High expression of inositol 1,4,5-trisphosphate receptor, type 2 (ITPR2) as a novel biomarker for worse prognosis in cytogenetically normal acute myeloid leukemia.

Inositol 1,4,5-trisphosphate receptor, type 2 (ITPR2) is a key regulator for the activity of calcium ion transmembrane transportation, which plays a critical role in cell cycle and proliferation. However, the clinical impact of ITPR2 in cytogenetically normal acute myeloid leukemia (CN-AML) remained unknown. Several microarray datasets were used to evaluate the association between ITPR2 expression and clinical and molecular characteristics. ITPR2 showed a higher expression in CN-AML patients than normal persons. In a cohort of 157 CN-AML patients, high ITPR2 expression (ITPR2high) was associated with dramatically shorter overall survival (OS; P = 0.004) and event-free survival (EFS; P = 0.01), which were also shown in the European Leukemia Net (ELN) intermediate-I genetic category (OS: P = 0.0066; EFS: P = 0.009). Multivariable analyses adjusting for known prognostic factors confirmed ITPR2high to be associated with shorter OS (P = 0.0019) and EFS (P = 0.012). The prognostic value of ITPR2 was further validated in another cohort of 162 CN-AML patients (P = 0.007). In addition, first gene/microRNA expression signatures were derived that associated with ITPR2high on the genome-wide scale, which provided many indications to illustrate the possible mechanisms why ITPR2 could function. These results could aid to identify new targets and design novel therapeutic strategies for CN-AML patients.


INTRODUCTION
Cytogenetically normal acute myeloid leukemia (CN-AML) constituting 40-50% of all AML patients is the largest cytogenetic group [1,2]. It is characterized with the rapid accumulation of abnormal white blood cell in the bone marrow and the interface with normal blood cells' production without any chromosomal aberrations. CN-AML patients are usually categorized to be intermediate risk, yet their clinical and molecular characteristics are sharply heterogeneous. Finding proper markers has been an active area in order to obtain a more refined stratification and deeper understanding for CN-AML. Mutations of NPM1 [3] and CEBPA [4] have been used as favorably prognostic biomarkers, while FLT3-ITD [5] and MLL [6] mutations have been associated with worse prognosis in the European Leukemia Net (ELN) reporting system [7]. Also, a recent report combined mutations of DNMT3A, FLT3 and NPM1 to predict clinical features for CN-AML [8]. Besides, high expression of several genes/microRNAs have been associated with adverse prognosis, including BAALC [9], ERG [9], WT1 [10], MN1 [11], DNMT3B [12], TCF4 [13], miR-155 [14] and miR-3151 [15], while high expression of LEF1 [16] has been regarded as favorable prognostic factors. Because the mechanisms of leukemogenesis are still unknown, finding new prognostic biomarkers is critical for obtaining refined risk-stratification and designing novel therapeutic strategies of CN-AML.
Activity of calcium ion transmembrane transportation is a critical biological process for maintenance and regulation of cell cycle, and plays an important role in cell proliferation, differentiation and senescence [17]. Inositol 1,4,5-trisphosphate receptor, type 2 (ITPR2) is a an essential www.impactjournals.com/oncotarget regulator for mediating the mobilization of intracellular Ca 2+ stores, and acts as an pivotal role in intracellular Ca 2+ signaling in a variety of cell types. Early report showed that ITPR2 participated in Ca 2+ -calpain and Caspasemitochondria dependent pathways and regulated the apoptosis of U937 cell [18], also ITPR2 might act as target of CEBPB and all-trans retinoic acid (ATRA) in NB4 cells [19]. Recent reports identified ITPR2 variations as novel susceptibility loci for renal cell carcinoma via a genomewide association study [20]. Another report identified ITPR2 as a susceptiable gene for Kashin-Beck disease in Han Chinese [21]. However, the prognostic impact of ITPR2 expression has not been reported in CN-AML.
This manuscript provided consolidated evidence for the first time that, high ITPR2 expression (ITPR2 high ) was associated with worse prognosis in CN-AML. Firstly, ITPR2 high was shown in CN-AML patients compared to normal bone marrow (NBM) measured with microarrays. These microarray data was confirmed by qPCR [13,16]. Secondly, the prognostic value of ITPR2 was determined with 2 independent, relative large CN-AML cohorts, with respect to clinical, molecular characters and analysis of OS and EFS. Multivariable analysis further confirmed ITPR2 high as a worse prognostic marker. Finally, underlying mechanisms of why ITPR2 functioned as a worse prognosticator was investigated using microarray or high-throughput sequencing data of gene/ microRNA expression, genome-wide DNA methylation, combining with the known annotation and pathway information. These results may potentially facilitate our understanding of leukemogenesis and provide new criterions for risk-stratification in CN-AML, which will be finally exploited and lead to new treatment strategies.

