Making sense of artificial intelligence and large language models-including ChatGPT-in pediatric hematology/oncology.

ChatGPT and other artificial intelligence (AI) systems have captivated the attention of healthcare providers and researchers for their potential to improve care processes and outcomes. While these technologies hold promise to automate processes, increase efficiency, and reduce cognitive burden, their use also carries risks. In this commentary, we review basic concepts of AI, outline some of the capabilities and limitations of currently available tools, discuss current and future applications in pediatric hematology/oncology, and provide an evaluation and implementation framework that can be used by pediatric hematologist/oncologists considering the use of AI in clinical practice.

Directed evolution-based discovery of ligands for in vivo restimulation of CAR-T cells.

Chimeric antigen receptor (CAR) T cell therapy targeting CD19 elicits remarkable clinical efficacy in B-cell malignancies, but many patients relapse due to failed expansion and/or progressive loss of CAR-T cells. We recently reported a strategy to potently restimulate CAR-T cells in vivo, enhancing their functionality by administration of a vaccine-like stimulus comprised of surrogate peptide ligands for a CAR linked to a lymph node-targeting amphiphilic PEG-lipid (termed CAR-T-vax). Here, we demonstrate a general strategy to generate and optimize peptide mimotopes enabling CAR-T-vax generation for any CAR. Using the clinical CD19 CAR FMC63 as a test case, we employed yeast surface display to identify peptide binders to soluble IgG versions of FMC63, which were subsequently affinity matured by directed evolution. CAR-T vaccines using these optimized mimotopes triggered marked expansion of both murine CD19 CAR-T cells in a syngeneic model and human CAR-T cells in a humanized mouse model of B cell acute lymphoblastic leukemia (B-ALL), and enhanced control of leukemia progression. This approach thus enables vaccine boosting to be applied to any clinically-relevant CAR-T cell product.

Reinfusion of CD19 CAR T cells for relapse prevention and treatment in children with acute lymphoblastic leukemia.

ABSTRACT: Relapse after CD19-directed chimeric antigen receptor (CAR)-modified T cells remains a substantial challenge. Short CAR T-cell persistence contributes to relapse risk, necessitating novel approaches to prolong durability. CAR T-cell reinfusion (CARTr) represents a potential strategy to reduce the risk of or treat relapsed disease after initial CAR T-cell infusion (CARTi). We conducted a retrospective review of reinfusion of murine (CTL019) or humanized (huCART19) anti-CD19/4-1BB CAR T cells across 3 clinical trials or commercial tisagenlecleucel for relapse prevention (peripheral B-cell recovery [BCR] or marrow hematogones ≤6 months after CARTi), minimal residual disease (MRD) or relapse, or nonresponse to CARTi. The primary endpoint was complete response (CR) at day 28 after CARTr, defined as complete remission with B-cell aplasia. Of 262 primary treatments, 81 were followed by ≥1 reinfusion (investigational CTL019, n = 44; huCART19, n = 26; tisagenlecleucel, n = 11), representing 79 patients. Of 63 reinfusions for relapse prevention, 52% achieved CR (BCR, 15/40 [38%]; hematogones, 18/23 [78%]). Lymphodepletion was associated with response to CARTr for BCR (odds ratio [OR], 33.57; P = .015) but not hematogones (OR, 0.30; P = .291). The cumulative incidence of relapse was 29% at 24 months for CR vs 61% for nonresponse to CARTr (P = .259). For MRD/relapse, CR rate to CARTr was 50% (5/10), but 0/8 for nonresponse to CARTi. Toxicity was generally mild, with the only grade ≥3 cytokine release syndrome (n = 6) or neurotoxicity (n = 1) observed in MRD/relapse treatment. Reinfusion of CTL019/tisagenlecleucel or huCART19 is safe, may reduce relapse risk in a subset of patients, and can reinduce remission in CD19+ relapse.

Chimeric Antigen Receptor T Cell Manufacturing on an Automated Cell Processor.

