Unbiased Cluster Analysis Research Articles

Abstract 4694: Evaluation of tumor-associated antigen expression with the MACSimaTM high-content imaging platform

Abstract Here we report the use of the MACSima™ imaging platform to perform high-content imaging for the pre-clinical validation of tumor target expression on tumor and healthy human tissues indicative for potential on-target/off-tumor toxicity in vivo. Major advances have been achieved in cancer therapy in the past decade. In particular, targeted-immunotherapy has progressed and clinical benefits for patients have been achieved. However, on-target/off-tumor toxicity is a potential threat which has been shown to be more pronounced in solid than in liquid tumors. These findings highlight the need for better pre-clinical assays to improve the safety profile of immunotherapies. On-target/off-tumor toxicity is mainly based on the expression of tumor-associated antigens (TAA) in healthy tissues under physiological conditions. Currently, most prediction methods for on-target/off-tumor expression are based on bulk mRNA expression data of healthy tissue. These prediction models, however, have limitations, mainly poor predictable relation between RNA and protein level. Moreover, it is frequently unclear which cell types are affected by on-target/off-tumor effects. To overcome these limitations we employ multi-parameter imaging to analyze the expression of TAAs directly at the protein level. Additionally, we gain spatial information about tissue and cellular distribution of TAAs. Notably, our novel high-content imaging technology potentially allows for the analysis of hundreds of markers in a single experiment, paving the way for high-dimensional characterization of cells within complex solid tissues. We performed high-content imaging with the MACSima™ platform to validate the expression of known TAAs across tumor and healthy tissue samples. Subsequently, we employed unbiased cluster analysis revealing correlation patterns within the datasets to identify cell types at risk. In detail, we analyzed several high-grade serous ovarian carcinoma and pancreatic ductal adenocarcinoma for the expression of known TAAs, described tumor markers, and tissue lineage markers. Additionally, we evaluated the expression of these TAAs across several healthy human tissues. Next we performed pixel- and object-based data analysis for unbiased cluster analysis. Thereby we identified cell clusters that express TAAs in ovarian and pancreatic cancers, as well as in healthy tissues. We found that primarily epithelial cells express the analyzed TAAs in different tissues and that TAA expression shows inter- and intra-tumor heterogeneity. Taken together, we established a novel workflow for multi-parameter characterization of tissues employing the MACSima™ imaging platform. Our work enables efficient identification of potential tissues for on-target/off-tumor toxicity, paving the way for improved pre-clinical evaluation of TAAs as new targets. Citation Format: Christoph Herbel, Vera Dittmer, Manuel Martinez-Osuna, Laura Nadine Kuester, Daniel Schaefer, Jan Drewes, Jutta Kollet, Werner Mueller, Michael Mallmann, Peter Mallmann, Philipp Stroebel, Olaf Hardt, Dominik Eckardt, Andreas Bosio. Evaluation of tumor-associated antigen expression with the MACSimaTM high-content imaging platform [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 4694.

20-LB: Endothelial Cell Metabolomics Identifies ?-glutamyl Cycle Alterations Associated with the 1q25 T2D-Specific Cardiovascular Locus

