Hematology › Acute Myeloid Leukemia

AI-Driven Benefit-Risk Analysis for Acute Myeloid Leukemia

AML BRA spans venetoclax-based combinations, FLT3 inhibitors (midostaurin, gilteritinib), IDH1/2 inhibitors (ivosidenib, enasidenib), and gemtuzumab ozogamicin. ArcaScience delivers mutation-directed safety profiling across fit and unfit patient populations, with specialized models for tumor lysis syndrome, differentiation syndrome, and treatment-related mortality in elderly patients.

See Platform Request Demo

160,000+

New AML cases per year globally

1,500+

AML clinical trials analyzed

$4B

Global AML therapeutics market

10+

Targeted AML agents approved since 2017

Disease Overview

Why AML Demands Specialized BRA

AML is a genetically heterogeneous malignancy where molecular profiling (FLT3, IDH1/2, NPM1, TP53, RUNX1) dictates treatment selection. The median age at diagnosis is 68 years, and the majority of patients are ineligible for intensive chemotherapy, creating a critical need for BRA that balances efficacy against treatment-related mortality in frail and comorbid populations.

Tumor Lysis Syndrome with Venetoclax

Venetoclax combined with azacitidine or low-dose cytarabine has become standard of care for unfit AML patients, but carries significant tumor lysis syndrome (TLS) risk requiring mandatory dose ramp-up over 3–4 days with hospitalization. BRA must model TLS incidence by baseline white blood cell count, renal function, and tumor burden, alongside the ramp-up protocol's impact on early mortality and treatment delay.

QTc Prolongation & FLT3 Inhibitor Safety

Gilteritinib carries QTc prolongation risk requiring ECG monitoring, while midostaurin combined with intensive chemotherapy increases febrile neutropenia severity and duration. BRA for FLT3-mutated AML must account for drug-drug interactions with azole antifungals (CYP3A4), the cardiac safety profile during prolonged maintenance therapy, and comparative safety between type I and type II FLT3 inhibitors.

Differentiation Syndrome with IDH Inhibitors

Ivosidenib (IDH1) and enasidenib (IDH2) can induce differentiation syndrome in 10–20% of patients, a potentially fatal complication resembling retinoic acid syndrome in APL. BRA must model early recognition, dexamethasone intervention, and the benefit-risk of continuing therapy through differentiation syndrome events versus treatment discontinuation, alongside hepatotoxicity monitoring for ivosidenib.

Platform Capabilities

How ArcaScience Addresses AML BRA

Our platform integrates mutation-specific AML data, elderly/unfit patient risk models, and regulatory templates aligned with hematologic oncology submission requirements across FDA, EMA, and accelerated approval pathways.

Data Intelligence

AML Data Coverage

1,500+ AML clinical trials including VIALE-A, ADMIRAL, RATIFY, AGILE, and BRIGHT AML 1003 datasets. Adverse event data covering venetoclax TLS, gilteritinib QTc prolongation, IDH inhibitor differentiation syndrome, gemtuzumab ozogamicin hepatic sinusoidal obstruction syndrome, and intensive chemotherapy treatment-related mortality across all cytogenetic and molecular risk groups.

Explore Data Engine →

Decision Intelligence

Mutation-Directed AI Models

AI models for TLS risk stratification during venetoclax ramp-up, differentiation syndrome early detection for IDH-mutated patients, QTc modeling during FLT3 inhibitor therapy, and treatment-related mortality prediction using fitness assessment scores (ECOG, HCT-CI). BRAT framework application with AML-specific regulatory precedent from ODAC deliberations on accelerated approvals and confirmatory trial design.

Explore AI Models →

Automated Outputs

AML Regulatory Outputs

PSURs with TLS and differentiation syndrome deep-dives, RMPs with cardiac monitoring protocols for FLT3 inhibitors and hepatotoxicity surveillance for GO and IDH inhibitors, CTD 2.5 with CR/CRi, OS, and MRD endpoint summaries, and HEOR reports supporting reimbursement submissions for targeted therapies in mutation-defined AML subpopulations.

Explore Outputs →

AML Intelligence

Platform Performance in AML

6,200,000,000+

AML adverse event data points

65%

Faster TLS risk signal identification

Molecular subtypes modeled

AML submissions supported

Case Evidence — AML

Venetoclax-Based Regimen Safety Optimization in Unfit Elderly AML

Challenge

A pharma company developing a venetoclax combination regimen for elderly AML patients (median age 76, ECOG 2–3) needed to model the benefit-risk of their TLS prophylaxis protocol alongside treatment-related mortality rates across comorbidity strata. Regulatory agencies required evidence that the ramp-up schedule and hospitalization requirements were optimized for this fragile population.

