Myelodysplastic syndrome (MDS) is a heterogeneous group of clonal hematopoietic stem-cell disorders mainly characterized by significant morphologic dysplasia, ineffective hematopoiesis, and peripheral blood cytopenias. MDS patients also have an increased probability of leukemic transformation [1, 2, 3, 4]. Several mechanisms have been recognized to be involved in MDS development, including recurrent cytogenetic abnormalities, genetic mutations, and aberrant epigenetic modifications, which are also variable throughout the different MDS subtypes. Epigenetic modifications are very common in cancer and play a central role in MDS pathogenesis [5, 6, 7]. This process is a significant gene transcription regulator that occurs mainly in the promoter region of genes, particularly in tumor suppressor genes (TSG), leading to the silencing and inactivation of numerous TSG genes [8].

DNA methylation appears to have great potential as a nucleic acid-based biomarker for assessing cancer risk and early detection, prognosis, targeted therapy, and therapeutic responses [9, 10, 11]. There are several advantages to using hypermethylated genes as biomarkers: hypermethylation is a mechanism characteristic of neoplastic cells, different tumors show different methylation profiles with different clinical impact, and it can be measured by quick and sensitive techniques. Another advantage is the high stability of the DNA molecule that can be obtained from a wide variety of sources [12], such as peripheral blood where its recovery is highly efficient.

The peripheral blood has a rich content of several cellular and molecular elements that provide different information relative to health and disease status, making it an ideal sample source to develop noninvasive biomarkers for cancer detection and monitoring [13]. DNA methylation biomarkers from whole blood and blood leucocytes have been identified in patients with different solid tumors, such as breast, gastric, and colorectal cancer [9, 14, 15, 16]. However, despite the important role of DNA methylation in MDS pathogenesis and other hematopoietic cancers, methylation profiles of peripheral blood cells still largely unknown.

In the present study, we analyzed the DNA methylation status in peripheral blood (PB) and in the corresponding bone marrow (BM) samples from MDS patients in order to explore the existence of a correlation between these two samples. This analysis will provide the relevance of PB samples for DNA methylation detection in MDS patients.

Noninvasive molecular tests for biomarker detection have become one of the most important issues in everyday clinical practice, not just for detecting and monitoring cancer but also for the diagnosis and follow-up of several other diseases. Peripheral blood samples and other body fluids seem to have great potential, as they can be obtained by noninvasive procedures with almost no harm to the patients. Utilizing these tests will allow the clinicians to perform a tighter follow-up with the patients since they can apply these methods more frequently along the disease progression. Several studies have already addressed the potential use of peripheral blood markers in myelodysplastic syndrome patients using different approaches. For instance, WT1 mRNA expression levels in PB leucocytes have been shown to be potential markers for MDS diagnosis and risk evaluation [23, 24] and for the prediction of early relapse [25]. Fakhr et al. demonstrated that FISH analysis in PB for common MDS cytogenetic abnormalities could be useful for clinical monitoring of these patients [26]. Other groups have demonstrated that flow cytometry in PB testing might also be useful for diagnosing low-risk MDS patients based on neutrophil and monocyte immunophenotyping [27] and could also replace BM aspiration in some suspected cases of MDS [28].

Aberrant DNA methylation has been reported in many types of cancers and has been implicated in MDS pathogenesis, where disruption of several genes is known to affect different cellular processes [6, 29, 30, 31]. DNA methylation analysis is performed mainly in bone marrow samples, but in pathologies like MDS, methylation analysis in peripheral blood might give clinicians a large amount different information while avoiding invasive medullary aspirates.

The present results revealed that DNA methylation is a common process in MDS patients, with more than 75% of MDS patients presenting more than two hypermethylated genes. P15 and DAPK presented a higher methylation status among our MDS group. P15 gene is one of the most studied MDS genes, as its hypermethylation is present in 50 to 80% of MDS cases [32, 33, 34, 35]. The inactivation of this cell cycle regulator plays a crucial role in MDS pathogenesis, leading to inappropriate progression of the cell cycle which induces the growth advantage of neoplastic cells that can result in leukemic transformation [35, 36]. DAPK promotor inactivation leads to apoptosis inactivation and a higher metastatic capacity in solid tumors, and is associated with a poor prognosis, for instance, in lung cancer [37, 38]. In hematological cancers such as monoclonal gammopathies, it was observed that DAPK methylation might be a marker of disease progression with a negative impact on disease prognosis [39]. A more recent study in lymphoma patients also demonstrated that DAPK promoter methylation might serve as a negative prognostic biomarker even though it is not associated with lymphoma progression [40]. Despite the impact of DNA methylation in several neoplasias, it is important to point out that there is also a link between DNA methylation and aging, showing the establishment of global hypomethylation and regions of CpG-island methylation with age [41]. The presence of comorbidities in some older patients and controls, such as type 2 diabetes, can also influence DNA methylation profile [42]. In our study, both analyzed groups are age-matched; however, information about the presence of comorbidities in all individuals in both patient and control groups were not available and might explain some differences between our results and other studies.