A variant on chromosome 1q25 (rs10911021) has been previously associated with coronary artery disease (CAD) in individuals with T2D. In human umbilical vein endothelial cells (HUVECs), the 1q25 risk allele (‘C’) was found to be associated with decreased expression of the adjacent gene GLUL encoding the enzyme glutamine synthase converting glutamate to glutamine. To further investigate the mechanisms through which the 1q25 locus modulates CAD risk in T2D, we performed a global metabolomics study (LC-MS, Metabolon) in 62 HUVECs (TT=11, CT=21, CC=30). Among 597 metabolites detected in HUVEC lysates, 23 attained p<0.005 for association with the 1q25 risk allele. Through an unbiased hierarchical cluster analysis, these 23 metabolites could be subdivided into four clusters - two upregulated and two downregulated in carriers of the risk allele. The two upregulated clusters included a group of γ-glutamyl amino acids (γ-glutamylthreonine, γ-glutamylleucine, and γ-glutamylvaline) and a heterogeneous group of metabolites including isoleucylglycine and tyrosylglycine. The two downregulated clusters included a group of polyunsaturated fatty acid (PUFA) derivatives, such as DHA-choline and N-arachidonoyl taurine, and a cluster of several sphingomyelins. Consistent with the higher γ-glutamyl amino acids, the risk allele was associated with increased ophthalmate (p=0.04), a marker of γ-glutamyl cycle malfunction, and decreased S-lactoylglutathione (p=0.016), a possible sign of defective detoxification of the AGE-precursor methylglyoxal. In summary, we have identified several metabolic pathways potentially mediating the 1q25 locus effect on CAD risk in T2D. These include impairment of the γ-glutamyl cycle, possibly increasing susceptibility to oxidative stress, and downregulation of PUFA derivatives and sphingomyelins. Further studies of these pathways may lead to the identification of novel therapeutic targets to reduce CAD risk in T2D. Disclosure C. Pipino: None. H. Shah: None. K. Park: None. S. Monti: None. C. Mendonca: None. G.L. King: Research Support; Self; Sanofi. A. Pandolfi: None. A. Doria: Research Support; Self; Sanofi-Aventis. Funding Sanofi-Aventis

Abstract 101: Single-cell Profiling of Atherosclerotic Tissue Identifies T Cell Subsets Associated with Cerebrovascular Events

Atherogenesis is driven by the infiltration of immune cells into the arterial wall where inflammation occurs and promotes plaque growth. While some plaques are stable, others are vulnerable and may rupture and lead to stroke or myocardial infarction. The contribution of specific immune cell types to plaque stability is obscure. In our present study, we aim to characterize the immune cell repertoire of human atherosclerosis using innovative single cell methods. We used mass cytometry (CyTOF) to study plaque tissue and paired blood from symptomatic (TIA or stoke < 6 months) or asymptomatic (no TIA or stroke) patients undergoing carotid endarterectomy. Using unbiased MetaLouvain clustering analysis, we characterized cell populations from two independent cohorts to (1) define major immune populations (cohort1; n=15), and (2) detail the adaptive immune compartment (cohort2; n=23), which was predominant in cohort 1. Moreover, we used single cell RNAseq to obtain the transcriptional signatures of immune infiltrates of atherosclerotic tissue. We found that T cells dominate advanced plaques. Atherosclerotic tissue was enriched in CD8 + Effector Memory (EM) T cells that were CD69 hi , CCR5 hi and PD-1 hi vs. their blood counterparts. Our transcriptional profiling confirmed that T cells were the majority of all immune populations, and that T cells expressed genes that represented an activated and largely exhausted phenotype. Our CyTOF results identified differences in T cell subsets between patient types. Symptomatic patients had a higher frequency of a subset of CD4 + EM T cells (CD69 hi CCR5 hi PD1 hi CD27 lo CD28 lo ), and displayed a more differentiated phenotype of CD27 lo GZMB lo CD38 lo CD8 + EM T cells. Our study is the first to provide a single cell immune atlas of advanced atherosclerotic lesions and to identify specific subsets associated with clinical outcomes of atherosclerotic disease. This study will be valuable for the future design of precise tissue-targeted immunotherapies.

Adventitial Cell Atlas of wt (Wild Type) and ApoE (Apolipoprotein E)-Deficient Mice Defined by Single-Cell RNA Sequencing.

