Expression Pattern and Prognostic Significance of EVI1 Gene in Adult Acute Myeloid Leukemia Patients with Normal Karyotype
Irena Marjanovic1 • Teodora Karan-Djurasevic1 • Tatjana Kostic1 • Marijana Virijevic2,3 • Nada Suvajdzic-Vukovic2,3 • Sonja Pavlovic1 • Natasa Tosic1
Abstract According to current criteria, patients with acute myeloid leukemia with normal karyotype (AML-NK) are classified as intermediate risk patients. There is a constant need for additional molecular markers that will help in substratification into more precise prognostic groups. One of the potential new markers is Ecotropic viral integration 1 site (EVI1) transcriptional factor, whose expression is dis- sregulated in abnormal hematopoietic process. The purpose of this study was to examine EVI1 gene expression in 104 adult AML-NK patients and on 10 healthy bone marrow donors using real-time polymerase chain reaction method, and to evaluate association between EVI1 expression level and other molecular and clinical features, and to examine its potential influence on the prognosis of the disease. Overexpression of EVI1 gene (EVI1+ status) was present in 17% of patients. Increased EVI1 expression was predomi- nantly found in patients with lower WBC count (P = 0.003) and lower bone marrow blast percentage (P = 0.005). EVI1+ patients had lower WT1 expression level (P = 0.041), and were negative for FLT3-ITD and NPM1 mutations (P = 0.036 and P = 0.003). Patients with EVI1+ status had higher complete remission rate (P = 0.047), but EVI1 expression didn’t influence overall and disease free survival. EVI1 expression status alone, cannot be used as a new marker for more precise substratification of AML-NK patients. Further investiga- tions conducted on larger number of patients may indicate how EVI1 expression could influence the prognosis and outcome of AML-NK patients, by itself, or in the context of other molecular and clinical parameters.
Introduction
Adult acute myeloid leukemia with normal karyotype (AML-NK) represents the largest group of adult AML patients classified as intermediate risk [1]. Still, their out- come, as well as their clinical and molecular characteristics are very heterogeneous, implying to the need of sub- stratification of these patients into more precise risk groups. Some molecular features, like the presence of mutations in NPM1, CEBPA gene (bialelic) or RUNX1 gene, have been incorporated into revised World Health Organization (WHO) classification of myeloid neoplasms and acute leukemia [2]. Furthermore, there is an ongoing investiga- tion about the prognostic impact that expression level of some individual genes might have [3]. Ecotropic viral integration 1 site (EVI1) gene located on chromosome 3q26, encodes complex transcriptional factor. It can act as a co-activator, or as a co-repressor, reacting to numerous stimuli [4]. This transcriptional factor is neces- sary for normal hematopoietic development with its expression level low, but still detectible in normal bone marrow cells [5]. It was found that EVI1 expression is at its highest level in the least differentiated cells (CD34+- CD38- cells), rapidly declining in more mature cells [6]. In in vitro experiments, overexpression of EVI1 in transformed myeloid progenitors blocks myeloid differen- tiation, and affects survival and proliferation of these progenitors, in that way exhibiting its leukemogenic potential [7, 8].
EVI1 gene is characterized by the presence of several 5‘ splice variants resulting from the use of alternative transcriptional initiation sites and alternative splicing. Also, numerous reports have found an intergenic splice variant (MDS1/EVI1) in both normal and leukemic cells [4, 9]. In AML, aberrant expression of EVI1 is present in 8–10% of cases. One-third of high EVI1 expressing AML patients are characterized by the existence of 3q26 rear- rangements, like inv(3)(q21:q26) and t(3;3)(q21:q26). Also, about 20% of patients with 11q23 rearrangements show increased EVI1 expression [10–12]. In patients without these aberrations, EVI1 overexpression can also be detected but in fewer cases [13–15]. In many studies, high EVI1 expression is a feature of aggressive leukemia, while in intermediate risk AML, like AML-NK, its adverse effect on prognosis is not so straightforward [10, 11, 16]. In this study we have analyzed the expression pattern of EVI1 in de novo AML-NK patients. Also, we evaluated association between EVI1 expression level and other molecular and clinical features, and examined its potential influence on the prognosis of the disease.
