The majority of patients with acute myeloid leukemia (AML) will die of their disease. Nevertheless, the prognosis of AML varies widely. Some AML patients may be cured by chemotherapy alone, while others require approaches such as allogeneic stem cell transplantation to have the best chance of long-term survival. As physicians, we are often asked by our AML patients: “How likely is this treatment going to work, and how long do I have to live?”1
Prognostication in AML has evolved over time. Initially, models for prediction of response to therapy were based on patient’s parameters such as age and performance status in combination with cell characteristics such as morphology and chromosomal karyotype. With technological advancements, our understanding of disease biology has evolved and factors including molecular mutations and minimal residual disease have been integrated into prognostication schemes. Recently, an international expert panel on behalf of the European LeukemiaNet (ELN) published a revised version of a widely utilized prognostication scheme that categorizes AML patients into three risk groups (Favorable, Intermediate, and Adverse) based on genetic abnormalities (incorporating chromosomal karyotype and specific molecular mutations).2 These AML risk groups have profound clinical implications, particularly with regard to post-remission therapy for younger fit patients. In general, fit Favorable-risk AML patients who achieve a first complete remission after induction chemotherapy go on to consolidation chemotherapy with curative intent. However, even fit patients with Adverse- and Intermediate-risk AML are unlikely to be cured by chemotherapy alone, and therefore it is reasonable to consider allogeneic stem cell transplantation for Intermediate- and Adverse-risk patients upon achievement of first complete remission.
Why is AML so often resistant to chemotherapy? The biology of AML chemoresistance is complex. However, at a basic level, adverse-risk AML cells are more likely to evade conventional chemotherapeutics that target the cell cycle. It has therefore been hypothesized that one powerful driver of adverse prognosis in AML may be the properties of the leukemia stem cell (LSC), a type of cell that exhibits cell cycle quiescence, self-renewal, and chemoresistance.63 Although AML LSC remain challenging to isolate, assessment of AML LSC gene expression signatures has been proposed as a method to further refine prognosis – with LSC-like AML phenotypes contributing to adverse risk.
A study by Ng et al. recently defined a list of genes differentially expressed between LSC and non-LSC fractions (validated by xenotransplantation) from 78 AML patients.7 The list of genes highly expressed in LSC was subjected to statistical regression analysis to relate the expression profile to patients’ survival, which yielded an optimal “17-gene LSC score” prognostic signature. When the scoring algorithm was applied to five cohorts of AML patients, high scores consistently correlated with poor prognostic factors such as older age, higher initial white blood cell count, and unfavorable cytogenetics. High scores also correlated with resistance to standard induction chemotherapy, higher rates of relapse, and poor outcomes including inability to achieve complete response, decreased overall survival, and shorter event-free and relapse-free survival. Ng et al. proposed that this scoring tool could be applied to guide selection of initial therapy in newly diagnosed patients, specifically to identify high-risk patients not likely to benefit from standard induction chemotherapy.
In this issue of Haematologica, Bill et al. provide an impressive validation of the 17-gene LSC scoring system using RNA-sequencing data from a large number of patients treated in cooperative group (CALGB) trials.8 This work confirms and expands upon the insights published in 2016 by Ng et al., showing the prognostic value of LSC gene expression signatures in an independent large cohort of AML patients. Here, Bill et al. apply the 17-gene LSC score to 934 de novo AML patients and report the association of the 17-gene LSC score with prognostic clinical parameters, specific AML mutations, and ELN risk classification.
Using unsorted pre-treatment bone marrow and/or peripheral blood specimens, the group conducted transcriptome analysis via RNA-sequencing. The 17-gene LSC score was calculated as the weighted sum of the normalized expression values of the 17 genes included in the signature panel defined by Ng et al. The scores derived were then divided into two groups using the median as the cutoff to define “17-gene” (more LSC-like) and “17-gene” (less LSC-like) (Figure 1A, B).