Objective- Vascular adventitia encompasses progenitors and is getting recognized as the major site of inflammation in early stage of atherosclerosis. However, the cellular atlas of the heterogeneous adventitial cells, the intercellular communication, the cellular response of adventitia to hyperlipidemia, and its contribution to atherosclerosis have been elusive. Approach and Results- Single-cell RNA sequencing was applied to wt (wild type) and ApoE (apolipoprotein E)-deficient aortic adventitia from 12-week-old C57BL/6J mice fed on normal laboratory diet with early stage of atherosclerosis. Unbiased clustering analysis revealed that the landscape of adventitial cells encompassed adventitial mesenchyme cells, immune cells (macrophages, T cells, and B cells), and some types of rare cells, for example, neuron, lymphatic endothelial cells, and innate lymphoid cells. Seurat clustering analysis singled out 6 nonimmune clusters with distinct transcriptomic profiles, in which there predominantly were stem/progenitor cell-like and proinflammatory population (Mesen II). In ApoE-deficient adventitia, resident macrophages were activated and related to increased myeloid cell infiltration in the adventitia. Cell communication analysis further elucidated enhanced interaction between a mesenchyme cluster and inflammatory macrophages in ApoE-deficient adventitia. In vitro transwell assay confirmed the proinflammatory role of SCA1+ (stem cell antigen 1 positive) Mesen II population with increased CCL2 (chemokine [C-C motif] ligand 2) secretion and thus increased capacity to attract immune cells in ApoE-deficient adventitia. Conclusions- Cell atlas defined by single-cell RNA sequencing depicted the heterogeneous cellular landscape of the adventitia and uncovered several types of cell populations. Furthermore, resident cell interaction with immune cells appears crucial at the early stage of atherosclerosis.

Revealing cellular and molecular transitions in neonatal germ cell differentiation using single cell RNA sequencing.

Neonatal germ cell development provides the foundation of spermatogenesis. However, a systematic understanding of this process is still limited. To resolve cellular and molecular heterogeneity in this process, we profiled single cell transcriptomes of undifferentiated germ cells from neonatal mouse testes and employed unbiased clustering and pseudotime ordering analysis to assign cells to distinct cell states in the developmental continuum. We defined the unique transcriptional programs underlying migratory capacity, resting cellular states and apoptosis regulation in transitional gonocytes. We also identified a subpopulation of primitive spermatogonia marked by CD87 (plasminogen activator, urokinase receptor), which exhibited a higher level of self-renewal gene expression and migration potential. We further revealed a differentiation-primed state within the undifferentiated compartment, in which elevated Oct4 expression correlates with lower expression of self-renewal pathway factors, higher Rarg expression, and enhanced retinoic acid responsiveness. Lastly, a knockdown experiment revealed the role of Oct4 in the regulation of gene expression related to the MAPK pathway and cell adhesion, which may contribute to stem cell differentiation. Our study thus provides novel insights into cellular and molecular regulation during early germ cell development.

Novel Trajectories for Identifying Asthma Phenotypes: A Longitudinal Study in Korean Asthma Cohort, COREA

Unbiased cluster analysis has identified several asthma phenotypes. However, these phenotypes did not consistently predict disease prognosis and reflect temporal variability in airway inflammation. We aimed to identify longitudinal trajectories in terms of pulmonary function parameters and investigated whether the trajectories are associated with prognosis. Data were extracted from the Cohort for Reality and Evolution of Adult Asthma in Korea (COREA). Three-year pulmonary function test results were used to apply finite mixture models for group-based trajectory in 486 patients with eligible data set. Two main sets of longitudinal trajectories were identified in terms of FEV1% predicted, and FEV1 variability. In the 4 trajectories determined with FEV1% predicted, the pulmonary function showed a consistent course in 4 stratified levels during 3 years of follow-up, which was associated with unexpected hospital visits and the use of steroid bursts due to exacerbation. The variability in pulmonary function showed 3 different patterns, and we found that higher blood and sputum eosinophil levels were associated with the higher variability in pulmonary function and more exacerbations. Trajectory analysis is a novel method that provides longitudinal asthma phenotypes and aids in prediction of future risk of exacerbation. Further analysis is needed to validate the usefulness of these trajectories in an independent population.

The organization of leukotriene biosynthesis on the nuclear envelope revealed by single molecule localization microscopy and computational analyses.