ITPR2 expression discriminates between CN-AML and normal bone marrow
Microarray analysis was used to analyze expression alteration of ITPR2 in CN-AML (n = 116) and normal bone marrow (NBM) (n = 5). ITPR2 was positive in these two populations. However, there existed a significant difference, ITPR2 showed a remarkably higher expression in CN-AML than NBM (P = 0.014, Figure 1A and 1B). Higher expression in CN-AML made the detection of ITPR2 more easily, and discriminated patients from normal population, these two characteristics were important for clinical applications.

Association of ITPR2 expression with clinical and molecular characteristics
In the 157 de novo CN-AML cohort, we found that more ITPR2 high patients fell into M1 (P < 0.001) and M2 (P = 0.018), less fell into M4 (P = 0.044) and M5 (P < 0.001) in FAB subtype, compared to ITPR2 low patients. FLT3-ITD mutation occurred more in ITPR2 high (P < 0.001) than ITPR2 low patients. In addition, ITPR2 high patients seemed more likely to have high expression of ERG (P < 0.001), BAALC (P = 0.004), WT1 (P < 0.001) and DNMT3B (P < 0.001). Based on the European Leukemia Net (ELN) genetic categories, more Intermediate-I patients fell into ITPR2 high group (P < 0.001) ( Table 1).  Table 2). When all 157 patients were dichotomized according to the ITPR2 expression levels, we found that ITPR2 high group showed a significantly shorter OS (P = 0.0039, Figure 2A) and EFS (P = 0.01, Figure 2B) than ITPR2 low patient group.

Prognostic value of ITPR2 in ELN genetic subgroups
European Leukemia Net (ELN) divided CN-AML patients into the ELN favorable or ELN intermediate-I genetic categories, according to the mutation of CEBPA, NPM1 and FLT3-ITD [7]. In our analysis to the primary cohort of 157 CN-AML patients, expression of ITPR2 differed dramatically between different ELN genetic groups. More patients of the ELN Intermediate-I  Table 1). Thus, we further investigated the impact of ITPR2 expression within the two ELN genetic categories separately. ITPR2 expression showed no association with OS (P > 0.9, Figure 3A) or EFS (P > 0.9, Figure 3B) in the ELN favorable category. However, in ELN intermediate-I genetic category, ITPR2 high patients showed significantly shorter OS (P = 0.0066, Figure 3C) and EFS (P = 0.009, Figure 3D) than ITPR2 low patients.

Multivariable analysis of ITPR2 expression associated with OS and EFS
To further determine the prognostic value of ITPR2 expression, multivariable analysis was preformed after adjusting for the impact of other known risk factors, including commonly known mutations and several recently published prognostic factors such as ERG [22], BAALC [9], LEF1 [16] and WT1 [10]. In the multiple model for OS, ITPR2 high patients had 2.44 times increase of risk to death, other factors associated with longer OS included mutations of NPM1, CEBPA, and high   (Table 3).

Validation in an independent cohort of 162 CN-AML patients
An independent cohort of 162 de novo CN-AML patients was exploited to validate our findings. The third quartile (Q3) of ITPR2 expression was used as the cutoff. Patients with FAB M1 were more likely to have a higher expression of ITPR2 (P < 0.001), while FAB M5 seemed to be significantly associated with lower ITPR2 expression levels (P = 0.0015). Further, patients with higher expression of ITPR2 were more likely to have shorter OS (P = 0.011), and have higher expression of ERG, WT1, DNMT3B and TCF4 (All with P < 0.001) and lower expression of LEF1 (P = 0.004). In addition, mean OS showed significant difference between ITPR2 high and ITPR2 low groups (P = 0.011), and ITPR2 high patients seemed to have a shorter OS (P = 0.007, Figure S1.) Noticeably, mean age of ITPR2 low patients was older than that of ITPR2 high patients (P = 0.034), which consolidated the prognostic value of ITPR2 expression to some extent. (All statistics were listed in Table S1.)

