P425: a multidimensional and spatial analysis reveals distinct immune phenotypes and secondary lymphoid organ-resembling structures in the bone marrow of pediatric acute myeloid leukemia

Background: Pediatric acute myeloid leukemia (AML) is a cancer with a particularly low mutational burden in comparison to other pediatric and adult cancers and therefore, thought to be a poor candidate for T cell-engaging immunotherapies. Interestingly, recent evidence suggests that considerable variation in immune cell infiltration exists among immunologically ‘cold’ cancers, with subsets of patients showing a ‘hot’ tumor microenvironment and improved responses to immunotherapy in comparison to their ‘cold’ counterparts. However, little is known about the immune compartment in the bone marrow (BM) of pediatric AML patients and therefore, it is unclear whether children with AML may be successfully treated with T cell-engaging immunotherapies. Aims: To perform a multidimensional characterization of the tumor immune microenvironment in newly diagnosed children with AML and non-leukemic controls. Methods: For this study, we obtained 53 formalin-fixed and paraffin-embedded bone biopsies from a representative cohort of pediatric patients with treatment-naïve de novo AML (n=43) and from age- and sex-matched pediatric patients with treatment-naïve early-stage rhabdomyosarcoma without malignant BM infiltration (non-leukemic controls, n=10). To characterize the immune compartment and its spatial organization in the BM, we employed (multiplex) immunohistochemistry (IHC) alongside immune-related gene expression profiling and spatial transcriptomics. Results: Using IHC, we found a decreased abundance of both the overall number of T cells and cytotoxic T cells in the BM of pediatric AML patients in comparison to non-leukemic controls (P=0.006 and P=0.003, respectively). However, the extent of overall T cell and cytotoxic T cell infiltration could differ up to 90-fold between individual AML patients, with several patients showing remarkably high BM T cell infiltration. These immune-infiltrated samples showed gene expression indicative of tumor-specific T cell responses and a significantly lower abundance of anti-inflammatory macrophages in comparison to immune-depleted samples (P<0.001). Intriguingly, KMT2A-rearranged immune-infiltrated samples harbored large networks of T and B cells that were devoid of AML cells despite full-blown leukemia in other areas of the BM. Using spatial transcriptomics, we dissected the composition of these lymphoid aggregates and revealed localized anti-tumor immunity in the BM of AML. In comparison to tumor areas, these aggregates showed a higher abundance of activated cytotoxic T cells, memory B cells, and plasma cells, along with class-switched immunoglobulin gene expression, suggesting the presence of secondary lymphoid organ-resembling structures in the BM of AML. Summary/Conclusion: Among pediatric AML patients, we identified a subset with remarkably high T cell infiltration and a relatively low abundance of anti-inflammatory macrophages in the BM. In addition, for the first time, we unraveled the composition of secondary lymphoid organ-resembling structures in the BM of AML, with implications for future immunotherapy approaches. For instance, the presence of these structures and associated anti-tumor antibody production may lead to novel targets for antibody-based therapies. Moreover, the tumor-specific nature of T cells in these lymphoid structures may pave the way for new adoptive cell therapies with tumor infiltrating T cells. Altogether, our work raises the possibility that a subset of pediatric AML patients may benefit from T cell engaging immunotherapies and encourages further study of these lymphoid structures in the context of immunotherapy in AML. Keywords: Children, KMT2A, Antigen presentation, Adoptive immunotherapy