The initial steps in the synthesis of leukotrienes are the translocation of 5-lipoxygenase (5-LO) to the nuclear envelope and its subsequent association with its scaffold protein 5-lipoxygenase-activating protein (FLAP). A major gap in our understanding of this process is the knowledge of how the organization of 5-LO and FLAP on the nuclear envelope regulates leukotriene synthesis. We combined single molecule localization microscopy with Clus-DoC cluster analysis, and also a novel unbiased cluster analysis to analyze changes in the relationships between 5-LO and FLAP in response to activation of RBL-2H3 cells to generate leukotriene C4. We identified the time-dependent reorganization of both 5-LO and FLAP into higher-order assemblies or clusters in response to cell activation via the IgE receptor. Clus-DoC analysis identified a subset of these clusters with a high degree of interaction between 5-LO and FLAP that specifically correlates with the time course of LTC4 synthesis, strongly suggesting their role in the initiation of leukotriene biosynthesis.

In-depth serum proteomics reveals biomarkers of psoriasis severity and response to traditional Chinese medicine.

Serum and plasma contain abundant biological information that reflect the body's physiological and pathological conditions and are therefore a valuable sample type for disease biomarkers. However, comprehensive profiling of the serological proteome is challenging due to the wide range of protein concentrations in serum.Methods: To address this challenge, we developed a novel in-depth serum proteomics platform capable of analyzing the serum proteome across ~10 orders or magnitude by combining data obtained from Data Independent Acquisition Mass Spectrometry (DIA-MS) and customizable antibody microarrays.Results: Using psoriasis as a proof-of-concept disease model, we screened 50 serum proteomes from healthy controls and psoriasis patients before and after treatment with traditional Chinese medicine (YinXieLing) on our in-depth serum proteomics platform. We identified 106 differentially-expressed proteins in psoriasis patients involved in psoriasis-relevant biological processes, such as blood coagulation, inflammation, apoptosis and angiogenesis signaling pathways. In addition, unbiased clustering and principle component analysis revealed 58 proteins discriminating healthy volunteers from psoriasis patients and 12 proteins distinguishing responders from non-responders to YinXieLing. To further demonstrate the clinical utility of our platform, we performed correlation analyses between serum proteomes and psoriasis activity and found a positive association between the psoriasis area and severity index (PASI) score with three serum proteins (PI3, CCL22, IL-12B).Conclusion: Taken together, these results demonstrate the clinical utility of our in-depth serum proteomics platform to identify specific diagnostic and predictive biomarkers of psoriasis and other immune-mediated diseases.

Extracellular vesicle microRNA cargo is correlated with HPV status in oropharyngeal carcinoma.

The incidence of human papilloma virus positive (HPV+ ) oropharyngeal squamous cell carcinoma (OPSCC) has increased rapidly in recent decades. These tumours have a favourable outcome compared to HPV-negative (HPV- ) OPSCC. However, HPV+ tumours are more likely to metastasise to distant sites, suggesting a difference in how these tumour subtypes interact with the metastatic niche. Extracellular vesicles (EVs) have emerged as important players in cell-to-cell communication and are a potential source of biomarkers for cancer diagnosis. This study aims to characterise the microRNA cargo of small EVs released by HPV+ and HPV- OPSCC cell lines. Extracellular vesicles produced by HPV+ (SCC2 and SCC90) and HPV- (SCC72 an SCC89) OPSCC cells were characterised by tunable resistive pulse sensing (TRPS) and western blotting. RNA was extracted from EVs and analysed by small RNA sequencing. A bioinformatics approach was used to identify EV miRNA signatures associated with HPV status. HPV- OPSCC cells produced more EVs than HPV+ OPSCC cells. EVs were positive for the common EV markers CD63, CD9 and TSG101. Unbiased hierarchical clustering analysis of EV miRNA cargo revealed that samples clustered based on HPV status. 14 miRNA were enriched in HPV+ cell-derived EVs, whereas 19 miRNA were enriched in EVs derived from HPV- cell lines. Here, we identify EV miRNA signatures indicative of the HPV status of the parent cell. This may provide a platform from which to validate salivary or blood-based biomarkers with utility for early detection and stratifying risk in OPSCC patients.

