BACKGROUND: The patient moving average (MA) is a QC strategy using the mean patient result to continuously monitor assay performance. Developing sensitive MA protocols that rapidly detect systematic error (SE) is challenging. We compare MA protocols established using a previously published report as a guide and demonstrate the use of a simulated annealing (SA) algorithm to optimize MA protocol performance. METHODS: Using 400 days of patient data, we developed MA protocols for 23 assays. MA protocols developed using a previously published report and our SA algorithm were compared using the average number of patient samples affected until error detection (ANPed). RESULTS: Comparison of the strategies demonstrated that protocols developed using the SA algorithm generally proved superior. Some analytes such as total protein showed considerable improvement, with positive SE equal to 0.8 g/dL detected with an ANPed of 135 samples using the previously published method whereas the SA algorithm detected this SE with an ANPed of 18. Not all analytes demonstrated similar improvement with the SA algorithm. Phosphorus, for instance, demonstrated only minor improvements, with a positive SE of 0.9 mg/dL detected with an ANPed of 34 using the previously published method vs an ANPed of 29 using the SA algorithm. We also demonstrate an example of SE detection in a live environment using the SA algorithm derived MA protocols. CONCLUSIONS: The SA algorithm–developed MA protocols are currently in use in our laboratory and they rapidly detect SE, reducing the number of samples requiring correction and improving patient safety.

Objective: To evaluate the implementation of telecytology in an academic cytology service for immediate assessment of endoscopic ultrasound (EUS) and endobronchoscopic ultrasound (EBUS) fine-needle aspiration (FNA). Study Design: Telecytology was evaluated over a 10-month period. Using an Olympus BX41® microscope and an Olympus DP72® camera with Olympus cellSens® software, real-time dynamic images of air-dried Diff-Quik®-stained smears were transmitted by a cytopathology fellow or cytotechnologist using a secure internet connection. The cytopathologists remotely accessed the real-time images on a computer in their office and rendered immediate assessments. Mean procedure times, and preliminary and final diagnoses were compared between telecytology and conventional on-site evaluation. Results: Two hundred and forty consecutive EUS-FNA and EBUS-FNA procedures with immediate assessments were performed during the evaluation period, of which 158 (66%) utilized telecytology and 82 (34%) did not utilize telecytology. The mean procedure time required for cytotechnologists and cytology fellows was 1.1 h for both conventional on-site and telecytology evaluations. The mean procedure time for cytopathologists was 0.74 h for conventional on-site evaluations and 0.2 h for telecytology. Conclusions: Incorporation of telecytology for immediate assessment of EUS-FNA increased cytopathologist efficiency.

Objectives: Clinical flow cytometry is laborious, time-consuming, and expensive given the need for data review by highly trained personnel such as technologists and pathologists as well as the significant number of normal cases. Given these issues, automation in analysis and diagnosis holds the key to major efficiency gains. The objective was to design an automated pipeline for the diagnosis of B-cell malignancies in flow cytometry and evaluate its performance against our standard clinical diagnostic flow cytometry process.Methods: Using 3,417 cases of peripheral blood data over 6 months from our 10-color B-cell screening tube, we used a newly described method for feature extraction and dimensionality reduction called UMAP on the raw flow cytometry data followed by random forest classification to classify cases without gating on specific population.Results: Our automated classifier was able to achieve greater than 95% accuracy in diagnosing all B-cell malignancies, and even better performance for specific malignancies for which the panel was designed, such as chronic lymphocytic leukemia. By adjusting classifier cutoffs, 100% sensitivity could be achieved with an albeit low 14% specificity. Hypothetically, this would allow 11% of the cases to be autoverified without human intervention.Conclusions: These results suggest that a clinical implementation of this pipeline can greatly assist in quality control, improve turnaround time, and decrease staff workloads.

T cells expressing chimeric antigen receptors (CARs) have shown remarkable therapeutic activity against different types of cancer. However, the wider use of CAR T cells has been hindered by the potential for life-threatening toxicities due to on-target off-tumor killing of cells expressing low amounts of the target antigen. CD229, a signaling lymphocyte-activation molecule (SLAM) family member, has previously been identified as a target for CAR T cell–mediated treatment of multiple myeloma (MM) due to its high expression on the surfaces of MM cells. CD229 CAR T cells have shown effective clearance of MM cells in vitro and in vivo. However, healthy lymphocytes also express CD229, albeit at lower amounts than MM cells, causing their unintended targeting by CD229 CAR T cells. To increase the selectivity of CD229 CAR T cells for MM cells, we used a single amino acid substitution approach of the CAR binding domain to reduce CAR affinity. To identify CARs with increased selectivity, we screened variant binding domains using solid-phase binding assays and biolayer interferometry and determined the cytotoxic activity of variant CAR T cells against MM cells and healthy lymphocytes. We identified a CD229 CAR binding domain with micromolar affinity that, when combined with overexpression of c-Jun, confers antitumor activity comparable to parental CD229 CAR T cells but lacks the parental cells’ cytotoxic activity toward healthy lymphocytes in vitro and in vivo. The results represent a promising strategy to improve the efficacy and safety of CAR T cell therapy that requires clinical validation.

This review will explore the integration of machine learning (ML) techniques to enhance the analysis of increasingly complex and voluminous flow cytometry data, as traditional manual methods are insufficient for handling this data. We attempt to provide a comprehensive introduction to ML in flow cytometry, detailing the transition from manual gating to computational methods and emphasizing the importance of data quality. Key ML techniques are discussed, including supervised learning methods like logistic regression, support vector machines, and neural networks, which rely on labeled data to classify disease states. Unsupervised methods, such as k-means clustering, FlowSOM, UMAP, and t-SNE, are highlighted for their ability to identify novel cell populations without predefined labels. We also delve into newer semi-supervised and weakly supervised methods, which leverage partial labeling to improve model performance. Practical aspects of implementing ML in clinical settings are addressed, including regulatory considerations, data preprocessing, model training, validation, and the importance of generalizability, and we underscore the collaborative effort required among pathologists, data scientists, and laboratory professionals to ensure robust model development and deployment. Finally, we show the transformative potential of ML in flow cytometry in uncovering new biological insights through advanced computational techniques.