Multiomics Used To Stratify Multiple Myeloma Patients Into Affinity Groups
By Deborah Borfitz
January 6, 2022 | All new multiple myeloma patients being treated at the Tisch Cancer Institute at Mount Sinai are now being assigned to one of 12 subgroups based on shared genetic and molecular features of their disease. The multiple myeloma patient similarity network (MM-PSN) is one of the latest computational modeling tools being used here to translate clinical data into better treatment decision-making in the clinic, and the first to apply multiomics to the stratification of patients with the blood cancer, according to Alessandro Lagana, PhD, assistant professor of oncological sciences.
The basic premise is not unlike most social media networks, where users are partitioned into communities with others sharing the same interests, he says. Multiple myeloma, like the broader digital outlets Facebook and Twitter, is also highly heterogenous.
Only with MM-PSN, patients serve as the nodes and the connection drivers are in their DNA and RNA profiles, Lagana explains. This expands well beyond gene expression—the standard basis of comparison over the past 15 years—to also include data on fusion genes, somatic single-nucleotide variations, copy number alterations, and translocations.
The patient similarity network was built from data generated by whole-exome sequencing, whole-genome sequencing, and RNA sequencing of tumor samples from 655 newly diagnosed myeloma patients. As recently reported in Science Advances (DOI: 10.1126/sciadv.abg9551), it was used to identify specific genes and genetic alterations responsible for never-before-defined subtypes of the blood cancer as well as potential targeted treatments.
Lagana was previously involved in the analysis of gene expression, copy number variation, and clinical data for a multiple myeloma network model where the nodes were genes. The similarity network fusion method instead combines multiomics data as a way of better describing the disease.
In the latest study, MM-PSN succeeded in finding communities of patients who were similar to one another, rediscovering some of what was already known, says Lagana. Past studies had established that roughly half of multiple myeloma patients have trisomies (three copies of a chromosome instead of two), for instance, and about the same proportion have translocations (rearrangements in the genome) creating “super-oncogenes” driven by overexpression of immunoglobulin.
One of the most important clinical contributions of the new study is the discovery of two novel disease subgroups among patients with hyperdiploidy (more than two complete sets of chromosomes), he notes, which statistically corresponds to better prognosis. But among the subset of hyperdiploidy patients who also have amplification of the long arm (q) of chromosome 1, their prognosis is “significantly” worse.
1q has been recognized for many years as a bad prognostic factor not only in multiple myeloma but other cancers as well as myelodysplastic syndromes, Lagana says. But the underlying mechanisms have never been fully understood.
Clinicians anywhere can take full advantage of MM-PSN—the multiomics classifier is freely available—by doing DNA and RNA profiling on their patients and plugging in the data to learn in which of the three main groups, and their subgroups, they belong.
The primary groups are patients with hyperdiploidy and the translocation of chromosomes 8 and 14 on the MYC oncogene; translocations of chromosomes 4 and 14 dysregulating genes FGFR3 and MMSET or translocation of chromosomes 14 and 16 on the MAF gene; translocation of chromosomes 11 and 14 on the CCND1 gene. Most patients cluster into one the three groups, Lagana says.
Within the 12 subgroups, patients are further sorted by the pattern of these alterations, and many of these features are easily obtainable using typical cytogenetic approaches (e.g., aspiration and biopsy of bone marrow) since the alterations “colorizing” each subgroup—based on significant up-regulation or down-regulation of the pathway—are so well defined in MM-PSN.
Spectral clustering revealed novel insights into the co-occurrence of primary translocation events and secondary adverse lesions such as gain of 1q and whole-arm deletions of chromosomes 16q and 17p (the short arm), which harbors the tumor suppressor gene TP53.
Overall, about 6% of multiple myeloma cases will ultimately be in the subgroup involving overexpression of MMSET and amplification of chromosome 1 and, consequently, given the worst prognosis, Lagana says. The handful of outliers, patients who are hard to classify but MM-PSN nevertheless assigns to one of the 12 subgroups, are the subject of ongoing study.
As part of the published study, the Mount Sinai team used multiomics drug repurposing to identify candidate therapeutic options and potential immuno-oncology targets in the different MM-PSN subgroups. The analysis was based on published clinical trials identifying markers with good response to drugs that are significantly present in certain patients.
It is an admittedly “optimistic” treatment approach, Lagana says, but typical of precision medicine where a drug used for one type of cancer may prove useful for another based on shared genomic alterations. Researchers at Mount Sinai now plan to formally investigate some of these leads with a formal clinical trial, but some of the identified associations can be applied in clinical practice immediately, he adds.
For example, the widely prescribed BCL2 inhibitor venetoclax might logically be given to multiple myeloma patients with translocations on the CCND1 gene who are relapsing—and most especially those in the subgroup with amplification of chromosome 11 and deletion of chromosome 13 who appear to be at higher sensitivity to the chemotherapy drug. The discovery was the result of computational analysis of cell lines which, like patents in the study, were classified by MM-PSN, says Lagana.
1q is, once again, a bad player among patients in the primary CCND1 group. “According to all the cell line models, [patients with both these features] are not sensitive to venetoclax.”
Genomic analyses directly translate into clinical practices at Mount Sinai, says Lagana, citing ongoing collaboration with his medical colleagues and study co-authors Samir Parekh, M.D., and Sundar Jagannath, M.D. over the past few years.
Mount Sinai researchers plan to launch their clinical trial in 2022 and it will be informed by the genomic-guided approach of MM-PSN, says Lagana. It is a follow-up of sorts to a small pilot study published a few years ago in the JCO Precision Oncology (DOI: 10.1200/po.18.00019) where they showed how a comprehensive sequencing approach can identify viable drug options for patients with relapsed or refractory multiple myeloma by targeting specific genomic alterations.
The new study will likely focus on patients with an actionable alteration as well as how to overcome potential resistance by assigning specific drug combinations, he says.
Meanwhile, work on the MM-PSN model will continue, Lagana says, noting his interest in understanding why patients are so different genomically and molecularly, and what it means in terms of disease pathogenesis and progression as well as resistance to drugs and relapse and what to do next. There may be specific vulnerabilities that can be exploited to improve treatment of multiple myeloma—for example, immunotherapies like CAR T-cell that train the body’s immune system to recognize tumors and kill them.
The transcript expression of up- and down-regulated genes in the different subgroups may be an important indication of potential actionable targets, he says. Particularly relevant are the gene-encoding proteins that are most enriched on the surface of multiple myeloma cells.
Moving forward, the number of subgroups in the MM-PSN classification system is likely to grow once the research team factors in genomic and transcriptomic data from the bone marrow biopsies of patients who relapsed from treatment or have refractory disease. What differences will be seen relative to treatment-naïve patients is currently an open question.
But it is known that targeted anticancer therapies almost uniformly select for drug-resistant subclones within these heterogenous tumors, eventually and often rapidly making patients resistant to treatment. “Multiple myeloma patients many times have three, four, five, even almost 20 lines of therapy [over the course of their disease],” Lagana says.
That makes their genome look a lot different than it did at the point of initial diagnosis. The stage of their disease also advanced with time, further changing the clonal landscape.