Stenosis in Mature Arteriovenous Fistulas Characterized by Contractile Smooth Muscle Cells and Versican Authors: Yichi Zhang, Nagla Elzinad, Saran Lotfollahzadeh, Adam Lazowski, Murad Elsadawi, Wenqing Yin, Mostafa Belghasem, Joyita Bharti, Najia Idrees, Chris Zavarro, Tarek Shazly, Aala Jaberi, Jean Francis, Alik Farber, Roberto Vazquez-Padron, Laura M Dember, Jeffrey Siracuse, Vijaya B Kolachalama, Vipul C Chitalia Published: 2026 ABSTRACT AbstractIntroductionStenosis occurs in 20% to 40% of mature arteriovenous fistulas (AVFs). Alterations of the vessel wall composition in stenosis should provide insights into its pathogenesis. However, there is a dearth of studies examining the vessel wall composition of AVF with secondary failure.MethodsTwenty-one AVF explants were used. Cellular (vascular smooth muscle cells [vSMCs]) and noncellular components (collagen and extracellular matrix [ECM]) were quantified using a customized machine-learning approach.ResultsAVFs were explanted for aneurysms (16/21), pseudoaneurysms (2/21), or cosmetic reasons (3/21). Fifteen of the AVFs had a stenotic lesion. Proteoglycans and collagen predominated over vSMCs in the AVF wall and showed disruptions along the needle tracks. The stenotic wall showed compact collagen organization and higher expression of perlecan, versican, and versikine, and a linear correlation between versican and versikine. vSMCs in stenosis showed higher myosin heavy chain 11 (MYH11), Calponin, and smooth muscle actin (contractile and/or quiescent markers) than Ki67, a proliferative marker, without an increase in osteopontin (a secretory marker). Proliferating cells were located > 50 to 100 microns from the subendothelium.ConclusionAdvanced stenosis in mature AVFs are characterized by contractile vSMC phenotype and predominated by extracellular and proteoglycan-rich remodeling raising a possibility of AVF stenotic lesions as a temporal continuum in which vSMCs initially adopt a synthetic and/or proliferative phenotype during stenosis formation and subsequently contractile phenotype as the stenosis stabilizes. Although providing potential insights into the heterogeneous response of paclitaxel-coated devices in AVF stenosis, this study underscores the need for a deeper understanding of the biology of stenosis in mature AVFs and the use of targeted therapies for specific components in AVF stenosis.Keywords: arteriovenous fistula, chronic kidney disease, hemodialysis, inflammation, maturation rate, stenosis An AVF is the preferred type of vascular access for hemodialysis in patients with end-stage kidney disease (ESKD).1 AVFs develop frequent complications; studies have shown primary AVF failure rates of ∼ 20% to 50%.2,3 For AVFs that successfully mature to support dialysis, data from the Dialysis Outcomes and Quality Initiative reported primary patency rates of 85% at 1 year (15% AVF failure) and 75% at 2 years (25% AVF failure).4 Recent studies suggest that stenosis in mature AVFs occurs in approximately 16% to 28% by 1 year and 26% to 36% by 2 years,5, 6, 7, 8, 9 and that a large proportion of mature AVFs require repeated endovascular or surgical reinterventions to maintain usability. Medicare spent $2.8 billion on maintaining functional dialysis access for patients. These studies underscore the high burden of AVF complications in patients with ESKD. Complications in AVFs that render them nonfunctional despite their successful hemodialysis use are classified as secondary AVF failures. Secondary failure can result from various complications, such as aneurysms and pseudoaneurysms, both of which result in part from the stenosis.10AVF stenosis reduces blood flow through the vascular access, compromises solute clearance, increases postcannulation bleeding, depending on the location, can contribute to aneurysm or pseudoaneurysm formation, and can lead to access thrombosis. Although several studies have examined risk factors and the morphology of primary AVF failure,11, 12, 13, 14, 15 there is a dearth of studies examining the morphology of a mature AVF wall and of secondary failure. Deciphering the morphology and composition of mature AVFs and how they are altered across various AVF pathologies is essential for improving our understanding of the mechanisms driving secondary AVF failure. Our objective was to interrogate the remodeling biology underlying stenosis in mature AVFs, given the central role of stenoses in access dysfunction and the potential for pharmacologic targeting. Direct study of stenotic lesions in mature AVFs is challenging because stenosis alone is not an indication for AVF aneurysmectomy, limiting tissue availability. However, stenosis is