Leukemia Prediction Using Sparse Logistic Regression

We describe a supervised prediction method for diagnosis of acute myeloid leukemia (AML) from patient samples based on flow cytometry measurements. We use a data driven approach with machine learning methods to train a computational model that takes in flow cytometry measurements from a single patient and gives a confidence score of the patient being AML-positive. Our solution is based on an regularized logistic regression model that aggregates AML test statistics calculated from individual test tubes with different cell populations and fluorescent markers. The model construction is entirely data driven and no prior biological knowledge is used. The described solution scored a 100% classification accuracy in the DREAM6/FlowCAP2 Molecular Classification of Acute Myeloid Leukaemia Challenge against a golden standard consisting of 20 AML-positive and 160 healthy patients. Here we perform a more extensive validation of the prediction model performance and further improve and simplify our original method showing that statistically equal results can be obtained by using simple average marker intensities as features in the logistic regression model. In addition to the logistic regression based model, we also present other classification models and compare their performance quantitatively. The key benefit in our prediction method compared to other solutions with similar performance is that our model only uses a small fraction of the flow cytometry measurements making our solution highly economical.     

Analysis of Flow Cytometry Data by Matrix Relevance Learning Vector Quantization

Flow cytometry is a widely used technique for the analysis of cell populations in the study and diagnosis of human diseases. It yields large amounts of high-dimensional data, the analysis of which would clearly benefit from efficient computational approaches aiming at automated diagnosis and decision support. This article presents our analysis of flow cytometry data in the framework of the DREAM6/FlowCAP2 Molecular Classification of Acute Myeloid Leukemia (AML) Challenge, 2011. In the challenge, example data was provided for a set of 179 subjects, comprising healthy donors and 23 cases of AML. The participants were asked to provide predictions with respect to the condition of 180 patients in a test set. We extracted feature vectors from the data in terms of single marker statistics, including characteristic moments, median and interquartile range of the observed values. Subsequently, we applied Generalized Matrix Relevance Learning Vector Quantization (GMLVQ), a machine learning technique which extends standard LVQ by an adaptive distance measure. Our method achieved the best possible performance with respect to the diagnoses of test set patients. The extraction of features from the flow cytometry data is outlined in detail, the machine learning approach is discussed and classification results are presented. In addition, we illustrate how GMLVQ can provide deeper insight into the problem by allowing to infer the relevance of specific markers and features for the diagnosis.     

High throughput, real-time detection of Naegleria lovaniensis in natural river water using LED-illuminated Fountain Flow Cytometry

AIMS: To test Fountain Flow Cytometry (FFC) for the rapid and sensitive detection of Naegleria lovaniensis amoebae (an analogue for Naegleria fowleri) in natural river waters. METHODS AND RESULTS: Samples were incubated with one of two fluorescent labels to facilitate detection: ChemChrome V6, a viability indicator, and an R-phycoerytherin (RPE) immunolabel to detect N. lovaniensis specifically. The resulting aqueous sample was passed as a stream in front of a light-emitting diode, which excited the fluorescent labels. The fluorescence was detected with a digital camera as the sample flowed toward the imager. Detections of N. lovaniensis were made in inoculated samples of natural water from eight rivers in France and the United States. FFC enumeration yielded results that are consistent with other counting methods: solid-phase cytometry, flow cytometry, and hemocytometry, down to concentrations of 0.06 amoebae ml(-1), using a flow rate of 15 ml min(-1). CONCLUSIONS: This study supports the efficacy of using FFC for the detection of viable protozoa in natural waters and indicates that use of RPE illuminated at 530 nm and detected at 585 nm provides a satisfactory means of attenuating background. SIGNIFICANCE AND IMPACT OF THE STUDY: Because of the severe global public health issues with drinking water and sanitation, there is an urgent need to develop a technique for the real-time detection of viable pathogens in environmental samples at low concentrations. FFC addresses this need.     