Materials and Methods
Patients and Therapy Protocol
Bone marrow samples from the 104 AML-NK patients at diagnosis were collected. All of the patients came from the Clinic of Hematology, Clinical Center of Serbia. Research was conducted in accordance with the ethical standards of the World Medical Association’s Declaration of Helsinki. The study was approved by the Ethics Committee of the Clinical Center of Serbia, and written informed consent was obtained for all patients. All patients received induction and consolidation chemotherapy with daunorubicin and cytarabine according to the protocol 3 + 7 (Ara-C 200 mg/m2 for 7 days in continuous infusion and daunorubicin 45 mg/m2 for 3 days (cycle I), followed by a second cycle of protocol ADE in which etoposide 200 mg/m2 was added up for 5 days. After the ADE protocol, the patients received two courses of consolidation therapy of MACE (amsacrine, cytarabine, etoposide) and MIDAC (mitozantrone, cytarabine) according to the MRC10 protocol.
Gene Expression and Mutational Analyses
Patient’s bone marrow mononuclear cells (BMMCs) as well as BMMCs from 10 healthy controls (bone marrow donors), were purified on Ficoll-PaqueTM Plus (GE Healthcare) density gradient, suspended in TRI Reagent (Ambion) and total RNA was extracted. One microgram of total RNA was used for the cDNA synthesis using Rev- ertAid Reverse Transcriptase (Thermo Scientific). Real time-PCR was performed by 7900 HT Fast Real-Time PCR System (Applied Biosystems). We have performed PCR using 1 ll of cDNA (50 ng RNA equivalent) with
TaqMan® Universal Master Mix II (Applied Biosystems), and EVI1 primers and probe [11]. This combination of primers and probe enables detection of all EVI1 splice variants. ABL gene was used as endogenous control [17]. All samples were run in duplicates. Relative quantification analysis was performed using comparative ddCt method, using healthy controls as calibrator, meaning ddCt = dCtsample – dCthealthy control (medijan). ROC curve analyses using on-line program ‘‘Cut-off finder’’ [18] was applied for the definition of the minimum expression level above which patient sample is considered to be positive for EVI1 expression. WT1 gene expression level as well as mutational anal- yses of FLT3, NPM1, and IDH1 and IDH2 genes were also analyzed as previously described [19–22].
Statistical Analysis and Definition of Clinical Endpoints
Data are presented as medians with range, mean ± SD, or as absolute numbers with percentages. Differences in continuous variables were analyzed using Mann–Whitney U test for distribution between 2 groups. Analyses of fre- quencies were performed using the v2 test for 2 × 2 tables or the Fisher exact test for larger tables. Survival probabilities were estimated by the Kaplan–Meier method, and differences in survival distributions were evaluated using the Log-Rank test. Overall survival (OS) was calculated from the first day of therapy to death or last visit. Patients undergoing hematopoietic stem cell transplantation (HSCT) were censored at the time of transplantation (12 patients under- went HSCT). Disease free survival (DFS) for patients who had achieved CR was measured from the date of CR to relapse/death/last follow-up. The statistical analyses were performed using the SPSS computer software 21.0 (IBM). For all analyses, the P values were 2-tailed, and P \ 0.05 was considered sta- tistically significant.
Results
EVI1 Expression Levels and Clinical Characteristics
The expression level of EVI1 gene was evaluated in a cohort of 104 newly diagnosed AML-NK patients, and in healthy controls. The median expression level of EVI1 transcript in AML-NK patients at diagnosis was 0.06 (range 0.00–60.50), which was significantly different compared to the expression level found among healthy controls (median 1.00 ± 0.59, range 0.06–2.18; P = 0.001) (Fig. 1). We used ROC curve analysis to calculate ‘‘cut- off’’ value for the discrimination between AML patients and healthy controls. The EVI1 expression level of 0.51 was determined as the most predictive one (AUC = 0.82, Specificity = 82.7%, Sensitivity = 90%, P \ 0.001). Using this value, 104 de novo AML patients were divided into EVI1 positive (EVI1+) and EVI1 negative (EVI1-) group. At diagnosis 17% of patients (18/104) were EVI1+ (Table 1). Analyzing the clinical characteristic of EVI1+ patients, we detected that the presence of EVI1 expression is strongly correlated with lower WBC count (P = 0.003), lower LDH levels (P = 0.005), and lower bone marrow blast percentage (P = 0.005). EVI1 positive patients expressed CD34 antigen (P = 0.016), and majority of them were detected in the M4 and M0/M1 FAB subgroups (P = 0.048) (Table 1).