Consistent with prior data, allocation into the 17-gene and 17-gene groups correlated with known prognostic factors. 17-gene patients were more often younger (age <60 years) and predominantly had favorable cytogenetic profiles. Bill et al. also correlated the 17-gene score with known AML mutations.8 Favorable mutations in genes such as CEBPA, GATA2, and KIT were more frequent in patients with a 17-gene score (Figure 1A) while unfavorable mutations in genes including ASXL1, RUNX1, and TP53 occurred more frequently in patients with a 17-gene score (Figure 1B). Patients with extremely high-risk EVI1 rearrangements inv(3)/t(3;3) were exclusively found in the 17-gene score group. With respect to mutation burden, more LSC-like AML harbored slightly more mutations, with a median of two among patients with a 17-gene score and of three among those with a 17-gene score.
Next, Bill et al. assessed outcomes in the groups with 17-gene and 17-gene scores. Both groups followed known associations for favorable and poor outcomes (complete remission rate, longer disease-free and overall survival) in, respectively, the younger (Figure 1C) and older cohorts of patients. In addition to validating the prognostic impact of the 17-gene LSC score in a large independent cohort and adding correlations with AML mutations, Bill et al. also compared the 17-gene LSC score to AML ELN risk stratification.8
When patients were classified according to the ELN stratification into Favorable-, Intermediate-, and Adverse-risk groups, there were significant differences in ELN risk distribution between the 17-gene and 17-gene LSC score patients of different ages. In younger patients with a 17-gene score, most (66%) were classified as having Favorable-risk, with 14% and 17% classified as having Intermediate- and Adverse-risk, respectively. However, younger patients with a 17-gene score were spread across the ELN classification: Adverse-risk (41%), Intermediate-risk (32%), and Favorable-risk (26%). In older patients with a 17-gene score, only 36% were classified in the Favorable-risk group, while 24% had an Intermediate risk and 40% an Adverse risk. By comparison, older patients with a 17-gene score clustered mainly into the Adverse-risk group (63%), with fewer in the Intermediate- (18%), and Favorable-risk (18%) groups. When assessing outcomes, the 17-gene LSC score failed to add significant prognostic information to ELN classification in older AML patients, in whom prognosis remains poor across prognostic groups with conventional chemotherapy.
Intriguingly, the data suggest that the 17-gene LSC score can provide additional prognostic value particularly for younger patients who may be currently misclassified as having a favorable risk. Younger patients with an ELN Favorable-risk classification with a high 17-gene LSC score (20% of ELN Favorable-risk patients) have a worse prognosis than would otherwise be expected from the ELN classification alone. This unexpectedly high-risk group of patients epitomizes the rationale for using refined prognostication schemes such as the 17-gene scoring tool, with the goal of tailoring first-line therapy more precisely and identifying populations of patients in need of prospective clinical trials.
The comprehensive RNA-sequencing approach described by Bill et al. does have some limitations. From a practical point of view, while pre-treatment cytogenetics as well as genomic profiling for mutations in specific genes have become standards of care for patients with AML, it is premature to recommend universal pre-treatment RNA-sequencing. Future studies in adult AML may validate the prognostic significance of pre-treatment profiling of a limited list of LSC-related genes using more targeted gene expression analysis, as was recently shown using Nanostring technology in pediatric AML.9
In a broader perspective, prognosis in any disease is shaped by the efficacy of available therapy. All patients evaluated in the current study by Bill et al. received cytarabine/anthracycline-based induction chemotherapy.8 Although AML prognosis has traditionally been evaluated in response to cytotoxic chemotherapy, the prognostic impact of ELN genetic risk classification and LSC gene expression signatures will need to be re-evaluated in the context of novel and more targeted therapeutics.