Extracting a cellular hierarchy from high-dimensional cytometry data with SPADE

The ability to analyze multiple single-cell parameters is critical for understanding cellular heterogeneity. Despite recent advances in measurement technology, methods for analyzing high-dimensional single-cell data are often subjective, labor intensive and require prior knowledge of the biological system. To objectively uncover cellular heterogeneity from single-cell measurements, we present a versatile computational approach, spanning-tree progression analysis of density-normalized events (SPADE). We applied SPADE to flow cytometry data of mouse bone marrow and to mass cytometry data of human bone marrow. In both cases, SPADE organized cells in a hierarchy of related phenotypes that partially recapitulated well-described patterns of hematopoiesis. We demonstrate that SPADE is robust to measurement noise and to the choice of cellular markers. SPADE facilitates the analysis of cellular heterogeneity, the identification of cell types and comparison of functional markers in response to perturbations.     

RNA sequencing: new technologies and applications in cancer research

Acute myeloid leukemia (AML) is a fatal hematopoietic malignancy and has a prognosis that varies with its genetic complexity. However, there has been no appropriate integrative analysis on the hierarchy of different AML subtypes. Using Microwell-seq, a high-throughput single-cell mRNA sequencing platform, we analyzed the cellular hierarchy of bone marrow samples from 40 patients and 3 healthy donors. We also used single-cell single-molecule real-time (SMRT) sequencing to investigate the clonal heterogeneity of AML cells. From the integrative analysis of 191727 AML cells, we established a single-cell AML landscape and identified an AML progenitor cell cluster with novel AML markers. Patients with ribosomal protein high progenitor cells had a low remission rate. We deduced two types of AML with diverse clinical outcomes. We traced mitochondrial mutations in the AML landscape by combining Microwell-seq with SMRT sequencing. We propose the existence of a phenotypic “cancer attractor” that might help to define a common phenotype for AML progenitor cells. Finally, we explored the potential drug targets by making comparisons between the AML landscape and the Human Cell Landscape. We identified a key AML progenitor cell cluster. A high ribosomal protein gene level indicates the poor prognosis. We deduced two types of AML and explored the potential drug targets. Our results suggest the existence of a cancer attractor.     

Single-cell vs. bulk activity properties of coastal bacterioplankton over an annual cycle in a temperate ecosystem

The connections between single-cell activity properties of heterotrophic planktonic bacteria and whole community metabolism are still poorly understood. Here, we show flow cytometry single-cell analysis of membrane-intact (live), high nucleic acid (HNA) content and actively respiring (CTC+) bacteria with samples collected monthly during 2006 in northern Spain coastal waters. Bulk activity was assessed by measuring ³H-Leucine incorporation and specific growth rates. Consistently, different single-cell relative abundances were found, with 60–100% for live, 30–84% for HNA and 0.2–12% for CTC+ cells. Leucine incorporation rates (2–153 pmol L⁻¹ h⁻¹), specific growth rates (0.01–0.29 day⁻¹) and the total and relative abundances of the three single-cell groups showed marked seasonal patterns. Distinct depth distributions during summer stratification and different relations with temperature, chlorophyll and bacterial biovolume suggest the existence of different controlling factors on each single-cell property. Pooled leucine incorporation rates were similarly correlated with the abundance of all physiological groups, while specific growth rates were only substantially explained by the percentage of CTC+ cells. However, the ability to reduce CTC proved notably better than the other two single-cell properties at predicting bacterial bulk rates within seasons, suggesting a tight linkage between bacterial individual respiration and biomass production at the community level.     

Cryptomonad algal phycobiliproteins as fluorochromes for extracellular and intracellular antigen detection by flow cytometry

