Current intensive treatment of pediatric T-cell acute lymphoblastic leukemia (T-ALL) has substantial side-effects, highlighting a need for novel biomarkers to improve risk stratification. Canonical biomarkers such as genetics and immunophenotype are largely not used in pediatric T-ALL stratification. This study aimed to validate the prognostic relevance of DNA methylation CpG island methylator phenotype (CIMP) risk stratification in two pediatric T-ALL patient cohorts: the Nordic NOPHO ALL2008 T-ALL study cohort (n=192) and the Dutch DCOG ALL-10/ALL-11 validation cohorts (n=156). Both cohorts showed that combining CIMP classification at diagnosis with measurable residual disease (MRD) at treatment day 29 or 33 significantly improved outcome prediction. The poor prognosis subgroup, characterized by CIMP low/D29 or D33 MRD≥0.1%, showed a cumulative incidence of relapse (pCIR5yr) of 29.0% and 23%, and overall survival (pOS5yr) of 59.7% and 65.4%, in NOPHO and DCOG, respectively. Conversely, a good prognosis subgroup was also identified representing CIMP high/D29 or D33 MRD<0.1% with pCIR5yr of 0% and 3.4%, and pOS5yr of 98.2% and 94.8%, in NOPHO and DCOG, respectively. For NOPHO, MRD was also evaluated on D15, and the relapse prediction accuracy of CIMP/D29 MRD (0.79) and CIMP/D15 MRD (0.75) classification was comparable, indicating potential for earlier stratification. The evaluation of the biology behind the CIMP subgroups revealed associations with transcriptome profiles, genomic aberrations, and mitotic history, suggesting distinct routes for leukemia development. In conclusion, integrating MRD assessment with the novel CIMP biomarker has the potential to improve risk stratification in pediatric T-ALL and guide future therapeutic decisions.

Age and sex have historically been associated with differences in the survival of patients with acute lymphoblastic leukemia (ALL). The NOPHO ALL2008 trial included patients aged 1-45 years with BCR::ABL1-negative B-precursor and T-cell ALL, but neither sex nor age was integrated into risk group allocation. Among 1,771 trial patients stratified into protocol- appropriate risk groups, we estimated the impact of age and sex on survival (even after relapse) and toxicities prospectively registered at 3-month intervals. In multivariate Cox regression analysis adjusted by sex, age group, and risk group, age but not sex was an independent risk factor for reduced 5-year event-free survival (EFS): hazard ratio=1.57 (95% confidence interval: 1.15-2.14) for patients 10-17.9 years, and 2.70 (95% confidence interval: 2.03-3.58) for patients 18-45 years, compared to patients <10 years old at diagnosis. The overall 5-year pEFS was 0.83. For standard-risk patients (B-lineage, white cell count <100x109/L, no risk genetics, minimal residual disease day 29 <0.1%), an inferior 5-year EFS was observed among patients 18-45 years (pEFS 0.78, P<0.001) and 10-17.9 years (pEFS 0.86, P=0.002) compared to patients <10 years at diagnosis (pEFS 0.93). For the intermediate-risk and high-risk groups, EFS was worse for patients 18-45 years compared to patients <10 years: pEFS 0.69 versus 0.89 (P<0.001) and pEFS 0.55 versus 0.71 (P=0.005), respectively. Osteonecrosis and veno-occlusive disease were associated with female sex in the standard-risk group, and age ≥10 years was associated with osteonecrosis, thrombosis, and pancreatitis in sex- and treatment-group-adjusted analyses. In conclusion, this study indicates that risk-grouping and/or treatment-intensity criteria should differ across age groups and that age-adapted strategies to mitigate toxicities are needed.

INTRODUCTION:Infection remains the most common treatment-related toxicity of childhood ALL, emphasizing the need to identify patients at risk and to tailor treatment strategies accordingly. AIMS:The primary aim was to compare infectious toxicity during early treatment for childhood ALL following the ALLTogether and NOPHO ALL-2008 (ALL-2008) protocols, and second, to identify risk factors for infectious toxicity. METHODS:A national retrospective matched cohort study was conducted, including 345 patients aged 1-17 years diagnosed with ALL and treated in Sweden according to the ALLTogether or ALL-2008 protocols. Nonparametric tests were used to compare infectious outcomes between protocols, and regression modeling was used to identify risk factors of the infectious outcomes. RESULTS:Treatment following ALL-2008 showed higher infectious toxicity during induction, whereas treatment following ALLTogether showed increased infectious toxicity during consolidation 1. Overall, treatment according to ALL-2008 was associated with a higher incidence of infections. Anthracycline use and young age (1-9 years) were associated with both higher infectious incidence and more severe infectious complications. Dexamethasone was associated with both lower incidence and lower severity of infectious complications as compared to prednisone. CONCLUSIONS:A notable shift in the timing of infectious toxicity was observed between the two treatment protocols. Risk factors for infectious toxicity during early treatment include anthracycline use and young age. Dexamethasone as an induction steroid was associated with lower infectious burden, although its effect is difficult to isolate from the simultaneous anthracycline effect. The findings suggest that treatment composition plays a central role in determining both the extent and timing of infectious complications.

Normal karyotype acute myeloid leukaemia (NK-AML) in children is a heterogeneous subgroup with scarce data on characteristics and prognosis. We investigated NK-AML in a large paediatric AML cohort from four trials of the Nordic Society for Paediatric Haematology and Oncology-Dutch Belgian Hongkong (NOPHO-DBH) group. Among 1476 AML patients, we identified 316 NK-AML patients (21%). NK-AML was characterized by high frequencies of FLT3 internal tandem duplications (ITD, 33%), mutated NPM1 (28%), WT1 (25%) and CEBPA (21%). Five-year event-free survival (EFS) and overall survival (OS) in NK-AML were 52% (95% confidence interval [CI]: 46-58) and 70% (CI: 65-75) respectively. Restricted to NPM1wt cases only (n = 959), NK-AML was associated with unfavourable outcome (relative risk [RR] of EFS = 0.80, p = 0.014; RR of OS = 0.87, p = 0.022). NK-AML with mutated NPM1 had excellent EFS (79%, CI: 66-88) and OS (97%, CI: 88-99), which was not influenced by concomitant FLT3-ITD. In multivariable analysis, mutated NPM1 in NK-AML was associated with favourable EFS (hazard ratio [HR]: 0.24, CI: 0.13-0.43, p < 0.001) and OS (HR: 0.10, CI: 0.03-0.35, p < 0.001). FLT3-ITD was associated with inferior EFS (HR: 1.56, CI: 1.03-2.35, p = 0.035) and OS (HR: 1.91, CI: 1.11-3.31, p = 0.02). We conclude that prognosis in paediatric NK-AML is independently affected by NPM1 and FLT3-ITD status. Further molecular characterization of NK-AML is needed, especially for NPM1wt NK-AML.

To study the mechanisms of relapse in KMT2A-rearranged (KMT2A-r) acute lymphoblastic (ALL) and acute myeloid leukemia (AML), we performed whole-genome and exome sequencing of infants and children with relapsed ALL/AML (n = 36), and longitudinal deep-sequencing of 257 samples in 30 patients. Somatic alterations in drug-response genes, most commonly in TP53 and IKZF1 (64%), were highly enriched in early relapse ALL (79%, 9-36 months after diagnosis), but rare in very early relapse ALL (<9 months, 9%). A marked chemotherapy-exposure signature was detected for mutations in early relapse ALL but not in very early ALL or AML relapse, in line with different mechanisms of relapse. Longitudinal analyses could track residual leukemia cells, clonal drug responses, and the upcoming relapse. These results highlight that KMT2A-r ALL and AML evade therapy differently and provide insights into the mechanisms of relapse in this highly lethal form of pediatric acute leukemia.

