Almost all solid tumors are treated with multiple drugs because the combinations have better efficacy than single-agent treatments.
"We currently do this empirically, and predictive algorithms really have not made it to the point where they can be used to identify useful drug combinations," James Korkola, assistant professor at Oregon Health & Science University, told the audience at a session on data-driven approaches to drug combinations during the American Association for Cancer Research Virtual Annual Meeting II.
But finding those combinations can be difficult in an ever-growing sea of pair-wise combinations. Testing 100 drugs in a pair-wise fashion using the Korkola lab's standard cell-survival assay in 96 well plates would require 4,950 plates for each cell line tested. It gets exponentially worse when adding additional drugs.
To help reduce the workload, rather than testing multiple doses for each combination, Korkola decided to only test a single midrange dose because most true synergistic effects are seen at midrange doses without many false positives. At a single dose, Korkola is able to test 14 drugs on a single 96 well plate, increasing the ability to test multiple cancer cell lines.
In a pilot program of 22 breast cancer cell lines, Korkola found broad synergism with combinations that included the kinase inhibitor Sutent (sunitinib, Pfizer Inc.) and antagonism with other drug combinations. To confirm the results, his lab tested eight of the combinations in a full dose-response analysis and found the dose response was generally in agreement with the pilot study, especially for the drugs that were synergistic.
The Beat AML program tested 900 samples of patients with acute myeloid leukemia (AML), including DNA and RNA sequencing, examination of the immune microenvironment and ex vivo drug sensitivity assays.
The combination drug sensitivity screen run by Jeffrey Tyner, professor at Oregon Health & Science University, found combinations with the Bcl2 inhibitor Venclexta (venetoclax, Roche Holding AG/Abbvie Inc.) were synergistic in slowing the growth or killing cell samples from AML patients.
Venclexta plus the cytotoxic chemotherapy azacitidine, which is approved by the FDA for the treatment of AML, showed a synergy in area under the curve (AUC) measurements with the combination performing better than the AUC for azacitidine (p<0.001) and Venclexta (p=0.021) as single agents. Using the genomic data, Tyner found that samples with IDH mutations were more sensitive, while mutations in the RAS pathway were more resistant to the combination, which was also seen in clinical trials testing the combination.
While it's nice to see the clinical data supporting the screen, Tyner said he found "at least half a dozen other venetoclax combinations that appear to have greater activity and greater synergy than the one that's FDA-approved."
One of those combinations, Venclexta plus the JAK1/2 inhibitor Jakafi (ruxolitinib, Incyte Corp.), showed synergy in initial AML diagnoses and had continued efficacy in seven of eight patient samples after they relapsed or were refractory to their treatment. The combination also had synergy in AML cell lines. "This and all the other data has made us so excited about this combination that we have already begun a clinical trial directed by Dr. Uma Borate, which is currently enrolling relapsed/refractory AML patients to be tested with this combination," Tyner said.
Sequential vs. combination
During the session, Lisa Tucker-Kellogg, assistant professor at Duke National University of Singapore, presented mathematical modeling and theories of evolution to predict how different dosing regimens might help increase the likelihood of success with combination treatments.
When using two drugs, it may make the most sense to stagger the dosing where the maximum exposure to one drug coincides with the trough for the other drug, so cells that are sensitive to both drugs are exposed to one of the drugs close to the maximum dosage for the longest time. Optimizing dose and schedule will need to be personalized for drugs with different clearance rates in patients.
For drugs that can't be given at full strength when combined due to toxicity, it may not be advisable to treat in combination, even if there is synergism at the lower dose, because heterogeneous tumors may have resistance to one drug in the combination, which will be more heavily selected for at the lower dose.
"We cannot automatically infer long-term winners based on short-term winners. We should choose combinations by trying to maximize long-term efficacy – long-term outcomes – not by trying to maximize short-term synergism," Tucker-Kellogg explained.
The first-strike, second-strike strategy suggests that it may be more beneficial to treat with drugs sequentially to get the tumor population below a minimal viable population at which point the patient's immune system could wipe out the remaining tumor cells or there may not be enough cells to change the environment to allow the tumor to survive.
To get below the minimal viable population, a second drug is used after a successful first drug is able to bring down the tumor population below the detectable level. While the idea makes sense, it requires giving the second drug when it isn't clear whether the first drug may have been fully successful. "In practice, there would be some considerable challenges. As human beings, are we ready to pay high costs and endure side effects and go through drug treatments to fight something that is an invisible hypothetical?" Tucker-Kellogg asked.
Finally, another strategy for sequential dosing suggests that a second drug may have an easier time killing cells that have become resistant to the first drug if there is a cost to developing that resistance. While the situation depends on the fitness cost and the likelihood for development of doubly-resistant cells, "there are a lot of cases where sequential is superior to simultaneous," Tucker-Kellogg concluded.