Manual Gating Research Articles

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

SummaryMulti-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.

Transformation of multicolour flow cytometry data with OTflow prevents misleading multivariate analysis results and incorrect immunological conclusions.

The rapid evolution of the flow cytometry field, currently allowing the measurement of 30-50 parameters per cell, has led to a marked increase in deep multivariate information. Manual gating is insufficient to extract all this information. Therefore, multivariate analysis (MVA) methods have been developed to extract information and efficiently analyze the high-density multicolour flow cytometry (MFC) data. To aid interpretation, MFC data are often logarithmically transformed before MVA. We studied the consequences of different transformations of flow cytometry data in datasets containing negative intensities caused by background subtractions and spreading error, as logarithmic transformation of negative data is impossible. Transformations such as logicle or hyperbolic arcsine transformations allow linearity around zero, whereas higher (positive and negative) intensities are logarithmically transformed. To define the linear range, a parameter (or cofactor) must be chosen. We show how the chosen transformation parameter has great impact on the MVA results. In some cases, peak splitting is observed, producing two distributions around zero in an actual homogeneous population. This may be misinterpreted as the presence of multiple cell populations. Moreover, when performing arbitrary transformation before MVA analysis, biologically relevant and statistically significant information might be missed. We present a new algorithm, Optimal Transformation for flow cytometry data (OTflow), which uses various statistical methods to optimally choose the parameter of the transformation and prevent artifacts such as peak splitting. Arbitrary or unconsidered transformation can lead to wrong conclusions for the MVA cluster methods, dimensionality reduction methods, and classification methods. We recommend transformation of flow cytometry data by using OTflow-defined parameters estimated per channel, in order to prevent peak splitting and other artifacts in the data.

Measurement of hematopoietic progenitor cells using XN2000 hematology analyzer.

Stem cell enumeration by the hematopoietic progenitor cells (HPC) mode is a novel method available from Sysmex XN2000 hematology analyzer. A small amount of blood (190μL) is required, and the results are available in a few minutes without manual gating or presample treatment. The present study compares stem cell measurements using XN2000 analyzer HPC mode and FC500 flow cytometry analyzer using peripheral blood (PB) specimens and apheresis products. In this prospective study, CD34-positive cell counts were enumerated using an FC500 flow cytometry analyzer and compared with XN2000 Sysmex analyzer (XN-HPC mode) in the same samples. Results were compared using Bland-Altman plots. A total of 103 samples were used. In the PB samples, the median HPC count and CD34-positive cells were 83.5×106 /L and 78.0×106 /L, respectively. The mean Bland-Altman difference was 4.5×106 /L (Limits: -51.7 to 60.7×106 /L), with a Pearson's correlation of 0.79. In the apheresis products, the median HPC count and CD34-positive cells were 1468×106 /L (IQR: 1049 - 1960×106 /L) and 1327×106 /L (IQR: 910 - 2001×106 /L), respectively. The mean Bland-Altman difference was 179.0×106 /L (Limits: -2022.2 - 2380.2×106 /L), with a Pearson's correlation of 0.58. The XN-HPC mode has an excellent correlation and minimal disagreement for stem cell enumeration in PB compared with flow cytometry and could replace it. There is high disagreement in apheresis products, and therefore, the XN-HPC mode cannot be recommended.

Immune Profiling Mass Cytometry Assay Harmonization: Multicenter Experience from CIMAC-CIDC.

The Cancer Immune Monitoring and Analysis Centers - Cancer Immunologic Data Commons (CIMAC-CIDC) Network is supported by the NCI to identify biomarkers of response to cancer immunotherapies across clinical trials using state-of-the-art assays. A primary platform for CIMAC-CIDC studies is cytometry by time of flight (CyTOF), performed at all CIMAC laboratories. To ensure the ability to generate comparable CyTOF data across labs, a multistep cross-site harmonization effort was undertaken. We first harmonized standard operating procedures (SOPs) across the CIMAC sites. Because of a new acquisition protocol comparing original narrow- or new wide-bore injector introduced by the vendor (Fluidigm), we also tested this protocol across sites before finalizing the harmonized SOP. We then performed cross-site assay harmonization experiments using five shared cryopreserved and one lyophilized internal control peripheral blood mononuclear cell (PBMC) with a shared lyophilized antibody cocktail consisting of 14 isotype-tagged antibodies previously validated, plus additional liquid antibodies. These reagents and samples were distributed to the CIMAC sites and the data were centrally analyzed by manual gating and automated methods (Astrolabe). Average coefficients of variation (CV) across sites for each cell population were reported and compared with a previous multisite CyTOF study. We reached an intersite CV of under 20% for most cell subsets, very similar to a previously published study. These results establish the ability to reproduce CyTOF data across sites in multicenter clinical trials, and also highlight the importance of quality control procedures, such as the use of spike-in control samples, for tracking variability in this assay.

