Introduction

Myelodysplastic syndromes (MDS) are a group of heterogeneous hematopoietic clonal stem cell disorders characterized by peripheral cytopenia(s), myeloid dysplasia, and an increased risk of progression to acute myeloid leukemia (AML) [1,2,3,4,5]. Despite the recent advances in the last 5 years in describing a panel of molecular abnormalities that might contribute to the multistep process in the pathogenesis of MDS [6], the pathologic sequence remains poorly understood and likely depends on an interplay between aberrant hematopoietic cells and their microenvironment [7,8,9].

Several findings account for an immune dysregulation whose nature differs between MDS-risk groups. In the lower-risk population, the immune system is characterized by an activated proinflammatory state leading to high levels of apoptosis of hematopoietic progenitors, thus releasing dysplasia-associated antigens that may evoke an adaptive immune response. In advanced MDS, an immunosuppressive environment with an increased number of regulatory T cells occurs, enabling dysplastic clones to expand and evade immune surveillance. The disturbed differentiation is accompanied by apoptosis resistance and an increased proliferation [9]. Accumulated evidence has indicated that marrow failure, in some MDS, is associated with autoimmunity, T-cell mediated myelosuppression, and cytokine-induced cytopenias [8]. In addition, a previous history in patients of any autoimmune disease has been associated with an increased risk of MDS, and –vice versa– clinical manifestations of autoimmune disease are frequently described in patients with MDS and chronic myelomonocytic leukemia [10].

Cytokines are key elements in the interaction of the malignant clone with the microenvironment along with the ineffective hematopoiesis in MDS and may influence a variety of clinical manifestations that characterize the disease. The role of the tumor necrosis factor-α (TNF-α) in affecting the rates of intramedullary apoptosis [11], the outcome of MDS patients [12], and the response to erythropoietin (EPO) treatment has been widely confirmed [13]. Higher levels of interferon-γ (IFN-γ) have also been demonstrated in trephine biopsies and in the bone marrow (BM) plasma [14]. Adherent cell layers from MDS patients in culture produced higher levels of interleukin-6 (IL-6) and TNF-α [7], while increased plasma levels of IL-6 were independently associated with an inferior outcome [15]. A higher production of the transforming growth factor-β (TGF-β) by mesenchymal stem cells in the high-risk MDS group contributes to the increase in regulatory T cells in those patients [16].

Several reports have identified polymorphic variants in human leukocyte antigen genes and in cytokine and cytokine-receptor genes that are linked to susceptibility to cancer and autoimmunity and that contribute differentially to disease development, progression, and the response to therapy [17,18,19,20,21,22,23,24,25,26,27]. Functional variants such as single nucleotide polymorphisms (SNPs) localized in the promoter or regulatory regions of relevant cytokine genes may influence their genetic expression, thus skewing the functioning of immune cells through the action of their effector molecules, the cytokines and chemokines. Previous studies have proven that the presence of the TNF –308A, IFNG +875 12 CA-repeat, IL6 –174G, and TGFB1 +869C or +915G alleles increase the expression of the transcript and/or the protein of the encoded cytokines. Although the influence of these variants has been confirmed in vitro (i.e., mitogen stimulation of cells in culture, isolation of individual alleles of gene promoters by cloning adjacent to a reporter gene, and/or performing a mobility shift assay in the presence of nuclear extracts), certain data reinforce the notion of an in vivo modulation as well [28,29,30].

Since the level of cytokine secretion is under the influence of genetics and environmental conditions, in the present study, we were interested in evaluating a possible association between the above detailed cytokine gene polymorphisms and both disease susceptibility and clinic-pathologic features in a series of Argentine MDS patients.

Subjects, materials and methods

Study population

The study was a multicenter retrospective and prospective analysis of 153 MDS patients from the registry of the Laboratorio de Genética Hematológica, Instituto de Medicina Experimental (IMEX-CONICET)/ Academia Nacional de Medicina, Buenos Aires. Hematologists from the participating institutions completed a standard registration form for each patient detailing the clinical and hematologic features at presentation and during the follow-up. The ethics committee of the Institutos de la Academia Nacional de Medicina has evaluated and approved this research.

