Flotetuzumab (FLZ), an Investigational CD123 x CD3 Bispecific Dart® Protein-Induced Clustering of CD3+ T Cells and CD123+ AML Cells in Bone Marrow Biopsies Is Associated with Response to Treatment in Primary Refractory AML Patients

Introduction

The infiltration of immune cells into tumors has been associated with therapeutic effects in preclinical models and patients with cancer. In AML, we have previously reported that immune infiltrated TME is predictive of failure to cytotoxic chemotherapy, but associated with response to immunotherapy, specifically FLZ (Uy ASH 2018, Rutella ASH 2018). Furthermore, FLZ also affects immune infiltration in the TME (Rutella ASH 2018). NK cells play an important role in AML control (Ruggieri Science 2012). FLZ (MGD006/S80880) is a humanized DART^® molecule that bridges CD123 on AML with CD3 on T cells and mediates anticancer activity via T-cell activation and cytolytic activity against the bound cancer cell. While this is well described in vitro, little evidence of this interaction is available in vivo.

Methods

Patients (pts) were treated on the recommended phase 2 dose (RP2D) of FLZ (multi-step lead-in dose followed by 500ng/kg/day, in 28-day cycles). We studied the bone marrow (BM) tissue samples for 6 primary refractory pts at baseline and after treatment. Response assessment was performed at day 25±3 days of each cycle. Serial BM samples were evaluated using 2 different staining panels (PD-L1, FoxP3, CD8, CD3, CD103 / CD123, CD3, CD57, CD16) on consecutive slides. Slides were stained using a Leica BondRx autostainer and fluorescence imaged using a Polaris Vectra 3 and analyzed using inForm software. A density-based clustering algorithm developed and run in QuPath was used to quantify CD3+ T cell clusters.

Results

Six pts with primary refractory AML were included in this report. Pts were heavily pretreated (median prior lines of therapy was 3, range 2-9), and had adverse cytogenetic risk (ELN 2017). Three pts had a complete remission (CR) after 1 cycle of therapy (CR, CRh, CRi), two went on the receive allogeneic stem cell transplant (HSCT). In baseline BM samples, CD3 and CD8 cell infiltrates were higher in CR vs non-responders (CD3+ 18.3% ±6.9 vs 9.3% ±1.8; CD8+ 9.4% ±3.5 vs 4.8% ±1.2; mean±SEM). Two of the three CR patients, who underwent HSCT, developed clusters (Figure 1) in their on-treatment biopsies with 65 and 22 clusters of an average of 34 and 17 T cells per cluster, respectively. All clusters in CR pts were found on or adjacent to CD123+ cells. The BM biopsy of the CR pt with no detected clusters had no unequivocal evidence of residual/recurrent leukemic blasts. This pt had their dose interrupted early due to non-treatment related AE (infectious complication) and did not receive a full cycle of treatment; the response was transient and the pt relapsed shortly thereafter. NK cells (CD57+CD16+) were increased in post treatment biopsies of CR vs non-responders (0.93 ±0.31 vs 0.27 ±0.13; mean±SEM) with the largest fold increase in CR (28 vs 9). Lastly, post treatment biopsy PD-L1 expression was higher in non-responders than CR (23% vs 16%) with non-responders exhibiting the largest fold change in total PD-L1+ cells (10.9 vs 2.2).

Summary

Consistent with its proposed mechanism of action, these data highlight for the first time, the dynamic induction of an increase in T-cell infiltration, and clustering around CD123 AML cells in the bone marrow microenvironment of two AML patients that responded to FLZ. In pts with resistance to FLZ (non-responders) PD-L1 induction was significantly higher indicating that in some pts treatment with sequential check point inhibitor could obviate this mechanism of resistance A trial combining FLZ with sequential administration of a PD-1 inhibitor (MGA012) is currently recruiting pts.

Figure 1. Baseline and on-treatment IHC of BM biopsies of a FLZ-treated CR pt showing cluster formation following treatment.

View largeDownload slide

View largeDownload slide

Close modal