Psoriasis is a chronic inflammatory disease affecting 2% to 3% of the adult population and is associated with an increased risk of developing other inflammatory diseases, such as cardiovascular disease (CVD), inflammatory bowel disease, and diabetes [1,2]. Moderate-to-severe psoriasis is associated with increased risks of myocardial infarction and stroke independent of traditional risk factors, such as smoking, BMI, diabetes, hypertension, and dyslipidemia [3,4]. In addition, patients with psoriasis display increased prevalence of subclinical CVD, for example, endothelial dysfunction and increased carotid artery intimamedia thickness (CIMT), compared to controls [5-7]. Although the molecular mechanisms underlying the link between psoriasis and CVD remain to be identified in detail, they may rely, in part, on shared inflammatory pathways [8]. Indeed, psoriasis and atherosclerosis are both immune-driven chronic inflammatory diseases with an overlap of inflammatory mediators, including T-helper 1 (Th1) and Th17 cells [9]. It remains unclear, however, if treatment of psoriasis with systemic antipsoriatic medications can decrease the risk of CVD, but studies have indicated that tumor necrosis factor inhibitors and methotrexate may carry this potential [10,11].

Basic, translational, and clinical research directed separately at psoriasis or CVD has led to the identification of the molecular disease mechanisms and the development of new therapies aimed at psoriasis or CVD. However, this traditional paradigm is built on studies that do not capture deep phenotypes of individual patients, and to identify the central drivers of diseases and enable precision therapy, systems biology–oriented approaches are needed that seek to integrate all relevant available data. Advances in high-throughput molecular profiling and computational analysis techniques offer such novel opportunities to improve the understanding of disease connections and accelerate the discovery of new therapeutic strategies [12-15]. A multiscale biology approach has not, to our knowledge, been applied previously to provide comprehensive insights into the association between psoriasis and CVD. Therefore, we will use these techniques to investigate the pathophysiological links between psoriasis and CVD by integrating deep phenotypic data with multiomic data and use systems biology techniques to interrogate the molecular mechanisms connecting the two diseases (Figure 1). This approach will provide information on the fundamental regulatory circuits that drive psoriasis and CVD, enable the discovery and testing of novel biomarkers and therapeutics, and deliver other insights relevant for future prevention and therapy in these patients.

This study aims to investigate the complex multiscale interactions that drive cardiovascular risks in individual patients with psoriasis, involving specific gene transcripts, protein and lipid markers, signaling pathways, immune cell types, organ systems, and microbiota. Moreover, the study will examine multiscale differences across patient groups with psoriasis having distinct phenotypic traits, for example, with or without CVD, systemic antipsoriatic treatment or psoriatic arthritis, and early or late onset of psoriasis.

The study will include 120 patients (aged ≥30 years) with moderate-to-severe plaque psoriasis, with 60 having prior (over 6 months before inclusion) atherosclerotic CVD, including myocardial infarction, coronary revascularization, ischemic stroke, and/or peripheral artery disease, and 60 not having this history. Furthermore, half of the patients receive systemic antipsoriatic treatment (unchanged therapy in the preceding 3 months), while the other half of the patients do not receive systemic antipsoriatic treatment. The exclusion criteria are shown in Textbox 1.

Patients are recruited at the Department of Dermatology and Allergy, Herlev and Gentofte Hospital, when attending regularly scheduled visits for psoriasis control. Recruitment began in January 2018 and is also at the Department of Dermatology, Bispebjerg Hospital, at selected private dermatology clinics in the Copenhagen area, and through public advertisement of the research project on the home page of Herlev and Gentofte Hospital, social media outlets, and the member magazine of the Danish Psoriasis Association. All patient-related examinations are performed at Herlev and Gentofte Hospital. Because of the extensive examination program, the study visits are extended over 2 days of attendance within a period of 2 weeks.

Each participant completes a questionnaire containing questions regarding civil status, occupation, educational level, annual household income, psychological stress, depression, self-rated health, diet, and exercise. The questionnaire also contains questions to establish the Dermatology Quality of Life Index. An interview is conducted by study physicians with questions regarding history of psoriasis, current and previous medical treatment of psoriasis, history of CVD, medical treatment of CVD, family history of CVD, comorbidities (diabetes, hypertension, hypercholesterolemia, etc), odontological status, smoking status, alcohol consumption, other medications including use of antibiotics (exclusion if taken within 1 month before consideration for inclusion), and ethnicity.

A full skin examination is performed in each patient, and the severity of psoriasis is measured according to the Psoriasis Area and Severity Index and body surface area [16]. In addition, fingernails are examined for signs of psoriasis. Blood pressure is measured on each arm after the patient has rested for 5 minutes in the sitting position and is registered as the mean of these two measurements. BMI and waist-to-hip ratio are registered. The tongue is examined for the presence of a geographical tongue [17].

