The mechanisms underlying T-ALL relapse remain essentially unknown. Multilevel-omics in 38 matched pairs of initial and relapsed T-ALL revealed 18 (47%) type-1 (defined by being derived from the major ancestral clone) and 20 (53%) type-2 relapses (derived from a minor ancestral clone). In both types of relapse, we observed known and novel drivers of multidrug resistance including MDR1 and MVP, NT5C2 and JAK-STAT activators. Patients with type-1 relapses were specifically characterized by IL7R upregulation. In remarkable contrast, type-2 relapses demonstrated (1) enrichment of constitutional cancer predisposition gene mutations, (2) divergent genetic and epigenetic remodeling, and (3) enrichment of somatic hypermutator phenotypes, related to BLM, BUB1B/PMS2 and TP53 mutations. T-ALLs that later progressed to type-2 relapses exhibited a complex subclonal architecture, unexpectedly, already at the time of initial diagnosis. Deconvolution analysis of ATAC-Seq profiles showed that T-ALLs later developing into type-1 relapses resembled a predominant immature thymic T-cell population, whereas T-ALLs developing into type-2 relapses resembled a mixture of normal T-cell precursors. In sum, our analyses revealed fundamentally different mechanisms driving either type-1 or type-2 T-ALL relapse and indicate that differential capacities of disease evolution are already inherent to the molecular setup of the initial leukemia.

Long-term treatment with 6-thioguanine (6-TG) for pediatric acute lymphoblastic leukemia (ALL) is associated with high rates of hepatic sinusoidal obstruction syndrome (SOS). Nevertheless, current treatment continues to use short-term applications of 6-TG with only sparse information on toxicity. 6-TG is metabolized by thiopurine methyltransferase (TPMT) which underlies clinically relevant genetic polymorphism. We analyzed the association between hepatic SOS reported as a serious adverse event (SAE) and short-term 6-TG application in 3983 pediatric ALL patients treated on trial AIEOP-BFM ALL 2000 (derivation cohort) and defined the role of TPMT genotype in this relationship. We identified 17 patients (0.43%) with hepatic SOS, 13 of which with short-term exposure to 6-TG (P < 0.0001). Eight of the 13 patients were heterozygous for low-activity TPMT variants, resulting in a 22.4-fold (95% confidence interval 7.1-70.7; P ≤ 0.0001) increased risk of hepatic SOS for heterozygotes in comparison to TPMT wild-type patients. Results were supported by independent replication analysis. All patients with hepatic SOS after short-term 6-TG recovered and did not demonstrate residual symptoms. Thus, hepatic SOS is associated with short-term exposure to 6-TG during treatment of pediatric ALL and SOS risk is increased for patients with low-activity TPMT genotypes.