Modelling the Effects of MCM7 Variants, Somatic Mutations, and Clinical Features on Acute Myeloid Leukemia Susceptibility and Prognosis

The main objective of the study was to evaluate the associations between MCM7 rs2070215, rs1527423, and rs1534309 single nucleotide polymorphisms (SNPs) and acute myeloid leukemia (AML) risk and prognosis. The secondary objectives were to assess if any relationships existed between the mentioned SNPs and FLT3, DNMT3A, NPM1 mutations with clinical outcomes and overall survival (OS) in AML patients. We investigated 281 AML cases and 405 healthy subjects. The results showed a significant association between a variant allele of rs2070215 (p = 0.007), CAT haplotype (p = 0.012), and AML susceptibility. No significant association was found between MCM7 variant genotypes and overall survival of AML patients (p > 0.05), while several associations between somatic mutations, clinical and biological features, and poor OS were noticed. Lactate dehydrogenase (LDH) level ≥ 600 IU/L had a significant effect on the hazard of death (p = 0.004, HR = 1.49, 95% CI: 1.13–1.95). Our study showed that the variant allele of rs2070215, in the allelic model, and CAT haplotype were associated with AML susceptibility. The investigated FLT3, DNMT3A, and NPM1 mutations were associated with the clinical and biological features and poor OS. LDH level ≥ 600 IU/L was associated with an increased hazard of death and this association remained significant when quantifying for effect modification by FLT3 mutation status.


Introduction
Acute myeloid leukemia (AML) is characterized by a phenotypic and genetic heterogeneity. Several chromosomal rearrangements, somatic mutations, and epigenetic aberrations have been described to be implicated in the AML susceptibility, pathogenesis, prognosis, evolution, overall survival (OS), or in the treatment response of AML patients [1][2][3][4][5].
Recurrent structural and numerical chromosomal aberration and molecular markers, such as Fms like tyrosine kinase 3 (FLT3), Nucleophosmin 1 (NPM1), and DNA methyltransferase 3 alpha (DNMT3A) mutations, were proved to have an impact on AML treatment response and prognosis [6][7][8][9][10][11]. High lactate dehydrogenase (LDH) value or hyperleukocytosis has been reported to be independent of poor prognostic factors in AML [12][13][14]. Associations between hyperleukocytosis and somatic mutations have also been reported [12]. The available biomarkers may explain the poor prognosis for just a part of all AML cases; therefore, it was supposed that other new markers might be involved [15], such as minichromosome maintenance complex component 7 (MCM7) gene variants, as they have been reported to be associated with relapse in pediatric and adult AML patients and their overall poor survival rate [9,15,16].
MCM7 gene encodes an essential protein (Mcm7) of the mentioned complex with an important role in the control of cell cycle progression and initiation of DNA replication [15,16]. Mutations or single nucleotide polymorphisms (SNPs) that may affect DNA replication are a key step in the progression of many cancers [17,18]. DNA replication is an essential process for the survival; therefore, deficient DNA replication (or cell cycle) may affect the integrity of the genome (genomic instability is known to be a specific feature of cancer), becoming predisposed to different types of neoplasms, including hematological malignancies [19]. Thus, we may hypothesize that the MCM7 gene may be a candidate marker involved in the development of various types of neoplasms [15].
Overexpression of this gene was correlated (in the case of leukemia) with a negative prognosis, patients having multiple relapses, and a short overall survival [15,24], whereas suppression of MCM7 gene expression was proposed as a new potential therapy for leukemia [17].
Different SNPs were reported to be associated with overexpression of the MCM7 gene [25,26]. For example, the variant allele of rs2070215 increased the expression level of MCM7 in T cell leukemia cell line and prostate cancer, resulting in increased cell proliferation and relapse of prostate cancer [15,25,26].
Previously, one study indicated that MCM7 rs2070215, rs1527423, and rs1534309 SNPs might be used as prognostic markers for AML patients [15]. According to Ensembl genome browser (ensembl.org), rs2070215 consists of substitution of T with C, which leads to a missense variant, being a gain-of-function SNP, while rs1527423 and rs1534309 are intronic SNPs of the MCM7 gene. No data were available regarding the association between MCM7 SNPs and FLT3, DNMT3A, and NPM1 mutations.
The main objective of this case-control study was to evaluate the relationship of SNPs of the MCM7 gene, namely rs2070215, rs1527423, and rs1534309, with AML susceptibility and overall survival of AML patients. Subsequently, associations between the mentioned SNPs, somatic mutations such as FLT3 ITD, FLT3 D835, DNMT3A (codon R882), and NPM1 (4-bp dup/ins, rs587776806, rs1554138188, rs1554128189), and clinical features of AML patients were evaluated.
This study was approved by the Ethics Committee of the Clinical and Emergency Hospital of Târgu Mureş, Romania (10665/2019), and was performed in accordance with the Declaration of Helsinki in order to respect the ethical principles. All subjects included in this study, signed a written informed consent form.