BACKGROUND: Phycobiliproteins play an important role in fluorescent labeling, particularly for flow cytometry. The spectral properties of R-phycoerythrin (R-PE) and allophycocyanin (APC) have made them the dominant reagents in this class of fluorochromes. In this study, we evaluate a lesser-known but potentially important series of low-molecular weight cryptomonad-derived phycobiliproteins (commercially termed the CryptoFluortrade mark dyes) for their applicability to flow cytometry, both in extracellular and intracellular labeling applications. METHODS: Several cell lines were labeled with biotin-conjugated antibodies against expressed extracellular surface proteins, followed by streptavidin conjugates of three cryptomonad phycobiliproteins (CryptoFluor-2, CryptoFluor-4, and CryptoFluor-5). Cells were then analyzed by flow cytometry using a variety of laser lines and emission filters to establish the optimal excitation/emission characteristics for each fluorochrome. Some cells were permeabilized and labeled for intracellular antigens, also using the cryptomonad fluorochromes. Where appropriate, parallel samples were labeled with other fluorochromes (including R-PE, APC, the cyanin dyes Cy3 and Cy5, and others) to gauge the performance of the cryptomonad fluorochromes against fluorescent labels previously evaluated for flow cytometry. RESULTS: CryptoFluor-2 possessed excitation/emission maxima similar to those of APC and Cy5, with good excitation in the red (HeNe laser 632 nm) and strong emission in the far red (660 nm). CryptoFluor-4 possessed excitation/emission maxima similar to those of Cy3, with optimal excitation in the green (Kr 530 nm) and strong emission in the yellow/orange (585 nm). CryptoFluor-5 possessed excitation/emission maxima similar to those of lissamine rhodamine, with optimal excitation in the yellow (Kr 568 nm) and emission in the orange (610 nm). All cryptomonad fluorochromes gave satisfactory results for both intracellular and extracellular labeling, with detection sensitivities that were comparable or better than traditional phycobiliproteins and low- molecular weight synthetic fluorochromes such as the cyanin dyes. CONCLUSIONS: The CryptoFluor fluorochromes were applicable to flow cytometric immunodetection, with excitation and emission conditions commonly found on multilaser instruments. Performance of several of these dyes was at least comparable to existing fluorescent labels. The low molecular weights (30-60 kd) of phycobiliproteins may make them particularly useful in intracellular antigen detection. Cytometry 44:16-23, 2001. Published 2001 Wiley-Liss, Inc.     

Simultaneous detection of mitochondrial respiratory chain activity and reactive oxygen in digitonin-permeabilized cells using flow cytometry

BACKGROUND: Increased mitochondrial generation of reactive oxygen intermediates (ROI) due to defective respiratory chain activity has been implicated in physiological processes such as apoptosis, in the pathogenesis of mitochondrial diseases, and as part of the normal aging process. Established methods addressing activity of the respiratory chain complexes have been limited to bulk assays for single parameters. This study describes a flow cytometry-based method and its validation for the detection of respiratory chain function in single cells permeabilized by digitonin. METHODS: Flow cytometry was used to measure mitochondrial membrane potential (DeltaPsi(m)) and reactive oxygen generation under differing conditions of respiration. This was brought about by the addition of substrates and inhibitors to digitonin-permeabilized cells. This method was validated by measurement of oxygen consumption and ATP production and by confocal microscopy. RESULTS: Activity of the respiratory chain complexes assessed by DeltaPsi(m) responded to substrates and inhibitors as predicted from assessment by oxygen consumption and ATP synthesis. In addition, the flow cytometry method allows the simultaneous assessment of mitochondrial ROI generation. This was confirmed by the localization of the ROI probe, carboxy-DCF, to the same site as the mitochondrial probe observed by confocal microscopy. CONCLUSIONS: This method allows the functional integrity of the respiratory chain complexes to be studied at the single-cell level, thus addressing the relationship between disordered function of respiratory chain complexes and mitochondrial ROI generation.     

A flow cytometric technique for quantification and differentiation of bacteria in bulk tank milk

