Background
Acute myeloid leukemia (AML) is a well-known heterogeneous hematological malignancy with a broad range of prognosis, which is greatly impacted by clinical factors, cytogenetics and molecular characteristics [
1]. In the current risk stratification system, recurrent genetic abnormalities stratify AML into three risk status categories including favourable-risk, intermediate-risk and poor-risk [
1]. This risk stratification system, together with clinical characteristics of AML patients such as age and medical comorbid, dictates the prognosis of each individual patient, as well as guide physicians to decide appropriate treatment regimens [
1]. For instance, in younger adult patients high-dose cytarabine-based therapy has been recommended as the conventional consolidation regimen for AML of favourable-risk, whereas allogeneic hematopoietic stem cell transplantation (HCT) for AML of poor-risk [
1]. Nevertheless, either regimen can be considered for AML of intermediate-risk [
1]. While patients have benefited greatly from current risk stratification strategies of AML, the prognosis of AML in each risk category is still quite variable. Further improvement of prognostic tools is needed to better stratify these patients and guide treatments accordingly. In addition to the well-established recurrent cytogenetic aberrations and molecular mutations, genes with aberrant expression at protein or mRNA level have also been shown to have significant prognostic values in AML over the last decade [
2]. These gene expression biomarkers not only help shed light on mechanisms of development and progression of this largely heterogeneous malignancy, but more importantly help clinicians to refine prognostic tools to improve patient care in clinical practice.
Elevated protein expressions of various cluster of differentiation (CD) marker genes such as
interleukin 2 receptor subunit alpha (
IL2RA/CD25) [
3‐
7],
C-
X-
C chemokine receptor type 4 (
CXCR4/CD184) [
8‐
10],
CD34 [
11,
12] and
CD56 [
13,
14] have been shown to predict poor clinical outcome in AML. Most studies used flow cytometry (FCM) technology to quantify protein expression of these CD markers. Meanwhile, many studies based on mRNA quantification platforms identified mRNA expression biomarkers in AML, such as
BAALC,
ERG,
MECOM/EVI1 and
WT1, which could offer additional prognostic values to improve current stratification system [
15‐
21]. RNA quantitative methodology, in comparison to FCM method, have several unique advantages. Firstly, stored RNA or bone marrow (BM) samples could be used to accurately quantify RNA gene expression when fresh samples are not available. This enhanced tissue flexibility would potentially improve patient care in practice. In addition, this approach makes retrospective analysis of samples feasible, which also enhance clinical care and facilitate research. Secondly, only limited tissue is needed for RNA quantitation that could significantly improves tissue efficiency. Lastly, many RNA quantitative platforms are designed to perform multiplex gene testing to improve consistency and efficiency.
By far, the prognostic values of mRNA expression of most CD markers in AML remain elusive. To our knowledge, the only prognostic CD marker, of which the mRNA expression prognostic value has been indicated, is CD34 [
22,
23]. However, reports on prognostic values of
CD34 mRNA and protein expressions were not consistent [
24‐
26]. In general, gene expression levels at mRNA level do not necessarily correlate well with those at protein level [
27,
28] which are subject to multiple layers of regulation [
27‐
29]. In addition, the protein levels of known CD biomarkers in AML are generally quantitated by FCM assay on blast cells, whereas mRNA expression levels of CD markers are quantitated in bulk tissue using different platforms. Therefore, it is important to investigate and validate the prognostic value of these CD biomarkers at mRNA level independently.
In our study, we initially sought to investigate the prognostic value of mRNA expressions of various CD biomarkers in AML and study if they can add independent prognostic value to the established prognostic factors. We conducted a pilot study to evaluate correlations of mRNA/protein expressions of four prognostic CD marker genes (IL2RA, CXCR4, CD34 and CD56), analyzed the prognostic values of their mRNA expressions in the TCGA-LAML cohort and determined IL2RA as best candidate gene for further study in larger cohort. Subsequently, in our clinical cohorts, we aimed to systemically evaluate the prognostic value of IL2RA mRNA expression in AML in the context of clinical and laboratory factors with prognostic relevance. We further characterized its prognostic role in core binding factor (CBF) AML in the context of established prognostic factors, as well as in intermediate-risk AML in the context of other mRNA expression prognostic factors (ERG, ID1, WT1, FLT3, WT1, BAALC, CDKN1B, MECOM/EVI1, MN1). Our study not only consistently reveals independent prognostic value of mRNA expression of IL2RA in AML, particularly in CBF and intermediate-risk AML, but also serves as a proof-of-concept study for future research endeavors to investigate prognostic roles of mRNA expression of other CD biomarkers.