Inclusion in the study required a confirmed diagnosis of de novo MDS, excluding those patients with treatment-related MDS or a previous history of toxic substance exposure. MDS patients were diagnosed from September 1981 through May 2014 with a median overall survival of 44.4 months. During the follow-up period of 26.1 (1.0–174.6) months, 45 (29%) evolved to AML and 87 (57%) died including 44 patients with previous progression to AML. Patients were classified according to the French-American-British (FAB) [1] and the World-Health-Organization (WHO) [3] diagnostic criteria and were distributed as defined by the International Prognostic Scoring System (IPSS) and the System’s revised version (IPSS-R) (Table 1) [2, 4].

Table 1 Clinical characteristics and IPSS, IPSS-R, FAB, and WHO classifications of the MDS population studied
Full size table

Samples from 131 unrelated healthy blood donors who donated between 2000 and 2003 in a different ward from the one used for the patients’ blood withdrawal (Transfusion Blood Bank from the Servicio de Hemoterapia y Hemofilia, Instituto de Investigaciones Hematológicas, Buenos Aires) were studied as the control population. The donors presented with a median age of 38 years (range: 20–64) and a male/female ratio of 1.6 not being statistically different from the patients’ sex ratio (Fisher’s exact test, p = 0.539).

Genotyping of cytokine polymorphisms

Genomic DNA was isolated from peripheral blood or BM mononuclear cells by a standard phenol-chloroform method.

The TNF (−308 G/A, rs1800629) study was performed by polymerase chain reaction (PCR)-restriction fragment length polymorphisms (PCR-RFLP, NcoI digestion) and visualized by silver staining [31]. The IFNG microsatellite region (+875 CAn, rs587776821) was amplified by PCR, resolved by nondenaturing polyacrylamide gel electrophoresis (PAGE-PCR), and visualized by silver staining [20]. The IFNG (+874 A/T, rs2430561), IL6 (−174 G/C, rs1800795), and TGFB1 (+869 C/T, rs1800470; +915 G/C, rs1800471) genotypes were determined by PCR through the use of sequence-specific primers (SSP-PCR) and resolved on ethidium bromide stained agarose gels [19, 29, 32]. Table 2 lists the PCR primers used. The genotypes obtained for each cytokine polymorphism were confirmed by automated DNA sequencing.

Table 2 PCR primers used for genotyping of cytokine polymorphisms
Full size table

Statistical analysis

Differences in genotypic or allelic frequencies and in genotype-related clinical parameters at diagnosis regarding genotype were evaluated by the chi-square or Fisher’s exact test, and the Mann-Whitney test or Student t test for continuous data. In the multivariate analysis, variables with a p value <0.2 were included, and the logistic regression model was performed to calculate the odds ratio (OR) and the respective 95% confidence interval (CI). The logistic regression model was internally validated by a bootstrap algorithm based on 1000 replications. The results were presented as bootstrap CIs and significance tests for the correlation coefficient (B). Pairwise linkage disequilibrium (D’ value) between the rs2430561 SNP and the rs587776821 microsatellite were estimated according to Maynard [33].

Overall survival (OS), leukemia-free survival (LFS) and transfusion dependence-free survival (TDFS) were defined as the interval from diagnosis to the date of death, progression to AML, and transfusion dependence, respectively. Patients undergoing hematopoietic stem cell transplantation (3.9%) were censored up to the moment of the procedure. OS, LFS, and TDFS were estimated by the Kaplan-Meier method and compared with the log-rank test (Mantel-Cox). Analyses were performed with SPSS V24 software (IBM, New York, USA) and InfoStat Version 2016 (Grupo InfoStat, Universidad Nacional de Córdoba, Argentina). And, the correspondence analysis was performed according to Version 1.1, Data Theory Scaling System Group (DTSS) (DTSS, Faculty of Social and Behavioral Sciences, Leiden University, The Netherlands). All statistical tests were two-sided, and the level of statistical significance was defined as a p value <0.05.

Results

Analysis in the overall MDS population

Genotypic and allelic frequencies of cytokine polymorphisms

As described in Table 3, both the genotypic and allelic frequencies of the TNF −308 G/A and IL6 –174 G/C SNPs were different between the MDS patients and the control subjects. That the high-expressing TNF genotype (G/A + A/A) was overrepresented in patients suggests an increased susceptibility of 2.7-fold to develop MDS (p = 0.005). The high-expressing IL6 genotype (G/G) was less frequently observed, thus inferring a protective role of 0.5-fold (p = 0.017). With respect to IFNG and TGFB1 polymorphisms, only the +869 C/T heterozygote genotype tended to be increased (p = 0.094, OR: 1.658). These results were confirmed by bootstrap analysis as the method of internal validation (Online Resource Table S2a). The genotype distributions for cytokine polymorphisms in MDS patients and controls were in Hardy-Weinberg equilibrium, except for the IL6 polymorphism.

