PAX5-altered acute lymphoblastic leukemia (PAX5-alt ALL) is a recently recognized molecular subtype of B-ALL characterized by a distinct transcriptional signature and frequent PAX5 fusions (PAX5-r). While PAX5-alt ALL has been associated with intermediate outcomes, clinical data specifically investigating pediatric PAX5-r ALL remain limited. We collected and analyzed 159 pediatric cases of PAX5-r ALL treated between 2001-2024 within AIEOP-BFM ALL studies across Italy, Germany and Austria, revealing high-risk clinical features. Comparative analyses of patients consecutively enrolled in the AIEOP-BFM ALL 2017 study (PAX5-r, n=96 vs. non-PAX5-r, n=1948) confirmed higher rates of hyperleukocytosis at diagnosis (22.9% vs. 8%, p<0.001) and enrichment of IKZF1plus profile (14.7% vs. 7.8%, p=0.0015) in PAX5-r patients. No relevant differences in terms of minimal/measurable residual disease (MRD)-based treatment response and risk-group stratification were observed. PAX5-r patients had a 4-year EFS and OS of 72.6±5.7% and 95.4±2.3%, respectively. Four-year EFS were 100%, 63.4±9.7% and 63.3±10% for standard-risk, medium-risk (MR) and high-risk (HR) groups, respectively, indicating that the poor prognostic impact of PAX5-r applies only when end-of-induction MRD is positive (MR/HR). Whole transcriptome sequencing revealed high FLT3 median expression in PAX5-r ALL and high-throughput drug screening of patient-derived xenografts (PDXs) showed marked sensitivity to several FLT3 inhibitors. Gilteritinib showed potent ex vivo cytotoxicity and synergism with dexamethasone in PAX5-r PDX models. Collectively, this study investigated clinical and biological features of pediatric PAX5-r ALL highlighting its MRD-dependent unfavorable prognosis and identifying FLT3 overexpression as a novel, potential therapeutic target.
Abstract Children with relapsed or refractory acute lymphoblastic leukemia (ALL) require more effective and less toxic therapies. We established a prospective, multicenter Drug Response Profiling (DRP) registry ( NCT06550102 ) integrating functional testing into precision-guided treatment. DRP was performed for 340 patients from 17 European countries with a turn-around time of two-weeks. Image-based drug screening with over 135’000 unique perturbations revealed a heterogeneous landscape of ex vivo responses to 88 drugs on average. Ranking drug responses across the patient cohort defined individual drug fingerprints, identifying “DRP twins” by similarity in sensitivity and resistance independent of genetic ALL subtypes. Of 239 high-risk patients with follow-up, DRP-informed interventions were reported for 63 patients (26%). Patients received combination therapies based on venetoclax, tyrosine kinase inhibitors, trametinib, bortezomib or selinexor, resulting in objective clinical responses in 43 cases (68%). Precision-guided treatments allowed bridging to cellular therapies in 42 patients among whom 28 (67%) were still alive with a median follow-up of 21 months after DRP (IQR: 14.7-26.6 months). Top responders to venetoclax, ranked within the first tertile of the cohort, had superior 1-year event-survival compared to venetoclax non-responders (0.57 [95% CI, 0.39-0.85] vs. 0.25 [95% CI, 0.11-0.58]). Collectively, these findings demonstrate the feasibility and clinical relevance of functional profiling within an international network. This scalable framework enables individualized therapy selection for enrolment in adaptive precision trials for high-risk pediatric ALL.
