CyTOFmerge: integrating mass cytometry data across multiple panels

Computational flow cytometry immunophenotyping at diagnosis is unable to predict relapse in childhood B-cell Acute Lymphoblastic Leukemia

B-cell Acute Lymphoblastic Leukemia is the most prevalent form of childhood cancer, with approximately 15% of patients undergoing relapse after initial treatment. Further advancements depend on novel therapies and more precise risk stratification criteria. In the context of computational flow cytometry and machine learning, this paper aims to explore the potential prognostic value of flow cytometry data at diagnosis, a relatively unexplored direction for relapse prediction in this disease. To this end, we collected a dataset of 252 patients from three hospitals and implemented a comprehensive pipeline for multicenter data integration, feature extraction, and patient classification, comparing the results with existing algorithms from the literature. The analysis revealed no significant differences in immunophenotypic patterns between relapse and non-relapse patients and suggests the need for alternative approaches to handle flow cytometry data in relapse prediction.

A backbone-based flow cytometry approach to decipher regulatory T cell trajectories in the human thymus

Thymus-committed regulatory T cells (Tregs) are essential for immune homeostasis. Recent findings stress their heterogeneity, suggesting possible alternate routes for thymic Treg development with unique features in humans, namely the clear evidence of Treg commitment at the double-positive (DP) stage and the presence of a significant population of CD8 single-positive (SP) FOXP3pos Tregs. Here, we present a dedicated analysis strategy to a spectral flow cytometry-based study of thymus from children and aged adults (≥ 74-years-old), to further elucidate Treg development and heterogeneity in the human thymus. We applied an unsupervised analysis pipeline to data generated from 6 high-dimensional panels, taking advantage of a common backbone of 11 markers, and we were able to map thymocytes along T cell maturation stages. Generating UMAP and FlowSOM cluster coordinates from the backbone, we projected all other markers onto these, characterizing clusters with the information of all markers. Focusing this analysis on events inside a putative total Treg gate, we could portray rarer subsets of human thymic Tregs and investigate their trajectories using pseudotime analysis. We uncover clusters within human DP thymocytes uniquely expressing FOXP3 or CD25, a DP-branching trajectory towards a CD103posCD8SP Tregs endpoint, and define trajectories towards CD4SP Tregs, including towards a cluster of CXCR3posCD4SP Tregs, that may consist of thymic resident or recirculating Tregs, and do not expand in the elderly. Our flow cytometry approach separates Treg populations with likely distinct functions and facilitates the design of future studies to unravel the complexity of human regulatory T cells.

RUCova: Removal of Unwanted Covariance in mass cytometry data

MOTIVATION

High dimensional single-cell mass cytometry data are confounded by unwanted covariance due to variations in cell size and staining efficiency, making analysis, and interpretation challenging.

RESULTS

We present RUCova, a novel method designed to address confounding factors in mass cytometry data. RUCova removes unwanted covariance from measured markers applying multivariate linear regression based on surrogates of sources of unwanted covariance (SUCs) and principal component analysis (PCA). We exemplify the use of RUCova and show that it effectively removes unwanted covariance while preserving genuine biological signals. Our results demonstrate the efficacy of RUCova in elucidating complex data patterns, facilitating the identification of activated signalling pathways, and improving the classification of important cell populations such as apoptotic cells. By providing a robust framework for data normalization and interpretation, RUCova enhances the accuracy and reliability of mass cytometry analyses, contributing to advances in our understanding of cellular biology and disease mechanisms.

AVAILABILITY AND IMPLEMENTATION

The R package is available on https://github.com/molsysbio/RUCova. Detailed documentation, data, and the code required to reproduce the results are available on https://doi.org/10.5281/zenodo.10913464.

FlowAtlas: an interactive tool for high-dimensional immunophenotyping analysis bridging FlowJo with computational tools in Julia

As the dimensionality, throughput and complexity of cytometry data increases, so does the demand for user-friendly, interactive analysis tools that leverage high-performance machine learning frameworks. Here we introduce FlowAtlas: an interactive web application that enables dimensionality reduction of cytometry data without down-sampling and that is compatible with datasets stained with non-identical panels. FlowAtlas bridges the user-friendly environment of FlowJo and computational tools in Julia developed by the scientific machine learning community, eliminating the need for coding and bioinformatics expertise. New population discovery and detection of rare populations in FlowAtlas is intuitive and rapid. We demonstrate the capabilities of FlowAtlas using a human multi-tissue, multi-donor immune cell dataset, highlighting key immunological findings. FlowAtlas is available at https://github.com/gszep/FlowAtlas.jl.git.

