Background Outcomes for adolescents and young adults (AYAs) with acute lymphoblastic leukemia (ALL) are improving, yet little is known specifically about their outcomes after relapse. A recent study from the Children's Oncology Group (COG) identified risk factors for overall survival (OS) after relapse for patients with ALL enrolled on frontline clinical trials from 1996 to 2014 [Rheingold et al Leuk (2024) 38: 2382]. Here we report relapse outcomes for the AYAs enrolled on these trials. Methods Patients with de novo B- or T-ALL enrolled on twelve frontline clinical trials were reviewed retrospectively. AYA patients (aged ≥15 years at initial diagnosis) who relapsed were analyzed. Univariate and multivariate Cox regression models assessed the impact of patient and disease characteristics on OS following relapse. Statistical significance was set at p <0.05. Analyses were performed using SAS software (version 9.4). Results Among 16,115 patients enrolled, 2,053 (12.7%) relapsed, including 257 AYAs between the ages of 15 and 29 at initial diagnosis. The OS rate at 5 years (± standard error) after relapse for AYAs was 27.7 ± 3.0%, versus 37.9 ± 2.8% for 10-14 year-olds and 57.9 ± 1.4% for 1-9 year-olds (p <0.001). In multivariable analysis, neither race, white blood cell count (WBC) >100,000/µL at initial diagnosis, nor minimal residual disease (MRD) at the end of initial induction were significantly associated with OS post relapse. However, traditional relapse risk stratification measures, such as isolated central nervous system (CNS) relapse and longer time from initial diagnosis to relapse correlated with better outcomes in AYAs (p<0.001). The 217 AYAs with B-ALL had a 5-year post-relapse OS of 27.7 ± 3.3% compared to 40.9 ± 3.0% for 10-14 year-olds and 59.7 ± 1.5% for 1-9 year-olds (p <0.001). B-ALL AYAs with isolated CNS relapse had better outcomes compared to AYAs with an isolated bone marrow (BM) relapse (hazard ratio (HR) [95% confidence interval (CI)] = 0.38 [0.23, 0.66], p <0.001). For AYAs with B-ALL, time from initial diagnosis to relapse was also significantly associated with OS post relapse, with a longer duration of initial remission (18-35 months or ≥36 months) associated with improved outcomes (HR [95% CI] = 0.37 [0.24, 0.58] and 0.18 [0.11, 0.28], respectively, p <0.0001 in both cases). There was a trend towards worse 5-year post-relapse OS for B-ALL AYAs of Hispanic ethnicity of all races (19.9 ± 5.4%), end of initial induction (EOI) MRD ≥0.1% (17.8±4.4%), and presenting WBC >100,000/µL (12.4 ± 5.2%), but these factors did not reach statistical significance. T-ALL data are limited by small sample size (40 patients), but similar trends emerged. The 5-year post-relapse OS for AYAs was 29.0 ± 7.7%, while the OS for 10-14 year-olds was 18.1 ± 6.0% and for 1-9 year-olds was 43.0 ± 4.3% (p <0.001). An initial WBC >100,000/µL correlated with worse OS post relapse compared to an initial WBC <50,000/µL (HR [95% CI] = 8.23 [1.68, 40.17], p=0.009). T-ALL AYAs with an isolated CNS relapse again tended to have better outcomes compared to AYAs with an isolated BM relapse (HR [95% CI] = 0.16 [0.02, 1.09], p=0.062), but there was not a significant difference in outcomes for AYAs with either combined or isolated (with or without CNS involvement) BM relapse (HR [95% CI] = 1.76 (0.32, 9.58), p = 0.51). While not statistically significant, Hispanic ethnicity and non-White race, MRD at the end of initial induction ≥0.1%, and time <36 months from initial diagnosis to relapse appeared to be associated with inferior outcomes. Conclusion AYA patients with ALL have poor outcomes after relapse, and current COG and other cooperative group studies stratify patients age ≥18 years at the time of relapse as high risk. Our data indicate that age at the time of diagnosis is also correlated with post-relapse outcomes, with patients ≥15 years at the time of diagnosis faring poorly after relapse. This suggests that perhaps age at the time of diagnosis should be considered in relapse risk stratification as well, and further work with this data set can focus on not only age at diagnosis, but also age at the time of relapse. Current frontline and relapse treatment studies are now focusing on the incorporation of immunotherapies which may lead to improved outcomes, but our findings highlight the need for continued attention to AYAs as a particularly high-risk population.

Abstract Introduction Minimal residual disease (MRD) detection by flow cytometry is a powerful prognostic indicator in B acute lymphoblastic leukemia (B-ALL), but accurately differentiating normal precursor B cells from leukemic cells requires significant training and experience. This study presents an integrated machine learning pipeline that combines cell- and sample-level assessments to automatically classify B-ALL MRD samples. Materials and Methods List mode data from 772 bone marrow samples analyzed for residual B-ALL using a two-tube 6-color panel were collected from Johns Hopkins Hospital cases under an IRB approved protocol; 386 were classified as MRD positive and 386 negative by an expert (MJB). Manual cell-type annotations from a subset of cases were used to train a hierarchical cell-level classification model, which served as a preprocessing step to exclude non-lymphocytes and mature B cells from the full dataset. This facilitated the training of a sample-level classification model. The integrated model development comprises three main stages: First, data Curation and quality check: Raw flow cytometry data files (.fcs or .lmd), clinical interpretations (MRD-negative, MRD-positive <0.1%, 0.1–1%, and >1%), and expert cell-gating results were collected and reviewed to ensure eligibility for analysis. Second, model training: Approved datasets and annotations were used to train and validate machine learning models. The sample-level model utilized previously established frameworks [1,2], such as GMM-SVM, for classifying MRD status. The hierarchical cell-level model employed algorithms like XGBoost for binary classifications: first distinguishing lymphocytes from non-lymphocytes, then separating immature B cells within the lymphocyte population. The trained cell model was then applied to filter irrelevant cell populations prior to sample-level classification. Third, expert review and validation: Model predictions were reviewed by experts using custom visualization and annotation tools, facilitating an efficient feedback cycle for model improvement. Results The cell-level classification models demonstrated strong performance, achieving an area under the curve (AUC) of 99.96% and an accuracy of 99.19% for distinguishing lymphocytes from non-lymphocytes, and an AUC of 99.94% with an accuracy of 99.43% for differentiating immature B cells from mature lymphocytes. At the sample level, we evaluated MRD classification performance using different cell-filtering strategies. With initial cell filtering, the AUC increased from 76.5% to 82.45%, and accuracy improved from 68.79% to 74.74%. Further enhancement was achieved by removing the downsampling step from the previous framework and excluding mature lymphocytes based on cell-type classification, resulting in an AUC of 91.13% and an accuracy of 83.16%. The best performance was observed with the addition of a channel-wise transformation step and excluding mature lymphocytes for data preprocessing, reaching an AUC of 94.06% and an accuracy of 86.40%. Discussion These results demonstrate how integrated machine learning methods could deliver consistent automated analysis of B-ALL MRD. We next created an interactive review tool for evaluation of discordant cases that will enable iterative model performance improvements, particularly in challenging scenarios with low-level MRD or atypical immunophenotypes.

