Investigation of plasma-derived extracellular vesicles in hypertrophic cardiomyopathy patients Authors: Alessandra Stefania Rizzuto, Andrea Faggiano, Chiara Macchi, Isabella Fichtner, Daniele Bianchi, Alice Mallia, Eleonora Gnan, Francesco Tafuri, Andrea Pandolfi, Michele Baroni, Stefania Paganini, Roberto Meazza, Ilaria Giusti, Vincenza Dolo, Paola Castronovo, Palma Finelli, Alberto Corsini, Marco Vicenzi, Roberto Sassi, Cristina Banfi, Massimo Walter Rivolta, Margherita Calcagnino, Stefano Carugo, Massimiliano Ruscica Published: 2026 ABSTRACT SummaryHypertrophic cardiomyopathy (HCM) is a genetically determined heart disease characterized by marked clinical heterogeneity and limited availability of specific circulating biomarkers that reflect underlying disease mechanisms. This study investigated whether plasma-derived extracellular vesicles (EVs) capture disease-related molecular features in patients with HCM. Plasma EVs were characterized for size, cellular origin, and protein cargo using flow cytometry and targeted and untargeted proteomic analyses. While size and concentration of EVs were comparable between groups, patients with HCM exhibited an enrichment of platelet-derived EVs (CD41a+) and reductions in neutrophil (CD66b+)- and lymphatic-endothelial-cell (CD310+)-derived EVs. Proteomic profiling revealed EV-associated proteins linked to platelet activation and thrombo-inflammation. Multivariable protein signatures derived from EVs discriminated patients from controls with performance comparable to plasma-based models and were not influenced by age. These findings indicate that plasma-derived EVs capture key biological pathways involved in HCM and may complement biomarker discovery strategies.Keywords: biomarkers, extracellular vesicles, cardiomyopathies, heart diseases Introduction Hypertrophic cardiomyopathy (HCM), characterized by abnormal thickening of the heart, is the most prevalent Mendelian monogenic cardiomyopathy, affecting upward of 1 in 500 individuals.1 It is primarily caused by dominantly inherited actionable variants in genes that encode proteins of the cardiac sarcomere or Z disk.2 The manifestations of the disease, including symptoms, vary widely among patients with HCM, and sometimes the initial presentation is a major adverse cardiac event, including sudden cardiac death (SCD).3 Despite significant advancements in understanding the genetic basis of HCM, this knowledge has had limited impact on treatment and prognosis, mainly due to a lack of understanding of the factors driving disease progression.4 Currently, clinical management for HCM combines genetic testing to identify known actionable causative variants with electrocardiography, exercise tests—optionally complemented by cardiopulmonary exercise testing—to assess functional capacity and unmask dynamic left ventricular outflow obstruction (LVOTO), cardiac imaging (echocardiography and cardiac magnetic resonance), and clinical red flags (such as syncope) to refine the overall phenotype, monitor disease progression, and stratify prognosis.5 At present, the clinical utility of circulating biomarkers in HCM remains limited. Although N-terminal pro-B-type natriuretic peptide (NT-proBNP) and high-sensitivity cardiac troponins (hs-cTns) are associated with disease severity and adverse outcomes, their integration into routine clinical decision-making is still evolving. Additional biomarkers reflecting myocardial fibrosis, inflammation, and endothelial dysfunction also show potential but require more robust validation before they can be implemented clinically.6 Therefore, there is a need for new biomarkers to better define diagnosis, and track disease progression, enabling the assessment of both existing and novel therapies.7 Extracellular vesicles (EVs) are a heterogeneous population of cell-derived membranous nanoparticles released into virtually all biological fluids, including blood, saliva, lymph, pericardial fluid, and urine.8 During their biogenesis, EVs acquire surface molecules and intraluminal cargo—comprising proteins, nucleic acids, and lipids—that closely mirror the physiological or pathological state of their donor cell.9,10 Hence, the molecular composition of EVs is not static: it dynamically changes in response to cellular stress, activation, or disease, resulting in qualitative and quantitative alterations of their cargo. Importantly, the repertoire of surface markers expressed on EV membranes provides valuable information about their cellular origin, enabling the tracing of EVs back to specific tissues or cell types.11 Through their capacity to transfer a wide array of biomolecules (including proteins, lipids, nucleic acids, and metabolites) between cells,12 EVs play a pivotal role in mediating intercellular and inter-organ communication, influencing a broad range of physiological and pathological processes.13 These unique properties position EVs as promising non-invasive indicators of tissue and organ status, offering a means to monitor otherwise inaccessible biological systems.14 In the context of cardiovascular diseases, numerous studies have demonstrated that circulating EV levels are increased in patients with heart failure15 and that the abundance of specific EV subsets differs between patients with chronic and acute heart disease.16 Beyond their quantitative changes, recent findings indicate that EVs are not only promising biomarkers but also active molecular mediators in the pathophysiology of cardiovascular disorders. For instance, epicardial-adipose-tissue-derived EVs have been shown to participate in myocardial and vascular remodeling, highlighting their involvement in heart-fat crosstalk and in modulating local inflammatory and metabolic processes.17 Similarly, EV-associated miRNAs have been implicated in regulating atherosclerotic burden, further supporting their contribution to disease progression beyond their diagnostic potential.18 Together, these findings underscore the role of EVs as both messengers and effectors in cardiovascular pathology, reinforcing their promise as valuable biomarker elements in clinical settings. In HCM, growing attention has focused on the contribution of EVs to disease pathogenesis.19 At least in in vitro models, EVs carrying specific proteins and miRNAs appear to promote cardiomyocyte hypertrophy, a mechanism corroborated by transcriptomic analyses of EVs isolated from cardiomyocytes derived from human-induced pluripotent stem cells of HCM patients.20 Few studies, however, have addressed the circulating profile of plasma-derived EVs in patients with HCM.19 Nevertheless, multiple reports show that alterations in EV features—such as circulating concentration, size, and protein content—associate with disease presence and severity in cardiac disorders.21 Given the current lack of precise diagnostic and prognostic biomarkers for HCM, in-depth characterization of circulating EVs and their molecular cargo in this condition could be clinically transformative. To pursue this aim, the primary objective of this single-center, cross-sectional case-control study was to comprehensively characterize the subcellular origin of EVs, in individuals with a diagnosis of HCM compared with healthy volunteers. As a secondary objective, we characterized the protein cargo of EVs to explore their potential as a diagnostic tool. Results Clinical characteristics of the study population A total of 46 patients with hypertrophic cardiomyopathy (HCM; median age: 61 years [interquartile range [IQR]: 50–69.5]) and 31 control subjects (CTRL; median age: 48 years [IQR: 44–56]) were enrolled in the study. The proportion of males did not differ significantly between the groups (HCM, 69.5%; CTRL, 51.6%; p = 0.18). As reported in Table 1, body surface area and most routine hematological parameters, including red blood cell count, hemoglobin levels, platelet count, lymphocytes, monocytes, and eosinophils, were comparable between groups. In contrast, HCM patients exhibited significantly higher total white blood cell counts and neutrophil counts compared with CTRL, resulting in an increased neutrophil-to-lymphocyte ratio (NLR) (all p < 0.05). Echocardiographic parameters related to chamber size, including end-diastolic volume index (EDVI), end-systolic volume index (ESVI), or LV end-diastolic diameter (LVIDd), did not differ between groups. HCM patients exhibited markedly increased interventricular septal thickness in diastole (IVSd), posterior wall thickness in diastole (PWTD), maximum wall thickness (MWT), and left ventricular mass indexed to body surface area compared with CTRL (IVSd: 16 mm compared to 8 mm; PWTD: 11 mm compared to 8 mm; MWT: 17 mm compared to 9 mm; 120.6 g/m2 compared to 60.8 g/m2; all p < 0.0001). When available (n = 25), cardiac magnetic resonance (CMR) confirmed increased LV (left ventricular) mass in HCM patients (median ∼77 g/m2 [IQR: 67-103.5]). Finally, LVEF (left ventricular ejection fraction) was significantly lower in HCM compared with CTRL (58% compared to 62%; p = 0.014), whereas estimated LV filling pressures, as assessed by ratio of early mitral inflow velocity to early mitral annular tissue Doppler velocity (E/e’ ratio), were significantly higher (8.54 [IQR: 7.15–11.14] compared to 5.5 [4.9–6.8]). CMR-detected late gadolinium enhancement (LGE), reflecting myocardial fibrosis, was present in 19 out of 25 HCM patients (76%), whereas LVOTO was identified in about 6.5% of the cohort. Genetic testing was performed in 43 patients (93%), while the remaining three either declined testing or lacked living relatives. As detailed in Table 2, an actionable causative variant (pathogenic or likely pathogenic) was identified in 20 (46.5%) of the 43 HCM patients who underwent genetic testing, most commonly in the sarcomeric genes MYBPC3 (myosin binding protein C3) and MYH7 (myosin heavy chain 7) (n = 10 and n = 7, respectively). Variants of uncertain significance in the absence of a pathogenic or likely pathogenic variant were observed in 11 patients (25.6%), who were classified as VUS-only, whereas 12 patients (27.9%) had no identifiable disease-causing variant