The Composition of Macrophages Has Prognostic Impact in Diffuse Large B-Cell Lymphoma By Spatial Omic Studies

Diffuse Large B-cell lymphoma (DLBCL) has a diverse response to treatment. Better understanding of the immune microenvironment may identify novel therapeutic strategies and improve stratification of patients. Lately, it has been shown that histological features not previously measured, such as distance between specific immune cells and tumor cells impact the biology and clinical outcome of disease. Such spatial metrics can be used to describe distinct local cell neighbourhoods in the tumor immune microenvironment (TIME). Thus, spatial metrics can provide a deeper understanding of the complexity of the disease. Spatial metrics are also expected to contribute to prognostic value beyond measurement of degree of immune cell infiltration that commonly is used today to assess the TIME and may provide pathologists with important novel parameters to assess clinical risk, orthogonal to the established patient and genetic risk factors.

With the aim to identify relevant immunotherapeutic strategies for high risk DLBCL, we have applied spatially guided transcriptional profiling in combination with image analysis of T-cells and subsets of myeloid cells.

The population-based cohort includes tissue from 650 patients diagnosed with DLBCL during 2000-2014 in southern Sweden. Tissue was mounted in tissue microarrays and stained using routine immunohistochemistry for diagnostic markers such as CD20, CD3, CD5, CD10, BCL-2, MUM1, BCL-6, MYC, P53 and EBNA and went through pathology validation. Information on clinicopathological parameters and treatment regimen were collected, and tissue from 448 patients where treatment information was available were selected for digital spatial profiling analysis (Nanostring Inc.). The Cancer Transcriptome Atlas panel with ~1850 transcripts was applied, and analysis was guided by staining CD20+ cells, CD163+ cells and CD3+ cells. This allowed transcriptional profiling of tumor cells (445 patients), CD163+ cells (286 patients) and CD3+ cells (396 patients) separately in each patient. In addition, multiplex staining of CD163, CD11c and CD20 was performed to assess differences in infiltration of myeloid subsets.

Pathology review, that was performed through digital pathology, revealed that 44% of patients were GCB and 56% non-GCB, using Hans algorithm. High risk patients, based on IPI constituted 13.7 % of the full cohort, and 12.9% of the selected cohort that was used for transcriptional profiling. Molecular high-risk group based on degree of MYC and BCL-2 overexpression, so called double protein expressors, was 20 %. The double expressors were in 55% of the cases classified as non-GCB. P53 overexpression (>20%) was found in 11.7% of patients.

Survival analysis showed that patients with double expression of MYC and BCL-2 have inferior 5-year overall survival (OS) and p53 expression alone was associated with worse 5-years OS also when adjusted for gender and age (HR = 3.8, p-value = 0.014).

Image analysis showed expression of CD163 and CD11c to be mutually exclusive, and divide patients into distinct prognostic sub-groups, independent on degree of p53 overexpression. Survival analysis shows that CD11c-dominated TIMEs were associated with favourable 5-year OS (HR = 0.42, p-value = 0.0047). To further understand the molecular features associated with differences in TIME, spatially guided analyses of tumor and immune cells are ongoing.

Furthermore, the connection between tumor-cell biology and established risk-factors such as p53 overexpression and double protein expressors are explored.

In summary, combined image- and spatially guided analysis of lymphoma and individual immune-cells provide orthogonal risk stratification strategies for patients with DLBCL.