Flow cytometry is an essential component of routine hematological lab testing. Many computational methods have been proposed for the analysis of flow cytometry data, but most have focused on supervised learning for just one or a few specific disorders. To maximize clinical utility, we develop a method that enables identification of multiple common disorders and quality indicators. Our method includes a self-supervised pretraining component as well as a new, transformer-based model architecture. The self-supervised training algorithm is based on the DINO method while the model architecture is a relatively simple transformer encoder stack that includes a class (CLS) token, similar to BERT or vision-transformer models. Using a dataset of 52,625 samples obtained during routine clinical testing at our laboratory, we show that our pretraining method develops informative tube-level representations that clearly separate important diagnostic classes. We then evaluate performance on multiple downstream tasks, including sample viability estimation and five common hematological disorders. We compare our method to self-organizing maps, convolutional neural networks, attention-based multiple-instance learning models, and two varieties of set-transformer-based models, and demonstrate that our method delivers higher classification performance than other approaches.
Hereditary spherocytosis (HS) is the most common congenital red blood cell membrane disorder, characterized by structural protein defects that lead to hemolytic anemia. Although several diagnostic tests exist, including osmotic fragility tests (OFTs), acidified glycerol lysis test (AGLT), and the EMA-binding test (EMA), each presents specific limitations regarding sensitivity, specificity, or technical requirements. Flow cytometric osmotic fragility testing (OFT-FCM) emerges as a promising complementary assay, offering a standardized workflow and rapid turnaround time. We conducted a retrospective study including 106 subjects (20 HS patients and 86 healthy controls) recruited at Hospital Clínic de Barcelona between September 2024 and September 2025. Clinical and laboratory data were collected, and all participants underwent OFT, AGLT, EMA, and OFT-FCM using two acquisition protocols (300 and 214 s). Logistic regression and receiver operating characteristic curve analysis were performed to evaluate diagnostic performance and determine optimal cut-off values. HS patients exhibited significantly altered hematologic parameters compared with controls, including higher reticulocyte counts, red cell distribution width, and mean corpuscular hemoglobin. The EMA-binding test demonstrated high specificity (100%) but lower sensitivity (57.9%). OFT achieved high sensitivity (>97%) but low specificity (<47%). AGLT showed balanced accuracy (sensitivity 68.4%, specificity 96.1%). OFT-FCM yielded areas under the curve of 0.85 for both protocols, with optimal thresholds providing specificities of 95-100% and sensitivities of 57-59%. No significant differences were observed between OFT-FCM and EMA performance. OFT-FCM effectively discriminates HS patients from healthy controls and showed diagnostic performance comparable to EMA and favorable relative to classical OFT and AGLT in this cohort, while offering practical advantages in terms of workflow simplicity and turnaround time, and supporting its use as a complementary flow-cytometric assay within the diagnostic work-up of HS.
Our knowledge of the immune system continues to expand at a rapid pace, and this coupled with technological advances now enables us to interrogate both the breadth and the depth of the immune response at levels without precedent. This has also facilitated rapidly integrating some of this carefully vetted knowledge into clinical practice. Notable examples of these advances include successfully harnessing the therapeutic potential of the immune system (immunotherapy), as well as an expanding menu of clinical flow-cytometry laboratory tests to assess the phenotype and function of the cellular immune response. This has also given rise to an emerging sub-discipline called "Immune-Health", with its premise undergirded by the notion that the surveillance capacity and sentinel nature of the immune response might enable the immune system to serve as a reliable barometer of overall health of the individual. At its core, immune-health entails defining baseline immune characteristics for each individual so that perturbations in this baseline signature can serve as clinically actionable biomarkers that might predict the onset, help monitor the progression and potentially mitigate the effects of the underlying disease process. Defining appropriate reference-ranges (RR) for key cellular immune parameters constitutes one of the essential building-blocks of the concept of immune-health. Establishing pediatric RR for cellular correlates of immune-health and disease is a time-consuming and labor-intensive process, and consequently only a select few specialty laboratories at some children's hospitals (with a well-established immunodeficiency/immunedysregulation clinical service) and a couple of large national reference laboratories in the United States (US) have invested their time and effort into this endeavor. Furthermore, the lack of standardization in the definition of immune subsets has also complicated this effort. In 2012, the Human Immunophenotyping Consortium (HIPC) established by the National Institute of Allergy and Infectious Diseases (NIAID)-Division of Allergy, Immunology and Transplantation, published a benchmark study that attempted to standardize the definitions for several cellular immune parameters. In general, Europe has led the charge in advancing these standardization efforts, while similar efforts in the US have been rather sparse, and currently there's a marked paucity of US based studies describing the establishment of pediatric reference ranges (RR) for T and B cell subsets based on the HIPC standardization initiative. In this report, we describe the results of the endeavor, at a large, free-standing children's hospital in the US with a busy clinical immunology service, to establish pediatric reference ranges for naïve and memory T and B cell subsets largely adapted from the subset definitions outlined by the HIPC.
X-linked agammaglobulinemia (XLA) is most often caused by impaired Bruton's tyrosine kinase (BTK) function. Rapid laboratory support is needed when genetic findings are inconclusive or pending. We developed a flow-cytometric assay to quantify intracellular BTK protein in monocytes, aiding the diagnosis of XLA. A stain index (SI) reference range was established using healthy donors (n = 25; SI median 1.14, range 1.01-1.66). ROC analysis defined a diagnostic cutoff of SI = 1.79, providing 95% sensitivity and 100% specificity compared to patients with pathogenic BTK variants (n = 40). Using this cutoff, abnormal BTK expression was detected in 38/40 genetically confirmed XLA patients. All patients with variants of uncertain/likely pathogenic significance (n = 17) had SI values above the cutoff. Bone marrow analysis in a patient subset showed a developmental arrest at the Pre-BI stage. The assay offers a rapid, clinically applicable readout to support XLA diagnosis and carrier evaluation in females, effectively complementing genetic testing.