Abstract Introduction: While advancements in treatment of pediatric B-cell acute lymphoblastic leukemia (B-ALL), through incorporation of novel agents such as blinatumomab, have resulted in remarkable improvements in disease-free survival (Gupta, NEJM 2024), central nervous system (CNS) relapses remain a barrier. As immunotherapies which provide excellent medullary disease control but lack CNS penetration are incorporated into frontline treatment, there is a need to identify patients at risk of CNS relapse in whom deintensification of CNS directed therapy may be inappropriate, and whether these patients differ from those at risk of bone marrow (BM) relapse. Methods: Data from Children's Oncology Group (COG) trials for patients with de novo B-ALL (AALL0331, AALL0232, AALL0932 and AALL1131) were analyzed to examine factors associated with isolated CNS (iCNS), isolated BM (iBM) relapse and any CNS-involved relapse (CNS with any other site). Demographic and clinical characteristics were examined. Cumulative incidence of iCNS, iBM and any CNS-involved relapse was evaluated, treating death, induction failure, subsequent neoplasm and other relapses as competing events. Proportional cause-specific hazards models were used to estimate the association of covariates with the hazard of iCNS, iBM and any CNS relapse. Results: Among the 21,830 patients enrolled across the four trials between 01/2004 and 08/2019 (54.6% male; median age: 4 years [y] [range, 1-30y]), 2,238 (10.3%) patients relapsed (473 [21.2%] iCNS, 1,341 [60.1%] iBM, and 671 [30.0%] any CNS). The 5-year cumulative incidence of relapse among all patients was 9.5%±0.2% and for iCNS, iBM and any CNS-involved relapse was 2.2%±0.1%, 5.6%±0.2% and 3.2%±0.1%, respectively. Median time to relapse was earlier for iCNS compared to iBM: 24.6 vs. 38 months. Multivariable analysis demonstrated that age at diagnosis was not associated with iCNS relapse (10-15y: Hazard ratio [HR]=1.2, 95% confidence interval [CI]=0.9-1.5; ≥16y: HR=1.3, 95%CI=0.9-1.7; ref=1-9y) but was associated with iBM relapse (10-15y: HR=1.6, 95%CI=1.4-1.8; ≥16y: HR=1.7, 95%CI=1.5-2.1; ref=1-9y). Compared to males, females were at a lower hazard of iCNS relapse (HR=0.5, 95%CI=0.4-0.6) but greater hazard of iBM relapse (HR=1.2, 95%CI=1.0-1.3). Compared to Non-Hispanic White patients, Non-Hispanic Black race/ethnicity was associated with greater hazard of both iCNS (HR=1.8, 95%CI=1.3-2.5) and iBM relapse (HR=1.7, 95%CI=1.4-2.1) with similar magnitudes of risk, whereas Hispanic ethnicity was associated with greater hazard of iBM relapse only (HR=1.2, 95%CI=1.0-1.3). The findings for any CNS-involved relapse were similar to iCNS relapse, with the addition of Hispanic ethnicity associated with higher risk (HR=1.2, 95%CI=1.0-1.5). Favorable blast genetics were associated with reduced hazard of both iCNS and iBM relapse. Unfavorable blast genetics were associated with greater hazard of iBM relapse (hypodiploidy: HR=4.6, 95%CI=3.6-5.7; BCR::ABL1: HR=1.9, 95%CI=1.5-2.5; ref=neutral cytogenetics) but not iCNS relapse (hypodiploidy: HR=0.3, 95%CI=0.1-1.1; BCR::ABL1: HR=1.3, 95%CI=0.8-2.2; ref=neutral cytogenetics), with the exception of intrachromosomal amplification of chromosome 21 (iAMP21), which was associated with both (iBM relapse: HR=1.8, 95%CI=1.5-2.3; iCNS relapse: HR=1.8, 95%CI=1.2-2.7). CNS status at diagnosis was associated with greater hazard of iCNS relapse (CNS2: HR=2.2, 95%CI=1.7-2.7; CNS3: HR=2.5, 95%CI=1.5-4.0; ref=CNS1) as well as any CNS relapse but not iBM relapse (CNS2: HR=1.1, 95%CI=0.9-1.2; CNS3: HR=1.1, 95%CI=0.8-1.7; ref=CNS1). EOI MRD was also associated with both, however magnitude of association was substantially higher for iBM relapse (0.01%-0.099%: HR=3.0, 95%CI=2.6-3.5; 0.1%-0.99%: HR=4.2, 95%CI=3.6-4.8; ≥1.0%: HR=4.5, 95%CI=3.7-5.4; ref=<0.01%) than iCNS relapse (0.01%-0.099%: HR=1.5, 95%CI=1.2-2.0; 0.1%-0.99%: HR=1.5, 95%CI=1.1-2.0; ≥1.0%: HR=1.6, 95%CI=1.0-2.3; ref=<0.01%). The findings for any CNS-involved relapse were similar to iCNS. Conclusions: To our knowledge, this is the largest analysis of factors associated with CNS relapse to date. We find that predictors of CNS relapse differ from those of BM relapse, with implications for future trials testing treatment strategies which rely on immunotherapy to de-intensify traditional chemotherapy. Novel and more powerful predictors of CNS relapses and more effective CNS-directed therapies are needed to fully optimize the potential of immunotherapy.