Chimeric antigen receptor (CAR)-T cells represent a promising immunotherapeutic approach for the treatment of various malignant and non-malignant diseases. CAR-T cells are genetically modified T cells that express a chimeric protein that recognizes and binds to a cell surface target, resulting in the killing of the target cell. Traditional CAR-T cell manufacturing methods are labor-intensive, expensive, and may carry the risk of contamination. The CliniMACS Prodigy, an automated cell processor, allows for manufacturing cell therapy products at a clinical scale in a closed system, minimizing the risk of contamination. Processing occurs semi-automatically under the control of a computer and thus minimizes human involvement in the process, which saves time and reduces variability and errors. This manuscript and video describes the T cell transduction (TCT) process for manufacturing CAR-T cells using this processor. The TCT process involves CD4+/CD8+ T cell enrichment, activation, transduction with a viral vector, expansion, and harvest. Using the Activity Matrix, a functionality that allows ordering and timing of these steps, the TCT process can be customized extensively. We provide a walk-through of CAR-T cell manufacturing in compliance with current Good Manufacturing Practice (cGMP) and discuss required release testing and preclinical experiments that will support an Investigational New Drug (IND) application. We demonstrate the feasibility and discuss the advantages and disadvantages of using a semi-automatic process for clinical CAR-T cell manufacturing. Finally, we describe an ongoing investigator-initiated clinical trial that targets pediatric B-cell malignancies [NCT05480449] as an example of how this manufacturing process can be applied in a clinical setting.

A machine-learning model that incorporates CD45 surface expression predicts hematopoietic progenitor cell recovery after freeze-thaw.

BACKGROUND AIMS: Sufficient doses of viable CD34+ (vCD34) hematopoietic progenitor cells (HPCs) are crucial for engraftment. Additional-day apheresis collections can compensate for potential loss during cryopreservation but incur high cost and additional risk. To aid predicting such losses for clinical decision support, we developed a machine-learning model using variables obtainable on the day of collection. METHODS: In total, 370 consecutive autologous HPCs, apheresis-collected since 2014 at the Children's Hospital of Philadelphia, were retrospectively reviewed. Flow cytometry was used to assess vCD34% on fresh products and thawed quality control vials. The ratio of vCD34% thawed to fresh, which we call "post-thaw index," was used as an outcome measure, with a "poor" post-thaw index defined as <70%. HPC CD45 normalized mean fluorescence intensity (MFI) was calculated by dividing CD45 MFI of HPCs to the CD45 MFI of lymphocytes in the same sample. We trained XGBoost, k-nearest neighbor and random forest models for the prediction and calibrated the best model to minimize falsely-reassuring predictions. RESULTS: In total, 63 of 370 (17%) products had a poor post-thaw index. The best model was XGBoost, with an area under the receiver operator curve of 0.83 evaluated on an independent test data set. The most important predictor for a poor post-thaw index was the HPC CD45 normalized MFI. Transplants after 2015, based on the lower of the two vCD34% values, showed faster engraftment than older transplants, which were based on fresh vCD34% only (average 10.6 vs 11.7 days, P = 0.0006). CONCLUSIONS: Transplants taking into account post-thaw vCD34% improved engraftment time in our patients; however, it came at the cost of unnecessary multi-day collections. The results from applying our predictive algorithm retrospectively to our data suggest that more than one-third of additional-day collections could have been avoided. Our investigation also identified CD45 nMFI as a novel marker for assessing HPC health post-thaw.

Automation of Hematopoietic Cell Transplantation Outcomes Reporting Leads to Dramatic Reduction in Errors Reported to Real-World Data Registry.

Institutions that perform hematopoietic cell transplantation (HCT) are required by law to report standardized, structured data on transplantation outcomes. A key post-transplantation outcome is engraftment, the time between HCT infusion and reemergence of circulating neutrophils and platelets. At our center, we found that manual chart abstraction for engraftment data was highly error-prone. We developed a custom R/Shiny application that automatically calculates engraftment dates and displays them in an intuitive format to augment the manual chart review. Our hypothesis was that use of the application to assist with calculating and reporting engraftment dates would be associated with a decreased error rate. The study was conducted at a single tertiary care institution. The application was developed in a collaborative, multidisciplinary fashion by members of an embedded cellular therapy informatics team. Retrospective validation of the application's accuracy was conducted on all malignant HCTs from February 2016 to December 2020 (n = 198). Real-world use of the application was evaluated prospectively from April 2021 through April 2022 (n = 53). The Welch 2-sample t test was used to compare error rates preimplementation and postimplementation. Data were visualized using p charts, and standard special cause variation rules were applied. The accuracy of reported data postdeployment increased dramatically; the engraftment error rate decreased from 15% to 3.8% for neutrophils (P = .003) and from 28% to 1.9% for platelets (P < .001). This study demonstrates the effective deployment of a custom R/Shiny application that was associated with significantly reduced error rates in HCT engraftment reporting for operational, research, and regulatory purposes. Users reported subjective satisfaction with the application and that it addressed difficulties with the legacy manual process. Identifying and correcting erroneous data in engraftment reporting could lead to a more efficient and accurate nationwide assessment of transplantation success. Furthermore, we show that it is possible and practical for academic medical centers to create and support embedded informatics teams that can quickly build applications for clinical operations in a manner compliant with regulatory requirements.

