Introduction: Acute leukemia is the most common malignancy in children and develops within the bone marrow. Consequently, bone marrow derived T cells of leukemia patients can be defined as tumor infiltrating lymphocytes (TILs). Dysfunctional TILs have been described in several other malignancies. However, in pediatric patients the interaction between leukemic blasts and TILs remains largely unknown. In order to understand the impact of leukemic blasts on bone marrow T cells we profiled T cells in the bone marrow of pediatric leukemia patients by surface marker and transcriptome wide analysis.

Methods: First, artificial changes in marker expression due to cryopreservation and thawing were ruled out (n=5). Then, cryopreserved bone marrow samples from both pediatric patients with acute leukemia (n= 77; BCP-ALL: 18, TCP-ALL: 23, AML: 36) and age-matched healthy bone marrow donors (HD, n=23) were identified in a local biobank. Multicolor flow cytometry was performed to quantify co-inhibitory markers on CD4 and CD8 T cells in primary (n=49) and relapse leukemia samples (n=28). As we could not detect surface CTLA4 expression on T cells, CTLA4 was stained intracellularly. Additionally, RNA-Seq on sorted bone marrow derived CD8 T cells (n=48; TCP-ALL: 12, AML: 20, HD: 16) was performed. Analysis of RNA-Seq data was based on Reads Per Kilobase Million (RPKM) normalization and False Discovery Rate (FDR, Benjamini-Hochberg) statistics. 172 differentially expressed genes were found when comparing bone marrow derived CD8 T cells from healthy donors (n=16) and leukemia patients (n=32) using the following criteria: RPKM>2 in both groups, fold change>2 and FDR<0.05).

Results: The frequency of bone marrow T cells was reduced in patients with acute leukemia in comparison with healthy controls (5.9% vs. 24.4%, mean values, p<0.001). This reduction was more pronounced in BCP-ALL than in AML (0.9% vs. 8.4%, p<0.001). LAG3 and CTLA4 protein expression of T cells was increased in leukemia patients vs. healthy controls (LAG3: CD4: 2.6% vs. 0.7%, p<0.001; CD8: 8.6% vs. 2.2%, p<0.001; CTLA4: CD4: 7.3% vs. 3.8%, p=0.001; CD8: 1.2 vs. 0.3%, p<0.001). For CD8 T cells, those findings could be confirmed by RNA-Seq of sorted CD8 T cells (LAG3: 60.4 vs. 23.3 (RPKM), FDR=0.0044; CTLA4: 28.7 vs 4.7 (RPKM), FDR=0.046). Equally, TIM3 on T cells showed higher expression in leukemia patients vs. healthy controls (CD4: 3.7% vs. 1.3%, p=0.002; CD8: 8.5% vs. 3.3%, p<0.001). However, the same analysis of RNA-Seq data on sorted CD8 T cells did not yield a significant difference (18.1 vs. 5.6 (RPKM), FDR=0.29). PD1 was the only surface marker found to be more highly expressed in relapse samples than in primary diagnosis samples than in healthy controls (CD4: 42.3% vs. 28.9% vs. 19.8%, p<0.001; CD8: 45.2% vs. 33.3% vs. 26.5%, p=0.002). For CD8 T cells, RNA-Seq did not recapitulate this finding as no significant difference of PD1 transcript abundancy could be observed between leukemia patients and healthy donors by RNA-Seq (21.4 vs. 16.9 (RPKM), FDR=0.92). Finally, RNA-Seq on sorted CD8 T cells showed a pronounced overexpression of genes that are involved in the cytotoxic granule machinery in leukemia patients indicating an increase of effector phenotype in those cells. Contrarily, genes crucial for T cell function and memory formation were significantly downregulated in CD8 T cells from leukemia patients.

Conclusion: By analyzing bone marrow samples from pediatric leukemia patients and healthy controls we confirm that bone marrow T cells of leukemia patients show signs of exhaustion compared to healthy individuals. Importantly, PD1 surface expression on T cells was identified as a marker that correlates with disease status (relapse > primary > healthy). A significant increase of exhaustion markers could be demonstrated both on protein and transcriptome level (LAG3, CTLA4) or on protein level only (TIM3, PD1). Moreover, we observed an increase of many elements of the cytotoxic granule machinery which is compatible with a loss of naïve/memory CD8 T cells. Additionally, genes essential for T cell memory formation were found to be downregulated in CD8 T cells from leukemia patients. These findings reflect an insufficient immune surveillance of pediatric leukemia by bone marrow T cells and may provide a rationale for future therapeutic interventions.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.