Review of flow cytometry as a tool for cell and gene therapy

Quality control testing and analytics are critical for the development and manufacture of cell and gene therapies, and flow cytometry is a key quality control and analytical assay that is used extensively. However, the technical scope of characterization assays and safety assays must keep apace as the breadth of cell therapy products continues to expand beyond hematopoietic stem cell products into producing novel adoptive immune therapies and gene therapy products.  Flow cytometry services are uniquely positioned to support the evolving needs of cell therapy facilities, as access to flow cytometers, new antibody clones and improved fluorochrome reagents becomes more egalitarian. This report will outline the features, logistics, limitations and the current state of flow cytometry within the context of cellular therapy.

Cytomulate: accurate and efficient simulation of CyTOF data

Recently, many analysis tools have been devised to offer insights into data generated via cytometry by time-of-flight (CyTOF). However, objective evaluations of these methods remain absent as most evaluations are conducted against real data where the ground truth is generally unknown. In this paper, we develop Cytomulate, a reproducible and accurate simulation algorithm of CyTOF data, which could serve as a foundation for future method development and evaluation. We demonstrate that Cytomulate can capture various characteristics of CyTOF data and is superior in learning overall data distributions than single-cell RNA-seq-oriented methods such as scDesign2, Splatter, and generative models like LAMBDA.

Combining Mass Cytometry Data by CyTOFmerge Reveals Additional Cell Phenotypes in the Heterogeneous Ovarian Cancer Tumor Microenvironment: A Pilot Study

The prognosis of high-grade serous ovarian carcinoma (HGSOC) is poor, and treatment selection is challenging. A heterogeneous tumor microenvironment (TME) characterizes HGSOC and influences tumor growth, progression, and therapy response. Better characterization with multidimensional approaches for simultaneous identification and categorization of the various cell populations is needed to map the TME complexity. While mass cytometry allows the simultaneous detection of around 40 proteins, the CyTOFmerge MATLAB algorithm integrates data sets and extends the phenotyping. This pilot study explored the potential of combining two datasets for improved TME phenotyping by profiling single-cell suspensions from ten chemo-naïve HGSOC tumors by mass cytometry. A 35-marker pan-tumor dataset and a 34-marker pan-immune dataset were analyzed separately and combined with the CyTOFmerge, merging 18 shared markers. While the merged analysis confirmed heterogeneity across patients, it also identified a main tumor cell subset, additionally to the nine identified by the pan-tumor panel. Furthermore, the expression of traditional immune cell markers on tumor and stromal cells was revealed, as were marker combinations that have rarely been examined on individual cells. This study demonstrates the potential of merging mass cytometry data to generate new hypotheses on tumor biology and predictive biomarker research in HGSOC that could improve treatment effectiveness.

Merging and imputation of flow cytometry data: A critical assessment

Although most modern techniques and analysis methods in multiparameter flow cytometry (MFC) allow for increased dimensionality for the characterization and quantification of cell populations, most MFC applications depend on flow cytometers measuring relatively small (<16) numbers of parameters. When more markers than the available parameters need to be acquired, these are commonly distributed over multiple independent measurements that include a backbone of common markers. Several methods have been proposed to impute values for combinations of markers that were not measured simultaneously. These imputation methods are frequently used without proper validation and knowledge of their effects on data analysis. We evaluated the performance of existing imputation software (Infinicyt, CyTOFmerge, CytoBackBone, and cyCombine) in approximating known measured expression data in terms of similarity in visual appearance, cell expression, and gating in different datasets by splitting MFC samples into separate measurements with partially overlapping markers and re-calculating missing marker expression. Out of the assessed packages, CyTOFmerge showed the most accurate approximation of the known expression in terms of similar expression values and concordance with manual gating, with a mean F-score between 0.53 and 0.87 when retrieving cell populations in different datasets. Performance remained inadequate for all methods, with only limited similarity at the cell level. In conclusion, the use of imputed MFC data should take such limitations into account and include independent validation of results to justify conclusions.