Background:Children with relapsed/refractory acute leukaemias have lower response rates to reinduction and decreased overall survival. Mitoxantrone and clofarabine both have proven efficacy in acute leukaemia. We present our final results utilising this novel reinduction platform as a bridge to haematopoietic stem cell transplantation (HSCT) in children with high-risk pediatric leukaemias. Methods:From 2013 to 2021, patients 0-30.99 yr old with acute lymphoblastic leukaemia (ALL) or acute myelogenous leukaemia (AML) with relapse/refractory disease were given 1 to 3 cycles of clofarabine (escalating doses 20, 30, 35 and 40 mg/m2/day to establish the maximal tolerated dose [MTD]) days 1-5, in combination with mitoxantrone 12 mg/m2/day on days 3-6. The primary objective was to determine the maximal tolerated dose (MTD) or maximal acceptable dose of clofarabine in combination with mitoxantrone 12 mg/m2/day. The key secondary objective was to determine the overall response rate of the combination of mitoxantrone and clofarabine. The protocol was registered with clinicaltrials.gov (NCT01842672). Findings:Forty patients enrolled (18 phase I, 22 phase II) with median age 13 yrs (8 months-23 yrs). Demographics: 22 ALL (10 = induction failure [IF]/minimal residual disease [MRD], 9 = Relapse 1, 3 = Relapse 2), 17 AML (9 = IF/MRD, 6 = Relapse 1, 2 = Relapse 2). During phase I, there were 2 dose-limiting toxicities (DLTs) at dose level 4 requiring de-escalation to dose level 3. The phase I MTD was established at 35 mg/m2/dose clofarabine and continued in phase II. One patient with Burkitt Lymphoma was not included in this efficacy analysis. Thirty-three of 39 (85%) leukaemia patients achieved a complete response (CR). Of these, 88% achieved MRD negativity. Thirty-two of 33 patients went on to allogeneic HSCT. The event free survival/overall survival (EFS/OS) at 1 year was 74% for the entire cohort and 85% for responding/bridging patients at a median follow-up time of >75 months (range 30-120). Interpretation:The MTD of clofarabine in combination with mitoxantrone reinduction therapy was 35 mg/m2/dose x 5 days and was safe, well-tolerated and resulted in an 85% CR rate with 88% MRD negativity and, following HSCT, a 1 yr EFS/OS of 85%. Funding:Pediatric Cancer Research Foundation, Pediatric Cancer Foundation, Children's Cancer Fund, St. Baldrick's Foundation, Carolinas Healthcare, and NCCF subcontract #16252.

Immunophenotyping by flow cytometry is a valuable test providing important information in a timely manner. In clinical laboratories, it is performed using validated antibody panels designed to ensure consistent and accurate results. However, unforeseen situations, such as unique or unusual immunophenotypes, or supply chain issues, may necessitate ad hoc modifications to these panels. This manuscript provides guidance for performing minor modifications, such as substituting or adding one or two antibodies, while maintaining the integrity of the assay. These modifications are intended for rare clinical situations and are not substitutes for the full validation protocols outlined in CLSI H62. An example of this would be a patient with a rare, but not uncommon, situation in which a B cell lymphoma lacks expression of CD19, CD20, and surface light chains, such that the lineage of the neoplastic cells cannot be determined without a straightforward addition or substitution of another marker into a laboratory's available panel. The recommendations and best practices herein aim to optimize patient care by allowing laboratories to adapt to unique clinical scenarios without compromising assay performance and are not a way to permanently modify the assay. Key considerations include assessing the impact on fluorescence compensation, antibody binding, assay sensitivity, and overall assay performance. The manuscript provides limitations for the extent of modifications, examples, and troubleshooting strategies to ensure reliable results when ad hoc changes are made. Proper documentation with review and approval by laboratory medical directors is recommended to mitigate risks associated with these modifications.

Context.— Generative artificial intelligence (AI) technologies are rapidly transforming numerous fields, including pathology, and hold significant potential to revolutionize educational approaches. Objective.— To explore the application of generative AI, particularly large language models and multimodal tools, for enhancing pathology education. We describe their potential to create personalized learning experiences, streamline content development, expand access to educational resources, and support both learners and educators throughout the training and practice continuum. Data Sources.— We draw on insights from existing literature on AI in education and the collective expertise of the coauthors within this rapidly evolving field. Case studies highlight practical applications of large language models, demonstrating both the potential benefits and unique challenges associated with implementing these technologies in pathology education. Conclusions.— Generative AI presents a powerful tool kit for enriching pathology education, offering opportunities for greater engagement, accessibility, and personalization. Careful consideration of ethical implications, potential risks, and appropriate mitigation strategies is essential for the responsible and effective integration of these technologies. Future success lies in fostering collaborative development between AI experts and medical educators, prioritizing ongoing human oversight and transparency to ensure that generative AI augments, rather than supplants, the vital role of educators in pathology training and practice.