Efficiency of Smart Sluice Gate in Agricultural Water Distribution on Paddy (Oryza sativa L.)

The development of irrigation sluice gate is to optimize the use of water in agriculture, the development is carried out on a laboratory scale in the hydraulics and environment laboratory, then the implementation was located at PIAT UGM in block A covering an area of 6.93 ha, Kali Tirto Village. The objective of this research is investigate the efficiency of smart sluices with manual sluices using the Bernoulli equation. Based on the analysis, the efficiency in manual sluice openings with 10 variations obtained an average of 49%, while the smart sluice openings with a 5 cm open scenario, variations in gate openings can be carried out continuously, the average result of the smart sluice efficiency analysis is 72%, so the smart sluice has a greater efficiency value than the manual sluice gate.

Abstract 2279: Improving cell type and state classification in CITE-seq using probabilistic models and hierarchical cell ID matrix

Abstract Accurate classification of cells in a population is important for most immune phenotyping applications. Surface protein marker expression measured with multiplexed platforms like Flow Cytometry and sequential gating strategies are typically employed to classify cells. However, new single-cell RNA sequencing (scRNA-seq) methods that measure RNA expression alone or Cellular Indexing of Transcriptomes and Epitopes by sequencing (CITE-seq) methods that concurrently measure both surface protein and RNA expression in single cells are providing new data and opportunities to develop methods both to classify cells and to assess their states of activation. Traditional manual gating methods using biaxial plots, where one designates a threshold separating a bimodal distribution through visual approximation, are subjective and can vary widely between individuals. Instead of manual, sequential gating for cell classification, we have implemented a probabilistic Gaussian Mixture Model (GMM) and K-means clustering to separate cell populations based on a library of marker thresholds for a variety of cell types present in human peripheral blood mononuclear cells (PBMCs) using a CITE-seq ADT expression matrix that has been preprocessed with centered-log-ratio (CLR) normalization. Furthermore, we coupled this GMM method with a hierarchical approach for cell typing. We have developed a Cell ID Matrix that outlines over 40 different immune cell types using consensus markers applicable to immune cell types across various studies. In this typing method, major cell types are identified following positive and negative marker expression as described in the Cell ID Matrix. Hierarchical gating of cells can then be used to identify subpopulations in a principled manner. The information in the Cell ID matrix is input as a binary matrix for each hierarchy level and is matched to the binarized expression values for classifying the cells. This approach can also be extended to CITE-seq datasets from disassociated tumor cells (DTCs). We conclude that automated cell gating with CITE-seq data may enhance the classification of numerous cell types in addition to significantly decreasing the amount of human intervention and time allocated for gating cells. Identifying cell states is a work in progress and will supplement this technique in the future. Citation Format: Lila Fakharzadeh, Julien Tessier, Shannon McGrath, Emma Wang, Angelique Biancotto, Alexei Protopopov, Jack Pollard, Joon Sang Lee. Improving cell type and state classification in CITE-seq using probabilistic models and hierarchical cell ID matrix [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 2279.

Abstract 164: Development of DeepCytoFlow: A novel deep learning pipeline enabling automated gating of flow cytometry data

Abstract Introduction Following the advent of cellular therapies, flow cytometry has become a central tool for simultaneous assessment of drug levels and disease burden due to its ability for high dimensional cellular analysis. However, identification and monitoring of myriad immune- and leukemic cell phenotypes requires a manual gating process by subject matter experts. Although this process is the current gold standard for flow data analysis, it is slow, tedious, and prone to subjectivity. Herein, we describe our approach to automate flow data analysis using novel strategies that reduces the reliance on clustering methods that are unable to filter challenging noisy data sets inherent in flow data. Methods We developed DeepCytoFlow, a novel deep learning framework adapted for flow cytometry data analysis. DeepCytoFlow utilizes a convoluted neural network structure to analyze flow cytometry data files to generate appropriately gated cell populations in a matter of minutes significantly reducing the time required for manual analysis. The neural networks were first trained using a unique data set generated manually by multiple operators and tested using an independent data set to establish concordance in assessments. In addition, we evaluated the performance of our method using both accuracy and F1-score. Results Our integrated DeepCytoFlow pipeline employing neural network design shows promise on clustering-based analysis methods in terms of accuracy, precision and recall. More specifically, DeepCytoFlow is capable of correctly classifying various immune cell lineages (e.g., CD4 and CD8 T-cells). However, a limitation was observed on its ability to classify rare cell populations with a very limited prevalence. In an effort to improve performance, we are currently designing a new network architecture, and increasing our training and test dataset's size and quality. Conclusions DeepCytoFlow provides proof of concept of the applicability of deep learning for automation of gating in flow cytometry. The high adaptability of deep learning to a large set of problems suggests that this technique can be broadly applied to a larger set of biomarkers for the purpose for automated gating. With additional improvements, we believe DeepCytoFlow can enable a fast, accurate, and robust platform for analysts to perform consistent and, repeatable flow analysis for biomarker monitoring in clinical trials investigating novel immunotherapies. Citation Format: Arjun Chandran, Jennifer Tsau, Dalia Gaddis, Shyam Sarikonda, Naveen Dakappagari, Christian Laing, Swetha Ann Thomas. Development of DeepCytoFlow: A novel deep learning pipeline enabling automated gating of flow cytometry data [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 164.