frequently identified as an underlying pathological substrate in AVFs explanted for aneurysm or pseudoaneurysm, which are recognized indications for surgical removal under predefined clinical criteria.16,17 Accordingly, we analyzed explanted mature AVFs removed for aneurysm, pseudoaneurysm, or cosmetic reasons, to specifically examine coexisting stenotic and nonstenotic regions. These specimens underwent detailed morphological assessment and were processed using a customized machine-learning–based image segmentation pipeline to quantitatively characterize the cellular and noncellular components of the AVF wall across distinct anatomical regions. Methods Study Design and Histological Characterization Twenty-one explanted AVFs from patients with ESKD on hemodialysis from 2016 to 2020 at Boston Medical Center were analyzed following Institutional Review Board approval at Boston University Medical Campus (H-26367). The inclusion criteria were patients with ESKD who underwent AVF explantation for aneurysms, pseudoaneurysms, cosmetic reasons, or persistent pain. Patients’ medical records and baseline characteristics were extracted from an electronic health record system (EPIC Systems, Verona, WI). We leveraged these clinically indicated AVF explants to systematically characterize the coexisting stenotic pathology within mature AVFs. Accordingly, the study population represents patients undergoing standard-of-care AVF aneurysmectomy.16 Stenotic and nonstenotic regions were identified by a board-certified pathologist (Murad Elsadawi) before fixing the biospecimen and were annotated accordingly. A similar number of stenoses and nonstenoses were retrieved from each AVF. Hyperplastic regions were defined by neo-intimal hyperplasia in the nonstenotic AVF wall. Paraffin Embedding and Slide Cutting Tissues were processed by the Department of Pathology at Boston Medical Center. After embedding in paraffin, tissue blocks were cut into 4 to 5-μm-thick sections. Hematoxylin & Eosin Staining Hematoxylin and eosin staining of AVF tissue slides was performed according to the protocol described previously.18,19 Picro Sirius Red and Movat Pentachrome Staining Sirius Red staining was performed using the Picro Sirius Red Stain kit from Abcam (#AB150681, Abcam, Cambridge, United Kingdom), as per the manufacturer’s instructions. Similarly, Pentachrome staining was performed using the Movat Pentachrome Stain (modified Russel-Movat) kit from ScyTek Laboratories (#MPS-2; ScyTek Laboratories, Logan, UT), according to the manufacturer’s instructions. vSMC ImmunoHistochemistry Staining Immunohistochemistry (IHC) staining was performed using the Mouse and Rabbit Specific Horseradish Peroxidase and 3,3'-Diaminobenzidine, Horseradish Peroxidase/3,3'-Diaminobenzidine IHC Detection Kit from Abcam (#AB236466), as per the manufacturer’s instructions. MYH11 (#NBP2-44533; Novus Biologicals, Centennial, CO), Calponin (#1789; Cell Signaling Technology, Danvers, MA), alpha SMA (Novus Biologicals, #NBP2-33006), and Ki67 (Cell Signaling, #17189), Osteopontin Polyclonal antibody (Cat No. 22952-1-AP Proteintech, Rosemont, IL) Concentration: 1:100 antibodies were diluted 1:100 in protein block before overnight incubation at 4 °C. Ki67 is a proliferation marker,20 and Calponin, smooth muscle actin, and MYH 11 are markers of differentiated, secretory vSMCs.21 Proteoglycan IHC Staining Following deparaffinization and antigen retrieval, blocking was performed. Glypican (Novus Biologicals, #NBP1-33197), Aggrecan (Abcam, #ab85174), Syndecan (Abcam, #ab128936), Perlecan (Novus Biologicals, #NBP3-17840), and Versikine (#PA1-1748A ; Invitrogen, Thermo Fisher Scientific, Waltham, MA) antibodies were diluted 1:50 in protein block and incubated overnight at 4°C. Anti-Rabbit secondary antibody incubation, 3,3'-Diaminobenzidine chromogen incubation, and hematoxylin counterstain were performed. Versican IHC Staining Following deparaffinization, slides were treated with chondroitinase ABC (#C3667; Sigma-Aldrich, St. Louis, MO), diluted to 0.2 to 0.5 units/ml in 18 mM Tris-acetate buffer containing 0.1% bovine serum albumin, at 37 °C for 1 hour. Endogenous peroxidase activity was quenched and blocked as above. The primary antibody was rabbit antiversican (Abcam #Ab177480, 1:100), which was applied overnight at 4 °C. Quantitative Histology and Machine Learning Approach A color-based image-processing pipeline (Supplementary Figure S1) was developed to segment and quantify, as percentages, proteoglycans, collagen, and versican from Sirius Red- and Movat Pentachrome-stained AVF cross-sections. Our analytical pipeline uses a classical unsupervised machine learning approach based