Loss of CD20 Expression and Exhaustion of Effector Cells Limit ADCC in CLL Patients Treated with Rituximab.

Abstract 1610

Poster Board I-636

A major mechanism how the chimeric anti-CD20 monoclonal antibody rituximab (RTX) depletes B-cells is antibody-dependent cellular cytotoxicity (ADCC). ADCC has been modeled in-vitro and in mouse models. However, investigations on ADCC directly in patients treated with RTX are scarce. Recent efforts have focused on improving ADCC through modifications in the Fc binding portion of novel antibodies or through stimulation of effector cell functions with GM-CSF. A more detailed understanding of ADCC as a therapeutic process is needed to optimize such strategies and to identify biomarkers of improved efficacy. Here we report a comprehensive analysis of ADCC in previously untreated CLL patients during the first two RTX infusions (375mg/m2) given in combination with fludarabine every 4 weeks. Following the initial infusion of RTX the absolute lymphocyte count (ALC) decreased by a median of 74% at 2h, followed by a partial recrudescence of cells so that by 24h the median decrease in ALC reached 39% (n=11). ADCC is mediated by effector cells that include NK cells, monocytes/macrophages, and granulocytes. First, we investigated changes in NK cell function: consistent with NK cell activation we found an increase in CD69 at 2, 6 and up to 24h (median 4.2-fold, p=0.005, n=10) after RTX administration and increased expression of the degranulation marker CD107a/b (median 1.9-fold, p<0.001, n=5) and down-regulation of perforin expression (median decrease 63%, p<0.001, n=5) at 4h from treatment start. Activation of NK cells is triggered by the engagement of CD16/FcγRIIIa by RTX coated CLL cells. Interestingly, CD16 expression on NK cells was rapidly lost, already apparent at 2h and maximal at 6h from the start of the RTX infusion (median decrease 82%, p=0.02, n=10) and was not completely recovered by 24h. We also found a significant decrease in expression of CD16 on granulocytes (78%, p<0.001, n=5) but an increase in monocytes (3.9-fold, p<0.001, n=5). In addition to loss of CD16, we found that the cytotoxic capacity of the effector cells was rapidly exhausted: in an oxidative-burst assay, monocytes showed a significant decrease in the production of reactive oxygen species 4h after initiation of RTX infusion (median 60% decrease, p=0.043) and at 6h from the start of the RTX infusion NK cell-mediated killing of K562 target cells was reduced by half (p<0.001, n=3). Interestingly, both the acute reaction to RTX infusions that manifest as a cytokine release syndrome and changes in effector cell function peaked during the first hours of the RTX infusion. We hypothesized that this might be due to the process of CD20 shaving, a rapid and pronounced decrease of CD20 cell surface expression modeled in-vitro and in mice as the result of a mechanism called trogocytosis that relies on the direct and rapid exchange of cell membrane fragments and associated molecules between effectors and target cells (Beum, J Immunol, 2008). First, we used western blot analysis of total CD20 protein in CLL cells and found a rapid loss of CD20 that was apparent already at 2h resulting in virtually complete loss of expression at 24h. Next, we used ImageStream technology to directly visualize ADCC interactions in-vivo. We indeed detected transfer of CD20 from CLL cells to NK cells and monocytes, resulting in complete CD20 loss in circulating CLL cells. While we detected transfer of CD20 into both cell types, monocytes were much more engaged in trogocytosis than NK cells. Consistently, 4h post RTX infusion we found a significant increase in intracellular RTX in granulocytes and monocytes using intracellular staining for human IgG. CD20 shaving appears to be of particular importance given that immunohistochemical analyses revealed that persistent disease in the bone marrow aspirates after 4 cycles of RTX treatment was mostly CD20 negative. Collectively, our results identify loss of CD20 from CLL cells by trogocytosis and exhaustion of immune effector mechanisms as limitations for anti-CD20 immunotherapy. These data identify possible avenues for improving CD20 mediated immunotherapy and characterize endpoints on which different anti-CD20 antibodies can be compared. Given that trogocytosis appears to be a common occurrence our findings likely have general importance to immunotherapy of hematologic malignancies.