Citations

Abstract Objectives Traditional laboratory regulatory frameworks provide robust guidance for conventional clinical testing but lack requirements for data documentation for artificial intelligence and machine learning (AI/ML) solutions. This gap creates significant risks since AI solutions directly inherit patterns, biases, and limitations from their training data. Here, we describe the development and implementation of a comprehensive data documentation checklist with accompanying data lineage for use within an AI lifecycle framework for clinical laboratories. Methods Building on the previously established Clinical AI Readiness Evaluator (CARE) framework, we developed a comprehensive data checklist and data lineage methodology through a multiphase process, including (1) comprehensive review of existing AI/ML data documentation frameworks, (2) focused meetings with 3 institutional AI operations teams, and (3) 3 rounds of iterative refinement by our multidisciplinary team. The checklist’s effectiveness was then assessed using 3 diverse AI/ML projects moving through the CARE framework. Results The CARE Data Checklist and Data Lineage provide a structured approach to documenting critical aspects of datasets used in AI/ML projects, including data composition, demographics, collection methods, labeling processes, usage constraints, maintenance requirements, and a data readiness assessment. The checklist addresses unique data-centric challenges of AI/ML applications, facilitating transparency, reproducibility, and regulatory compliance. Conclusions The CARE Data Checklist and Data Lineage serve as both a technical guide and a communication tool bridging gaps between technical and clinical stakeholders. By working on these documents early in the AI lifecycle, laboratories can anticipate and address data-related challenges, ultimately saving time, optimizing resources, and improving the reliability of AI-augmented laboratory solutions.

Abstract Deep learning models enable the prediction of clinical endpoints from whole-slide images (WSIs), but many such models function as “black boxes”, lacking transparency about whether and which histomorphological patterns drive their predictions, hindering interpretability and clinical adoption. Here we propose a human-in-the-loop explanation framework, MorphoXAI, which provides both local and global interpretability for deep learning models by incorporating human-expert interpretations. At the global level, it reveals the histomorphological patterns on which the model consistently relies to distinguish between classes of WSIs, as well as the patterns associated with confusion between classes. At the local level, it indicates which of these patterns are used in the prediction of an individual WSI and which regions within the slide correspond to such patterns. We validated our method on a deep learning model trained for ovarian tumor histologic subtype prediction. The results show that our framework generates explanations that accurately reflect the histomorphology underlying the model’s predictions at both global and local levels. For interpretability and clinical utility in diagnostic contexts, human evaluation results showed that our explanations were easy to interpret, rich in diagnostic features, and directly helpful for diagnostic decision-making, thereby enhancing pathologist-AI collaboration. Our work highlights that unifying global and local explanations and grounding them in expert-interpreted morphology enhances the interpretability and verifiability of deep learning models, thereby facilitating the transparent deployment of such models in clinical practice.

Abstract Introduction:A comprehensive integration of significantly mutated genes and pathways with molecular subgroups is lacking in adult B-ALL. Furthermore, the genomic drivers underlying each subgroup remain to be identified.We aimed to perform genomic analysis on a large cohort of patients with adult B-ALL, and to correlate genomic markers with achievement of MRD negativity enabling blinatumomab randomization, and with subsequent response to blinatumomab. Methods:For genomic analysis, we studied 569 adults with newly diagnosed B-ALL registered for initial screening on the ECOG-ACRIN-led E1910 trial (NCT02003222), which evaluated the addition of blinatumomab to standard consolidation chemotherapy. The median age was 52 yrs (range 30 to 71 yrs), with 52% males. Analysis was performed on tumor and matched-normal samples using whole transcriptome sequencing (RNA-seq; tumor only; n=569), whole exome sequencing (n=490), whole genome sequencing (n=131), and single nucleotide polymorphism array (n=445). B-ALL cases were classified into 22 molecular groups. Outcome analysis was limited to patients enrolled and treated on E1910 (n=319). Results: Driver genes (n=268) were identified by the mutation-significance detection tool dNdScv or by the presence of pathogenic