Myeloma Molecular Profiling Reveals Unique Copy Number and Expression Subtypes

Sagar Lonial, MD, FACP

Findings from a molecular analysis of the baseline cohort of the Relating Clinical Outcomes in Multiple Myeloma to Personal Assessment of Genetic Profile (CoMMpass) study (NCT01454297) revealed target genes for recurrent loss-of-function (LOF) and gain-of-function (GOF) events and identified unique copy number and expression subtypes of high-risk patients with newly diagnosed multiple myeloma.

Study results were published in Nature Genetics. Investigators developed an integrated model to comprehensively identify LOF and GOF events and address the limitations of analyzing singular data types. The model combined measurements from whole-exome sequencing (WES), whole-genome sequencing (WGS), and RNA sequencing (RNA-seq) and assigned a functional state to each gene.

At diagnosis, among patients who had undergone all 3 sequencing assays (n = 592), a complete LOF event was identified in at least 5 patients across 70 genes, the most common being TRAF3 (10.1%), DIS3 (6.9%), TENT5C/FAM46C (5.1%), CTLD (4.7%), TP53 (4.1%), MAX (3.5%), WWOX (3.2%), RB1 (3.2%), HUWE1 (2.7%), PVT1 (2.5%), MAGEC1 (2.0%), and CDC42BPB (2.0%).

Questions remain regarding the target gene(s) of chromosome 13 (chr13) loss, according to the authors. WGS detected a 13q14 deletion (del(13q14)) in 52.0% of patients, whereas LOF analysis showed that 26.5% of patients had complete LOF in at least 1 gene on chr13. Complete LOF in DIS3 and RB1 were observed in 6.9% and 3.2% of patients, respectively, and additional genes were independently knocked out.

Two continuous gene regions with complete LOF were seen. The first was composed of PSPC1 (1.5%), MPHOSPH8 (1.4%), ZMYM5 (1.4%), and ZMYM2 (1.0%). The second included TGDS (1.9%) and GPR180 (0.8%). The minimal regions of deletion and LOF frequency indicate that the targets in these regions are PSPC1 and TGDS, respectively.

Other complete LOF events targeted LATS2 (1.4%), BRCA2 (1.2%), MYCBP2 (1.0%), PARP4 (1.0%), TPP2 (1%), ARHGEF7 (0.8%), TSC22D1 (0.8%), and CDK8 (0.8%). These findings indicate that monosomy 13 is associated with several independent gene inactivation events.

GOF events were observed in 92% of patients at diagnosis across 27 genes, the most common being KRAS (23.6%), NRAS (21.6%)WHSC1 (10.3%), BRAF (7.1%), FGFR3 (4.9%), HIST1H1E (3.2%) and EGR1 (2.5%).

“Despite substantial efforts to understand the molecular basis of the disease, predicting patient outcomes and identifying high-risk patients remains a challenge,” study author Sagar Lonial, MD, FACP, and coauthors, wrote. “Although previous genomic studies were instrumental in deconvoluting the genetic heterogeneity of myeloma, they are mostly limited by small cohort sizes, the number and types of assays performed, a lack of longitudinal sampling, clinical follow-up and biased inclusion of heavily pretreated patients, limiting our comprehensive understanding of the disease.”

Lonial serves as chief medical officer of the Winship Cancer Institute of Emory University. He is also the Anne and Bernard Gray Family Chair in Cancer, as well as a professor and chair in the Department of Hematology and Medical Oncology at the Emory University School of Medicine in Atlanta, Georgia.

An Overview of CoMMpass: Design and Demographics

This prospective, longitudinal, observational study included 1143 treatment-naive patients who were diagnosed with multiple myeloma from sites across the United States (US), Canada, Spain, and Italy between 2011 and 2016. Tumor samples were collected at diagnosis and each progression event and were characterized using WGS, WES, and RNA-seq. Clinical data were collected every 3 months throughout the 8-year observational period.

The demographics and clinical parameters of the baseline cohort were distributed as expected, the authors noted. The median age of patients at diagnosis was 63 years (range, 27-93), and most patients were male (60.4%). In total, 35.1%, 35.1%, and 27.2% of patients had international staging system (ISS) I (ISSI), ISSII, and ISSIII disease, respectively. Patients were predominantly from the US, and 80.6%, 17.5%, and 1.9% of patients were Caucasian, Black, and Asian, respectively.

