Antigen-dependent clonal sweep with sequential or combinatory anti-myeloma T cell engagers

Abstract

Abstract Bispecific T cell engagers (TCE) have led to significant therapeutic advances in multiple myeloma (MM), however with finite durability of response. Growing evidence implicates TNFRSF17 (BCMA) and GPRC5D antigenic escape as key mechanism of TCE failure. However, more data is needed to i) accurately define the true prevalence of antigen escape, ii) delineate the phylogenetic processes and mutational drivers of these events and iii) elucidate the role of sequential versus combinatorial targeting of myeloma antigens in mitigating immune evasion. To comprehensively profile the genomic events underlying TCE relapse, we collected serial bone marrow biopsies from MM patients (n=64) treated with anti-BCMA (n=38), -GPRC5D (n=23), or -FcRL5 (n=4) TCEs sequentially with and without anti-CD38 antibodies. Sorted CD138+ cells were subjected to bulk whole genome sequencing (WGS 100x), scRNA-seq and flow cytometry analysis. Among 33 cases of post anti-BCMA TCE relapse and excluding primary refractory disease, 67.7% demonstrated TNFRSF17 genomic alterations. TNFRSF17 events included i) BCMA extracellular domain (ECD) mutations + monoallelic TNFRSF17 deletion (n=6), ii) BCMA ECD mutations only (n=5), iii) TNFRSF17 biallelic deletion (n=8), and iv) TNFRSF17 monoallelic deletion (n=2). Detected TNFRSF17 mutations localized exclusively to BCMA extracellular domains and revealed recurrent mutational hotspots including p.Pro34del (n=6), p.Ser30Phe (n=3), p.Gln25ter (n=2), and single cases of p.Arg27Pro, p.Arg27Gln, p.Ser30del, p.Gln7ter, p.Ser48ter, and p.Thr32Ser. Notably, genomically divergent BCMA ECD mutant subclones were simultaneously detected in 4 cases including MM-17 (p.Pro34del + p.Ser30del), MM-51 (p.Gln25ter + p.Gln7ter + p.Ser30Phe), MM-52 (p.Gln25ter + p.Ser30Phe + p.Pro34del), and MM-79 (p.Ser30Phe + p.Pro34del). Similarly, antigenic drift involving GPRC5D through copy number deletions, SNVs/indels, or epigenetic regulation was observed in 68.4% of relapsed cases following talquetamab treatment. Dual antigen escape was observed in patients treated with sequential anti-BCMA and GPRC5D TCEs, as well as those who also received TCE in combination with anti-CD38 antibodies. Phylogenetic studies of these cases often revealed a clonal sweep driven by the TCE immunotherapeutic bottleneck with the emergence of single clone disease at progression. This is illustrated in patient MM-03, with a clonal TNFRSF17 p.Arg27Pro mutation detected at progression from elranatamab, followed by two phased mutations in GPRC5D including p.Asp239Asn (VAF 47%) and p.Trp237Ter (VAF 36%) on progression from talquetamab, daratumumab and pomalidomide (Talq-DP) in the immediate next line of therapy. Functional profiling confirmed the disruption of elranatmab and teclistamab binding to mutant BCMA p.Arg27Pro, and analysis of the GPRC5D p.Asp239Asn mutant demonstrated ~275 fold increase in talquetamab logIC50 relative to wtGPRC5D. Notably, this GPRC5D mutant clone retained the clonal TNFRSF17 p.Arg27Pro mutation consistent with a common single ancestral phylogeny and a clonal sweep post TCEs. Of interest, mutational signature analysis (mmsig) of CD138+ tumor cells prior to and post each TCE therapy demonstrated a dominant hyper-APOBEC signature (SBS2/SBS13 73%) consistent with its likely mutagenic contribution to antigenic escape. Double and triple antigenic loss involving BCMA, GPRC5D, and CD38 were also observed in patients (n=4) treated sequentially or in combination immunotherapies targeting these antigens. Serial WGS analysis consistently demonstrated an increase in mutational burden at relapse post TCE-based therapies. dN/dScv analysis was performed at progression to define drivers of TCE resistance. Surprisingly, this analysis revealed a strong positive enrichment only for nonsynonymous mutations of targeted antigens rather than that of common oncogenic drivers (with the exception of TP53 and NRAS), consistent with an antigen escape model of TCE clonal sweep. High prevalence of target antigen escape at progression is confirmed in this large cohort of TCE-treated MM patients. Importantly, multi-target antigen loss is also noted with sequential or combination immunotherapy-based regimens. These observations, marked by the absence of enrichment for hallmark oncogenic driver mutations at relapse but rather the emergence of ‘antigen-null’ clonal sweeps, support a finite, response adapted, TCE administration to limit immune sculpting.