F. E. G. Beaudry, D. Yendi, D. Arshinoff, N. Light, S. Perrotti, E. Winter, L. R. Cristant, A. Xu, J. Wilson, A. Dodd, R. Bucur, E. X. Chen, E. Elimova, R. Wong, A. Mesci, A. Hosni, A. Ghanekar, R. Jang, C. G. Shwaartz, T. Reichman, C.-A. Moulton, E. Sanz Garcia, G. M. O'Kane, E. S. Tsang, X. Wang, I. McGilvray, S. Gallinger, T. J. Pugh, G. Sapisochin, A. Vogel, J. J. Knox, F. Notta, R. C. Grant
Introduction Biliary tract cancer (BTC) comprises a family of rare malignancies subclassified by anatomy and pathology. However, this scheme may obscure shared biology and limit patient stratification. Objectives We tested whether BTC heterogeneity can be explained by coherent latent axes and evaluated the potential to unify diverse clinical and genomic factors under a tractable biological framework Methods We performed whole-genome and transcriptome sequencing of 180 tumors enriched for tumor cells by laser capture microdissection to identify shared programs in BTC. Results Network integration across transcriptomic classes identified two consensus cancer subtypes (CCS). CCS segregated with anatomical location of primary tumor and gene expression marker analyses suggest subtypes reflect tumor cell of origin differences. CCS displayed strikingly divergent molecular landscapes, explaining more variance than anatomical location of primary tumor. CCS-B tumors were mutationally loaded with clock-like and APOBEC signatures and extrachromosomal DNA, whereas CCS-A tumors were characterized by chromosome-arm deletions. Conclusion Our approach showed that harnessing the genomic and transcriptomic diversity of BTC uncovers novel biology and improves stratification.