Multiple Myeloma Differentiation-State Stratification
Stratify myeloma patients by maturation arrest state along the B-to-plasma cell differentiation axis
The Challenge
Heterogeneous Biology Across Patients
Myeloma cells arrest at various stages along B-cell to plasma cell differentiation. This biological heterogeneity, compounded by variable bone marrow microenvironment interactions, is poorly captured by standard clinical or cytogenetic classifiers.
Biomarker Signals Confounded by Maturation State
Biomarkers discovered in bulk cohorts often perform inconsistently because they may be influenced by unrecognized maturation-state variation within the plasma cell compartment. A signature enriched in plasmablast-like patients may not generalize to plasma cell-like patients, and vice versa.
Pathway Dependencies Shift Along the Differentiation Axis
Proliferative programs dominate in early-arrest (pre-plasmablast and plasmablast-like) states, while proteostatic and microenvironmental dependencies emerge in late-arrest (plasma cell-like) states. Without state-resolved analysis, these distinct biological dependencies are diluted in aggregate data.
Clinical Classifiers Miss the Functional Differentiation Axis
ISS staging and cytogenetic risk categories are prognostic but do not capture where a tumor sits along the differentiation trajectory. Two patients with identical staging can have fundamentally different maturation states and pathway dependencies.
"Our myeloma cohort looked uniform by staging and cytogenetics, but biomarkers kept failing in validation. It wasn't until we resolved differentiation states that we saw why — plasmablast-like and plasma cell-like patients had completely different pathway dependencies."
The Helomnix Solution
Helomnix projects patient molecular profiles onto a reference differentiation trajectory spanning B-cell to plasma cell maturation — from pre-plasmablast through plasmablast to mature plasma cell states. Each patient is assigned a maturation-state label reflecting where their tumor is arrested along this axis.
This applies the platform's core Patient Stratification capability to myeloma-specific differentiation biology. Patients within the same maturation state share pathway dependencies, regulatory programs, and functional characteristics that are poorly captured by conventional classifiers.
The resulting stratification supports downstream analyses including state-specific biomarker discovery, pathway characterization, and translational subgroup definition for cohort enrichment — feeding directly into target validation, companion diagnostic development, and state-informed trial design.
Unique Differentiator
Unlike cytogenetic or staging-based classifiers, differentiation-state stratification resolves functional heterogeneity along a defined biological axis — providing a mechanistic foundation for subgroup definition and translational interpretation.
How It Works
Multi-Omics Profiling
Obtain transcriptomic or multi-omics data from patient bone marrow samples. Profiles are harmonized and integrated into a unified molecular representation.
Differentiation-State Assignment
Project patient profiles onto the reference differentiation trajectory using the Digital Twin Map. Each patient receives a maturation-state label along the B-to-plasma cell differentiation axis.
State-Specific Characterization
Characterize pathway activity, regulatory programs, and molecular dependencies within each differentiation state to provide biological context for the assigned labels.
Translational Readouts & Reporting
Deliver state labels, state-specific signatures, pathway context, cohort composition summaries, and hypothesis lists for downstream translational workflows.
Real-World Application
A myeloma cohort stratified by standard risk classifiers shows heterogeneous pathway activity and inconsistent biomarker performance across patients.
Before
Standard classifiers group patients by cytogenetics and staging, but pathway activity and biomarker signals are inconsistent within each group. Candidate signatures fail to replicate, and the source of heterogeneity is unclear.
After
Differentiation-state assignment reveals that early-arrest (plasmablast-like) patients cluster with proliferative pathway dependencies, while late-arrest (plasma cell-like) patients show distinct proteostatic and microenvironmental signatures. State-resolved analysis clarifies subgroup boundaries and identifies state-specific biomarker candidates.
Outcome
State-informed stratification enables cohort enrichment criteria based on differentiation biology, supports state-specific biomarker refinement, and provides mechanistic context for downstream target validation and companion diagnostic development.
Value to Your Organization
Clearer Stratification
Resolve confounded signals by stratifying along a defined biological axis rather than clinical staging alone, reducing biomarker attrition from unrecognized maturation-state variation.
Subgroup Definition
Define translational subgroups based on shared differentiation biology, enabling state-informed cohort enrichment and biomarker refinement.
Interpretability
Provide pathway-level context for each differentiation state, supporting mechanistic interpretation of biomarker and target candidates.
Our Methodology
Data Inputs
- Bulk RNA-seq or scRNA-seq from bone marrow plasma cells
- Clinical metadata (stage, cytogenetics, prior treatments)
- Additional omics layers (optional, enhances resolution)
AI/ML Techniques
- Projection onto reference B-to-PC differentiation trajectory
- Differentiation-state signature scoring and label assignment
- Digital Twin Map projection for maturation-state positioning
- State-specific pathway and dependency characterization
- Cohort composition analysis across differentiation states
Deliverables
- Differentiation-state label per patient
- State-specific molecular signatures
- Pathway and dependency context per state
- Cohort composition summary across states
- Digital Twin Map visualization showing patient positioning
- Translational hypothesis list for downstream workflows
Discuss a Translational Application
We welcome discussions about how this approach can support your translational research.