Immune System Deregulation Predicts Outcome in Diffuse Large B Cell Lymphoma (DLBCL)

Background: Treatment failure after RCHOP is associated with poor outcome in DLBCL. Despite considerable immune system variation among individual patients, the diagnostic absolute monocyte (AMC) and absolute lymphocyte (ALC) counts can reproducibly separate DLBCL patients with different outcomes. In addition, polymorphisms in immune regulating cytokines are associated with NHL risk and cytokine deregulation can predict response to RCHOP. This suggests distinct patterns of immune system deregulation in DLBCL which may serve to guide prognosis and therapy.

Aim: To investigate immune cell and cytokine deregulation in DLBCL.

Methods: Diagnostic AMC and ALC from 116 DLBCL patients and 111 healthy stem cell donors were assessed. 20 cytokines were measured in duplicate in serum samples from 34 DLBCL patients with known outcome and 11 health donors using the mesoscale discovery (MSD) electrochemiluminescence platform. The 10 most deregulated cytokines were measured in duplicate in 30 additional DLBCL patients. Peripheral blood mononuclear cells (PBMCs) were analysed by mass cytometry (MC) from 33 of these patients with a myeloid focused panel (29 markers) and from 27 with a T cell directed panel (28 markers). A standard control healthy donor sample was acquired with each MC batch. The OMIQ platform was used for gating and high dimensional data analysis.

Results: In DLBCL median AMC was higher (p<0.0001) and ALC lower (p=0.0036). The AMC and ALC prognostic score separated DLBCL into 3 groups with distinct outcomes (p<0.0001), independent of cell of origin (COO). 8/20 cytokines were significantly elevated in DLBCL (p<0.05). IL-1RA (p=0.025), IL-6 (p=0.0002), IL-10 (p=0.01), MIP-1a (p=0.003) and TNF (p=0.0056) were significantly more elevated in DLBCL patients who subsequently had relapse/refractory (R/R) disease within 12 months of RCHOP (n=15) compared to those with ongoing remission > 24 months (CR) (n=46).

After MC acquisition and normalization, PBMC populations were assessed as a percentage of CD45+ events by outcome (n=33). CD14+ monocytes were higher and CD3+ T cells lower at diagnosis in DLBCL with R/R disease (p<0.05). CD4+ T cells, CD8+ T cells, CD19+ B cells, NK, NKT and DC were not significantly different. Unsupervised clustering with FlowSOM followed by differential abundance analysis using edgeR was performed to dissect the PBMC populations above into subpopulations. This identified 3 classical monocyte clusters and 1 CD8+ T cell cluster associated with R/R DLBCL (-log10 FDR >1.3). The classical monocyte clusters all expressed the myeloid markers CD11b, CD31, CD33, SIRPa (CD172a) and CCR2 (CD192), the Fc gamma receptors CD32 and CD64 and the activation marker CD38. 2 clusters expressed HLA-DR and CD206 and were differentiated by CD13+ (cluster09) or CD40/CD86+ (cluster23). The other cluster had a suppressor phenotype with low HLA-DR and expressed CD163 and CD184 (cluster12). The differentiating T cell cluster was a CD45RA-CCR7- CD8 effector memory population (Tem) with an EM3 phenotype (CD27-, CD28-) and T-bet, cytotoxic marker (granzyme, perforin) and PD-1 expression (cluster10).

To identify co-deregulated immune cell / cytokine relationships in DLBCL contributing to R/R disease we used correlation analysis across PBMC populations, differentiating clusters and cytokine levels. CD14+ correlated positively with cluster9 (classical monocyte), cluster10 (Tem), IL-1RA, IL-6, and MIP-1a (p<0.05). CD3+ and CD4+, but not CD8+ correlated negatively with IL-1RA, IL-6 and MIP-1a (p<0.05). Cluster9 and cluster10 both correlated negatively with CD3+ and CD4+, but not with each other or cytokine levels. PBMC populations, differentiating clusters, and cytokine levels were not significantly different based on cell COO, except for IL-1RA which was elevated in non-GCB type (p=0.0148).

Conclusion: We find deregulated immune cell and cytokine production in DLBCL, which is associated with R/R disease and largely independent of COO. Immune profiling with MC identified 3 distinct classical monocyte clusters and 1 CD8 Tem cluster associated with subsequent relapse. These differentiating clusters likely reflect immune dysfunction and contribute to the prognostic nature of the AMC / ALC. Further ongoing work is assessing which immune are cells secreting the most deregulated cytokines, with the aim of identifying biomarkers and targetable pathways.