Background: PAX5-altered acute lymphoblastic leukemia (PAX5alt ALL) is a recently defined molecular subtype of BCP-ALL, which displays a specific gene expression and is frequently associated with PAX5 aberrations, including rearrangements (PAX5r). Poor/intermediate outcomes were reported for PAX5alt ALL. However, evidence on clinical features and prognosis of pediatric PAX5r ALL remains limited. Since FLT3 is directly repressed by wild-type PAX5, we hypothesized a potential role for FLT3 in sustaining blast survival in PAX5r ALL and investigated the efficacy of FLT3 targeting in this ALL subtype. Methods: Clinical data of all newly diagnosed PAX5r ALL patients (pts) (<18 years) enrolled in the consecutive AIEOP-BFM ALL trials (ALL 2000, NCT0061345; ALL 2009, NCT01117441; ALL 2017, NCT03643276) or observational studies in Italy (AIEOP), Germany and Austria between 2001-2024, were retrospectively collected. PAX5 gene fusions were identified either by cytogenetics, FISH, targeted-RNA-seq or whole-RNA-seq (WTS). FLT3 expression was assessed by WTS in AIEOP pts and in PAX5r patient derived xenografts (PDXs). A high-throughput drug screening (HTS) platform was used to evaluate FLT3 inhibitors (FLT3i) in PAX5r ALL PDXs. Apoptosis experiments using AnnexinV/7AAD staining, were conducted on PAX5r cell line (NALL-1) and on PAX5r PDX blasts in co-culture with human bone marrow stromal cells, to test the ex vivo efficacy of gilteritinib (GILT). Results: Overall, 165 pts with a confirmed PAX5r ALL were included in the study. Sixty-five different fusion gene partners were identified, with JAK2 (n=22), NOL4L (N=18), AUTS2 (n=13) and ETV6 (n=12) as the most frequent ones. Median age was 3.4 yrs (0.4-17.8) and 66.1% of the pts were males. Low incidence of CNS involvement (CNS3, 3.8%) and a high rate of hyperleukocytosis (WBC>100x109/L) (22%) was found. According to NCI criteria, 53.3% had high-risk (HR) status. IKZF1plus profile was found in 15.1% of the pts. A tendency to slow treatment response was observed: 13% of pts had prednisone poor response (PPR), 74.7% were PCR-MRD+ at the end of induction (EOI), with 28.6% showing MRD≥5x10-4, and 31.5% were still PCR-MRD+ at the end of consolidation. Per protocol final risk stratification: 25.8%, 42.9% and 31.3% of the pts were standard-risk (SR), medium-risk (MR) and HR, respectively. Further, we restricted the analyses solely to the consecutive PAX5r pts (n=99) enrolled in the AIEOP-BFM ALL 2017 study and compared their clinical features with all non-PAX5r pts (n=1945) enrolled in the same protocol. ETV6::RUNX1 pts were excluded from the analysis. Lower median age at diagnosis (3.4 vs 4.7 yrs), higher hyperleukocytosis rate (23.2% vs 8%, p<0.001), increasedrate of IKZF1plus profile (14.4% vs 7.8%, p=0.02) and NCI HR status (57.6% vs 39.1%, p<0.001) were observed in the PAX5r group. No relevant differences in terms of treatment response and risk-stratification distribution were found between the two groups, except for a higher frequency of PPR (14.4% vs 5.7%, p=0.001) in the PAX5r ALL. With a median follow-up of 3.6 yrs, 5-yr EFS and OS for the AIEOP-BFM ALL 2009/2017 PAX5r pts (n=116) were 66.8±5.0% and 95.3±2.1%, respectively; 5-yr EFS were 100%, 59.8±9.3% and 43.8±12.7% for standard-risk (SR), medium-risk (MR) and high-risk (HR) groups, respectively, indicating that the poor prognostic impact of PAX5r applies only when EOI MRD is positive (MR/HR). FLT3 expression was profiled in 465 consecutive BCP-ALL AIEOP pts. PAX5r pts (n=16) showed 20% higher median FLT3 expression than the whole cohort (5.3 vs 4.4 relative expression units, RU p=0.04, unpaired t-test), similar to ZNF384r (5.8 RU), KMT2Ar (5.6 RU) and hyperdiploid ALL (5.4 RU). High FLT3 expression was preserved also in PAX5r PDX samples. HTS revealed sensitivity to 8 FLT3i for PAX5r ALL, showing a Drug Sensitivity Score (DSS) ranging from 14.7 to 47.9, with GILT (median DSS=26.7) emerging as a promising agent in terms of efficacy and hematological toxicity in vitro profile. Apoptosis experiments with GILT in NALL-1 and PDX blasts (n=4) showed single-agent cytotoxicity at nanomolar doses (IC50, range 250-550nM) and a synergistic effect in combination with dexamethasone. Conclusions: This multinational retrospective study indicates that PAX5r ALL is frequently associated with HR clinical features and unfavorable outcomes in non-SR pts. FLT3 is highly expressed in PAX5r ALL and ex vivo treatment with GILT showed promising efficacy.