In most study protocols for children with acute myeloid leukemia (AML), treatment response is assessed with flow cytometry (FCM) after one and/or two courses of induction treatment, impacting risk stratification. In AML with RUNX1::RUNX1T1, which is the most common subtype of pediatric AML and generally associated with good prognosis, assessment of treatment response is complicated by two factors: 1) Reverse transcription quantitative PCR (RT-qPCR-MRD) often shows high levels of RUNX1::RUNX1T1 fusion transcripts indicating residual disease. 2) FCM-MRD sometimes detects cells with immature markers such as CD34 or CD117 and atypical immunophenotype, leading to an interpretation as MRD. However, we have observed that such cells lack the diagnostic immunophenotype and instead display basophilic or mast cell markers. In this study, we aimed to improve interpretation of FCM-MRD by characterization of AML with RUNX1::RUNX1T1 with focus on basophils and mast cells. Our study included 45 of 46 children with AML with RUNX1::RUNX1T1 in Sweden, Finland, Norway, Denmark, Hong Kong and Israel during 2013-2020, comprising 22 females and 23 males with median age of 10 years (range 4-17). They were treated according to the NOPHO-DBH AML2012 protocol. FCM-MRD analysis utilized antibodies allowing for identification of basophils as CD123+, HLA-DR-, CD33+ and mast cells as CD117++, HLA-DR-, CD33+. FCM-MRD was performed at diagnosis, day 22 after induction 1, before start of induction 2, and before start of consolidation. Results were compared with 70 children treated in Sweden with the same protocol during the same time period for AML without RUNX1::RUNX1T1. When analyzing children with AML with RUNX1::RUNX1T1 on day 22 after induction 1, both basophils (median 0.64%, range 0-40%) and mast cells (0.44%, 0-29%) were elevated compared with regenerating bone marrow (basophils 0.32%, 0.04-0.69%; mast cells 0.01%, 0.006-0.06%). In 80% of cases, the basophils displayed atypical immunophenotype with lower expression of CD123, CD38, CD11b and CD13 and higher CD34 and CD117, and in 38% mast cells exhibited lower CD117. Levels normalized before consolidation treatment (basophils 0.12%, 0-0.7%; mast cells 0.04%, 0-1.2%). To investigate the impact of these cells, we divided children into those with notably high basophils on day 22, defined as ≥2% (n=16), and those with low basophils, <2% (n=29). The group with high basophils day 22 had more atypical basophils and mast cells already at diagnosis, as well as high levels of mast cells day 22 (median 1.87% vs 0.27%, p=0.007). Leukemic origin of atypical basophils and mast cells at diagnosis was verified using FCM cell sorting and RT-qPCR of RUNX1::RUNX1T1. Children with high basophils did not differ from low basophil cases regarding age, white blood cell count, leukemia-associated immunophenotype, KIT or FLT3 mutation status, or treatment intensity. The clinical outcome was favorable, with only five relapses (11%), with no difference between children with high and low basophils. Similar findings were seen when children were divided into groups of high (≥0.9%, n=18) and low (<0.9%, n=27) mast cells day 22, partly overlapping with basophil groups. This suggests that both basophils and mast cells have leukemic origin but not relapse potential. A comparison with children with other AML (n=70) showed that elevated levels and atypical immunophenotype of basophils and mast cells day 22 was unique to AML with RUNX1::RUNX1T1. To understand the reason for this phenomenon, we investigated the gene expression profile of cases with AML with RUNX1::RUNX1T1 using the TARGET (children), Beat AML 1.0 and TCGA-LAML (both adults) datasets. In all three datasets, AML with RUNX1::RUNX1T1 showed an enrichment of the signature genes of the common myeloid progenitor with eosinophil/mast cell/basophil potential, namely CSF2RB (CD131), CLC, HDC, EPX and IL5RA, compared to other types of AML. This suggests that RUNX1:::RUNX1T1 leukemia can originate in this progenitor. In conclusion, during induction treatment of children with AML with RUNX1::RUNX1T1, basophils and mast cells are often increased. Since they are leukemia-related and often have atypical immature immunophenotype, they might cause concern. However, such cells do not seem to be associated with a worse prognosis but rather reflect an inherent feature of this leukemia and should not be interpreted as MRD.

Abstract BACKGROUND Whole genome sequencing (WGS) is the most informative singular molecular assay in cancer diagnosis. Recent evidence demonstrates that WGS can add diagnostic information and change the management of childhood cancer, and thus is being increasingly employed in clinical settings globally. However, it remains unknown whether WGS can accurately recapitulate existing multi-assay standard-of-care (SOC) genomic testing used in pediatric cancer diagnostics. In this study we evaluate the concordance between WGS and SOC findings from an unselected cohort of children across 8 centres from two healthcare systems (England and Sweden) that offer routine WGS. METHODS We compared WGS and SOC genomic test reports for children under 18 years presenting with new or relapsed cancer between January 2021 and November 2023 across 2 English centres; Cambridge University Hospital and Great Ormond Street Hospital, and 6 Swedish centres; Gothenburg, Karolinska, Linköping, Lund, Umeå and Uppsala. Only WGS findings reported to clinicians were evaluated, without re-analysis of genomes. Tests were described as ‘concordant’ where WGS and SOC were in complete concordance (positive or negative) for all SOC-detected variants. Discordance described occasions where SOC detected findings not identified by WGS. ‘Additional findings’ described cases where WGS provided disease-relevant findings above SOC testing. Only disease-relevant variants were considered in the analysis. RESULTS A cohort of 1032 patients with 1841 SOC molecular tests was included – 436 with haematological malignancies and 596 with solid tumor malignancies (528 from England, and 504 from Sweden). WGS recapitulated 99.3% of SOC tests performed, across all types of genomic alteration (1829/1841). Of the 12 instances of discordance, 3 related to poor WGS sample purity, 5 were gene fusions, 1 low variant allele frequency (0.02) internal tandem duplication, 2 single nucleotide variants and 1 copy-number aberration. WGS provided additional disease relevant findings in 19.7% of cases (203/1032). DISCUSSION Deployment of available SOC genomic testing for cancer diagnostics is highly variable across nations, individual centres and disease entities, and is usually dictated by test availability, cost and likely clinical yield, in a non-agnostic manner. For the first time we demonstrate, across two national systems, that WGS faithfully recapitulates the vast majority of SOC findings irrespective of mutation class, cancer type and variant calling algorithm. Sample quality and intra-tumoral heterogeneity likely account for the few discrepancies observed. Barriers to implementation of routine WGS as the only molecular diagnostic assay for pediatric cancer are cost, analytical expertise, and turnaround time (TAT). Our group is systematically studying the health economic benefits of WGS as a single assay to replace all SOC testing. Finally, an ongoing collaborative project aimed at reducing TAT to under 48 hours using novel technology has demonstrated feasibility in a small number of patients to date. Citation Format: Jonathan Kennedy, Sarah M. Leiter, Angus Hodder, Sheng-Yuan Kan, Jack Bartram, Giuseppe Barone, Michael Gattens, Matthew J. Murray, Sam Behjati, Patrick Tarpey, Matthew Cullen, Antony Ceraulo, Karin Langenberg, Jan Molenaar, Sandra Wessman, Frida Abel, Gustaf Ljungman, Geraldine Giraud, Hakon Anderson Blomstrand, Zdenek Rohan, Anna Staffas, Christina Orsmark-Pietras, Tatjana Pandzic, Irina Golovleva, Linda Fogelstrand, Jonas Abrahamsson, Ulrika Norèn-Nyström, Josefine Palle, Thoas Fioretos, Lucia Cavelier Franco, Gisela Barbany, Nadège Corradini, Gudrun Schleirmacher, Richard Rosenquist, David Gisselsson, Aditi Vedi. Whole genome sequencing can reproduce all standard-of-care diagnostics for childhood cancer: Results from two national systems [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Advances in Pediatric Cancer Research; 2024 Sep 5-8; Toronto, Ontario, Canada. Philadelphia (PA): AACR; Cancer Res 2024;84(17 Suppl):Abstract nr B011.

