should be able to reasonably precisely identify persons with AML in first remission (usually for a few months) at high risk for relapse.
- (2) A transplant in first remission should have a therapeutic advantage over conventional therapy (if any) of a sufficient magnitude to overcome the adverse biologic features of the high-risk leukemia.
- (3) A transplant in first remission should confer a substantial survival or leukemia-free survival advantage over waiting to determine if or when a person relapses and then doing a transplant (with or without additional therapy).
- (4) We cause little harm to persons in first remission misidentified as likely to relapse and who receive a transplant.
Unfortunately, few data support any of these conditions. For example, we are poor at identifying, at the subject level, persons in first remission for a few months likely to relapse subsequently. Receiver-operator curves of prediction accuracy show we exhaust most of our predictive value by this time. A person with high-risk cytogenetics at diagnosis who rapidly achieves a remission and does not relapse in the first 3 to 6 months has a markedly different prognosis than he or she had at diagnosis. This is because the best predictor of remission is remission and the best predictor of relapse is relapse, not cytogenetics. Although the a priori odds may be stacked against you, if you make it to kilometer 40, you are likely to finish the marathon. (Bayesians arise!)
As for a therapeutic advantage, most data show that variables (or combinations thereof) predicting an unfavorable outcome with conventional therapy also operate after transplants. This should come as no surprise to hematologists and oncologists. Unfortunately, the adverse impact of bad cancer biology is usually stronger than the favorable impact of new therapies. Even if we assume transplants have a slight antileukemia advantage over conventional therapy, some or all of this benefit is lost to treatment-related mortality. And other bad things can happen, such as chronic graft-vs-host-disease, which may be worse than death from leukemia.
To know whether bad cancer biology can be overcome in this setting, we need a study in which high-risk subjects are randomly assigned to receive or not receive the new intervention being tested. Such a study has not been reported, but, fortunately, several are in progress.
Early vs Late Transplant
The third assumption is critical. Are there really convincing data from a randomized study showing a benefit for transplants in first remission vs waiting for relapse and performing transplants only in these persons? No. The correct answer to whether there is a benefit of transplants over conventional therapy can only be derived from a trial starting at a reasonable time into first remission (say, 3 months), in which eligible subjects are randomly assigned to what is essentially an early vs late transplant in those who relapse.
Such a study, of course, needs to be analyzed on an intent-to-treat basis, as some (many?) persons who relapse may not make it to a transplant and some persons assigned to a transplant may relapse before it can be done. Also, being in second remission should not be a requirement to proceed to a transplant in a person who relapses.
Few studies attempting to address this question, even tangentially, are reported. The largest found comparable outcomes with a strategy of delaying transplants in persons with intermediate-risk AML until relapse with the advantage of having to do fewer transplants.2 Let’s call it a wash for now.
As for the final condition—primum non nocere—there is little question that performing a transplant in someone already cured by conventional therapy is hardly in his/her interest.
Dubious Nomenclature
Consider the nomenclature of AML risk assignment. Saying someone with newly diagnosed AML is “good-risk” seems a stretch. Good-risk compared to what? A 65-year-old with good-risk AML has a 5-year survival probability of less than 20%. Good-risk in this context ranks close to “military intelligence” and “parental guidance” as a euphemism. People with good-risk AML should not buy an extended warranty on their new washer/dryer. (Consumer Reports suggests none of us do; free advice. Perhaps they know something we don’t.)
Finally, some will say our discussion is akin to debating how many angels can dance on the head of a pin—medieval. Now that we have tests of measurable (not minimal) residual disease (MRD),3 can’t we simply reserve transplants for persons in first remission who are MRD-positive? Sadly, this is not so simple. Specificity and sensitivity of MRD tests for AML are quite imperfect. And although there is an association between a positive MRD test and an increased relapse risk, 20% to 40% of persons who are MRD-positive do not relapse, whereas 20% to 30% of persons who are MRD-negative do.
In a recent large, well-done study of MRD testing in persons with AML in first remission, test results were informative in making a therapeutic decision in only about 20% of subjects.4 There were also high rates of false-negative and false-positive correlations between MRD test results and clinical outcomes.
So asking whether it is better to be MRD-negative than MRD-positive is a bit like asking is it better to be young, beautiful, and rich, or old, ugly, and poor. The answer is simple, but it’s not the whole story. One is reminded of when George Bernard Shaw received this offer from Isadora Duncan: “Will you be the father of my next child? A combination of my beauty and your brains would startle the world.” Shaw replied, “I must decline your offer with thanks, for the child might have my beauty and your brains.”
Complex Biology
None of this is surprising given several technical issues including sampling error at low leukemia cell frequencies.5 And then there is the fundamental biologic issue of the clonal diversity of AML at diagnosis, revealed by whole-exome and whole-genome sequencing.6,7 Because persons with AML at diagnosis have multiple clones with diverse mutations and with unknown capacities to proliferate, it seems rather unlikely we can devise an MRD test that will capture this complex biology. Moreover, it might be necessary to develop a different MRD test for each person. Added to this is our lack of knowledge of the phenotype and genotype of the cell(s) with the biologic potential to cause recurrent AML (rather than progeny of the clone).
We may be blasphemers but we are not nihilists, and we support clinical trials to test whether allotransplants in persons with AML in first remission—even those with high-risk leukemia (whatever that is)—are appropriate and needed for us to move forward. ■
Disclosure: Dr. Gale acknowledges support from the NIHR Biomedical Research Centre funding scheme and is a part-time employee of Celgene. Drs. Lazarus and Wiernik reported no potential conflicts of interest.
References
1. Gale RP, Wiernik PH, Lazarus HM: Should persons with acute myeloid leukemia have a transplant in first remission? Leukemia 28:1949-1952, 2014.
2. Burnett AK, Goldstone A, Hills RK, et al: Curability of patients with acute myeloid leukemia who did not undergo transplantation in first remission. J Clin Oncol 31:1293-1301, 2013.
3. Goldman JM, Gale RP: What does MRD in leukemia really mean? Leukemia 28:1131, 2014.
4. Terwijn M, van Putten WL, Kelder A, et al: High prognostic impact of flow cytometric minimal residual disease detection in acute myeloid leukemia: Data from the HOVON/SAKK AML 42A study. J Clin Oncol 31:3889-3897, 2013.
5. Butturini A, Klein J, Gale RP: Modeling minimal residual disease (MRD) testing. Leuk Res 27:293-300, 2002.
6. Ding L, Ley TJ, Larson DE, et al: Clonal evolution in relapsed acute myeloid leukemia revealed by whole-genome sequencing. Nature 481:506-510, 2012.
7. Cancer Genome Atlas Research Network: Genomic and epigenomic landscapes of adult de novo acute myeloid leukemia. N Engl J Med 368:2059-2074, 2013.
Dr. Gale is Visiting Professor of Haematology, Haematology Research Centre, Division of Experimental Medicine, Department of Medicine, Imperial College London and is a part-time employee of Celgene Corporation. Dr. Lazarus is Professor of Medicine at Case Western Reserve University School of Medicine, Director of Novel Cell Therapy at University Hospitals Case Medical Center, Case Western Reserve University, Cleveland, and consultant for Actinium Pharmaceuticals. Dr. Wiernik is a medical oncologist at Cancer Research Foundation of New York, Chappaqua.