PURPOSE:Modern ALL therapy aims to reduce toxicity, while maintaining and improving the current high cure rates. Acute and late sequelae of anthracyclines are of major concern. The AIEOP-BFM ALL 2009 trial aimed to clarify the need for anthracyclines in low-risk patients. PATIENTS AND METHODS:After 2 weeks of induction therapy, which included two daunorubicin (DNR) doses once weekly (30 mg/m2 each) as part of a 4-drug therapy, patients age 1-17 years with newly diagnosed non-high-risk B-ALL either positive for ETV6::RUNX1 or with rapid treatment response, as assessed by induction day-15 evaluation, were randomly assigned to receive either two additional doses of DNR (control arm [CA]) or no further DNR during induction (experimental arm [EA]). Patients treated as randomly assigned were included in the primary analysis on noninferiority in event-free survival (EFS). Adverse reactions of special interest (ARSI) were analyzed in the as-treated population. RESULTS:Of 6,136 patients enrolled in AIEOP-BFM ALL 2009, 2,514 patients (41.0%) were eligible for this random assignment, with 82.7% actually randomly assigned (EA: n = 1,040 and CA: n = 1,039). The 5-year EFS was 92.5% (SE 0.8%) in CA and 92.2% (SE 0.9%) in EA. Accordingly, cumulative incidence of relapse was 5.8% (SE 0.7%) and 5.7% (SE 0.7%), and overall survival was 97.6% (SE 0.5%) and 97.4% (SE 0.5%) in CA and EA, respectively. Life-threatening and fatal ARSI were similar in the two arms, but there was a three times lower incidence of invasive fungal infections in the EA (0.5% v 1.5%, P = .02). CONCLUSION:A reduced DNR dose during induction did not compromise the outcome of patients with favorable prognostic factors but did diminish infectious toxicity indicated by the lower rate of invasive fungal infections.
To improve the outcome of pediatric T-cell acute lymphoblastic leukemia (T-ALL) patients, the AIEOP-BFM ALL 2009 trial modified T-ALL stratification and treatment based on AIEOP-BFM ALL 2000 and other pediatric ALL groups' results. This report aims to describe the outcome of T-ALL patients in trial AIEOP-BFM ALL 2009 and evaluate prognostic features defined within the end of induction (EOI) therapy, for future protocols stratification and interventions. From 06/2010 to 02/2017, 872 T-ALL patients, aged 1-17, were enrolled. High risk (HR) criteria were prednisone poor response (PPR), Day 15 flow cytometry minimal residual disease (MRD) ≥ 10%, no complete remission at EOI, or polymerase chain reaction (PCR)-MRD ≥ 5 × 10-4 at end of consolidation (EOC). Three Cox regression models on event-free survival (EFS) evaluated prognostic factors. Overall, 5-year EFS and survival were 79.9% ± 1.4% and 84.9% ± 1.2% with cumulative incidence of relapse (CIR) and death of 13.0% ± 1.2% and 5.9% ± 0.8%. Five-year EFS and CIR were 86.8% ± 1.6% and 8.7% ± 1.3% in non-HR patients (n = 470); 71.9% ± 2.3% and 18.0% ± 1.9% in HR patients (n = 402). High PCR-MRD levels at EOI and EOC were prognostic in all models, with EOC-MRD ≥ 5 × 10-3 related to a hazard ratio of 6.22 (P < 0.001). When a model considered factors identified at EOI only, central nervous system (CNS)3 (hazard ratio = 2.3, P < 0.001), PPR (hazard ratio = 1.74, P = 0.02), and high EOI-MRD (hazard ratio 4.71 for ≥5 × 10-2 vs. negative, P < 0.001) significantly impacted EFS. Results of T-ALL patients in AIEOP-BFM ALL 2009 were favorable. While EOC-MRD remained the strongest prognostic predictor, PPR, CNS3 disease, and EOI-MRD showed relevant prognostic value, with CNS3 and EOI-MRD ≥ 5 × 10-2 being candidate criteria for early stratification and intervention modifications.
