The study, published in Frontiers in Oncology, set out to find transcriptomic signatures — patterns of gene activity — that could distinguish patients destined for remission from those destined for resistance. The researchers note that transcriptional profiling can capture biological phenotypes beyond genetic mutations, but that regimen-specific predictive biomarkers for 7+3 therapy had not been established due to methodological limitations in existing studies. Their strategy was to interrogate two large, independent AML cohorts drawn from different populations, reasoning that any signal strong enough to survive both would be more robust.

The first cohort, OHSU BeatAML2.0, comprised 264 patients, while the second, designated SIH/JIH/ZIH AML, included 582 patients. In each dataset, the team compared gene expression profiles between patients who achieved complete remission after 7+3 induction and those who proved refractory. To sift through thousands of candidate genes without overfitting the models to statistical noise, the researchers employed machine learning algorithms, most notably the least absolute shrinkage and selection operator, or LASSO, a technique that penalizes complexity and forces the model to retain only genes with genuine predictive power for therapy response.

When the gene lists generated from the two cohorts were intersected, seven genes emerged as consistently associated with 7+3 therapy response across both ethnically diverse patient groups. That convergence was the study’s first important result: a small set of transcripts whose expression patterns tracked treatment outcome regardless of the population being examined. But a predictive biomarker is only useful if it can be measured reliably, so the team moved to an independent validation step using quantitative reverse transcription PCR on samples from 58 additional AML patients.

The validation narrowed the field dramatically. Of the seven candidate genes, only two — TMEM150C and F2R — were reliably detectable in the independent patient samples, and both were significantly elevated in patients whose leukemia proved refractory to induction. TMEM150C expression was 4.2-fold higher in refractory cases, while F2R was 2.9-fold higher. The consistency of those differences, measured in a patient group entirely separate from the discovery cohorts, suggested that the two transcripts were not statistical artifacts but genuine molecular correlates of treatment resistance.

To probe the functional relevance of the two genes, the researchers turned to in vitro assays using AML cell lines. When the team examined cell lines with elevated TMEM150C and F2R expression, they found that these cells exhibited reduced sensitivity to cytarabine, one of the two backbone drugs of the 7+3 regimen. The result provides a functional bridge between the correlative transcriptomic data and the behavior of leukemia cells in laboratory models: it is one thing for a gene to be switched on in resistant patients, and another for its heightened expression to coincide with measurable drug resistance in cell culture.

Taken together, the study’s three pillars — discovery across two ethnically diverse cohorts, independent validation in 58 patient samples, and functional assays in cell lines — form a converging case that TMEM150C and F2R may serve as potential predictive biomarkers for 7+3 therapy response in AML. The authors describe the genes as potential rather than proven biomarkers, and the findings will require further testing before they can inform clinical decisions.

The work also underscores a broader point made by the researchers: transcriptional profiling can capture biological phenotypes that genetic mutations alone do not explain, and regimen-specific biomarkers have been lacking for the most widely used induction therapy in AML. For patients who currently undergo 7+3 induction with no way of knowing in advance whether it will succeed, the two overexpressed transcripts identified here offer a candidate molecular signal — one that, if validated in future studies, could support more personalized treatment selection in this heterogeneous malignancy.

Subject of Research: Transcriptomic biomarkers of refractoriness to 7+3 induction therapy in acute myeloid leukemia

Article Title: TMEM150C and F2R as transcriptomic signatures for refractoriness to 7 + 3 induction therapy in acute myeloid leukemia

Article References: Choi, Y.-J., Kim, J., Mai, X. L., de Guzman, A. C. V., An, Y. J., Kim, H. S., Cho, B.-S., & Park, S. (2026). TMEM150C and F2R as transcriptomic signatures for refractoriness to 7 + 3 induction therapy in acute myeloid leukemia. Frontiers in Oncology, 16, Article 1926332. https://doi.org/10.3389/fonc.2026.1926332

Image Credits: AI Generated

DOI: 10.3389/fonc.2026.1926332

Keywords: acute myeloid leukemia, 7+3 induction therapy, TMEM150C, F2R, transcriptomics, predictive biomarkers, refractory disease, LASSO, machine learning, cytarabine, qRT-PCR, hematologic malignancies