Inferring direction of associations between histone modifications using a neural processes-based framework

Current technologies do not allow predicting interactions between histone post-translational modifications (HPTMs) at a system-level. We describe a computational framework, imputation-followed-by-inference, to predict directed association between two HPTMs using EpiTOF, a mass cytometry-based platform that allows profiling multiple HPTMs at a single-cell resolution. Using EpiTOF profiles of >55,000,000 peripheral mononuclear blood cells from 158 healthy human subjects, we show that neural processes (NP) have significantly higher accuracy than linear regression and k-nearest neighbors models to impute the abundance of an HPTM. Next, we infer the direction of association to show we recapitulate known HPTM associations and identify several previously unidentified ones in healthy individuals. Using this framework in an influenza vaccine cohort, we identify changes in associations between 6 pairs of HPTMs 30 days following vaccination, of which several have been shown to be involved in innate memory. These results demonstrate the utility of our framework in identifying directed interactions between HPTMs.

CyTOF® for the Masses

Mass cytometry has revolutionized immunophenotyping, particularly in exploratory settings where simultaneous breadth and depth of characterization of immune populations is needed with limited samples such as in preclinical and clinical tumor immunotherapy. Mass cytometry is also a powerful tool for single-cell immunological assays, especially for complex and simultaneous characterization of diverse intratumoral immune subsets or immunotherapeutic cell populations. Through the elimination of spectral overlap seen in optical flow cytometry by replacement of fluorescent labels with metal isotopes, mass cytometry allows, on average, robust analysis of 60 individual parameters simultaneously. This is, however, associated with significantly increased complexity in the design, execution, and interpretation of mass cytometry experiments. To address the key pitfalls associated with the fragmentation, complexity, and analysis of data in mass cytometry for immunologists who are novices to these techniques, we have developed a comprehensive resource guide. Included in this review are experiment and panel design, antibody conjugations, sample staining, sample acquisition, and data pre-processing and analysis. Where feasible multiple resources for the same process are compared, allowing researchers experienced in flow cytometry but with minimal mass cytometry expertise to develop a data-driven and streamlined project workflow. It is our hope that this manuscript will prove a useful resource for both beginning and advanced users of mass cytometry.

UMAP Based Anomaly Detection for Minimal Residual Disease Quantification within Acute Myeloid Leukemia

Leukemia is the most frequent malignancy in children and adolescents, with acute lymphoblastic leukemia (ALL) and acute myeloid leukemia (AML) as the most common subtypes. Minimal residual disease (MRD) measured by flow cytometry (FCM) has proven to be a strong prognostic factor in ALL as well as in AML. Machine learning techniques have been emerging in the field of automated MRD quantification with the objective of superseding subjective and time-consuming manual analysis of FCM-MRD data. In contrast to ALL, where supervised multi-class classification methods have been successfully deployed for MRD detection, AML poses new challenges: AML is rarer (with fewer available training data) than ALL and much more heterogeneous in its immunophenotypic appearance, where one-class classification (anomaly detection) methods seem more suitable. In this work, a new semi-supervised approach based on the UMAP algorithm for MRD detection utilizing only labels of blast free FCM samples is presented. The method is tested on a newly gathered set of AML FCM samples and results are compared to state-of-the-art methods. We reach a median F1-score of 0.794, while providing a transparent classification pipeline with explainable results that facilitates inter-disciplinary work between medical and technical experts. This work shows that despite several issues yet to overcome, the merits of automated MRD quantification can be fully exploited also in AML.

Knowledge transfer to enhance the performance of deep learning models for automated classification of B cell neoplasms

Multi-parameter flow cytometry (MFC) is a cornerstone in clinical decision making for leukemia and lymphoma. MFC data analysis requires manual gating of cell populations, which is time-consuming, subjective, and often limited to a two-dimensional space. In recent years, deep learning models have been successfully used to analyze data in high-dimensional space and are highly accurate. However, AI models used for disease classification with MFC data are limited to the panel they were trained on. Thus, a key challenge in deploying AI into routine diagnostics is the robustness and adaptability of such models. This study demonstrates how transfer learning can be applied to boost the performance of models with smaller datasets acquired with different MFC panels. We trained models for four additional datasets by transferring the features learned from our base model. Our workflow increased the model's overall performance and, more prominently, improved the learning rate for small training sizes.