Association between pre-operative biological phenotypes and postoperative pulmonary complications: An unbiased cluster analysis.

Biological phenotypes have been identified within several heterogeneous pulmonary diseases, with potential therapeutic consequences. To assess whether distinct biological phenotypes exist within surgical patients, and whether development of postoperative pulmonary complications (PPCs) and subsequent dependence of intra-operative positive end-expiratory pressure (PEEP) differ between such phenotypes. Operating rooms of six hospitals in Europe and USA. Secondary analysis of the 'PROtective Ventilation with HIgh or LOw PEEP' trial. Adult patients scheduled for abdominal surgery who are at risk of PPCs. Measurement of pre-operative concentrations of seven plasma biomarkers associated with inflammation and lung injury. We applied unbiased cluster analysis to identify biological phenotypes. We then compared the proportion of patients developing PPCs within each phenotype, and associations between intra-operative PEEP levels and development of PPCs among phenotypes. In total, 242 patients were included. Unbiased cluster analysis clustered the patients within two biological phenotypes. Patients with phenotype 1 had lower plasma concentrations of TNF-α (3.8 [2.4 to 5.9] vs. 10.2 [8.0 to 12.1] pg ml; P < 0.001), IL-6 (2.3 [1.5 to 4.0] vs. 4.0 [2.9 to 6.5] pg ml; P < 0.001) and IL-8 (4.7 [3.1 to 8.1] vs. 8.1 [6.0 to 13.9] pg ml; P < 0.001). Phenotype 2 patients had the highest incidence of PPC (69.8 vs. 34.2% in type 1; P < 0.001). There was no interaction between phenotype and PEEP level for the development of PPCs (43.2% in high PEEP vs. 25.6% in low PEEP in phenotype 1, and 73.6% in high PEEP and 65.7% in low PEEP in phenotype 2; P for interaction = 0.503). Patients at risk of PPCs and undergoing open abdominal surgery can be clustered based on pre-operative plasma biomarker concentrations. The two identified phenotypes have different incidences of PPCs. Biologic phenotyping could be useful in future randomised controlled trials of intra-operative ventilation. The PROtective Ventilation with HIgh or LOw PEEP trial, including the substudy from which data were used for the present analysis, was registered at ClinicalTrials.gov (NCT01441791).

Time resolved quantitative phospho-tyrosine analysis reveals Bruton's Tyrosine kinase mediated signaling downstream of the mutated granulocyte-colony stimulating factor receptors.

Granulocyte-colony stimulating factor receptor (G-CSFR) controls myeloid progenitor proliferation and differentiation to neutrophils. Mutations in CSF3R (encoding G-CSFR) have been reported in patients with chronic neutrophilic leukemia (CNL) and acute myeloid leukemia (AML); however, despite years of research, the malignant downstream signaling of the mutated G-CSFRs is not well understood. Here, we utilized a quantitative phospho-tyrosine analysis to generate a comprehensive signaling map of G-CSF induced tyrosine phosphorylation in the normal versus mutated (proximal: T618I and truncated: Q741x) G-CSFRs. Unbiased clustering and kinase enrichment analysis identified rapid induction of phospho-proteins associated with endocytosis by the wild-type G-CSFR only; while G-CSFR mutants showed abnormal kinetics of canonical STAT3, STAT5 and MAPK phosphorylation, and aberrant activation of Bruton’s Tyrosine Kinase (Btk). Mutant-G-CSFR-expressing cells displayed enhanced sensitivity (3 to 5-fold lower IC50) for Ibrutinib-based chemical inhibition of Btk. Primary murine progenitor cells from G-CSFR-Q741x knock-in mice validated activation of Btk by the mutant receptor and retrovirally-transduced human CD34+ umbilical cord blood cells expressing mutant receptors displayed enhanced sensitivity to Ibrutinib. A significantly lower clonogenic potential was displayed by both murine and human primary cells expressing mutated receptors upon ibrutinib treatment. Finally, a dramatic synergy was observed between ibrutinib and ruxolinitib at lower dose of the individual drug. Together, these data demonstrate the strength of unsupervised proteomics analyses in dissecting oncogenic pathways, and suggest repositioning Ibrutinib for therapy of myeloid leukemia bearing CSF3R mutations. Phospho-tyrosine data is available via ProteomeXchange with identifier PXD009662.