Altered Peripheral Immune Phenotypes in Mantle Cell Lymphoma Survivors

Background: Advances in lymphoma therapy have led to improved survival rates for both Hodgkin and Non-Hodgkin lymphoma patients. This has resulted in a large and ever increasing population of lymphoma survivors. Depending on their disease and treatment history, these patients are known to be at risk for late cardiopulmonary effects, endocrine abnormalities, second malignancies, and psychosocial sequelae. However, little is known about the immune health of lymphoma survivors. Lymphoma itself perturbs the immune system. Additionally, lymphoma patients receive many immune-modulating therapies, such as rituximab, B- and T-cell signaling inhibitors, and stem cell transplants. We hypothesize that these interventions produce detectable long-term immune alterations in lymphoma survivors.Methods: To determine whether altered immune phenotypes are detectable in lymphoma survivors, we used mass cytometry by time of flight (CyTOF) to broadly profile peripheral blood mononuclear cells (PBMCs) of a homogenously treated population of 8 mantle cell lymphoma survivors and compared these to PBMCs from 9 age-matched healthy donors. The data was normalized and analyzed visually using manual gating and viSNE for dimensionality reduction and subsequently with the CITRUS algorithm for unbiased statistical comparisons.Results: All of the mantle cell lymphoma patients had undergone intensive induction chemotherapy followed by autologous stem cell transplant. Peripheral blood was collected on average 2.3 years after the end of therapy from patients confirmed to be free of disease. There were no significant differences between the proportions of T-cells, B-cells, monocytes, NK cells, or NK/T-cells among the peripheral blood mononuclear cells (PBMCs) of lymphoma survivors and healthy donors, but the heterogeneity among patients was notably higher than among controls. On average, lymphoma survivors had significantly fewer CD4 T-cells compared to healthy donors (p=0.006; Figure 1A) and these cells were skewed toward a memory phenotype compared to the CD4+ population in controls (p=0.0001 comparing proportion naïve, p=0.0034 comparing proportion effector memory; Figure 1B), a finding that correlated with time from end of therapy. CD8+ T-cells of lymphoma survivors were more frequently CD28-CD57+ (p=0.028), a phenotype that has been associated with chronic antigen experience. Unbiased analysis using CITRUS confirmed the above findings and additionally found IL7R-alpha to be differentially expressed between patient and control T-cells. This finding was confirmed by manual gating, where IL7R-alpha expression was found to be significantly decreased on both CD4+ T-cells (p=0.0007) and CD8+ T-cells (p=0.0001) in patients compared to controls (Figure 1C).Conclusions: Survivors of mantle cell lymphoma have evidence of altered peripheral blood immunophenotypes years after completing treatment, possibly reflecting long-term effects of their exposure to lymphoma and its immune modulating treatments. Such derangements may lead to clinical immune dysfunction, a focus of planned future investigations. [Display omitted] DisclosuresNo relevant conflicts of interest to declare.

Gaussian Mixture Models and Machine Learning Predict Megakaryocytic Growth and Differentiation Potential Ex Vivo

