Abstract
Background
Neuroblastoma is the most common extracranial solid tumor among pediatric patients, with MYCN amplification marking a significant increase in mortality risk. The utilization of neuroblastoma cell lines serves as the first-line model for studying the phenotypic outcomes of novel therapeutic interventions. Selecting cell lines that closely mimic the molecular characteristics of neuroblastomas derived from patients is crucial for enhancing the efficacy of drug discovery. We hypothesized that the similarity of transcriptomic profiles between patient-derived neuroblastoma tumors and neuroblastoma cell lines would enable the identification of cell model representatives (CMRs) that accurately reflect the molecular features of these patient tissues, particularly focusing on MYCN amplification status.
Methods
To test this hypothesis, we conducted a rank-based correlation analysis comparing the transcriptomic profiles of 643 patient-derived tumor tissues (493 from the SEQC/MAQC-III cohort and 150 from the Target cohort) with those of 39 neuroblastoma cell lines (sourced from GSE89413). Pharmacogenomic connectivity analysis was performed using the differentially expressed gene signatures (MYCN-amp vs. MYCN-NA) derived from the CMRs and patient-derived tumor tissues to predict candidate drugs that reverse MYCN-driven oncogenic signals. Finally, a Drug-gene network was performed to narrow down the list of drug candidates that potentially target MYCN interactome.
Results
Our analysis identified COG-N-557, SMS-KAN, and NB-SD as the top three CMRs for MYCN-amplified neuroblastoma, and COG-N-549, FELIX, and SK-N-SH emerged as the top CMRs for MYCN-non-amplified neuroblastoma. Further, pharmacogenomic connectivity map and drug-gene network analyses suggested different drug classes targeting the first- and second-level interactors of MYCN, i.e., kinase inhibitors, topoisomerase inhibitors, and chemotherapeutic agents, might serve as potential therapeutic candidates for MYCN-amplified neuroblastoma. In contrast, all predicted candidates from the poor representative cell lines showed no connection with MYCN interactome.
Conclusion
This bioinformatic study proposes a set of CMRs of patient-derived neuroblastoma tissues, offering a valuable resource for future research aimed at the discovery and validation of novel treatments for pediatric neuroblastoma. Our findings also underscore the potential of targeting the undruggable MYCN through its interactome.