AUTHORS
Ross Firestone, MD/PhD, Mansour Poorebrahim, PhD2, Juan-Jose Garces, PhD3, Huihui Tang, PhD4, Kylee Maclachlan, MD, PhD5, Bachisio Ziccheddu6, Holly Lee, MD, PhD7, Tereza Novakova8, Tereza Sevcikova9, Roman Hajek, MD, PhD10, Veronika Kapustova11, Alexander M. Lesokhin, MD12, Paola Neri, MD, PhD13, Issam S Hamadeh, Pharm.D.14, Lawrence H Boise, PhD15, Constantine Mitsiades, MD, PhD16, Tomas Jelinek, MD, PhD17, Saad Z. Usmani, MD18, Nizar Bahlis, MD19, Francesco Maura, MD20
AFFILIATIONS
1 Memorial Sloan Kettering Cancer Center, New York, New York, United States
2 University of Calgary, Calgary, Alberta, Canada
3 Memorial Sloan Kettering Cancer Center, New York, New York, United States
4 Dana Farber Cancer Institute, Boston, Massachusetts, United States
5 Memorial Sloan Kettering Cancer Center, New York, New York, United States
6 Memorial Sloan Kettering Cancer Center, New York, New York, United States
7 University of Calgary, Calgary, Alberta, Canada
8 University Hospital Ostrava, Ostrava, Czech Republic
9 University Hospital Ostrava, Ostrava, Czech Republic
10 University Hospital Ostrava, Ostrava, Czech Republic
11 University Hospital Ostrava, Ostrava, Czech Republic
12 Memorial Sloan Kettering Cancer Center, New York, New York, United States
13 University of Calgary, Calgary, Alberta, Canada
14 Memorial Sloan Kettering Cancer Center, New York, New York, United States
15 Emory University, Atlanta, United States
16 Dana Farber Cancer Institute, Boston, Massachusetts, United States
17 University Hospital Ostrava, Ostrava, Czech Republic
18 Memorial Sloan Kettering Cancer Center, New York, New York, United States
19 University of Calgary, Calgary, Alberta, Canada
20 Memorial Sloan Kettering Cancer Center, New York, New York, United States
ABSTRACT
Fc receptor-like 5 (FCRL5) is a highly expressed surface protein on plasma cells. While its function remains poorly understood, it is emerging as an important target for the novel FCRL5 × CD3 bispecific antibody (BsAb) cevostamab (Cevo). However, responses are heterogeneous, with some patients experiencing primary refractoriness and others achieving prolonged remission. Mechanisms underlying this clinical variability remain largely unknown.
To define mechanisms of resistance to Cevo in relapsed/refractory multiple myeloma (RRMM) patients, we performed whole-genome sequencing (WGS) on pretreatment bone marrow samples from 27 patients enrolled in Cevo clinical trials. The median progression-free survival (PFS) of the entire cohort was 16 months, and 12 patients showed primary refractoriness, progressing within 6 months of therapy initiation.
Although the FCRL5 locus resides on chromosome 1q, 1q gain did not predict improved PFS with Cevo (p=0.1). Using the CoMMpass dataset (n=767), we demonstrated that FCRL5 expression increases with copy number gain, but that additional copies selectively enriched soluble FCRL5 isoforms rather than the membrane-bound, Cevo-engageable form.
Four key genomic events were associated with shorter PFS: structural variants (SVs) involving MYC (p=0.01) and TXNDC5 (p=0.016), focal deletion of CCSER1 (p=0.0099), and TP53 loss (p=0.04). Patients without any of these genomic events experienced excellent outcomes, with 8 of 12 still in remission after 2 years. While TP53 loss and CCSER1 deletion are associated with poor outcomes across other immunotherapies, including anti-BCMA BsAbs, CAR T cell therapies, and anti-CD38 monoclonal antibodies, SVs in MYC and TXNDC5 are typically prognostically neutral in MM. However, FCRL5 is known to suppress ERK signaling, suggesting it may function as a tumor suppressor rather than an oncogene. As MYC is downstream of ERK, RRMM harboring MYC SVs may bypass FCRL5-mediated growth inhibition, enabling unchecked tumor proliferation. Notably, in an in vitro genome-scale open-reading frame (ORF) screen, MM1.S cells transduced with FCRL5 ORF were depleted (relative to the control distribution of ORFs), providing further support for an anti-proliferative role of this gene.
One Cevo patient remained primary refractory in the absence of the genomic drivers described above and harbored a rare germline variant, N427K. This variant was also accompanied by a deletion potentially disrupting the wild-type allele. Functional validation in K562 cells demonstrated that the N427K variant did not alter Cevo binding, but instead drove proteasomal degradation of FCRL5, with ubiquitination approximately twice that of wild-type FCRL5. This progressively reduced surface FCRL5 density over prolonged culture and subsequently impaired Cevo-mediated cell killing, supporting N427K as a mechanism of Cevo failure. Cell surface FCRL5 was not restored upon proteasome inhibition with either bortezomib or carfilzomib, though carfilzomib did lead to increased total FCRL5 expression. Using gnomAD, we found that N427K was the third most common germline missense variant in FCRL5, with an allele frequency of 0.005% in individuals with European ancestry, but with ~1100-fold relative enrichment in individuals with African ancestry (~5.5% allele frequency), suggesting ancestry-linked germline variants may contribute to disparities in Cevo outcomes.
To further investigate events potentially affecting Cevo-mediated killing, we extended our analysis to a larger cohort of newly diagnosed MM (n=907) and additional RRMM (n=411) patients, identifying 15 tumors carrying 14 different nonsynonymous mutations in FCRL5 (9 NDMM, 6 RRMM). Next, we cloned 6 of these FCRL5 variants, those encompassing residues at or near the Cevo-binding epitope, into K562 cells. Two of these variants, S623fs and Y817C, led to reduced FCRL5 expression and Cevo binding in vitro; however, only Y817C impaired Cevo-mediated cell killing when co-cultured with healthy donor T cells.
These findings establish a genomic and functional framework for predicting primary refractoriness to Cevo, suggesting FCRL5 acts as a tumor suppressor gene. Furthermore, a recurrent germline FCRL5 variant enriched in patients of African ancestry may also affect response. Pretreatment genomic profiling may enable patient selection for FCRL5-directed therapy and inform rational combination strategies aimed at overcoming resistance.
DISCLOSURES
Ross Firestone, MD/PhD: No relevant conflicts of interest to declare.