Genotyping Investigation
DNA was extracted from leukocytes using the PureLink Genomic DNA kit (Thermo Fisher Scientific, Waltham, MA USA) according to the recommendations of the manufacturer. In cases with low white blood cell (WBC) count, blood quantity and protocol were adjusted.
Wild-type (reference) and variant alleles (alternatives) of mentioned SNPs and somatic mutations were indicated according to the Ensembl genome browser.

Statistical Analysis
Descriptive statistics as arithmetic mean ± standard deviation or absolute (number of cases) and relative frequencies (%) were used for describing genetic and clinical variables. The difference in distribution of the studied SNPs genotypes and alleles among AML patients and controls was tested using Chi-square (χ 2 ). The bivariate associations between studied MCM7 SNPs, somatic mutations (FLT3, NPM1, DNMT3A), combined somatic mutations (defined by "nonmutated", "single-mutated", and "double-/triple-mutated" status), and demographic or clinical variables were tested using the Chi-square (χ 2 ) or Fisher exact test. In the case of a significant association between combined mutations in FLT3/DNMT3A, FLT3/NPM, and NPM1/DNMT3A or combined mutations in FLT3/NPM1/DNMT3A and clinical variables, a post hoc test using a pairwise Fisher's exact test was performed, and adjusted p-values (pFDR) were reported in order to control false discovery rate (FDR).
Unconditional binomial logistic regression with main effects was used to assess the effect of investigated MCM7 SNPs on the AML risk, using R software (version 3.6.0) [27]. We applied the Benjamini-Hochberg FDR procedure to the p-values of our family of genetic models (additive, dominant, and allelic) in order to account for multiple comparisons in the case of significant results.
The nonparametric Kaplan-Meier method with the Log-Rank test was performed for evaluating differences between survival rate for all potential predictors (somatic mutations, clinical factors). The Cox PH regression was used to assess the effects of MCM7 variant genotypes and somatic mutations on survival time. The effect size was measured by crude Hazard ratios (HR) and adjusted HR for age group, gender, Eastern Cooperative Oncologic Group Scale (ECOG) performance status, cytogenetic risk group, type of AML, combined mutations (double-or triple-mutated versus one single-or nonmutated) in FLT3/NPM1/DNMT3A. An estimated p < 0.05 was set as the level of statistical significance, and all statistical tests were two-sided. Linkage disequilibrium and Hardy-Weinberg equilibrium in AML cases and control groups were also determined by a "genetics" R package.
The haplo.stats R package (version 1.7.9) was used to test the association between statistically inferred haplotypes and AML risk [28,29]. The method implemented in this package assumed that haplotypes were ambiguous because of an unknown linkage phase of the genetic variants, and in addition, it could estimate the crude and adjusted magnitude of the effect of each haplotype on AML risk (crude OR and adjusted OR) with associated 95% confidence interval (CI).

Description of AML and Control Groups
AML cases and control groups were similar in age (p = 0.484) and gender distribution (p = 0.348). According to European Leukemia Net (ELN) 2017 risk stratification, 55 (19.6%) of our patients were included in the low-risk category, 146 (52%) in the intermediate, 71 (25.3%) in the high-risk category, and for 9 (3.2%), we were unable to determinate the risk category. Regarding treatment and the response to therapy of our AML patients, we summarized that 122 (43.4%) were treated with low-dose (LD) therapy, 143 (50.9%) with high-dose (HD), and 16 (5.7%) with a combination of high-dose therapy and hematopoietic stem cell transplantation (HD + HSCT). The majority of them showed resistance to treatment (71, 25.3%) or did not respond and deceased a short time after diagnosis (69, 24.6%). On the other side, 43 cases (15.3%) had a complete remission (CR) without relapse, 49 (17.4%) had a partial remission (PR), and 49 (17.4%) relapsed after CR or PR.