•

Device capabilities and diagnostic approaches differ greatly in lymphoma MFC panels

•

Single laboratories generate too little data to train an AI model with high accuracy

•

Transfer learning across panels increases classification performance significantly

•

Merging MFC data from multiple tubes per sample increases the model's transferability

Multi-parameter flow cytometry (MFC) is a critical tool in leukemia and lymphoma diagnostics. Advances in cytometry technology and diagnostic standardization efforts have led to an ever-increasing volume of data, presenting an opportunity to use artificial intelligence (AI) in diagnostics. However, the MFC protocol is prone to changes depending on the diagnostic workflow and the available cytometer. The changes to the MFC protocol limit the deployment of AI in routine diagnostics settings. We present a workflow that allows existing AI to adapt to multiple MFC protocols. We combine transfer learning (TL) with MFC data merging to increase the robustness of AI. Our results show that TL improves the performance of AI and allows models to achieve higher performance with less training data. This gain in performance for smaller training data allows for an already deployed AI to adapt to changes without the need for retraining a new model that requires more training data.

High-throughput single-cell quantification of hundreds of proteins using conventional flow cytometry and machine learning

Modern immunologic research increasingly requires high-dimensional analyses to understand the complex milieu of cell types that comprise the tissue microenvironments of disease. To achieve this, we developed Infinity Flow combining hundreds of overlapping flow cytometry panels using machine learning to enable the simultaneous analysis of the coexpression patterns of hundreds of surface-expressed proteins across millions of individual cells. In this study, we demonstrate that this approach allows the comprehensive analysis of the cellular constituency of the steady-state murine lung and the identification of previously unknown cellular heterogeneity in the lungs of melanoma metastasis–bearing mice. We show that by using supervised machine learning, Infinity Flow enhances the accuracy and depth of clustering or dimensionality reduction algorithms. Infinity Flow is a highly scalable, low-cost, and accessible solution to single-cell proteomics in complex tissues.

Restoration of NK Cell Cytotoxic Function With Elotuzumab and Daratumumab Promotes Elimination of Circulating Plasma Cells in Patients With SLE

Systemic lupus erythematosus (SLE) is a multisystem autoimmune disease characterized by multiple cellular and molecular dysfunctions of the innate and adaptive immunity. Cytotoxic function of NK cells is compromised in patients with SLE. Herein, we characterized the phenotypic alterations of SLE NK cells in a comprehensive manner to further delineate the mechanisms underlying the cytotoxic dysfunction of SLE NK cells and identify novel potential therapeutic targets. Therefore, we examined PBMC from SLE patients and matched healthy controls by single-cell mass cytometry to assess the phenotype of NK cells. In addition, we evaluated the cell function of NK cells (degranulation and cytokine production) and the killing of B cell subpopulations in a B cell-NK cell in vitro co-culture model. We found that SLE NK cells expressed higher levels of CD38 and were not able to adequately upregulate SLAMF1 and SLAMF7 following activation. In addition, ligation of SLAMF7 with elotuzumab or of CD38 with daratumumab on SLE NK cells enhanced degranulation of both healthy and SLE NK cells and primed them to kill circulating plasma cells in an in vitro co-culture system. Overall, our data indicated that dysregulated expression of CD38, SLAMF1 and SLAMF7 on SLE NK cells is associated with an altered interplay between SLE NK cells and plasma cells, thus suggesting their contribution to the accumulation of (auto)antibody producing cells. Accordingly, targeting SLAMF7 and CD38 may represent novel therapeutic approaches in SLE by enhancing NK cell function and promoting elimination of circulating plasma cell.

Phenotypic Characterization by Mass Cytometry of the Microenvironment in Ovarian Cancer and Impact of Tumor Dissociation Methods

Simple Summary

High-grade serous ovarian cancer (HGSOC) is the deadliest gynecological malignancy. Despite increasing research on HGSOC, biomarkers for individualized selection of therapy are scarce. In this study, we develop a multiparametric mass cytometry antibody panel to identify differences in the cellular composition of the microenvironment of tumor tissues dissociated to single-cell suspensions. We also investigate how dissociation methods impact results. Application of our antibody panel to HGSOC tissues showed its ability to identify established main cell subsets and subpopulations of these cells. Comparisons between dissociation methods revealed differences in cell fractions for one immune, two stromal, and three tumor cell subpopulations, while functional marker expression was not affected by the dissociation method. The interpatient disparities identified in the tumor microenvironment were more significant than those identified between differently dissociated tissues from one patient, indicating that the panel facilitates the mapping of individual tumor microenvironments in HGSOC patients.