Table 3 Distribution of genotypic and allelic frequencies of cytokine polymorphisms in MDS patients and control subjects
Full size table

Association of cytokine polymorphisms with clinical variables

Relevant MDS clinical variables were analyzed according the respective level of expression associated with each cytokine polymorphism (Online Resource Table S3). As we previously described [22], the presence of the genotype of high (G/A + A/A) as opposed to low (G/G) TNF expression was associated with a lower hemoglobin level (8.8 vs. 9.7 g/dL; p = 0.034), reduced platelet counts (90,000 vs. 131,000/μL; p = 0.010), and a shorter TDFS (7.9 vs. 27.0 months; p = 0.016). In addition, this group ran a 3.9-fold higher risk to succumb to non-AML related death (73 vs. 41%; p = 0.013). Similarly, the genotype of high (G/G) opposed to intermediate + low (G/C + C/C) IL6 expression was correlated with a lower hemoglobin level (8.8 vs. 9.9 g/dL; p = 0.009), with a 3.9-fold risk of a lower basal hemoglobin level <8 g/dL (40 vs. 14%; p = 0.001), and accordingly, with a shorter TDFS (6.4 vs. 39.4 months; p = 0.044). Therefore, we found a significant association between the presence of high-producing TNF and IL6 variants and a severer anemic phenotype within the context of the disease.

The analysis of the IFNG +875 CAn microsatellite polymorphism revealed that MDS patients carrying the high-expressing homozygote IFNG genotype presented with lower neutrophil counts [12/12 CA (high): 1140 vs. 12/non-12CA + non-12/non-12 CA (intermediate + low): 1777/μL; p = 0.029]. A similar association with a more profound neutropenia was manifested with the IFNG +874 T/T genotype (data not shown) owing to the strong linkage disequilibrium between the two functional alleles (D′ = 0.99), as we previously described [25].

As to TGFB1 polymorphisms, the high-producing +869 genotype was associated with a younger age at diagnosis [C/C (high): 62 vs. C/T + T/T (intermediate + low): 68 years; p = 0.034], while the high-producing +915 genotype with <5% BM blasts [G/G (high): 72% vs. G/C + C/C (intermediate + low): 48%; p = 0.032, OR: 2.790]. No other relevant associations were found (Online Resource Table S3).

Relevant combined genotypes affecting clinical features

Since TNF and IL6 SNPs were associated with abnormalities in the hemoglobin level in the univariate analysis, we evaluated the influence of the combined genotypes on the severity of the anemia at diagnosis along with other clinical variables. The presence of each or both high-producing genotypes [TNF (high): p = 0.048, OR: 3.979; IL6 (high): p = 0.001, OR: 6.835; TNF (high)_IL6 (high): p = 0.010, OR: 6.068] and platelet counts <50,000/μL (p = 0.004, OR: 4.857) were independently associated with an increased risk to manifest MDS with hemoglobin levels <8 g/dL at diagnosis (Table 4). These results were internally validated by bootstrap analysis (Online Resource Table S2b). Patients with the TNF (high)_IL6 (high) genotype presented more frequently with severe bicytopenia at diagnosis than those with other genotypic combinations (p = 0.010, Fig. 1) along with an associated risk of 8.4-fold (5/15, 33% vs. 7/124, 6%; p = 0.004, 95%CI: 2.240–31.185) and consistently became transfusion dependent earlier (2.9 vs. 34.6 months; p = 0.001) (Fig. 2a). The TNF (high)_IL6 (high) genotype manifested a 3.4-fold higher susceptibility to develop MDS with respect to the control group (p = 0.015).