The ability to analyze single cells via flow cytometry has resulted in a wide range of biological and medical applications, such as monitoring populations of developing immune and hematopoietic cells over time. Currently, there is no established framework to compare and interpret time-series flow cytometry data for cellular engineering applications. Manual analysis of temporal trends is time-consuming and subjective for large-scale datasets. We resolved this bottleneck by developing TEmporal Gaussian Mixture models (TEGM), an unbiased computational strategy to quantify and predict temporal trends of developing cell subpopulations indicative of cellular phenotype. We validated its utility by accurately predicting peak CD41+/CD42+ % of megakaryocyte (MK) cultures derived from human CD34+-selected umbilical cord blood units (CB).TEGM applies Gaussian mixture modeling for feature extraction and gradient boosted trees for prediction. Our modeling strategy allows for data to be gated and clustered on multi-dimensional planes, enabling rapid processing of flow cytometry data. TEGM enables the extraction of subtle features, such as the dispersion and rate of change of surface marker expression for each subpopulation over time. These critical, yet hard-to-discern, features are fed into machine-learning algorithms that predict underlying cellular classes and phenotypes. Our framework can be flexibly applied to conventional flow cytometry sampling schemes, and allows for faster and more consistent processing of time-series flow cytometry data.As proof-of-concept, we applied our method to the analysis of ex vivo MK differentiation and maturation of hematopoietic cells from donors with varying potential to generate CD41+/CD42+ cells. Our computational approach consists of three major steps: preprocessing, feature extraction, and prediction (Figure 1). We illustrate the major steps of the computational approach by predicting peak CD41+/CD42+ % MK maturation of CD34+-selected CB cells from 16 independent donors. Cells were cultured over a 19-day multi-phase differentiation culture, consisting of a pre-expansion phase and a differentiation phase. The novel dataset comprised 720 measurements from 80 perturbations of 16 individual donors, with 9 time-point measurements sampled every 2-5 days for each donor. In the preprocessing step, we filtered out non-uniform events, margin events, and doublets. We then constructed an automated gating strategy to extract surface marker expression of various clusters of DAPIlow/CD41+ MK cells. Notably, we demonstrated that estimation of the CD34+ and CD42+ % was within 1% of manual gating estimates, thus illustrating the consistency and accuracy of the technique. We then performed feature extraction for each flow cytometry time-course dataset on several descriptors, such as growth rate, viability, production, percentage positivity of each surface marker, covariance of mean fluorescence intensity, rate of change, and bifurcation of each subpopulation. A gradient boosted tree model was trained using an explanatory matrix describing early characteristics (Day 0-Day 9) and tested to predict peak CD41+/CD42+ marker expression, which typically occurs on Day 14 to Day 17.Overall, we identified several influential early culture factors that are predictive of peak CD42+ % expression. We showed that CD41+ % on Day 5 and Day 7 is highly predictive of peak CD42+ % expression. Cell viability and CD34+ % were comparatively less predictive of peak CD42+ %. Based on our analysis, we were able to identify the best performing cultures with high sensitivity and specificity (AUROC = 0.92, where 1 denotes perfect accuracy). Predicted and actual CD41+/CD42+ responses were highly correlated (p=7.4e09, R = 0.87) using 3 independently selected partitions of test/training sets of our data. We demonstrated that the resulting model captured apparent cell development processes represented by cell surface markers including CD34, CD41, and CD42, and was able to predict MK differentiation and maturation potential of a given CB unit in culture. Identifying CB units with high and low MK potential early in the 19-day culture process can save expensive resources and time, and provides the potential to intervene during the culture process.Figure 1. Manual gating strategy (left) vs. TEGM gating strategy (right) for time-series flow cytometry data. [Display omitted] DisclosuresNo relevant conflicts of interest to declare.

An Artificial Intelligence Approach for B Lymphoblastic Leukemia Minimal Residual Disease Detection and Clinical Prognosis Prediction Using Flow Cytometry Data

IntroductionDetection of minimal residual disease (MRD) of B-cell acute lymphoblastic leukemia (B-ALL) measured by flow cytometry (FC) has been shown to be an important prognostic indicator in B-ALL. However, the current analysis is based on the manual gating by the reviewers; therefore, objectivity could be compromised, and decentralization of the B-ALL MRD detection is difficult to be established. We then aimed to develop an objective diagnostic tool via artificial intelligence (AI) technique for FC data interpretation in order to provide efficient and objective B-ALL MRD detection.MethodsRetrospective FC data of patients with adult B-ALL from 2009 to 2016 in National Taiwan University Hospital were enrolled. A total of 1,073 unique bone marrow FC samples were collected for analysis. Each of the samples had a 10-tube test done on 100,000 cells - each cell was measured in 6 fluorescent channels (FSC, SSC, FITC, PE, PerCP, APC) within one tube.The flow conclusion of each FC sample was categorized into “positive MRD” and “no MRD” groups according to previous manual interpretation, and the whole dataset was randomly divided into 3 groups: Training set (n=718), Accuracy set (n=141) and Validation set (n=178). The Training set was used to train the algorithm, and the Accuracy set was used to evaluate the accuracies achieved and optimize algorithmic parameters. The final optimized algorithm was then evaluated blindly in Validation set, with the concordance rate comparing the manual and the AI conclusions in distinguishing “positive MRD” from “no MRD”.For the algorithm development, the recorded numerical values of the 6 fluorescent channels (FSC, SSC, FITC, PE, PerCP, APC) of each tube (100,000 cells) were considered as raw feature attributes. The probabilistic distributions of these attributes were first modeled as multivariate Gaussian mixture model using a sub-dictionary learning approach. Then a probabilistic derivation was exploited to compute per-sample L2-normalized vectorized representations. Lastly, representations of each tube were then concatenated to be the final feature input to the supervised machine learning classifier, in this case, a support vector machine with linear kernel. In addition, ANOVA-based feature selection was also conducted throughout the experiments.For outcome correlation analysis, we included 35 adult B-ALL patients with available FC data after induction therapy, record their clinical parameters and measured their overall survival (OS) and progression-free survival (PFS), with median follow-up 40.0 months. The Kaplan-Meier curves were constructed to estimated OS and RFS.ResultsThe primary algorithm concordance estimated using the Accuracy set was 90.8% and the area under the receiver operating characteristic curve (AUC) is 0.944 (Table1). The calibrated algorithm, adjusted with training in Accuracy set, can produce a concordance rate of 88.2% with manual interpretation on Validation set, and the AUC is 0.919 (Table 1).It is noteworthy that it only took 7 seconds to analyze one FC sample using the AI algorithm while it took approximately 15 to 30 minutes for manual gating by trained professionals. Furthermore, the concordance of the algorithm developed from 10 tube-data remained almost unchanged when the tube number decreased. In fact, algorithm from 1 tube still yields an AUC of 0.874 (vs. 0.919 from 12 tubes), indicating that AI algorithm may reduce laboratory testing currently applied in standard FC processing.For outcome correlation, B-ALL patients with no MRD by AI algorithm had significant longer OS compared to those with positive MRD (NR vs 20.0 months, p=0.0051). Moreover, patients with no MRD also had significant longer RFS compared to those with positive MRD (8.9 vs 5.2 months, p=0.019), (Figure.1).ConclusionsThis is the first study using AI approach to develop FC data diagnostic algorithms with prognosis implications for B-ALL, demonstrating the advantages of reliability, high efficacy and time saving. Incorporating other test results into AI algorithm may lead to more precise prognostic prediction in the future and potentially guide therapeutic strategy for B-ALL. [Display omitted] DisclosuresKo:Celgene International Sàrl: Research Funding. Li:Celgene International Sàrl: Research Funding. Tien:Celgene International Sàrl: Research Funding. Tang:Celgene International Sàrl: Research Funding.