solely on digital histology images to compute the area fraction of each tissue class within the AVF wall. The pipeline comprises the following steps: basic image preprocessing, including downsampling, background purification, and removal of unwanted tissues; color feature amplification using a combination of filters; color-based segmentation; and valid-area computation for selected classes within each image. The region of interests were precisely segmented; however, in some region of interests, they heavily coexisted with other components. The focus of tissue boundary determination was switched from texture and morphology to color. The skin and fat tissue were not included in our calculations, nor were artifacts or broken tissue fragments. All red, green, and blue color images were first aimed at achieving color constancy using global white-balancing normalization, followed by max-red, green, blue color normalization (Supplementary Figure S1B). This procedure detected colors independently of the illuminance. Outputs were then converted to L∗a∗b color space (Supplementary Figure S1C), and a contrast-limited adaptive histogram equalization was applied to create a distance between colors of similar wavelengths. The k-means clustering algorithm (k = 5) was used on each transformed color image for all stains. A board-certified surgical pathologist identified areas corresponding to the respective clusters of proteoglycan, collagen, and versican (Supplementary Figure S1D). Regions predominantly rich in collagen were quantified by placing a single cluster covering areas with highly compacted fiber tissue (Supplementary Figure S1F), a texture feature that is reflected in density but is well converted into color intensity, even in neighboring tissues, which showed colors of similar wavelength after staining (Supplementary Figure S1G). Each component was quantified as a percentage of pixels of the total tissue pixels. Bright blue pixels were interpreted as proteoglycans. Because of a color-mismatched stain caused by potential overexposure, the surgical pathologist annotated the presence of vSMCs in QuPath using the Pentachrome stain. An object detection algorithm was then applied to get the total area of tissue.22 Statistical Analysis Statistical parameters were presented as the mean and SD, and calculations were performed using MATLAB and Prism. Averages were presented as mean and median, and variance was presented as SD or SEM. Analysis of Variance and independent paired Student’s t-tests were performed for stenotic and nonstenotic lesions coming from the same AVF, followed by adjustment for multiple comparisons. A paired t-test was performed for vSMC abundance (proteoglycan, versican, versikine, aggrecan, perlecan, and syndecan). The stenotic and nonstenotic regions were derived from the same AVF specimens. Results We examined 21 AVF explants. The baseline characteristics of the patients are provided in Table 1. The average age was 56 years, and 77% were male. The cohort included 9 Black, 6 White, 4 Hispanic, and 2 patients of unknown race and/or ethnicity. Diabetes and hypertension were the predominant causes of ESKD among the patients, followed by glomerular disease (29%). Additional comorbidities are presented in Table 1. Table 1Patient characteristicsParametersn (%)Total (21)Age (yrs) average58Sex (male, female) Male16 (76 %) Female5 (24 %)Race Blacks12 (57 %) White4 (19%) Hispanic3 (14 %) Other2 (10 %)Cause of ESRDDiabetes11 (52%)Hypertension21 (100%)Polycystic kidney disease2 (5%)Glomerular disease9 (23 %)Comorbidities Diabetes11 (52%) CAD5 (24 %)COPD3 (14%)Hypertension21 (100 %)GI bleeding5 (24%)Heart failure5 (24%)PVD3 (14%)Stroke4 (19 %)HIV1 (5%)Hepatitis C2 (10%)Cancer3 (14 %)Autoimmune4 (19 %)Hyperlipidemia13 (62%)DVT/PE5 (24%)Smoking0 (0%)Other1 (5 %)Valvular heart disease3 (14 %)Renal transplantation9 (43 %)Atrial fibrillation2 (10%)CAD, coronary artery disease; COPD, chronic obstructive pulmonary disease; DVT-PE, deep vein thrombosis/pulmonary embolism; ESRD, end-stage renal disease; PVD, peripheral vascular disease.Autoimmune disease – systemic lupus erythrometosis, rheumatoid arthritis, and vasculitis. Patient characteristics CAD, coronary artery disease; COPD, chronic obstructive pulmonary disease; DVT-PE, deep vein thrombosis/pulmonary embolism; ESRD, end-stage renal disease; PVD, peripheral vascular disease. Autoimmune disease – systemic lupus erythrometosis, rheumatoid arthritis, and vasculitis. Eighty percent of AVFs were of the brachiocephalic type (18 of 21), and the remaining were of the radiocephalic type. The average time since AVF creation was 7.9 years. Nearly 60% of the AVFs had at