AIMS: The present study describes a flow cytometric technique for quantification and differentiation of bacteria in bulk tank milk according to the main cause of elevated counts. METHODS AND RESULTS: A total of 75 Danish bulk tank milk samples exceeding the grading level of 3.0 x 10(4) CFU ml(-1) were examined by both flow cytometry and traditional microbiological analyses. The correlation coefficient (r) between the two methods was 0.71. For the differential analyses of the dominant bacterial populations four different parameters were used to give a species-characteristic pattern. The four parameters were as follows: staining with Oregon Green conjugated wheat germ agglutinin that binds to the cell wall of bacteria, staining with hexidium iodide that binds to all bacterial DNA, the flow cytometric forward scatter and the flow cytometric side scatter. Three regions in the flow cytometric plot were defined: region 1 includes bacteria mainly associated with poor hygiene, region 2 includes psychrotrophic hygiene bacteria and region 3 includes bacteria mainly related to mastitis. The ability of the flow cytometric technique to predict the main cause of elevated bacterial counts on routine samples was examined. Comparing these results with results obtained by traditional microbiological analyses for identification showed that for 81% of the samples the two techniques agreed on the main cause of an elevated bacterial count. CONCLUSIONS: The ability of the presented flow cytometric technique to enumerate and differentiate bacteria in bulk tank milk according to the main cause of elevated counts was demonstrated. SIGNIFICANCE AND IMPACT OF THE STUDY: This study described the first step in development of a technique suitable for routine analyses of bulk tank milk samples. A technique indicating the main cause of an elevated count will enable the farmer to eliminate the contamination source within a short time limit.     

Cytokine flow cytometry: multiparametric approach to immune function analysis

More precise quantitation of cellular immune responses has become possible with the advent of single-cell assays of immune function, such as cytokine flow cytometry, enzyme-linked immunospot (ELISPOT), and MHC-peptide multimers. Cytokine flow cytometry is an attractive technique because it allows the detection of responses to whole antigens without regard to MHC restriction, while also collecting additional information on responding cells via multiparameter flow cytometry. In this review, we compare cytokine flow cytometry with other assays of immune function, summarize some of that data that have been collected in various disease states using cytokine flow cytometry, and describe some methodological improvements designed to increase the robustness, throughput, and information content of this technique. We hypothesize that a new generation of automated cytokine flow cytometry assays will allow elucidation of the correlates of protection for diseases involving cellular immunity, through application of these assays in more and large clinical trials.     

High-Dimensional Analysis of Acute Myeloid Leukemia Reveals Phenotypic Changes in Persistent Cells during Induction Therapy

The plasticity of AML drives poor clinical outcomes and confounds its longitudinal detection. However, the immediate impact of treatment on the leukemic and non-leukemic cells of the bone marrow and blood remains relatively understudied. Here, we conducted a pilot study of high dimensional longitudinal monitoring of immunophenotype in AML. To characterize changes in cell phenotype before, during, and immediately after induction treatment, we developed a 27-antibody panel for mass cytometry focused on surface diagnostic markers and applied it to 46 samples of blood or bone marrow tissue collected over time from 5 AML patients. Central goals were to determine whether changes in AML phenotype would be captured effectively by cytomic tools and to implement methods for describing the evolving phenotypes of AML cell subsets. Mass cytometry data were analyzed using established computational techniques. Within this pilot study, longitudinal immune monitoring with mass cytometry revealed fundamental changes in leukemia phenotypes that occurred over time during and after induction in the refractory disease setting. Persisting AML blasts became more phenotypically distinct from stem and progenitor cells due to expression of novel marker patterns that differed from pre-treatment AML cells and from all cell types observed in healthy bone marrow. This pilot study of single cell immune monitoring in AML represents a powerful tool for precision characterization and targeting of resistant disease.     

Chlorophyll a fluorescence measurements of isolated spinach thylakoids obtained by using single-laser-based flow cytometry

Flow cytometry data of spinach thylakoid membrane preparations indicate the presence of a homogeneous thylakoid population. Fluorescence data from a flow cytometer and comparison with data from two other fluorometers show that chlorophyll a fluorescence detected with a flow cytometer has the character of maximum fluorescence (Fmax), not of the constant component (Fo). This conclusion is important since Fo measures fluorescence that is affected mostly by changes in excitation energy transfer and Fmax-Fo (the variable fluorescence) by changes in photochemistry. This was demonstrated by: 1) The light intensity as well as diffusion rate dependence of the quenching effect of various quinones (p-benzoquinone, phenyl-benzoquinone, and 2,5-dibromo-3-methyl-6-isopropyl-p-benzoquinone, DBMIB) on fluorescence yield; quenching for the same concentration of these quinones was lower at the higher than at the lower light intensities. 2) Temperature dependence of the fluorescence yield; increasing the temperature from 20 to 70 degrees C did not show an increase in fluorescence yield using a flow cytometer in contrast to measurements with weak excitation light, but similar to those obtained for Fmax. 3) Addition of an inhibitor diuron up to 100 microM did not change the fluorescence intensity. A comparison of quenching of fluorescence by various quinones obtained by flow cytometry with those by other fluorometers suggests that the high intensity used in the cytometry produces unique results: the rate of reduction of quinones in much larger than the rate of equilibration with the bulk quinones.     