An Artificial Intelligence Approach for Predicting Allogeneic Hematopoietic Stem Cell Transplantation Outcome By Detecting Pre-Transplant Minimal Residual Disease in Acute Myeloid Leukemia Using Flow Cytometry Data

IntroductionDisease status before allogeneic hematopoietic stem cell transplantation (allo-HSCT) is an important prognostic marker for recipients with acute myeloid leukemia (AML), and flow cytometry (FC)-assisted minimal residual disease (MRD) detection is of well-known clinical significance in this setting. However, current interpretation requires experienced manual gating, which is time consuming as well as suffering from inter-physician idiosyncrasies. The reproducibility and objectivity could then be compromised. In this study, we aimed to develop automated FC data interpretation algorithm using artificial intelligence (AI) technology in supporting physicians to conduct rapid and reliable pre-HSCT MRD detection for AML patients.MethodsRetrospective FC data of AML or myelodysplastic syndrome (MDS) from 2009 to 2017 for MRD detection at the National Taiwan University Hospital were used for establishing AI diagnostic strategy. Totally 4350 data samples from two different FC machines (3090 from FASCalibur and 1260 from FASCantoII-Senior) were included. Each of the samples had a 12-tube test on 100000 cells, measured in 6 fluorescent channels (FSC, SSC, FITC, PE, PerCP, APC) within 1 tube.The whole dataset was randomly divided to the Training set and the Validation set, taking 80% and 20% samples respectively. The Training set was used to develop and tune the AI algorithm, and final concordance was estimated on the blinded Validation set. Algorithms for pair-wise recognition (AML vs. normal, MDS vs. normal and abnormal (AML+MDS) vs. normal) were developed independently, according to previous manual interpretation of FC data.The recorded numerical values of the 6 fluorescent channels of each tube were considered as raw feature attributes. The data was first statistically-modelled with a sub-dictionary approach in learning a multivariate Gaussian mixture model. Then a probabilistic derivation based on Fisher scoring was exploited to compute the L2-normalized fixed-dimensional input representations. The raw feature attributes were then encoded into a Fisher representational vector. Lastly, vectors of each tube were concatenated as the final high-dimensional input to the supervised machine learning classifier, i.e., a support vector machine with linear kernel. In addition, ANOVA-based feature selection was also conducted throughout the experiments.For outcome predicting analysis, we included 94 AML patients with available pre-allo-HSCT FC data. Their clinical parameters, progression-free survival (PFS) and overall survival (OS) after allo-HSCT are recorded and analyzed with a median follow-up of 39.9 months.ResultsFor the Calibur Training set (n=2,649), the concordance rate of AI algorithm with manual analysis in differentiating AML vs. normal, MDS vs. normal, and abnormal vs normal was as high as 91.8%, 94.0%, and 90.8%, respectively. Similar rates were noted for the CantoII-Senior Training set (n=1,041): 88.0%, 85.5%, and 84.4%. The final concordance rates for the overall Validation set was 87.9%, 87.9% and 85.1%, respectively. Surprisingly, the algorithm developed from only single tube (CD13, CD16, CD45, FSC and SSC) can achieve almost identical concordance rates as to that from all 12 tubes, both for Calibur data (estimated by the area under receiver operating characteristic curve (AUC), 0.932 vs. 0.935) and CantoII-Senior data (AUC 0.790 vs. 0.832). Another point to note is that the AI system is 100 times faster than the trained professionals in interpreting one FC data (around 7 secs vs 15-30 mins).The clinical parameters for 94 AML patients underwent HSCT were illustrated (Table 1). As noted, 38 were classified to have abnormal pre-HSCT FC data by AI system (indicating residual disease) and 56 were normal. Those with normal FC had significantly longer post-HSCT OS compared to those with abnormal FC (median OS NR vs 6.5 months, p<0.001) (Figure 1a). Pre-HSCT normal FC also predicted longer post-HSCT PFS (median PFS NR vs 4.5 months, p<0.0001) (Figure. 1b). Multi-variate analysis with Cox Hazard Proportional model confirmed the prognostic significance of positive MRD by AI for OS and RFS.ConclusionsThis study demonstrated that AI could be an efficient and reliable diagnostic and prognosis prediction tool for AML. In the future, we like to incorporate other test results simultaneously measured for those patients as our next phase of advancing the AI system. [Display omitted] DisclosuresLi:Celgene International Sàrl: Research Funding. Ko:Celgene International Sàrl: Research Funding. Hou:Celgene International Sàrl: Research Funding. Lin:Celgene International Sàrl: Research Funding. Tien:Celgene International Sàrl: Research Funding. Tang:Celgene International Sàrl: Research Funding.