Manufacture and Characterization of Good Manufacturing Practice-Compliant SARS-COV-2 Cytotoxic T Lymphocytes.

BACKGROUND: Coronavirus disease 2019 (COVID-19) is caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). SARS-CoV-2 virus-specific cytotoxic T-cell lymphocytes (vCTLs) could provide a promising modality in COVID-19 treatment. We aimed to screen, manufacture, and characterize SARS-CoV-2-vCTLs generated from convalescent COVID-19 donors using the CliniMACS Cytokine Capture System (CCS). METHODS: Donor screening was done by stimulation of convalescent COVID-19 donor peripheral blood mononuclear cells with viral peptides and identification of interferonγ (IFN-γ)+ CD4 and CD8 T cells using flow cytometry. Clinical-grade SARS-CoV-2-vCTLs were manufactured using the CliniMACS CCS. The enriched SARS-CoV-2-vCTLs were characterized by T-cell receptor sequencing, mass cytometry, and transcriptome analysis. RESULTS: Of the convalescent donor blood samples, 93% passed the screening criteria for clinical manufacture. Three validation runs resulted in enriched T cells that were 79% (standard error of the mean 21%) IFN-γ+ T cells. SARS-CoV-2-vCTLs displayed a highly diverse T-cell receptor repertoire with enhancement of both memory CD8 and CD4 T cells, especially in CD8 TEM, CD4 TCM, and CD4 TEMRA cell subsets. SARS-CoV-2-vCTLs were polyfunctional with increased gene expression in T-cell function, interleukin, pathogen defense, and tumor necrosis factor superfamily pathways. CONCLUSIONS: Highly functional SARS-CoV-2-vCTLs can be rapidly generated by direct cytokine enrichment (12 hours) from convalescent donors. CLINICAL TRIALS REGISTRATION: NCT04896606.

Impact of poverty and neighborhood opportunity on outcomes for children treated with CD19-directed CAR T-cell therapy.

Children living in poverty experience excessive relapse and death from newly diagnosed acute lymphoblastic leukemia (ALL). The influence of household poverty and neighborhood social determinants on outcomes from chimeric antigen receptor (CAR) T-cell therapy for relapsed/refractory (r/r) leukemia is poorly described. We identified patients with r/r CD19+ ALL/lymphoblastic lymphoma treated on CD19-directed CAR T-cell clinical trials or with commercial tisagenlecleucel from 2012 to 2020. Socioeconomic status (SES) was proxied at the household level, with poverty exposure defined as Medicaid-only insurance. Low-neighborhood opportunity was defined by the Childhood Opportunity Index. Among 206 patients aged 1 to 29, 35.9% were exposed to household poverty, and 24.9% had low-neighborhood opportunity. Patients unexposed to household poverty or low-opportunity neighborhoods were more likely to receive CAR T-cell therapy with a high disease burden (>25%), a disease characteristic associated with inferior outcomes, as compared with less advantaged patients (38% vs 30%; 37% vs 26%). Complete remission (CR) rate was 93%, with no significant differences by household poverty (P = .334) or neighborhood opportunity (P = .504). In multivariate analysis, patients from low-opportunity neighborhoods experienced an increased hazard of relapse as compared with others (P = .006; adjusted hazard ratio [HR], 2.3; 95% confidence interval [CI], 1.3-4.1). There was no difference in hazard of death (P = .545; adjusted HR, 1.2; 95% CI, 0.6-2.4). Among children who successfully receive CAR T-cell therapy, CR and overall survival are equitable regardless of proxied SES and neighborhood opportunity. Children from more advantaged households and neighborhoods receive CAR T-cell therapy with a higher disease burden. Investigation of multicenter outcomes and access disparities outside of clinical trial settings is warranted.