Genome-wide gene-expression profiles associated with high expression of ITPR2
To further investigate the biological role of ITPR2, we performed a genome-wide differential analysis based on the sample division of median ITPR2 expression. 768 upregulated and 1136 down-regulated genes were identified dramatically associated with high ITPR2 expression (False Discovery Rate, FDR < 0.01; Fold Change, FC >= 1.5, data not shown). Among the 768 up-regulated genes, excitedly, we found WT1, ERG and DNMT3B, high expression of which were all clearly reported as worse prognostic biomarkers in CN-AML [9,10,12]. Other up-regulated genes included a cluster of genes that control cell cycle and differentiation (CDK6, CDKN1C, CCND2 et.al), several genes that function as tyrosine kinase (MAP4K3, PTK7 and c-Kit), and genes that previously known to be related with leukemogenesis (MLLT11, MPL, MYCN, MSI2). Noteworthy down-regulated genes included members of LILR family (LILRB1, LILRB3, LILRB4, LILRA1, LILRA2, LILRA6), members of TLR family (TLR2, TLR4, TLR5, TLR7) and the immune molecule CD86, which indicated the possible mechanisms of immune evasion that led to the worse outcome for CN-AML patients with high ITPR2 expression. (See Figure 4A and 4B.) In addition, cell signaling pathways that were associated with ITPR2 expression alteration were identified. Based on the pathway data provided by MSigDB [23], expressions of genes that participated in a pathway were averaged to represent expression of the pathway. We found that pathways of apoptosis, antigen processing and natural killer cells mediated cytotoxicity were all significantly down-regulated (all P < 0.001), associated with high expression of ITPR2. This result, just being consistent to previously presented dysregulated genes, possibly illustrated why high expression of ITPR2 was associated with adverse outcome in CN-AML. (See Table 4 for the statistics of those pathways.)

Genome-wide microRNA profiles associated with ITPR2 expression
To gain further insights into biological differences associated with varying ITPR2 expression levels, genomewide analysis of microRNA expression profiles were exploited using high throughput sequencing of 79 CN-AML patients from The Cancer Genome Atlas (TCGA) [24]. Sixty microRNAs were significantly associated with ITPR2 expression (P < 0.01, data not shown). Of these microRNAs, miR-155 and two members of miR-181 family(miR-181-a-1 and miR-181c) were positively associated with ITPR2 expression, the former of which has been validated to predict worse outcome for CN-AML patients [14], and the latter were recently shown to be potential targets and associated with adverse outcomes for AML patients [25]. MiR-193a was the most significant microRNA negatively associated with ITPR2 expression, which was reported to target c-Kit and up-regulation of miR-193a predicted favorable outcome in our group recently [26,27]. (See Figure 4C and 4D.)

Genome-wide methylation profiles associated with ITPR2 expression
Because ITPR2 expression was positively correlated with DNMT3B expression, differential analysis for methylation was performed to find different DNA methylation patterns at genome-wide scale and within main cell signaling pathways. But no clear difference was found significantly with respect to ITPR2 expression status at the whole genome-wide level and the 186 known pathways in MSigDB [23]. (Figure S2, pathway results were not shown.)

DISCUSSION
For the first time, we evaluated the prognostic value of ITPR2 expression, high expression of which was associated with shorter OS and EFS in two independent, large cohorts of de novo CN-AML patients. In our study, patients with ITPR2 high are significantly more classified in the M1 or M2 FAB subgroups than with ITPR2 low , suggesting that the leukemic cells of the ITPR2 high patients derive from relatively more immature cells. In addition, high ITPR2 expression was associated with the presence of FLT3-ITD and high expression of ERG, BAALC and WT1, which were all associated with worse outcomes. Besides, more patients of high ITPR2 expression  belonged to ELN intermediate-I group (P < 0.001) and less to ELN-favorable group, which also showed high ITPR2 expression as a worse biomarker. This will improve the ELN intermediate-I group risk classification and suggest that these patients may be turned candidates for alternative therapies. Furthermore, the association of high ITPR2 expression with shorter OS and EFS was confirmed in log-rank test and multivariable analyses adjusting for almost all known molecular prognosticators in CN-AML. Considering the fact that these two cohorts of CN-AML patients received uniformed therapeutic treatments separately, these results validated ITPR2 as an independent prognostic factor. Moreover, the fact that ITPR2 showed higher expression in CN-AML than normal bone marrow indicated its power of discrimination and easy access, which were important characteristics for clinical application. The mechanisms why high ITPR2 expression is associated with adverse treatment response are unknown. However, our exploration in genome-wide gene/microRNA expression analysis provided possible interpretations. We found that expression of WT1, ERG and DNMT3B were significantly associated with ITPR2, and several genes that were active in cell cycle and tyrosine kinase process were all up-regulated, while genes that function as immune factors are down-regulated. In addition, several important cell signaling pathways showed aberrant expression associated with high expression of ITPR2, including the significantly down-regulation of apoptosis, natural killer cell mediated cytotoxicity and antigen processing and presentation. These changes may lead to out of control for cell death and immune escape, which might contribute to worse outcomes.
The microRNA profiles associated with ITPR2 expression also made sense for the worse outcome. MiR-155 was found to be positively, while miR-193a to be the most negatively correlated with ITPR2 expression, because miR-155 was clearly confirmed worse prognosticator in CN-AML and miR-193a was validated as a favorable biomarker by our group.
Epigenetic regulation is an important mechanism that can alter the genomic expression, however we observed no significant association between ITPR2 expression and DNA methylation changes via our analysis of HELP array data. So, these CN-AML patients might show no sensitive to drugs eliminating methylation such as decitabine.
In conclusion, we show that high expression of ITPR2 is associated with shorter OS and EFS in CN-AML patients. ITPR2 shows higher expression in CN-AML than healthy persons, which indicates its easy access by qPCR and potential using in clinical application. However, future studies are needed to establish a standardized protocol of quantification, before it can be used for risk-stratification of CN-AML patients. Furthermore, the derived genomewide analysis of gene/microRNA expression and DNA methylation shed light on the underlying biologic mechanisms of leukemogenesis, and might help to develop new therapeutic strategies for CN-AML disease.