Flow cytometry data mining by cytoChain identifiesdeterminants of exhaustion and stemness in TCR-engineered T cells.

The phenotype of infused cells is a major determinant of Adoptive T-cell therapy (ACT) efficacy. Yet, the difficulty in deciphering multiparametric cytometry data limited the fine characterization of cellular products. To allow the analysis of dynamic and complex flow cytometry samples, we developed cytoChain, a novel dataset mining tool and a new analytical workflow. CytoChain was challenged to compare state-of-the-art and innovative culture conditions to generate stem-like memory cells (TSCM ) suitable for ACT. Noticeably, the combination of IL-7/15 and superoxides scavenging sustained the emergence of a previously unidentified nonexhausted Fit-TSCM signature, overlooked by manual gating and endowed with superior expansion potential. CytoChain proficiently traced back this population in independent datasets, and in T-cell receptor engineered lymphocytes. CytoChain flexibility and function were then further validated on a published dataset from circulating T cells in COVID-19 patients. Collectively, our results support the use of cytoChain to identify novel, functionally critical immunophenotypes for ACT and patients immunomonitoring.

Defining T Cell Subsets in Human Tonsils Using ChipCytometry.

ChipCytometry is a multiplex imaging method that can be used to analyze either cell suspensions or tissue sections. Images are acquired by iterative cycles of immunostaining with fluorescently labeled Abs, followed by photobleaching, which allows the accumulation of multiple markers on a single sample. In this study, we explored the feasibility of using ChipCytometry to identify and phenotype cell subsets, including rare cell types, using a combination of tissue sections and single-cell suspensions. Using ChipCytometry of tissue sections, we successfully demonstrated the architecture of human palatine tonsils, including the B and T cell zones, and characterized subcompartments such as the B cell mantle and germinal center zone, as well as intrafollicular PD1-expressing CD4+ T cells. Additionally, we were able to identify the rare tonsillar T cell subsets, mucosal-associated invariant T (MAIT) and γδ-T cells, within tonsil tissue. Using single-cell suspension ChipCytometry, we further dissected human tonsillar T cell subsets via unsupervised clustering analysis as well as supervised traditional manual gating. We were able to show that PD1+CD4+ T cells are comprised of CXCR5+BCL6high follicular Th cells and CXCR5-BCL6mid pre-follicular Th cells. Both supervised and unsupervised analysis approaches identified MAIT cells in single-cell suspensions, confirming a phenotype similar to that of blood-derived MAIT cells. In this study, we demonstrate that ChipCytometry is a viable method for single-cell suspension cytometry and analysis, with the additional benefit of allowing phenotyping in a spatial context using tissue sections.

T lymphocyte and monocyte subsets are dysregulated in type 1 diabetes patients with peripheral neuropathic pain

Diabetic neuropathic pain is a common and devastating complication of type 1 diabetes, but the mechanism by which it develops and persists is yet to be fully elucidated. This study utilised high-dimensional suspension mass cytometry in a pilot cohort to investigate differences in peripheral blood immunophenotypes between type 1 diabetes patients with (n ​= ​9) and without (n ​= ​9) peripheral neuropathic pain. The abundance and activation of several leukocyte subsets were investigated with unsupervised clustering approaches FlowSOM and SPADE, as well as by manual gating. Major findings included a proportional increase in CD4+ central memory T cells and an absolute increase in classical monocytes, non-classical monocytes, and mature natural killer cells in type 1 diabetes patients with pain compared to those without pain. The expression of CD27, CD127, and CD39 was upregulated on select T cell populations, and the phosphorylated form of pro-inflammatory transcription factor MK2 was upregulated across most populations. These results provide evidence that distinct immunological signatures are associated with painful neuropathy in type 1 diabetes patients. Further research may link these changes to mechanisms by which pain in type 1 diabetes is initiated and maintained, paving the way for much needed targeted treatments.

