An international team led from The University of Bergen and Haukeland University Hospital Norway, has found that blood samples taken within hours of a patient's first leukaemia tablet can predict how well the treatment will be working a year later.
The researchers analysed blood from patients with chronic myeloid leukaemia one cell at a time. Measurements taken just before and a few hours after the first dose predicted each patient's response a year later. The predictive signals differed between two widely used drugs.
The study, published in Signal Transduction and Targeted Therapy, could one day let doctors adjust treatment within days rather than months.
The work was led by Stein-Erik Gullaksen and Professor Bjørn Tore Gjertsen, at the K.G. Jebsen Centre for Myeloid Blood Cancer, Haukeland University Hospital and the University of Bergen. Researchers from six countries took part, including statisticians at Stanford University in the United States.
Months of waiting
Chronic myeloid leukaemia (CML) is a blood cancer treated with daily tablets called tyrosine kinase inhibitors. These drugs block the overactive enzyme that drives the disease, and most patients have exceptional therapy response.
However, 30–40 per cent of patients end up switching to another drug: two in three switch because of side effects. One of three patients switch medication because the leukaemia does not respond optimally. Response to therapy is checked with a blood test every three months, so it can take months to see that a drug is falling short.
Reading the blood one cell at a time
The team analysed blood from 81 newly diagnosed patients in two clinical trials: NordCML007, a clinical trial from the nordic CML study group, and the European ENEST1st trial from Novartis. Blood samples were collected just before the first tablet and again around the time each drug peaks in the blood: after one hour for dasatinib and three hours for nilotinib.
Using mass cytometry, which measures more than 40 proteins in every single cell, the researchers identified 37 types of blood and immune cells and the chemical signals active inside them. In total, they analysed 44 million cells from patients and healthy volunteers.
Within hours of the first dose, both drugs had sharply reduced the overactive signalling in leukaemia cells. This included rare cells that resemble leukemic stem cells – the cells thought to let the disease survive treatment.
Different drugs, different predictors
A machine-learning model combined each patient's measurements into a personal profile. These profiles predicted how much leukaemia would remain in the blood after 12 months. They did as well as or better than the risk score doctors use at diagnosis today.
The signals that predicted success differed between the two drugs. Dasatinib also acts on the immune system. In patients taking it, the best predictors were the amount of leukaemia at the start and the activity of T cells and natural killer cells.
Nilotinib acts more selectively on the leukaemia. In patients taking it, the best predictors were survival and cell-death signals inside the rare stem-cell-like leukaemia cells. This suggests such profiles could one day help doctors choose the best drug for each patient from the start – an idea that now needs testing in clinical trials.
“Today, patients and doctors have to wait months to learn whether a treatment is working,” says Gullaksen. “Our results show that important clues are already in the blood within hours of the first tablet. Single-cell profiling like this could become a next-generation white blood cell count – a test as precise as the targeted drugs it guides.”
What comes next
The results need to be confirmed in larger, independent groups of patients before the method can be used in routine care. Mass cytometry is a research instrument currently available mainly in specialist laboratories, so the researchers suggest a simplified test for the routine flow cytometers available at university hospitals hospitals treating CML.
The findings form the scientific basis for SWITCH, an upcoming clinical trial led by the K.G. Jebsen Centre for Myeloid Blood Cancer in Bergen. It will use an early single-cell readout to guide the choice of drug in real time. The researchers believe the approach could also work in other cancers treated with similar targeted drugs.
The paper: Gullaksen, SE., Brevik, M., Craig, E., Hellesøy, M., Sefland, Ø., Olsson-Strömberg, U., Stentoft, J., Richter, J., Gedde-Dahl, T., Tibshirani, R., Plevritis, S.K., Wolf, D., Mustjoki, S., Hjorth-Hansen, H. & Gjertsen, B.T. Deep single-cell immune- and signaling profiles predict therapy response in chronic myeloid leukemia within hours. Sig Transduct Target Ther 11, 499 (2026). https://doi.org/10.1038/s41392-026-03070-z
Web link to the publication: https://rdcu.be/Nwjf8oXsHsga
About the team:
https://cmyc.w.uib.no