least 1 previous interventional procedure, including angioplasty or stent placement. Sixty-two percent of the patients had previously used central venous hemodialysis catheters. Aneurysms and pseudoaneurysms were the primary reasons for explantation in most cases (18 out of 21). The rest of the AVF explantation was because of severe pain at the site of the fistula, fistula thrombosis, or cosmetic reasons following renal transplant, where the AVF remained redundant. AVFs removed for cosmetic reasons were considered normal, as they had no history of AVF stenosis or abnormalities. One patient with an AVF aneurysm had ulceration and pain at the site of the aneurysm with impending bleeding. Of 18 patients with aneurysms and pseudoaneurysms, 15 showed evidence of stenosis on previous venography, which was also evident in the explanted AVFs (Figure 1a). The gross examination of AVF explants revealed a dilated aneurysmal segment near the normal AVF wall (Figure 1a - white asterisk depicts an aneurysmal segment, and white arrowhead points to the AVF wall). Figure 1b shows a representative image of a stenosis from an explant with a small segment of AVF (a white asterisk represents the prestenotic AVF segment, and black arrowheads point to the stenosis). AVF explants with different pathologies. (a) and (b) show 2 surgically explanted AVFs. The white asterisk represents the aneurysmal segment, whereas the white arrowhead points to the AVF wall. (b) depicts an AVF (white arrowhead) next to a stenotic region, indicated with two black arrowheads. (c–j). Shown are hematoxylin and eosin-stained paraffin-embedded AVF explant sections with images captured at 40x and 200x magnification. scale bar = 100 microns. B, bleeding; C, calcification; L, lumen; T, thrombus. All the AVFs were annotated by a surgical pathologist to identify different layers and wall components, as shown in Supplementary Figure S2. Figures 1c and 1d depict a segment of the hematoxylin and eosin-stained normal AVF with evidence of bleeding (marked as B, Figure 1d) and inflammatory infiltrate (yellow asterisk, Figure 1d). Intimal hyperplasia is shown by a black asterisk in the subendothelial region in Figure 1c. Figures 1e and 1f show a segment of AVF removed with a pseudoaneurysm (a black asterisk is placed on the flanks of the breached wall, Figure 1e). Figures 1g and 1h show a segment from an aneurysm with an uneven vessel wall thickness (a black asterisk points to a thicker wall, and a black arrowhead shows a thin aneurysmal wall). Figures 1i and 1j show images of stenosis in the AVF, as shown in Figure 1b. The stenotic segment showed a lumen narrowing, with an organized thrombus adherent to the AVF wall (marked by T). Regions of microcalcification were noted in the AVF wall (marked as C in Figures 1i and 1j). All fistulas showed varying degrees of intimal hyperplasia. Multiple cross-sections for each AVF were examined to characterize the intimal hyperplasia. Seven of 21 AVFs showed intimal hyperplasia uniformly encircling the entire lumen of the fistula (concentric hyperplasia), and the remaining AVFs showed eccentric hyperplasia. Thirty percent of fistulas showed evidence of intraluminal thrombus formation. Sirius Red was used to examine the collagen fiber deposition (red/orange staining) and its orientation (Figure 2). Collagen was noted predominantly in the subintimal and medial regions of normal AVF walls (black asterisk, Figure 2a and 2b). The collagen was disorganized in regions (indicated with a black asterisk in Figure 2a) closer to the cannulation track in the AVF wall (black lines in Figure 2b). In aneurysms and pseudoaneurysms, collagen uniformly circled the vessel lumen (black asterisks in Figure 2e and 2f). Collagen deposition was nonuniform in some areas of the aneurysm wall (Figure 2e and 2f, black arrowhead points to a region with less collagen deposition; the black asterisk marks an area with higher collagen content). In contrast, stenosis was characterized by striking circumferential collagen deposition, in a circumferential manner, throughout the AVF wall (Figure 2g and 2h). Collagen and proteoglycans constitute the wall of AVF. (a–h) Representative Sirus red stained paraffin-embedded AVF explant sections with images captured at 40x and 200x magnification. scale bar = 100 microns. Dotted lines correspond to the needle marks. (i–p) Representative Movat Pentachrome-stained paraffin-embedded AVF explant sections with images captured at 40x and 200x magnification. scale bar = 100 microns. (q) AVF wall component quantified using a customized image analysis pipeline. N = 21. ANOVA compared all the groups. P = 0.003. The students’ t-test