Leveraging high-dimensional flow cytometry to reveal the human immune system at a population-wide scale

Abstract Population-wide assessment of the human immune system is needed to better understand immune homeostasis, immune responses against pathogens or the efficacy of vaccines. High-dimensional flow cytometry is the gold standard for in-depth immunophenotyping but the assessment of large cohorts requires multiple experiments over a prolonged period of time and bear multiple sources of error, reducing precision required for sensitive analyses. We performed a large immunophenotyping study using 28-color flow cytometry to investigate the heritability and homeostasis of the immune system and T cell functions in 3000 individuals. Here, we describe our high-throughput sample and data processing pipeline (200 samples/day) in combination with high dimensional flow cytometry to dramatically reduce the number of experiment-days in our study, resulting in low variation and high reproducibility. We demonstrate several carefully optimized strategies to reduce experimental errors and inter-assay variation at every level. In addition, we discuss strategies for pre-processing and analysis of large flow cytometry datasets using manual gating and unsupervised clustering analyses, such as FlowSOM, in view of the fact that most current strategies are highly limited for large datasets. In conclusion, we established a high-throughput high-dimensional flow cytometry pipeline enabling the analysis of thousands of samples with low technical and experimental variation. Our pipeline in combination with our three 28-color flow cytometry panels is adaptable to other large cohort studies for which precision and accuracy over time are important.

High-dimensional analysis defines multicytokine T-cell subsets and supports a role for IL-21 in atopic dermatitis.

Flow cytometry is a well-accepted approach for immune profiling; however, its value is restricted by the limited number of markers that can be analyzed simultaneously. Mass cytometry/CyTOF offers broad-scale immune characterization integrating large number of parameters. While partial blood phenotyping was reported in atopic dermatitis (AD), patients' comprehensive profiling, critical for leveraging new targeted treatments, is not available. IL-21 may be involved in inflammatory skin diseases but its role in AD is not well established. We studied T-cell polarization in the blood of 20 moderate-to-severe AD and 15 controls. Using CyTOF and an unsupervised analysis, we measured the frequencies and mean metal intensities of activated polar CD4+ /CD8+ T-cell subsets. Immunohistochemistry, immunofluorescence, and qRT-PCR were used to analyze skin samples. Examining 24 surface, intracellular markers, and transcription factors, we identified six CD4+ and five CD8+ T-cell metaclusters. A CD4+ skin-homing IL-13+ monocytokine and a novel IL-13+ IL-21+ multicytokine metaclusters were increased in AD vs. controls (p<.01). While IL-13 signature characterized both clusters, levels were significantly higher in the IL-21+ group. Both clusters correlated with AD severity (r = 0.49, p = .029). Manual gating corroborated these results and identified additional multicytokine subsets in AD. Immunohistochemistry and immunofluorescence, validated by mRNA expression, displayed significantly increasedIL-21 counts and colocalization with IL-13/IL-4R in AD skin. A multicytokine signature characterizes moderate-to-severe AD, possibly explaining partial therapeutic responses to one cytokine targeting, particularly in severe patients. Prominent IL-21 signature in blood and skin hints for a potential pathogenic role of IL-21 in AD.

Dissecting Response to Cancer Immunotherapy by Applying Bayesian Network Analysis to Flow Cytometry Data

Cancer immunotherapy, specifically immune checkpoint blockade, has been found to be effective in the treatment of metastatic cancers. However, only a subset of patients achieve clinical responses. Elucidating pretreatment biomarkers predictive of sustained clinical response is a major research priority. Another research priority is evaluating changes in the immune system before and after treatment in responders vs. nonresponders. Our group has been studying immune networks as an accurate reflection of the global immune state. Flow cytometry (FACS, fluorescence-activated cell sorting) data characterizing immune cell panels in peripheral blood mononuclear cells (PBMC) from gastroesophageal adenocarcinoma (GEA) patients were used to analyze changes in immune networks in this setting. Here, we describe a novel computational pipeline to perform secondary analyses of FACS data using systems biology/machine learning techniques and concepts. The pipeline is centered around comparative Bayesian network analyses of immune networks and is capable of detecting strong signals that conventional methods (such as FlowJo manual gating) might miss. Future studies are planned to validate and follow up the immune biomarkers (and combinations/interactions thereof) associated with clinical responses identified with this computational pipeline.

Current trends in flow cytometry automated data analysis software.

