AUTHORS
Safiyyah Ziyad, PhD, Kristina Jordahl, PhD2, Carolyn Tarpey, MS3, Nicholas Stong, PhD4, Jesús G. Berdeja, MD5, Susan Bal, MD6, Omar Nadeem, MD7, Larry D Anderson Jr, MD, PhD8, Myo Htut, MD9, Mehmet Kocoglu, MD10, Tara K Gregory, MD11, Adriana Rossi, MD12, Thomas Martin, MD13, Wei-Ming Kao, MD, PhD14, Ethan Thompson, PhD15, Shari Kaiser, PhD16
AFFILIATIONS
1 Bristol Myers Squibb, Princeton, New Jersey, United States
2 Bristol Myers Squibb, Princeton, New Jersey, United States
3 Bristol Myers Squibb, Princeton, New Jersey, United States
4 Bristol Myers Squibb, Princeton, New Jersey, United States
5 Greco-Hainsworth Center for Research at Tennessee Oncology, Chattanooga, Tennessee, United States
6 University of Alabama at Birmingham, Birmingham, Alabama, United States
7 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
8 Hematologic Malignancies and Cellular Therapy Program, Simmons Comprehensive Cancer Center, UT Southwestern Medical Center, Dallas, Texas, United States
9 City of Hope Comprehensive Cancer Center, Duarte, California, United States
10 University of Maryland Marlene and Greenebaum Comprehensive Cancer Center, Baltimore, Maryland, United States
11 Colorado Blood Cancer Institute, Sarah Cannon Cancer Network, Denver, Colorado, United States
12 Icahn School of Medicine at Mount Sinai, New York, New York, United States
13 University of California, San Francisco, California, United States
14 Bristol Myers Squibb, Princeton, New Jersey, United States
15 Bristol Myers Squibb, Princeton, New Jersey, United States
16 Bristol Myers Squibb, Princeton, New Jersey, United States
ABSTRACT
Introduction: GPRC5D-targeted immunotherapies have demonstrated robust antimyeloma activity in relapsed/refractory multiple myeloma (RRMM), supporting interest in their use earlier in the treatment paradigm and following other therapies, such as BCMA-directed therapies. In a first-in-human (FIH) phase 1 trial, arlocabtagene autoleucel (arlo-cel), a GPRC5D-targeted CAR T cell therapy, demonstrated deep and durable activity in patients (pts) with RRMM across early- (1-3 prior lines of therapy [pLOT]) and late-lines (≥3 pLOT), regardless of prior BCMA exposure (Bal et al., Haematologica. 2026; Bal et al., Blood. 2026). Emerging data suggest antigen loss may be a cause of resistance to immunotherapies. In a multi-institutional cohort of 21 pts with MM who relapsed after GPRC5D T-cell engager (TCE) therapy, GPRC5D mutations, primarily single-nucleotide variants (SNVs) and indels, were reported in 68.4% of pts (Lee et al., Nat Med. 2026); data in relapse settings following GPRC5D-targeted CAR T cell therapy are limited (Mailankody et al., 2022 NEJM; Ma et al., 2025 Blood). Here we report an exploratory analysis from the largest phase 1 trial (NCT04674813) characterizing baseline GPRC5D expression by pLOT and BCMA exposure. Associations with clinical outcomes and post-progression alterations in the GPRC5D genetic locus were evaluated.
Methods: Pretreatment bone marrow aspirates (BMAs) were centrally analyzed by flow cytometry to quantify cell surface GPRC5D positivity and density on plasma cells (PCs; CD138+CD38+CD45dim). CD138+ PCs from BMA samples, with matched peripheral blood mononuclear cell (PBMC) controls, underwent DNA isolation. Whole-genome sequencing (WGS) was performed on DNA from PCs and PBMCs using 2×150-bp paired-end sequencing, generating 960 million (M) and 320M reads, respectively. GPRC5D locus integrity was assessed using WGS of tumor and matched-normal samples to identify somatic SNVs, indels, and copy-number (CN) alterations using Sentieon TNhaplotyper2/TNfilter, Manta, and Battenberg. Variants were annotated with SnpEff/SnpSift and public databases (dbSNP, COSMIC, ClinVar, dbNSFP, and gnomAD).
Results: Among 88 pts with pretreatment protein data, no differences in GPRC5D protein levels were observed by prior BCMA status or pLOT subgroup (1-3 vs >3). Pretreatment GPRC5D protein expression was not associated with best overall response. Of 50 pts with pretreatment samples that met WGS quality criteria, none had pretreatment deleterious mutations (SNV/indels) at the GPRC5D locus and 15 (4 in 1-3 pLOT; 11 in ≥3 pLOT) had a single CN loss spanning the GPRC5D gene; the remainder either had a copy-neutral loss of heterozygosity or retained 2 copies. Eight samples had clonal CN variation (CNV) (2 in 1-3 pLOT; 6 in ≥3 pLOT) and 7 had subclonal CNV (2 in 1-3 pLOT; 5 in ≥3 pLOT).
Among 65 patients with progressive disease (PD), 6 PD samples met WGS quality criteria of which 5 had a pretreatment sample from the same patient. From pretreatment to PD, 3 of the 5 samples retained a clonal hemizygous loss overlapping the GPRC5D gene, and 2 had both GPRC5D alleles intact. No new CNV losses were observed between pretreatment and PD. The PD sample without a corresponding pretreatment sample had a hemizygous GPRC5D loss overlapping GPRC5D.
Two of the 3 pts with pretreatment hemizygous GPRC5D deletion acquired deleterious GPRC5D mutations in the remaining allele at PD, resulting in biallelic GPRC5D alterations; both were in pts from the ≥3 pLOT cohort and samples were collected 9-10 months post-infusion. Investigations are underway to evaluate whether these mutations drive structural alterations of the GPRC5D protein or impact surface expression.
Conclusions: Baseline GPRC5D protein was expressed similarly regardless of BCMA exposure and pLOT, supporting broad target availability in RRMM, and deep and durable responses were observed across the range of GPRC5D expression. Given the limited PD WGS sample size, these findings are hypothesis generating. Among the 6 pts with evaluable PD samples, 2 pts acquired deleterious mutations post-infusion in the remaining GPRC5D allele at PD, that resulted in biallelic GPRC5D alterations; this was numerically lower than reported in pts progressing after GPRC5D-directed TCE therapy.
DISCLOSURES
Safiyyah Ziyad, PhD: BMS - Current equity holder in publicly-traded company; BMS - Current Employment