Table 4 Associations of TNF and IL6 gene polymorphisms and clinical variables with the degree of anemia at diagnosis in MDS patients
Full size table
Fig. 1
figure 1

Simple correspondence analysis. Distribution of MDS patients according to the TNF_IL6 combined genotypes and bicytopenia at diagnosis, the latter defined as hemoglobin level (Hb) <8 g/dL and platelet counts (Plt) <50,000/μL. Dimension 1 (abscissa) explains 90.7% of variance, while dimension 2 (ordinate) 9.3%. The TNF (high)_IL6 (high) and <8 g/dL_<50,000/μL categories are positively associated owing to the greater distance from the origin

Full size image
Fig. 2
figure 2

Transfusion dependence-free survival (TDFS) and combined genotypes. Kaplan-Meier curves indicating the median TDFS with respect to TNF_IL6 combined genotypes for (a) the overall MDS population and (b) patients with <5% bone marrow blasts

Full size image

TNF-α and IFN-γ are typical cytokines mediating a Th1 immune response. The TNF (high)_IFNG (high) genotype, observed in eight patients, was associated with a reduced platelet count relative to those of other genetic combinations (88,000 vs. 124,000/μL; p = 0.042) along with lower neutrophil counts (702 vs. 1763/μL; p = 0.056) at diagnosis.

Analysis in MDS patients with <5% BM blasts

The relationship between these cytokine gene polymorphisms and MDS susceptibility in the group of patients that presented with <5% BM blasts (n = 102) was similar to that of the overall MDS population (logistic regression, TNF: p = 0.003, OR: 3.080, 95%CI: 1.468–6.461; IL6: p = 0.037, OR: 0.547, 95%CI: 0.311–0.963, and TGFB1 +869: p = 0.099, OR: 1.754, 95%CI: 0.899–3.422, as internally validated through bootstrap analysis).

In this restricted population, the presence of the genotype of high (G/A + A/A) as opposed to low (G/G) TNF expression was also associated with a lower hemoglobin level (8.9 vs. 10.1 g/dL; p = 0.020), reduced platelet counts (100,500 vs. 141,500/μL; p = 0.021), a shorter TDFS (8.5 vs. 173.6 months; p < 0.001), and with a 9.5-fold increased risk to succumb to death unrelated to AML evolution (12/13, 92% vs. 19/34, 56%; p = 0.036, 95%CI: 1.535–58.484). Furthermore, the genotype of high (G/G) as opposed to intermediate + low (G/C + C/C) IL6 expression was associated with a 2.9-fold increased risk of presenting with hemoglobin levels <8 g/dL (12/36, 33% vs. 9/61, 15%; p = 0.042, 95%CI: 1.073–7.778). The TNF (high)_IL6 (high) vs. the other combined genotypes was associated with a 4.5-fold higher risk of manifesting hemoglobin levels <8 g/dL (logistic regression, p = 0.037, 95%CI: 1.091–18.408, as internally validated by bootstrap analysis), a 16-fold proclivity to manifest a severe bicytopenia (4/14, 29% vs. 2/88, 2%; p = 0.004, 95%CI: 2.592–98.773) at diagnosis, and a shorter TDFS (2.9 vs. 173.6 months; p < 0.001) (Fig. 2b).

Because of the relevance of IL-6 to promote Th17 development within the context of TGF-β1, we combined both genotypes. The IL6 (high)_TGFB1 +869 (high) vs. the other combined genotypes was associated with increased risks of 4.4- and 5.6-fold of manifesting hemoglobin levels <8 g/dL (5/10, 50% vs. 16/86, 19%; p = 0.038, 95%CI: 1.195–16.018) and platelet counts <50,000/μL (4/9, 44% vs. 10/82, 12%; p = 0.029, 95%CI: 1.415–23.450), respectively. This genotypic combination was also associated with a 7-fold increased risk of presenting with a severe bicytopenia at diagnosis (3/10, 30% vs. 5/86, 6%; p = 0.035, 95%CI: 1.496–32.216).

Discussion

Polymorphisms in the immunomodulatory genes are known to affect the immune functions and to be closely correlated with the susceptibility to autoimmune disease. Although an immune dysregulation seems to be a key player in the pathologic process in MDS development, to the best of our knowledge, there are few articles to date have been focused on isolated cytokine polymorphisms and their relationship to susceptibility. Furthermore, studies that evaluate possible associations with the clinical characteristics in MDS patients are infrequent, restricted to only TNF variants [21, 34], or are otherwise absent.

Since our previous work was focused on only the TNF –308 G/A SNP, we increased the population size and extended the study to five other candidate polymorphisms in the IFNG, IL6, and TGFB1 genes based on published data that indicated the relevance of these cytokines to the MDS aberrant microenvironment. We were then able to confirm that the high-expressing TNF genotype (A/A + G/A) was related to an increased chance of developing MDS, in accord with other previous reports [23, 35], and that this genotype was associated with anemia and thrombocytopenia at diagnosis [22]. Consistent with the cytopenia, these patients proved to be transfusion dependent and died of non-AML related causes. MDS patients express high levels of TNF-α in the BM [7, 12], a phenotype related to the rates of intramedullary apoptosis and a worse patient outcome [12]. Because TNF-α exerts a negative regulation on endogenous EPO, the expression of that cytokine directly affects hemoglobin levels [12] and the response to EPO treatment [13].