We also evaluated the correlation of DNA methylation in BM and PB samples of MDS patients. We observed the higher correlation rates in P15 and TRAIL DcR1 gene methylation. TRAIL DcR1 gene is an anti-apoptotic TRAIL receptor similar to TRAIL DcR2; neither have the death domain, which makes them unable to transduce the death signal. For this reason, they are considered oncogenes [43, 44]. The methylation levels of TRAIL DcR1 gene are not yet described in MDS, as well as the comparison between BM and PB samples. However, Shivapurkar (2004) demonstrated the hypermethylation of this TRAIL receptor in several solid neoplasias, as well as in lymphomas and multiple myelomas [44].

For P15 and TRAIL DcR1, the level of agreement between samples was fair according to the kappa statistics (P15: k = 0.344; TRAIL DcR1: k = 0.243), and the concordance between the peripheral blood and bone marrow were also considerably high for these two genes (P15: 70.6%; TRAIL DcR1: 75.0%). Although this type of correlation had never been done in MDS patients, there are a few studies involving a simple comparison of P15 gene promoter methylation between these different types of samples. One of these studies demonstrated that all patients with P15 methylation in BM (n = 4) also presented the same result in PB samples [45]. Moreover, an analysis performed in our investigation group, using samples from monoclonal gammopathy patients, reported a strong correlation between BM and PB samples when evaluating the methylation status of P15 and DAPK genes [46]. Even though we observed a good level of agreement between the two analyzed samples, a higher agreement would be expected. The agreement level obtained might result from the small sampling studied or even due to false-positive and/or false-negative results that are a direct consequence of using a very sensitive detection method such as the MS-PCR [47]. P15 also presented the highest sensitivity (79.5%), and TRAIL DcR1 showed high specificity (82.5%). Although both of these parameters should present higher rates, these results might be explained by the number of methylated samples for each gene promoter, with P15 having a high percentage of methylated samples (positive results) and TRAIL DcR1 a high percentage of unmethylated samples (negative results). This fact is also supported by PPV and NPV values.

The other genes did not present any significant results when we compared their methylation status, though the level of methylated samples was relatively high in some of those genes, like DAPK. This outcome may result from the number of patients enrolled in this study. It is also important to note that none of the patients presented promoter methylation for TP53 and MGMT genes. The lack of TP53 promoter methylation was expected as its inactivation is usually due to genetic mutations [48] and it is more frequently associated with del(5q) cases [49, 50].

In conclusion, our study showed that promoter methylation is a common event in MDS patients, in both peripheral blood and bone marrow samples. P15 and TRAIL DcR1 methylation status might be used as potential cancer biomarkers in peripheral blood samples from MDS patients, mainly as part of a biomarkers panel. DNA methylation patterns measured in peripheral blood might have great potential as informative biomarkers of cancer risk and prognosis; however, large systematic and prospective studies must be performed, and other genes should be evaluated in single or combined approaches.

JJ and ABSR conceived of the study. EC, CG, AP, and LR recruited and provided the clinical information of the participants. JJ, ACP, and RA conducted the experiments. JJ and ACG performed the statistical analysis. RA, ACG, and JJ wrote the paper. JNNC and ABSR revised the manuscript. All authors read the final version and approved its submission.

The procedures performed in the present study were approved by the Ethics Committee of the Faculty of Medicine, University of Coimbra, Portugal. Samples were collected after informed consent was obtained from the involved patients in accordance with the Helsinki Declaration.

A sincere thank you to Lara Mouttham (Cornell University, NY, USA) for revising the manuscript.

The present work was supported by CIMAGO - Center of Investigation on Environment, Genetics and Oncobiology, Faculty of Medicine, University of Coimbra, Portugal, and by National Funds via the Portuguese Foundation to Science and Technology (FCT) through the Strategic Project UID/NEU/04539/2019, COMPETE-FEDER (POCI-01-0145-FEDER-007440) and UIDB/04539/2020 and UIDP/04539/2020 (CIBB). FCT supported RA and JJ with a PhD Grant (SFRH/BD/51994/2012 and SFRH/BD/145531/2019, respectively).

The authors declare no competing interests.