The study proposes a novel machine learning (ML) paradigm for cardiovascular disease (CVD) detection in individuals at medium to high cardiovascular risk using data from a Greek cohort of 542 individuals with rheumatoid arthritis, or diabetes mellitus, and/or arterial hypertension, using conventional or office-based, laboratory-based blood biomarkers and carotid/ femoral ultrasound image-based phenotypes. Two kinds of data (CVD risk factors and presence of CVD-defined as stroke, or myocardial infarction, or coronary artery syndrome, or peripheral artery disease, or coronary heart disease) as ground truth, were collected at two-time points: (i) at visit 1 and (ii) at visit 2 after 3 years. The CVD risk factors were divided into three clusters (conventional or office-based, laboratory-based blood biomarkers, carotid ultrasound image-based phenotypes) to study their effect on the ML classifiers. Three kinds of ML classifiers (Random Forest, Support Vector Machine, and Linear Discriminant Analysis) were applied in a two-fold cross-validation framework using the data augmented by synthetic minority over-sampling technique (SMOTE) strategy. The performance of the ML classifiers was recorded. In this cohort with overall 46 CVD risk factors (covariates) implemented in an online cardiovascular framework, that requires calculation time less than 1 s per patient, a mean accuracy and area-under-the-curve (AUC) of 98.40% and 0.98 (p < 0.0001) for CVD presence detection at visit 1, and 98.39% and 0.98 (p < 0.0001) at visit 2, respectively. The performance of the cardiovascular framework was significantly better than the classical CVD risk score. The ML paradigm proved to be powerful for CVD prediction in individuals at medium to high cardiovascular risk. KeywordsCardiovascular risk estimation • Cardiovascular disease • Three-year follow-up • Conventional risk factors • Ultrasound • And machine learning Abbreviations ANOVA Analysis of variance ASCVD Atherosclerotic cardiovascular disease AUC Area-under-the-curve BMI Body mass index CAD Coronary artery disease CCVRC Conventional cardiovascular risk calculators Cluster 1 Conventional office-based biomarkers Cluster 2 Fusion of office-based biomarker and laboratory-based biomarkers Cluster 3 Fusion of office-based biomarker, laboratory-based biomarker, and carotid ultrasound image phenotypes CUSIP Carotid ultrasound image phenotype CV Cross-validation CVD Cardiovascular disease CVD-3YFU Cardiovascular disease risk-three-year follow-up CVD-CR Cardiovascular disease-current risk CVE Cardiovascular events DM Diabetes mellitus FH Family history FNR False-negative rate FPR False-positive rate FRS Framingham risk score HTN Hypertension

Waldenström macroglobulinemia (WM) is a low-grade B- cell lymphoproliferative disorder characterized by bone marrow (BM) infiltration by small lymphoplasmacytic lymphoma (LPL) cells that secrete monoclonal IgM immunoglobulin. Our understanding of the pathogenesis of WM has improved significantly by the discovery that the vast majority of patients with WM harbor the somatic mutation L265P in MYD88 gene on their clonal cells. However, the small fraction of patients (3-10%) that lack mutations in MYD88 is characterized by a different clinical course, with an increased risk of disease transformation and probably shorter overall survival. WM is a clonal disease but there may be significant heterogeneity within the clone, which may be associated with clinical outcomes, however, this has not been extensively explored. Recent advances in the area of single-cell RNA sequencing (scRNA-Seq) technologies have improved so that it is feasible to sequence and analyze thousands of cells per tumor delivering significant insights into a tumor's cellular heterogeneity and the biological features that distinguish different cell populations. To better understand the differences occurring in the B cell composition between WM patients harboring the MYD88 L265Pmutation and those with MYD88 WT respectively, we performed single-cell RNA sequencing on CD19+ sorted cells from BM aspirates of these patients. Both patients were wild type to CXCR4 mutations. Overall, we characterized 8,517 cells from one MYD88 L265P WM patient (3,116 cells), one MYD88 WT patient (3,348 cells) and one healthy donor as control (2,053 cells). Clustering of these cells based on their gene expression profile, revealed that the majority of the cells form a large cluster corresponding to B cells (naïve, immature and mature;6852 cells) while the rest of the cells were classified into four broad cell types ranging from pro-B cells (1243 cells) to only few cells belonging to mature populations engaged in the immune response such as plasma cells (190 cells), T cells (124 cells) and monocytes (53 cells) indicating a high