Myelodysplastic neoplasms (MDSs) are bone marrow cancers with significantly heterogeneous characteristics, accompanied by an increased risk of progression to acute myeloid leukemia (AML).1 The MDS pathogenesis is associated with genetic and epigenetic defects and immune dysregulation.2,3

The overexpression of immune-related genes in the MDS was linked to its pathogenesis in more than 50% of patients.4 Recent findings showed the role of immune alterations in the MDS, where immunological mechanisms can modulate the environment by favoring the expansion of malignant clones.5 The innate and adaptive immune systems are active in the MDS niche and participate in the abnormal hematopoiesis mechanisms. The immune system and the immune evasion mechanism can act in the cell death, proliferation, or suppression of hematopoiesis.6

The immune evasion occurs by the expressing of immune checkpoint molecules, such as the Protein Death 1 (PD-1) and Protein Death Ligand 1 (PD-L1). These molecules participate in the co-inhibition signaling of lymphocyte responses, preventing the cell death. Furthermore, the induction of the PD-L1 expression in tumor cells occurs in response to specific cytokines, such as the tumor necrosis factor alpha (TNF-α) and interferon gamma (IFN-γ).7 The immunotherapy with checkpoint inhibitors is already performed in several cancers, but the treatment success depends on the type of tumor and the patient's immune system.8–10

In this context, the measurement of PD-L1 expression levels emerged as a possible biomarker to screen patients more susceptible to anti-PD-L1 drugs.11,12 In the MDS, studies reported contradictory findings and the role of the PD-L1 in the development and progression of the disease remains to be elucidated.13,14 We aimed to evaluate if the PD-L1 gene expression would be linked to clinical and prognostic features of patients with myelodysplastic neoplasms.

The immune dysregulation began to be explored in recent decades as an essential component in the pathogenesis of the MDS.18,6 The MDS heterogeneity makes it difficult to determine an immunological profile that can be explored in immunotherapy studies. Several drugs targeting the PD-1/PD-L1 pathway have been tested in the MDS and the results are still contradictory.19–21 In this study, we sought to associate the PD-L1 gene expression levels with clinical and prognostic characteristics to contribute to the research directing patients to treatment with checkpoint inhibitor drugs.

The PD-L1 overexpression in bone marrow cells of MDS patients was directly linked to unfavorable prognostic markers, such as the dyserythropoiesis and the most advanced subtype according to the WHO 2016 Classification.22 The dyserythropoiesis occurs when there are 10% or more dysplastic erythroid cells in the bone marrow and affects about 80% of patients with MDS.23 Recent work by our group has suggested the hyperactivation of the toll-like receptor 3 (TLR-3) by endogenous retroviruses (ERVs) as the trigger of the dyserythropoiesis in MDS.24 In addition, the TLR-3-ERV axis has also been identified as an interferon response inducer through the interferon regulatory factor 3 (IRF3), IRF7 and nuclear factor kappa B (NF-kB). In cancers, NF-kB directly induces the expression of the PD-L1 through binding to its promoter and post-transcriptionally by indirect pathways.25 Thus, we speculate that this would be a line between dysplasias, ERVs and PD-L1.

Hypomethylating agents (HMAs) are the primary drugs used for high-risk MDS, such as the MDS-EB2 subtype. These drugs, including the 5-azacytidine (AZA) and 5-aza-2′deoxycytidine (DEC), act as inhibitors of DNA methyltransferase and lead to DNA demethylaton. The DNA demethylation restores the transcription of many silenced genes, including tumor suppressors.26 Evidence has already shown, however, that patients in treatment with HMAs have a loss of response after approximately two years of treatment.27,28 Chiappinelli et al. showed that PD-L1 levels are upregulated after using HMAs.29 An increase in PD-L1 levels after using HMAs is possibly associated with the tumor escape and loss of response to the treatment.30 Based on our findings, we hypothesize that these high-risk patients with higher levels of PD-L1 may be more resistant to HMA therapy and an alternative management could be an association between HMAs and PD-L1 checkpoint inhibitors.

Another recent study by Sallam et al. found that bone marrow stem cells from MDS patients with mutations in the TP53 gene had a higher expression of the PD-L1 than those with the wild-type TP53 gene.13 Mutations in the TP53 tumor suppressor gene confer a poor prognosis in the myelodysplastic syndrome. Even in patients with a good prognostic karyotype, such as the deletion in the long arm of chromosome 5 (5q-), a second mutation in the TP53 completely transforms the course of the disease, leading the patient to a high-risk condition with a lower survival.31 Different from the findings by Salman et al., which did not find significant differences between disease subtypes and the PD-L1 expression, our results showed a statistically significant increase in the MDS-EB2 versus MDS-EB1 subtypes and, despite the statistics, patients classified in the MDS-EB2 showed a higher expression of the PD-L1 than all other subtypes in our cohort.

A study performed on AML patients demonstrated that the loss of the p53 leads to the PD-L1 expression through the p53/miR34/PD-1 pathway.32 The p53 molecule induces the expression of the microRNA-34, inhibiting the expression of the PD-L1 by binding to the 3′UTR regulatory region of the gene.33 In MDS, we believe that, similar to AML, mutations in the TP53 that directly reflect the p53 protein deregulate the miR-34 pathway and increase the expression of the PD-L1. Unfortunately, our data lacks a mutational analysis of the TP53 gene and microRNA expression. Nevertheless, we believe that our data indicate that the PD-L1 expression is a biomarker of a worse prognosis in the MDS.

Regarding our predominant group of outliers, we had a complex karyotype result in one of the three. Williams et al. observed a higher rate of the PD-L1 expression in bone marrow blasts of AML patients with complex karyotypes.34 Chromosomal instability and defects in DNA repair pathways are involved in the pathogenesis of high-risk MDS, which often has an accumulation of mutations.35,36 Studies suggest that tumors with defects in DNA repair pathways present a higher expression of the PD-L1 due to the context of genomic instability that favors the generation of neoantigens and, consequently, activation of the INF and PD-L1.37,38 Possibly, targeting anti-PD-L1 therapies to patients with MDS who have cytogenetic abnormalities and a high mutational load, particularly in DNA repair pathways, are more susceptible to a positive therapeutic response.

In conclusion, we believe that the PD-L1 expression data presented here in our cohort were associated with worse prognosis markers, such as the dyserythropoiesis and its worst subtype, MDS-EB 2. We speculate that the increase in PD-L1 levels after using hypomethylating agents is one of the factors responsible for the loss of response to this treatment. An initial combination of HMAs with PD-L1 inhibitors should be tested in clinical trials. In the group of outliers, the idea remains that the high expression of the PD-L1 is a biomarker of a worse prognosis in MDS, but with great possibilities of targeting anti-PD-L1 therapy.

All procedures were approved by the Ethics Committee of the UFC (No. 69366217.6.0000.5054) and are in accordance with the 1964 Helsinki Declaration and its later amendments. Informed consent was obtained from all individual participants included in the study.

LRS, BFV, SMMM and RFP designed the study, provided patient materials and were responsible for collection and assembly of data. MAV, VSO, MMLM, RTGO and DPB performed the molecular procedures and analyzed the data. All authors drafted and edited the manuscript and have approved the final version of the manuscript before publication.