Gene Signature May Help Predict Treatment-Free Remission in CML
Key Clinical Summary:
- Researchers developed a 50-gene transcriptomic signature that distinguished patients with chronic myeloid leukemia (CML) who maintained treatment-free remission (TFR) for 2 years after tyrosine kinase inhibitor (TKI) discontinuation from those who experienced molecular relapse.
- The model demonstrated an area under the receiver operating characteristic curve (AUROC) of 0.83 in the training cohort and retained predictive performance in an independent real-world cohort.
- Patients with a high TFR signature demonstrated distinct immune characteristics, including greater representation of myeloid immune and natural killer T cells and greater immune repertoire polyclonality.
A transcriptome-based biomarker may help identify patients with CML who are more likely to maintain TFR after stopping TKI therapy, potentially addressing an important gap in selecting patients for treatment discontinuation.
Although TFR has become an important therapeutic objective in CML, approximately half of patients experience molecular relapse after stopping TKI therapy. Researchers therefore evaluated whether gene expression patterns in peripheral blood at the time of treatment discontinuation could predict sustained remission.
50-Gene Signature Predicts TFR
Investigators analyzed peripheral blood cell transcriptomes from 96 patients enrolled in the multicenter STIM2 trial who discontinued imatinib.
Using a DESeq2-based machine learning approach, researchers identified a 50-gene signature capable of distinguishing patients who maintained TFR for 2 years from those who subsequently experienced molecular relapse.
model achieved an AUROC of 0.83 (95% CI, 0.73-0.93) in the training cohort and 0.75 (95% CI, 0.55-1.00) in the internal validation cohort.
Researchers then tested the signature in an independent real-world cohort of 70 patients attempting discontinuation of imatinib or nilotinib. In this external cohort, the model achieved an AUROC of 0.71 (95% CI, 0.58-0.83) for predicting 2-year TFR.
Performance was higher when the external analysis was restricted to patients previously treated with imatinib, with an AUROC of 0.77 (95% CI, 0.61-0.92).
The signature also differentiated patients according to time to molecular relapse, with a significant difference between signature groups (log-rank P = .0042).
Immune Profiles Differed by TFR Signature
Beyond predicting treatment outcomes, the transcriptomic analysis identified biologic differences that may provide insight into mechanisms underlying sustained TFR.
Patients classified in the high TFR-signature group had a greater proportion of myeloid immune cells and natural killer T cells. Their transcriptomic profiles were also enriched for Hedgehog signaling.
In contrast, patients in the low TFR-signature group had a greater proportion of lymphoid cells, enrichment of mammalian target of rapamycin signaling, and a trend toward increased oxidative phosphorylation.
T-cell receptor and immunoglobulin heavy-chain repertoire analyses also demonstrated significantly greater polyclonality among patients in the high TFR-signature group, suggesting that differences in immune composition and diversity may be associated with the ability to maintain remission without continued TKI therapy.
Implications for Managed Care
Accurately identifying patients who can successfully discontinue TKI therapy has potential implications for both patient care and health care utilization.
For patients who sustain TFR, successful discontinuation can eliminate the ongoing treatment burden associated with long-term TKI therapy. A validated biomarker capable of better estimating the likelihood of sustained TFR could potentially support more individualized discussions around treatment cessation and follow-up.
However, the findings do not establish the 50-gene signature as a clinical decision tool. Although predictive performance persisted in an independent real-world cohort, the external analysis included only 70 patients, and performance differed between the overall population and the subgroup previously treated with imatinib.
Additional validation will therefore be necessary before transcriptomic profiling can be incorporated into routine CML management.
Conclusion
A 50-gene peripheral blood signature demonstrated the ability to predict 2-year TFR following TKI discontinuation in patients with CML and was associated with distinct immune profiles. Further validation could clarify whether transcriptomic testing can help identify patients most likely to maintain molecular remission after stopping therapy.
Reference
Alcazer V, Dulucq S, Mosnier I, et al. Development and external validation of a transcriptome-based multivariable prediction model for treatment-free remission in chronic myeloid leukemia. J Clin Oncol. 2026;44(21). doi:10.1200/JCO-25-02948
