Immune checkpoint gene VSIR predicts patient prognosis in acute myeloid leukemia and myelodysplastic syndromes

Abstract

Background: Immune checkpoint proteins play critical functions during the immune response to cancer and have been targeted by immune checkpoint blockade therapy. V-domain Ig suppressor of T cell activation (VSIR) is one of these immune checkpoint genes and has been investigated extensively in recent years due to its conflicting roles in cancer immunity. Specifically, in acute myeloid leukemia (AML), the prognostic value of VSIR is debated.

Results: In both patient tumor samples and cancer cell lines we find that VSIR has the highest expression in AML out of all cancer types and, in AML, has the highest expression out of all other immune checkpoint genes. Survival analysis indicated that AML patients with higher VSIR expression have significantly shorter survival than those patients with lower expression, even within established AML subgroups (e.g., FAB subtypes). Importantly, VSIR expression is predictive of progression from myelodysplastic syndromes (MDS) patients into AML, suggesting its potential role during the very early stage of AML development and progression. In addition to AML, VSIR also demonstrates prognostic values in other cancer types, including multiple myeloma and mesothelioma.

Conclusion: In summary, our analyses revealed the prognostic value of VSIR and its potential as a target for immunotherapy, especially in AML.

Keywords: VSIR; AML; MDS; prognosis.

FIGURE 1

VSIR has highest expression and is prognostic in AML. (A) VSIR has the highest expression out of all cancer types in both the TCGA and CCLE datasets. (B) VSIR has the highest expression out of other known immune checkpoint genes in AML samples for both the TCGA and CCLE datasets. (C) Within TCGA AML samples, VSIR expression has the highest statistical significance and hazard ratio when predicting survival compared to other immune checkpoint genes. (D) Kaplan–Meier plot showing that TCGA AML patients with higher‐than‐median VSIR expression have significantly worser prognosis. (E) Forest plot validating the prognostic value of VSIR expression after considering the contributions of other molecular and genetic variables such as cytogenetic risk and FAB subtype in a multivariate Cox proportional hazards model.

FIGURE 2

Validation of VSIR prognostic value in GSE14468. (A) Kaplan–Meier plot shows that patients with higher VSIR expression have a significantly worse prognosis. (B) Forest plot showing that VSIR expression remains significant in predicting prognosis after adjusting for molecular and genetic variables like FAB score and cytogenetic risk. (C) VSIR expression is different depending on the FAB subtype (ANOVA p‐value = 3 e‐39). (D–F) AML patients with the FAB M1, M2, and M4 subtypes, respectively, have significantly poorer prognoses if they have high VSIR expression. (G) VSIR expression is different depending on cytogenetic risk (ANOVA p‐value = 4 e‐17). (H) AML patients with an intermediate cytogenetic risk have a significantly poorer prognosis if they have high VSIR expression.

FIGURE 3

VSIR association with the genomic landscape. (A) Volcano plot showing that NPM1 mutation is the only common genomic mutation in AML that has significantly different VSIR expressions between the group of patients with the mutation versus those without it. (B, C) Patients with the NPM1 mutation have significantly higher VSIR expression versus those without the mutation in both the TCGA (Wilcoxon p‐value = 1 e‐4) and GSE14468 (Wilcoxon p‐value = 1 e‐23) datasets. (D, E) Kaplan–Meier plots showing that patients are significantly dichotomized by their VSIR expression if they are NPM1 WT, but not if they have the NPM1 mutation, in both the TCGA and GSE14468 datasets.

FIGURE 4

VSIR expression in single cells associated with leukemia. (A) VSIR expression in various hematopoietic cells of different cell lineages. Myeloid cells appear to have the highest expression of VSIR. (B) Single‐cell RNA sequencing data show that monocytes have the highest nonzero percent expression of VSIR. (C) Boxplots showing the distribution of VSIR expression for cDC, GMP, HSC, monocytes, progenitor cells, and pro‐monocyte cells based on whether the cell is healthy or cancerous. Malignant cells have consistently higher VSIR expression than normal cells. (D) VSIR expression correlation with monocyte marker genes such as ITGAM and CCR2. Any point above the horizontal red line has a significant nonzero Pearson correlation coefficient (p‐value <0.05).

FIGURE 5

Prognostic value of VSIR in other blood cancer types. (A) Barplot showing the meta‐p values when predicting survival using VSIR in multiple myeloma, AML, CLL, and other blood cancers using the PRECOG dataset. (B, C) Kaplan–Meier plot showing that patients with higher VSIR expression have significantly better prognosis for both multiple myeloma and CLL, respectively. (D) Volcano plot showing that VSIR expression has high statistical significance when predicting survival in AML compared to other cancer types. (E, F) Kaplan–Meier plot showing that patients with higher VSIR expression have significantly better prognosis for both mesothelioma and cervical squamous cell carcinoma, respectively.