Adaptation to changes in amino acid availability is crucial for cellular homeostasis, which requires an intricate orchestration of involved pathways. Some cancer cells can maintain cellular fitness upon amino acid shortage, which has a poorly understood mechanistic basis. Leveraging a genome-wide CRISPR-Cas9 screen, we find that superoxide dismutase 2 (SOD2) has a previously unrecognized dismutase-independent function. We demonstrate that SOD2 regulates global proteasomal protein degradation and promotes cell survival under conditions of metabolic stress in malignant cells through the E3 ubiquitin ligases UBR1 and UBR2. Consequently, inhibition of SOD2-mediated protein degradation highly sensitizes different cancer entities, including patient-derived xenografts, to amino acid depletion, highlighting the pathophysiological relevance of our findings. Our study reveals that SOD2 is a regulator of proteasomal protein breakdown upon starvation, which serves as an independent catabolic source of amino acids, a mechanism co-opted by cancer cells to maintain cellular fitness.
Variations in the TP53 and KRAS genes indicate a particularly adverse prognosis in relapsed pediatric T-ALL. We hypothesized that these variations might be subclonally present at disease onset and contribute to relapse risk. To test this, we examined two cohorts of children diagnosed with T-ALL: one with 81 patients who relapsed and 79 matched non-relapsing controls, and another with 226 consecutive patients, 30 of whom relapsed. In cohort 1, targeted sequencing revealed TP53 clonal and subclonal variants in 6 of 81 relapsing patients but none in the non-relapsing group (p=0.014). KRAS alterations were found in 9 of 81 relapsing patients compared to 2 of 79 non-relapsing patients (p=0.032). Survival analysis showed that none of the relapsed patients with TP53 and/or KRAS alterations survived, whereas 19 of 67 relapsed patients without such variants did with a minimum follow-up time of 3 years (p=0.023). In cohort 2, none of the relapsing patients but 10 of 196 non-relapsing patients carried TP53 or KRAS variants, indicating that mutation status alone does not predict poor prognosis. All ten non-relapsing patients with mutations had a favorable early treatment response. Among the total cohort of 386 patients, 188 showed poor treatment response of whom 69 relapsed. Nine of these poor responders harbored TP53 or KRAS variants. In conclusion, subclonal TP53 and KRAS alterations identified at the time of initial diagnosis, along with a poor treatment response, characterize a subset of children with T-ALL who face a dismal prognosis and may benefit from alternative treatment approaches.