variants in known cancer genes, with NUP188 identified as a novel driver gene. Recurrently mutated pathways included: B-cell development (46%), cell cycle (44%), epigenetic regulation (30%), other transcriptional regulation (27%), Ras signaling (25%), RNA machinery (17%) and JAK-STAT signaling (8%). A high frequency of high-risk subtypes was observed including BCR::ABL1 (20%), BCR::ABL1-like (18%), low hypodiploid (14%) and KMT2A (12%). Using tSNE analysis, we identified a new cluster of cases (n=20, 3.5%) lacking a known subgroup driver, with overexpression of CEBPA (n=13) or CEBPB (n=7), distinct from CEBPE/ZEB2, termed “CEBP-altered” ALL (CEBPalt). Five of 13 cases with high CEBPA expression harbored an IGH::CEBPA rearrangement and one harbored a RXRA::CEBPA enhancer hijacking alteration. Of the 7 cases with high CEBPB expression, 5 harbored IGH::CEBPB, and two of these had concomitant clonal BCR::ABL1 fusions. To identify additional mechanisms of CEBP deregulation, in situ Hi-C coupled to H3K27 acetylation immunoprecipitation (HiChIP) was performed on 5 cases. Four cases showed evidence of de novo CEBPA enhancer activity. In two cases, we identified novel insertions (12 and 34nt) ~2kb downstream of CEBPA. A novel translocation and enhancer hijacking event between LINC00426 and the 5' UTR of CEBPB was identified in the remaining case. Overall, we confirmed genomic alterations of CEBPA or CEBPB in 16 of 20 cases with available material. Thus, we have identified a new molecular subgroup of adult B-ALL characterized by genomic alterations that drive enhancer hijacking and oncogenic deregulation of CEBPA/CEBPB. Several subgroups were enriched in patients that failed induction chemotherapy (n=62) or were MRD-positive (n=63) compared to those that achieved MRD-negative status (n=194): BCR::ABL1-like (30 vs 14%, p<0.001), KMT2A (18 vs 9%, p=0.03) and BCL2/MYC (5 vs 1%, p=0.06). Conversely, the following subgroups were enriched in patients that achieved MRD-negativity: PAX5alt (6 vs 18%, p=0.002), TCF3::PBX1 (0 vs 5%, p=0.01) and ZNF384 (2 vs 6%, p=0.08) Within each molecular subgroup we compared the survival of patients who achieved MRD-negative status after induction and were randomized to receive blinatumomab plus chemotherapy (n=93) or chemotherapy only (n=91). Although numbers were low, blinatumomab improved relapse-free survival for patients with hyperdiploid, PAX5alt, BCR::ABL1-like and KMT2A ALL compared to chemotherapy alone. Clonal hematopoiesis of indeterminate potential (CHIP)-related gene mutations were identified in 121 of 417 cases analyzed (29% total: somatic 74%, remission 31%, both 5%). TP53 mutations were most frequent (n=65 patients), with the majority identified in patients with low hypodiploid (n=52). Interestingly, patients with CRLF2 rearrangements collectively harbored the highest number of other CHIP gene mutations (n=13), including DNMT3A (n=4), TET2 (n=3) and ASXL1 (n=3). Conclusions:We provide a comprehensive landscape of genomic alterations in adult B-ALL and identify a new group characterized by deregulation of CEBPA/CEBPB (CEBPalt). We also provide insights into the efficacy of blinatumomab in different molecular subgroups of adult B-ALL.

Abstract Background: Recent advances in computer vision have given rise to image foundation models (FMs), large-scale neural networks pre-trained on millions of unlabeled histopathology patches, which can learn generalizable representations of tissue morphology. In many cancers, these FMs have predicted genetic mutations and prognoses directly from histology, suggesting that morphology encodes “hidden” molecular information. In lymphoma, this field is nascent but growing.General-purpose histopathology FMs have limited representation of lymphoid morphology. We developed and evaluated LymphoVision, a lymphoma-specialized image foundation model pretrained with self-supervised learning, to extract clinically meaningful features from routine hematoxylin and eosin (H&E) histology. Methods: LymphoVision was pre-trained using knowledge distillation and consistency (DINOv2) on over 37 million multi-resolution patches (5x, 10x, 20x, 40x) sampled using a density-aware strategy from 31,211 H&E whole slide images (WSIs) from 9,155 archival lymphoma and reactive lymphoproliferative cases. To ensure diverse, representative samples, our patch sampling method uses an efficient graph-based clustering algorithm and prioritizes morphologically informative regions by clustering patch embeddings and performing density sampling within each cluster. The pretrained encoder was evaluated using a clustering-constrained attention multiple instance learning (MIL) framework on three diagnostic tasks: (1) cell-of-origin (COO) classification in diffuse large B-cell lymphoma (DLBCL), distinguishing germinal center B-cell (GCB) from non-GCB subtypes based on the Hans immunohistochemistry algorithm; (2) follicular lymphoma (FL) grading, differentiating low-grade (grades 1–2) from high-grade (grade 3); and (3) multiclass lymphoma subtype classification across six categories: DLBCL, FL, marginal zone lymphoma (MZL), mantle cell lymphoma (MCL), classical Hodgkin lymphoma (CHL), and benign/reactive lymphoid conditions. WSI-level labels were derived from pathology reports, and none of the evaluation cases were included in pretraining. Models used a ViT-Giant LymphoVision backbone to generate 20x patch embeddings, followed by MIL using 50% training data and 2-fold cross-validation (25% validation, 25% test). Performance was assessed using area under the receiver operating characteristic curve (AUC), accuracy, recall (sensitivity) and F1 score. Results: For DLBCL COO classification (total n=358 cases), LymphoVision achieved a mean test AUC of 0.93 (range 0.92–0.94), accuracy of 84.5% and F1 of 0.85 across two independent test folds (n=91 per fold, 50% GCB). These results outperformed general-purpose FMs on the same task and cases, including Virchow-v2 (AUC 0.85) and UNI-v2 (AUC 0.88), despite those models being pretrained on 10–100 times more images. For FL grading (total n=660 cases), the test set in each fold included 166 cases (18% grade 3), yielding a mean test AUC of 0.88 (range 0.87–0.90), accuracy of 89.2% and F1 of 0.80. In multiclass lymphoma subtype classification (total n=1,176 cases), the test set included 296 cases per fold. LymphoVision achieved a mean test AUC of 0.98 (range 0.97-0.99), mean test accuracy of 87.1% and mean F1 of 0.84. Class-wise recall was 92.0% for DLBCL (n=112), 88.2% for FL (n=72), 77.2% for MZL (n=29), 70.8% for MCL (n=35), 87.5% for CHL (n=24), and 95.8% for benign/reactive lymphoid conditions (n=24). Conclusion: LymphoVision is a lymphoma-specialized FM trained using self-supervision that achieves high diagnostic accuracy across multiple clinically meaningful tasks using weak supervision and standard H&E images, suggesting the encoder has learned fine-grained lymphoma morphology. Based on historical benchmarks from the 1997 International Lymphoma Study Group classification project, expert hematopathologists achieved diagnostic agreement rates of 55% to 84% using H&E morphology alone for the non-Hodgkin lymphoma (NHL) entities included in our multiclass task. LymphoVision approached or exceeded these benchmarks. These findings underscore the value of our approach using disease-specialized pretraining and intelligent patch sampling for data-efficient histopathology learning. LymphoVision has the potential to augment diagnostics, enable scalable AI-powered classification in real-world workflows, and serve as a foundation for prognostic and theragnostic applications. External validation and multimodal integration are ongoing.

Abstract BACKGROUND Bevacizumab is an anti-angiogenic monoclonal antibody that inhibits vascular endothelial growth factor (VEGF) used in the treatment of various cancers including recurrent glioblastoma (rGBM) and is particularly useful in mitigating cerebral edema and radiation necrosis. Complications associated with bevacizumab that are of concern for patients requiring urgent surgical procedures include: impaired wound healing, increased bleeding risk and infection. Recent bevacizumab use (e.g. <28 days) is considered a contraindication to surgical procedures. Therapeutic plasma exchange (TPE) is a procedure that nonspecifically removes substances from plasma such as immunoglobulins. We reviewed our institutional experience with the use of TPE to remove bevacizumab for patients who required urgent neurosurgical intervention. METHODS Single center retrospective study from: 01/01/2022 to 12/31/2024 (first patient treated for this indication was in 2022). Inclusion criteria: patients who underwent perioperatively TPE to remove bevacizumab. Data collected from electronic medical record included age, gender, pertinent medical history, indication for bevacizumab and TPE for its removal, TPE procedure parameters, and clinical outcomes. RESULTS Four patients met study inclusion criteria, all were male. Three patients had a diagnosis of rGBM, one patient had a diagnosis of metastatic melanoma. All patients had received bevacizumab within 21 days of TPE (range 8-21). All subjects underwent 3 daily, 1 plasma volume TPE. Eleven of 12 procedures used the Fenwal Amicus® Separator. Neurosurgical outcomes were favorable. In one patient, bevacizumab levels measured by ELISA decreased from 1087.63 ng/mL to 145.35 ng/mL after the third procedure. No post-surgical wound complications (e.g., wound dehiscence, bleeding, infection) were observed in four patients requiring urgent neurosurgery despite recent administration of bevacizumab. CONCLUSIONS TPE was successful and well-tolerated by subjects in this study. Perioperative bevacizumab removal by TPE may be considered to reduce the risk of wound healing complications associated with bevacizumab.