TNF-α and IFN-γ are the main cytokines involved in Th1-lymphocyte immune-mediated suppression of hematopoiesis through different pathways. In agreement with the previous study of Serio et al. analyzing the IFNG +874 A/T SNP [23], which variation shows a strong linkage disequilibrium with IFNG +875 CAn microsatellite (D′ = 0.99) [25], no relationship of these two polymorphisms to susceptibility was found in our population. Nevertheless, MDS patients carrying the high-expressing IFNG genotype (12/12 CA) presented with a severer neutropenia at diagnosis. IFN-γ reduces the neutrophil differentiation induced by the granulocyte-colony stimulating factor [36] and is also overexpressed in MDS patients [14] whose CD34+ cells manifest an up-regulation of antiproliferative IFNG-stimulated genes [37]. Although the TNF (high)_IFNG (high) combined genotype may contribute to the cytopenic state, that genetic pairing was observed in only eight patients.

We were able to observe a direct relationship between the IL6 –174 G/C SNP and MDS susceptibility, in contrast to previous reports [23, 26]; furthermore, the high-expressing IL6 genotype (G/G) was associated with a severe anemia and a shorter TDFS. The contribution to MDS pathophysiology of IL-6, another multifunctional cytokine, is not completely understood: whether elevated levels of IL-6 represent a reactive response or an intrinsic pathophysiologic component. This cytokine is the principal mechanism underlying the anemia of inflammation that induces the synthesis of hepcidin, which dysregulation constitutes a complication in MDS [38]. Fishman et al. demonstrated that a genetically determined difference exists in the degree of the IL-6 response between individuals within an in vitro inflammatory context [19], and that the influence of the high-expressing variant may perturb the hemoglobin levels of MDS patients. An overproduction of IL-6 has been described in MDS [7, 15] that was associated with a lower OS and LFS [15]. Furthermore, the presence of IL-6, within the context of TGF-β1, induces differentiation in favor of the Th17 development, thus contributing to the immune dysregulation as an additional characteristic of low-risk MDS [9]. In this regard, our results indicated that the IL6 (high)_TGFB1 +869 (high) combined genotype may intensify the severity of bicytopenia in MDS patients with low (<5%) BM blast counts.

As defined above, TNF and IL6 polymorphisms were independently associated with the hemoglobin level at diagnosis in our series of patients, thus influencing the MDS phenotype. The present analysis of combined genotypes confirmed the association between the high-producing TNF and IL6 genotypes and the severity of anemia. Especially, patients carrying the TNF (high)_IL6 (high) combined genotype presented with a severer bicytopenia and consistently experienced a shorter TDFS. Since both cytokines are directly linked to hematopoietic failure in MDS, the two have been targeted through the production of their respective inhibitors. Approved TNF-α inhibitors have shown a varying degree of efficacy in improving blood cell counts in MDS patients [39,40,41]. Despite the therapeutic efficiency observed in Castleman’s disease and rheumatoid arthritis, an anti-IL-6 therapy did not reduce the need for transfusions in transfusion-dependent patients with low-risk MDS [42]. An association has been recognized between genetic polymorphisms and the response to anti-TNF-α of patients with inflammatory diseases. López-Hernández et al. found an increased frequency of the high-expressing TNF allele (−308A) and genotype (−308G/A) in patients not responding to TNF-α inhibitors [24]. The relationship between TNF –308A and anti-TNF-α refractoriness as well as the influence of this allele on the clinical features of our patients may constitute the rational bases for designing future clinical trials in MDS.

Our results point out that a patient’s genetic background likely contributes to the T cell-mediated inhibition of hematopoiesis that influences the degree of cytopenias in MDS, which effect becomes more evident in patients with lower BM blast counts. TNF and IL6 polymorphisms, as underlying host factors, are likely to play an essential role in influencing the susceptibility and clinical characteristics in MDS patients. These findings would argue that the genotyping of functional cytokine polymorphisms might be instrumental to the tailoring of cytokine-target therapies in MDS.