Patients and treatment
A cohort of 157 patients with previously untreated CN-AML (median age, 50 years, range: 16-77 years), who were collected at Erasmus University Medical Center (Rotterdam) between 1990 and 2008 [28], One hundred thirty patients (83%) were aged < 60 years (younger patients) and 27 patients (17%) were ≥ 60 years (older patients). The patients had been treated on study protocols of Dutch-Belgian Hematology-Oncology Cooperative Group (HOVON, http://www.hovon.nl). The detailed therapeutic protocol was shown in Figure S3. All samples were collected at diagnosis, with bone marrow aspirates or peripheral-blood, containing 80-100 percent blast cells after thawing [29]. Conventional cytogenetic examination of more than 20 metaphases from BM was used to determine the diagnosis of a normal karyotype. Patients were assessed for NPM1, CEBPA, IDH1, and IDH2 mutations, FLT3-internaltandem duplications (FLT3-ITD), N-RAS, K-RAS, and FLT3-tyrosine kinase domain mutations (FLT3-TKD [D835]). This research were approved by the institutional review boards at Weill Cornell Medical College and Erasmus University Medical Center, and all subjects provided written informed consent in accordance with the Declaration of Helsinki [30]. To validate our results, another independent cohort of 162 CN-AML patients was exploited, which was provided by the multicenter AMLCG-1999 trial of the German AML Cooperative Group between 1999 and 2003 [31]. These patients received intensive double induction and consolidation chemotherapy [31]. The AMLCG-1999 clinical trials were approved by the local institutional review boards, and informed consent from all patients obtained in accordance with the Declaration of Helsinki.

Microarray for gene expression and methylation, RNA/microRNA sequencing data
For the primary cohort of 157 CN-AML patients, pretreatment samples were studied using Affymetrix HG-U133Plus 2.0 expression GeneChips [28] and HELP methylation arrays [30], while the validating 162 patients were with Affymetrix HG-U133A expression GeneChips [31]. Experimental designs, quality control and normalization of data were carried out according to the standard Affymetrix protocols. Microarray data are available at the (GEO: accession no. GSE1159, GSE6891 and GSE12417 for expression, GSE18700 for methylation) including clinical, cytogenetic and molecular characteristics [32]. To further identify microRNAs correlatively expressed with ITPR2, RNA-Sequencing and microRNA-Sequencing data from The Cancer Genome Atlas (TCGA) were exploited [24], which provided 79 CN-AML patients. Pretreatment and clinical characteristics can be publicly downloaded from TCGA data portal (https://tcga-data.nci.nih.gov/tcga).

Statistical analyses
This study tried to evaluate the prognostic value of ITPR2 expression in CN-AML, and further explored the underlying mechanisms why it functioned, based on the gene/microRNA expression and methylation data. Samples were divided into two groups, high ITPR2 expression (ITPR2 high , n = 78) and low ITPR2 expression (ITPR2 low , n = 79), based on the median expression value of ITPR2. Also, high and low classifications of ERG, BAALC, WT1, LEF1, MN1, EVI1, DNMT3B and TCF4 were determined according to the median expression of corresponding genes. Pretreatment clinical and molecular characteristics were compared between ITPR2 high and ITPR2 low patients groups using the Fisher exact test for categories variables and the Wilcoxon rank-sum test for continuous variables. Association between ITPR2 expression and clinical outcomes was analyzed using Kaplan-Meier method, and difference was estimated with log-rank test. Multivariable Cox proportional hazards models were used to study the time-to-event factors associated with survival endpoints.
Differential analysis was conducted with Student's t-test with multiple hypothesis correction (False Discovery Rate, FDR), to identify genes and pathways whose expression or methylation levels were associated with ITPR2 expression. Pearson correlation test was performed to determine the correlated expression between ITPR2 and microRNA sequencing profiles. All analysis was performed on the platform of R 3.1.1 software package.