Discussion
The prognostic value of CD biomarkers has been increasingly recognized in the AML research community over the last decade to help improve the current standard prognostic tools in AML clinical practice [
2,
5,
14]. These CD markers usually predict inferior clinical outcome in AML. Specifically, some of them were shown to offer additional adverse prognostic value to current stratification strategy, such as CD25 and CD56 in cytogenetically intermediate-risk AML [
5,
48] and CD56 in AML with t(8;21);
AML1/ETO or with t(15;17);
PML/RARA [
14,
49]. While the majority of these CD biomarkers were investigated at protein level using flow cytometry technology, the prognostic value of these CD biomarkers at mRNA level remains largely unknown. Our study is a first study that emphasized on prognostic value of mRNA expressions of CD marker genes in a systemic manner and we established that mRNA expression of
IL2RA gene, among a selected list of CD marker genes, is a significant and independent prognostic biomarker in AML, in particular, in CBF and intermediate-risk AML subtypes.
Our study quantified
IL2RA mRNA expression in 239 AML excluding APL and showed that
IL2RA mRNA is differentially expressed in groups classified by
FLT3-
ITD,
CEBPADM or
NPM1+
FLT3-
ITD− status, the recurrent genetic mutations that are prognostic relevant. Through a minimal p-value approach, we determined the optimal cutoff value for
IL2RA mRNA expression that facilitate its use as a prognostic tool. We found that high
IL2RA mRNA level is positively or negatively correlated with recurrent mutational and cytogenetic aberrations including
FLT3-
ITD,
NPM1+
FLT3-
ITD−, and t(16;21);
FUS/ERG, which was consistent with previous reports on CD25 [
3,
5]. Our results by uni- and multivariable analyses clearly showed that high
IL2RA mRNA expression was correlated with lower CR rate, shorter RFS and OS in AML and the effect is independent on other prognostic factors such as age, cytogenetic,
FLT3-
ITD and
c-
KIT D816V status. Therefore, we established the significant and independent role of high
IL2RA mRNA expression as an adverse prognostic factor in AML.
While
IL2RA could provided additional prognostic information to cytogenetics in AML, we further studied its prognostic value in AML of distinct cytogenetics risk statuses. CBF AML is a group of AML defined by carrying transcripts t(8;21);
AML1/ETO or inv(16);
CBF/MYH11. Although CBF AML belongs to the favorable-risk group, it still demonstrated considerable clinical, pathophysiological and molecular heterogeneity. In CBF AML, around 85% cases achieve CR after induction therapy, about 40–50% cases relapse after CR and around 50% cases remain alive at 5 years [
50,
51]. In previous studies, the prognostic value of CD25 protein expression in favorable-risk AML or CBF AML has not been reported probably due to the low frequency of CD25 positive cases in favorable-risk group in the cohorts [
5,
7]. Although the frequency of high expression
IL2RA cases was also low in APL (2 out of 63, Additional file
1: Table S8) and in all favourable-risk cases in our study, the frequency of high
IL2RA mRNA expression cases in CBF AML (16.7%) was not significantly different from that in intermediate-risk and poor-risk AML. Further survival analyses within this subgroup of patients demonstrated strong association between high
IL2RA expression and shorter RFS and OS, and it is the first time that elevated expression of CD25 biomarker gene are indicated in predicting adverse outcome in CBF AML.