Automated flow cytometry (FC) data analysis tools for cell population identification and characterization are increasingly being used in academic, biotechnology, pharmaceutical, and clinical laboratories. The development of these computational methods is designed to overcome reproducibility and process bottleneck issues in manual gating, however, the take-up of these tools remains (anecdotally) low. Here, we performed a comprehensive literature survey of state-of-the-art computational tools typically published by research, clinical, and biomanufacturing laboratories for automated FC data analysis and identified popular tools based on literature citation counts. Dimensionality reduction methods ranked highly, such as generic t-distributed stochastic neighbor embedding (t-SNE) and its initial Matlab-based implementation for cytometry data viSNE. Software with graphical user interfaces also ranked highly, including PhenoGraph, SPADE1, FlowSOM, and Citrus, with unsupervised learning methods outnumbering supervised learning methods, and algorithm type popularity spread across K-Means, hierarchical, density-based, model-based, and other classes of clustering algorithms. Additionally, to illustrate the actual use typically within clinical spaces alongside frequent citations, a survey issued by UK NEQAS Leucocyte Immunophenotyping to identify software usage trends among clinical laboratories was completed. The survey revealed 53% of laboratories have not yet taken up automated cell population identification methods, though among those that have, Infinicyt software is the most frequently identified. Survey respondents considered data output quality to be the most important factor when using automated FC data analysis software, followed by software speed and level of technical support. This review found differences in software usage between biomedical institutions, with tools for discovery, data exploration, and visualization more popular in academia, whereas automated tools for specialized targeted analysis that apply supervised learning methods were more used in clinical settings.

Evaluating the performances of a flexible mechanism of water diversion: application on a northern Italy gravity-driven irrigation channel

Improving the management of irrigation schemes is a priority in the Mediterranean countries of the EU to achieve the objectives of the Water Framework Directive 2000/60EC. In the case of schemes based on open-channel networks for water conveyance and delivery, a limitation to the management efficiency is the huge number of gates that need to be controlled and regulated manually to adapt the flow to the actual irrigation water demand. Automated and coordinated gate regulation of large irrigation schemes has been successfully implemented in some areas of the world (e.g., NSW, Australia), but it implies huge public investments that are often unavailable. Opposite to this top–down approach, the Lombardy region (the largest irrigated region in the EU) has explored an approach based on funding small projects, which should demonstrate the potential of innovation in irrigation practice and foster the expansion of the most effective measures. In this paper, we analyze the effects of one of these projects, consisting in the substitution of a manual gate, controlling the supply of irrigation to a 150 ha irrigation district, with an automated, remotely controlled gate. Nine years of daily flow measurements at the district inlet, provided by the irrigation consortium that manages the gate, were compared with the rigid flow regulation that was applied when the inlet gate was manually regulated and with the irrigation water requirements simulated with a distributed agro-hydrological model widely used in the region for irrigation accounting and planning. The results show that the flexible regulation allowed by the new gate provided an average water saving of 12,000 m3/ha/year compared to the rigid regulation system. A further, though smaller, margin of improvement of 5000 m3/ha/year can be achieved with an optimal regulation that follows exactly the crop water requirements. A further interesting empirical evidence is that the simulation model fits very well the irrigation water requirements of the district using only easily accessible meteorological input data, without exploiting any information from ground or remote sensors to update soil moisture or crop developments during the season. This indicates that it could be effectively used, with very limited costs and effort, to support and improve the gate management. Finally, a preliminary economic analysis shows that the cost of the gate installation is sustainable, but the upscaling of this type of intervention to larger areas requires the support of public funding to cover approximately half of the cost of investment needed for the networking infrastructures.

Characteristic pancreatic and splenic immune cell infiltration patterns in mouse acute pancreatitis

BackgroundA systemic evaluation of immune cell infiltration patterns in experimental acute pancreatitis (AP) is lacking. Using multi-dimensional flow cytometry, this study profiled infiltrating immune cell types in multiple AP mouse models.MethodsThree AP models were generated in C57BL/6 mice via cerulein (CAE) injection, alcohol and palmitoleic acid (EtOH + POA) injection, and alcohol diet feeding and cerulein (EtOH + CAE) injection. Primary pancreatic cells and splenocytes were prepared, and multi-dimensional flow cytometry was performed and analyzed by manual gating and computerized PhenoGraph, followed by visualization with t-distributed stochastic neighbor embedding (t-SNE).ResultsCAE treatment induced a time-dependent increase of major innate immune cells and a decrease of follicular B cells, and TCD4+ cells and the subtypes in the pancreas, whereas elicited a reversed pattern in the spleen. EtOH + POA treatment resulted in weaker effects than CAE treatment. EtOH feeding enhanced CAE-induced amylase secretion, but unexpectedly attenuated CAE-induced immune cell regulation. In comparison with manual gating analysis, computerized analysis demonstrated a remarkable time efficiency and reproducibility on the innate immune cells and B cells.ConclusionsThe reverse pattern of increased innate and decreased adaptive immune cells was consistent in the pancreas in CAE and EtOH + POA treatments. Alcohol feeding opposed the CAE effect on immune cell regulation. Together, the immune profiling approach utilized in this study provides a better understanding of overall immune responses in AP, which may facilitate the identification of intervention windows and new therapeutic strategies. Computerized analysis is superior to manual gating by dramatically reducing analysis time.