ABSTRACT:Pediatric Hispanic and Black patients with newly diagnosed B-cell acute lymphoblastic leukemia (B-ALL) experience worse overall survival (OS). We hypothesized that differential outcomes by race and ethnicity following relapse may contribute to disparities. We examined 2053 patients with ALL enrolled in frontline Children's Oncology Group trials from 1996 to 2014 who relapsed. We assessed the association of race and ethnicity, disease characteristics, and socioeconomic status with relapse survival predictors and postrelapse OS. For noninfant B-ALL, postrelapse OS (P = .002) and disease-related prognosticators such as time to relapse (P = .0002) differed by race and ethnicity. After adjusting for disease and patient characteristics, the OS association with overall race and ethnicity was attenuated, and lost statistical significance; Hispanic ethnicity specifically remained associated with worse OS (hazard ratio [HR], 1.19; 95% confidence interval [CI], 1.01-1.41). Patients from highest annual median household income ZIP codes (>$85 000, approximately the highest quartile of patients) had better 5-year OS than those from the lowest (<$50 000; HR, 0.79; 95% CI, 0.63-0.99). Non-Hispanic Black and Hispanic patients more commonly lived in lower-income ZIP codes. For T-cell ALL, race, ethnicity, and socioeconomic status were not associated with OS. Worse postrelapse outcomes among racial and ethnic minority patients are largely driven by the prevalence of adverse disease-related factors at time of relapse, with a persistent disparity observed in Hispanic patients. The greatest impact in decreasing racial and ethnic B-ALL outcome disparities may be achieved by targeting frontline treatment interventions to address increased relapse among Black and Hispanic patients, and by developing and enabling equitable access to effective relapse treatments such as novel immunotherapies.

Identifying aberrant T lymphocytes in peripheral blood is essential for diagnosing Sezary syndrome (SS) and is a prognostic indicator in mycosis fungoides (MF). Flow cytometry using a T-cell receptor constant beta-1 chain (TRBC1)-targeting antibody provides a refined approach for detecting T-cell clonality. We evaluated the performance of the TRBC1 antibody assay (TRBC1-aa) in 164 patients, compared to standard flow cytometry methods for assessing T-cell aberrancy, demonstrating 92.3% sensitivity and 83.5% specificity. TRBC1-aa accurately excluded clonality in 100% of benign inflammatory dermatoses and improved the detection of residual blood involvement earlier than standard flow cytometry in 66.7% of SS and 25.0% of advanced-stage MF patients on systemic therapy.

Introduction End of induction (EOI) positive minimal residual disease (MRD) and unfavorable sentinel genetic lesions predict poor outcomes for patients with B-cell acute lymphoblastic leukemia (B-ALL). We examined outcomes for patients with newly diagnosed NCI standard risk (SR) B-ALL enrolled on AALL0932 who were stratified as high risk (HR) or very high risk (VHR) post-induction based on MRD and/or unfavorable genetic features. Patients and methods Children (n = 9277) with newly diagnosed NCI SR B-ALL (age 1-9.99 years and initial white blood cell count <50,000/microliter) were enrolled on AALL0932 (2010-2018). They received a 3-drug, 4-week induction and were subsequently classified into different risk groups for post-induction therapy. We examined disease-free (DFS) and overall survival (OS) of 2295 HR or VHR patients. HR patients had favorable cytogenetics (ETV6::RUNX1 or double trisomies of chromosomes 4+10) and EOI MRD ≥0.01% or no favorable cytogenetics with day 8 peripheral blood MRD ≥1% and EOI MRD < 0.01%. VHR patients had unfavorable cytogenetics (intrachromosomal amplification of chromosome 21, KMT2A-rearrangement, hypodiploidy), and/or induction failure (M3) and/or no favorable cytogenetics and EOI MRD ≥0.01%. These HR/VHR patients were ineligible to continue AALL0932 therapy but were eligible to enroll on the COG AALL1131 trial (2012- 2019) post-induction. Intensified post-induction AALL1131 therapy utilized an augmented BFM therapy backbone with randomizations testing different intrathecal therapy strategies (HR) and consolidation chemotherapy regimens (VHR). Those not enrolled on AALL1131 were treated off protocol therapy at physician discretion. Patients with Down syndrome or BCR::ABL1 were not eligible for post-induction enrollment on AALL1131. Results At AALL0932 EOI, 1262 (13.7%) patients were classified as HR and 1033 (11.2%) patients as VHR. Most patients had CNS1 status at diagnosis (HR patients 1153, 91.4%; and VHR patients 908, 87.9%) and the remainder were CNS2 (CNS3 patients were ineligible for AALL0932). 708 HR patients (56.1%) had favorable cytogenetics, 311 VHR patients (30.1%) had unfavorable cytogenetics, and the remainder in both groups had neutral cytogenetics. EOI MRD ≥ 0.01% was detected in 705 (55.9%) HR and 820 (79.4%) VHR patients. Overall, these HR and VHR NCI SR B-ALL patients had 7-year post-induction DFS and OS rates (±SE) of 85.3 ± 1.0% and 94.1 ± 0.6%. In outcome analyses stratified by risk group, no differences in outcomes were seen for the HR or VHR patients enrolled or not on AALL1131. For HR patients treated on (693, 54.9%) vs not on AALL1131: 7-year DFS 92.1 ± 1.3% vs. 91.2 ± 1.5% (P = 0.70) and 7-year OS 97.8 ± 0.7% vs. 97.1 ± 0.9% (P = 0.58). For VHR patients enrolled on (217, 21.0%) vs not on AALL1131: 7-year DFS 76.1 ± 3.9% vs. 77.4 ± 1.9% (P = 0.95) and 7-year OS 91.9 ± 2.5% vs. 89.3 ± 1.4% (P = 0.32). Temporary suspensions of AALL1131, in part due to unexpected toxicities on the VHR arm, contributed to non-enrollment of AALL0932 HR/VHR patients for AALL1131 post-induction therapy. Conclusion Following treatment with a 3-drug induction, clinical outcomes for children with NCI SR B-ALL that were EOI MRD-positive or had unfavorable genetics, are excellent with intensified, high risk augmented BFM post-induction therapy. We anticipate that outcomes for these patients will improve further with addition of blinatumomab in ongoing and future clinical trials.

