Stopping Leukemia in Its Tracks

Should Preemptive Hematopoietic Stem-cell Transplantation Be Offered to Patients at Increased Genetic Risk for Acute Myeloid Leukemia?

Kayla V. Hamilton, MS; Luke Maese, DO; Jonathan M. Marron, MD, MPH; Michael A. Pulsipher, MD; Christopher C. Porter, MD; Kim E. Nichols, MD


J Clin Oncol. 2019;37(24):2098-2104. 

Although it has been recognized for decades that leukemia can develop in the context of rare inherited disorders, only recently has the underlying genetic etiology for many of these disorders been revealed. This is particularly true for myeloid neoplasms, in which genomic investigations have identified several new genes and associated genetic syndromes that increase the risk for acute myeloid leukemia (AML). Among these recently identified syndromes and their associated genes are CEBPA-associated predisposition to AML (fAML-CEBPA);[1,2] familial platelet disorder with associated myeloid malignancy (RUNX1);[3,4]GATA2-associated predisposition to myelodysplastic syndrome (MDS)/AML (GATA2-MDS/AML);[5] myelodysplasia, restriction of growth, adrenal hypoplasia, genital phenotypes, and enteropathy (SAMD9);[6,7] ataxia pancytopenia (SAMD9L);[8,9] thrombocytopenia 2 (ANKRD26);[10,11] and DDX41-associated predisposition to AML[12,13] (Table 1). In some of these syndromes, AML occurs acutely and may be the primary manifestation (eg, fAML-CEBPA),[14] whereas in others, AML arises in the context of other hematologic abnormalities, such as antecedent thrombocytopenia (RUNX1, ETV6, and ANKRD26) and/or MDS (GATA2, SAMD9, SAMD9L, ANKRD26, DDX41, and ETV6, among others).[10,12,16,23,26,28–34] Finally, for some syndromes, individuals are also at risk for development of nononcologic manifestations, such as neurologic and other organ dysfunctions or immunodeficiency (eg, SAMD9, SAMD9L, and GATA2).[8,19,24,25,27,35]

Thanks to the growing awareness of these syndromes and the expanding availability of clinical genetic counseling and testing options, more individuals with an underlying predisposition are being identified. Although the identification of a genetic predisposition provides an explanation for leukemia development and enables counseling and testing of family members, such information brings many challenges. Perhaps one of the greatest is how best to treat at-risk individuals, particularly those who are most likely to develop acute-onset AML. For these individuals, AML generally occurs rapidly and without warning, and there are currently no effective means to predict its onset.[36,37] This is in contrast to individuals with conditions that predispose to AML in the setting of an underlying MDS (eg, Fanconi anemia), for whom there is mounting evidence that preemptive allogeneic hematopoietic stem-cell transplantation (pHSCT) can improve overall outcomes.[36,38–42] For these individuals, AML often evolves gradually and manifests with progressive cytopenias or bone marrow cytogenetic abnormalities. This allows time to prepare for and carry out pHSCT before the onset of overt AML. In so doing, patients are spared exposure to intensive chemotherapeutic regimens, which are associated with significant morbidity and mortality. Of note, pHSCT has recently been reported to correct immunodeficiency in a patient with ataxia telangiectasia, a hereditary condition with significantly increased risk for acute lymphoid malignancies.[43]

Given this information, the question arises as to whether pHSCT will hold the same medical benefit for patients with predisposition to acute-onset myeloid leukemia. Here, we provide a discussion of the clinical and ethical variables that must be considered when contemplating pHSCT for individuals at increased genetic risk for AML, considering emerging data regarding the myeloid leukemia predisposition syndromes and their genetic penetrance, and we provide a framework for handling these decisions.