Isohelix DNA/RNA buccal swabs (SK-1S, Cell Projects Ltd) are used to collect samples from the skin. One swab is taken from a psoriasis plaque (if applicable), where the patient has at least 4 to 6 cm² of affected skin. A second swab is taken from adjacent clinically healthy skin. The sample is taken by rubbing the skin with the swab for 60 seconds and is stored in DNA/RNA shield (Zymo Research Corp). Site-specific skin microbiomes vary between different body regions [18], and therefore, we attempt to collect samples from the same body site among patients in the following priority: the lumbar area and buttocks, arms, and legs. Sampling controls are collected by holding the swabs freely in the examination room for 3 minutes. The skin microbiome is analyzed by shotgun metagenomics with untargeted sequencing of all microbial genomes [19].

Two 4-mm skin punch biopsies are taken under local anesthesia, including one from active psoriatic skin and one from clinically healthy skin neighboring the psoriatic biopsy area with a minimum distance of 2 cm from the psoriasis biopsy. The biopsy locations have the same priority for body sites as the skin swabs. After collection, biopsies are immediately cut in two parts and processed separately for RNA sequencing and immunohistochemistry.

Biopsies for RNA sequencing are placed directly into RNAprotect Tubes (QIAGEN), which are stored at 4°C overnight and thereafter at −80°C. RNA is isolated from biopsies, and its quality is assessed using the 2100 Bioanalyzer (Agilent). Sequencing libraries are prepared using SureSelect XT RNA Direct (Illumina) for samples with an RNA integrity number (RIN) score >8 and are sequenced at the Genomics Core Facility at Icahn School of Medicine at Mount Sinai.

Biopsies for immunohistochemistry are placed into a cryomold (Tissue Tek, Sakura Finetek) with optimal cutting temperature (OCT) solution (Tissue Tek, Sakura Finetek) that is subsequently snap-frozen with liquid nitrogen and immediately stored at −80^°C. OCT samples are cut, mounted, fixed, and stained with hematoxylin and eosin, and with antibodies to selected target antigens using standard immunohistochemical techniques.

Routine hematological and biochemical parameters, including high-sensitive C-reactive protein and N-terminal pro-brain natriuretic peptide, are analyzed at the Department of Clinical Biochemistry, Herlev and Gentofte Hospital, Denmark.

For mass cytometry profiling, blood is drawn in an ACD-A (acid citrate dextrose, Hettich Lab) tube, and 1 mL of blood is aliquoted into a tube containing 2 µL of Cell Activation Cocktail (phorbol 12-myristate-13-acetate [40.5 µM], ionomycin [669.3 µM], and Brefeldin A [2.5 mg/mL]; Biolegend). A second tube containing 1 mL of blood without the Cell Activation Cocktail is also collected. The tubes are mixed gently and then placed in an incubator at 37°C (5% CO₂)_. After 6 hours, 1.4 mL of proteomic stabilizer (PROT1, Smart Tube Inc) is added to each tube and mixed. The tubes are incubated for 10 minutes at room temperature and are then immediately placed at −80°C. Immune cell populations and cytokines are analyzed via mass cytometry by time-of-flight (CyTOF) [20] using a panel of 42 antibodies to profile various immune cells, including dendritic cells and T-cell subtypes, such as Th1, Th2, Th9, Th17, Th22, and Treg cells, in addition to relevant cytokines, such as interleukin (IL)-17 and IL-23.

For profiling of plasma proteins, blood is collected in ethylenediaminetetraacetic acid (EDTA) tubes and centrifuged for 10 minutes at 2000 rpm. The plasma is aliquoted into 1-mL tubes and immediately stored at −80^°C. Plasma concentrations of proteins are measured using the Olink Proseek multiplex assay (Olink Bioscience), which uses proximity extension assay technology to detect protein biomarkers in liquid samples [21]. In brief, pairs of antibodies linked to oligonucleotides bind to their target protein in close proximity so that the oligonucleotides can hybridize and generate a unique sequence that can be detected and quantified by subsequent quantitative real-time polymerase chain reaction [22,23]. Predesigned Olink multiplex biomarker panels (Inflammation, Cardiovascular II, and Cardiovascular III) are used to determine protein profiles [24].

Lipidomic and metabolomic profiling of plasma are assessed by Metabolon (Morrisville), where approximately 1100 lipids and 5200 metabolites are measured by ultra-high performance liquid chromatography/tandem mass spectrometry [25].

For transcriptomic profiling, blood (2.5 mL) is drawn into PAXgene RNA Tubes (BD Bioscience), placed at −20°C overnight, and then stored at −80°C. RNA is isolated using the PAXgene Blood RNA Kit (QIAGEN), and its quality is assessed using the 2100 Bioanalyzer (Agilent). Sequencing libraries are prepared using TruSeq Stranded Total RNA kits (Illumina) for samples with a RIN score >8 and are sequenced at the Genomics Core Facility at Icahn School of Medicine at Mount Sinai. Owing to the relatively small number of study subjects, genomics are currently not planned to be assessed in this work.

All patients receive an OMNIgene GUT kit (DNA Genotek) for collection of feces at a maximum of 14 days after the blood samples. The gut microbiome is analyzed by shotgun metagenomics [19].