10014 Background: Minimal (measurable) residual disease (MRD) at end of induction (EOI) therapy is a strong predictor of outcome in pediatric B-ALL. Currently, EOI MRD is assessed in bone marrow (BM). We hypothesized that the highly sensitive assay, high-throughput sequencing (HTS) of immunoglobulin loci, can effectively monitor MRD in peripheral blood (PB) and may provide a less invasive way to track therapy response. Methods: We conducted HTS MRD on paired EOI BM and PB samples from 808 NCI standard risk (SR) pediatric B-ALL patients enrolled on Children’s Oncology Group study AALL1731 (NCT03914625). We determined the correlation between BM and PB HTS MRD via Spearman’s rank correlations. We calculated the BM/PB MRD ratio and compared these by subgroup using Kruskal-Wallis tests. We defined subgroups by cytogenetics (cyto) ( ETV6::RUNX1, double trisomies of chromosome 4 and 10 (DT), Unfavorable (hypodiploidy, iAMP21, or KMT2A-rearranged), or Neutral (lacking ETV6::RUNX1, DT, or unfavorable)), and risk group (SR-average (AVG) and SR-High). Flow cytometry-defined EOI BM MRD was < 0.01% for all SR-AVG patients (N = 623) and ≥0.01% for selected SR-High patients (N = 185). Results: There was strong correlation between PB and BM HTS MRD with an overall correlation coefficient of 0.75 (P < 0.001). Correlation was similar by cytogenetics: ETV6::RUNX1, 0.69 (N = 63; P < 0.001), DT, 0.75 (N = 147; P < 0.001), Neutral, 0.74 (N = 580; P < 0.001), and Unfavorable, 0.66(N = 18; P = 0.003). For risk groups, correlation for SR-AVG was 0.67 (p < 0.001) and SR-High, 0.64 (p < 0.001). Of the 591 SR-AVG patients with detectable BM HTS MRD, PB HTS MRD was detectable in 474 (80.2%), undetectable in 94 (15.9%) and indeterminate (no leukemic cell detected and < 500,000 total cells in sample) in 23 (3.9%). Among 182 SR-High patients with detectable BM HTS MRD, 175 (96.2%) had detectable PB HTS MRD. Disease burden was higher in the BM than PB with a significantly higher BM/PB ratio in SR-High compared to SR-AVG patients (median 16.5 vs 2.6, P < 0.001). The median BM/PB ratio also varied by cytogenetics with those with Unfavorable cyto having the highest ratio (15.3 vs 6.3 in DT, 3.8 in ETV6::RUNX1, 3.1 in Neutral; P = 0.013). Conclusions: This is the largest analysis of paired B-ALL BM/PB HTS MRD to date. We show strong correlation between PB and BM across risk and cytogenetic groups. The ratio of BM/PB MRD varied and was highest among patients with Unfavorable cyto suggesting BM tropism. Importantly, PB MRD was detectable in nearly all patients with flow EOI BM MRD ≥0.01%, a threshold warranting therapy intensification. However, most patients with EOI BM flow MRD < 0.01% also had detectable PB HTS MRD. Thus, PB HTS MRD may provide a useful adjunct for screening and clinical management of B-ALL patients. Defining a PB HTS MRD threshold useful for risk stratification will require correlation with outcome.

10015 Background: While cure rates for childhood acute lymphoblastic leukemia (ALL) exceed 90%, half of relapses arise in those originally classified with standard risk (SR) disease. Methods: We performed genome/transcriptome sequencing of diagnostic and germline samples of children with SR (n=1381) B-ALL or high-risk (HR) B-ALL with favorable cytogenetics ( ETV6: RUNX1 or double trisomy (DT) of chromosomes (chr) 4+10; n=115) to identify predictors of relapse. We used a case-control study to analyze 439 patients who relapsed and 1057 who remained in complete remission for > 5 years. Results: Genomic subtype was associated with relapse. Unbalanced ETV6:RUNX1 translocations were more common than balanced in relapse patients (OR=2.01, CI=1.25-3.20, P=0.002). Conversely, balanced TCF3:PBX1 translocations were more often associated with relapse than unbalanced in TCF3:PBX1 ALL (OR=0.11, CI=0.01-0.50, P=0.003). A striking finding was the high relapse rate in PAX5 altered ALL (57 of 116 cases (49%); OR=3.29, CI=2.16-5.01, P=3.49x10 -8 ). The nature of the heterogeneous PAX5 driver alterations of this subtype influenced relapse risk, with internal PAX5 amplifications and biallelic PAX5 alterations associated with the highest risk. Specific chr gains influenced outcome in hyperdiploid ALL, with gain of chr 10 and disomy of chr 7 associated with favorable outcome (OR=0.27, CI=0.17-0.42, P=8.02x10 -10 , St Jude Children’s Research Hospital (SJCRH) validation cohort: OR=0.22, CI=0.05-0.80, P=0.009), while disomy of chr 10 and 17 and gain of chr 6 were enriched in patients that relapsed (OR=7.16, CI=2.63-21.51, P=2.19x10 -5 ; SJCRH cohort: OR=21.32, CI=3.62-119.30, P=0.0004). Genomic alterations were also associated with relapse in a subtype-dependent manner, including alterations of INO80 in ETV6:RUNX1, IKZF1 and CREBBP in hyperdiploid, and FHIT in Ph-like ALL. Conclusions: Genetic subtype, aneuploidy patterns, and secondary genomic alterations influence risk of relapse in children otherwise classified with SR ALL, or HR ALL with favorable genetics. Comprehensive genomic analysis is required for optimal risk stratification and treatment allocation, and particularly to study reduction of therapy in the lowest risk patients. [Table: see text]