Background: Flow cytometric detection of minimal residual disease (MRD) in chronic lymphocytic leukemia (CLL) is complex, time-consuming, and subject to inter-operator variability. Deep neural networks (DNNs) offer potential for standardization and efficiency improvement, but require rigorous validation and monitoring for safe clinical implementation. Methods: We evaluated a DNN-assisted human-in-the-loop approach for CLL MRD detection. Initial validation included method comparison against manual analysis (n = 240), precision studies, and analytical sensitivity verification. Post-implementation monitoring comprised four components: daily electronic quality control, input data drift detection, error analysis, and attribute acceptance sampling. Laboratory efficiency was assessed through a timing study of 161 cases analyzed by five technologists. Results: Method comparison demonstrated 97.5% concordance with manual analysis for qualitative classification (sensitivity 100%, specificity 95%) and excellent correlation for quantitative assessment (r = 0.99, Deming slope = 0.99). Precision studies confirmed high repeatability and within-laboratory precision across multiple operators. Analytical sensitivity was verified at 0.002% MRD. Post-implementation monitoring identified 2.97% of cases (26/874) with input data drift, primarily high-burden CLL and non-CLL neoplasms. Error analysis showed the DNN alone achieved 97% sensitivity compared to human-in-the-loop-reviewed results, with 13 missed cases (1.5%) showing atypical immunophenotypes. Attribute acceptance sampling confirmed 98.8% of reported negative cases were true negatives. The DNN-assisted workflow reduced average analysis time by 60.3% compared to manual analysis (4.2 ± 2.3 vs. 10.5 ± 5.8 min). Conclusions: The implementation of a DNN-assisted approach for CLL MRD detection in a clinical laboratory provides diagnostic performance equivalent to expert manual analysis while substantially reducing analysis time. Comprehensive performance monitoring ensures ongoing safety and effectiveness in routine clinical practice. This approach provides a model for responsible AI integration in clinical laboratories, balancing automation benefits with expert oversight.

AbstractBreast cancer affects millions globally, necessitating precise biomarker testing for effective treatment. HER2 testing is crucial for guiding therapy, particularly with novel antibody‐drug conjugates (ADCs) like trastuzumab deruxtecan, which shows promise for breast cancers with low HER2 expression. Current HER2 testing methods, including immunohistochemistry (IHC) and in situ hybridization (ISH), have limitations. IHC, a semi‐quantitative assay, is prone to interobserver variability. While ISH provides higher precision than IHC, it remains more resource‐intensive in terms of cost and workflow. However, turnaround time is typically faster than that of other advanced molecular methods such as next‐generation sequencing. We adapted the clustering‐constrained‐attention multiple‐instance deep learning model to improve IHC testing and reduce dependence on reflex fluorescence ISH (FISH) tests. Using 5,731 HER2 IHC images, including 592 cases with FISH testing, we trained two models: one for predicting HER2 scores from IHC images and another for predicting FISH scores from equivocal cases. The HER2 IHC score prediction model achieved 91% ± 0.01 overall accuracy and a receiver operating characteristic (ROC) area under the curve (AUC) of 0.98 ± 0.01. The FISH score prediction model had an ROC AUC of 0.84 ± 0.07, with sensitivity at 0.37 ± 0.13 and specificity at 0.96 ± 0.03. External validation on cases from 203 institutions showed similar performance. The HER2 IHC model maintained a 91% ± 0.01 accuracy and an ROC AUC of 0.98 ± 0.01, while the FISH model had an ROC AUC of 0.75 ± 0.03, with sensitivity at 0.28 ± 0.04 and specificity at 0.93 ± 0.01. Our model advances HER2 scoring by reducing subjectivity and variability in current scoring methods. Despite lower accuracy and sensitivity in the FISH prediction model, it may be a beneficial option for settings where reflex FISH testing is unavailable or prohibitive. With high specificity, our model can serve as an effective screening tool, enhancing breast cancer diagnosis and treatment selection.