Abstract

Improved molecular dissection of the tumor microenvironment (TME) holds promise for treating high-grade serous ovarian cancer (HGSOC), a gynecological malignancy with high mortality. Reliable disease-related biomarkers are scarce, but single-cell mapping of the TME could identify patient-specific prognostic differences. To avoid technical variation effects, however, tissue dissociation effects on single cells must be considered. We present a novel Cytometry by Time-of-Flight antibody panel for single-cell suspensions to identify individual TME profiles of HGSOC patients and evaluate the effects of dissociation methods on results. The panel was developed utilizing cell lines, healthy donor blood, and stem cells and was applied to HGSOC tissues dissociated by six methods. Data were analyzed using Cytobank and X-shift and illustrated by t-distributed stochastic neighbor embedding plots, heatmaps, and stacked bar and error plots. The panel distinguishes the main cellular subsets and subpopulations, enabling characterization of individual TME profiles. The dissociation method affected some immune (n = 1), stromal (n = 2), and tumor (n = 3) subsets, while functional marker expressions remained comparable. In conclusion, the panel can identify subsets of the HGSOC TME and can be used for in-depth profiling. This panel represents a promising profiling tool for HGSOC when tissue handling is considered.

Stabilization of Human Whole Blood Samples for Multicenter and Retrospective Immunophenotyping Studies

Whole blood is often collected for large-scale immune monitoring studies to track changes in cell frequencies and responses using flow (FC) or mass cytometry (MC). In order to preserve sample composition and phenotype, blood samples should be analyzed within 24 h after bleeding, restricting the recruitment, analysis protocols, as well as biobanking. Herein, we have evaluated two whole blood preservation protocols that allow rapid sample processing and long-term stability. Two fixation buffers were used, Phosphoflow Fix and Lyse (BD) and Proteomic Stabilizer (PROT) to fix and freeze whole blood samples for up to 6 months. After analysis by an 8-plex panel by FC and a 26-plex panel by MC, manual gating of circulating leukocyte populations and cytokines was performed. Additionally, we tested the stability of a single sample over a 13-months period using 45 consecutive aliquots and a 34-plex panel by MC. We observed high correlation and low bias toward any cell population when comparing fresh and 6 months frozen blood with FC and MC. This correlation was confirmed by hierarchical clustering. Low coefficients of variation (CV) across studied time points indicate good sample preservation for up to 6 months. Cytokine detection stability was confirmed by low CVs, with some differences between fresh and fixed conditions. Thirteen months regular follow-up of PROT samples showed remarkable sample stability. Whole blood can be preserved for phenotyping and cytokine-response studies provided the careful selection of a compatible antibody panel. However, possible changes in cell morphology, differences in antibody affinity, and changes in cytokine-positive cell frequencies when compared to fresh blood should be considered. Our setting constitutes a valuable tool for multicentric and retrospective studies. © 2020 International Society for Advancement of Cytometry.

Unraveling the Complexity of the Cancer Microenvironment With Multidimensional Genomic and Cytometric Technologies

Cancers are characterized by extensive heterogeneity that occurs intratumorally, between lesions, and across patients. To study cancer as a complex biological system, multidimensional analyses of the tumor microenvironment are paramount. Single-cell technologies such as flow cytometry, mass cytometry, or single-cell RNA-sequencing have revolutionized our ability to characterize individual cells in great detail and, with that, shed light on the complexity of cancer microenvironments. However, a key limitation of these single-cell technologies is the lack of information on spatial context and multicellular interactions. Investigating spatial contexts of cells requires the incorporation of tissue-based techniques such as multiparameter immunofluorescence, imaging mass cytometry, or in situ detection of transcripts. In this Review, we describe the rise of multidimensional single-cell technologies and provide an overview of their strengths and weaknesses. In addition, we discuss the integration of transcriptomic, genomic, epigenomic, proteomic, and spatially-resolved data in the context of human cancers. Lastly, we will deliberate on how the integration of multi-omics data will help to shed light on the complex role of cell types present within the human tumor microenvironment, and how such system-wide approaches may pave the way toward more effective therapies for the treatment of cancer.