Intensification of chemotherapy for pediatric B-cell acute lymphoblastic leukemia (B-ALL) has improved survival. In recently completed Children's Oncology Group (COG) frontline trials for B-ALL, a second interim maintenance (IM2) phase including escalating dose methotrexate, vincristine, ±pegaspargase was utilized to intensify therapy. On AALL1131 for high risk B-ALL (HR-ALL), very high risk (VHR) patients received IM2. On AALL0932 for standard risk B-ALL (SR-ALL), patients were randomized to steroid/vincristine pulses every 4 (q4weeks) or 12 weeks (q12weeks); all patients received IM2 to maintain treatment intensity in the context of dose-reduced Maintenance therapy. However, the efficacy of IM2 has never been established in either HR-ALL or SR-ALL. Thus, we performed secondary analyses comparing outcomes for matched cohorts treated ±IM2 in sequential COG trials in HR-ALL treated on AALL0232 (no IM2) and AALL1131 (+IM2) and in SR-ALL treated on AALL0331 (no IM2) and AALL0932 (+IM2). The primary objective was to compare disease-free (DFS) and overall survival (OS) in patients receiving identical systemic COG ALL therapy ±IM2. Secondary objectives were to examine the efficacy of IM2 in SR-ALL with the current standard q12week steroid/vincristine pulses and within cytogenetic subsets. Eligibility criteria for these analyses were designed to match treatment arms between trials. For HR-ALL, inclusion criteria were age ≥13 years old and end of induction (EOI) minimal residual disease (MRD) <0.01%, based on eligibility for the AALL1131 VHR arm. For SR-ALL, patients with neutral or favorable cytogenetics (ETV6::RUNX1 or double trisomies 4 and 10) and EOI MRD <0.01% were included. Patients with BCR::ABL1, KMT2A rearrangement, hypodiploidy, or EOI MRD ≥0.01% were excluded due to treatment variation among trials. Statistical analyses compared DFS and OS via log-rank test and were reported at five years with standard errors. Cox multivariable models analyzed hazard ratios (HR) adjusted for presenting features (demographics, white blood cell count, cytogenetics, central nervous system status). In HR-ALL, 249 patients from AALL0232 and 65 patients from AALL1131 VHR were eligible. For AALL1131 VHR versus AALL0232, the addition of IM2 was not associated with a difference in DFS (85%±4.7 vs 85%±2.5, p=0.91) or OS (90%±4.0 vs 90%±2.1, p=0.63). In SR-ALL patients receiving q4week pulses in Maintenance, 1,317 patients were eligible from AALL0331 and 1,186 patients from AALL0932. As compared to AALL0331, addition of IM2 in AALL0932 was associated with significantly higher DFS (94%±0.7 vs 91%±0.8, p= 0.0024) and OS (99%±0.4 vs 97%±0.5, p<0.0001). In the Cox multivariable model comparing SR-ALL patients receiving q4week pulses, IM2 significantly decreased risk for an event (HR 0.67, p=0.0048). In SR-ALL patients enrolled in AALL0932 who were randomized to receive q12week pulses (n=1,178), addition of IM2 was associated with higher DFS and OS versus AALL0331 (95%±0.7 vs 91%±0.8, p<0.0001 and 99%±0.3 vs 97%±0.5, p<0.0001, respectively). However, adjusted Cox models demonstrated the decreased event risk was largely in those with neutral cytogenetics (HR=0.62, p=0.0179), with OS (HR=0.58, p=0.0783). No difference in DFS/OS was noted in SR-ALL with favorable cytogenetics. Within the limitations of unplanned subset analyses, we determined that IM2 may be able to be safely omitted in MRD-negative patients with HR-ALL or with SR-ALL with favorable cytogenetics and that IM2 may provide a survival advantage for patients with SR-ALL, neutral cytogenetics, and MRD <0.01%. Whether IM2 is necessary in patients with higher risk B-ALL due to adverse cytogenetics or persistent MRD was not addressed in these analyses.

Current strategies to treat pediatric acute lymphoblastic leukemia rely on risk stratification algorithms using categorical data. We investigated whether using continuous variables assigned different weights would improve risk stratification. We developed and validated a multivariable Cox model for relapse-free survival (RFS) using information from 21199 patients. We constructed risk groups by identifying cutoffs of the COG Prognostic Index (PICOG) that maximized discrimination of the predictive model. Patients with higher PICOG have higher predicted relapse risk. The PICOG reliably discriminates patients with low vs. high relapse risk. For those with moderate relapse risk using current COG risk classification, the PICOG identifies subgroups with varying 5-year RFS. Among current COG standard-risk average patients, PICOG identifies low and intermediate risk groups with 96% and 90% RFS, respectively. Similarly, amongst current COG high-risk patients, PICOG identifies four groups ranging from 96% to 66% RFS, providing additional discrimination for future treatment stratification. When coupled with traditional algorithms, the novel PICOG can more accurately risk stratify patients, identifying groups with better outcomes who may benefit from less intensive therapy, and those who have high relapse risk needing innovative approaches for cure.

Introduction: Pediatric Hispanic and non-Hispanic (NH) Black patients with newly diagnosed acute lymphoblastic leukemia (ALL) experience worse overall survival (OS) compared to NH White patients (Gupta, Lancet Haematol 2023). We hypothesized that differential outcomes by race and ethnicity (R/E) following relapse may contribute to these disparities. Methods: We examined children and young adults with ALL enrolled on 12 frontline Children's Oncology Group (COG) trials from 1996-2014 who subsequently relapsed. We assessed association of R/E with relapse survival predictors including time-to-relapse, relapse site, ALL risk group, and cytogenetics. We examined association of R/E with post-relapse 5-year OS and assessed the effect of disease characteristics and socioeconomic status. Socioeconomic status was evaluated using US ZIP code-based median household yearly income from the 2020 Census and US insurance status. Analyses of OS used univariate (crude) and multivariable (adjusted) Cox regression models. Results: Among 16,115 patients with ALL treated on frontline COG trials, 2,053 (1,147 NH White, 492 Hispanic, 145 NH Black, 65 NH Asian, 184 other/unknown) relapsed and formed our primary cohort. For B-ALL, post-relapse OS differed by R/E (p=0.002), and specifically, Hispanic patients had worse survival (46.2±2.4%, crude hazard ratio [cHR] 1.39, 95% confidence interval [CI] 1.19-1.63) compared to NH White patients (55.7±1.7%). Disease-related prognosticators, including time-to relapse (p=0.0002), white blood cell count at initial diagnosis (p=0.03), and presence of central nervous system disease at initial diagnosis (p=0.03), varied by R/E. The overall association of OS with R/E was substantially attenuated when adjusted for disease-related prognosticators and ZIP-based income (p=0.53). However, Hispanic ethnicity still associated with worse OS but by a lower magnitude (adjusted HR [aHR] 1.19, 95% CI 1.01-1.41). Post-relapse OS in B-ALL also differed based on ZIP-based income on univariate (p=0.008) but not multivariable analysis. Focusing on the highest and lowest ZIP-based income, patients with ZIP-based income >$85,000 had better OS (56.4±2.9%) compared to those with <$50,000 (48.4±2.7%, aHR 0.77, 95% CI 0.61-0.96). NH Black patients most commonly had based income <$50,000 (46.8%), followed by Hispanic (30.6%), NH White (15.8%), and NH Asian (6.3%) patients. US insurance status also differed based on R/E (univariate p<0.0001), with NH Black patients most commonly being Medicaid insured (42.0%), followed by Hispanic (41.1%), NH White (18.3%), and NH Asian (18.0%) patients. OS was not associated with US insurance status. For T-ALL, neither R/E nor socioeconomic status were associated with OS. For infant ALL, R/E was associated with OS on multivariable analysis (p=0.03). Conclusions: In this large retrospective cohort of patients with relapsed ALL, we found that although R/E were associated with post relapse-OS, multivariable analyses suggest that inferior post-relapse outcomes among Hispanic and NH Black patients are in large part driven by a higher prevalence of adverse disease-related risk factors present at the time of relapse. The persistent disparity observed in Hispanic patients may be related to unmeasured underlying disease biology (such as the higher prevalence CRLF2-rearranged/Philadelphia chromosome-like B-ALL in Hispanic patients). Differing toxicities, supportive care, and adherence to treatment in the frontline setting for R/E minorities may also contribute to the differences in time-to-relapse. Our findings suggest that while post-relapse interventions are needed, the greatest impact in decreasing R/E-based ALL outcome disparities will come through identifying and targeting mechanisms in the frontline treatment setting that contribute to increased high-risk relapse among Hispanic and NH Black patients.