Result

ArcaScience deployed TLS risk stratification models integrating baseline WBC count, LDH, renal function, and bone marrow blast percentage, enabling patient-level risk scoring that guided individualized ramp-up durations. The BRA identified a subpopulation where outpatient ramp-up was feasible, reducing hospitalization burden while maintaining TLS safety. Early mortality modeling supported optimized dose modification guidelines accepted by both FDA and EMA.

2.5x

Faster TLS risk stratification

35%

Reduction in early treatment-related mortality uncertainty

View All Case Studies →

The ability to model treatment-related mortality across comorbidity strata was exactly what we needed for our elderly AML population. ArcaScience's TLS risk stratification directly informed our revised ramp-up protocol, and the data-driven approach gave regulators confidence in our outpatient dosing proposal for lower-risk patients.

Head of Hematology Clinical Safety

Global Pharma Company

Frequently Asked Questions

AML Benefit-Risk Analysis

Our platform stratifies BRA by molecular subtype. For FLT3-ITD and FLT3-TKD mutations, we model midostaurin (induction) and gilteritinib (relapsed/refractory) safety including QTc prolongation and CYP3A4-mediated drug interactions. For IDH1-mutated AML, ivosidenib BRA covers differentiation syndrome and hepatotoxicity. For IDH2 mutations, enasidenib BRA includes differentiation syndrome and indirect hyperbilirubinemia. Each molecular subgroup receives tailored benefit metrics (CR/CRi rate, OS, MRD negativity) weighed against mutation-specific and regimen-specific adverse events, reflecting the ELN 2022 risk classification framework.

ArcaScience integrates fitness assessment scores (ECOG PS, HCT-CI, Wheatley index) with age, organ function, and cytogenetic risk to model treatment-related mortality (TRM) across regimen intensity levels. For venetoclax + azacitidine in patients ≥75 years with ECOG 2–3, the platform models 30-day and 60-day TRM alongside CR/CRi rates and OS. For patients on the cusp of fitness for intensive chemotherapy, we provide comparative BRA between intensive 7+3 and lower-intensity venetoclax-based approaches, helping clinicians and regulators assess whether intensive therapy's higher TRM is justified by superior remission depth.

Yes. Our platform covers allogeneic stem cell transplant conditioning BRA including myeloablative (MAC) versus reduced-intensity conditioning (RIC) safety profiles. We model graft-versus-host disease (GVHD) incidence and severity, veno-occlusive disease/sinusoidal obstruction syndrome risk, non-relapse mortality, and the benefit-risk of novel GVHD prophylaxis regimens (post-transplant cyclophosphamide, abatacept). The BRA integrates pre-transplant fitness, donor type, and conditioning intensity to provide individualized transplant safety assessments.

AML maintenance therapy is an evolving paradigm with oral azacitidine (CC-486/ONUREG) now established post-intensive chemotherapy. Our platform models the benefit-risk of prolonged maintenance including cumulative cytopenias, gastrointestinal toxicity, quality-of-life impact, and relapse-free survival benefit. For FLT3-mutated AML, we model post-transplant sorafenib or gilteritinib maintenance safety including prolonged QTc exposure risk and drug interaction management with immunosuppressants. The BRA accounts for the critical question of maintenance duration optimization.

Yes. Measurable residual disease (MRD) is increasingly used as an early efficacy endpoint and treatment decision tool in AML. ArcaScience models the benefit-risk implications of MRD-guided therapy escalation (e.g., proceeding to transplant for MRD-positive patients) versus de-escalation (e.g., omitting transplant for MRD-negative favorable-risk patients). The platform integrates multiparameter flow cytometry and molecular MRD data (NPM1, CBF) to assess whether MRD-directed treatment modifications alter the safety profile acceptably relative to relapse risk reduction, supporting both clinical trial design and regulatory discussions around MRD as a surrogate endpoint.

See ArcaScience Applied to AML

Request a demonstration of ArcaScience's platform configured for acute myeloid leukemia benefit-risk analysis. Our hematology scientists will walk through mutation-specific data coverage, TLS and differentiation syndrome models, elderly patient risk assessment, and regulatory output examples.

Request AML Briefing Explore the Platform →