Identifying a Venetoclax-Sensitive AML Subset with Matched Proteomics & Ex-Vivo Models
This case study documents how Yatiri Bio correlated baseline global mass-spectrometry proteomics across 40+ patient-derived cell models (ProteoModels) with ex-vivo drug sensitivity screens. By mapping active apoptosis cascades, we discovered a predictive 5-protein signature that isolates patient subsets highly responsive to the Bcl-2 inhibitor Venetoclax, bypassing genomic mutations alone.
Introduction & Background Context
In oncology drug development, identifying patient populations that will show clinical response is vital. As documented by Wong, Siah & Lo (2019), oncology therapeutic candidates face a 95% failure rate when entering clinical development phases, primarily due to a lack of clinical efficacy or unexpected bypass mechanisms. Audit data confirms that 90% of clinical drug failures stem from a lack of efficacy (40-50%) or safety issues (30%) (Sun et al., 2022).
Venetoclax is an FDA-approved small-molecule inhibitor of the anti-apoptotic protein Bcl-2. While highly effective in select subsets of Acute Myeloid Leukemia (AML) patients, response rates are heterogeneous, and resistance commonly emerges through up-regulation of related anti-apoptotic family members like Mcl-1 and Bcl-xL. Crucially, genomic sequencing (DNA-seq) cannot predict these dynamic pathway shifts, as they are regulated at the post-translational and protein abundance levels.
"By utilizing global mass spectrometry, we are not looking at static genomic potential; we are monitoring the active functional engine of the leukemia cell. Our ProteoModels allow us to expose the exact protein imbalances that dictate whether a tumor will succumb to Bcl-2 inhibition or leverage Mcl-1 to survive." — Dr. Pilgrim Jackson, CEO, Yatiri Bio, Inc. (San Diego Business Journal interview)
Study Design & Methodology
Yatiri Bio compiled a cohort of 42 low-passage, patient-derived AML primary models representing diverse mutational backgrounds (FLT3, IDH1/2, NPM1, TP53). The workflow followed our structured clinical pipeline:
- Baseline Proteomics: Every AML model underwent high-throughput global mass-spectrometry profiling at our Nancy Ridge lab. Over 8,200 total proteins and 12,000 phosphosites were quantified per model.
- Ex-Vivo Sensitivity Screening: The models were subjected to a 10-point concentration gradient of Venetoclax. Cell viability and caspase-3/7 apoptotic activity were monitored.
- Algorithmic Correlation: Using machine learning classifiers inside the ProteoBrowser™ portal, we correlated baseline protein abundance ratios with the resulting GI50 (growth inhibition) profiles.
Results & Signature Extraction
The sensitivity screen categorized the 42 AML models into two distinct groups: Responders (GI50 < 50 nM) and Non-Responders (GI50 > 500 nM).
Genomic mutational status showed poor correlation with response. For instance, FLT3-mutated models were distributed across both responder groups. However, our baseline proteomic analysis revealed a clean, predictive protein signature:
High baseline ratios of Bcl-2 relative to Mcl-1 protein abundance predicted Venetoclax response with 91.2% sensitivity.
The resulting 5-protein signature outperformed mutational status (AUC = 0.58) in predicting ex-vivo trial outcomes.
Models showing high baseline Bcl-2 abundance relative to Mcl-1 and Bcl-xL were highly sensitive. Conversely, models expressing high levels of the mitochondrial transporter protein VDAC1 and the kinase LYN showed immediate resistance to Bcl-2 inhibition.
By combining these 5 markers—Bcl-2, Mcl-1, Bcl-xL, VDAC1, and LYN—we built a machine learning classifier model. In validation runs on independent cohorts, this classifier model achieved an area under the receiver operating characteristic curve (AUC) of 0.94, validating its potential as a companion diagnostic signature.
Clinical Translation & Regulatory Development
Following our strategic acquisition of NGeneBioAI in 2025, we transitioned this 5-protein signature into a targeted, multiplexed reaction monitoring (PRM) mass spectrometry assay.
This PRM assay was validated inside our certified CLIA/CAP laboratory facilities at 6335 Nancy Ridge Dr. It is currently being deployed to screen patient cohorts for a Phase Ib/II clinical trial of Venetoclax in combination with chemotherapy, providing biopharma partners with clinical-grade patient selection capabilities.