S124: PHOSPHOPROTEOMICS ACCURATELY PREDICTS RESPONSES TO MIDOSTAURIN PLUS CHEMOTHERAPY IN TWO INDEPENDENT COHORTS OF FLT3 MUTANT-POSITIVE ACUTE MYELOID LEUKAEMIA
Bibliographic record
Abstract
Background: Midostaurin plus intensive chemotherapy (M+IC) is approved for FLT3 mutant-positive (FLT3-MP) acute myeloid leukaemia (AML). The presence of refractory/early relapse (R/ER) disease following M+IC treatment suggests the existence of FLT3-independent determinants of M+IC response (Stone et al. NEJM 2017). We have previously reported a phosphoproteomic signature significantly elevated in primary AML blasts that responded to midostaurin ex vivo (Casado et al., 2018, Leukemia). Aims: To build and test a phosphoproteomics-based model to predict M+IC response from FLT3-MP AML patient samples collected at diagnosis. Methods: We retrospectively analysed peripheral blood (PB, n=37) and/or bone marrow (BM, n=34) diagnosis samples of 47 FLT3-MP AML patients subsequently treated with M+IC (median age at diagnosis 61, range 19-79y) using liquid chromatography-tandem mass spectrometry and MS1-based peptide quantification for phosphoproteomics analysis. Data from patients with extreme response profiles were used for model building; the “good-responder” (GR) group had a disease-free survival (DFS)>24 months (n=20), whereas the R/ER group had DFS<6 months (n=14, including refractory patients). Multivariate analysis and machine learning were used to build a phosphoproteomic signature-based model capable of predicting M+IC response from diagnosis samples. The model was validated on an independent, blinded retrospective set of 13 diagnosis FLT3-MP AML samples (median age 60, age range 33-73y, 9xPB and 4xBM). Results: In this study, we identify a highly-predictive phosphoproteomic signature of M+IC response in FLT3-MP AML diagnosis samples, and test it on an independent, blinded patient cohort. First, multivariate analysis of phosphoproteomic data identified several biochemically different groups of AML cases (Fig. 1A), highlighting potential distinct mechanisms of drug response. GR1 and GR2 groups showed upregulation of DNA damage response (DDR), and downregulation of receptor tyrosine kinase (RTK) signalling, and either downregulation of immune response (IR) pathways (GR1), or upregulation of chromatin remodellers (GR2). GR3 showed upregulation of RTK signalling and IR pathways, and downregulation of DDR. A phosphoproteomic signature made of a subset of more than a hundred phosphopeptides discriminating between at least two of these four patient groups (R/ER, GR1-GR3) was used to build a response-prediction model. On the expanded training dataset, including patients with DFS between 6 months and 24 months (n=13), response stratification was achieved with log rank p<1x10-9 (not shown); median DFS was 17.7 weeks for the signature-negative patients, and was not reached for signature-positive patients. The model was then tested on a blinded independent cohort of 13 FLT3-MP patients (Fig. 1B and C), with those positive for our signature showing markedly increased survival than signature-negative patients (median DFS 0 weeks vs not reached, log-rank p<0.0008). The overall model accuracy, with “response” defined as DFS>6 months, was 100% for signature-negative samples (5/5) and 85% for signature-positive samples (6/7, data was censored before 6 months for one patient). Summary/Conclusion: Using MS1-based quantitation of phosphoproteomic data, we identified several potential mechanisms of sensitivity to M+IC. Accounting for response heterogeneity enabled the creation of a model based on a highly-predictive phosphoproteomic signature of M+IC response. In an independent blinded patient cohort of 13 FLT3-MP patients this model predicted M+IC response with 92% accuracy.Keywords: Survival prediction, Acute myeloid leukemia, flt3 inhibitor, Phosphorylation
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.002 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
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 teacher head, 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".