Introduction Firstly described in 1967, hyperdiploidy (HD) is the most frequent genetic abnormality in B-cell- precursor acute lymphoblastic leukemia (BCP-ALL) in children, comprising about 25% of all cases. A not yet exactly defined proportion have predisposing pathogenic germline variants in DNA repair pathway genes, chromatin remodeling factors, transcription factors regulating B-cell development (particularly ETV6) or receptor tyrosine kinases pathway genes like RAS/RAF. Among the latter, mutations have mostly been described in PTPN11 and SOS1, but not yet in other components of this central regulatory hub of cellular communication. Germline loss-of-function (LOF) LZTR1 mutations are typically linked to hereditary nerve sheath tumors. However, it remains largely unknown, if other tumor entities are associated with LZTR1 LOF germline mutations, potentially broadening the spectrum of malignancies associated with RASopathies. Aim Our aim is to understand the frequency and the impact of LZTR1 germline variants in HD BCP-ALL of childhood. Material and Methods We analyzed WES data of 283 children with BCP-ALL for the presence of pathogenic variants in the LZTR1 cancer predisposition gene. For a part of the detected variants, we performed further functional analyses to investigate the effect of the alteration on protein function. We applied a Drosophila model that is particularly suited for functional evaluation of Ras pathway activity. Results We identified LZTR1 germline variants in 10/283 (3.5%) patients (Figure 1). Interestingly, in 6 out of 283 (2.1%) children the LZTR1 variants were classified as pathogenic/likely pathogenic (P/LP). The majority of patients (5/6) harboring a P/LP germline LZTR1 variant presented with a HD BCP-ALL. Only two of our patients showed concomitant phenotypic features indicative of an underlying syndromic condition. Patient P1 presented with a HD BCP-ALL at the age of 9 years and mild psychomotor delay. Sequencing revealed a likely pathogenic variant (p.Arg283Trp) that was not yet reported in association with RASopathy syndromes. The second child P8, also diagnosed with a HD BCP-ALL, carried a well-known autosomal dominant mutation described in Noonan syndrome (NS) variant (p.Gly248Arg). The child was characterized by facial dysmorphism, as well as mild developmental delay, although a diagnosis of NS was not established prior to development of the BCP-ALL. The other three patients had no clinical peculiarities other than HD BCP-ALL. We functionally characterized patient derived LZTR1 variants in dLztr1-depleted ISC. Like wild-type hLZTR1 wt, hLZTR1 p.Arg283Trp and hLZTR1 p.Lys761Arg restored control-like ISC lineage production, suggesting that hLZTR1 p.Arg283Trp and hLZTR1 p.Lys761Arg still regulate Ras ubiquitination. In contrast, hLZTR1 p.Tyr535Ter results in even 1.8-fold higher ISC lineage production than dLztr1-RNAi. Beyond ISC production as a readout, we employed a second experimental paradigm directly addressing Ras signalling activity with a translocating modified ERK sensor. In line with Ras activity control of LZTR1, hLZTR1 wt significantly reduced ERK activity, while the putatively LOF variant hLZTR1 p.Tyr535Ter variant significantly increased Ras pathway activation by 1.4-fold over controls. We also noticed that hLZTR1 p.Tyr535Ter induced extensive membrane blebbing and nuclear fragmentation of GFP positive cells likely indicating programmed cell death. Quantification of apoptotic cells revealed an 11.1-fold increase compared to controls, which was not observed for hLZTR1 p.Arg283Trp and hLZTR1 p.Lys761Arg. Most notably, forced expression of hLZTR1 p.Arg283Trp andhLZTR1 p.Tyr535Ter increased mitotic recombination by 4-fold and 3.4-fold, respectively, which was not observed for the hLZTR1 p. Lys761Arg variant. Conclusion In our cohort, approximately 2% of all children with BCP-ALL harboured a P/LP LZTR1 germline variation, not necessarily linked with clinical appearance of Noonan-syndrome-like features, but to the development of a high hyperdiploid karyotype. By applying a Drosophila model, we demonstrated that patient-derived LZTR1 germline variants affect RAS pathway activation, ERK accumulation, cell proliferation, DNA recombination and apoptosis. Figure 1: Depiction of the ten LZTR1 variants detected in an unselected pediatric cohort of 283 BCP-ALL patients
BackgroundThe increasing volume and intricacy of sequencing data, along with other clinical and diagnostic data, like drug responses and measurable residual disease, creates challenges for efficient clinical comprehension and interpretation. Using paediatric B-cell precursor acute lymphoblastic leukaemia (BCP-ALL) as a use case, we present an artificial intelligence (AI)-assisted clinical framework clinALL that integrates genomic and clinical data into a user-friendly interface to support routine diagnostics and reveal translational insights for hematologic neoplasia.MethodsWe performed targeted RNA sequencing in 1365 cases with haematological neoplasms, primarily paediatric B-cell precursor acute lymphoblastic leukaemia (BCP-ALL) from the AIEOP-BFM ALL study. We carried out fluorescence in situ hybridization (FISH), karyotyping and arrayCGH as part of the routine diagnostics. The analysis results of these assays as well as additional clinical information were integrated into an interactive web interface using Bokeh, where the main graph is based on Uniform Manifold Approximation and Projection (UMAP) analysis of the gene expression data. At the backend of the clinALL, we built both shallow machine learning models and a deep neural network using Scikit-learn and PyTorch respectively.FindingsBy applying clinALL, 78% of undetermined patients under the current diagnostic protocol were stratified, and ambiguous cases were investigated. Translational insights were discovered, including IKZF1plus status dependent subpopulations of BCR::ABL1 positive patients, and a subpopulation within ETV6::RUNX1 positive patients that has a high relapse frequency. Our best machine learning models, LDA and PASNET-like neural network models, achieve F1 scores above 97% in predicting patients' subgroups.InterpretationAn AI-assisted clinical framework that integrates both genomic and clinical data can take full advantage of the available data, improve point-of-care decision-making and reveal clinically relevant insights promptly. Such a lightweight and easily transferable framework works for both whole transcriptome data as well as the cost-effective targeted RNA-seq, enabling efficient and equitable delivery of personalized medicine in small clinics in developing countries.FundingGerman Ministry of Education and Research (BMBF), German Research Foundation (DFG) and Foundation for Polish Science.