Introduction Allopurinol can be used in maintenance therapy (MT) for acute lymphoblastic leukemia (ALL) to mitigate hepatic toxicity in patients with skewed 6-mercaptopurine (6MP) metabolism. These patients have high erythrocyte levels of methylated mercaptopurine metabolites (e-MeMP) associated with liver toxicity, including transaminitis and hypoglycemia, and low erythrocyte levels of thioguanine nucleotides (e-TGN), the key intermediate metabolites mediating the antileukemic effect. Retrospective studies and case reports show that the unfavorable metabolite ratio (high e-MeMP/TGN) can be modified by adding allopurinol leading to lower MeMP and higher TGN. In a recent publication https://doi.org/10.3324/haematol.2023.284390 we showed that allopurinol leads to a similar metabolic shift also in unselected pediatric ALL-patients, without previous severe liver toxicity or other signs of skewed 6MP metabolism. We have now analysed the levels of DNA incorporated TGN (DNA-TG) in our cohort since more recent studies, e.g. https://doi.org/10.1016/s1470-2045(17)30154-7, suggests that relapse-free survival is better correlated to DNA-TG than to e-TGN. Methods Pediatric ALL patients on NOPHO ALL-2008 non-high risk protocols with thiopurine methyltransferase wild-type were studied in a prospective before-after trial, NCT03022747. 6MP metabolites were measured, in total 9 blood samples per patient, during 12 weeks of standard MT, followed by 12 weeks of MT with addition of allopurinol 50 mg/m2 and finally 4 weeks of MT without allopurinol. Mean DNA-TG for each patient was calculated for all study phases separately. DNA-TG/e-TGN ratio was calculated as there was a concern that a decrease in e-MeMP might reduce the DNA incorporation of TGN, since e-MeMP is known to inhibit de novo purine synthesis. The 6MP dose was reduced by 50% when allopurinol was initiated to prevent excessive myelosuppression. DNA-TG was quantified using the same method as in the study cited above, with 1−2 µg DNA purified from whole blood and thioguanine measured with ultra-performance liquid chromatography tandem mass spectrometry. Results 51 patients from Sweden and Finland, age 0-15 (median 4) years, were included, of whom 48 completed the study. In paired analysis DNA-TG was 393 fmol/mg DNA higher (1248 vs 855) (p<0.001), e-TGN 193 nmol/mmol Hb higher (457 vs 264) (p<0.001) and e-MeMP 6324 nmol/mmol Hb lower (2614 vs 8938) (p<0.001) when allopurinol was added, Mean DNA-TG/e-TGN ratio did not change significantly, 3.15 on allopurinol compared to 3.51 before, (p=0.10). Four weeks after allopurinol was discontinued, mean DNA-TG decreased 564 fmol/mg DNA (p<0.001), e-TGN decreased 271 nmol/mmol Hb (p<0.001) whereas e-MeMP showed a non-significant increase of 777 nmol/mmol Hb (p=0.17). Alanine aminotransferase decreased by 39% a few weeks after the e-MeMP-levels decreased. Mean absolute neutrophil count (ANC) was lower on allopurinol, 1.35 x 109/L, compared to 1.72 before (p<0.001). Allopurinol did not increase the number of severe adverse events (SAE) nor cause any life-threatening episodes. In accordance with study protocol the mean given dose 6MP was halved during the weeks on allopurinol, 187 mg/m2/week compared to 379 (p<0.001) for the weeks before allopurinol. Conclusions Addition of allopurinol to standard MT with 6MP and methotrexate leads to substantially increased DNA-TG levels. DNA-TG/e-TGN ratio was unchanged indicating that DNA incorporation of TGN was not significantly affected during allopurinol treatment. The 46 % rise in DNA-TG, together with our previously published data with higher proportion of ANC levels within target, lower ALT and no increase in SAE, indicate that addition of allopurinol could be an effective strategy to optimize ALL MT.

Background: Central nervous system disease (CNS3) in pediatric acute myeloid leukemia (pAML) is reported in 6% to 29% of cases. However, its impact on event-free survival (EFS) and overall survival (OS) remains uncertain. This study evaluates the effect of CNS involvement at diagnosis on relapse and survival in patients treated on the NOPHO-DBH AML2012 protocol. Methods: Data from 931 pediatric AML patients in the NOPHO-DBH AML2012 protocol were analyzed, comparing outcomes, relapse rates, and survival between those with and without CNS disease. CNS-directed therapy included intensified intrathecal chemotherapy without irradiation. Results: Of 922 patients with available CNS status, 10.9% had CNS3 at diagnosis. CNS3 patients were younger (median age 3.5 vs. 8 years,P=0.001) with higher white blood cell counts (56.1x109/L vs 18.7x109/L,P<0.001) and higher frequency of other extramedullary disease (30.4% vs 11.3%,P<0.001) and inv(16)(P<0.001). EFS5y was 71.5% for CNS-positive patients vs. 62.7% for CNS-negative (p=0.14), and OS5y was 81.4% vs. 79.3% (p=0.54). Patients with CNS disease had a lower cumulative incidence of relapse (15.6% vs 26.5%,P=0.023), and CNS relapse occurred in 0.9% and 0.7% of patients with and without CNS disease. Conclusion: CNS disease at diagnosis in pAML does not adversely affect survival or treatment response.

PURPOSE Several genomic subsets of NPM1 mutations with varying sequences (type A, B, D, etc) have been identified. Despite molecular heterogeneity, NPM1 mutations cumulatively portend a more favorable outcome, but biology and prognostic implications of different genomic subsets have not been extensively studied. In this multicentric study, we investigated the impact of NPM1 genotypes on patient's outcomes and interrogated the underlying biology of the different subtypes. MATERIALS AND METHODS Of more than 4,000 patients enrolled in multiple pediatric cooperative (AIEOP, BFM, ELAM02, NOPHO, DCOG, and COG trials), or adult (SWOG) trials, 348 pediatric and 75 adult AML patients with known NPM1 genotype and available outcome were selected for this study. Diverse NPM1 variants were correlated with the probabilities of overall survival (OS) and event-free survival. Nuclear localization and translational efficiency of the NPM1 variants was studied. RESULTS Evaluation of clinical outcome on the basis of NPM1 genotypes showed that patients with type A, B, and other rare variants had similarly favorable outcomes, whereas those with type D had a significantly worse outcome (OS of 63% for type D v 86% for type non-D, P = .005). Multivariate analysis confirmed type D as an independent prognostic factor associated with inferior OS (hazard ratio, 3; P = .005). In vitro, we demonstrated that in type D versus type A synonymous variants, codon optimality plays major roles in determining gene expression levels, and translation efficiency, which resulted in a more expressed NPM1-D mRNA and protein, mediating peculiar mitochondrial gene expression. CONCLUSION The evaluation of specific NPM1 genotypes identified AML patients with type D mutations being significantly associated with inferior outcomes, suggesting a reclassification of D cases to higher-risk groups.