Limited prognostic factors have been associated with overall survival (OS) post-relapse in childhood Acute Lymphoblastic Leukemia (ALL). Patients enrolled on 12 Children’s Oncology Group frontline ALL trials (1996–2014) were analyzed to assess for additional prognostic factors associated with OS post-relapse. Among 16,115 patients, 2053 (12.7%) relapsed. Relapse rates were similar for B-ALL (12.5%) and T-ALL (11.2%) while higher for infants (34.2%). Approximately 50% of B-ALL relapses occurred late (≥36 months) and 72.5% involved the marrow. Conversely, 64.8% of T-ALL relapses occurred early (<18 months) and 47.1% involved the central nervous system. The 5-year OS post-relapse for the entire cohort was 48.9 ± 1.2%; B-ALL:52.5 ± 1.3%, T-ALL:35.5 ± 3.3%, and infant ALL:21.5 ± 3.9%. OS varied by early, intermediate and late time-to-relapse; 25.8 ± 2.4%, 49.5 ± 2.2%, and 66.4 ± 1.8% respectively for B-ALL and 29.8 ± 3.9%, 33.3 ± 7.6%, 58 ± 9.8% for T-ALL. Patients with ETV6::RUNX1 or Trisomy 4 + 10 had median time-to-relapse of 43 months and higher OS post-relapse 74.4 ± 3.1% and 70.2 ± 3.6%, respectively. Patients with hypodiploidy, KMT2A-rearrangement, and TCF3::PBX1 had short median time-to-relapse (12.5-18 months) and poor OS post-relapse (14.2 ± 6.1%, 31.9 ± 7.7%, 36.8 ± 6.6%). Site-of-relapse varied by cytogenetic subtype. This large dataset provided the opportunity to identify risk factors for OS post-relapse to inform trial design and highlight populations with dismal outcomes post-relapse.

BACKGROUND:Previous studies have identified racial and ethnic disparities in childhood acute lymphocytic leukaemia survival. We aimed to establish whether disparities persist in contemporaneous cohorts and, if present, are attributable to differences in leukaemia biology or insurance status. METHODS:Patients with newly diagnosed acute lymphocytic leukaemia in inpatient and outpatient centres in the USA, Canada, Australia, and New Zealand, aged 0-30 years, who had race or ethnicity data available, enrolled on eight completed Children's Oncology Group trials (NCT00103285, NCT00075725, NCT00408005, NCT01190930, NCT02883049, NCT02112916, NCT02828358, and NCT00557193) were included in this secondary analysis. Race and ethnicity were categorised as non-Hispanic White, Hispanic, non-Hispanic Black, non-Hispanic Asian, and non-Hispanic other. Event-free survival and overall survival were compared across race and ethnicity groups. The relative contribution of clinical and biological disease prognosticators and insurance status was examined through multivariable regression models, both among the entire cohort and among those with B-cell lineage versus T-cell lineage disease. FINDINGS:Between Jan 1, 2004, and Dec 31, 2019, 24 979 eligible children, adolescents, and young adults with acute lymphocytic leukaemia were enrolled, of which 21 152 had race or ethnicity data available. 11 849 (56·0%) were male and 9303 (44·0%) were female. Non-Hispanic White patients comprised the largest racial or ethnic group (13 872 [65·6%]), followed by Hispanic patients (4354 [20·6%]), non-Hispanic Black patients (1517 [7·2%]), non-Hispanic Asian (n=1071 [5·1%]), and non-Hispanic other (n=338 [1·6%]). 5-year event-free survival was 87·4% (95% CI 86·7-88·0%) among non-Hispanic White patients compared with 82·8% (81·4-84·1%; hazard ratio [HR] 1·37, 95% CI 1·26-1·49; p<0·0001) among Hispanic patients and 81·8% (79·3-84·0; HR 1·45, 1·28-1·65; p<0·0001) among non-Hispanic Black patients. Non-hispanic Asian patients had a 5-year event-free survival of 88·1% (95% CI 85·5-90·3%) and non-Hispanic other patients had a survival of 82·8% (76·4-87·6%). Inferior event-free survival among Hispanic patients was substantially attenuated by disease prognosticators and insurance status (HR decreased from 1·37 [1·26-1·49; p<0·0001] to 1·11 [1·00-1·22; p=0·045]). The increased risk among non-Hispanic Black patients was minimally attenuated (HR 1·45 [1·28-1·65; p<0·0001] to 1·32 [1·14-1·52; p<0·0001]). 5-year overall survival was 93·6% (91·5-95·1%) in non-Hispanic Asian patients, 93·3% (92·8-93·7%) in non-Hispanic White patients, 89·9% (88·7-90·9%) in Hispanic, 89·7% (87·6-91·4%) in non-Hispanic Black patients, 88·9% (83·2-92·7%) in non-Hispanic other patients. Disparities in overall survival were wider than event-free survival (eg, among non-Hispanic other patients, the HR for event-free survival was 1·43 [1·10-1·85] compared with 1·74 [1·27-2·40] for overall survival). Disparities were restricted to patients with B-cell acute lymphocytic leukaemia, no differences in event-free survival or overall survival were seen in the T-cell acute lymphocytic leukaemia group. INTERPRETATION:Substantial disparities in outcome for B-cell acute lymphocytic leukaemia persist by race and ethnicity, but are not observed in T-cell acute lymphocytic leukaemia. Future studies of relapsed patients, access to and quality of care, and other potential aspects of structural racism are warranted to inform interventions aimed at dismantling racial and ethnic disparities. FUNDING:National Cancer Institute and St Baldrick's Foundation.

