Immune Cell Profiling in CML Bone Marrow By Multiplex Immunohistochemistry

Background

In most solid tumors, CD8+ cytotoxic T-cells and type 1 T-helper cells are associated with a positive prognosis, but a strong immunosuppressive microenvironment may hamper their effectiveness. This notion has contributed to the development of new immune-activating therapies, such as immune checkpoint inhibitors. Although having demonstrated long-term remissions in many different solid tumor types, immune checkpoint inhibitors have not been evaluated comprehensively in hematological malignancies. In this study, we aimed to characterize the cellular and molecular immunological profiles of chronic myeloid leukemia (CML) patients' bone marrow (BM) samples.

Methods

BM biopsies were taken at the time of diagnosis from chronic phase CML patients (n=57) treated in the Helsinki University Hospital during years 2005-2015. We used non-leukemic (NL) BM biopsies (n=10) as controls. Using hematopathologic expertise, we constructed tissue microarray (TMA) blocks from duplicate BM spots characterized with high leukemic cell infiltration.

We stained TMA slides using multiplexed immunohistochemistry (IHC) combining fluorescent and chromogenic staining allowing detection of up to six markers and nuclei simultaneously. Marker panels included T and B-lymphoid (CD3, CD4, CD8, CD20), myeloid dendritic (CD11c, BDCA-1, BDCA-3), macrophage (CD68, pSTAT1, c-MAF), natural killer cell (CD3 and CD56) and leukemia cell (CD34) markers. In addition, we examined immune checkpoint molecules (PD1, CTLA4, OX40, LAG3, TIM3) and their ligands in leukemic cells (HLA-G, PD-L1, PD-L2, HLA-ABC), as well as activation markers (CD25, CD27, CD57, Granzyme B and CD45RO). We analyzed leukemia patients' immune checkpoint expression profiles quantitatively using the image analysis software Cell Profiler and cell analysis software FlowJo and compared results with NL BMs' immune cell profiles.

Results

The proportion of CD3+ T cells of all cells was significantly higher in CML BM vs. NL BM (median 6.0% [interquartile range (IQR) 3.6-10.7] vs. 2.1% [IQR 1.5-4.5], p=0.001). There was no significant difference in CD8+ cytotoxic T cell levels, but CD4+ helper T cells were 8-fold more abundant in CML as compared to non-leukemic BM (p<0.0001).

The proportion of both memory CD45RO+CD8+ T cells (62.2% [IQR 47.4-69.8] vs. 47.3% [IQR 27.9-56.2] of CD8+ T cells, p=0.03) and memory CD45RO+CD4+ T cells (61.8% [IQR 51.8-68.5] vs. 40.0% [IQR 25.6-57.9] of CD4+ T cells, p=0.004) were significantly higher in leukemic patients.

Although the proportion of PD1+CD8+ T cells did not differ between CML and NL BM, there was a significantly lower proportion of PD1+CD4+ T cells in CML BM vs. NL BM (25.1% [IQR 17.0-38.7] vs. 69.5% [IQR 50.7-77.9], p<0.0001). However, as the number of CD4+ T cells was increased in CML, the absolute number of CD4+PD1+ T cells of total cell population was 3-fold higher in CML BM than in NL BM (p=0.02).

Both the proportion of OX40+CD4+ T cells (42.3% [IQR 28.7-51.6] vs. 18.1% [IQR 13.2-22.9], p=0.001) and OX40+CD8+ T cells (42.6% [IQR 25.8-60.7] vs. 12.7% [IQR 5.0-15.8], p<0.0001) were increased in leukemic patients. Interestingly, also the proportion of OX40+PD1+CD8+ T cells (25.7% [IQR 15.4-36.4] vs. 11.9% [IQR 5.0-15.8], p=0.0019) was higher in CML samples.

Conclusion

Multiplex IHC allows detailed characterization of immune cell subtypes and their phenotypes in BM biopsy samples. Our data show significant heterogeneity in immune cell subsets between individual patients. The CML BM is characterized with an increase in CD3+ T cells, especially helper T cells and CD45RO+ memory T cells, when compared to non-leukemic BM. Phenotypically, OX40+PD1neg T cells and OX40+PD1+ cytotoxic T cells were elevated in CML patients. The analysis of other immune cell subclasses, including inhibitory immune cells, and the correlation of histologic findings to prognostic data are ongoing. Together, they will provide a detailed understanding of BM immune cell composition in CML.