EVI1 Expression Level and Other Molecular Markers
We have examined the association between the EVI1 expression level and other molecular markers, such as FLT3, NPM1 and IDH1 and IDH2 mutational status, as well as WT1 expression level. The presence of EVI1 expression was significantly associated with the absence of FLT3-ITD mutations (P = 0.036), and NPM1 mutations (P = 0.003). More precisely, EVI1+ status was predomi- nantly found in FLT3-ITD-/NPM1- ‘‘double negative’’ group with intermediate prognostic risk (ELN classifica- tion) (P = 0.011). Only 2 patients with EVI1 expression had IDH1/IDH2 mutations, and also, EVI1+ patients dis- played low WT1 expression level (P = 0.041) (Table 1).
Prognostic Significance of EVI1 Expression Level at Diagnosis
In our cohort of AML-NK patients 61% of them experi- enced CR. EVI1+ patients had even higher CR rate (82%), which was significantly superior compared to the EVI1- group (56%) (P = 0.046) (Table 1). None of the EVI1+ patients experienced early death, and refractory disease. Contrary to this, EVI1+ patients were more prone to the relapse of the disease (11/14 EVI1+ patients in CR relapsed, P = 0.058). However, Kaplan– Meier survival analysis showed no statistically significant Fig. 1 Box-plot chart of relative EVI1 expression level detected in control subjects and in AML-NK patients difference in DFS duration between EV1+ and EVI1- patients (12 vs. 15 months median, Log-Rank = 0.429, P = 0.512) (Fig. 2a). Also, we didn’t detect any statistically significant difference in OS time between EVI1+ and EVI1- patients (14 vs. 5 months median, Log- Rank = 0.765, P = 0.382) (Fig. 2b).
Fig. 2 a Kaplan–Meier analysis DFS of AML-NK patients according to prognostic group and EVI1 expression status. There were no significant difference between DFS of EVI1- patients (n = 48, 24 censured), and EVI1+ (n = 14, 3 censured); 15 months versus 12 months (v2 = 0.429, Log-Rank = 0.429, P = 0.512). b Kaplan– Meier analysis OS of AML-NK patients according to prognostic group and EVI1 expression status. There were no significant difference between OS of EVI1- patients (n = 86, 17 censured), and EVI1+ (n = 18, 2 censured); 5 months versus 14 months (v2 = 0.765, Log-Rank = 0.765, P = 0.382)
Discussion
Acute myeloid leukemia with normal karyotype (AML- NK) is a very large, genetically and clinically heteroge- neous subtype of AML, in constant need for discovery of additional molecular markers that can be used for prog- nosis and risk-stratification purposes. In this study we have analyzed the expression pattern and prognostic significance of new molecular maker, EVI1 gene, in 104 AML-NK patients.
We have found that EVI1 expression was present in 17% of patients at diagnosis. In previously published studies, of which only a few have been dedicated entirely to AML-NK subgroup, EVI1+ patients were present in 0.1–25% of cases [10, 11, 13–15, 23]. This wide range of frequency of EVI1+ patients can be attributed to different approaches that were used in determining the ‘‘cut-off’’ values for EVI1 expression, which varied from study to study. Namely, in our study we have decided to use the ‘‘cut-off’’ value that clearly discriminated between healthy and AML samples. Although this approach has been applied in other surveys [13, 16, 23], in a larger number of studies the ‘‘cut-off’’ value has been chosen arbitrarily between multiple values [10, 11, 15].