Despite overall survival (OS) rates of approximately 80% in recent collaborative trials, relapse remains the predominant cause of death in childhood acute myeloid leukemia (AML). We present disease characteristics, prognostic factors, and outcome in children with first relapse of AML after primary therapy according to the NOPHO-DBH AML 2012 protocol. Since 2013, patients with AML from the Nordic and Baltic countries, Hong Kong, the Netherlands, Belgium, Spain, and Israel have been included in the 2012 protocol. Patients experiencing first relapse before January 1st, 2023, were included. Disease characteristics, treatment details from both primary therapy and relapse, response assessments, and outcome were retrieved from the NOPHO-AML database and case report forms. Relapses after NOPHO-AML 2004 therapy were used as a comparison cohort (n=137). OS and 95% confidence intervals (CI) were assessed using the Kaplan-Meier method. Cox proportional models, including core binding factor (CBF) aberrations, time to relapse, and stem cell transplantation (SCT) in first complete remission (CR1) as co-variates were used for multivariate analyses. Relapse occurred in 190 patients. Distributions of sex and age were comparable for patients with relapse after treatment on the 2012 and 2004 protocols. AML harbouring KMT2A-rearrangements was more common in patients with relapse after 2012 therapy (18% vs. 7%, P=0.003), whereas t(8;21) was less common (8% vs. 17%, P=0.01) compared to the 2004 protocol. There was no difference in 3-year OS between relapsed patients treated on the 2012 (45%; CI 38-53) and 2004 protocols (43%; CI 34-51, P=0.9). Patients with CBF AML performed exceptionally well with an OS of 85% (CI 65-94) in the 2012 cohort. Patients with non-CBF AML had an OS of 37% (CI 30-45). Patients with early relapse (<12 months from primary diagnosis) had a significantly lower OS (31%; CI 22-40) than those with late relapse (62%; CI 50-72, P<0.001). Patients who underwent SCT in CR1 had a significantly lower OS (26%; CI 12-43) compared to patients who received chemotherapy only (48%; CI 40-56, P=0.004). Patients who were multiparameter flow cytometrymeasurable residual disease (MRD) positive (≥0.1% leukemic blasts) after first course of therapy at initial AML diagnosis showed an OS of 36% (CI 25-48) compared to 50% (CI 40-60) in MRD negative patients (P=0.08). Data on relapse treatment was available for 184 of 190 patients. Curative treatment was initiated in 175 (95%) patients with relapse, 119 (65%) patients received a fludarabine and high-dose cytarabine (FLA)-based regimen including an anthracycline (84/119; 69%), 44 (25%) patients received other myelosuppressive regimens, and 12 (7%) received experimental therapy as first relapse therapy. Second complete remission was achieved in 120 (69%) patients of whom 107 (89%) proceeded to SCT with an OS of 69% (Cl 59-77). MRD after first reinduction course was available for 106/171 (62%) patients with bone marrow relapse. MRD negative patients showed a higher OS of 68% (CI 52-80) compared to MRD positive patients (OS 34%; CI 22-45, P=0.0009). MRD before SCT was available for 77 (72%) patients and demonstrated an improved OS in MRD negative patients compared to MRD positive patients (75%; CI 60-85 vs 36%; CI 18-54, P=0.003). MRD negativity after first reinduction course (Hazard ratio (HR) 0.47; CI 0.26-0.85, P=0.01), late relapse (HR 0.29; CI 0.17-0.52, P<0.001), and CBF AML (HR 0.06; CI 0.01-0.47, P=0.007) were significant predictors of higher OS. With the exception of CBF AML, OS after relapse has not improved in the last decades in our experience. In addition to established prognosticators, we identify MRD during reinduction therapy as a significant predictor of outcome.

Treatment intensification of children with acute myeloid leukemia (AML) depends on measurable residual disease (MRD) analysis assessed by multiparameter flow cytometry (MFC). MFC is also increasingly used for MRD analysis in adults. However, a substantial proportion of patients with no detectable MRD by MFC eventually relapse. Next-generation sequencing of leukemia-specific mutations, targeted ultra-deep sequencing (UDS), is a promising tool to provide highly sensitive detection of MRD. UDS is designed for the mutations present in the individual patient and has a robust limit of detection of 0.02% variant allele frequency. As opposed to qPCR of patient-specific genomic breakpoints, it does not require a standard curve of diagnostic sample. Here we aimed to investigate the feasibility of UDS for analysis of MRD in children with AML, and to characterize residual mutated cells. We assessed the sensitivity of UDS as compared to MFC and characterized the genetic and protein signature on the single cell level. Whole exome sequencing (WES) of sorted leukemic cells of bone marrow (BM) samples collected from 26 children with AML at diagnosis was used to identify leukemia-specific mutations. For UDS tracking, we selected 1-2 clonal mutations (present in all leukemic cells) in any gene, and 1-2 subclonal mutations in cancer-associated genes. BM samples (n=96) from day 22, before start of second induction and first consolidation courses from 24/26 patients were assessed with patient-tailored UDS. In total, 68 mutations were assessed. Results of all samples were compared to parallel MRD analyses with 8 color MFC with a limit of detection of 0.1% cells, and when applicable RT-qPCR (8/24 patients). Single cell multi-omics analysis was performed on 15 BM samples from 7/24 patients at diagnosis and the same follow-up time-points using the Tapestri platform. The analysis included 45 primer-barcoded antibodies for cell surface protein and customized mutation-specific primers for DNA, enabling detection of mutations and copy number aberrations, followed by sequencing. In the diagnostic samples, WES identified 1-16 leukemia-specific mutations/case (median 9 mutations). Samples during treatment were assessed with UDS and compared to MFC results. Interestingly, 51 of 96 analyzed samples (53.1%) were positive for at least one mutation with UDS but negative with MFC, 35/96 samples (36.5%) were positive with both methods, and 10/96 samples (10.4%) were negative with both methods. In the 8 patients that were also assessed with RT-qPCR, UDS and RT-qPCR showed similar performance. Thus, UDS was more sensitive than MFC, and importantly, leukemia-specific mutations were detectable before first consolidation course in patients that eventually relapsed, but also in some patients not experiencing relapse. To clarify this observation, we aimed to identify the origin of residual mutated cells using single cell multi-omics. First, a non-leukemic BM sample was used to verify single cell protein expression signatures by performing MFC in parallel. Using 42 protein markers and 957 single cells, UMAP clustering identified 16 human BM cell types. Overall, there was a good consistency between single cell analysis and MFC, with slightly higher abundance of mononuclear cell types, e.g. progenitors and lymphocytes, with single cell analysis. Next, we analyzed 15 AML samples with single cell multi-omics. At diagnosis, we could characterize leukemic clones, with good concordance with findings from WES. The detected patterns were patient-specific, and in addition to subclonality also revealed the presence of founding clone in stem/progenitor cells. In follow-up samples, single cell multi-omics provided detection of MRD on a single cell level. Single cell results confirmed findings with UDS in the respective children. Furthermore, they revealed the genomic and phenotypic characterization of residual leukemic cells, which in some cases represented the founding clone in progenitors, and in some cases represented leukemic clone in mature cells. In conclusion, targeted UDS can enable a more sensitive detection of MRD than MFC and is applicable to most children with AML. Residual mutated cells during treatment can be stem/progenitor cells representing the founder clone, or mature cells originating from the leukemic clone. Further studies are needed to address the prognostic value of highly sensitive molecular MRD in childhood AML.