The t(1;19) translocation, which codes for the oncogenic fusion protein E2A (TCF3)-PBX1, is involved in acute lymphoblastic leukemia (ALL) and associated with a pre-B cell receptor (preBCR+) phenotype. Relapse in E2A-PBX1+ ALL patients frequently occurs in the central nervous system (CNS). Therefore, there is a medical need for the identification of CNS active regimens for the treatment of E2A-PBX1+/preBCR+ ALL. Using unbiased shRNA library screening approaches, we identified Bruton's tyrosine kinase (BTK) as a key gene involved in both proliferation and dasatinib sensitivity of E2A-PBX1+/preBCR+ ALL. Depletion of BTK by shRNAs resulted in decreased proliferation of dasatinib-treated E2A-PBX1+/preBCR+ cells compared with control-transduced cells. Moreover, combination of dasatinib with BTK inhibitors (BTKi) (ibrutinib, acalabrutinib or zanubrutinib) significantly decreased E2A-PBX1+/preBCR+ human and murine cell proliferation, reduced PLCG2 and BTK phosphorylation and total protein levels and increased disease-free survival of mice in secondary transplantation assays, reducing particularly CNS-leukemic infiltration. Hence, dasatinib with ibrutinib reduced pPLCG2 and pBTK in primary ALL patient samples, including E2A-PBX1+ ALLs. In summary, genetic depletion and pharmacological inhibition of BTK increase dasatinib effects in human and mouse E2A-PBX1+/preBCR+ ALL in most of performed assays, and the combination of dasatinib and BTKi is very effective in reducing CNS-infiltration of E2A-PBX1+/preBCR+ ALL cells in vivo.
Next Generation sequencing (NGS) has greatly advanced precision oncology. The growing amount and complexity of NGS data, along with other clinical data such as drug responses and measurable residual disease (MRD), present a big challenge for data integration and interpretation. Machine learning, especially deep neural networks, offer a promising approach for efficiently analyzing intricate relationships within large datasets, with the potential of improving clinical outcomes. We performed targeted RNA sequencing in routine diagnostics and sequenced 1849 cases with hematological neoplasms, primarily pediatric B-cell precursor acute lymphoblastic leukemia (BCP-ALL). We used featureCounts in megSAP pipeline and Uniform Manifold Approximation and Projection (UMAP) to analyze the gene expression data, and Bokeh to create the web interface for data integration and visualization. Scikit-learn and PyTorch were used to develop shallow machine learning models and deep learning neural networks (NN), respectively. Our platform integrates various genetic and clinical data based on UMAP analysis of gene expression patterns. It improves the point-of-care decision making and facilitates the discovery of new patterns such as subpopulations with different genetic and clinical features. The platform is supported by machine learning algorithms for cancer subtyping. Six basic machine learning algorithms were coupled with feature selection methods and the best F1 score achieves 98%. We also built a biologically informed deep NN that can accurately predict BCP-ALL subtypes (F1=97%) with a good interpretability, which helps to identify crucial genetic aberrations associated with disease subtypes. Our machine learning based platform can not only provide support for clinical decision-making but also bring novel translational insights for hematological malignancies.
