Pediatric acute lymphoblastic leukemia (ALL) is a cancer entity of minimal mutational load and low immunogenicity. The interaction of ALL cells with bone marrow (BM) T cells has not been investigated as a pathogenic driver or prognostic marker for pediatric ALL. We defined BM T cells of pediatric ALL patients as tumor-infiltrating lymphocytes (TILs) and investigated the prognostic relevance of co-stimulatory and co-inhibitory signals between ALL and BM T cells.


BM samples of 100 pediatric ALL patients were analyzed at time of initial diagnosis. T-cell subpopulations and expression of co-stimulatory and co-inhibitory molecules were defined by flow cytometry and correlated with clinical outcome of the patients. To investigate the role of TIM-3 for the interaction between T cells and leukemic cells, CRISPR/Cas9-mediated TIM-3 knockout (KO) was performed in primary T cells by ribonucleoprotein electroporation. T-cell activation and proliferation after contact with leukemic target cells were analyzed in TIM-3 KO cells and compared to wildtype T cells and T cells with retroviral TIM-3 overexpression. Interaction of T cells with leukemic target cells was induced by addition of anti-CD19/-CD3 bispecific T-cell engager (BiTE). Fold change (FC) of T-cell activation and proliferation was analyzed before and after co-culture. BM expression levels of known TIM-3 inducers were identified by RNA next generation sequencing of the bone marrow samples.


Multivariate analyses identified high TIM-3 expression on CD4+ BM T cells at initial diagnosis as strong predictor for relapse of pediatric acute lymphoblastic leukemia (relapse free survival (RFS) 94.6% vs. 70.3%). The risk to develop ALL relapse was 7.1-fold higher in the group of TIM-3 high expressing patients (n=37) compared to TIM-3 low expressing patients (n=37). Expression levels of known TIM-3 ligands and inducers in the bone marrow of the patients were analyzed by RNA next generation sequencing and compared between patients with high TIM-3 expression (n=12) and low TIM-3 expression (n=15) on BM T cells. Presence of known TIM-3 ligands HMGB1 (High-Mobility-Group-Protein B1) and Galectin-9 was confirmed, but expression levels did not show significant differences. Known TIM-3 inducers IL-2, -7, -15 and -21 were not expressed on RNA level indicating that another mechanism must be responsible for TIM-3 overexpression. In vitro experiments showed that the interaction with leukemic cells induces TIM-3 expression on the surface of T cells (mean TIM-3 expression 51.1% vs. 29.7% on T cells with vs. without addition of leukemic cells, n=3). To investigate the functional relevance of TIM-3 expression in pediatric leukemia, TIM-3 KO and overexpression was performed on primary T cells. TIM-3 KO T cells showed higher activation levels after co-culture with leukemic cell lines plus CD3-/CD19-specific BiTE compared to wildtype (WT) T cells (FC of CD69 surface expression 5.0 vs. 3.2, n=3). FC of anti-leukemic proliferation was impaired in TIM-3 overexpressing T cells compared to WT T cells (FC 1.6 vs. 2.3, n=3) whereas TIM-3 KO T cells showed a higher proliferation FC compared to controls (FC 6.5 vs. 2.4, n=3).


Our study identifies TIM-3 expression on CD4+ bone marrow T cells at initial diagnosis as a strong predictor for pediatric ALL relapse. TIM-3 expression is induced by interaction of T cells with leukemic cells and results in impaired anti-leukemic T-cell activation and proliferation. TIM-3-mediated T-cell inhibition represents a new mechanism of impaired immune surveillance in pediatric ALL and blockade of this axis may be of importance for future immunotherapy in ALL.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.