Background: Approximately 2-3% of B-ALL patients have Philadelphia chromosome like (Ph-like) B-ALL with an ABL-class fusion ( ABL1, ABL2, PDGFRB, or CSF1R fusions), but lacking BCR::ABL1. These patients are predicted to be sensitive to ABL-class tyrosine kinase inhibitors, such as imatinib or dasatinib. Ph-like B-ALL is associated with male sex, older age, higher initial white blood cell (WBC) count, elevated end of induction (EOI) minimal residual disease (MRD), and poor outcome. The AALL1131 Dasatinib arm was designed to evaluate response to therapy following induction of patients with ABL-class fusion B-ALL, when given dasatinib continuously on a modified Berlin-Frankfurt-Münster (MBFM) backbone. Methods: Between February 2012 and March 2019, AALL1131 enrolled patients 1-30 years with newly diagnosed high risk (HR) B-ALL. AALL1131 was amended in August 2016, to include the Dasatinib arm for patients with ABL-class fusions involving ABL1, ABL2, PDGFRB, and CSF1R. Patients with Down Syndrome were not eligible for the Dasatinib arm. ABL-class fusion B-ALL patients with a predicted TKI-sensitive mutation were screened by low density array (LDA) PCR and confirmed by additional molecular testing. Following 4-drug induction, patients with HR B-ALL and ABL-class fusions received MBFM with dasatinib (60 mg/m 2, maximum 140 mg) daily from start of consolidation through end of maintenance. Dasatinib was held only for toxicity. Results: Twenty-two evaluable patients with HR B-ALL were non-randomly assigned to the Dasatinib arm. ABL-class 3' partners included ABL1 (n=4), ABL2 (n=4), PDGFRB (n=12), and CSF1R (n=2). Compared to all patients on AALL1131, patients treated on the Dasatinib arm were older (median age at diagnosis 14 versus 10 years, p=0.008), more often male (77% versus 56%, p=0.046), and had a trend towards an increased initial WBC (median WBC at diagnosis 44,000/µL versus 19,000/µL, p=0.086). Patients with ABL-class fusions had EOI MRD > 0.01% in 20 of 22 patients and available end of consolidation MRD >0.01% in 6 of 10 patients. Only 5 of 22 (22.7%) patients completed prescribed protocol therapy. Reasons for therapy discontinuation included induction failure (n=2), relapse (n=1), alternate therapy (n=1), determined to be in the patient's best interest (n=10), death (n=1), unknown (n=2). Compared to all patients on AALL1131, four-year disease-free survival was 52.5+18.1% versus 86.8+0.7% (p<0.0001) and overall survival 79.4+13.6% versus 89.2+0.4% (p<0.0001). Dasatinib was well tolerated with no unexpected treatment related toxicities. Conclusion: Patients with ABL-class fusions were more likely male, EOI MRD+, and had poorer outcomes. Seventy-seven percent of patients enrolled on the Dasatinib arm did not complete prescribed therapy. While dasatinib was well tolerated, treatment failures occurred early, indicating alternate strategies are needed.

The early thymic precursor (ETP) immunophenotype was previously reported to confer poor outcome in T-cell acute lymphoblastic leukemia (T-ALL). Between 2009 and 2014, 1256 newly diagnosed children and young adults enrolled in Children's Oncology Group (COG) AALL0434 were assessed for ETP status and minimal residual disease (MRD) using flow cytometry at a central reference laboratory. The subject phenotypes were categorized as ETP (n = 145; 11.5%), near-ETP (n = 209; 16.7%), or non-ETP (n = 902; 71.8%). Despite higher rates of induction failure for ETP (6.2%) and near-ETP (6.2%) than non-ETP (1.2%; P < .0001), all 3 groups showed excellent 5-year event-free survival (EFS) and overall survival (OS): ETP (80.4% +/- 3.9% and 86.8 +/- 3.4%, respectively), near-ETP (81.1% +/- 3.3% and 89.6% +/- 2.6%, respectively), and non-ETP (85.3% +/- 1.4% and 90.0% +/- 1.2%, respectively; P = .1679 and P = .3297, respectively). There was no difference in EFS or OS for subjects with a day-29 MRD <0.01% vs 0.01% to 0.1%. However, day-29 MRD >= 0.1% was associated with inferior EFS and OS for patients with near-ETP and non-ETP, but not for those with ETP. For subjects with day-29 MRD >= 1%, end-consolidation MRD >= 0.01% was a striking predictor of inferior EFS (80.9% +/- 4.1% vs 52.4% +/- 8.1%, respectively; P = .0001). When considered as a single variable, subjects with all 3 T-ALL phenotypes had similar outcomes and subjects with persistent postinduction disease had inferior outcomes, regardless of their ETP phenotype.