Background Characterization of clinical phenotypes in context with tumor and host genomic information can aid in the development of more effective and less toxic risk-adapted and targeted treatment strategies. To analyze the impact of therapy-related hyperbilirubinemia on treatment outcome and to identify contributing genetic risk factors of this well-recognized adverse effect we evaluated serum bilirubin levels in 1547 pediatric patients with acute lymphoblastic leukemia (ALL) and conducted a genome-wide association study (GWAS). Patients and methods Patients were treated in multicenter trial AIEOP-BFM ALL 2000 for pediatric ALL. Bilirubin toxicity was graded 0 to 4 according to the Common Toxicity Criteria (CTC) of the National Cancer Institute. In the GWAS discovery cohort, including 650 of the 1547 individuals, genotype frequencies of 745,895 single nucleotide variants were compared between 435 patients with hyperbilirubinemia (CTC grades 1-4) during induction/consolidation treatment and 215 patients without it (grade 0). Replication analyses included 224 patients from the same trial. Results Compared to patients with no (grade 0) or moderate hyperbilirubinemia (grades 1-2) during induction/consolidation, patients with grades 3-4 had a poorer 5-year event free survival (76.6 ± 3% versus 87.7 ± 1% for grades 1-2, P = 0.003; 85.2 ± 2% for grade 0, P < 0.001) and a higher cumulative incidence of relapse (15.6 ± 3% versus 9.0 ± 1% for grades 1-2, P = 0.08; 11.1 ± 1% for grade 0, P = 0.007). GWAS identified a strong association of the rs6744284 variant T allele in the UGT1A gene cluster with risk of hyperbilirubinemia (allelic odds ratio (OR) = 2.1, P = 7 × 10 − 8 ). TT-homozygotes had a 6.5-fold increased risk of hyperbilirubinemia (grades 1-4; 95% confidence interval (CI) = 2.9-14.6, P = 7 × 10 − 6 ) and a 16.4-fold higher risk of grade 3-4 hyperbilirubinemia (95% CI 6.1-43.8, P = 2 × 10 − 8 ). Replication analyses confirmed these associations with joint analysis yielding genome-wide significance (allelic OR = 2.1, P = 6 × 10 − 11 ; 95% CI 1.7-2.7). Moreover, rs6744284 genotypes were strongly linked to the Gilbert’s syndrome-associated UGT1A1 *28/*37 allele (r 2 = 0.70), providing functional support for study findings. Of clinical importance, the rs6744284 TT genotype counterbalanced the adverse prognostic impact of high hyperbilirubinemia on therapy outcome. Conclusions Chemotherapy-related hyperbilirubinemia is a prognostic factor for treatment outcome in pediatric ALL and genetic variation in UGT1A aids in predicting the clinical impact of hyperbilirubinemia. Trial registration http://www.clinicaltrials.gov ; #NCT00430118.
Tolerance of amino acid starvation is crucial for cancer cells to promote cell fitness upon nutrient deprivation. In response to starvation of asparagine, leucine, and valine, GSK3α - a key regulator of the amino acid starvation response - undergoes supramolecular assembly. This response triggers efficient protein degradation as an alternative nutrient supply by concentrating GSK3α with the ubiquitin-proteasome system into cytoplasmic GSK3α bodies. However, the mechanistic underpinnings of how cancer cells can regulate GSK3α assembly have yet to be understood. We here identified the protein PRUNE, the only known mammalian exopolyphosphatase (PPX), as a novel regulator in this adaptive cellular response. PRUNE selectively binds to GSK3α in the absence of asparagine, leucine and valine. Notably, loss of PRUNE completely impairs formation of GSK3α bodies, and highly sensitizes resistant cancer cells to amino acid starvation in dependence of its PPX activity. Thus, for the first time we here identify polyphosphate levels as a fundamental regulator of the response to amino acid shortage in leukemia cells, opening an avenue for therapeutic intervention.