Single-cell blood profiles separated future responders from non-responders within hours of treatment in CML

Chronic myeloid leukemia treatment has long relied on a slow readout: residual disease measured months into therapy. This study moves the clock forward dramatically. In de novo chronic phase chronic myeloid leukemia, deep mass cytometry of peripheral blood taken on the first treatment day captured drug-specific signaling changes within 1 to 3 hours of the first oral dose and linked those early shifts to response over the first 12 months of therapy. The findings came from two multicenter clinical trials, NordCML007 and ENEST1st, and they point to a practical biologic truth in leukemia care: the first dose is already telling a story about whether the disease and the host immune system are moving in the right direction.

The approach relied on high-dimensional single-cell analysis of peripheral blood leukocytes from 81 patients and 44 million single cells, with samples taken before treatment and again within hours after dasatinib or nilotinib. Dasatinib was assessed at 100 mg once daily, with post-dose sampling 1 hour later, while nilotinib was given at 300 mg twice daily, with post-dose sampling at 3 hours. Those windows were not arbitrary. They tracked the different pharmacokinetic peaks of the two agents and allowed investigators to capture the earliest measurable intracellular effects before clinical blood counts or molecular response data would normally have changed.

One hour after dasatinib, leukocytes were already showing mobilization patterns consistent with rapid kinase inhibition and immune redistribution.

Drug-specific signaling changes emerged almost immediately after the first dose

The two tyrosine kinase inhibitors behaved differently, and that difference mattered. In leukocytes, dasatinib and nilotinib inhibited intracellular signaling within 1 or 3 hours, respectively, but each drug produced a unique signaling signature that extended beyond simple blockade of BCR::ABL1. Dasatinib’s pattern fit its broader kinase activity, including SRC-family inhibition, which plausibly explains the immune mobilization observed in blood. Nilotinib, by contrast, was more selective and did not induce the same mobilization of major immune populations. The only clear exception was a modest reduction in non-classical monocytes after the first nilotinib dose.

This divergence is clinically meaningful because it shows that early response biology is not uniform across TKIs. A drug with broader off-target kinase effects can reshape circulating immune compartments in ways that are visible almost immediately, while a more selective agent may reveal its impact through intracellular signaling states rather than overt cell trafficking. In practice, that means a single early blood draw can contain enough information to distinguish drug-specific pharmacology from patient-specific disease sensitivity.

Baseline myeloid abnormalities were extensive in untreated CML

Before treatment started, untreated patients in both trials showed expanded myeloid compartments compared with healthy donor blood. Basophils, eosinophils, neutrophil subsets, classical monocytes, and dendritic cells were all increased. Neutrophil subsets also showed reduced pSTAT3 Y705, a pattern consistent with less mature myeloid biology despite surface features that could appear relatively differentiated. Basophils showed increased global phosphotyrosine signal, reinforcing the idea that the myeloid compartment was not merely overrepresented but also biochemically abnormal.

After three months of dasatinib, many of those abnormalities had normalized. Signaling states and cell counts in most myeloid populations moved toward healthy donor ranges. A smaller residual reduction persisted in neutrophils, basophils, monocytes, and dendritic cells, aligning with the known myelosuppressive profile of TKI therapy. Neutropenia, including grade 2 and grade 3 to 4 events, occurred in 20% of patients during the first 12 months in NordCML007. The early blood signature therefore did not just reflect leukemia biology; it also captured the tension between therapeutic suppression of the malignant clone and collateral pressure on normal hematopoiesis.

Predictive features differed by TKI and by cell compartment

The most important result was not simply that early single-cell profiling changed after treatment. It was that the profile predicted treatment response during the first 12 months, measured by BCR::ABL1IS. The best predictive features were drug specific. For dasatinib, the predictive profile was dominated by leukemic burden, neutrophil maturity measured through pSTAT3 Y705, and immune effector activity in T and NK cells measured by pERK 1/2. For nilotinib, prediction depended more on intrinsic signaling states within CD34+CD38lowCD25+ leukemic stem cell-like cells, including cleaved caspase-3, pCRKL Y207, and total and Y705 STAT3.

The biology is coherent. Dasatinib’s broader kinase profile can modify immune effector compartments as well as the leukemic milieu, so response prediction naturally draws on both tumor burden and immune competence. Nilotinib’s more selective pattern appears to place heavier weight on the internal signaling tone of stem-cell-like leukemic populations, which are the cells most likely to sustain disease if they remain biochemically resilient. In other words, one drug’s early fingerprint lives in the host-response interface, while the other’s sits deeper inside the leukemic stem compartment.

A response model built from hours, not months

The practical appeal of this work lies in timing. Standard response assessment in CML still depends on molecular monitoring that unfolds over months, yet the cells already exposed to therapy on day one can reveal whether the pathway blockade is engaging the right biology. Here, single-cell immune and signaling profiles measured within hours of first dose anticipated treatment response over the next year. That opens the door to earlier optimization of kinase inhibitor choice, potentially within days rather than months, if the approach can be translated into a clinically workable assay.

The study also clarifies the limitations of relying solely on bulk molecular measurements early in therapy. A drop in aggregate BCR::ABL1 transcript burden is informative, but it arrives late and compresses distinct biologic processes into one number. Single-cell profiling separates those processes: leukemic burden, myeloid maturation, immune effector activity, and stem-cell signaling can each be tracked independently, giving clinicians a more detailed map of who is likely to do well and who may need a different strategy.

The next step is not conceptual; it is operational. This work still needs validation in a setting that can deliver the same depth of immune and signaling information quickly enough to influence frontline treatment decisions, while preserving the ability to predict true 12-month response rather than short-lived pharmacodynamic noise.


Reporting by Dr. Revathy Mohan V.R., MBBS for Medicine World Foundation, with additional data provided by Nature Medical Research.
PRIMARY CLINICAL EVIDENCE & REGISTRY DATA