Introduction E1910 randomized pts with ALL aged 30-70 years who were measurable residual disease negative (MRD-) by multi-color flow cytometry (FC) after intensification to conventional chemotherapy (chemo) with or without blinatumomab (blina). The addition of blina resulted in improved overall survival (OS) (Litzow, NEJM 2024) and led to a new standard of care. Older ALL pts have inferior outcomes when compared to younger pts in part due to increased treatment toxicity. In E1910, exposure to dexamethasone and pegasparagase was reduced for older pts (age >55 years). We present results of subgroup analyses describing outcomes of older pts enrolled in E1910. Methods Details of the E1910 protocol remission induction (step 1), high dose methotrexate with pegaspargase intensification (step 2), blina randomization (step 3), and maintenance (step 4) or allogeneic hematopoietic stem cell transplant (HCT) were previously presented (Litzow, NEJM 2024). MRD status was determined after step 2 centrally with 6-color FC with MRD- defined as <0.01% in the bone marrow (BM). Following approval of blina in the US for MRD positive (MRD+) disease in 2018, MRD+ pts were assigned to the blina arm and no longer randomized. The primary objective was to compare the OS and relapse free survival (RFS) among older pts aged >55 years (pre-specified stratification factor) who received chemo + blina to that of pts who received chemo alone (step 3 treatment). Estimates of OS and RFS were calculated using the Kaplan-Meier method. Comparison of OS and RFS between treatment arms was conducted using the two-sided stratified log-rank test and Cox model with CD20 status, rituximab use, and whether pts intended to receive HCT or not as stratification factors. Stratified Multivariate Cox models of OS and RFS were used to further evaluate the effect of step 3 treatment, adjusted by gender, white blood cell count (WBC), platelets, hemoglobin, peripheral blood blasts, BM blasts, performance status and combined molecular risk category. Results Among the 488 enrolled pts, 211 were ≥55 years old with median age of 61 (range 55-70). Among older pts, 49.3% were female and 83.4% were white, 17 (8.1%) pts had WBC >30.000/uL at the time of diagnosis. 156 (73.9%) pts had CD10+ early B-ALL and 46 (21.8%) had CD10- B-ALL immunophenotype. The most common genetic subgroups were low hypodiploid (n=58, 27.5%), BCR::ABL1-like (n=34, 16.1%) and KMT2A rearranged (n=26, 12.3%). 8.1% of pts had low, 9.0% intermediate, 63.5% high combined molecular risk disease and in 19.4% of pts risk could not be assigned. Baseline characteristics were not significantly different in chemo + blina and chemo only pts. The median follow-up for this analysis from step 1 registration was 54.6 months. Complete response with (CR) or without (CRi) count recovery after induction was observed in 174 (82.5%) pts (167 CR, 7 CRi). 93 (44%) pts were MRD-. 32 (15.1%) pts received HCT on study and 24 (11.4%) pts received it off study. Among MRD- pts (n=93), 46 were randomized to and 45 received blina: 6 (13.3%) pts had 1, 14 (31.1%) had 2, 2 (4.4%) had 3 and 23 (51.1%) had 4 cycles. 3-year OS were 71% and 67% for the blina and the chemo only arm (univariate model: HR 0.75, 95% CI: 0.37-1.50; multivariate model: HR 0.56, 95% CI: 0.25-1.27). 3-year RFS were 69% and 59% for the blina and the chemo only arm (univariate model: HR 0.75, 95% CI: 0.39-1.45; multivariate model: HR 0.67, 95% CI: 0.32-1.42). Among MRD+ pts (n=25), 14 were randomized to and 11 received blina: 7 (63.6%) pts had 2, and 4 (36.4%) had 4 cycles. 3-year OS were 66% and 44% for the blina vs the chemo only arm (univariate model: HR 0.66, 95% CI: 0.12-3.64). 3-year RFS were 64% and 31% for the blina vs the chemo only arm (univariate model: HR 0.64, 95% CI: 0.15-2.74). Frequency of treatment-related toxicities among older pts during different study phases and treatment assignments will be presented. Conclusion Despite evidence of improved outcomes in the whole study cohort, the addition of blina to consolidation chemo for older pts with ALL was not associated with statistically significant improvement of OS or RFS in this exploratory analysis not powered to detect the difference in subgroups. This is possibly due to sample size but may be due to biologic differences in the older adult population. Further studies are needed to definitively determine tolerance and benefit of blina addition to consolidation for older pts with ALL.

