NEW YORK – California-based biotech company Notable Labs and researchers from Stanford University showed in a 21-patient feasibility study that Notable's drug sensitivity assay was able to identify treatment options for blood cancer patients who would not have gotten a precision drug based on next-generation sequencing results.
The study, published last month in Blood Advances, showed that Notable's ex vivo drug sensitivity platform was able to correctly predict response or nonresponse in patients with "an overall accuracy of 85 percent." Although these patients were sequenced, and several patients had mutations that characterized their cancers, these mutations could not be targeted by available drugs. As such, Notable believes the data from this study shows that its drug sensitivity assay can help identify treatment options that blood cancer patients may benefit from even when NGS can't.
"This is something that can be useful to physicians in the future to help inform treatment decisions. … In those instances where there aren't mutations targetable by current medicines, we might still be able to find viable treatment options for those patients," said Hiroomi Tada, chief medical officer at Notable.
Even when genomic testing identifies tumor mutations that may be targeted by marketed drugs, patients don't always respond as predicted. Indeed, patients have varying responses to targeted agents. For example, larotrectinib (Bayer's Vitrakvi) for refractory solid tumors with NTRK fusions, had a 75 percent response rate in studies that led to its approval by the US Food and Drug Administration.
On the other hand, "there are many drugs [with response rates] in the 40 percent range," Tada said, adding that functional testing with Notable's platform has the potential to fill that gap.
In blood cancers such as myelodysplastic syndrome, Notable's assay uses flow cytometry to look at multiple cell surface markers that identify different cell populations within a sample. These cells are then incubated with various single or combination treatments, and evaluated for indications of response, by tracking cell viability, cell cycle arrest, apoptosis, and differentiation.
For the prospective feasibility study, researchers led by Peter Greenberg from the Stanford University Cancer Center enrolled 21 patients with myeloid neoplasms, including those with myelodysplastic syndrome and acute myeloid leukemia, who were refractory to hypomethylating agents. Researchers collected blood and bone marrow samples from patients and sent a portion of the samples to Notable for ex vivo drug sensitivity testing and a portion for next-generation sequencing with Tempus' 596-gene panel.
Stanford researchers chose the panel of 74 drugs and 36 drug combinations they wanted Notable to test on patients' samples, including targeted therapies, cytotoxic chemotherapy, differentiative agents, hypomethylating agents, and androgen therapies.
"They wanted to be sure that these were drugs patients could potentially get access to," Tada explained. Many of the selected cancer drugs and combinations have been approved for other cancer indications. But researchers also included therapies such as calcitriol or dexamethasone that are not characterized as cancer drugs but are known to cause changes in cancer cells.
Stanford also ranked the order in which they wanted the drugs and drug combinations tested, so that the ones that were of highest interest were generally tested first, according to Tada.
The turnaround times for the NGS and drug sensitivity assay were similar. On average, drug sensitivity results were available after 15 days following sample collection, and NGS results were available after 14.5 days. Tumor board review was conducted within two days after receiving the results.
Once the genomic test results on patients' mutation status and the drug sensitivity assay results were available, four of the study authors met as a tumor board to review the results and recommended treatments for the patient.
In total, eight patients received a therapy recommended by the tumor board. The 13 patients who did not receive a tumor board-recommended therapy had either chosen best supportive care, hospice, other approved therapies, a clinical trial, or allogeneic hematopoietic cell transplantation. Of the eight patients who received tumor board recommended therapies, six patients responded to treatment and two had stable disease. The six responding patients included three patients who had a complete response, two who had a partial response, and one patient who had hematologic improvement.
The most commonly recommended treatment was venetoclax (AbbVie/Genentech's Venclexta) with a hypomethylating chemotherapy, such as azacitadine or decitabine, but this is not given to patients with any particular gene mutations, Tada noted. Another recommended chemotherapy some patients received was bortezomib, which also does not have a molecularly targeted indication. Tada said that in MDS, the majority of currently available therapies do not target specific gene mutations.