Unrelated donor α/ß T cell- and B cell-depleted HSCT for the treatment of pediatric acute leukemia.

Unrelated donor (URD) hematopoietic stem cell transplant (HSCT) is associated with an increased risk of severe graft-versus-host disease (GVHD). TCRαß/CD19 depletion may reduce this risk, whereas maintaining graft-versus-leukemia. Outcome data with TCRαß/CD19 depletion generally describe haploidentical donors, with relatively few URDs. We hypothesized that TCRαß/CD19-depletion would attenuate the risks of GVHD and relapse for URD HSCT. Sixty pediatric and young adult (YA) patients with hematologic malignancies who lacked a matched-related donor were enrolled at 2 large pediatric transplantation centers between October 2014 and September 2019. All patients with acute leukemia had minimal residual disease testing, and DP typing was available for 77%. All patients received myeloablative total body irradiation- or busulfan-based conditioning with no posttransplant immune suppression. Engraftment occurred in 98%. Four-year overall survival was 69% (95% confidence interval [CI], 52%-81%), and leukemia-free survival was 64% (95% CI, 48%-76%), with no difference between lymphoid and myeloid malignancies (P = .6297 and P = .5441, respectively). One patient (1.7%) experienced primary graft failure. Relapse occurred in 11 patients (3-year cumulative incidence, 21%; 95% CI, 11-34), and 8 patients (cumulative incidence, 15%; 95% CI, 6.7-26) experienced nonrelapse mortality. Grade III to IV acute GVHD was seen in 8 patients (13%), and 14 patients (26%) developed chronic GVHD, of which 6 (11%) had extensive disease. Nonpermissive DP mismatch was associated with higher likelihood of acute GVHD (odds ratio, 16.50; 95% CI, 1.67-163.42; P = .0166) but not with the development of chronic GVHD. URD TCRαß/CD19-depleted peripheral HSCT is a safe and effective approach to transplantation for children/YAs with leukemia. This trial was registered at www.clinicaltrials.gov as #NCT02323867.

Measures to reduce red cell use in patients with sickle cell disease requiring red cell exchange during a blood shortage.

The COVID-19 pandemic has created major disruptions in health care delivery, including a severe blood shortage. The inventory of Rh and K antigen-negative red cell units recommended for patients with hemoglobinopathies became alarmingly low and continues to be strained. Because patients with sickle cell disease requiring chronic red cell exchange (RCE) incur a large demand for red cell units, we hypothesized that implementation of 2 measures could reduce blood use. First, obtaining the pretransfusion hemoglobin S (HbS) results by procedure start time would facilitate calculation of exact red cell volume needed to achieve the desired post-RCE HbS. Second, as a short-term conservation method, we identified patients for whom increasing the targeted end procedure hematocrit up to 5 percentage points higher than the pretransfusion level (no higher than 36%) was not medically contraindicated. The goal was to enhance suppression of endogenous erythropoiesis and thereby reduce the red cell unit number needed to maintain the same target HbS%. These 2 measures resulted in an 18% reduction of red cell units transfused to 50 patients undergoing chronic RCE during the first 6 months of the COVID-19 pandemic. Despite reduction of blood use, pretransfusion HbS% target goals were maintained and net iron accumulation was low. Both strategies can help alleviate a shortage of Rh and K antigen-negative red cells, and, more generally, transfusing red cell units based on precise red cell volume required can optimize patient care and judicious use of blood resources.

Risk-Adapted Preemptive Tocilizumab to Prevent Severe Cytokine Release Syndrome After CTL019 for Pediatric B-Cell Acute Lymphoblastic Leukemia: A Prospective Clinical Trial.