Abstract 1003: Improving the functional taxonomy of breast cancer-associated fibroblasts by single-cell RNA sequencing reveals clinically relevant cellular subsets

Abstract Aim: The traditional tumor cell-centric view of cancer has been replaced in the past years with the appreciation of the importance of the tumor microenvironment for the malignant phenotype. Cancer-associated fibroblasts (CAFs) are the main constituent of the breast tumor microenvironment, although the exact origin and role of CAF subpopulations in disease initiation, progression and response to treatment remain unclear. Therefore, cancer therapies targeting CAFs are not available. Methods: Here, we delineate the heterogeneity of CAFs through single-cell RNA sequencing of 768 EPCAM-CD31-CD45-NG2- cells from the MMTV-PyMT mouse model of breast cancer. Results and discussion: Unbiased transcriptional clustering and differential gene expression analysis revealed four populations, with functionally distinct transcriptional profiles. The existence of these cell populations was further validated by immunostaining in experimental and human tumors revealing a preferential localization of each CAF subtype within the tumor microenvironment. Gene profiles specific to each CAF subpopulation held prognostic capability in clinical cohorts and correlated to known functional gene profiles. Our high-resolution comparison of single-cell transcriptomes in CAFs reveals previously unappreciated functional diversity within the tumor microenvironment. Conclusions: Ultimately, our delineation of an improved taxonomy of breast CAFs reveals a previously unappreciated functional diversity within the tumor microenvironment and opens the possibility for biomarker-driven development of new drugs for precision medicine. Citation Format: Michael Bartoschek, Kristian Pietras. Improving the functional taxonomy of breast cancer-associated fibroblasts by single-cell RNA sequencing reveals clinically relevant cellular subsets [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 1003.

Social stress shortens lifespan in mice.

SummaryStress and low socioeconomic status in humans confer increased vulnerability to morbidity and mortality. However, this association is not mechanistically understood nor has its causation been explored in animal models thus far. Recently, cellular senescence has been suggested as a potential mechanism linking lifelong stress to age‐related diseases and shorter life expectancy in humans. Here, we established a causal role for lifelong social stress on shortening lifespan and increasing the risk of cardiovascular disease in mice. Specifically, we developed a lifelong chronic psychosocial stress model in which male mouse aggressive behavior is used to study the impact of negative social confrontations on healthspan and lifespan. C57BL/6J mice identified through unbiased cluster analysis for receiving high while exhibiting low aggression, or identified as subordinate based on an ethologic criterion, had lower median and maximal lifespan, and developed earlier onset of several organ pathologies in the presence of a cellular senescence signature. Critically, subordinate mice developed spontaneous early‐stage atherosclerotic lesions of the aortic sinuses characterized by significant immune cells infiltration and sporadic rupture and calcification, none of which was found in dominant subjects. In conclusion, we present here the first rodent model to study and mechanistically dissect the impact of chronic stress on lifespan and disease of aging. These data highlight a conserved role for social stress and low social status on shortening lifespan and increasing the risk of cardiovascular disease in mammals and identify a potential mechanistic link for this complex phenomenon.