Recently, the BCL-2 inhibitor venetoclax in combination with hypomethylating agents has become a new standard of care for adult patients with AML who are unfit, by virtue of age or comorbidities, to receive intensive chemotherapy.10 Although many patients still relapse, this combination shows activity in disease often refractory to standard induction chemotherapy, including secondary AML, therapy-related AML, and AML with high-risk cytogenetic and mutation profiles. One explanation for the relatively mutation-agnostic efficacy of venetoclax + azacitidine is the combination’s suppression of oxidative phosphorylation and disruption of energy metabolism in LSC.11 The impact of LSC gene expression signatures on prognosis in patients treated with hypomethylating agents + venetoclax has yet to be determined. Similarly, the impact of LSC gene expression signatures on prognosis in FLT3-mutated patients may also need to be re-evaluated, as more effective and specific FLT3 inhibitors enter clinical practice.12 In general, as more effective therapies are developed that target the fundamental biology of AML, prognostic factors and even post-remission therapies will need to be re-examined.
- Liesveld J. Management of AML: who do we really cure?. Leuk Res. 2012; 36(12):1475-1480. PubMedhttps://doi.org/10.1016/j.leukres.2012.08.006Google Scholar
- Dohner H, Estey E, Grimwade D. Diagnosis and management of AML in adults: 2017 ELN recommendations from an international expert panel. Blood. 2017; 129(4):424-447. PubMedhttps://doi.org/10.1182/blood-2016-08-733196Google Scholar
- Thomas D, Majeti R. Biology and relevance of human acute myeloid leukemia stem cells. Blood. 2017; 129(12):1577-1585. PubMedhttps://doi.org/10.1182/blood-2016-10-696054Google Scholar
- Lapidot T, Sirard C, Vormoor J. A cell initiating human acute myeloid leukaemia after transplantation into SCID mice. Nature. 1994; 367(6464):645-648. PubMedhttps://doi.org/10.1038/367645a0Google Scholar
- Shlush LI, Zandi S, Mitchell A. Identification of pre-leukaemic haematopoietic stem cells in acute leukaemia. Nature. 2014; 506(7488):328-333. PubMedhttps://doi.org/10.1038/nature13038Google Scholar
- Pollyea DA, Gutman JA, Gore L, Smith CA, Jordan CT. Targeting acute myeloid leukemia stem cells: a review and principles for the development of clinical trials. Haematologica. 2014; 99(8):1277-1284. PubMedhttps://doi.org/10.3324/haematol.2013.085209Google Scholar
- Ng SW, Mitchell A, Kennedy JA. A 17-gene stemness score for rapid determination of risk in acute leukaemia. Nature. 2016; 540(7633):433-437. https://doi.org/10.1038/nature20598Google Scholar
- Bill M, Nicolet D, Kohlschmidt J. Mutations associated with a 17-gene leukemia stem cell score and the score’s prognostic relevance in the context of the European LeukemiaNet classification for acute myeloid leukemia. Haematologica. 2020; 105(3):721-729. PubMedhttps://doi.org/10.3324/haematol.2019.225003Google Scholar
- Duployez N, Marceau-Renaut A, Villenet C. The stem cell-associated gene expression signature allows risk stratification in pediatric acute myeloid leukemia. Leukemia. 2019; 33(2):348-357. Google Scholar
- DiNardo CD, Pratz K, Pullarkat V. Venetoclax combined with decitabine or azacitidine in treatment-naive, elderly patients with acute myeloid leukemia. Blood. 2019; 133(1):7-17. PubMedhttps://doi.org/10.1182/blood-2018-08-868752Google Scholar
- Pollyea DA, Stevens BM, Jones CL. Venetoclax with azacitidine disrupts energy metabolism and targets leukemia stem cells in patients with acute myeloid leukemia. Nat Med. 2018; 24(12):1859-1866. https://doi.org/10.1038/s41591-018-0233-1Google Scholar
- Perl AE, Martinelli G, Cortes JE. Gilteritinib or chemotherapy for relapsed or refractory FLT3-mutated AML. N Engl J Med. 2019; 381(18):1728-1740. Google Scholar