and were classified as genotype-negative. Table S1 presents a patient-level summary of all detected variants along with their American College of Medical Genetics and Genomics (ACMG) classification.22 Risk stratification based on the HCM risk-SCD score ranged from 0.9% to 8.2%. Most patients were classified as low risk (<4%; N = 41, 89.1%), while four patients (8.7%) fell into the intermediate-risk category (4%–6%). Only one patient (2.2%) was identified as being at high risk of SCD (>6%). In addition, 12 patients (26.1%) already carried an implantable cardioverter-defibrillator (ICD) for primary prevention of SCD, based on clinical judgment, most commonly associated by a single major risk factor (e.g., family history of juvenile SCD, LVEF <50%, or extensive LGE). Table 1Clinical characteristics, blood work parameters, and echocardiographic findings of CTRL subjects and patients with HCMParameterCTRL (N = 31)HCM (N = 46)p valueAge (years)48 (44.0–56.0)61 (50.0–69.5)0.0001BSA (m2)1.8 (1.5–1.9)1.8 (1.7–2.1)0.1847Male (%)16 (51.6%)32 (69.5%)0.1760Blood workWBC (109/L)5.97 (5.1–6.5)6.815 (5.1–8.2)0.0246RBC (1012/L)4.86 (4.3–5.0)4.895 (4.4–5.1)0.4126Hb (g/dL)14.4 (13.1–15.1)14.3 (13.0–15.0)0.6318PLT (109/L)257 (209.3–277.0)229 (194.0–262.0)0.1171Neutrophils (109/L)2.94 (2.3–3.4)3.7 (2.8–5.5)0.0200Lymphocytes (109/L)1.99 (1.5–2.2)1.8 (1.5–2.2)0.3590Monocytes (109/L)0.50 (0.4–0.6)0.5 (0.4–0.6)0.5160Eosinophils (109/L)0.17 (0.1–0.2)0.16 (0.10–0.2)0.8000NLR1.59 (1.31–1.77)1.98 (1.40–3.0)0.0421NT-proBNP (pg/mL)57 (30–100)401 (161.8–1114)<0.001Echocardiographic assessmentEDVI (mL/m2)47 (43.0–55.0)51.0 (38.7–64.0)0.8043ESVI (mL/m2)17 (15.0–21.0)19.7 (15.0–29.0)0.0807IVSD (mm)8 (7.0–10.0)16.0 (14.0–18.0)<0.0001LVID (mm)44 (42.0–47.0)44.0 (40.0–49.0)0.9440PWTD (mm)8 (7.0–8.0)11.0 (10.0–13.0)<0.0001MWT (mm)9 (8.0–10.0)17.0 (15.0–19.0)<0.0001LV Mass/BSA (g/m2)60.8 (55.4–68.0)120.6 (101.8–148.6)<0.0001LVEF (%)62 (59.0–63.0)58.0 (54.7–61.2)0.0140LAVi(mL/m2)24.5 (19.2–30.7)42.5 (30.0–56.7)<0.0001E/e'ratio5.5 (4.9–6.8)8.5 (7.2–11.1)<0.0001E/aratio1.2 (1–1.5)0.95 (0.7–1.3)0.0398CMR-LGE (%)–76–LVOTO present, n (%)0 (0%)3 (6.5%)–Comprehensive comparison of demographic characteristics, hematological parameters, and echocardiographic features between control subjects (CTRL) and patients with HCM. The table summarizes baseline clinical data, complete blood count variables, and detailed echocardiographic measurements obtained from 31 CTRL individuals and 46 HCM patients. Demographic variables include age, body surface area (BSA), and sex distribution. Hematological assessment comprises total white blood cell count (WBC), red blood cell count (RBC), hemoglobin concentration (Hb), platelet count (PLT), and differential leukocyte counts, including neutrophils, lymphocytes, monocytes, eosinophils, and neutrophil-to-lymphocyte ratio (NLR). N-terminal pro-B-type natriuretic peptide (NT-proBNP).Echocardiographic evaluation includes indexed left ventricular end-diastolic and end-systolic volumes (EDVI and ESVI), interventricular septal thickness in diastole (IVSD), left ventricular internal diameter in diastole (LVID), posterior wall thickness in diastole (PWTD), maximal wall thickness (MWT), left ventricular mass indexed to body surface area (LV mass/BSA), left ventricular ejection fraction (LVEF), left atrial volume index (LAVi), and Doppler-derived indices of diastolic function, including E/e′ and E/A ratios. The presence of cardiac-magnetic-resonance-detected late gadolinium enhancement (CMR-LGE) and left ventricular outflow tract obstruction (LVOTO) is reported for the HCM cohort only. Continuous variables are presented as median and interquartile range (IQR), while categorical variables are expressed as absolute number and percentage. Statistical comparisons between CTRL and HCM groups were performed using the Mann-Whitney U test for continuous variables, with corresponding p values reported. A p value ≤ 0.05 was considered statistically significant. Bold values indicate statistically significant differences between groups. Clinical characteristics, blood work parameters, and echocardiographic findings of CTRL subjects and patients with HCM Comprehensive comparison of demographic characteristics, hematological parameters, and echocardiographic features between control subjects (CTRL) and patients with HCM. The table summarizes baseline clinical data, complete blood count variables, and detailed echocardiographic measurements obtained from 31 CTRL individuals and 46 HCM patients. Demographic variables include age, body surface area (BSA), and sex distribution. Hematological assessment comprises total white blood cell count (WBC), red blood cell count (RBC), hemoglobin concentration (Hb), platelet count (PLT), and differential leukocyte counts, including neutrophils, lymphocytes, monocytes, eosinophils, and neutrophil-to-lymphocyte ratio (NLR). N-terminal pro-B-type natriuretic peptide (NT-proBNP). Echocardiographic evaluation includes indexed left ventricular end-diastolic and end-systolic volumes (EDVI and ESVI), interventricular septal thickness in diastole (IVSD), left ventricular internal diameter in diastole (LVID), posterior wall thickness in diastole (PWTD), maximal wall thickness (MWT), left ventricular mass indexed to body surface area (LV mass/BSA), left ventricular ejection fraction (LVEF), left atrial volume index (LAVi), and Doppler-derived indices of diastolic function, including E/e′ and E/A ratios. The presence of cardiac-magnetic-resonance-detected late gadolinium enhancement (CMR-LGE) and left ventricular outflow tract obstruction (LVOTO) is reported for the HCM cohort only. Continuous variables are presented as median and interquartile range (IQR), while categorical variables are expressed as absolute number and percentage. Statistical comparisons between CTRL and HCM groups were performed using the Mann-Whitney U test for continuous variables, with corresponding p values reported. A p value ≤ 0.05 was considered statistically significant. Bold values indicate statistically significant differences between groups. Table 2Genetic variant classification in the HCM cohortVariantN%Pathogenic (P, class V)1227.9└─ MYBPC35–└─ MYH75–└─ CACNA1C1–└─ MYL31–Likely pathogenic (LP, class IV)818.6└─ MYBPC35–└─ MYH72–└─ FHOD31–VUS-only (class III)1125.6Genotype-negative1227.9Detailed distribution and classification of genetic variants identified in HCM cohort. The table summarizes the results of genetic testing performed in patients with HCM, reporting the number and relative frequency of individuals carrying variants classified according to established pathogenicity criteria. Genetic variants were categorized into four groups: pathogenic variants (class V), likely pathogenic variants (class IV), variants of uncertain significance (VUS; class III), and genotype-negative individuals, defined as patients in whom no reportable variants were identified. Among patients harboring pathogenic or likely pathogenic variants, the distribution of variants across the main sarcomeric genes implicated in HCM, including MYBPC3 (myosin binding protein C3) and MYH7 (myosin heavy chain 7), as well as other less frequently involved genes (CACNA1C, L-type voltage-gated calcium channel; FHOD3, formin homology 2 domain containing 3) is reported. VUS-only indicates the presence of one or more variants of uncertain significance without a pathogenic or likely pathogenic variant. Genotype-negative indicates that no reportable variant was identified. Data are expressed as absolute numbers and percentages relative to the 43 HCM patients who underwent genetic testing, providing an overview of the genetic architecture of the study population and the relative contribution of different variant classes and sarcomeric genes to disease burden. Genetic variant classification in the HCM cohort Detailed distribution and classification of genetic variants identified in HCM cohort. The table summarizes the results of genetic testing performed in patients with HCM, reporting the number and relative frequency of individuals carrying variants classified according to established pathogenicity criteria. Genetic variants were categorized into four groups: pathogenic variants (class V), likely pathogenic variants (class IV), variants of uncertain significance (VUS; class III), and genotype-negative individuals, defined as patients in whom no reportable variants were identified. Among patients harboring pathogenic or likely pathogenic variants, the distribution of variants across the main sarcomeric genes implicated in HCM, including MYBPC3 (myosin binding protein C3) and MYH7 (myosin heavy chain 7), as well as other less frequently involved genes (CACNA1C, L-type voltage-gated calcium channel; FHOD3, formin homology 2 domain containing 3) is reported. VUS-only indicates the presence of one or more variants of uncertain significance without a pathogenic or likely pathogenic variant. Genotype-negative indicates that no reportable variant was identified. Data are expressed as absolute numbers and percentages relative to the 43 HCM patients who underwent genetic testing, providing an overview of the genetic architecture of the study population and the relative contribution of different variant classes and sarcomeric genes to disease burden. Characterization of plasma-derived EVs EVs were isolated from plasma samples of study participants using differential ultracentrifugation. Nanoparticle tracking analysis (NTA) was performed to assess the size distribution and concentration of plasma-derived EVs. Median EV size in HCM did not differ significantly from CTRL (HCM, 181.5 nm; [IQR: 166.7–194.8]; CTRL, 180.6 nm [IQR: 163.5–193.1]). Similarly, EV concentrations were comparable between groups (HCM, 3.13 × 109 particles/mL [IQR: 2.14 × 109–4.90×109]; CTRL, 3.26 × 109 particles/mL [IQR: 1.97 × 109–4.86 × 109]; Figure 1A). The presence of canonical EV markers was confirmed by Western Blot (WB) analysis, which demonstrated expression of the cytosolic protein Alix and Flotillin-1 and the membrane-associated protein β1-integrin in EV preparations from CTRL