PURPOSE: To prospectively evaluate the effectiveness of risk-adapted preemptive tocilizumab (PT) administration in preventing severe cytokine release syndrome (CRS) after CTL019, a CD19 chimeric antigen receptor T-cell therapy. METHODS: Children and young adults with CD19-positive relapsed or refractory B-cell acute lymphoblastic leukemia were assigned to high- (≥ 40%) or low- (< 40%) tumor burden cohorts (HTBC or LTBC) based on a bone marrow aspirate or biopsy before infusion. HTBC patients received a single dose of tocilizumab (8-12 mg/kg) after development of high, persistent fevers. LTBC patients received standard CRS management. The primary end point was the frequency of grade 4 CRS (Penn scale), with an observed rate of ≤ 5 of 15 patients in the HTBC pre-defined as clinically meaningful. In post hoc analyses, the HTBC was compared with a historical cohort of high-tumor burden patients from the initial phase I CTL019 trial. RESULTS: The primary end point was met. Seventy patients were infused with CTL019, 15 in the HTBC and 55 in the LTBC. All HTBC patients received the PT intervention. The incidence of grade 4 CRS was 27% (95% CI, 8 to 55) in the HTBC and 3.6% (95% CI, 0.4 to 13) in the LTBC. The best overall response rate was 87% in the HTBC and 100% in the LTBC. Initial CTL019 expansion was greater in the HTBC than the LTBC (P < .001), but persistence was not different (P = .73). Event-free and overall survival were worse in the HTBC (P = .004, P < .001, respectively). In the post hoc analysis, grade 4 CRS was observed in 27% versus 50% of patients in the PT and prior phase I cohorts, respectively (P = .18). CONCLUSION: Risk-adapted PT administration resulted in a decrease in the expected incidence of grade 4 CRS, meeting the study end point, without adversely impacting the antitumor efficacy or safety of CTL019.

New directions in chimeric antigen receptor T cell [CAR-T] therapy and related flow cytometry.

Chimeric antigen receptor (CAR) T cells provide a promising approach to the treatment of hematologic malignancies and solid tumors. Flow cytometry is a powerful analytical modality, which plays an expanding role in all stages of CAR T therapy, from lymphocyte collection, to CAR T cell manufacturing, to in vivo monitoring of the infused cells and evaluation of their function in the tumor environment. Therefore, a thorough understanding of the new directions is important for designing and implementing CAR T-related flow cytometry assays in the clinical and investigational settings. However, the speed of new discoveries and the multitude of clinical and preclinical trials make it challenging to keep up to date in this complex field. In this review, we summarize the current state of CAR T therapy, highlight the areas of emergent research, discuss applications of flow cytometry in modern cell therapy, and touch upon several considerations particular to CAR detection and assessing the effectiveness of CAR T therapy.

Machine Learning Models Improve the Diagnostic Yield of Peripheral Blood Flow Cytometry.

OBJECTIVES: Peripheral blood flow cytometry (PBFC) is useful for evaluating circulating hematologic malignancies (HM) but has limited diagnostic value for screening. We used machine learning to evaluate whether clinical history and CBC/differential parameters could improve PBFC utilization. METHODS: PBFC cases with concurrent/recent CBC/differential were split into training (n = 626) and test (n = 159) cohorts. We classified PBFC results with abnormal blast/lymphoid populations as positive and used two models to predict results. RESULTS: Positive PBFC results were seen in 58% and 21% of training cases with and without prior HM (P < .001). % neutrophils, absolute lymphocyte count, and % blasts/other cells differed significantly between positive and negative PBFC groups (areas under the curve [AUC] > 0.7). Among test cases, a decision tree model achieved 98% sensitivity and 65% specificity (AUC = 0.906). A logistic regression model achieved 100% sensitivity and 54% specificity (AUC = 0.919). CONCLUSIONS: We outline machine learning-based triaging strategies to decrease unnecessary utilization of PBFC by 35% to 40%.

Red blood cell alloantibodies are associated with increased alloimmunization against human leukocyte antigens.