Stanford investigators were particularly interested in patients with mutations in high-risk genes, including TP53, RUNX1, ASXL1, ETV6, EZH2, and RAS. However, even the mutations with the highest prevalence in the study (occurring in TET2, ASXL1, and STAG2) appeared in only 24 percent of patients. For the majority of mutations detected by NGS, there was no targeted drug while the study was being conducted. More recently, the FDA approved tazemetostat (Epizyme's Tazverik) for epithelioid sarcoma, follicular lymphoma with no alternative treatment options, and EZH2-mutated follicular lymphoma, but not for MDS or AML.
Two patients who had a complete response both received venetoclax and azacitidine and the third patient with a complete response received venetoclax and the cytotoxic drug cytarabine. One of these patients had mutations in ASXL1, STAG2, SF3B1, KRAS, NRAS, CEBPA, the second had a mutation in PBRM1, and the third had a TET2 mutation.
According to Tada, these data suggest that the value of Notable's ex vivo platform is specifically for those patients, such as the majority of MDS patients, who may have mutations characterizing their cancer, but not any treatments that target those mutations.
In oncology more broadly, despite the high hopes that NGS would be able to get more cancer patients to precision drugs, expectations have been tempered by the results from studies, such as NCI-MATCH, Tada noted. That study, which is testing the efficacy of genomically informed treatment strategies, has garnered great interest and investigators have screened thousands of patients, but relatively few patients have matched to a treatment arm (see here and here).
Notable is hoping that its platform will be able to predict beneficial treatment options for cancer patients who come up empty after NGS, but the data on whether its predictions actually lead to improved outcomes is still early and evolving.
Out of the 21 patients in the feasibility study, there were 10 patients who did not respond to their given treatment, and Notable correctly predicted this in all patients. In these patients, physicians had ultimately prescribed a non-tumor board-recommended therapy either because it was standard of care, or because they felt the recommended drug was inaccessible or would be unmanageable due to the fitness of the patient.
The platform predicted correctly that seven out of 11 patients would respond to their given treatment, and incorrectly predicted that four patients would not respond.
Greenberg and colleagues estimated that the one-year progression-free survival rate for the eight patients who received tumor board recommended therapies is 73 percent with an estimated one-year overall survival rate of 88 percent. However, they cautioned that the dataset was too small for reliable conclusions around clinical benefit.
To further evaluate the clinical utility of this approach, the investigators are planning a multicenter randomized trial comparing tumor board-guided therapy using Notable’s ex vivo drug sensitivity screening platform to a standard-of-care arm, in which patients will receive the clinician's choice of therapy for patients with HMA-refractory myeloid neoplasms.
Tada added that Notable has preliminary, unpublished data indicating that patients who received a therapy that Notable's assay predicted they would respond to tended to survive longer than patients who received a therapy that the assay predicted nonresponse for. Now, Notable is enrolling 25 to 50 patients in a follow-up validation cohort to solidify the assay's current cut-off for response and non-response, which correlates to the amount of blasts killed by the drug.
While Notable is working to demonstrate the value of its platform, its drug sensitivity screening approach must exist alongside NGS. Toward this end, the study authors recognized that combining drug sensitivity screening and genomic analysis can also benefit patients. They pointed out in the paper that drug sensitivity screening was most useful in identifying beneficial therapies for higher-risk MDS patients with excess blasts and less clinically valuable for those with low blast counts.
For the latter patient population, the investigators' primary therapeutic goal was to increase the number of mature blood cells rather than reduce the blast count. As such, the tumor board tended to recommend differentiative agents such as tretinoin and calcitriol.
"Greater ex vivo differentiation with calcitriol has been reported for patients with monosomy 7 and del(7q), whereas greater differentiation with tretinoin has been reported in FLT3 wild-type NPM1-mutated AML patients, again suggesting an important relationship between genotype and drug sensitivity," the authors wrote. "Combining both genomics-based and ex vivo functional data may further refine precision therapy, enhance the selection of rational drug combinations, and ultimately improve outcomes for patients with myeloid neoplasms."