Alternative Lengthening of Telomeres is not synonymous with mutations in ATRX/DAXX

The PCAWG Consortium has recently released an unprecedented set of tumor whole genome sequence (WGS) data from 2,658 cancer patients across 38 different primary tumor sites 1 . WGS is able to document the quantity and distribution of telomeric repeats 2 . In one of the papers analyzing the PCAWG dataset, Sieverling et al. 3 confirmed previous data 4 indicating that tumors with truncating ATRX or DAXX alterations, referred to as ATRX/DAXX trunc , have an aberrant telomere variant repeat (TVR) distribution. By regarding these mutations, vs. TERT modifications (TERT mod ; i.e. promoter mutations +/− amplifications +/− structural variations), as indicators of Alternative Lengthening of Telomeres (ALT) vs. telomerase, they built a random forest classifier for ALT-probability, and then associated genomic characteristics with the putative Telomere Maintenance Mechanism (TMM) 3 . However, we show here that equating ATRX/DAXX trunc and TERT mod with ALT and telomerase, respectively, results in TMM predictions which correlate poorly with TMM assay data. ATRX/DAXX trunc mutations are heterogeneously distributed in ALT-positive (ALT+) tumors of different types, as are TERT mod in telomerase-positive tumors 4 . Although these mutations are strongly associated with TMM, most tumors do not harbor them, making them an inadequate basis for building a classifier in a large-scale pan-cancer study 4–7 . Here, we provide a new analysis of the PCAWG data, based on C-circle assay (CCA) 8 that is available for a subset of these tumors 4,9,10 . We show that the Sieverling et al. score overestimates the proportion of ALT associated with ATRX/DAXX trunc and misclassifies ALT tumors when these mutations are absent. We also show some TVR correlate with ATRX/DAXX trunc mutations, regardless of TMM. Finally, we propose a new classifier to identify ALT tumors in the PCAWG cohort.