MCM7 SNPs rs2070215, rs1527423, and rs1534309 and AML Risk
The genotype frequency for all three studied SNPs was consistent with Hardy-Weinberg equilibrium in AML cases and control groups (p > 0.05). The MCM7 rs1534309 and rs1527423, rs2070215 and rs1527423, rs1534309 and rs2070215 SNPs were in linkage disequilibrium in both groups (p < 0.001).
There was no significant difference in genotype distribution for any of the three studied SNPs between AML patients and controls (Table 1). Regarding allelic model, we noticed that only variant C allele of rs2070215 had a significantly different distribution between groups (p = 0.007), observing a higher frequency of this allele in AML patients (197 cases, 32% vs. 207 controls, 25.6%). We also investigated the distribution of MCM7 haplotypes between AML patients and controls (Table 2). Eight different haplotypes were obtained, of which 2 haplotypes with frequency <0.4% among AML patients or controls were not included in the estimation of AML risk because of their rarity. When haplotype 4 (GGT) was the considered reference, haplotype 1 (GAT) had a protective effect that remained significant after adjusting for conventional covariates like age and gender (OR = 0.71, 95% CI: 0.52-0.97). In the case of haplotype 6 (CAT), there was a significant positive association with AML risk, and it remained an independent risk factor after adjustment for age and gender of the patients (OR = 9.55, 95% CI: 1.13-80.40). In Table 3 we illustrated the distribution of wild-type and variant genotypes of the investigated SNPs according to the clinical features of the patients. No significant association was observed between the genotypes and investigated features (such as WBC count, platelet (PLT) count, hemoglobin (Hgb), LDH level, blast percentage, AML subtypes, somatic mutations, treatment, and toxicity). In the case of rs2070215 and rs1527423 SNPs, distribution of variant genotypes differed between genders, the variant genotypes being more common in male AML cases (p = 0.026 and p = 0.05, respectively).

Aditional Analysis
We also investigated if the presence of FLT3 somatic mutations modified the effect of LDH ≥ 600IU/L and WBC count ≥ 10,000 cells/mm 3 on the hazard of death in AML patients. An unadjusted analysis of interaction between LDH, WBC, and FLT3 mutations was performed for study changes in the hazard of death, these models including only FLT3, LDH, or WBC as independent variables and the interaction terms. LDH level ≥600 IU/L had a significant positive effect on the hazard of death (p = 0.004, HR = 1.49, 95% CI: 1.13-1.95). We showed that hazard of death for patients with FLT3 gene mutation and LDH level ≥ 600 IU/L was about two times higher than in patients without FLT3 gene mutation with LDH level < 600 IU/L (HR = 1.80, 95% CI: 1.22-2.66), while an increased hazard of death was also observed for patients with LDH level ≥ 600 IU/L within FLT3 strata (patients with no FLT3 mutation: HR = 1.50, 95% CI: 1.10-2.03; patients carrying FLT3 mutation: HR = 1.15, 95% CI: 0.60-2.20). The multiplicative interaction between LDH and FLT3 mutations, measured on a relative scale, was HR interaction = 0.80, 95% CI: 0.39-1.62, so although we noticed that the estimated joint effect of LDH level ≥ 600 IU/L with the presence of FLT3 mutations was lower than estimated effect of LDH level ≥ 600 IU/L without FLT3 mutations, the interaction effect was not statistically significant (p = 0.527).

