Figure 2

Telomere analysis using whole-genome sequencing data of 235 pediatric cancers. (a) Bayesian information criterion (BIC) guided clustering, which divided the ΔT values in this cohort into two clusters with equal variance. The boundary of these clusters is marked in dark blue. Using 0.01 as the threshold for significance, we defined the lower and upper boundary of ΔT as 'gain' or 'loss' of telomeric DNA. Samples that fall within these boundaries are deemed to have 'no change' in telomere status. (b) The number of structural variations in tumors with 'gain', 'loss' or 'no change' of telomere status. Tumors with ΔT gains have significantly higher number of structural variations compared with the other two groups (Mann-Whitney P = 1.07e-10; brain tumors P = 0.013, solid tumors P = 0.0002, hematopoietic malignancies P = NA (Not Applicable - no telomeric content gains detected); M, median). (c) The number of non-silent mutations in tumors with 'gain', 'loss' or 'no change' of telomere status. Tumors with ΔT gains have significantly higher number of sequence mutations compared with the other two groups (Mann-Whitney P = 3.723e-07; brain tumors P = 0.061; solid tumors P = 0.013, hematopoietic malignancies P = NA; M, median). (d) ΔT values from 235 pediatric cancers. The dotted lines correspond to the lower and upper boundary of ΔT as 'gain' or 'loss'. CBF, core-binding factor ALL; HYPO, hypodiploid ALL; INF, infant ALL; TALL, ETP-ALL; EPD, ependymoma; HGG, high-grade glioma; LGG, low-grade glioma; MB, medulloblastoma, ACT, adrenocortical carcinoma; NBL, neuroblastoma; OS, osteosarcoma; RB, retinoblastoma; RHB, rhabdomyosarcoma.