The IRFs are transcription mediators of virus-, bacteria-, and IFN-induced signaling pathways which play a critical role in antiviral defense, immune response, cell growth regulation, and apoptosis [15]. In recent studies, the IRFs were reported to play roles in tumor biology, including in human leukemia [23, 24, 29]. Although the IRF family has been greatly explored in different tumors, the systematical bioinformatics analysis in human leukemia has not been carried out yet. We anticipated that the gene expression of IRFs relevant to the prognosis of AML will shed some light on expanding the knowledge on genetic factors of AML, promoting therapy, and improving the prognosis of AML.
In this study, we observed over-expression of the IRF1/2/4/5/7/8/9 genes and under-expression of the IRF3/6 genes in AML patients compared with healthy donors from the GEPIA database analysis. Furthermore, the IRF9 was correlated with poor prognosis in AML patients through the GEPIA database analysis, and IRF1 and IRF7 were correlated with poor prognosis in AML patients using the LinkedOmics and TCGA databases. The conclusion from the GEPIA database analysis was inconsistent with that from the LinkedOmics and TCGA databases, which may be related to statistical bias caused by the insufficient sample size used in the GEPIA database. From TCGA database analysis, we found that over-expressed IRF1, IRF6, and IRF9 were associated with poor prognosis in AML patients who received chemotherapy alone; the prognostic efficiencies of these markers were decreased in patients who received hematopoietic stem cell transplant, indicating that stem cells transplants may reverse the prognostic efficiency. The data from our center revealed that AML patients with high IRF1and IRF7 expression had a poor prognosis, but IRF9 expression was not associated with OS, which was consistent with the outcomes from the LinkedOmics and TCGA databases analyses.
A previous study reported that the IRF1 gene served as an interferon-induced regulator of transcription and was up-regulated during myeloid differentiation [13]. It was generally considered a tumor suppressor gene [22]. However, some studies showed that loss of IRF1 caused hyper-susceptibility to colitis-associated colorectal cancer [30], while high expression of IRF1 was involved in cell proliferation and invasion of pancreatic cancer by promoting decoy receptor 3 in pancreatic cancer [31], indicating that IRF1 played different roles in diverse cancers. In leukemia, IRF1 was inactivated in chronic and non-chronic myelogenous leukemia [32] and expressed at a similar level between normal individuals and AML patients, as reported in a small sample study [33]. In this study, the GEPIA database analysis showed that expression of IRF1 in AML patients was higher than that in normal individuals. In addition, we found that high expression of IRF1 was correlated with poor prognosis in AML patients from the LinkedOmics and TCGA databases analyses, and the result was also validated in our study. Then, we analyzed IRF1 in the chemotherapy-only and transplant subgroups using the TCGA database. The high expression of IRF1 showed a significant impact on OS; but in the transplant subgroup, IRF1 expression was irrelevant to survival, which showed that the IRF1 may serve as a potential prognostic marker for AML, and the transplants could reverse their poor prognosis. However, in the multivariate analysis, high expression of IRF1 was an independent unfavorable factor for OS and it was not significant from the TCGA database analysis. It is possible that the number of patients undergoing transplantation at our center is too small to reverse the prognosis of IRF1 with the treatment of transplantation. Based on the analysis of clinical data, high expression of IRF1 occurred in patients whose ages were over 60 years, and the high expression was positively correlated with high PB blasts percentage, high LDH level and less inv(16) karyotype. Guzman et al. documented a consistent overexpression of IRF1 in the primary AML cells with a primitive phenotype, indicating that IRF1 may play a role in the biological function of early leukemogenic cells [34]. One interpretation of these results may be that leukemic cells can undergo the onset of the apoptotic process but fail to complete the process of apoptosis. As is mentioned above, IRF1 is a factor for poor prognosis, which may be associated with leukemic stem/progenitor cells, and it may serve as a biomarker of risk stratification in AML, while treatment with transplants could reverse its poor efficiency in determining the prognosis.
