An Emerging Prognosis Prediction Model for Multiple Myeloma: Hypoxia-Immune Related Microenvironmental Gene Signature

Abstract

Background: Multiple myeloma (MM) is a hematologic malignancy characterized by clonal proliferation of malignant plasma cells. Hypoxia and the immune microenvironment play vital roles in the carcinogenicity and evolution of MM. In this study, we attempted to construct a microenvironmental gene marker associated with hypoxia immunity for effective prognostic prediction and risk stratification. Methods: MM-related datasets (GSE136324, GSE47552 and GSE136337) were downloaded from Gene Expression Omnibus (GEO). A Kaplan-Meier (K-M) survival curve was applied to excavate survival differences among different hypoxia clusters, and the hypoxia state was divided into high hypoxia and low hypoxia groups according to the prognosis. The expression data algorithm was used to estimate the immune score of each MM patient. Based on the optimal truncation immune score value, MM patients were divided into high and low immune score subgroups. The R packages were applied to identify the risk differentially expressed genes (DEGs) and protective DEGs. Univariate Cox and Least Absolute Shrinkage and Selection Operator (LASSO) regression analyses were applied to identify prognosis-related genes. Receiver Operating Characteristic curves and K-M survival curves were applied to verify the validity and agility of the prognostic signature. The cell type identification by estimating relative subsets of RNA transcripts algorithm was applied to analyze immune cell infiltration. Quantitative real-time PCR was used to test the signature in vitro. Cell proliferation was measured by a CCK-8 assay. The levels of apoptotic rate were analyzed by the flow cytometer. Cell migration and invasion ability were estimated by Transwell membranes and Matrigel. NOD/SCID mice were divided into four groups: normoxia control (CTL), normoxia shDDIT4 (CTL+shDDIT4), intermittent hypoxia control (IH), and IH shDDIT4 (IH+shDDIT4). CTL had an oxygen concentration of 21% while IH had a nadir of 6%~8%. Results: A total of 867 MM samples were classified according to hypoxia status and immune score and were classified into Hypoxia.low/Immune.high, Hypoxia.high/Immune, lowand mixed groups. K-M survival analysis showed patients in the Hypoxia.low/Immune.high group had a better prognosis than others. Overlapping the three groups of risk DEGs and protective DEGs. There were 472 protective DEGs and 205 risk DEGs with supporting evidence from hypoxia, immune and combing hypoxia/ immune groups. In GSE47522, 2633 DEGs were screened between normal and MM. 81 key hypoxia-immune-related genes were overlapped by the Venn diagram between MM and normal DEGs, protective DEGs and risk DEGs. Using GSE136324 as a training cohort, a univariate Cox regression analysis was performed on 81 key genes. We identified 44 hypoxia-immune-related prognostic DEGs statistically related to the overall survival time of MM patients. Then, an 8-gene signature (including CHRDL1, DDIT4, DNTT, FAM133A, MYB, PRR15, QTRT1, and ZNF275) was identified using the LASSO regression algorithm. MM patients were divided into high-risk and low-risk groups based on a threshold of the median risk score in the prognostic value of the 8-gene signature. K-M curves suggested that the MM patients in the low-risk group showed a statistically longer overall survival than those in the high-risk group. The AUC of 1 year in the training cohort was greater than 0.7. Univariate and multivariate Cox regression analyses showed that the risk score could be an independent prognostic risk factor. Moreover, we revealed that the proportions of 17 types of immune cells were significantly different between the two risk groups, . In the high-risk group, PD-L1 expression levels were higher than in the low-risk group. Finally, we found knockout of DDIT4 inhibited tumor formation in vitro under hypoxia. Further, the survival rate of the IH+shDDIT4 group was higher than that of the IH group. Moreover, both tumor volume and tumor weight in the IH+shDDIT4 group were reduced. Conclusions: We established a hypoxia-immune related prognostic signature. It could predict the prognosis of MM well and was related to the tumor microenvironment of MM. The downregulation of DDIT4 inhibited tumor formation and development in vitro and vivo under hypoxia. Meanwhile, our results suggest that the immune checkpoint inhibitors, PD-L1 inhibitors, may be helpful in the treatment of high-risk patients.