Mutations in genes activating tyrosine kinase signaling (including
c-
KIT,
N/KRAS, and
FLT3) have been shown as the most frequent additional mutations in CBF AML that confer worse prognosis [
51]. Of these mutations, the
c-
KIT D816V has been indicated to predict poor prognosis, particularly in AML with t(8;21) [
51,
52]. Correlation analysis in our study in CBF AML showed no evidence of correlation between high
IL2RA expression and
c-
KIT D816V status (p = 1.000). Further uni- and multivariable survival analysis indicateded that high
IL2RA expression and
c-
KIT D816V mutation remained as two independent prognostic factors to predict shorter RFS, whereas high
IL2RA expression alone remained significant in predicting shorter OS in CBF AML. Moreover, high
IL2RA expression significantly correlated with
FLT3-
ITD mutation in CBF AML (p = 0.023), similar as in intermediate-risk AML. However, it should be noted that since
FLT3-
ITD mutation was present at a much lower frequency in CBF-AML (4.7% in our study and 7% in literature) than in intermediate-risk AML [
51], the interaction between high
IL2RA expression and
FLT3-
ITD mutation should be rather limited. It remained to be studied if the prognostic significance of high
IL2RA mRNA expression in CBF AML is associated with gene mutations involved in activating tyrosine kinase signaling other than
FLT3-
ITD or
c-
KIT D816V mutation, or with other mechanisms. Nevertheless, our results suggested the significant potential of high
IL2RA mRNA, in coordination with
c-
KIT D816V mutation, to further refine prognostification in CBF AML and warrant future validation in larger cohort of CBF-AML.
The prognostic value of
IL2RA or CD25 in intermediate-AML has been consistently shown in our study and previous studies [
5‐
7]. However, it was unknown if
IL2RA prognostic value is associated with or dependent on other gene expression prognostic biomarkers which has been well studied in intermediate-risk or normal karyotype AML. In our intermediate-risk AML cohort, a panel of prognostic gene expression markers including
BAALC [
15],
CDKN1B [
15],
ERG [
15,
18],
FLT3 [
53],
ID1 [
17,
47],
IL2RA, MN1 [
15,
19],
MECOM/EVI1 [
46], and
WT1 [
54,
55] were quantified by NanoString technology. Despite the strong correlations of
IL2RA expression with other mRNA biomarkers were shown,
IL2RA and
FLT3 mRNA expressions remained significant in predicting shorter RFS, whereas
ERG and
IL2RA mRNA expressions remained significant in predicting shorter OS by uni- and multivariable analyses. Independent analysis on the same panel of genes in intermediate-risk AML cases in TCGA-LAML cohort consistently demonstrated the significant and independent value of
IL2RA mRNA expression in predicting inferior clinic outcome. So far, various mRNA expression gene-panels have been designed for better stratification in AML [
22,
47] and our results support the incorporation of
IL2RA gene into such multi-gene panel to improve prognostification within intermediate-risk AML.
The mechanisms by which
IL2RA is prognostic in AML can be implicated by its correlation results with other mRNA expression biomarkers. It is first shown by our study that upregulation of
IL2RA was correlated with upregulation of
FLT3 tyrosine kinase transcripts [
53] and of transcription factor
ID1, a key common target of oncogenic tyrosine kinases that contribute to transformation of leukemias [
56], indicating a strong association of
IL2RA expression with tyrosine kinases pathways. Our study also showed significant correlation of
IL2RA with
ERG and
CDKN1B, the expressions of which have been indicated in stem cell-like featured gene signatures [
57,
58]. Previously, it has been proposed that CD25 is a surrogate marker for leukemia stem cell (LSC) [
4,
5], our results again provided evidence that
IL2RA/CD25 is involved in the crosstalk of LSC related signalling and support that
IL2RA is an indicator of the LSC signature which has been shown as a fundamental adverse prognostic feature in AML [
59]. In addition to these cell autonomous mechanisms, there may be non-cell autonomous mechanisms that contribute to the prognostic role of
IL2RA mRNA expression such as
IL2RA expression regulatory T cells [
60] or IL2/IL3 interplay [
61], which warrant future studies.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.