In this study, the GEPIA database analysis showed that the expression level of IRF6 in AML patients was lower than that in normal donors. And the LinkedOmics and GEPIA databases showed that the expression of IRF6 was not correlated with the prognosis of AML patients. Next, we surprisingly found that high expression of IRF6 resulted in poor prognosis in the chemotherapy-only subgroup from the TCGA database analysis. However, in our center, we could not detect the IRF6 mRNA among our AML patients using quantitative RT-PCR. In addition, we found that the expression of IRF6 was significantly lower than the values from the TCGA database or even undetected in some cases. Furthermore, it was not expressed in the AML cell lines from the EMBL-EBI database analysis [35]. Accordingly, we considered that the expression of IRF6 may be too low to be detected and may cause some statistical bias.
IRF7 is expressed at low levels in most cells and is strongly induced by the RIG-I signaling pathway, including RIG-I, MDA5, IPS-1, TBK1, and IKKε [36], and play a distinct role in the induction of type I IFNs genes, whose aberrant expression is associated with many diseases such as cancers and autoimmune disorders [14, 37]. Another research revealed that the IRF7 and RIG-1 pathways activated by the AFAP1-AS1 gene were involved in migration and invasion in non-small cell lung cancer [38]. However, the study about IRF7 was rarely reported in leukemia. In our study, the GEPIA database analysis showed that the expression of IRF7 in AML patients was higher than that in normal donors. High IRF7 expression was significantly related to shorter survival in AML patients from the LinkedOmics and TCGA databases analyses, which was also validated from our data. Furthermore, the high expression of IRF7 in AML patients was characterized as more poor-risk stratification factors due to higher LDH levels, more complex karyotype, more RUNX1 mutation, and less NPM1 mutation [2–3], indicating that IRF7 may promote the progression of AML by interacting with some changes of genes and chromosomes. Multivariate analysis showed that IRF7 was an independent unfavorable prognostic factor both from the data of our center and the TCGA database. In addition, the reason for the association between poor prognosis and IRF7 may be related to the tumor microenvironment [39]. Next, we found that IRF7 in both transplant and chemotherapy-only subgroups from the TCGA database was irrelevant to survival. One explanation may be that IRF7 is correlated with EBV infection [37], which is easy to affect the survival of patients after transplantation. Thus, the IRF7 can be a valuable prognostic factor for risk stratification in AML. Further research with chromosome abnormalities and tumor microenvironment is needed to improve the survival in patients with high IRF7 expression.
According to a previous study, IRF9 was one of the components of interferon-stimulated gene factor 3 (ISGF3), essential for the antiviral response mediated by IFN-α/β and IFN-γ and promoted the tumor suppressor p53 gene expression through type I IFNs [13, 14, 40]. In our study, the GEPIA database analysis revealed that the IRF9 gene expression was increased in AML patients compared with healthy donors and was correlated with poor prognosis. Next, analyses of overall patients from the LinkedOmics database, TCGA database, and our center showed that the expression of IRF9 was not correlated with the prognosis of AML patients. However, from the GEPIA database and chemotherapy-only subgroup of the TCGA database, increased IRF9 expression was involved in poor survival in AML. Considering this outcome, IRF9 could regulate AML cell lines by SIRT1-p53 [41], and one explanation of the result may be that IRF9 regulated not only the wild tp53 gene but also the mutation of the tp53 gene. We further analyzed the relationship between IRF9 and TP53 mutation using TCGA database. The result showed that patients who over-expressed IRF9 was correlated with TP53 mutation (P = 0.042, no data shown). Another explanation may be that the sample size of the chemotherapy-only subgroup (n = 78) and the GEPIA database (n = 106) is relatively small. Hence, a study with a larger sample size of TP53 mutation is needed to clarify the prognostic role and to further elucidate the mechanism of IRF9 in AML.