Regarding WGS-defined cytogenetic phenotype, 57.2%, 42.8%, 24.3%, 35.2%, 52.0%, and 12.5% of patients had disease with hyperdiploid (HRD), non-HRD (NHRD), del(1p22), gain(1q21), del(13q14), and del(17p13), respectively. Common target genes from the 3 immunoglobulin loci that were involved in translocations included CCND1 (20.0%), CCND2 (1.2%), CCND3 (1.8%), MAF (4.0%), MAFA (0.7%), MAFB (1.3%), MYC (14.3%), and WHSC1 (12.8%). Eighty-three percent of these events involved the IgH locus; 11.7% and 5.3% of these events involved the λ and κ loci, respectively.

Survival Outcomes According to Copy Number Subtype

Regardless of treatment, the median time to second-line therapy across the entire cohort was 38.1 months, and the median overall survival (OS) was 103.6 months. The median OS among patients with ISSIII disease was 53.9 months; the median OS for patients with ISSI and ISSII disease could not be confidently predicted. The authors emphasized that patients with 1 or more high-risk cytogenetic features had worse OS outcomes, even with uniform use of novel agents.

Unsupervised consensus clustering was performed in copy number data from 871 patients to identify potential underlying myeloma phenotypes beyond HRD and NHRD subtypes. Eight copy number subtypes, including 5 HRD and 3 NHRD subtypes, were identified across consistent findings from 3 trials. The HRD groups included HRD classic subtype (defined by gains of classic HRD chromosomes); HRD, ++15 subtype (chr15 tetrasomy); HRD, diploid 7 subtype (chr7 absence); HRD, diploid 3,7 subtype (chr3 and chr7 absence); and HRD, +1q, diploid 11, –13 subtype (lack of chr11 trisomy, gain of chr1q, loss of chr13). NHRD groups included diploid subtype, which mostly lacked copy number events and was highly associated with translocations targeting a D-type cyclin (71.3%); the –13 subtype, which contained a subpopulation of patients with chr14 loss; and the +1q subtype, –13, which had 1q gains.

No difference in outcomes was observed between the HRD and NHRD patients. However, patients in either subtype with both chr13 loss and 1q gain had inferior OS outcomes compared with patients in other copy number subtypes (HR, 1.732; 95% CI, 1.354-2.215; P < .001). The difference in OS outcomes between NHRD patients in the +1q, –13 subtype and the –13 subtype was 35 months, indicating that 1q gain is a predictor of poor outcome. However, a Cox proportional hazard model did not identify gain(1q21) as an independent feature.

Breaking Down Myeloma Gene Expression Subtypes

Investigators performed consensus clustering on RNA-seq results from 714 baseline samples to identify myeloma gene expression subtypes. A total of four subtypes were identified: MMSET expressing (MS), which included patients with t(4;14); MAF family transcription factor expressing (MAF), which included those with t(14;16); cyclin D expressing group 1 (CD1), which included those with t(11;14); and proliferation (PR), which included those with a high proliferation index.

Of the total study population, 10.6% of patients had the MS subtype, with a t(4;14)-WHSC1 detected in 92.5% (n = 62/67) of samples by WGS. The MAF subtype included 6.4% of patients in the total population, 92.7% (n = 38/41) of whom had translocations. The 3 remaining patients in this subtype with indetectable immunoglobulin translocations had high expression of an MAF family gene.

The CD1 subtype included 4.3% of patients in the total population, 96% (n = 24/25) of whom had a D-type cyclin targeting translocation. One patient had a t(9;14) that resulted in PAX5 overexpression. Investigators also identified 2 related subtypes: CD2a and CD2b. The CD2a subtype consisted of 7.8% of patients in the total population and the CD2b subtype included 8.0% of patients in the total population. A total of 85.1% (n = 40/47) and 91.1% (n = 51/56) of patients with these respective subtypes had a detectable D-type cyclin IgH translocation, and both subtypes were associated with CD20 expression on the cell surface. Patients in the CD2b subtype had higher proliferative index scores than those in the CD2a subtype (P < .005), but no significant difference between the subtypes was observed regarding OS or time to second-line therapy. Patients in the CD2a subtype initiated second-line therapy 8.4 months earlier than those in the CD2b subtype.

The PR subtype included 7.1% of patients in the total population and had poor clinical outcomes, with a median OS of 21 months. In a comparison of current checkpoint and immunotherapy targets in non-PR vs PR patients, all 5 tested checkpoint targets (CD86, CD200, CD274/PDL1, LGALS9/GAL9, and TNFRSF14/HVEM) were decreased in the PR group, and TNFRSF17/BCMA was increased. No difference in the expression of the 4 tested immunotherapy targets (SLAMF7/CS1, CD38, FCRL5/FCRH5, and GPRC5D) was observed between the 2 groups.

A subtype consisting of 11.1% of patients most closely resembled the low bone subtype; however, a decrease in bone lesions was not observed among patients. This subtype included 59.2% of HRD patients and 40.8% of NHRD patients, among whom 74.0% had chr1q gain and 26.0% had at least 4 copies and were included in the 1q gain subtype.