Discussion
Previous studies suggested that MCM7 SNPs could be used as prognostic markers for cancer [15,25,26,32]. In the present study, the association between the mentioned SNPs (rs2070215, rs1527423, rs1534309) of the MCM7 gene and AML susceptibility were evaluated by analyzing the distribution of genotypes, alleles, and haplotypes between groups. According to our results, the variant allele (C) of the rs2070215 represented a risk factor for AML. It was reported that the variant allele of rs2070215 increases the expression of MCM7, therefore leading to cell proliferation, increasing the relapse rate and influencing cancer prognosis [25,26]. Our results seem to be similar to those reported in the literature regarding the risk of cancer for rs2070215 [21], but show no association between investigated MCM7 SNPs and AML prognosis, OS, relapse rate, or somatic mutations. The frequency of rs2070215, rs1527423, and rs1534309 observed in our study was similar to that reported by the Ensembl genome browser. Our findings, regarding the relapse frequency or OS, therefore partially contradict those reported earlier by Lee et al. [15]. One explanation for the different results is the number of investigated patients (we had a threefold larger AML group), and another is the ethnic origin (European as opposed to Asian).
However, in haplotype analysis, we noticed two opposite associations between AML risk and haplotype 1 (GAT, negative association, being a protective factor) and haplotype 6 (CAT, positive association, being a risk factor). The C and G alleles of the rs1534309 are the two different alleles of the mentioned haplotypes (GAT and CAT), suggesting the dual role of rs1534309.
None of the investigated SNPs were associated with OS, and according to our results, only the variant allele of rs2070215 seems to be important for AML susceptibility.
Regarding the association between MCM7 SNPs, somatic mutations, and clinical features of AML patients, none of the investigated SNPs were associated with the clinical features of AML patients or with FLT3, DNMT3A, and NPM1 mutations.
Furthermore, for the AML group, considered representative for our region, we found an association between the somatic mutations and high WBC count, high blast percentage, and high LDH level, as previously reported [12,33]. Therefore, patients with a high level of LDH, blast percentage, and WBC count need not only a faster investigation for FLT3 and NPM1 mutations, but also to be correctly classified and receive the benefit of personalized treatment (the goal of precision medicine) in the shortest time [34][35][36][37][38].
In the present study, associations between somatic mutations, WBC count ≥ 10,000 cells/mm 3 , LDH ≥600 IU/L, and poor OS at the univariable level were significant. At the same time, FLT3 mutations were associated with high WBC count (≥10,000 cells/mm 3 ) and LDH level (≥600 IU/L). Moreover, we observed that only the association between high LDH level (>600 IU/L) and poor OS remained significant, irrespective of FLT3 mutation status. Our findings suggest that LDH level ≥600 IU/L could be a useful biomarker for OS. Our observation is in line with the data reported by our group in a previous study [30], as well as with studies reported by Walaa et al. [13] and Hu et al. [14].
To the best of our knowledge, this is the first study to analyze the role of MCM7 SNPs in the development of AML in Caucasian patients and the second to investigate the mentioned SNPs in AML patients, but on a larger AML sample group than previously reported [15]. Furthermore, this is the first study to investigate the association between MCM7 SNPs and somatic mutations, combined somatic mutations, clinical features, and OS on a relatively large number of cases. In addition, the association between somatic mutations, clinical features, and OS were evaluated. One limitation of our study is the lack of investigation of MCM7 gene expression and Mcm7 protein level.

Conclusions
Our study indicates that variant genotypes of investigated MCM7 SNPs are not associated with the risk of AML, clinical features of AML patients, prognosis, or with FLT3, NPM1, and DNMT3A mutations. However, the variant allele of rs2070215, in allelic model, and CAT haplotype, were associated with AML susceptibility. The investigated somatic mutations were associated with high LDH level, blast percentage, and WBC count. From the clinical and biological features of AML (age at diagnosis over 65 years, ELN high-risk and intermediate-risk, presence of FLT3 and DNMT3A mutations, ECOG performance status ≥ 2, WBC count ≥ 10,000 cells/mm 3 , LDH level ≥ 600 IU/L, associated with a poor OS), only LDH level ≥ 600 IU/L was associated with an increased hazard of death, and this association remained significant when quantifying for effect modification by FLT3 (ITD, D835) mutation status. The association of high LDH level (≥600 IU/L) and WBC count (≥10,000 cells/mm 3 ) with poor OS showed the same magnitude in patients with or without FLT3 mutation. As a result of the mentioned associations between LDH level (≥600 IU/L), WBC count (≥10,000 cells/mm 3 ), and presence of FLT3 and NPM1 mutations, as well as their impact on OS, AML patients with high LDH level and WBC count need to be promptly investigated for FLT3 and NPM1 mutations.

Conflicts of Interest:
The authors declare no conflict of interest.