and HCM patients (Figures 1B and S1). Tetraspanins CD9 and CD63—typical transmembrane proteins enriched in EVs—were detected through flow cytometry analysis within 5(6)-carboxyfluorescein diacetate N-succinimidyl ester (CFSE)—positive events in all groups, providing additional evidence of vesicle identity (Figure 1C). The endoplasmic reticulum protein calnexin, commonly used as a marker of cellular contamination, was present only in THP-1 cell lysate (Figure 1D). Furthermore, albumin was detected only in the plasma sample. Transmission electron microscopy (TEM) analysis confirmed the structural integrity of the isolates. Representative micrographs from CTRL and HCM samples both showed rounded vesicular structures with defined phospholipid bilayers and sizes consistent with EV shape and size (Figure 1E). A lower magnification is provided in Figure S2. Phenotypic characterization of EVs (A) Quantification of EV size distribution and concentration by nanoparticle tracking analysis (NTA). Data are expressed as median and interquartile range (25th and 75th). CTRL represents healthy volunteers, and HCM refers to patients. Mann-Whitney U test. (B) Protein expression of β1-integrin, Alix, and Flotillin-1 in EVs assessed by Western Blotting (WB). (C) Flow cytometric analysis of the tetraspanins CD63 and CD9 in EVs. Representative dot plots from three independent experiments are shown. CFSE-positive events—operationally defined as membrane-enclosed EV-like events—were selected and displayed in R2. Among the CFSE-positive events (R2), R3 gates define EV subpopulations positive for CD63 or CD9 antibody staining. SSC-A, side scatter area; FITC-A, fluorescein isothiocyanate-area; APC-A, allophycocyanin-area. (D) Protein expression (by WB) of calnexin in a cell lysate sample of monocyte THP-1 cell, in a sample of EVs, and in a plasma sample. The band relative to albumin was detectable only in the plasma sample. (E) Representative transmission electron microscopy (TEM) images (magnification = 64,000×) showing EV ultrastructure. Scale bars, 100 nm. Assessment of relative cellular origin of plasma-derived EVs To ensure specificity for EVs, flow cytometry analysis was performed exclusively within CFSE-positive events (CFSE+), a gating strategy adopted to exclude non-vesicular particles. Triton X-100 was used as a negative control for CFSE positivity (i.e., a lysis control expected to eliminate CFSE signal), and we provide the corresponding data in Figure S3, showing that CFSE signal decreases to background following treatment with Triton. Within the CFSE+ EV population, the relative abundance of EV subpopulations was assessed based on antibody positivity and expressed as percentage of positive events (Figure 2). Compared with CTRL samples, HCM-derived EVs exhibited a reduction in neutrophil-derived EVs (CD66b+) from a median of 2.88% to a median of 1.55% (in HCM) (p = 0.045). A reduction in lymphatic-endothelial-cell-derived EVs (CD310+) was also observed in HCM samples (median CTRL: 2.13% vs. HCM: 1.21%; p = 0.020). Conversely, platelet-derived EVs (CD41a+) were significantly enriched in HCM (i.e., 23.7% [median value in CTRL] vs. 27.4% [median value in HCM]; p = 0.039). In contrast, no statistically significant differences were detected in the relative abundance of monocyte-derived EVs (CD14+; p = 0.854), vascular endothelial-derived EVs (CD202b+; p = 0.510), CD8+ T-lymphocyte-derived EVs (p = 0.139), or CD4+ T-lymphocyte-derived EVs (p = 0.147). Flow cytometric analysis of CFSE+ EVs Analysis revealed differences in the relative abundance of specific EV subpopulations between control and HCM samples. Platelet-derived EVs (CD41a+) were significantly increased in HCM, while neutrophil-derived (CD66b+) and lymphatic-endothelial-cell-derived EVs (CD310+) were reduced. No statistically significant differences were observed for EVs expressing CD14 (monocyte), CD202b (vascular endothelial cells), CD4 (CD4+ T lymphocyte), or CD8 (CD8+ T lymphocyte). Violin graphs represent median and interquartile range (25th and 75th). White and red violins represent healthy volunteers and HCM patients, respectively. Mann-Whitney U test. Statistical significance is indicated as follows: ∗p ≤ 0.05. Proteomic analysis of plasma and plasma-derived EVs To avoid the complications implicated by lipoprotein contamination, known to possess similar physical characteristics to those of EVs, size exclusion chromatography (SEC) was employed for isolation of EVs destined to undergo proteomic analyses. WB analysis for ApoA1 and ApoB across six fractions confirmed barely detectable lipoprotein contamination in fractions 1–3, whereas fractions 4–6 showed progressive HDL and LDL carryover. Accordingly, only fractions 1–3 were used for subsequent analyses (Figure S4). The presence of tetraspanins CD9, CD63, and CD81 was confirmed by fluorescence-activated cell sorting (FACS) analysis (Figure S5). Although SEC-derived EVs were smaller [HCM: 113 nm (IQR 103–128); CTRL: 120 nm (IQR 103.5–146)], no significant size differences were observed between CTRL- and HCM-derived EVs when isolation was performed by SEC (p = 0.62). Targeted Olink proteomics was performed on EV and plasma samples from 26 HCM patients and 18 matched CTRL (clinical characteristics are reported in Table S2; the full list of proteins and protein values are reported in Table S3). When analysis was run on the whole plasma, NT-proBNP was significantly increased in HCM patients compared to CTRL (+91%, p < 0.0001), accompanied by significant increases in tissue plasminogen activator (tPA; +16.2%, p = 0.015), growth differentiation factor 15 (GDF15; +10.9%, p = 0.009), CD163 (+5.1%, p = 0.043), von Willebrand factor (vWF; +10%, p = 0.007), surfactant protein D (SP-D; 16.8%, p = 0.038), and galectin-3 (+65%, p = 0.023) (Figure 3A; Table S4). Functional enrichment clustered these proteins within hemostasis and clot dissolution networks, underscoring the possible prominence of coagulation- and inflammation-related processes in HCM (Figure 3B). NT-proBNP was excluded from the subsequent functional enrichment analyses, as it represents a well-established rule-in and rule-out biomarker for heart failure, and its involvement in cardiac pathology is already extensively validated. Excluding NT-proBNP allowed the enrichment analysis to focus on less-characterized proteins, thereby highlighting novel or underappreciated biological pathways and preserving interpretative weight for newly emerging associations. Plasma and EV protein profiles and associated disease pathways in HCM (A) Plasma protein levels in patients with HCM (N = 26) and CTRL (N = 18). Protein expression is reported as normalized protein expression (NPX) values obtained using the Olink proximity extension assay (PEA), Cardiovascular Panel III. NPX values are log2-scaled relative units. Proteins shown include N-terminal pro-B-type natriuretic peptide (NT-proBNP), von Willebrand factor (vWF), tissue plasminogen activator (tPA), surfactant protein D (SP-D), cluster of differentiation 163 (CD163), growth differentiation factor 15 (GDF15), and galectin-3 (Gal-3). Galectin-3 levels were measured separately by enzyme-linked immunosorbent assay (ELISA). Each dot represents an individual subject; bars represent median and interquartile range (25th and 75th). Mann-Whitney U test. Statistical significance is indicated as follows: ∗p ≤ 0.05; ∗∗p ≤ 0.01; ∗∗∗∗p ≤ 0.0001. (B) Disease-gene association enrichment analysis of significantly altered plasma proteins shown in (A). Enrichment analysis was performed using STRING (Search Tool for the Retrieval of Interacting Genes/Proteins), with false discovery rate (FDR) correction applied automatically by the platform. Only proteins significantly increased in HCM were used as input. The bubble plot displays enriched disease terms ranked by signal strength, with dot size corresponding to gene count and color indicating FDR-adjusted significance. The network visualization illustrates functional associations among enriched plasma proteins. (C) Protein levels of plasminogen activator inhibitor-1 (PAI-1) and P-selectin measured in EVs isolated from plasma of CTRL (n = 18) and HCM (n = 26) subjects. EVs were isolated from plasma by SEC. Protein levels are reported as NPX values derived from Olink PEA. EV samples were adjusted to equal total protein concentration before Olink analysis. Data are shown as individual values, with bars representing median and interquartile range (25th and 75th). Mann-Whitney U test. Statistical significance is indicated as follows: ∗∗p ≤ 0.01. (D) Disease-gene association enrichment analysis of plasma and EV-associated proteins. Analysis was performed using STRING, with FDR correction applied. Only significantly changed proteins were included as input. Bubble plot and network representations follow the same conventions as in (B), highlighting disease pathways associated with EV protein alterations in HCM. (E) Relative abundance of selected proteins identified in plasma-derived EVs by untargeted proteomics analysis, including aminopeptidase N (ANPEP), alpha-1-acid glycoprotein 1 (ORM1), and serum amyloid A-4 protein (SAA4), in CTRL (N = 10) and CTRL (N = 10) subjects. Values represent relative protein abundance derived from proteomics data. Individual data points are shown with bars indicating median and interquartile range (25th and 75th). Mann-Whitney U test. Statistical significance is indicated as follows: ∗p ≤ 0.05; ∗∗p ≤ 0.01; ∗∗∗p ≤ 0.001. When the analysis was performed on EV-associated proteins, plasminogen activator inhibitor-1 (PAI-1) and P-selectin (CD62P) were significantly elevated in HCM compared to CTRL, respectively by 56% and 76% (both p < 0.01) (Figure 3C; Table S5). The subsequent disease-gene association enrichment analysis pointed to thrombosis-related processes (Figure 3C). A combined enrichment analysis of all proteins significantly upregulated in HCM in both plasma and EVs (vWF, tPA, CD163, GDF15, galectin-3, PAI-1, and CD62P) revealed a predominant association with thrombosis-related biological processes. Furthermore, disease-gene association