An illustrative example that underscores the complexities of making a recommendation for pHSCT is fAML-CEBPA. fAML-CEBPA is an autosomal-dominant condition of two subtypes that are dependent on whether the germline CEBPA mutation is located at the N- or C-terminus of the gene. Penetrance of AML is on the order of 90% for those with N-terminal CEBPA mutations versus approximately 50% for those with C-terminal mutations.[14,15] In both forms, patients have similar ages of leukemia onset, that is, at 24.5 and 29 years (median) for N- and C-terminal fAML-CEBPA, respectively, and neither form is associated with antecedent MDS or cytopenias.[14] For individuals with N-terminal CEBPAmutations, AML tends to respond favorably to chemotherapy;[14] however, there remains a lifelong risk for subsequent primary episodes of AML and accordingly, risk for additional exposure to chemotherapy and potentially HSCT. Because of the rarity of C-terminal fAML-CEBPA, disease response and outcomes have yet to be defined. Nonetheless, as shown in Figure 1, these two forms of fAML-CEBPArepresent distinct clinical entities to consider—one with an AML penetrance of 90% and one with an AML penetrance approximately 50%. If one assumes that disease-free survival after HSCT for the treatment of AML in individuals with N-terminal or C-terminal fAML-CEBPA is similar to disease-free survival after HSCT for individuals with sporadic AML (ie, 50%[44]), and transplantation-related mortality is approximately 15% for individuals undergoing pHSCT,[45] it is likely that pHSCT will exert a greater impact on survival for individuals with N-terminal fAML-CEBPA as compared with individuals with C-terminal fAML-CEBPA (Figure 1). Presumably, a similar argument can be made for other AML-predisposing genetic conditions with varying degrees of disease penetrance.

Figure 1.

Difference in survival rates for 90% penetrant versus 50% penetrant leukemia predisposition syndromes, comparing anticipated outcomes with the traditional approach of treatment after development of leukemia to anticipated outcomes with preemptive allogenic hematopoietic stem-cell transpantation (pHSCT). Figure demonstrates an increase in survival rates with the use of pHSCT, with a greater difference in survival rates for the more highly penetrant condition. AML, acute myeloid leukemia.

However, it is important to note that in addition to leukemia penetrance, there are several other factors that must be taken into account when considering pHSCT for a patient with an underlying predisposition. First, and as noted earlier in this article, the spectrum of tumors is variable across AML predisposing conditions. In some, individuals are primarily predisposed to myeloid neoplams.[37] Thus, the malignancy risk associated with these syndromes can theoretically be entirely mitigated with pHSCT. In others, affected individuals are also at risk for development of solid tumors (eg, Fanconi anemia,[46,47] dyskeratosis congenita[48]). Although pHSCT lessens the chance for AML, it does little to prevent these other cancers.[41,42] On the other hand, pHSCT can correct the immunologic abnormalities that accompany certain syndromes, such as GATA2-associated MDS/AML. Second, the age of AML onset can significantly influence prognosis. In this regard, older adults generally have a poorer response and less ability to tolerate AML chemotherapy and HSCT compared with children and young adults.[49,51] Therefore, for individuals at risk for AML as older adults (DDX41), should pHSCT be performed, and if so, at what age? These questions are particularly difficult to answer when penetrance is incomplete. Third, individuals with an underlying genetic predisposition may be at increased risk for toxicities after conventional AML therapies, which can preclude HSCT or worsen its outcomes.[52,53]Consequently, it is essential to intervene early, but when is the right time? Using genomic profiling to interrogate the pattern of somatic second "hits" in leukemia samples from individuals with familial MDS/AML, Churpek et al[54] recently identified recurrent mutations in several familial AML cases. Strikingly, the authors also detected clonal hematopoiesis in 67% of asymptomatic RUNX1 mutation carriers younger than 50 years of age, a prevalence significantly higher than in the general population. Currently, it remains unclear how best to incorporate genomic profiling such as this into the clinical care of individuals with familial predisposition to AML. However, as we learn more about clonal hematopoiesis and the patterns of repeated somatic second hits in familial AML cases, it is likely that in the future, serial monitoring through deep sequencing of blood samples from individuals with leukemia predisposition syndromes will inform decisions about indications for and timing of pHSCT. Finally, it is important to consider the possible outcome in the absence of HSCT. Unfortunately, because of the rarity and recent discovery of many of the AML-predisposing syndromes, we are only at the earliest stages of characterizing disease penetrance, clinical features, and longer-term outcomes, with or without HSCT.

When considering pHSCT, medical factors are only part of the decisional calculus. A strong ethical framework is also necessary to support at-risk individuals and clinicians in making these difficult decisions. Principlism is one such framework, which looks to individual principles to guide ethical decision-making.[55] Four commonly used principles are autonomy, beneficence, nonmaleficence, and justice.[55] Following a principle-based approach does not ensure that individuals and clinicians make the right decisions, but it can be helpful when approaching a challenging dilemma such as whether to pursue pHSCT in the setting of a leukemia predisposition.