A urine sample is collected, and the albumin-to-creatinine ratio is determined at the Department of Clinical Biochemistry, Herlev and Gentofte Hospital, Denmark.

Comprehensive two-dimensional resting transthoracic echocardiography, including tissue Doppler imaging, is performed using a Vivid E-95 ultrasound machine (GE Healthcare) with a M5Sc-D (1.4-4.6 MHz) transducer to determine myocardial structural and functional indices, such as left ventricular mass and systolic and diastolic function, right ventricular function, and left ventricular global longitudinal strain. Epicardial and pericardial adipose tissues are also measured in standard parasternal and short axis views [26,27]. All analyses are performed and stored in EchoPAC version 203.82 (GE Healthcare).

At a maximum of 14 days after the blood samples are taken, all patients undergo ¹⁸F-fluorodeoxyglucose (FDG) positron emission tomography/computed tomography (PET/CT). In brief, subjects are injected with 3.5 MBq (0.09 mCi) per kilogram FDG after fasting overnight. A whole-body FDG-PET/CT is performed 120 minutes after FDG injection on a GE Discovery 710 scanner (GE Medical Systems) using the proprietary Q.Clear PET reconstruction algorithm. Anatomic localization and attenuation correction are provided by an unenhanced low-dose CT scan. Regions of interest (ROIs) are delineated around the aorta in consecutive 3-mm-thick axial PET and CT slices using MIM 6.9.2 software (MIM Software Inc). FDG uptake in aortic segments is quantified by calculating the mean of the maximum activity in each ROI normalized to the mean activity in the superior vena cava (maximum target-to-background ratio [TBR_max]) to determine vascular inflammation according to established methodology [28]. Adipose tissue in brown, pericardial, visceral, and subcutaneous fat is measured by manually locating volumes of interest in the lateral neck, anterior to the pericardium at the level of the aortic root, caudal to the kidneys, and in the loin, respectively. Moreover, FDG uptakes in the bone marrow (lumbar vertebra L1-L5) and spleen are measured to assess activation of hematopoietic tissues. Separate electrocardiogram-gated low-dose CT acquisition is performed for determination of the coronary calcium score using a combined Agatston score for all coronary arteries using Siemens SyngoVia software VB40 (Siemens Healthcare) [29,30].

Ultrasound imaging of CIMT is performed in the right and left common carotid arteries by using an Affiniti 70G ultrasound system with a 5-12 MHz linear array transducer (Philips Ultrasound Inc) with Philips Q-App IMT software (version 3.03). CIMT is measured according to the Mannheim consensus on a 10-mm far wall segment of the distal common carotid artery during diastole, avoiding areas with focal thickening [31]. The presence of atherosclerotic plaques is also assessed in both carotid arteries.

For transcriptomics, RNA sequencing (RNA-seq) FASTQ sequence files are first subject to quality trim followed by alignment to the HG38 human genome and count summarization. Samples are normalized, and differential expression analysis is carried out to detect genes that are expressed at significantly different levels between groups. Gene set enrichment analysis is performed by the CAMERA method using reference data sets that include Hallmark, KEGG, Reactome, and Gene Ontology [32]. Protein-protein interaction network analysis is carried out using Cytoscape with the StringAPP plugin [33-35]. Functional enrichment analysis of the interaction network is performed using STRING enrichment against a collection of gene set databases [34,35].

For metagenomics, FASTQ files are quality trimmed before human genome mapping to filter out human reads. Microbial taxonomy classification is performed by the k-mer–based Kraken2 tool supplied with a microbial database including archaea, bacteria, fungi, protozoa, viruses, and vectors [36]. Relative taxonomy abundance is analyzed by LEfSe, and absolute abundance is analyzed by DESeq2 [37,38]. Alpha diversity is calculated using the phyloseq package in RStudio (R version 3.6) [39].

For high dimensional CyTOF, data are first preprocessed in Cytobank, and then, FCS data files are imported into RStudio [40]. Data are arcsinh-transformed and then scaled for heatmap presentation. Unscaled transformed data are used for FlowSOM hierarchical clustering [41]. The clusters are then manually annotated and visualized after UMAP dimension reduction [42]. Differential cell lineage markers are further quantified and cytokine levels are stratified by clustered cell populations.

For proteomics, plasma protein levels are measured using the Cardiovascular II, Cardiovascular III, and Inflammation panels (Olink Bioscience), and reported as normalized protein expression levels (NPX) in log2 scale. Proteins are filtered out when 40% of samples are below the limit of detection. Protein changes are analyzed with the limma-trend method for comparisons of interest. Functional enrichment analysis is performed as described above for transcriptomic data [43].

The strength of a systems biology approach is the potential to integrate data from multiple platforms. There are several examples of how this approach has been successfully applied in the disease areas of psoriasis and in particular CVD [44,45]. We will use previously described computational methods to integrate multiomics data sets to develop a comprehensive and integrated view of cardiovascular risk in patients with psoriasis [46-48].