analysis also identified carotid artery disease as a significantly enriched condition (Figure 3D). With the aim of pursuing a more exploratory analysis, untargeted proteomic profiling was performed on HCM-derived EVs. Patient selection was based on the cohort previously identified for targeted proteomic analysis. Among these, 10 genetically confirmed HCM patients were selected in light of the well-established association between genetic positivity and poorer clinical outcomes, and 10 age- and sex-matched healthy CTRLs from the same cohort were included in the analysis (clinical characteristics are reported in Table S6). Out of the proteins identified, three were significantly upregulated in HCM EVs (the full list of proteins is reported in Table S7). Specifically, these proteins consisted of aminopeptidase N, alpha-1-acid glycoprotein 1, and serum amyloid A-4 (Figure 3E). Figure 4 and Tables 3 and S8 show the cross-validated area under the curve (AUC) values obtained at each iteration of the greedy forward feature selection approach. The use of cross-validation at each iteration in this type of stepwise regression drives the selection of the optimal features through the predictive performance on data not used during model fitting, thus mitigating overfitting, especially on small populations, with the drawback of obtaining performance estimates with higher variance. Different from other stepwise regression approaches, where a statistical test is performed at each iteration involving the entire population to decide whether to add the feature or not, we did not perform such operation here to avoid being extremely conservative in our analysis, especially in presence of higher variance. We however mitigated the risk of selecting unnecessary features by constraining the method to provide only five proteins. The five selected plasma proteins, from the first to the fifth, were NT-proBNP, Q9NQ76 (matrix extracellular phosphoglycoprotein), O00175 (C-C motif chemokine 24), P05121 (plasminogen activator inhibitor 1), and P04275 (von Willebrand factor). The highest AUC was obtained when the first three proteins were considered together (NT-proBNP, Q9NQ76, and O00175), achieving an AUC of 0.95 (0.87, 1.00). For EV proteins, the selected ones were P16581 (E-selectin), P27930 (interleukin [IL]-1 receptor type 2), Q13867 (bleomycin hydrolase), P00533 (epidermal growth factor receptor), and Q9HCN6 (platelet glycoprotein VI). Also, for EVs, the highest performance was obtained when the first three proteins were considered (P16581, P27930, and Q13867), obtaining an AUC of 0.90 (0.79, 0.98). No statistically significant difference was found between the AUC of plasma and EV (p > 0.10). When considering plasma without NT-proBNP, the optimal proteins were, in order, P42574 (caspase-3), P04085 (platelet-derived growth factor subunit A), Q13867 (bleomycin hydrolase), P05107 (integrin beta-2), and P15086 (carboxypeptidase B). The highest AUC was obtained when the first four proteins were considered (caspase-3, platelet-derived growth factor subunit A, bleomycin hydrolase, and integrin beta-2), with an AUC of 0.82 (0.68, 0.93). Comparing the AUC of EVs with that of plasma without NT-proBNP, respectively considering three proteins each, achieved no statistical difference, even though the p value was borderline (p = 0.11). When comparing the AUC of the three optimal EV proteins with the four ones of plasma without NT-proBNP, we obtained no significant difference (p > 0.10). Greedy-forward-feature-selection-based classification performance of plasma and EV protein panels Cross-validated receiver operating characteristic (ROC) area under the curve (AUC) values obtained at each iteration of a greedy forward feature selection approach, using an increasing number of features (1–5). Classification performance is shown for plasma-derived protein panels (green line), plasma-derived panels excluding NT-proBNP (orange line), and EV-derived protein panels (red line). Data points represent mean AUC values, and error bars indicate variability across cross-validation runs. For both plasma and EV datasets, the highest classification performance was achieved using a limited number of features, with comparable AUCs observed between plasma and EV models. Correlations were assessed by Pearson’s correlation coefficients. Paired AUCs were compared by the De Long test. McNemar test was used to compare the paired predicted outcomes of the Logistic Regression model for classifying HCM vs. controls. Hypothesis tests and correlation coefficients were considered statistically significant when p < 0.05. NT-proBNP (N-terminal prohormone of brain natriuretic peptide), Q9NQ76 (matrix extracellular phosphoglycoprotein), O00175 (C-C motif chemokine 24), P05121 (plasminogen activator inhibitor 1), P04275 (von Willebrand factor), P16581 (E-selectin), P27930 (interleukin-1 receptor type 2), Q13867 (bleomycin hydrolase), P00533 (epidermal growth factor receptor), Q9HCN6 (platelet glycoprotein VI) P42574 (caspase-3), P04085 (platelet-derived growth factor subunit A), Q13867 (bleomycin hydrolase), P05107 (integrin beta-2), and P15086 (carboxypeptidase B). Table 3Discriminative performance of EV-based and plasma-based proteomic models for HCMModelNumber of featuresAUC (95% CI)EV proteome10.794 (0.600–0.964)20.869 (0.731–0.978)30.896 (0.787–0.978)40.869 (0.741–0.974)50.864 (0.709–0.977)Plasma proteome10.874 (0.756–0.964)20.959 (0.894–1.000)30.949 (0.873–0.996)40.908 (0.808–0.980)50.910 (0.810–0.975)Plasma proteome (excluding NT-proBNP)10.729 (0.567–0.874)20.688 (0.523–0.841)30.754 (0.601–0.884)40.821 (0.681–0.932)50.784 (0.631–0.913)Table 3 shows the cross-validated area under the curve (AUC) values obtained at each iteration of the greedy forward feature selection approach used to build the classification models. Models were constructed using extracellular-vesicle-based proteomic profiles, plasma proteomic profiles, and plasma proteomic profiles excluding NT-proBNP. Model performance was assessed using leave-one-out cross-validation. Discriminative performance of EV-based and plasma-based proteomic models for HCM Table 3 shows the cross-validated area under the curve (AUC) values obtained at each iteration of the greedy forward feature selection approach used to build the classification models. Models were constructed using extracellular-vesicle-based proteomic profiles, plasma proteomic profiles, and plasma proteomic profiles excluding NT-proBNP. Model performance was assessed using leave-one-out cross-validation. When comparing predictions with three proteins each, we obtained (1) no significant difference between EV and plasma (p > 0.10) and (2) a statistically significant difference between EV and plasma without NT-proBNP (p < 0.05). Repeating the McNemar test for EV and plasma without NT-proBNP at their highest AUC (i.e., three proteins vs. four proteins), we obtained a mild statistical difference (p = 0.08). Considering the age gap between HCM and CTRL (HCM; median age: 61 years [IQR: 50–69.5] and CTRL; median age: 48 years [IQR: 44–56]), we performed a correlation analysis between protein content and age, which could potentially explain the difference between the two populations found by the model. Among the EV proteins, none was found statistically significantly correlated with age (p > 0.10). Only P16581 was found mildly correlated with age (−0.27; p = 0.08). When considering all selected plasma proteins, NT-proBNP, P54760 (ephrin type-B receptor 4), and P15090 (fatty-acid-binding protein, adipocyte) were found moderately correlated with Pearson’s correlation coefficients of 0.40, 0.33, and 0.40 (p < 0.05), respectively, and P04275 achieved a correlation coefficient of 0.29 (p = 0.06). Impact of plasma-derived EVs on the expression of inflammatory cytokines in THP-1 monocytes Although not in the remit of the main aim of the present project, to provide biological insight into the molecular mechanisms associated with HCM, we preliminarily explored the impact of plasma-derived EVs on the inflammatory expression in THP-1 cells. As shown in Figure S6, 24-h treatment with SEC-isolated EV derived from HCM patients significantly increased the gene expression of the inflammatory cytokine IL-1α (+23%, p = 0.014) and IL-6 (approximately twofold; p = 0.01), without any change in the anti-inflammatory one, i.e., IL-10. Discussion In this single-center, cross-sectional case-control study, we performed EV subcellular characterization and proteomic profiling to identify unique HCM biomarkers in plasma-derived EVs in 46 patients with HCM and 31 healthy controls. The main findings indicate that, although plasma-derived EVs from HCM patients showed no differences in size or concentration compared with CTRL, they exhibited a shift in cellular origin (e.g., increased platelet-derived EVs) and an internal protein cargo profile consistent with platelet activation and thrombo-inflammatory processes. Targeted EV proteomics showed increased levels of PAI-1 and P-selectin, indicating activation of platelet α-granules and enhanced antifibrinolytic activity in patients with HCM. These findings were consistent with whole plasma proteomic analyses, which revealed a broader hemostatic, inflammatory, and fibrotic signature characterized by increased levels of vWF, tPA, the monocyte/macrophage marker CD163, GDF15, and galectin-3.23 When combining EV-associated and plasma protein content, enrichment analysis revealed thrombosis as the most significantly enriched disease pathway (Figure 3D). However, residual plasma/platelet-associated contamination during EV preparation cannot be completely excluded and remains an important limitation of the present study. The present results are consistent with previous evidence from a cohort of 266 HCM patients and 167 controls, in which plasma proteomic profiling of 1,681 proteins revealed upregulation of thrombo-inflammatory pathways, including complement and coagulation cascades.24 Increased platelet reactivity has also been reported in patients with biopsy-proven cardiomyopathy,25 contributing to pronounced platelet