BACKGROUND: Alloantibodies recognizing human leukocyte antigens (HLA) can cause immune-mediated refractoriness to platelet transfusion. An association between HLA alloimmunization and red blood cell (RBC) alloimmunization has been suggested but remains uncertain. STUDY DESIGN AND METHODS: We tested for HLA alloantibodies in 660 patients with and without RBC alloantibodies. Calculated panel reactive antibody (cPRA) values were determined for HLA alloimmunized patients. Current and historical diagnoses and blood product exposure were catalogued. Variables associated with high-level HLA alloimmunization (cPRA ≥ 90%) were evaluated. RESULTS: The cohort included 366 women and 294 men with median age of 66 years (interquartile range [IQR], 53-76). The number of patients with and without RBC alloantibodies was 447 and 213, respectively. Among patients with and without RBC alloantibodies, 20.6% and 8.5% had a cPRA ≥ 90%, respectively (p < 0.0001). In univariate analyses of men and nulliparous women and previously pregnant women, the median number of RBC alloantibodies was significantly higher in patients with a cPRA ≥ 90% (p < 0.0001). The number of RBC alloantibodies remained an independent predictor of a cPRA ≥ 90% in multivariate analysis (odds ratio [OR] 1.50, 95% confidence interval [CI] 1.22-1.85). Other independent predictors of a cPRA ≥ 90% were parity (OR 1.26, 95% CI 1.08-1.47), age (OR 0.98, 95% CI 0.97-1.00), history of renal disease (OR 1.88, 95% CI 1.02-3.48), and number of non-leukoreduced products transfused (OR 1.02, 95% CI 1.00-1.04). CONCLUSIONS: RBC alloimmunization was significantly associated with HLA alloimmunization with a cPRA ≥ 90%. RBC alloimmunization status combined with specific components of the clinical history may estimate the risk of high-level HLA alloimmunization.

Building a Robust Tumor Profiling Program: Synergy between Next-Generation Sequencing and Targeted Single-Gene Testing.

Next-generation sequencing (NGS) is a powerful platform for identifying cancer mutations. Routine clinical adoption of NGS requires optimized quality control metrics to ensure accurate results. To assess the robustness of our clinical NGS pipeline, we analyzed the results of 304 solid tumor and hematologic malignancy specimens tested simultaneously by NGS and one or more targeted single-gene tests (EGFR, KRAS, BRAF, NPM1, FLT3, and JAK2). For samples that passed our validated tumor percentage and DNA quality and quantity thresholds, there was perfect concordance between NGS and targeted single-gene tests with the exception of two FLT3 internal tandem duplications that fell below the stringent pre-established reporting threshold but were readily detected by manual inspection. In addition, NGS identified clinically significant mutations not covered by single-gene tests. These findings confirm NGS as a reliable platform for routine clinical use when appropriate quality control metrics, such as tumor percentage and DNA quality cutoffs, are in place. Based on our findings, we suggest a simple workflow that should facilitate adoption of clinical oncologic NGS services at other institutions.

Functions of BET proteins in erythroid gene expression.

Inhibitors of bromodomain and extraterminal motif proteins (BETs) are being evaluated for the treatment of cancer and other diseases, yet much remains to be learned about how BET proteins function during normal physiology. We used genomic and genetic approaches to examine BET function in a hematopoietic maturation system driven by GATA1, an acetylated transcription factor previously shown to interact with BETs. We found that BRD2, BRD3, and BRD4 were variably recruited to GATA1-regulated genes, with BRD3 binding the greatest number of GATA1-occupied sites. Pharmacologic BET inhibition impaired GATA1-mediated transcriptional activation, but not repression, genome-wide. Mechanistically, BETs promoted chromatin occupancy of GATA1 and subsequently supported transcriptional activation. Using a combination of CRISPR-Cas9-mediated genomic engineering and shRNA approaches, we observed that depletion of either BRD2 or BRD4 alone blunted erythroid gene activation. Surprisingly, depletion of BRD3 only affected erythroid transcription in the context of BRD2 deficiency. Consistent with functional overlap among BET proteins, forced BRD3 expression substantially rescued defects caused by BRD2 deficiency. These results suggest that pharmacologic BET inhibition should be interpreted in the context of distinct steps in transcriptional activation and overlapping functions among BET family members.

Activated protein C resistance testing for factor V Leiden.

Activated protein C resistance assays can detect factor V Leiden with high accuracy, depending on the method used. Factor Xa inhibitors such as rivaroxaban and direct thrombin inhibitors including dabigatran, argatroban, and bivalirudin can cause falsely normal results. Lupus anticoagulants can cause incorrect results in most current assays. Assays that include dilution into factor V-deficient plasma are needed to avoid interference from factor deficiencies or elevations, which can arise from a wide variety of conditions such as warfarin, liver dysfunction, or pregnancy. The pros and cons of the currently available assays are discussed.