Profiling human breast epithelial cells using single cell RNA sequencing identifies cell diversity

Breast cancer arises from breast epithelial cells that acquire genetic alterations leading to subsequent loss of tissue homeostasis. Several distinct epithelial subpopulations have been proposed, but complete understanding of the spectrum of heterogeneity and differentiation hierarchy in the human breast remains elusive. Here, we use single-cell mRNA sequencing (scRNAseq) to profile the transcriptomes of 25,790 primary human breast epithelial cells isolated from reduction mammoplasties of seven individuals. Unbiased clustering analysis reveals the existence of three distinct epithelial cell populations, one basal and two luminal cell types, which we identify as secretory L1- and hormone-responsive L2-type cells. Pseudotemporal reconstruction of differentiation trajectories produces one continuous lineage hierarchy that closely connects the basal lineage to the two differentiated luminal branches. Our comprehensive cell atlas provides insights into the cellular blueprint of the human breast epithelium and will form the foundation to understand how the system goes awry during breast cancer.

Differential abundance of mDC subsets predict response to Hepatitis B vaccination

Abstract Vaccination as a strategy for the prevention of infectious disease has been one of the greatest success stories in modern medicine, resulting in the complete eradication of smallpox, the near eradication of polio, and the drastic lowering in the incidences of measles, mumps, influenza and other common diseases. In contrast to these remarkable successes, recent failures in the development of effective vaccines against other major public health threats, including HIV, TB and malaria have highlighted the limitations to the current approach of empirical vaccine design and the need to better understand the basic principles of how to recognize and elicit immune responses that generate effective and durable protective immunity. Single cell transcriptional profiling by RNA sequencing (scRNAseq) is a powerful tool for exploring the phenotypic diversity of cell populations present in peripheral blood and other tissue specimens in an unbiased fashion. We applied scRNAseq to FACS sorted blood samples collected from participants before and after challenge with the licensed Hepatitis B vaccine, and identified several innate cell subtypes using unbiased clustering analysis. Marker genes specific for each cell subtype were identified using random forest machine learning for use in qPCR. Our results suggest that the relative levels of myeloid dendritic cells (mDCs) specifically expressing the N-myc target gene, NDRG2, prior to vaccination predict serum antibody responses to the HepB vaccine.

Predictors of post-operative cardiovascular events, focused on atrial fibrillation, after valve surgery for primary mitral regurgitation.

Primary mitral regurgitation (PMR) can be considered as a heterogeneous clinical disease. The optimal timing of valve surgery for severe PMR remains unknown. To determine whether unbiased clustering analysis using dense phenotypic data (phenomapping) could identify phenotypically distinct PMR categories of patients. One hundred and twenty-two patients who underwent surgery were analysed, excluding patients with pre-operative permanent atrial fibrillation (AF), were prospectively included before surgery. They were given an extensive echocardiographic evaluation before surgery, and clinical data were collected. These phenotypic variables were grouped in clusters using hierarchical clustering analysis. Then, different groups were created using a dedicated phenomapping algorithm. Post-operative outcomes were compared between the groups. The primary endpoint was post-operative cardiovascular events (PCE), defined as a composite of: deaths, AF, stroke, and rehospitalization. The secondary endpoint was post-operative AF. Data from three phenogroups with different characteristics and prognoses were identified. Phenogroup-1 (67 patients) was the reference group. Phenogroup-2 (33 patients) included intermediate-risk male and smoker patients with heart remodelling. Phenogroup-3 (22 patients) included older female patients with comorbidities (chronic renal failure, paroxysmal AF) and diastolic dysfunction. They had a higher risk of developing both PCE [(hazard ratio) HR = 3.57(1.72-7.44), P < 0.001] and post-operative AF [HR = 4.75(2.03-11.10), P < 0.001]. Pre-operative paroxysmal AF was identified as an independent risk factor for PCE. Classification of PMR can be improved using statistical learning algorithms to define therapeutically homogeneous patient subclasses. High-risk patients can be identified, and these patients should be carefully monitored and may even be treated earlier.

Distinct protein signatures of acute myeloid leukemia bone marrow-derived stromal cells are prognostic for patient survival.