dysfunction and a prothrombotic state.26 Although HCM often remains clinically silent for years, the presence of comorbidities substantially increases cardiovascular risk, particularly with respect to thromboembolic complications.27 Within this framework, coronary microvascular dysfunction cannot be overlooked. As reported by Pelliccia et al., non-invasive assessment of coronary microvascular dysfunction via transthoracic doppler echocardiography, positron emission tomography, and cardiac magnetic resonance should be strongly considered in all patients with an HCM diagnosis, particularly in those with angina or chest discomfort, even in the absence of obstructive epicardial coronary stenoses.28 Given that myocardial ischemia is believed to contribute to some of the most severe complications of HCM, including malignant arrhythmias and SCD, characterization of EV cargo and cellular origin may reveal novel biomarkers capable of identifying patients most likely to benefit from anticoagulant therapy, independent of atrial fibrillation status.29 Notably, as up to 35% of HCM patients with atrial fibrillation remain free of stroke during long-term follow-up, atrial fibrillation alone does not fully account for the thromboembolic burden in HCM. Consistently, platelet activation markers such as sCD40L, P-selectin, and β-TG have been reported to be elevated in HCM, at levels substantially higher than in the general population.30 In line with this, our study demonstrated an increase in platelet-derived EVs in HCM patients, occurring despite unchanged platelet counts, suggesting that platelet-derived EVs may serve as an early sensitive marker of platelet activation. This observation is consistent with evidence that cardiovascular stress, hypoxia, and inflammation increase circulating platelet EVs in humans.31 Coronary microvascular dysfunction leads to a reduction in coronary vasodilator reserve that extends beyond hypertrophied regions and affects the entire LV, indicating a primary microvascular involvement.32 Interestingly, impaired myocardial oxygenation has been detected in carriers of HCM-related actionable genetic variants even before the development of LV hypertrophy, supporting the concept that microcirculation impairment may occur early in the disease course.33 EV subcellular characterization further revealed a reduction in EVs released from neutrophils and lymphatic-endothelial cells. Neutrophils, the most abundant circulating leukocytes, have roles extending beyond antimicrobial defense and contribute to sterile inflammation and disease processes; however, their involvement in cardiac hypertrophy remains poorly understood.34 Previous studies showed that elevated neutrophil-to-lymphocyte ratio in HCM associates with an increased risk of SCD, prolonged QTc interval, and worse contractile function, all indicative of a more severe disease phenotype.35 Consistent with these findings, in our cohort, the percentage of neutrophil-derived EVs was positively correlated with the estimated 5-year HCM Risk-SCD score (Spearman correlation coefficient, r = 0.243), whereas the neutrophil count alone fails to correlate with SCD. To explain the inverse relationship found between neutrophil count and the percentage of derived EVs, it should be noted that EV release is influenced by cellular activation and environmental stress; thus, differences in activation state may substantially alter EV output per cell, potentially decoupling total EV abundance from cell counts.12 A possible explanation is that neutrophil-derived EVs are rapidly internalized by autologous neutrophils, with most membrane and intravesicular components transferred within minutes. This suggests that neutrophils constitutively both release and re-uptake their own EVs, thereby modulating their functional responses36; a second explanation is that EVs released from resting neutrophils exert anti-inflammatory effects, whereas those generated upon neutrophil interaction with opsonized particles promote proinflammatory responses.37 Thus, the quantity of neutrophil-derived EVs may depend on the nature of the stimuli triggering their release. In the context of the inflammatory milieu, the untargeted proteomics analysis identified three proteins significantly increased in EV derived from patients with HCM. Specifically, enrichment analysis indicated that these proteins are linked to the acute-phase response and have been independently associated with heart diseases (e.g., heart failure). Serum amyloid A and α-1-acid glycoprotein are well-established markers of chronic inflammation in heart failure and correlate with disease severity and prognosis, while aminopeptidase N is implicated in inflammatory-cell trafficking and cardiac repair after myocardial injury.38,39 In particular, positive correlations between serum amyloid A levels and echocardiographic measures of cardiac structure, including interventricular septal thickness and posterior wall thickness, have been previously described.40 To gain deeper biological insight into the molecular mechanisms associated with HCM and to further support the translational relevance of our findings, we performed an exploratory analysis of the effects of plasma-derived EVs on the inflammatory profile of THP-1 monocytes. Several studies have indeed reported a state of “chronic low-grade” inflammation in HCM, characterized by elevated levels of inflammatory cytokines, including ILs such as IL-6.41,42 Although preliminary and not linked to a specific cellular subpopulation of EV, our in vitro findings appear promising, as the expression of the inflammatory cytokines IL-1α and IL-6 was upregulated when monocytes were exposed to plasma-derived EVs isolated from HCM patients (Figure S6). The lymphatic system, which complements the blood vasculature and is lined by lymphatic-endothelial cells, undergoes lymphangiogenesis primarily through sprouting of pre-existing lymphatic-endothelial cells. VEGFR3 is essential for this process, and reduced VEGFR3 expression or signaling promotes lymphatic-endothelial-cell apoptosis, vessel destabilization, and functional impairment.43 Given the emerging role of cardiac lymphatics in cardiovascular diseases and in response to myocardial injury, our findings of reduced lymphatic-endothelial-cell-derived EVs (CD310+/VEGFR3+) may reflect altered lymphatic endothelial homeostasis in HCM. In particular, chronically elevated cardiac filling pressures and congestion may impose sustained mechanical and funtional stress on the lymphatic vasculature. Because lymphatic endothelial cells are highly responsive to changes in flow, shear stress, and tissue stretch, such chronic haemodynamic stress could conceivably modify their phenotype, including EV biogenesis and release. Thus, the reduction in circulating CD310+/VEGFR3+ EVs observed in HCM may represent a systemic response of altered lymphatic adaptation to chronic cardiac haemodynamic stress, although the mechanisms linking lymphatic dysfunction to EV release remain to be established.44 Given that, in the field of cardiovascular diseases, proteins predictive of adverse events have been identified through proteomics analyses in patients with various cardiomyopathies (e.g., dilated), but not in those with an HCM phenotype, we sought to assess whether EV-protein cargo could discriminate between healthy individuals and patients with HCM. In this context, our findings support this possibility, identifying an EV protein signature comprising E-selectin, IL-1 receptor type 2, and bleomycin hydrolase, which are involved in inflammation and fibrotic processes.45,46 In previous studies, bleomycin hydrolase was found to be associated with NT-proBNP (which reflects increased left ventricular wall stress)47 in dilated cardiomyopathies.48 Importantly, the EV-based proteomic signature was not significantly influenced by age, in contrast to plasma-derived biomarkers such as NT-proBNP, whose discriminative performance was partially driven by age-related effects.49 This observation supports the robustness of EV-associated proteins as disease-specific markers and suggests that EV cargo may capture HCM-related pathophysiological processes independently of chronological aging. It is worth noting that the feature selection and hypothesis testing were not performed using a nested cross-validation framework. Specifically, after feature sets were identified using leave-one-out (LOO) approach, hypothesis testing was performed on predictions computed in LOO with the selected feature sets. This procedure may lead to overoptimistic performance. Despite this can be generally true, considered the small sample size of our study, we were interested in the relative performance between different proteome modalities. Therefore, the models were all fitted using the same workflow and then compared with paired tests under the same conditions. In addition, considering the complexity of the problem, the use of logistic regression for classification purposes is likely underfitting the data, with only ≤5 (feature) + 1 (intercept) parameters, hence obtaining a lower performance than what can actually be achieved. Non-linear models are here expected to provide higher performance. With respect to the proteomic analyses, untargeted mass spectrometry and targeted proteomic analysis (via Olink PEA) should be considered complementary rather than overlapping, as they interrogate distinct layers of the plasma proteome. Mass spectrometry predominantly captures medium-to-high abundance proteins and peptide/proteoform-level information, whereas Olink PEA enables high-sensitivity quantification of predefined low-abundance analytes. Accordingly, only partial overlap between the two datasets is expected and should not be interpreted as discordance.50 In conclusion, despite the relatively small cohort size and heterogeneity of the HCM population, including a substantial proportion of patients without an actionable genetic variant, the combined subcellular and proteomic characterization of plasma-derived EVs identifies a distinct molecular pattern in HCM, predominantly associated with