Analysis
Clinical Flow
Blinatumomab response: what T-cell state and changing denominators reveal
A 34-patient retrospective B-ALL investigation links CD8 memory state and TIM-3 expansion to blinatumomab response, but different assays used 19, 13, 26 and 17 patients.
Blinatumomab recruits a patient's T cells to recognize CD19-positive B-lineage leukemia. It sounds as though the only question is whether sufficient T cells remain. A recent multi-omic investigation suggests a more interesting possibility: the functional state T cells bring into treatment may help explain why some patients respond and others develop exhausted cells during continuous CD3 engagement.
Corrado and colleagues investigated 34 patients with B-cell precursor acute lymphoblastic leukemia receiving blinatumomab at LMU University Hospital, reporting the results in Frontiers in Immunology on 20 August 2026 (10.3389/fimmu.2026.1895742). Their samples span a long clinical period (2010–2023). Because participation depended on stored peripheral-blood mononuclear cells or serum, the analysis is observational and cannot by itself establish a clinically validated prediction test.
The authors classified response after cycle one based on baseline disease: patients with morphological leukemia needed complete remission, while those starting with MRD-positive disease needed to become MRD-negative. That difference in entry state matters when interpreting immune biomarkers and comparing patients.

Four assays, four different denominators
The study integrated multiparameter flow cytometry in 19 patients, single-cell CITE-seq in 13, ex vivo redirected cytotoxicity in 26, and serum proteomics in 17. These are overlapping investigative subsets, not four tests completed in all 34 participants. Combining the results into a mechanistic hypothesis is legitimate; treating every association as if measured in a unified 34-patient multimodal dataset is not.
For flow analysis, peripheral-blood mononuclear cells were assessed before treatment and at approximately weekly intervals in the first cycle. The investigators used Navios, FlowJo and CD3/CD4/CD8 with CCR7 and CD45RA to identify naïve, central memory, effector memory and TEMRA populations. PD-1 was available throughout, whereas TIM-3 and LAG-3 antibodies were introduced later and evaluated in only 12 patients. Samples with fewer than 30 CD3-positive cells per microlitre were excluded from subsequent subset analysis. Such changing panels and denominators require careful handling in longitudinal comparisons.
The difference begins before treatment
Responders showed greater representation of CD8 effector-memory T cells with cytotoxic transcriptional features, including the transcription factor ZNF683. In non-responders, corresponding cells could appear activated at the transcript level without a matching cytotoxic commitment. This suggests that neither total CD8 abundance nor conventional activation status alone captures effective target-cell killing.
During the first cycle, non-responders exhibited expansion of TIM-3-positive CD8 cells and progressive deterioration of ex vivo cytotoxic function. The magnitude of TIM-3 expansion was inversely associated with baseline ZNF683 expression. The longitudinal pattern is biologically plausible, but it does not prove TIM-3 caused resistance: chronic stimulation, disease burden and changes in T-cell composition may all contribute.
The microenvironment matters too
The serum analysis found differences in factors associated with immunosuppressive macrophage polarization and checkpoint interactions, including CSF-1, HGF and the TIM-3 ligand Galectin-9. Their levels correlated with the expanding TIM-3-positive population. Together with single-cell transcriptional data, these observations provide a hypothesis connecting pre-existing cytotoxic programming, treatment-associated exhaustion and a suppressive milieu.
A correlation across modalities is not a causal rescue experiment. This small retrospective study does not establish that adding a TIM-3 inhibitor, blocking Galectin-9 or modifying macrophages will improve clinical outcomes. Such therapeutic implications remain to be tested in appropriately controlled studies.
What cytometry laboratories can take from it
For researchers designing longitudinal blinatumomab monitoring, the useful question is not whether to add every checkpoint marker to a diagnostic tube. It is whether a validated research panel, standardized sample time, adequate event collection and carefully defined memory/activation gates could distinguish numerical loss from phenotypic exhaustion. PD-1 and TIM-3 positivity should not be interpreted as a diagnosis of functional exhaustion without functional evidence and biological context.
Any attempt at clinical risk stratification would require prespecified markers, thresholds, analytical validation, external patient cohorts, harmonized sample handling and evaluation alongside leukemia burden and treatment factors. The paper offers a strong research hypothesis, not an approved decision algorithm and not a change to MRD thresholds.
Daily Learning Pill: In a bispecific T-cell engager assay, cell numbers, memory phenotype, checkpoint expression and functional cytotoxicity measure related but non-equivalent dimensions of response.
Read more on CFCM: How AI assists B-ALL MRD interpretation.
Source and evidence tier: Corrado F et al., Frontiers in Immunology, peer-reviewed observational original research, 20 August 2026; DOI 10.3389/fimmu.2026.1895742. Original CFCM illustration; no journal figure reproduced.