Mesenchymal stromal cells (MSC) support acute myeloid leukemia (AML) cell survival in the bone marrow (BM) microenvironment. Protein expression profiles of AML-derived MSC are unknown. Reverse phase protein array analysis was performed to compare expression of 151 proteins from AML-MSC (n=106) with MSC from healthy donors (n=71). Protein expression differed significantly between the two groups with 19 proteins over-expressed in leukemia stromal cells and 9 over-expressed in normal stromal cells. Unbiased hierarchical clustering analysis of the samples using these 28 proteins revealed three protein constellations whose variation in expression defined four MSC protein expression signatures: Class 1, Class 2, Class 3, and Class 4. These cell populations appear to have clinical relevance. Specifically, patients with Class 3 cells have longer survival and remission duration compared to other groups. Comparison of leukemia MSC at first diagnosis with those obtained at salvage (i.e. relapse/refractory) showed differential expression of 9 proteins reflecting a shift toward osteogenic differentiation. Leukemia MSC are more senescent compared to their normal counterparts, possibly due to the overexpressed p53/p21 axis as confirmed by high β-galactosidase staining. In addition, overexpression of BCL-XL in leukemia MSC might give survival advantage under conditions of senescence or stress and overexpressed galectin-3 exerts profound immunosuppression. Together, our findings suggest that the identification of specific populations of MSC in AML patients may be an important determinant of therapeutic response.

PULMONARY ARTERIAL HYPERTENSION CLINICAL PROFILES AND SURVIVAL USING LATENT CLASS ANALYSIS

We hypothesized that unbiased cluster analysis methods would identify distinct phenotypes of pulmonary arterial hypertension (PAH). We identified 191 patients with expert adjudicated PAH. A total of 47 clinical, laboratory, hemodynamic and echocardiographic variables were analyzed using Latent

Developmental Changes in HCN Channel Modulation of Neocortical Layer 1 Interneurons.

Layer 1 (L1) interneurons (INs) play a key role in modulating the integration of inputs to pyramidal neurons (PNs) and controlling cortical network activity. Hyperpolarization-activated, cyclic nucleotide-gated, non-specific cation (HCN) channels are known to alter the intrinsic and synaptic excitability of principal components (PCs) as well as select populations of GABAergic INs. However, the developmental profile and functional role of HCN channels in diverse L1 IN populations is not completely understood. In the present study, we used electrophysiological characterization, in conjunction with unbiased hierarchical cluster analysis, to examine developmental modulation of L1 INs by HCN channels in the rat medial agranular cortex (AGm). We identified three physiologically discrete IN populations which were classified as regular spiking (RS), burst accommodating (BA) and non-accommodating (NA). A distinct developmental pattern of excitability modulation by HCN channels was observed for each group. RS and NA cells displayed distinct morphologies with modulation of EPSPs increasing in RS cells and decreasing in NA cells across development. The results indicate a possible role of HCN channels in the formation and maintenance of cortical circuits through alteration of the excitability of distinct AGm L1 INs.

Big-Data Analysis, Cluster Analysis, and Machine-Learning Approaches.

Medicine will experience many changes in the coming years because the so-called "medicine of the future" will be increasingly proactive, featuring four basic elements: predictive, personalized, preventive, and participatory. Drivers for these changes include the digitization of data in medicine and the availability of computational tools that deal with massive volumes of data. Thus, the need to apply machine-learning methods to medicine has increased dramatically in recent years while facing challenges related to an unprecedented large number of clinically relevant features and highly specific diagnostic tests. Advances regarding data-storage technology and the progress concerning genome studies have enabled collecting vast amounts of patient clinical details, thus permitting the extraction of valuable information. In consequence, big-data analytics is becoming a mandatory technology to be used in the clinical domain.Machine learning and big-data analytics can be used in the field of cardiology, for example, for the prediction of individual risk factors for cardiovascular disease, for clinical decision support, and for practicing precision medicine using genomic information. Several projects employ machine-learning techniques to address the problem of classification and prediction of heart failure (HF) subtypes and unbiased clustering analysis using dense phenomapping to identify phenotypically distinct HF categories. In this chapter, these ideas are further presented, and a computerized model allowing the distinction between two major HF phenotypes on the basis of ventricular-volume data analysis is discussed in detail.