platelet activation and thrombo-inflammation. Compared to classical biomarkers (e.g., NT-proBNP), EVs may provide a more pathobiologically informative compartment, which is particularly relevant in disorders such as HCM where distinguishing disease-specific remodeling from generic hemodynamic stress is crucial. These findings open new avenues for the investigation of EV-mediated mechanisms and biomarker development in HCM. However, we do not propose EV profiling as an immediate replacement for conventional plasma biomarkers in routine care; rather, we view EV analysis as a complementary strategy, particularly valuable in the discovery phase and in settings where a deeper understanding of disease-specific biology is needed. Limitations of the study First and most importantly, these findings should nevertheless be interpreted in the context of the known limitations of EV isolation from platelet-poor plasma, since highly abundant circulating proteins and platelet-associated material may still contribute, at least in part, to the observed proteomic signatures. In support of a limited carryover, the untargeted proteomic analysis performed on the same SEC-isolated EV samples did not detect coagulation proteins such as fibrinogen chains (Table S7). Moreover, in an additional exploratory untargeted proteomic analysis of ultracentrifuged EV pellets, the combined concentration of fibrinogen α-, β-, and γ-chains corresponded to approximately 0.05%–0.13% of physiological plasma fibrinogen levels, while albumin corresponded to approximately 0.005%–0.008% of physiological plasma concentrations. Secondly, the single-center design and the relatively small sample size limit the generalizability of our findings and precluded stratified analyses by clinical phenotype, genotype, or disease severity, thereby warranting validation in larger, multicenter cohorts. However, our cohort closely reflects the real-world HCM population encountered in daily clinical practice, in which a purely Mendelian etiology cannot be identified in all cases.51 Furthermore, the absence of a dedicated positive control group consisting of patients with cardiac hypertrophy secondary to arterial hypertension limits the ability to fully disentangle HCM-specific EV signatures from those associated with pressure-overload-induced hypertrophy. In addition, regarding the choice of patients in whom untargeted proteomic analysis was performed (10 HCM vs. 10 CTRL), the intent was to focus on a more “pure” monogenic, Mendelian form of disease, thereby minimizing the influence of environmental factors and potential confounders. Moreover, with regard to the antibody panel used to identify EVs from specific cellular subpopulations, additional markers could also have been included—for example, monocyte subsets (e.g., CD14 expression within the CD16+ population) and endothelial cell markers such as VEGFR2. Finally, as the analyses were performed in plasma, it is not possible to causally attribute the observed changes in EV cellular origin or EV protein cargo exclusively to the cardiac disease process, as EV profiles may also be influenced by systemic, non-cardiac factors, warranting cautious interpretation of their association with hypertrophic cardiomyopathy. Resource availability Lead contact Further information and reasonable requests for resources and reagents should be directed to the lead contact, massimiliano.ruscica@unimi.it. Materials availability This study did not generate any unique materials. Data and code availability •The mass spectrometry proteomics data generated in this study have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository and are publicly available with the dataset identifier PXD080896 at the following webpage: https://www.ebi.ac.uk/pride/archive/projects/PXD080896. The mass spectrometry proteomics data generated in this study have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository and are publicly available with the dataset identifier PXD080896 at the following webpage: https://www.ebi.ac.uk/pride/archive/projects/PXD080896. •All other source data and analysis code have been deposited in the Universita degli Studi di Milano (UNIMI) repository and are publicly available at https://doi.org/10.13130/RD_UNIMI/HKXICG. All other source data and analysis code have been deposited in the Universita degli Studi di Milano (UNIMI) repository and are publicly available at https://doi.org/10.13130/RD_UNIMI/HKXICG. •Because this study includes human participants with clinical, echocardiographic, and genetic information, individual-level data that could compromise participant privacy are not publicly released. De-identified individual-level data may be made available by the lead contact upon reasonable request, subject to approval by the competent ethics committee of Fondazione IRCCS Ca Granda Ospedale Maggiore Policlinico and in accordance with applicable data protection regulations and the informed consent obtained from participants. Because this study includes human participants with clinical, echocardiographic, and genetic information, individual-level data that could compromise participant privacy are not publicly released. De-identified individual-level data may be made available by the lead contact upon reasonable request, subject to approval by the competent ethics committee of Fondazione IRCCS Ca Granda Ospedale Maggiore Policlinico and in accordance with applicable data protection regulations and the informed consent obtained from participants. •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request Acknowledgments The work of the authors was partially supported by Ministero dell’Istruzione, dell’Università e della Ricerca (progetto PNC0000003 – Anthem – AdvaNced Technologies for Human-centrEd Medicine) to M.R.; HORIZON EUROPE Marie Sklodowska-Curie Actions (101167421 – UNION) to M.R.; Fondo Beneficenza di Banca Intesa San Paolo (B/2025/0265) to M.R.; Ministero della Salute, Ricerca Finalizzata 2024 (RF-2024-12378718; ELUCIDATE) to M.R.; and Ministero della Salute, Ricerca Corrente, 2026 (Foundation IRCCS Ca' Granda Ospedale Maggiore Policlinico) to S.C. Author contributions M.R. conceived and supervised the study. A.S.R., A.F., and C.M. performed most experiments, analyzed data, and drafted the manuscript. I.F. and D.B. performed extracellular vesicle characterization and in vitro experiments. A.M. and C.B. performed and interpreted the proteomic analyses. A.F., E.G., F.T., A.P., M.B., R.M., and M.C. recruited participants and performed echocardiographic evaluations. S.P., P.C., and P.F. performed and interpreted the genetic analyses. I.G. and V.D. performed electron microscopy analyses. R.S. and M.W.R. developed and performed the logistic regression analyses used to combine the predictive effects of multiple proteins. M.V., A.C., S.C., and M.R. provided critical intellectual input and revised the manuscript. All authors reviewed and approved the final manuscript. Declaration of interests The authors declare no competing interests. STAR★Methods Key resources table REAGENT or RESOURCESOURCEIDENTIFIERAntibodiesβ1-integrinSanta Cruz BiotechnologyCat: sc-374429AlixCell Signaling TechnologyCat: 2171Flotillin 1ProteintechCat: 15571-1-APCalnexinAbcamCat: ab275018Apolipoprotein BMilliporeCat: AB742Apolipoprotein A1MilliporeCat: 178422-1 MLHuman anti-CD14Miltenyi BiotecCat: 130-110-520Human anti-CD202bMiltenyi BiotecCat: 130-109-399Human anti-CD8Miltenyi BiotecCat: 130-110-679Human anti-CD4Miltenyi BiotecCat: 130-114-534Human anti-CD66bMiltenyi BiotecCat: 130-117-692Human anti-CD41aMiltenyi BiotecCat: 130-128-830Human anti-CD310Miltenyi BiotecCat: 130-125-977Human anti-CD9Miltenyi BiotecCat: 130-118-808Human anti-CD63Miltenyi BiotecCat: 130-118-078Human anti-CD81Miltenyi BiotecCat: 130-119-787Peroxidase-AffiniPure Goat Anti-RabbitJacksonImmunoResearchJI111035003Biological samplesHuman plasma samplesThis studyEthical Committee - Approval No. 802_2021Chemicals, peptides, and recombinant proteinsCELLTRACE carboxyfluorescein succinimidyl ester (CFSE)Thermo FisherCat: C34554RPMI 1640 MediumThermo FisherCat: 11875093GlutaraldehydeMerckCat: G5882phosphotungstic acidMerckCat: P4006Triton™ X-100MerckCat: X-100RIPA, 10XMerckCat: 20-188Fetal Bovine SerumMerckCat: F7524Penicillin-StreptomycinMerckCat: P0781Clarity Western ECL chemiluminescent substratesBio-Rad LaboratoriesCat: 1705060SBSAMerckCat: A9418-50GTris-Buffered SalineBio-RadCat: 1706435Tween 20Molecular mass marker - Novex Sharp Protein StandardInvitrogenCat: LC5800cOmplete™, Mini, EDTA-freeMerckCat: 11836170001PhosSTOP™MerckCat: 049068450011,4-DithiothreitolMerckCat: 1.11474Trifluoroacetic acid solutionMerckCat: 4.80112Ammonium bicarbonateMerckCat: A6141Critical commercial assaysMaxima SYBR Green/Fluorescein qPCR Master Mix (2X)Thermo fisherCat: K0242Maxima First Strand cDNA Synthesis Kit for RT-qPCRThermo fisherCat: K1642RNeasy Mini Kit (50)QiagenCat: 74104Human Galectin-3 ELISA KitBio-TechneCat: DGAL30Olink® Target 96 Cardiovascular IIIOlink ProteomicsCat: 95611Deposited dataMass spectrometry dataThis paperData are available via ProteomeXchange with identifier Project accession: PXD080896; https://www.ebi.ac.uk/pride/archive/projects/PXD080896Source data and analysis codeThis paperDataverse repository: https://doi.org/10.13130/RD_UNIMI/HKXICGSTRING databasehttps://string-db.org/N/AKEGG PATHWAY Databasehttps://www.genome.jp/kegg/pathway.htmlN/AReactomehttps://reactome.org/11N/AExperimental models: Cell linesTHP-1 human monocyte. This cell line was not tested for mycoplasma contamination in this study.ATCCCat: TIB-202Experimental models: Study participantsHealthy controls (CTRL); sample size = 31.This paperEthical Committee - Approval No. 802_2021Patients with Hypertrophic cardiomyopathy (HCM); sample size = 46.This paperEthical Committee - Approval No. 802_2021OligonucleotidesInterleukin 1αFw: GTAGCAACCAACGGGAAGGTRev: AAGGTGCTGACCTAGGCTTGMerckN/AInterleukin 6Fw: ACCCCCAGGAGAAGATTCCARev: GGTTGTTTTCTGCCAGTGCCMerckN/AInterleukin 10Fw: CCTGCCTAACATGCTTCGAGARev: AAGAAATCGATGACAGCGCCMerckN/ASoftware and