Four RNA subtypes were associated with HRD, but were not uniquely associated with a previously defined subtype. Two HRD subtypes were closely associated with the HY subtype but differed due to tetrasomy 15 enrichment. In total, 75.5% (n = 37/49) of patients in the first subtype (designated HRD, ++15, MYC) and 30.3% (n = 23/76) of those in the second subtype (designated HRD, ++15) had MYC rearrangements.

A third HRD subtype, which included 8.3% of patients in the total population, was most closely associated with the PRL3 subtype. Investigators saw a MYC structural event in 71.4% (n = 35/49) of these patients. This subtype was named HRD MYC, low NF-kB.

The fourth HRD group consisted of 4.6% of patients in the total population and was associated with the NF-kB subtype, although this association could not be clearly defined. Overexpression of the TP53 translation inhibitor NINJ1 was deemed a predictor of this RNA subtype. The lowest median expression of TP53 was observed in this subtype vs all other RNA subtypes; accordingly, this subtype was named HRD, low TP53.

The final RNA subtype included 12.2% of patients in the total population. Disease features in this subtype correlated with the previously defined myeloid group, but it was dominated by lower-purity samples and was thus termed low purity.

Associations Between LOF and GOF Events and RNA Subtypes

Twenty-one genes with complete LOF or GOF were significantly associated with at least 1 RNA subtype. GOF was identified in translocation target genes associated with the MS, MAF, and CD subtypes. Complete WWOX loss was frequently detected (P < .001), indicating the potential role of WWOX in myeloma. NRAS GOF had diminished MS and 1q gain subtypes, and the 1q gain subtype was enriched for TRAF3 LOF. The CD2a subtype was enriched for GOF events in IRF4 (P < .005) and NRAS (P < .005), and the CD2b subtype was enriched for GOF events in EFTUD2 (P < .01) and IRF4 (P < .005), which may be subtype-specific therapeutic targets. The HRD RNA subtypes were generally not enriched for GOF or LOF events, except for the HRD, ++15, MYC subtype, which was enriched for FAM46D LOF events (P < .001).

The PR subtype was enriched for LOF in MAX (P < 0.01) and RB1 (P < 0.001) in del(13q14) (P < 0.001), gain(1q21) (P < 0.001), and ISSIII (P < .001) patients. In total, 50%, 22%, and 28% of PR patients were ISSIII, ISSI, and ISSII, respectively, which indicates that ISS underestimates disease severity in several high-risk patients. The LOF of RB1 and MAX are defining genetic features of the high-risk PR phenotype, the authors noted.

Investigators developed a predictive model to assign serial samples to the subtypes with the highest-class probability. Seventy-one patients were assigned to a subtype at 2 or more time points, and 55 patients were assigned to a subtype other than low purity at 2 or more time points. At diagnosis, 5 serial patients were assigned to the low purity subtype. However, at progression, all 5 patients were assigned to a subtype other than low purity, suggesting that this phenotype is characterized by sample purity instead of distinct disease biology. Most patients remained in the same subtype over their disease course, but 26.5% (n = 13/49) of patients who were not assigned to the low purity or PR subtypes at baseline transitioned into the PR subtype at progression. These patients had inferior outcomes compared with patients who also progressed, including a median OS of 88 days after detected progression.

Investigators compared gene functional status at the PR and prior non-PR time points to identify potential drivers of transition to the PR subtype among patients who transitioned to PR who had available molecular data from both time points (n = 9/13). At the time of progression, no patients acquired complete RB1 LOF; 3 patients had complete LOF of a CDK inhibitor; 2 patients had complete CDKN2C LOF due to homozygous deletions, 1 of whom acquired 2 independent deletions at progression, and 1 of whom had clonal deletion at diagnosis; and 1 patient acquired complete CDKN1B loss from a pre-existing deletion and had a clonal frameshift mutation that was detected only at progression. Overall, the authors noted that multiple genetic defects in G1/S checkpoint genes can result in the PR phenotype.

“The comprehensive nature of this dataset and our integrated analysis framework define both the overall frequency of gene alterations in myeloma and the genetic basis of a high-risk patient population that does not benefit from current therapies,” the authors concluded. “Given that patients frequently transition to the high-risk PR subtype at progression, it will be important to know the percentage of PR patients in clinical trial populations, particularly in the relapse/refractory setting, to understand if the arms are balanced and if there is a difference in response between these groups.”

Reference

Skerget S, Penaherrera D, Chari A, et al. Comprehensive molecular profiling of multiple myeloma identifies refined copy number and expression subtypes. Nat Genet. 2024;56(9):1878-1889. doi:10.1038/s41588-024-01853-0