The survival rate for childhood acute lymphoblastic leukaemia (ALL) has improved over time, due to uniform treatment with risk-adapted therapy developed through international collaboration. The international ALLTogether treatment protocol (EUDRACT 2018–001795-38; ClinicalTrials.gov: NCT03911128) replaced the NOPHO ALL2008 (ALL2008; EUDRACT NCT03911128) protocol in Sweden in 2019. ALLTogether introduced a high dose of intravenous asparaginase during early induction combined with dexamethasone. Both asparaginase and glucocorticoids lead to a transient coagulopathy. Asparagine depletion reduces liver synthesis both pro- and anticoagulants, most importantly fibrinogen and antithrombin,1 leading to coagulopathy, whereas dexamethasone induces hypercoagulability.2, 3 After the initiation of ALLTogether, we encountered patients with bleeding tendency during early treatment regardless of sufficient platelet levels. Further examination revealed activated partial thromboplastin time (aPTT) without clot formation and immeasurable low fibrinogen (<0.3 g/L). Replacement of fibrinogen corrected the bleeding tendency and aPTT. To address concerns on increased risk of bleeding, we collected antithrombin and fibrinogen values and clinical data on thromboses and bleeding events during early treatment, and compared the results between the two protocols. This national study was approved by the Swedish Ethical Review Authority (2021–03146). We identified all children (n = 120) between the age of 1 and 18 years at diagnosis and treated in the six Swedish childhood cancer centres according to ALLTogether protocols between August 2019 (initiation of the pilot in Sweden) and March 2021 (the first substantial amendment) from the Swedish Childhood Cancer Registry. Patients with Down syndrome (n = 2) or transferred to high-risk block therapy during consolidation 1 (n = 1) were excluded. We selected controls matched by NCI-risk group treated with the ALL2008 protocol (n = 185). Patient characteristics between groups were balanced (Table 1). Data on coagulation parameters, thromboses and bleeding events during induction and consolidation 1 were collected from medical charts. As clinical practice in measuring fibrinogen and antithrombin values during treatment varied between the paediatric oncology centres, we could not compare the values at specific timepoints and recorded only the lowest measured abnormal values. Otherwise, fibrinogen and antithrombin were classified as normal or below reference value. Fibrinogen values were available for 76/117 (65%) of the ALLTogether and 99/185 (53%) of the ALL2008 patients. During ALLTogether induction, 46% (29/63) had fibrinogen levels below the reference (<2.0 g/L), comparable to 54% (45/84) during ALL2008 (p = 0.295). Hypofibrinogenemia was more pronounced during ALLTogether induction, 86% (25/29) had fibrinogen values <1.0 g/L, and 35% (10/29) had fibrinogen values <0.5 g/L, p < 0.001 (Table 1). During consolidation 1, 58% (28/48) of the ALLTogether patients had fibrinogen levels below the reference, compared with 75% (51/68) in the ALL2008 group (p = 0.058). Extremely low fibrinogen values (<0.5 g/L) during consolidation 1 were uncommon (1/28 and 1/51, respectively). Antithrombin was measured in 56/117 (48%) in the ALLTogether and 86/185 (46%) in the ALL2008 groups. More patients treated according to ALLTogether (65%; 28/43) than ALL2008 (10%; 7/72) had antithrombin values under the reference (<0.85 IU/mL) during induction (p < 0.001), but the level of reduced antithrombin activity did not differ between the protocols. No differences between the groups were seen under consolidation 1 (Table 1). As the initial reason to compare the two protocols was the unexpectedly low fibrinogen levels, we collected both major (>20 g/L decrease in haemoglobin level due to the bleed and/or bleeding event involving critical organs such as the central nervous system) and non-major (defined as bleeding events leading to intervention by healthcare professionals, e.g. transfusions, hospital admission or other measures that would otherwise not have been taken) bleeding events. There were no significant differences in bleeding events between the protocols: 15/117 (13%) of the ALLTogether patients had 23 bleedings compared with 34/185 (18%) patients with 47 bleeding events in ALL2008 (p = 0.202). Both groups had two major bleedings: one epidural bleeding and one secondary to cerebral sinus venous thrombosis (CSVT; ALLTogether) and one haemopneumothorax and one intestinal bleeding (ALL2008). We then compared thrombosis during early treatment: 6% (7/117) of ALLTogether patients had a thrombotic event, defined as thrombosis leading to anticoagulant treatment (three CSVTs, two upper and two lower venous system). One CSVT was associated with secondary bleeding. By comparison, 4/182 (2%) had a thrombosis (three CSVTs and one lower venous system) in the ALL2008 group. In line with previous reports,3-5 combining asparaginase with dexamethasone led to a pronounced hypofibrinogenemia during induction in ALLTogether. Despite low fibrinogen during ALLTogether, the number of bleeding events did not increase. This is likely due to simultaneous decrease in anticoagulant proteins maintaining the delicate haemostatic balance. Non-major bleeds, expected but seldom reported side effects during ALL treatment, were common and led to outpatient visits, hospital admissions or transfusions. Our cohort was not powered to compare the incidence of thrombotic events between the two protocols, but we would expect the timing of thromboses to be earlier during ALLTogether than during ALL2008. Study limitations included the relatively low number of patients, retrospective approach, and differences in routines and the laboratory methods used for measuring coagulation parameters between centres and protocols. Many patients lacked fibrinogen and antithrombin values, especially during ALL2008 induction, before the start of asparaginase treatment. Due to these limitations, the results should be interpreted with caution. Study strengths were the national, population-based design and detailed patient chart review. In conclusion, our study showed no significant differences in the frequency of thrombosis or bleeding events between treatment protocols despite significantly decreased fibrinogen levels during ALLTogether. Yet, haemostatic reserve capacity during ALLTogether may be even more reduced than during ALL2008 and should be considered with clinical bleeding tendency and prior to surgery. As some patients had immeasurable fibrinogen levels, we encourage to consider the increased potential predisposition to bleeding in ALLTogether, particularly during induction, to optimise haemostasis in high-risk scenarios when managing these patients. Future studies should try to clarify whether novel methods including global haemostasis assays can help to identify children at risk of bleeding and those who may benefit from thromboprophylaxis. SR, AH, JA, MB, MH, ASH and HV contributed to the design and implementation of the study. JF, JJ, ID, LH, AV, AW and OZ collected data from paediatric cancer centres. JF, AH and SR performed data analysis and interpretation. JF wrote the original draft. All authors reviewed, edited and approved the original draft. We thank the Ellen Bachrach Memorial Fund (AH) and the Swedish Childhood Cancer Fund (MH, SR) for financial support. The authors have no conflict of interest to report.