Recent progress in mass spectrometry for single-cell metabolomics

Metabolites are the end products of cellular vital activities and can reflect the state of cellular to a certain extent. Rapid change of metabolites and the low abundance of signature metabolites cause difficulties in single-cell detection, which is a great challenge in single-cell metabolomics analysis. Mass spectrometry (MS) is a powerful tool that uniquely suited to detect intracellular small-molecule metabolites and has shown good application in single-cell metabolite analysis. In this mini-review, we describe three types of emerging technologies for MS-based single-cell metabolic analysis in recent years, including nano-ESI-MS based single-cell metabolomics analysis, high-throughput analysis via flow cytometry, and cellular metabolic imaging analysis. These techniques provide a large amount of single-cell metabolic data, allowing the potential of MS in single-cell metabolic analysis is gradually being explored and is of great importance in disease and life science research.     

Cell kinetics after high dose cytosine arabinoside in patients with acute myelocytic leukemia

In this study 10 patients with acute myelocytic leukemia (AML) each received a rapid intravenous injection of high dose cytosine arabinoside (HD Ara-C; 1 g/m2). Bone marrow aspirates were obtained before and after Ara-C administration to determine the percentage of cells in S-phase measured by flow cytometry. In 5 out of 10 cases synchronization of the leukemic cells in S-phase of the cell cycle was observed. However, the time of maximum synchronization turned out to be difficult to predict. Therefore, the strong correlation between percentage of cells in S-phase at diagnosis and the time of maximal accumulation of S-phase cells after Ara-C administration, as observed by others in childhood AML, could not be confirmed for adult AML patients. Although synchronization of AML cells after in vivo Ara-C administration could be demonstrated in at least half of the patients, the practical consequences are such that clinical application was hampered.     

Detection of anti-phosphatidylethanolamine antibodies using flow cytometry.

BACKGROUND: The finding that lupus anticoagulant (LA) is significantly associated with anti-phosphatidylethanolamine (PE) activity has led to great interest in its relation to the clinical features of the antiphospholipid syndrome (APS). Considerable variability has, however, been reported in the prevalence of anti-PE antibodies in APS patients using enzyme-linked immunosorbent assay (ELISA) methodology. The lack of standardization and differences in technique may in part explain these discrepancies. PE binds variably to different types of microtiter wells, reflected in the consequent detection, or lack of detection, of anti-PE antibodies. This study describes the use of flow cytometry as an alternative method for the detection of anti-PE antibodies. METHODS: Six LA-positive plasma samples were used in this original study. Polystyrene beads were coated with PE overnight. These were subsequently incubated with patient plasma. Both IgG and IgM binding were detected by flow cytometry using a cocktail of fluorescently labelled anti-human Ig isotypes. RESULTS: When these results were compared with those from ELISA, flow cytometric analysis provided an apparent enhanced detection of anti-PE antibodies. It was found that 6/6 were IgM anti-PE positive by flow cytometry, whereas 5/6 were IgM by ELISA; 2/6 negative for anti-cardiolipin antibodies by ELISA were positive by flow cytometry; and 2/6 positive for antiphosphatidylcholine antibodies in cytometry were negative by ELISA. CONCLUSIONS: With appropriate quantification, this method may be more sensitive than ELISA in detecting anti-PE antibodies in plasma samples of patients with APS.     

