Relapsed or refractory (R/R) acute myeloid leukemia (AML) remains difficult to treat. Median survival after an R/R diagnosis is about five months, and fewer than 13 percent of patients live beyond five years. These figures highlight the pressing need for better therapeutic options.
Combination therapies can improve patient outcomes, but the massive number of possible drug combinations vastly exceeds what could be tested in clinical trials or patient cells. The excessive inter- and intra-patient heterogeneity necessitate personalized treatment optimization approaches that effectively exploit patient-specific vulnerabilities to selectively target both the disease- and resistance-driving cell populations.
In a recent article published in Cancer Research, a team led by Professor Tero Aittokallio developed an experimental-computational approach that leverages single-cell transcriptomics and single-agent response profiles from primary patient samples. The method pinpoints targeted drug combinations that selectively co-inhibit treatment-resistant cancer cells in each AML patient sample.
Cancer cell-selectivity is important to avoid severe co-inhibition of non-malignant cells in the relapsed stage, and to de-prioritize broadly toxic combinations, with the aim of improving both combination efficacy and avoiding adverse drug effects.
The predictive approach provided a rational means to identify personalized combinatorial regimens for individual AML patients with R/R disease that synergistically and selectively target treatment-resistant leukemic cells, thereby increasing their likelihood for clinical translation.
In application to patient samples from the VenEx clinical trial, the approach predicted clinical outcomes of AML patients to the venetoclax-azacitidine combination, which is currently widely used for treatment of de novo AML in patients ineligible for intensive chemotherapy.
“We observed that cell type compositions evolve dynamically between the diagnostic and R/R stages uniquely in each patient, hence requiring personalized drug combination strategies to target therapy-resistant cancer cells”, says the first author of the study, Doctoral Researcher Yingjia Chen from the Institute for Molecular Medicine Finland (FIMM).
The machine learning modelling approach is patient-tailored, that is, it does not require large patient cohorts. It enables one to evaluate the selective synergy of more than 10,000 combinations in less than 10 seconds per patient on a standard desk-top machine.
Using flow cytometry-based single-cell drug combination assays, the researchers confirmed that the patient-specific combination predictions indeed led to an improved synergy and selectivity of co-inhibiting treatment-resistant AML cells in a patient-tailored manner
“This is the first systematic approach to prioritize selective and personalized drug combinations for relapsed disease using the transcriptomic history of the disease progression from longitudinal patient samples”, says professor Tero Aittokallio who led the study.
Since treatment resistance is a critical challenge for various treatments and cancer types, and combination therapies are often required to improve clinical benefits for most patients with advanced, relapsed or refractory malignancies, the predictive approach is expected to become of significance beyond AML.
Selective combinations with molecularly targeted drugs have the potential not only to overcome resistance, but to boost the response of first-line therapies, reduce dose-limiting single agent toxicity and hence expand the range of treatment possibilities.
The study was done in collaboration with researchers from Helsinki University Hospital, University of Helsinki (FIMM, HiLIFE and Faculty of Medicine), iCAN Digital Precision Cancer Medicine Flagship project, and Biotech Research and Innovation Centre (BRIC) of the University of Copenhagen.
Original publication: Chen Y, He L, Ianevski A, Nader K, Ruokoranta T, Linnavirta N, Miettinen JJ, Vähä-Koskela M, Vänttinen I, Kuusanmäki H, Kontro M, Porkka K, Wennerberg K, Heckman CA, Giri AK, Aittokallio T. A Machine Learning-Based Strategy Predicts Selective and Synergistic Drug Combinations for Relapsed Acute Myeloid Leukemia. Cancer Research 2025. doi: 10.1158/0008-5472.CAN-24-3840.