NEW YORK – A study presented at the American Society of Hematology's annual meeting on Sunday introduced a promising gene expression signature for identifying refractory multiple myeloma patients who are more likely to respond to selinexor (Karyopharm Therapeutics' Xpovio).
"We've identified the first robustly validated gene signature of response to selinexor in multiple myeloma," Paula Restrepo, a bioinformatician at Mount Sinai's Icahn School of Medicine, said in a presentation at the meeting describing her group's efforts to discover and validate genes differentially expressed in responders and non-responders to the drug. While researchers from Mount Sinai led the study, experts from selinexor-maker Karyopharm and Travera, a company with a ex vivo drug testing platform that researchers are using to further validate the three-gene expression signature, were also involved.
The study presented at the meeting showed that "upregulation of three genes — WNT10A, DUSP1, and ETV7 — is associated with both depth and duration of response," Restrepo said. The group is now conducting additional research to better understand why these genes are associated with therapeutic response.
Selinexor, an XPO1 inhibitor, causes cancer cells to die by blocking their ability to export tumor suppressor proteins and oncogenic mRNA and trapping them in the nucleus. Selinexor in combination with bortezomib (Takeda's Velcade) and dexamethasone is US Food and Drug Administration-approved as a second-line treatment for refractory multiple myeloma and as a fifth-line treatment for refractory multiple myeloma patients in combination with dexamethasone.
The agency last year approved the selinexor-bortezomib-dexamethasone second-line regimen based on the BOSTON trial, which included relapsed or refractory multiple myeloma patients who had received at least one and at most three prior therapies. Patients randomized to selinexor-bortezomib-dexamethasone had median progression-free survival of 13.9 months versus 9.5 months for those on bortezomib-dexamethasone.
"However, not all patients respond to selinexor-based therapy," Restrepo said at the meeting. "And many go on to experience drug-related adverse events." The FDA-approved label for selinexor warns patients about the risk of thrombocytopenia, neutropenia, gastrointestinal toxicity, hyponatremia, serious infection, neurological toxicity, embryo-fetal toxicity, and cataracts.
Although selinexor's mechanism of action is well described in the literature, there are currently no biomarkers that allow oncologists to identify which patients are likely to respond to the drug and which aren't. "There's a really great unmet need to identify biomarkers of response that can better guide the use of selinexor for the treatment of multiple myeloma," Restrepo said.
Her group analyzed RNA-seq data from CD138-positive cells from around 100 patients in the BOSTON trial that led to the drug's approval to discover differentially expressed genes associated with selinexor response. They then validated their findings in samples from 64 triple-class multiple myeloma patients who received selinexor and dexamethasone in the STORM trial and from another real-world cohort of 34 heavily pretreated, refractory patients who received selinexor outside of a clinical trial at Mount Sinai.
Comparing RNA-seq data from 53 patients in the selinexor arm of the BOSTON study, Restrepo's group identified differentially expressed genes and pathways that tracked with short and long progression-free survival, as well as response according to criteria from the International Myeloma Working Group, using 13 different cutoffs. The 47 patients who received bortezomib-dexamethasone in the comparator arm of the BOSTON study served as a negative control for the analysis.
Researchers selected the best-performing signature based on statistical analysis, which was an upregulation signature comprising three genes: WNT10A, DUSP1, and ETV7. The signature was associated with a progression-free survival of 120 days or more among selinexor-treated patients. The same signature could not similarly distinguish responders and responders in the comparator arm of the BOSTON study, which provided confidence that the signature "was specific to selinexor response," Restrepo said, and "not just a general prognostic marker associated with a less aggressive myeloma."
Her group then validated the performance of the signature in the STORM trial and real-world Mount Sinai cohorts. The gene signature's ability to separate responders from non-responders was "less dramatic" in the STORM cohort, Restrepo said, since these patients had much more advanced disease than those in the BOSTON trial. "Nevertheless, there's a statistically significant difference and the signature validates," she said, adding that the signature also validated in the Mount Sinai cohort, which involved heavily pretreated refractory multiple myeloma patients and likely better represents real-world use of the drug.
"We looked at these three genes individually by themselves," Restrepo added. "They are not predictive of selinexor response."
Researchers further confirmed that this three-gene expression signature was indeed associated with selinexor response and not merely prognostic, using 767 samples in the MMRF-COMMPASS study, in which multiple myeloma patients received various standard-of-care therapies other than selinexor.
Restrepo's group also considered the association of the three-gene expression signature with best overall response according to the IMWG's criteria. In the selinexor arm of the BOSTON trial, in the STORM trial, and in the Mount Sinai real-world data, "patients who had favorable responses [to selinexor] had higher expression of our three-gene expression signature," Restrepo said. "Overall, these data are showing that the signature not only tracks with the duration of response but also depth."
Given the performance of the three-gene expression signature in selinexor-treated multiple myeloma patients, researchers were curious to see if it would work in other tumor types treated with the drug. They conducted RNA-seq analysis on samples from 57 recurrent glioblastoma patients treated with selinexor. "Remarkably, we found those who had higher expression of the signature had significantly improved progression-free survival," Restrepo said. "We also found that patients who achieved a clinical benefit of a partial response or better had much higher expression of the signature."
Researchers also wanted to understand why the differential expression of these genes are predicting selinexor response. Looking at protein-protein interaction networks, they found that DUSP1 is involved in MAPK signaling, WNT10A is involved in WNT signaling, and ETV7 is involved in oncogenic transcriptional regulation. When the three genes were knocked down in multiple myeloma cell lines treated with selinexor in functional studies, Restrepo's group found the cells became resistant to treatment.
Recognizing that the three identified genes are not very well characterized in multiple myeloma or as part of selinexor's mechanism of action, "we're currently performing more validation and overexpression experiments in resistant cell lines to create a framework for mechanistic studies that can better interrogate why these three genes are associated with response," Restrepo said.
According to Samir Parekh, senior investigator of this analysis and director of translational myeloma research at Mount Sinai's Tisch Cancer Institute, his team also hopes to test out the performance of the signature prospectively in patients using "a qPCR-type or NanoString-type assay." At Mount Sinai, researchers are also starting to use Travera's ex vivo drug testing platform to see if the response in multiple myeloma cells track with the expression of the genes.
"Ultimately, a combination of platforms" to clinically validate this signature "would be even better than [using] an expression platform alone," Parekh said.