A comprehensive international consensus on the cytogenetic risk -group strati fication of KMT2A -rearranged ( KMT2A -r) pediatric acute myeloid leukemia (AML) is lacking. This retrospective (2005-2016) International Berlin-Frankfurt-M & uuml;nster Study Group study on 1256 children with KMT2A -r AML aims to validate the prognostic value of established recurring KMT2A fusions and additional cytogenetic aberrations (ACAs) and to de fine additional, recurring KMT2A fusions and ACAs, evaluating their prognostic relevance. Compared with our previous study, 3 additional, recurring KMT2A -r groups were de fined: Xq24/ KMT2A :: SEPT6 , 1p32/ KMT2A :: EPS15 , and 17q12/t(11;17)(q23;q12). Across 13 KMT2A -r groups, 5-year event-free survival probabilities varied signi ficantly (21.8%-76.2%; P < .01). ACAs occurred in 46.8% of 1200 patients with complete karyotypes, correlating with inferior overall survival (56.8% vs 67.9%; P < .01). Multivariable analyses con firmed independent associations of 4q21/ KMT2A :: AFF1 , 6q27/ KMT2A :: AFDN , 10p12/ KMT2A :: MLLT10 , 10p11.2/ KMT2A :: ABI1 , and 19p13.3/ KMT2A :: MLLT1 with adverse outcomes, but not those of 1q21/ KMT2A :: MLLT11 and trisomy 19 with favorable and adverse outcomes, respectively. Newly identi fied ACAs with independent adverse prognoses were monosomy 10, trisomies 1, 6, 16, and X, add(12p), and del(9q). Among patients with 9p22/ KMT2A :: MLLT3 , the independent association of French-American- British -type M5 with favorable outcomes was con firmed, and those of trisomy 6 and measurable residual disease at end of induction with adverse outcomes were identi fied. We provide evidence to incorporate 5 adverse -risk KMT2A fusions into the cytogenetic risk - group strati fication of KMT2A -r pediatric AML, to revise the favorable -risk classi fication of 1q21/ KMT2A :: MLLT11 to intermediate risk, and to re fine the risk -strati fication of 9p22/ KMT2A :: MLLT3 AML. Future studies should validate the associations between the newly identi fied ACAs and outcomes and unravel the underlying biological pathogenesis of KMT2A fusions and ACAs.

Acute myeloid leukemia (AML) is caused by genetic abnormalities in myeloid progenitors. These abnormalities are categorized in the 5th edition of the World Health Organization (WHO) Classification of Haematolymphoid Tumours and The International Consensus Classification (ICC) of Myeloid Neoplasms and Acute Leukemias. The WHO and ICC classifications are mainly derived from adult AML cohorts, but distinct patterns of cytogenetic abnormalities exist between children and adults with AML. We present a comparison of the WHO and ICC cytogenetic classifiers in a large cohort of children with AML. Children with AML were included from The Nordic Society for Pediatric and Hematology and Oncology (NOPHO) database between 2004 - 2021. NOPHO spans the Nordic and Baltic countries. Since 2012 NOPHO-DB-SHIP also includes Hong Kong, the Netherlands, Belgium, Spain, and Israel. Inclusion criteria were de novo non-M3-AML aged 0 - 18 years at diagnosis. Children with myeloid leukemia of Down syndrome, juvenile myelomonocytic leukemia, therapy-related AML and AML secondary to inherited bone marrow failure or predisposition syndromes were excluded. Children were treated as per the NOPHO AML 2004 or the NOPHO-DBH AML 2012 protocol. Classification was based on cytogenetics, i.e., cryptic or molecular variants were included in WHO, AML defined by differentiation or ICC, AML Not Otherwise specified (NOS). KMT2A-rearrangements (KMT2A-r) other than MLLT3::KMT2A were included in ICC, AML with other KMT2A-r. The Kaplan-Meier method was used to estimate overall survival (OS) with 95% confidence intervals (CI). We identified 1,080 children of whom 1,021 (95%) children had available diagnostic cytogenetics. More children (n = 694, 68%) were treated as per the NOPHO-DBH AML 2012 protocol. The three largest entities in WHO, AML with defining genetic abnormalities were AML with KMT2A-r (n = 259, 25%), AML with RUNX1::RUNX1T1 fusion (n = 142, 14%), and AML with CBFB::MYH11 fusion (n = 90, 9%). The three largest entities in ICC, AML with recurrent genetic abnormalities were AML with RUNX1::RUNX1T1 fusion (n = 142, 14%), AML with other KMT2A-r (n = 138, 14%), and AML with MLLT3::KMT2A fusion (n = 121, 12%). Rare entities in WHO, AML with defining genetic abnormalities and ICC, AML with recurrent genetic abnormalities were DEK::NUP214 fusion (n = 10, 1%), AML with BCR::ABL fusion (n = 2, <1%) and AML with MECOM rearrangements (n = 1, <1%). Categories consistently included in WHO, AML with defining genetic abnormalities (p = 0.002) and ICC, recurrent genetic abnormalities (p < 0.001) had significant differences in OS. AML with CBFB::MYH11 fusion had the highest 5-year OS (88%; CI: 82 - 95%) followed by AML with RUNX1::RUNX1T1 fusion (85%; CI 79 - 91%). WHO, AML with KMT2A-r had a 5-year OS of 75% (CI: 70 - 81%). We observed no difference in 5-year OS comparing ICC, AML with MLLT3::KMT2A (OS: 74%; CI: 66 - 83%, p = 0.8) to ICC, AML with other KMT2A-r (OS: 0.76%; CI: 69 - 84%). WHO, AML defined by differentiation included 367 (36%) children and ICC, AML NOS included 339 (33%) children. WHO, AML, Myelodysplasia-related (MR) included fewer children (n = 119, 12%) than ICC, AML with myelodysplasia-related cytogenetic abnormalities (MDSk) (n = 147, 14%). The only distinct cytogenetic classifier present in this cohort for WHO, AML MR was del(11q) (n = 1, <1%). Comparably, the distinct cytogenetic classifiers present in this cohort for ICC, AML MDSk were add(5q) (n = 1, <1%) and +8 (n = 28, 3%). Children with +8 only included in ICC, AML MDSk were significantly older at diagnosis (median: 13 years, range: 1 - 18 years, p < 0.001) but their 5-year OS was not significantly different (OS: 65%; CI: 47 - 90, p = 0.4) compared with other children included in AML MDSk (median: 4 years, range: 0 - 18 years, OS: 61%; CI: 51 - 72%). WHO, AML MR (OS: 61%; CI: 52 - 71%, p = 0.001) and ICC, AML MDSk (OS: 62%; CI: 54 - 71%, p = 0.002) had a significantly lower 5-year OS than WHO, AML defined by differentiation (OS: 74%; CI: 69 - 79%) and ICC, AML NOS (OS: 75%; CI: 70 - 80%), respectively. Using cytogenetic classifiers, the WHO and ICC had similar prognostic capabilities in childhood AML and were able to identify children with a poor prognosis despite differences in defining myelodysplasia-related cytogenetic abnormalities. Cytogenetic classification leaves one third of children with AML as WHO, defined by differentiation or ICC, NOS underlining the need to characterize molecular variants driving childhood AML.