purity of the samples (≥95%) (Fig 1A). Clonality assessment showed high lambda light chain expressing populations in both patients and as expected an evenly distributed polyclonal population in the healthy donor. To investigate the relative composition of the different cell types between the two genetically distinct WM patients (MYD88 L265P vs MYD88 WT) we searched for populations that were enriched in the B cell repertoire compared to the normal BM. From our data we observe a significant enrichment of pro-B cells in the MYD88 WT patient which is almost completely missing in the MYD88 L265P patient and an enhanced naïve B cell population in MYD88 L265P patient compared to both MYD88 WT patient and the healthy donor (Fig 1B, C). In addition, immature B cells are 3-fold higher in WM patients compared to the healthy donor. Differential gene expression analysis in B cells showed that among both patients several genes are highly expressed (&gt;1.5 log2 fold-change) compared to the healthy donor such as CD52, FCMR, CD74, CD37 and HLA-DPA1. Genes that were highly upregulated in the MYD88 mutant compared to the wild-type WM patient include CXCR4, TXNIP, ID3, LLT1, MTRNR2L8, ZFP36L2 and ZNF331. On the other hand, genes that are highly upregulated in the wild-type phenotype compared to the mutant phenotype include MS4A1, CD79A, CD82, CD1C, KIAA0040, CD72, EGR1, SYK, and LGALS1 (Fig 1D, E). Distinct pathways enriched by genes differentially expressed by the MYD88 mutant B cell population include response to corticosteroid and hypoxia whereas in the MYD88 wild type patient unique processes include B cell activation and receptor signaling pathways, immune response and NFκB signaling. In summary, our results show that there is a distinctive transcriptional profile in the B cells from the MYD88 WT patient where most differentially expressed genes do not overlap with the gene expression of the B cells from the MYD88 L265P patient, which may provide further mechanistic insights into the differences in biology and clinical presentation in patients with MYD88 WT versus those harboring MYD88 L265P. Further evaluation in additional patient samples is ongoing to gain additional insights in pathways that may be involved in disease transformation and response to therapy with targeted therapies, either targeting CD20 or BCR pathway. Figure 1 Figure 1. Disclosures Gavriatopoulou: Genesis: Honoraria; Sanofi: Honoraria; Karyopharm: Honoraria; Takeda: Honoraria; GSK: Honoraria; Janssen: Honoraria; Amgen: Honoraria. Terpos: Genesis: Consultancy, Honoraria, Research Funding; GSK: Honoraria, Research Funding; Janssen-Cilag: Consultancy, Honoraria, Research Funding; Novartis: Honoraria; Takeda: Consultancy, Honoraria, Research Funding; Sanofi: Consultancy, Honoraria, Research Funding; Celgene: Consultancy, Honoraria, Research Funding; BMS: Honoraria; Amgen: Consultancy, Honoraria, Research Funding. Dimopoulos: BMS: Honoraria; Takeda: Honoraria; Beigene: Honoraria; Janssen: Honoraria; Amgen: Honoraria. Kastritis: Amgen: Consultancy, Honoraria, Research Funding; Genesis Pharma: Honoraria; Takeda: Honoraria; Janssen: Consultancy, Honoraria, Research Funding; Pfizer: Consultancy, Honoraria, Research Funding.

A key unknown of the functional space in tumor immunity is whether physiologically relevant cancer antigen presentation occurs solely in draining lymph nodes versus tumors. Professional antigen presenting cells, i.e. the dendritic cells, are scarce and immature within tumors, greatly outnumbered by MHCII expressing non-hematopoietic cells, such as antigen-presenting cancer-associated fibroblasts (apCAFs). We hypothesized that after their exit from tumor-draining lymph nodes T cells depend on a second wave of antigen presentation provided in situ by structural cells. We show that dense apCAF regions in human lung tumors define hot immunological spots with increased numbers of CD4 T cells. The transcriptomic profile of human lung apCAFs aligned to that of pancreatic apCAFs across mice and humans and were both enriched for alveolar type II genes, suggesting an epithelial origin. Mechanistically, human apCAFs directly activated the TCRs of adjacent effector CD4 T cells and at the same time produced high levels of c1q, which acted on surface c1qbp on T cells to rescue them from apoptosis. Fibroblast-specific deletion of MHCII in mice impaired local MHCII immunity and accelerated tumor growth, while inducing c1qbp overexpression in adoptively transferred T cells expanded their numbers within tumors and reduced tumour burden. Collectively, our work shows that tumor T cell immunity post lymph node exit requires peripheral antigen presentation by a subset of CAFs and proposes a new conceptual framework upon which effective cancer immunotherapies can be built.