In recent years, much attention has been directed towards identification of the mechanism underlying EVI1 deregulation in AML. In an in vivo study done on mouse models, researchers have found that EVI1 performs its oncogenic role by totally altering cellular metabolism, causing suppression of erythropoiesis and lymphopoiessis, shifting differentiation process and causing an increase in expansion of preleukemic myeloid cells [24]. Therefore, targeted therapy against EVI1 could have major benefits for EVI1 expressing AML patients. There are ongoing studies related to such therapy and upon completion of clinical trials, their application in the treatment protocols of AML- NK patients may be expected [25]. To date, there is still ongoing discussion about the exact definition of aberrant EVI1 expression. As it is the case with some other genes that are expressed in healthy bone marrow as well as in leukemia cells, the ‘‘normal’’ value of EVI1 expression in AML patients is quite obscure. EVI1 is one of the markers of hematopoietic stamens with temporal pattern of expression restricted to CD34+ cells [26]. This was confirmed in our study where we detected strong association of EVI1+ and the presence of CD34 marker. In addition, our study showed that large proportion of AML- NK patients didn’t have detectable EVI1 expression level, or it was extremely low (\ 0.01).
Such EVI1 expression was detected in nearly 23% of patients. This was also observed in other studies, but the explanation for this finding is still lacking, mainly because the exact mecha- nism of abnormal EVI1 expression in AML-NK leukemia is not known [16, 27]. This mechanism is well character- ized in other subtypes of AML, like AML with 3q26 or with 11q23 chromosomal rearrangements [28, 29]. For the
AML-NK leukemia it is assumed that aberrant EVI1 expression is a consequence of histone modification mechanism [30]. There is also another possibility for ele- vated EVI1 expression level in AM-NK patients like the existence of cryptic 3q26 rearrangements. Namely, rear- rangements of this region could be missed during classical cytogenetic analysis, and thus, patients with 3q26 aberra- tions may be misclassified into those with a normal kary- otype. Yet, based on the few studies published, the percentage of these misclassified patients is about 1% in AML-NK patients [31, 32].
In the AML-NK patients, mutational status of NPM1 and FLT3 gene has great prognostic value. In this regard, we have analyzed the
association between the presence of these mutations and the level of EVI1 expression. As in previously published studies, we have also detected sig- nificant association between EVI1+ status and FLT3-ITD-/ NPM1- status [11, 13, 14, 27]. Patients belonging to this ‘‘double negative’’ group have intermediate prognosis according to ELN classification, and would benefit from additional molecular prognostic marker. Unlike our previ- ous study in which overexpression of WT1 in FLT3-ITD-/ NPM1- group of patients had affected their prognosis, the EVI1 expression level could not be used for additional risk stratification [19].
By examining the association between WT1 and EVI1 expression status, we have found that low WT1 expression was associated with EVI1+ status, but the combination of these two molecular markers had no effect on the prognosis and outcome of the disease either in the ‘‘double negative’’ group, or in the entire cohort of analyzed AML-NK patients. It is plausible that other molecular markers have stronger impact on the prognosis of these patients [27, 33–35].
In our cohort of AML-NK patients, EVI1 status had the most prominent impact on CR rate. Contrary to other published results, our EVI1+ patients had higher CR rate compared to EVI1- [11, 15, 16]. It should be emphasized that in the given studies the whole group of intermediate risk patients were analyzed, not only AML-NK patients. Also, our results on the prognostic impact of EVI1 expression on DFS and OS differed from the previously mentioned studies. In fact, we did not find a significant association between EVI1+ status and these prognostic parameters. In our cohort of AML-NK patients high EVI1 expression was not an independent factor of adverse prognosis, moreover EVI1+ patients had longer OS, but without statistical significance.
In conclusion, our study has shown that EVI1 expression status alone, cannot be used as a new marker for more precise substratification of AML-NK patients. Further investigations conducted on larger number of patients may indicate how EVI1 expression could influence the prognosis and outcome of AML-NK patients, by itself, or in the context of other molecular and clinical parameters.
Acknowledgements This work was funded by the Ministry of Edu- cation, Science and Technological Development, Republic of Serbia (Grant Number III 41004).
Author’s Contributions IM, TKD and TK performed experimental work and analysis; MV and NSV contributed patients samples and data; SP designed the study and contributed to the finalization of the manuscript; NT FLT3-IN-3 conducted the statistical analysis, wrote the manu- script. All of the authors approved the submitted version.