algorithmsNovoexpressAgilenthttps://www.agilent.com/en/product/research-flow-cytometry/flow-cytometry-software/novocyte-novoexpress-software-1320805NTA software version 3.4Malvern Panalyticalhttps://www.malvernpanalytical.com/en/products/product-range/nanosight-range/nanosight-pro#manualsandsoftwareGraphPad Prism v10GraphPadhttps://www.graphpad.com/scientific-software/prismOtherqEVoriginal 70 nm Gen 2 columnsIZONCat: ICO-70 Experimental model and study participant details Study population Patients diagnosed with HCM were consecutively enrolled (from December 2022 to October 2025) at the Cardiomyopathy Unit of Fondazione IRCCS Ca’ Granda Ospedale Maggiore Policlinico, Milan, Italy. Diagnosis was established according to the 2023 European Society of Cardiology (ESC) guidelines, which define HCM as an LV maximal wall thickness (MWT) ≥15 mm unexplained by loading conditions, or 13–14 mm with supportive evidence from family history, genotype, or electrocardiographic abnormalities.47 In HCM patients with an apical phenotype, recently refined diagnostic criteria were applied, defining apical HCM as an apical MWT >11 mm or >5.6 mm/m2 when indexed to body surface area.52 Exclusion criteria included age <18 or >80 years, pregnancy, and genotype-positive/phenotype-negative relatives. Healthy volunteers (from now on CTRL) with normal ECG and no personal or family history of cardiovascular disease served as controls. The study population included 46 patients with HCM, with a median age of 61 years (IQR: 50.0–69.5) and 32 males (69.5%), and 31 CTRL subjects, with a median age of 48 years (IQR: 44.0–56.0) and 16 males (51.6%). Age and sex distribution are reported in detail in Table 1. All participants underwent clinical evaluation, echocardiography, and venous blood collection on the same day. In HCM patients with an appropriate clinical indication and no contraindications, CMR was also performed. Demographic, genetic, echocardiographic, CMR and laboratory data were recorded in a dedicated database. To characterise each HCM patient’s intrinsic risk of life-threatening ventricular arrhythmias, we calculated the HCM-SCD risk score for all patients (incorporating age, maximal LV wall thickness, left atrial diameter, LV outflow tract gradient, family history of SCD, non-sustained ventricular tachycardia, and unexplained syncope). This tool estimates the 5-year risk of sudden cardiac death and classifies patients into high (>6%), intermediate (4–6%), and low (<4%) risk categories, in accordance with contemporary ESC and American guidelines.5,47 Ethical statement The study protocol conformed to the Declaration of Helsinki and received approval from the Institutional Review Board of Fondazione IRCCS Ca’ Granda Ospedale Maggiore Policlinico (Approval No. 802_2021). Written informed consent was obtained from all participants in accordance with the principles outlined in the Declaration of Helsinki. Echocardiographic and cardiac magnetic assessment Transthoracic echocardiography was performed using EPIC CVxi (Philips, The Netherlands) or VIVID E95 (GE Healthcare, USA) systems. Standard parameters included LV wall thickness, MWT, chamber volumes, ejection fraction (biplane Simpson’s method), LV mass (Devereux formula), biatrial size, and diastolic indices (E/A, E/e′ by tissue Doppler). The presence of LVOTO was defined as a peak instantaneous gradient ≥30 mmHg at rest or during provocation, in accordance with ESC guidelines.47 CMR studies were performed on Avanto (1.5T, Siemens Healthineers, Erlangen, Germany) scanners, with electrocardiogram gating at breath-hold periods. Conventional cine imaging sequences (SSFP sequences; matrix:256x256, FOV:300-370 mm; temporal resolution 50 ms) were acquired over multiple heartbeats and included: LV long-axis two-, three- and four-chamber views, as well as a stack of short-axis views. Late enhancement sequences (T1-weighted, inversion-recovery, fast gradient-echo pulse sequence) were acquired at identical slice positions as the cine images 10–15 min after intravenous administration of 0.2 mmol/kg of dimegluminegadopentetate Magnevist (matrix:256x256, FOV:300-370 mm; and optimal inversion time to suppress the myocardium signal). As previously described, short-axis cine images were evaluated to measure biventricular volumes and ejection fraction. The presence and pattern of distribution of late gadolinium enhancement (LGE) was visually assessed. Genetic analysis Genetic testing for HCM was conducted in accordance with local clinical and patient preference criteria. Next-generation sequencing (NGS) was performed using a custom Haloplex Target Enrichment kit (Agilent Technologies, USA) encompassing 54 genes associated with HCM. Library preparation, hybridization, capture, and sequencing were performed following the manufacturer’s protocol on an Illumina MiSeq platform. Variant calling and annotation were carried out using Agilent SureCall and wAnnovar software. Variant classification followed the ACMG recommendations, which integrate multiple lines of evidence—including population frequency, in silico predictions, segregation data, functional studies, and previously published reports—to categorise variants as pathogenic (class V) likely pathogenic (class IV), of uncertain significance (class III, VUS), likely benign (class II), or benign (class I).53 Variants classified as pathogenic or likely pathogenic in established HCM-associated genes were considered causative and clinically actionable. Patients harboring only VUS, likely benign/benign variants, or with no identifiable variant were classified as “genetically negative” or gene-elusive.54 Clinically relevant variants and regions of low coverage were validated by Sanger sequencing using the BigDye Terminator kit (Applied Biosystems, Thermo Fisher Scientific, USA). Method details Blood collection and EV isolation Venous blood was collected after overnight fasting into EDTA tubes and processed within 2 h. Platelet-poor plasma (PPP) was obtained by sequential centrifugation at 2000 × g (20 min, RT), followed by 1000-3000xg (15 min each, 4°C). Plasma aliquots of 500 μL were supplemented with sucrose (1M) to preserve vesicle integrity55 and stored at −80°C. For flow cytometry analyses, EVs were isolated by ultracentrifugation at 110,000xg for 90 min at 4°C using an Optima MAX-XP ultracentrifuge (Beckman Coulter, CA, USA) and resuspended in 0.10 μm-filtered (Merck, Millipore, MA, USA) phosphate buffered saline (PBS) without calcium chloride and magnesium chloride (PBS buffer PanReac AppliChem, Barcelona, Spain). For proteomics, EVs were isolated by size-exclusion chromatography (SEC) using qEVoriginal 70 nm Gen 2 columns (IZON, New Zealand), collecting fractions F1–F3, which were pooled and stored at −80°C until analysis. Nanoparticle tracking analysis (NTA) EV size and concentration were determined using a Nanosight LM10-HS system (NanoSight Ltd., UK). Samples were visualized by laser light scattering and 5 recordings of 30 s were performed for each sample. Data were processed using NTA software to generate particle size distribution profiles and concentrations.56 Transmission electron microscopy (TEM) EV morphology was assessed by TEM. Briefly, EVs were resuspended in PBS and adsorbed onto 300 mesh carbon-coated copper grids (Electron Microscopy Sciences, Hatfield, PA, USA) (5 min, room temperature, in a humidified chamber), fixed in 2% glutaraldehyde and stained with 2% phosphotungstic acid (pH 7.0). Grids were examined with a Philips CM100 TEM, and images were captured by a Kodak digital camera.57 Flow cytometry EVs were labeled with 0.2 μM CFSE (Thermo Fisher Scientific) for intact vesicle identification and with fluorophore-conjugated monoclonal antibodies (Miltenyi Biotec) against cell-type markers, labeled with Allophycocyanin (APC) or Vio Dyes (VioBright V423 and VioBlue). Acquisition was performed on a BD LSRFortessa flow cytometer at a low flow rate (12.5 μL/min) following the MIFlowCyt-EV framework.58 Data were analyzed using NovoExpress software (Agilent), and EVs were identified based on CFSE positivity. Marker expression was assessed by antibody staining, and EV subpopulations were quantified as the percentage of antibody-positive events within the CFSE+ EV population. This approach was used to assess the relative abundance of EVs expressing specific surface markers. Additional controls to correct for FITC-, APC-, Vio Bright V423- and VioBlue- fluorescence were also performed (unstained and single-stained controls), as well as serial dilutions to ensure proper event detection and swarming prevention. To assess the autofluorescence of each antibody, 250 μL of triple 0.10 μm pore size membrane-filtered PBS (control sample) were stained with CFSE, and antibodies. Sole analysis of EV suspension was performed to determine auto fluorescence of EVs. Triton X-100, used as a negative control for CFSE positivity, was used at a final concentration of 1% (vol/vol) for 30 min at room temperature.59 Western blot (WB) EV lysates were prepared in RIPA buffer with protease and phosphatase inhibitors. Equal protein amounts (20 μg) were separated by SDS-PAGE (4–12%) and transferred to nitrocellulose membranes. Human monocyte THP-1 cell lysate was used as a positive control for the protein expression of calnexin. Blots were probed with antibodies against Alix (1:1000), β1-Integrin (1:500) and Flotillin-1 (1:2000), followed by HRP-conjugated secondary antibodies (anti-rabbit, 1:10,000). Detection was performed with Clarity ECL (Bio-Rad Laboratories) using a ChemiDoc XRS System (Bio-Rad Laboratories). Targeted proteomics (Olink PEA) Proteomic profiling of 92 cardiovascular-related proteins was performed using the Olink CARDIOVASCULAR III panel (Olink Proteomics, Uppsala, Sweden) on both plasma and EV samples. Two μL of plasma and EV sample, adjusted to 0.5 mg/mL, were used for each assay. Data were expressed as normalized protein expression (NPX) values on a log2 scale.60 Untargeted proteomics EV samples isolated by SEC and resuspended in PBS were dissolved in 25 mmol/L ammonium bicarbonate (NH4HCO3) containing 0.1% RapiGest SF surfactant. Samples underwent sonication