Integrated Dependency and Primary-Patient Profiling Prioritize Nucleotide-Synthesis Vulnerabilities in High-Risk B-ALL
Relapsed B-cell acute lymphoblastic leukemia (B-ALL) remains difficult to treat, yet the space of therapeutic hypotheses is crowded with lineage markers, recurrent fusions, and poorly prioritized dependencies. We asked a narrower question: which noncanonical genes and pathways remain compelling therapeutic liabilities after integrating cell-line dependency data with primary-patient subtype and relapse information? We combined DepMap 24Q2 CRISPR knockout and expression profiles across 32 B-lymphoblastic leukemia/lymphoma models with two public patient cohorts: the DFCI 16-001 pediatric ALL RNA-seq cohort with molecular subtype calls and a paired diagnosis/relapse pediatric B-precursor ALL cohort. Rather than identifying a single dominant target, the analysis converged on a pathway-level program centered on de novo pyrimidine synthesis and connected one-carbon/purine support. UMPS, CAD, DHODH, ATIC, MTHFD1, and FPGS remained the most coherent axis after primary-cohort integration. Pathway-level scoring placed TCF3-PBX1 B-ALL near the top of the combined nucleotide program and highlighted KMT2A-rearranged disease as a strong pyrimidine-focused comparison subtype. Exploratory pharmacology using the public PRISM repurposing screen showed broad activity of the antifolate trimetrexate in annotated B-ALL lines, whereas the public release lacked a clean DHODH inhibitor. These results prioritize nucleotide-synthesis stress, rather than a single receptor or transcription factor, as the most defensible near-term therapeutic hypothesis for high-risk B-ALL and nominate antifolate-versus-pyrimidine-blockade experiments in TCF3-PBX1 and KMT2A-rearranged models as the next concrete step.
Reviews
This study’s main strength is a sensible, integrative prioritization strategy: starting from DepMap CRISPR dependencies in B-ALL lines and explicitly intersecting with primary pediatric ALL subtype/relapse expression patterns, it converges on a biologically coherent, pathway-level liability (de novo pyrimidine synthesis and linked one‑carbon/purine support). The convergence across multiple genes (e.g., CAD/DHODH/UMPS plus folate-cycle nodes) is more convincing than any single-gene hit, and the subtype framing (TCF3-PBX1 highlighted; KMT2A-r as a comparator) is plausible and potentially useful for hypothesis narrowing. Using PRISM to seek an “available handle” further supports pragmatic next steps, and the emphasis on pathway programs over isolated targets is an appropriately cautious interpretation given heterogeneity. The main uncertainty is that the evidence remains largely correlative/inferential and is limited by the small number of B-lymphoblastic models in DepMap and an even smaller annotated PRISM subset; this raises concerns about subtype representativeness, confounding by proliferation/essentiality of core metabolic genes, and whether the signal is truly selective (therapeutic window) versus reflecting generic nucleotide demand in fast-growing cells. The pharmacology layer is also weakly anchored: trimetrexate activity is consistent with antifolate sensitivity but does not specifically validate pyrimidine synthesis (and PRISM lacks a clean DHODH inhibitor), and no new functional experiments (genetic rescue, flux assays, synergy with standard agents, in vivo validation, or relapse-matched functional data) are presented in the excerpt. Overall, the conclusion that “nucleotide-synthesis stress is a defensible near-term hypothesis” is justified as a prioritization claim, but stronger claims about subtype-specific vulnerability or clinical tractability would require direct experimental validation and clearer controls for generic essential metabolism effects.