Introduction: Based on a donor versus no-donor analysis, UKALLXII/E2993 showed a survival benefit for adults with acute lymphoblastic leukemia (ALL) who underwent allogeneic hematopoietic cell transplantation (alloHCT) in first complete remission (CR1) compared to consolidation/maintenance chemotherapy alone. The follow-up phase 3 E1910 trial (NCT02003222) randomized BCR::ABL1-negative B-lineage ALL patients (pts) aged 30-70 years in measurable residual disease (MRD)-negative CR1 after intensification to conventional chemotherapy (CC) +/- blinatumomab (blin) and demonstrated significantly improved overall survival (OS) with blin (Litzow, NEJM 2024). The current exploratory subgroup analysis evaluates outcomes of MRD-negative pts randomized on E1910 who underwent alloHCT on trial. Methods: Among pts enrolled on E1910, pts in MRD-negative CR1 by 6-color flow cytometry after intensification who were randomized to CC +/- blin and subsequently underwent alloHCT were identified. We compared baseline characteristics, MRD status post induction, relapse-free survival (RFS) and OS using the Kaplan-Meier method and the two-sided log-rank test and Cox model. A multivariable analysis was performed with alloHCT included as a time-varying covariate as well as a landmark analysis comparing the outcome of pts with unfavorable risk on the CC + blin arm who underwent alloHCT versus those who did not. Results: Of 224 MRD-negative randomized pts, 44 underwent alloHCT on trial: 22 in each arm. The median time from randomization to transplant was 3.2 months (mos). The median follow-up from initial registration was 57.8 mos. Baseline demographics including age (median 55 years), gender (53% male), race (89% white) and ethnicity (89% non-Hispanic/Latino) were equally distributed. Compared to CC + blin, the median peripheral blood white cell (WBC) count and blast percentage at diagnosis were higher in the CC arm at 6,640 vs 3,740/uL and 55 vs 9%, respectively, while median marrow blast percentage and overall unfavorable risk as defined in E1910 (55% vs 64%) were similar. Of the 44 MRD-negative pts who proceeded to alloHCT on study, 77% in the CC arm and 64% in the CC+ blin arm were MRD-negative after induction. Myeloablative conditioning was given to the majority of pts (68% CC arm and 73% CC+blin arm) and the most common graft versus host disease prophylaxis was tacrolimus-based on both arms. Among the 44 pts who underwent alloHCT, the observed hazard ratio (HR) comparing relative risk of death between pts randomized to CC+blin or CC only was 0.59 (95% CI: 0.12-2.11) with a 3-year OS rate of 82% versus 71% and the HR comparing relative risk of relapse/death was 0.52 (CI: 0.15-1.79) with a 3-year RFS rate of 82% versus 67%. For patients with unfavorable risk, 3-year OS and 3-year RFS in each treatment arm were both 71% versus 65%. Among 50 pts with unfavorable risk in MRD-negative CR after intensification who were randomized to CC + blin, a landmark analysis performed at 5 mos post randomization showed 3-year OS and RFS rates of 71% versus 90% and 71% versus 86% between the 14 pts who underwent alloHCT and those who did not, respectively. Receipt of alloHCT was a time-varying covariate in a multivariable analysis with gender, WBC count, platelets, hemoglobin, peripheral blood blasts, bone marrow blasts, performance status, and combined molecular risk category. Age, CD20 status, rituximab use, and intention to receive alloHCT were used as stratification factors. In the entire cohort of MRD-negative pts who were randomized, there was a significant association between improved OS and the receipt of CC+blin vs CC alone (HR 0.46, 95% CI: 0.26 - 0.82; p=0.009), better performance status of 0-1 vs ≥2 and favorable vs unfavorable risk ALL. In contrast, the receipt of alloHCT had no significant effect on OS (HR 1.00, 95% CI: 0.49 - 2.01). Conclusions: Within the constraints of small numbers, this exploratory subgroup analysis demonstrates that in pts with newly diagnosed BCR::ABL1-negative B-ALL treated on E1910 who achieved MRD-negativity after intensification, alloHCT outcomes did not appear to be different with or without blin in consolidation. In a multivariate analysis of all MRD-negative pts, receipt of blin but not alloHCT was associated with improved OS. Further investigation is warranted to characterize the relative benefit of blinatumomab in pts receiving alloHCT in frontline consolidation.