Western medicine emphasizes respecting a patient's autonomous right to make medical decisions on the basis of his or her own values and preferences. Competent adult patients often make decisions that are not in their best interest, but such decisions are respected. When considering pHSCT, clinicians have an ethical obligation to provide patients with the information necessary for them to provide truly informed consent, including information about the risks and benefits of pHSCT as well as its uncertainties. Patients' perceptions of, and aversions to, risk strongly predict their choices regarding cancer prevention.[56] Therefore, it is quite plausible that some patients might choose to wait and watch rather than undergo pHSCT. Given the favorable outcomes for younger HSCT recipients, pHSCT would likely be considered by parents for at-risk children and adolescents. However, this can add even more complexity. Although adult patients provide consent (or refusal) for medical interventions, parents provide permission (or not) for the treatment of their minor children, who then provide assent (or not) after receiving age-appropriate information about planned interventions.[57] What if the parents and children do not agree? How to proceed in such a scenario is unclear, although consultation with ethics, psychosocial, and legal teams may prove beneficial.

Ultimately, most clinicians and families are motivated by the principle of beneficence, aiming to act in a patient's best interest. A strong driver in oncology is the "rule of rescue," the imperative to save those in danger of death, sometimes with limited regard for the costs of doing so.[58] For a patient with an AML predisposition syndrome, a clinician's intuition may be to act—in this case, perhaps to recommend pHSCT to prevent the development of leukemia. But is this truly in the patient's best interest? Although AML is not in any patient's best interest, the development of AML in individuals with an underlying predisposition is far from a certainty. For example, in Figure 1, scenario 2, 50% of pHSCT recipients represent presumably healthy but genetically affected individuals who would never develop leukemia. pHSCT would subject these patients to significant—and arguably unnecessary—risk. The situation is different in scenario 1, where 90% of patients develop leukemia. For these individuals, pHSCT would prevent the need for chemotherapy and eliminate the risk for second primary AML. Thus, the scales between beneficence and nonmaleficence ("do no harm"), tip based on complex probabilities and many uncertainties. One must also consider the ethical dilemma of testing related donors for the familial predisposition syndrome, especially in cases when the donor does not want to know his or her germline mutation status. In such cases, relatives might feel pressured to be tested so as not to disappoint other family members, although they might not be ready to handle the results should they test positive. This situation is even more challenging when the potential donor is an underage sibling who may not be old enough to make an informed and independent decision about testing.

A final ethical consideration is that of justice or treating like patients alike. Modern medicine arms clinicians with numerous tools and technologies, but we are only beginning to understand how best to harness and use these for the benefit of patients. Does available evidence support offering or even encouraging pHSCT for all patients at risk for development of AML? As providers and health care systems begin to focus more on value-based cancer care,[59] thought must be given to whether pHSCT falls in line with personal values for a given patient, and also whether it provides value for all such patients.

As can be seen, decision-making around pHSCT is a complicated process that must incorporate syndrome-specific factors (eg, disease penetrance, age of AML onset, presence of MDS, risks for therapy-associated toxicities), transplantation-related factors (eg, age of recipient; donor type and availability; risks for short- and longer-term toxicities, including acute and chronic graft v host disease and various end-organ dysfunctions) and patient-specific factors (eg, personal preferences, psychological status, family and other support mechanisms). Such decisions require the involvement of multiple disciplines, including oncology, genetics, stem-cell transplantation, and psychology, as well as the patients and their families. The diversity of patient and family backgrounds and heterogeneity of predisposition syndromes precludes a prescribed approach; thus, discussions should be driven by current knowledge of specific predisposing conditions and tailored to the unique circumstances of each individual patient.

Overall, despite its inherent risks and uncertainties, pHSCT may provide both medical and psychological benefits to families facing life-long and potentially debilitating or life-threatening risks related to certain blood disorders. This was recently discussed in the context of sickle cell anemia,[60,61] and we present pertinent arguments here for individuals with highly penetrant leukemia predisposition syndromes associated with the rapid onset of AML early in life and a high risk for multiple AML diagnoses. In contrast, pHSCT may be less beneficial for those with syndromes for which AML penetrance is lower and associated with later onset. Although this article focuses on predisposition to AML, many of the same guiding principles can be applied to consideration of pHSCT for individuals with predisposition to lymphoid malignancies or increased genetic susceptibility to therapy-induced MDS/leukemia.[62] As genetic testing for leukemia predisposing conditions becomes more commonplace, it is important that providers carefully weigh the risks and benefits of pHSCT for each patient. Providers must clearly communicate the risks, benefits, and uncertainties of pHSCT and encourage discussions to identify how patients and families conceptualize these risks and benefits. They must also take into account psychosocial considerations and help families make as informed a decision as possible that incorporates their personal beliefs, values, and experiences.