Analysis of the PD-1/PD-L1 Axis Points to Association of Unfavorable Recurrent Mutations with PD-L1 Expression in AML

Background

One of the potential mechanisms responsible for leukemic cells evading cytotoxic T lymphocytes might be the pathway of programmed death-1 receptor (PD-1). PD-1 and its ligand PD-L1 play a key role in tumor immune escape and the formation of tumor microenvironment, promoting tumor development.

Aims

As PD-L1 expression was broadly described in many types of human cancer cells including AML, we were focused on PD-1 expression, which until now was mostly characterized for T cells, where it inhibits proliferation and cytotoxic capabilities. Recently, Wang et al. described PD-L1 expression regulation through miR-34a molecules in AML patients. Overexpression of miR-34a blocked PD-L1 and reduced its surface expression. Moreover, Cortez et al. for the first time identified novel, complete mechanism of PD-L1 regulation by p53 via miR-34a in non-small cell lung cancer (NSCLC). In this study, our comprehensive analyses of PD-1, PD-L1, TP53and miR-34a expression in AML patients shed new light on the complex regulation of PD-1/PD-L1 axis during development of this disease.

Methods

We performed analysis of TP53, PD-1, PD-L1 and miR-34a expression in 197 AML patients available from The Cancer Genome Atlas (TCGA) database. Moreover, we assessed mRNA expression of TP53 and PD-1 in independent cohort of 54 primary AML patient samples using qRT-PCR method. We also characterized several SNPs for PD-1that demonstrate relevant associations with a higher risk of developing autoimmune diseases: PD-1.1 (rs36084323), PD-1.3 (rs11568821), PD-1.5 (rs2227981), PD-1.6 (rs10204525), PD -1.7 (rs41386349), PD-1.9 (rs2227982) in 54 AML, 64 MDS and 100 HVs samples.

Results

TCGA data analysis showed higher expression of PD-L1 in the AML group with low expression of TP53 (p=0.008). In contrast, PD-L1 expression was elevated in the group with TP53 mutations compared to unmutated TP53 (p<0.001).We also found a negative correlation of miR-34a and PD-L1 expression (r=-0.2; p=0.005), while there were no differences in PD-L1 expression between groups with or without following mutations: IDH1, TET2, RUNX1, NRAS, CEBPA, PTPN11, KIT, KRAS, FLT3, DNMT3, NPM1 and IDH2. The highest expression of PD-L1 was found in the poor prognosis group according to cytogenetic risk and molecular risk markers. We also observed an association between the expression level of PD-L1 and the number of recurrent mutations present in an AML case. Patients with more than 4 recurrent mutations were characterized with higher expression of PD-L1 compared to the group of patients with 0-3 recurrent mutations (p=0.01). Next, we observed significant differences in PD-1 expression in the group of 54 AML patients compared to HVs (p<0.001), and there were differences in the PD-1expression level regarding the PD-1. 5 polymorphism (p=0.07). Moreover, analysis of the PD-1. 3 polymorphism in HVs and MDS revealed that genotype GG was associated with nearly fivefold lower risk of disease (OR=4.93, p=0.009). We observed significant differences in OS in AML patients in case of presence of certain genotypes of PD-1. 6. Genotype AA was significant associated with higher risk of shorter OS compared to the rest of the genotypes (58 vs 333 days, HR=35; p=0.02).

Conclusions

Our analyses indicate that p53 might specifically modulate the tumor immune response by regulating PD-L1 via miR-34a which directly binds to the PD-L13'-UTR and blocks its expression. Moreover, we found that high PD-L1 expression is associated with the higher numbers of recurrent mutations as well as poor cytogenetic and molecular risk groups. We found significant differences in PD-1expression in AML patients compared to HVs that further support a deregulation of a signal transduction through the PD-1/PD-1L axis in AML.

While our SNP analysis in AML patients suggested a prognostic impact of PD-1. 6 polymorphism, further studies are warranted to evaluate the impact of the PD-1/PD-L1 axis in AML.

This work was supported by National Centre for Science Grant HARMONIA (UMO-2013/10/M/NZ5/00313).