Hypodiploidy, defined as modal numbers (MNs) 45 or lower, has not been independently investigated in pediatric acute myeloid leukemia (AML) but is a well-described high-risk factor in pediatric acute lymphoblastic leukemia. We aimed to characterize and study the prognostic impact of hypodiploidy in pediatric AML. In this retrospective cohort study, we included children below 18 years of age with de novo AML and a hypodiploid karyotype diagnosed from 2000 to 2015 in 14 childhood AML groups from the International Berlin-Frankfurt-Münster (I-BFM) framework. Exclusion criteria comprised constitutional hypodiploidy, monosomy 7, composite karyotype, and t(8;21) with concurring sex chromosome loss. Hypodiploidy occurred in 81 patients (1.3%) with MNs, 45 (n = 66); 44 (n = 10) and 43 (n = 5). The most frequently lost chromosomes were chromosome 9 and sex chromosomes. Five-year event-free survival (EFS) and overall survival (OS) were 34% and 52%, respectively, for the hypodiploid cohort. Children with MN≤44 (n = 15) had inferior EFS (21%) and OS (33%) compared with children with MN = 45 (n = 66; EFS, 37%; OS, 56%). Adjusted hazard ratios (HRs) were 4.9 (P = .001) and 6.1 (P = .003). Monosomal karyotype or monosomy 9 had particular poor OS (43% and 15%, respectively). Allogeneic stem cell transplantation (SCT) in first complete remission (CR1) (n = 18) did not mitigate the unfavorable outcome of hypodiploidy (adjusted HR for OS was 1.5; P = .42). We identified pediatric hypodiploid AML as a rare subgroup with an inferior prognosis even in the patients treated with SCT in CR1.

Background and Significance Pediatric AML is a heterogeneous disease with current rates for event-free survival (EFS) of 50-60% and for overall survival (OS) of 70-80%, for newly diagnosed patients. In patients with refractory and relapsed disease, and in subsets as defined by AML biology and treatment response, prognosis is dismal. To further improve EFS, the NOPHO-DB-SHIP consortium has initiated trial CHIP-AML22. Building on the successes achieved by the recent trial from this consortium (NOPHO-DBH AML-2012, preliminary data as per June 2023, n=878: 5-years pEFS 63%, 5-years pOS 78%), CHIP-AML22 will incorporate innovative elements, mainly the FLT3-inhibitor quizartinib (Vanflyta®) that was recently approved by the FDA and PMDA in combination with chemotherapy for the treatment of newly diagnosed adult AML patients with FLT3-ITD mutations, and gemtuzumab ozogamicin (GO, Mylotarg ®) that improved outcome in adult and pediatric AML when added to chemotherapy. CHIP-AML22 will maintain the approach of treatment-response driven risk-adapted treatment, intensified induction treatment for patients with ≥5% AML cells in the bone marrow (BM) after course 1, and consolidation with allogeneic stem cell transplantation (allo-SCT) for high-risk patients after the first consolidation course HAM. Study Design and Methods This clinical trial will enroll newly diagnosed de novo patients aged ≥1 day to ≤18 years to 2 randomized studies in the Master protocol, to the linked quizartinib trial, or the standard arm of the Master (Figure 1). Innovations include: (1) a phase II, single arm, open-label study on the safety, efficacy, pharmaco-kinetics and -dynamics of quizartinib in combination with chemotherapy, and as single-agent after allo-SCT, in FLT3-ITD/ NPM1wt AML; All patients will receive quizartinib monotherapy for 6 months post-SCT, except those that were negative for measurable residual disease determined with multiparameter flow cytometry (MFC-MRD) in BM1 and remained negative (study CHIP-AML22/Quizartinib); (2) a phase III, open-label randomization with gemtuzumab ozogamicin (GO; Mylotarg®) added to induction therapy MEC for pediatric CD33-positive newly diagnosed AML patients aimed at improved outcome (study Ri), (3) a phase III, open-label randomization with three (standard of care) versus two (investigational) courses of consolidation chemotherapy for standard-risk patients as non-inferiority trial (study Rc), and (4) refining risk-group adapted treatment, aimed at improved outcome. Novel high-risk categories are patients with RAM-phenotype and/or CBFA2T3::GLIS2, and patients with KMT2A-rearranged AML (excluding KMT2A::MLLT3) with ≥0.1% AML cells after course 1 in BM1. Patients with ≥15% AML cells on day 22 BM after course 1, or ≥0.1-5% AML cells by MFC-MRD after course 2 in BM2 (end-of-induction), or those with FLT3-ITD/NPM1wt remain to be high-risk, similar to AML-2012. Further, dexrazoxane is recommended to all patients before receiving daunorubicin or mitoxantrone, aiming to prevent late-onset cardiotoxicity. The quizartinib trial includes a safety run-in with a modified rolling-6 design and two dose levels with 6 evaluable patients per dose level. In study Ri, patients assigned to the experimental GO-arm will receive GO twice during induction with MEC at a dose of 3 mg/m 2 (max 5 mg) with an interim-analysis for superiority of any of these two arms. The investigational arm of the Rc randomization will omit HA 3E. Both the quizartinib trial and each randomization will also be compared to historical controls from the AML-2012 trial. CHIP-AML22 expects to enroll 130-140 patients per year and it will take 7-8 years to reach the total number of 905 patients needed to answer all study questions. Patients will be followed-up for at least 5 years from diagnosis and thus, the total study duration is expected to be 12-13 years. The Master protocol (EU CT: 2023-504999-25) and linked quizartinib trial (EU CT: 2023-505000-27-00) were approved by the Dutch Ethical Review Authority and as of July 2023, enrollment in the Master protocol including study Rc has started in The Netherlands. Study Ri and the linked quizartinib trial are expected to open in Q4 2023. At least 15 other countries with nearly 60 sites will participate in CHIP-AML22: Belgium, Denmark, Estonia, Finland, Hong Kong, Iceland, Israel, Latvia, Lithuania, Norway, Portugal, Spain, Sweden, Switzerland and Uruguay.

Book Citations: Authors, Title, HemaSphere, 2023;7(S3):pages. The individual abstract DOIs can be found at https://journals.lww.com/hemasphere/pages/default.aspx. Disclaimer: Articles published in the journal HemaSphere exclusively reflect the opinions of the authors. The authors are responsible for all content in their abstracts including accuracy of the facts, statements, citing resources, etc. 853 treatment-emergent adverse events (TEAEs) of all grades.