Mathematical modeling of a single-cell enzyme assay

A quantitative assay of beta-galactosidase activity in single cells of Saccharomyces cerevisiae has been developed using a fluorogenic substrate and flow cytometry [reported in Wittrup & Bailey, Cytometry, 9,394 (1988)]. The beta-galactosidase activity is expressed in yeast from the Escherichia coli lacZ gene under the control of the yeast GAL10 promoter, and is used as a marker for multicopy plasmid content. A nonfluorescent fluorogenic substrate is enzymatically cleaved by intracellular beta-galactosidase to form a fluorescent product. The accumulation of fluorescent product in single cells was found to depend on bulk substrate concentration and single-cell enzyme activity in a fashion that could not be described by a Michaelis-Menten kinetic rate form. It has been demonstrated that diffusion limitation rather than enzyme activity can determine the level of single-cell fluorescence under certain assay conditions, and a mathematical model has; been formulated which accounts for substrate and product diffusion. Guided by the mathematical model, the assay conditions were modified to allow measurement of single-cell enzyme activity rather than diffusion rates.     

Joint analysis of single-cell and bulk tissue sequencing data to infer intratumor heterogeneity

Many computational methods have been developed to discern intratumor heterogeneity (ITH) using DNA sequence data from bulk tumor samples. These methods share an assumption that two mutations arise from the same subclone if they have similar mutant allele-frequencies (MAFs), and thus it is difficult or impossible to distinguish two subclones with similar MAFs. Single-cell DNA sequencing (scDNA-seq) data can be very informative for ITH inference. However, due to the difficulty of DNA amplification, scDNA-seq data are often very noisy. A promising new study design is to collect both bulk and single-cell DNA-seq data and jointly analyze them to mitigate the limitations of each data type. To address the analytic challenges of this new study design, we propose a computational method named BaSiC (B ulk tumor a nd Si ngle C ell), to discern ITH by jointly analyzing DNA-seq data from bulk tumor and single cells. We demonstrate that BaSiC has comparable or better performance than the methods using either data type. We further evaluate BaSiC using bulk tumor and single-cell DNA-seq data from a breast cancer patient and several leukemia patients.     

Three-color versus four-color multiparameter cell cycle analyses of primary acute myeloid leukemia samples

Checkpoint alterations that impact cell cycle and apoptosis responses to therapeutic treatments may produce drug resistance in acute myeloid leukemia (AML). To study these, we have developed flow cytometry assays of checkpoint function that also allow quantitation of key molecular regulators of apoptosis and cell cycle. We have used three-color (3C) assays, with FITC-labeled anti-BCL-2 and PE-labeled anti-proliferating cell nuclear antigen (PCNA) antibodies, and the DNA dye 7-aminoactinomycin, to characterize primary leukemia cells identified in DNA x side light scatter (SSC) histograms. We showed that 3C assays are accurate and reproducible in analyses of leukemia cell lines and of primary AML and normal bone marrow samples (Banker et al.: Blood 89: 243-255, 1997; Banker et al.: Leukemia Res 22: 221-239, 1998; Banker et al.: Clin Cancer Res 4: 3051-3062, 1998). To further confirm the validity of our SSC leukemia cell gating and to address whether immunophenotypic AML subsets might have different biologic properties, we have now designed four-color (4C) flow assays to characterize checkpoint status in leukemic blasts specifically identified by surface immunostaining. In modeling this assay strategy, PE/Cy5-labeled anti-CD34 antibody was used to detect blasts, with FITC-labeled anti-BCL-2, PE-labeled anti-PCNA antibodies, and Hoechst 33342 (H33342) DNA dye. Four-color CD34-gated data was concordant with 3C, SSC-gated data for leukemia cell lines and for most primary AML samples with high and intermediate blast counts. BCL-2 and PCNA immunopositivity and sub-G1 apoptosis determinations were different in the CD34-gated versus SSC-gated blasts in particular samples with smaller CD34(+) subsets, suggesting that leukemia samples can contain blast subsets with different biologic properties. On the other hand, PCNA-gated cell-cycle distributions in untreated cells and G1 versus S phase cell-cycle arrests after cytosine arabinoside treatments were completely concordant in 4C and 3C assays. We conclude that both 3C and 4C assays can be used to characterize protein expression and cell-cycle drug response patterns in leukemia blasts, but that 4C assays may additionally allow discrimination of these properties in immunophenotypic leukemia subsets.