A rapid gene expression profiler predicts tumor responsiveness and patient outcome for standard-of-care therapies in acute myeloid leukemia
Bibliographic record
Abstract
Abstract Introduction: The molecular heterogeneity of acute myeloid leukemia (AML) complicates clinical assessments for optimal therapeutic selection and management of emergent resistance based on mutations and cytogenetics alone. The incorporation of gene expression profiles has the potential to augment current prognostic stratification. In patients treated with intensive chemotherapy, a 17-gene expression-based signature of leukemia stem cells (LSC17) is predictive of survival outcomes and responsiveness (Ng et al., 2016). For patients ineligible for intensive chemotherapy, less differentiated tumors are more sensitive to hypomethylating agents plus the BCL2 inhibitor venetoclax (HMA+Ven), whereas differentiated tumors exhibit reduced sensitivity (Pei et al., 2020; Kuusanmäki et al., 2020). Transcriptomic classification of AML tumors according to differentiation state (van Galen et al., 2019) aligns with ex vivo Ven sensitivity (Bottomly et al., 2022; Zeng et al., 2022). Methods: We developed a rapid, gene expression-based assay called Myelo-ID to simultaneously profile AML stemness (LSC17), tumor differentiation states, and apoptotic target gene levels. RNA extracted from freshly isolated primary AML patient mononuclear cells was used to synthesize cDNA. Samples were loaded onto a custom TaqMan Array Card panel for the targeted genes and amplified by quantitative PCR (qPCR). Expression data were analyzed for quality control, normalization, calculation of enrichment scores, and report generation in a custom R workflow. Banked AML patient samples with matched RNAseq and ex vivo Aza+Ven drug sensitivity data were used to validate tumor cell state scores. LSC17 scores were validated through an exchange of reference cohort patient samples analyzed using a clinical Nanostring-based assay (Ng et al., 2022). Clinical, genetic, and treatment information was curated from patient medical records. Features for HMA+Ven-treated patients were analyzed by univariable and multivariable regression models for predictors (odds ratio (OR) or Cox hazard ratio (HR), area under the receiver operator characteristic curve (ROC AUC)) for achievement of composite complete remission (cCR) and overall survival (OS). Results: Profiling of 100 newly diagnosed (ND) AML patient samples spanning a range of clinical and genetic features revealed diverse patterns of cell state signature enrichment, in contrast to balanced expression in healthy marrow. Cell state enrichment scores were concordant with RNAseq-derived scores from the same samples (n=40; Pearson r: 0.531 to 0.943). LSC17 scores were highly correlated with scores derived from a Nanostring-based clinical assay (n=50; Pearson r=0.931), with significantly lower scores in patients achieving CR following 7+3 induction (p=2.76E-6). Primitive tumor expression profiles and increased BCL2 gene expression aligned with Aza+Ven ex vivo sensitivity (n=53; Pearson r: -0.623). Univariable analysis of 108 clinical, genetic, and gene expression features among 58 ND-AML patients treated with HMA+Ven identified mutated DNMT3A associated with both achieving cCR (OR: 7.22, p=0.016) and improved OS (HR: 0.37, p=0.010). Leukocytosis (OR: 0.16, p=0.005) and high promonocyte-like gene score (OR: 0.19, p=0.009) tracked with reduced likelihood of response; antecedent hematologic disorder (HR: 2.38, p=0.019) and mutated TP53 (HR: 2.89, p=0.005) were associated with shorter OS. High BCL2 gene expression was the most significant predictor of survival among all features examined (median OS: 19.6 vs 6.0 months; HR: 0.35, p=0.002). Multivariable regression models of the top clinical/mutation (leukocytosis, FLT3-ITD, mutated DNMT3A and RUNX1) and gene expression (promonocyte-like high) predictors for cCR showed value separately (ROC AUC: 0.890 and 0.693, respectively) but improved discrimination for responsiveness as a combined clinical/genetic/expression model (ROC AUC 0.920). High BCL2 gene expression alone was comparably predictive of 1-yr OS as the top clinical/genetic features (mutated DNMT3A andTP53; ROC AUC: 0.702 vs 0.688, respectively), with improved predictive value in a model combining these features (ROC AUC: 0.808).Conclusion:Incorporation of a rapid gene expression-based profiler of stemness, cell state, and apoptotic target genes, such as Myelo-ID assay, into the prospective evaluation of newly diagnosed AML patients may augment current clinical/genetic labs to help guide treatment selection and improve outcomes.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".