was used to compare individual groups. AU, artificial unit. In addition to collagen, proteoglycans are essential components of the vessel wall and are identified using the Movat Pentachrome stain. Movat Pentachrome stains cellular and extracellular components differently, specifically (vSMCs, stained red) and proteoglycan (stained blue). Normal AVFs showed proteoglycan deposition in the subintimal and median regions (a black asterisk, Figure 2i and 2j). vSMCs in these samples are present in the media and intima, as indicated by a white asterisk (Figure 2i). AVFs with an aneurysm showed a proteoglycan deposition pattern similar to collagen. Regions rich in collagen (Figure 2e and 2f) showed prominent proteoglycan deposition (Figure 2k and 2l). A black asterisk points to an area with higher ECM deposition, and a black arrowhead points to a region with lower proteoglycan deposition (Figure 2m and 2n). The stenotic wall of all the AVF segments showed a heavy deposition of proteoglycan (indicated by a black asterisk in Figure 2p) interspersed with vSMCs (black arrowhead). To quantify the expression of different components in a heterogeneous tissue, such as AVF stained with different colors, we used a customized color-based image segmentation pipeline (given in Methods) on several high-power images of all AVFs. Proteoglycans and collagen were higher in the walls of AVFs than vSMCs (P < 0.001 and P = 0.0038, respectively, Figure 2q). Proteoglycans are a distinct class of macromolecules consisting of 1 or more glycosaminoglycans covalently attached to a core protein.23 There are several types of proteoglycans with distinct expression patterns, and some are expressed in the vasculature. The consecutive sections of AVFs were stained with glypican, aggrecan, perlecan, syndecan, and versican using immunohistochemistry in the stenotic and nonstenotic regions (Figure 3a–3f). Nonstenotic areas were selected from regions adjacent to the stenotic areas. All the proteoglycans were expressed in both the stenotic and nonstenotic segments, albeit to varying degrees. Every specimen was examined for cellular and noncellular components. The expression of glypican and aggrecan was low and similar across stenotic and nonstenotic segments (Figure 3a and 3b). Perlecan and syndecan expression was variable in the AVF wall and occasionally higher in the nonstenotic than in the stenotic areas (Figure 3c and 3d). Among the examined proteoglycans, versican and versikine were consistently higher in stenotic regions than in nonstenotic areas (Figure 3e). Vasa vasorum in the wall of AVFs showed prominent expressions of versikine (marked by blue arrowheads, Supplementary Figure S3) compared to glypican, aggrecan, and perlecan. Syndecan showed minimal expression in vasa vasorum. Versican and versikine show prominent expression in the AVF stenotic areas. (a–f) Representative IHC images of proteoglycan-stained paraffin-embedded AVF explants. Images captured at 40x and 200x magnification. scale bar = 100 microns. Since versican showed consistently high expression in the stenotic region, its expression was probed in different AVF pathologies (Figure 4). Versican was seen deposited in the subintimal region of normal AVFs and aneurysmal segments of AVFs (black asterisk in Figure 4b and aneurysm in Figure 4c and 4d). Versican showed prominent expression in the wall of stenotic segments in both the medial and intimal layers (black asterisk in Figures 4e and 4f). Versican correlates with versikine in AVF stenosis (a–f) Paraffin-embedded sections were stained via IHC with an antiversican antibody (brown stain on tissue). The box with the dotted line is imaged at 200X magnification. Scale bar = 100 microns. (g–h) Linear regression between the expressions of versican and versikine in the stenotic and nonstenotic regions. Proteoglycan from IHC was quantified using ImageJ and expressed as integrated density, as previously reported.19,24 Integrated density is the combination of pixel number and intensity of each pixel, normalized to the area (per micron2). It is used to quantify a monochromatic signal corresponding to the expression of a specific protein. Compared with the nonstenotic area, the stenotic area showed a 2.5- to 3-fold higher expression of versican (P = 0.001) and ∼2-fold higher expression of versikine (P= 0.0003; Table 2). Other proteoglycans showed no significant differences between the stenotic and nonstenotic regions. Versikine is a signaling molecule cleaved from versican by members of the metalloprotease family.25 Versican showed a strong and positive correlation with versikine in the stenotic region (R2 = 0.592, P = 0.0008) (Figure 4g) but