The genomic landscape and mechanisms driving relapse in KMT2A-rearranged ( KMT2A-r) infant and childhood acute lymphoblastic (ALL) and acute myeloid leukemia (AML) are not completely understood. We therefore studied 36 KMT2A-r ALL (n=19) and AML (n=17) patients of which 25 relapsed and 11 remained in remission. Twenty diagnose-relapse-germline trios and 5 multiple relapse samples were analyzed by whole genome (WGS) and whole exome sequencing (WES) and 30 patients longitudinally by using patient-specific mutations identified by WGS/WES, including the KMT2A-r (average coverage 3300X). The mutational burden increased from diagnosis to relapse and relapse evolved through branching evolution. Relapse was seeded by multiple diagnostic clones in 56%, by a single sweeping clone detected at diagnosis in 22%, and by a single sweeping clone not detected at diagnosis in 22%. Notably, the evolutionary patterns correlated to relapse time, where multiple diagnostic clones seeding relapse were connected to an earlier relapse with all very early relapse ALL (3/3, relapse <9 months from diagnosis) and half of the early AML relapse showing this pattern (2/4, relapse <1 year in complete remission, CR1). By contrast, later relapse was connected to a sweeping clone at relapse with 67% of early relapse ALL (>9 months from diagnosis) and 40% of late relapse AML (>1 year in CR1) showing this pattern. Pathway analysis showed that cell cycle genes, glucocorticoid signalling, purine metabolism, mismatch repair, and B-cell differentiation, were enriched in early relapse ALL (83%, 5/6) and included TP53, CREBBP, NT5C2 PMS2, PRPS2, NR3C1, IKZF1, with none of the very early relapse infant ALL harboring such alterations (n=4). Further, TP53 and IKZF1alterationsco-occurred (n=4/4). These results were validated in public data sets of 98 KMT2A-r ALL infants (n=84) and children (n=14) at diagnosis and relapse (n=24) and showed that 50% of early relapse ALL, and none of the 8 very early relapse ALL, had such alterations. Ultra-deep sequencing did not detect the CREBBP, NT5C2, PRPS2or TP53 mutations at diagnosis and manual inspection of the WGS reads failed to detect the PMS2 and NR3C1 deletions. In AML, TP53 and CCND3 alterations were maintained, and gain of WT1 was seen in late relapse AML. Signalling mutations were the most common type of mutations at diagnosis (64%) and relapse (56%) and the frequency was similar in patients that remained in remission and in those that relapsed (55% versus 60%). One infant ALL and four AML patients had multiple relapses, allowing us to study how the genetic landscape evolved across consecutive relapses. This showed a stepwise replacement of clones during treatment in agreement with a fitter clone that evolves under chemotherapeutic selective pressure. Longitudinal analysis allowed sensitive detection of residual leukemia cells and showed that the relapse clone could be detected at diagnosis in 64% of patients. Further, infants with >10% of molecularly detectable leukemia cells after induction therapy, had a high risk of a very early relapse. Ultra-deep sequencing allowed detection of the relapse clone up to 4 months before relapse. In 11 of the 30 patients (3 remission and 8 relapse), low-frequency KMT2A-fusion positive leukemic cells were found at remission outside of the MRD time points. Our longitudinal data also provided unique insights into clonal response to treatment by showing that 1) a change in therapy can favour the eradication of one clone and expansion of another, 2) a clone that initially was the most sensitive clone to therapy, was the one that eventually caused relapse, and 3), a diagnostic clone can be undetectable for a long time before expanding to cause relapse, suggesting that molecular monitoring with personal mutations is a powerful tool to follow response to therapy. These results provide new biological insights into the relapse mechanisms in KMT2A-r leukemia. The data shows different clonal evolution patters depending on when in time the patient relapsed, with very early relapse ALL being seeded by multiple diagnostic subclones and a paucity of acquired genetic alterations at relapse. By contrast, early relapse ALL was characterized by a single diagnostic clone seeding relapse by a clonal sweep along with acquired mutations in chemoresistance-associated genes. To validate and extend these findings, we are currently analyzing 11 additional infant relapse samples with WGS.

Background: In pediatric acute myeloid leukemia (AML), measurable residual disease (MRD) by flow cytometry (FCM) after induction therapy has emerged as one of the strongest prognostic factors. We designed a phase III study (NOPHO-DBH AML2012) using an intensified response-guided induction with mainly MRD based risk stratification. All high-risk patients were treated with allogeneic stem cell transplantation (SCT). The study contained two randomized questions: 1) a randomized comparison of mitoxantrone and liposomal daunorubicin (DNX) in first induction which is reported elsewhere and 2) the comparison of AD(x)E (low dose cytarabine, DNX, etoposide) and FLAD(x) (fludarabine, high dose cytarabine, DNX [60 mg/m 2 x III]) which was proven effective in refractory/relapsed AML in the second induction course Patients and methods: The study was population-based and conducted in the Nordic countries, Belgium, Hong Kong and the Netherlands (NOPHO-DBH collaboration). Between March 2013 and November 2017, 194 were randomized 1:1 to AD(x)E or FLAD(x). At this time DNX became unavailable and randomization was paused. After almost 18 months, when it became clear that the shortage was permanent, randomization was resumed (February 2019-July 2021) but with daunorubicin substituting DNX. In total 306 patients were included (152 AD(x)E, 154 FLAD(x)). Randomization was stratified according to treatment arm and response to course 1. The primary endpoint was the proportion with MRD <0.1% after the second induction. Patients with MRD ≥ 0.1% after course 2 or MRD ≥ 15% after course 1 or with FLT3-ITD and NPM1 wild type were assigned high-risk (HR) and consolidated with SCT whereas standard-risk (SR) received 3 chemotherapy courses ( CBFB:: MYH11 2 courses) (Fig 1). Estimates of overall survival (OS) and event-free survival (EFS) from date of diagnosis and given at 5 years were calculated by the Kaplan-Meier method and given as percentage± standard error with differences tested by log rank test. Results: Patient and disease characteristics were similar between treatment arms (Table 1). For all 306 randomized patients EFS was 66.8±2.8% and OS 80.5±2.4%. 232 (76%) had SR with EFS of 71.1±3.1% and OS 86.9±2.3% whereas 56 (18%) had HR with EFS 67.5±6.3% and OS 70.1±6.3%. Median observation time for patients without events was 68 months. There was no difference in proportion of patients with MRD < 0.1% in the randomized groups with 112/152 (74%) for AD(x)E and 120/154 (78%) for FLAD(x). CR rates were 93.4% for AD(x)E and 96.1% for FLAD(x). Survival outcomes for AD(x)E and FLAD(x) were very similar with EFS 67.9±3.8 vs 65.9±3.9 ( P=.93) and OS 82.0±3.2 vs 79.1±3.4 ( P=.60) (fig 1). Univariate Cox regression on all factors in table 1 showed statistical significance for RUNX1:: RUNX1T1 associated with better EFS (Hazard ratio[HzR] 0.26 CI 0.10-0.70) and “other” aberrations with worse EFS (HzR 2.21 CI 1.49-3.27). For OS, both RUNX1:: RUNX1T1 (HzR 0.12 CI 0.02-0.84) and CBFB:: MYH11 (HzR 0.12 CI 0.02-0.88) showed better outcome while the group “other” aberrations had increased HzR (1.77 CI 1.03-3.0). In multivariable analysis, RUNX1:: RUNX1T1 and CBFB:: MYH11 were the only factors significantly associated with EFS and OS (lower HzR) . One patient in each arm died from toxicity after course 2. There was a trend for increased toxicity in patients treated with FLAD(x), with 134 grade 3 (CTCAE4.0) toxicities registered after course 2 compared to 103 for AD(x)E and 37 vs 27 grade 4 toxicities. Also, 18 patients in the FLAD(x) group required ICU treatment compared to 9 for AD(x)E. Conclusion. Risk stratification based on MRD FCM and response-guided induction as used in the NOPHO-DBH AML 2012 protocol resulted in one of the best cure rates reported in newly diagnosed pediatric AML. The stratification allowed 75% of patients to receive standard-risk treatment without SCT in primary therapy. Patient with suboptimal response (ie our HR group) benefitted from SCT in first remission. Perhaps surprisingly, given the efficacy in relapsed AML, FLAD(x) did not improve outcome compared to AD(x)E. Since toxicity was higher we have opted to retain ADE in the upcoming protocol.