followed by centrifugation at 13,000×g for 10 min. A total of 50 μg of protein per sample was denatured at 80°C for 15 min, reduced with 5 mmol/L dithiothreitol (DTT) at 60°C for 15 min, and alkylated with 10 mmol/L iodoacetamide for 30 min at room temperature in the dark. Enzymatic digestion was carried out by overnight incubation at 37°C with 2.5 μg of sequencing-grade trypsin. Following digestion, 2% trifluoroacetic acid (TFA) was added to hydrolyze RapiGest and inactivate the trypsin. Mass spectrometry analysis was conducted using a hybrid time-of-flight instrument (Synapt XS, Waters Corporation) coupled to an M-Class UPLC system equipped with a nanospray ionization source. Peptides were first loaded onto a C18 trap column (Symmetry, 180 μm × 20 mm, 5 μm, 100 Å) and subsequently separated on an HSS T3 C18 analytical column (75 μm × 150 mm, 1.7 μm, 100 Å) using a linear gradient of organic solvent from 3% to 40% over 90 min. Label-free Liquid Chromatography – Mass Spectrometry with Elevated Energy (LC-MSE) acquisition was performed in triplicate using an ion mobility-enhanced data-independent acquisition (IMS-DIA) approach. Untargeted LC-MSE data were processed in Progenesis QI for Proteomics (v4.0, Waters) for peptide quantification and protein identification using an UniProt human protein sequence database (v 2025). Precursor and fragment mass tolerances were set as 5 ppm and 10 ppm, respectively. The peptide search parameters were set as follows: 1 missed cleavage allowed, carbamidomethylation (C) as a fixed modification, and oxidation (M) as a variable modification. Inclusion criteria required at least 2 unique peptides per protein for protein acceptance. Feature detection/quantification was performed using the software’s alignment and co-detection workflow. No separate missing-value imputation was applied. Data were normalized within Progenesis using the automatic normalization procedure, and differential abundance between groups was assessed on normalized protein abundances using the ANOVA p-value generated by the software. Patient selection was performed starting from the cohort selected for targeted proteomic analysis. Enzyme-linked immunosorbent assay (ELISA) Plasma concentrations of galectin-3 were measured using commercial human galectin-3 ELISA Kit – Quantikine (Bio-techne, USA). According to the manufacturer’s instructions, plasma samples were diluted 1:2. The detection limit was 0.016 ng/mL. Functional enrichment analysis Proteins with increased abundance in HCM compared to controls (p < 0.05) were analyzed separately for plasma and EV datasets using STRING v11.5 and g:Profiler. Gene Ontology (GO) and pathway enrichment analyses (KEGG, Reactome) were performed. Enriched terms were considered significant at a Benjamini-Hochberg false discovery rate (FDR) adjusted p < 0.05. Functional study on in vitro monocytes THP-1 monocyte cells were cultured in RPMI (Thermo Fisher Scientific, Milan, Italy) supplemented with 10% FBS and penicillin (10,000 U/mL), streptomycin (10 mg/mL) and β-mercaptoethanol 0.05 mM (Merk, Milan, Italy). Relative to treatments with isolated EVs, all cell lines were serum-starved overnight in RPMI supplemented with 0.1% bovine serum albumin (BSA). After starvation, media were replaced, and cells were treated with 5×108 EV/mL (isolated with SEC) for 24 h. Total RNA was extracted using spin column (Qiagen, Milan, Italy) according to the manufacturer’s instructions. Reverse transcription-polymerase first-strand cDNA synthesis was performed by using the Maxima First Strand cDNA synthesis kit (Thermo Fisher Scientific). Quantitative PCR was then performed by using the Thermo SYBR Green/ROX qPCR Master Mix kit (Thermo Fisher Scientific) and specific primers for selected genes Interleukin-1 (IL) alpha, IL-6 and IL-10. The analyses were performed with the 9600 Bio-Rad Real-Time PCR Detection Systems (Bio-Rad Laboratories, Milan, Italy). PCR cycling conditions were as follows: 95°C for 10 min, 40 cycles at 95°C for 15 s, and 55°C for 30 s. Data were expressed as Ct values and used for the relative quantification of targets with the 2−ΔΔCt calculation. 36B4 was used as the housekeeping gene. Quantification and statistical analysis Continuous variables were expressed as mean ± standard deviation (SD) or median and interquartile range (25th and 75th percentiles), according to distribution. Normality was assessed using the Shapiro–Wilk test and visual inspection of histograms and Q–Q plots. Group differences were assessed using the Mann–Whitney U test, and correlations with Spearman’s coefficient. Categorical variables were reported as counts and percentages and compared using the χ2 test or Fisher’s exact test, as appropriate. Data were analyzed using GraphPad Prism v10 (GraphPad Software, San Diego, USA). Regarding Olink data, we investigated whether the protein content of EVs could discriminate HCM patients from controls. Our goal was to verify whether a suitable set of EV proteins could be as efficient as plasma proteins in detecting HCM patients, and superior to plasma proteins without NT-ProBNP. To do so, we implemented a multivariable analysis (the overall workflow is depicted in the Figure S7). Specifically, we used a Logistic Regression model to combine the predictive effect of multiple proteins. To select a plausible set of optimal proteins for the task, we applied a greedy forward feature selection approach. Briefly, the method selected proteins through an iterative approach. At each iteration, the optimal protein was selected from a list of candidates. The optimal one was added to the set of selected proteins from the previous iterations. The optimal protein at a given iteration was the one maximizing the AUC but not having a Pearson’s correlation coefficient above 0.6 with respect to all previously selected proteins. The AUC was computed using the LOO approach, where all samples but one were used to fit the model and the one kept out was used for evaluation. The predictions on the kept-out samples were used to calculate the AUC. This approach provided a possible optimal set of proteins in cross-validation. The Logistic Regression model was fitted using a weighted binary cross-entropy loss where the weights were meant to compensate the class imbalance. Weights were computed for each LOO iteration. The model included the intercept. The maximum number of proteins allowed in this procedure was set to 5. The approach implemented is recommended by Salzberg.61 Correlations were assessed by Pearson’s correlation coefficients. Paired AUCs were compared by the DeLong test. McNemar's test was used to compare the paired predicted outcomes of the Logistic Regression model for classifying HCM vs. controls.61 Here, cross-validated predictions were obtained when logits>0. Ninety-five percent confidence intervals for AUCs were computed using bootstrapping with 10,000 resamplings. Hypothesis tests and correlation coefficients were considered statistically significant when p < 0.05 (we also specify when p < 0.10 to potentially account for the low sample size in our interpretation). ∗ asterisk stands for p ≤ 0.05; ∗∗ asterisks stand for p ≤ 0.01; ∗∗∗ asterisks stand for p ≤ 0.001; ∗∗∗∗ asterisks stand for p ≤ 0.0001. Footnotes Supplemental information can be found online at https://doi.org/10.1016/j.isci.2026.117301. Supplemental information Document S1. Figures S1–S7 and Tables S1, S2, S4–S6, and S8 Table S3. List of proteins included in the Olink Cardiovascular III panel used for targeted proteomic analysis Table S7. Full list of proteins identified by untargeted proteomic analysis of plasma-derived extracellular vesicles from HCM and control samples References Associated Data Supplementary Materials Document S1. Figures S1–S7 and Tables S1, S2, S4–S6, and S8 Table S3. List of proteins included in the Olink Cardiovascular III panel used for targeted proteomic analysis Table S7. Full list of proteins identified by untargeted proteomic analysis of plasma-derived extracellular vesicles from HCM and control samples Data Availability Statement •The mass spectrometry proteomics data generated in this study have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository and are publicly available with the dataset identifier PXD080896 at the following webpage: https://www.ebi.ac.uk/pride/archive/projects/PXD080896. The mass spectrometry proteomics data generated in this study have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository and are publicly available with the dataset identifier PXD080896 at the following webpage: https://www.ebi.ac.uk/pride/archive/projects/PXD080896. •All other source data and analysis code have been deposited in the Universita degli Studi di Milano (UNIMI) repository and are publicly available at https://doi.org/10.13130/RD_UNIMI/HKXICG. All other source data and analysis code have been deposited in the Universita degli Studi di Milano (UNIMI) repository and are publicly available at https://doi.org/10.13130/RD_UNIMI/HKXICG. •Because this study includes human participants with clinical, echocardiographic, and genetic information, individual-level data that could compromise participant privacy are not publicly released. De-identified individual-level data may be made available by the lead contact upon reasonable request, subject to approval by the competent ethics committee of Fondazione IRCCS Ca Granda Ospedale Maggiore Policlinico and in accordance with applicable data protection regulations and the informed consent obtained from participants. Because this study includes human participants with clinical, echocardiographic, and genetic information, individual-level data that could compromise participant privacy are not publicly released. De-identified individual-level data may be made available by the lead contact upon reasonable request, subject to approval by the competent ethics committee of Fondazione IRCCS Ca Granda Ospedale Maggiore Policlinico and in accordance with applicable data protection regulations and the informed consent obtained from participants. •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request