not in the nonstenotic segments (R2 = 0.113, P = 0.221) (Figure 4h), supporting the potential for active signal transduction by versican in the stenotic region. Versikine showed the highest expression in vasa vasorum compared with other proteoglycans (Supplementary Figure S3A and S3B). The above results show that vSMCs constitute a smaller fraction of the AVF wall. In contrast, proteoglycans, such as versican, constitute the dominant component of the AVF wall, especially in stenotic regions. Table 2Quantitation of proteoglycans in the stenotic and nonstenotic areasProteoglycansNonstenotic areaStenotic areaMeanSEMMeanSEMGlypican51.211.874.311.5Aggrecan69.05.0889.613.8Perlecan13430.311913.9Syndecan12345.220351.6Versicana16040.9703109Versikineb42.18.3131241.0aP = 0.000157.bP = 0.00003. Quantitation of proteoglycans in the stenotic and nonstenotic areas P = 0.000157. P = 0.00003. Proteoglycans are secreted proteins produced by various cell types, including vSMCs. We, therefore, posited that vSMCs demonstrate a secretory phenotype in the stenotic area. To test this notion, a well-established set of cellular markers of pro-proliferative vSMCs (Ki-67) and contractile and/or quiescent vSMCs (MYH11, Calponin, and alpha smooth muscle actin) and secretory (osteopontin) were used in the stenotic (Figure 5) and nonstenotic regions, and images were captured at 2 different magnifications. Ki67 was expressed throughout the AVF wall and showed 25% to 30% increase in the stenotic area (Figure 5a and Supplementary Figure S4A) compared with the nonstenotic region (Figure 5b). The expression of MYH11 and Calponin was prominent in the stenosis, especially in the subendothelial layers, and was higher than in nonstenotic regions (blue arrowheads, Figure 5a and 5b). However, the AVF stenotic and hyperplastic wall showed no expression of osteopontin (Figure 5a and 5b, bottom panels). vSMCs show secretory phenotype in the stenotic region of AVF explants. (a–b) Representative images of AVF nonstenotic (a) and stenotic areas (b) stained using IHC with indicated antibodies. The box with the dotted line is imaged at 100X magnification. Scale bar = 100 microns. (c) The stenotic region of the AVF is stained for Ki-67. The areas from the lumen are shown with the dotted lines. Scale bar = 100 microns. (d). The average integrated density of Ki67+ cells in between two groups is shown. N = 15/group. The contractile markers were increased by 1.8- to 3-fold in stenotic areas (MYH11: P = 0.002 and Calponin: P = 0.0004) compared with nonstenotic areas (Supplementary Figure S4B-S4C). Although there was a trend toward higher expression of smooth muscle actin in the stenotic segment, there was no statistical difference between the stenotic and nonstenotic segments (Supplementary Figure S4D). The localization of proliferating cells varied in the AVF wall (Figure 5c). The immediate subendothelial space (<100 microns from the lumen, delineated with a blue dotted line) had fewer Ki67-positive cells than the space beyond 100 microns, indicated with a yellow asterisk (∗∗P = 0.0052;Figure 5d). These data suggest that although vSMCs demonstrate the proliferative phenotype, they show a higher secretory phenotype in stenotic areas of AVFs. Proliferating cells are predominantly located beyond 100 microns within the fistula wall, away from the lumen. Discussion The current study probes the morphology and the cellular and noncellular composition of mature explanted AVFs, providing insight into key characteristics of stenoses. Mature AVF walls are composed of proteoglycans as the predominant component, followed by collagen and vSMCs. Among proteoglycans, perlecan, versican, and versikine are highly expressed in the stenotic areas along with dense collagen deposition. The vSMCs in the stenotic lesions exhibit a contractile phenotype, which is known to be associated with cellular quiescence. This study describes postcannulation collagen disruption, an aspect of postcannulation healing that is largely overlooked. We also developed and validated a customized analytical tool to quantify various elements in heterogeneous AVF tissue. This study advances our understanding of vascular remodeling associated with stenotic lesions in the setting of secondary AVF failure. Following AVF creation, the venous segment is exposed to markedly increased arterial blood flow and shear stress, which induces outward (expansive) remodeling through luminal dilation, accompanied by adaptive wall thickening driven by smooth muscle cell hypertrophy and/or hyperplasia and ECM reorganization (inward remodeling).22 Together, these processes promote venous arterialization and successful AVF maturation.26 Importantly, once an AVF has fully matured, outward remodeling is largely complete, and vessel caliber typically stabilizes. Therefore, in mature AVFs, stenosis is a form of pathologic inward remodeling, characterized by progressive wall thickening, ECM accumulation (including proteoglycans and collagen), and luminal encroachment occurring within an already dilated conduit. This process is mechanistically distinct from early maturation failure and reflects chronic injury responses to repeated cannulation, altered local hemodynamics, inflammation, and previous endovascular interventions. Our findings should therefore be interpreted within this framework of postmaturation maladaptive inward remodeling, rather than as a failure of initial outward remodeling. Smooth muscle cells in AVF play an important role in vascular remodeling. SMCs are described along a phenotypic continuum, classically categorized as contractile (quiescent) or synthetic and/or secretory. Contractile vSMCs are characterized by high expression of actin–myosin machinery genes, including MYH11, CNN1 (calponin), ACTA2 (αSMA), and TAGLN, and are considered terminally differentiated, quiescent cells with low proliferative and biosynthetic activity.27,28 These cells primarily regulate vascular tone and structural integrity and are not actively engaged in the ECM.28 In contrast, synthetic (secretory) vSMCs exhibit downregulation of contractile markers (e.g., MYH11, CNN1) and upregulation of genes associated with ECM production, migration, and inflammation, including osteopontin (SPP1), collagens, proteoglycans, and matrix-modifying enzymes.27,29 We observed higher integrated density of MYH11 and calponin without a corresponding increase in osteopontin, suggesting an apparent predominance of vSMCs retaining a contractile phenotype at this late, symptomatic stage of disease. However, we interpret these findings within the broader framework of vSMC phenotypic plasticity. Advanced stenotic lesions are characterized morphologically by ECM accumulation and proteoglycan-rich remodeling—features typically associated with earlier secretory and/or proliferative vSMC states. Thus, it is likely that the AVF stenotic lesion represents a temporal continuum in which vSMCs initially adopt a synthetic and/or proliferative phenotype during lesion formation and subsequently re-express contractile markers as the lesion stabilizes. Importantly, vSMC phenotypes are not binary, and cells may coexpress contractile and synthetic programs, including myofibroblast-like states. This notion is consistent with several studies showing that vSMCs exhibit marked phenotypic plasticity and do not exist as discrete contractile or synthetic states; rather, they transition along a continuum of phenotypes in response to local environmental and mechanical cues, with the capacity to coexpress contractile and synthetic programs.27,30,31 This temporal shift may have important implications for understanding the biology of mature AVF stenosis and for therapeutic strategies targeting late-stage access failure. Versican deposition in stenotic lesions is consistent with previous literature. Studies by Kenagy et al.32 using a baboon polytetrafluoroethylene aorto–iliac vascular graft model identified versican as a primary structural driver of inward vascular remodeling. Rather than acting as a passive bystander, versican accumulates in the ECM, creating a hydrated, permissive environment that facilitates smooth muscle cell retention and intimal thickening. Crucially, these neointimal lesions are highly dynamic and responsive to hemodynamic forces. This understanding further highlights the versican-versikine biochemical axis as a potential target for future therapeutic investigation to maintain vascular patency. Taken together, AVF stenosis in mature AVFs is likely driven by maladaptive ECM remodeling with proteoglycan accumulation, which contributes to luminal narrowing and altered wall mechanics, rather than by active cell growt.33,34 The vasa vasorum are microvessels providing oxygen and nutrients to the AVF walls. Even though no correlation was observed between primary AVF failure during maturation and vasa vasorum, the role of the latter in abnormal AVF remodeling in a matured AVF cannot be ruled out.35 In that regard, the higher expression of versikine around vasa vasorum raises the possibility of its involvement in angiogenesis within the AVF wall. Paclitaxel-coated balloons or stents are used for angioplasty of AVF stenosis. Paclitaxel stabilizes microtubules and inhibits cell division, thereby preferentially targeting actively proliferating cells. In a vascular environment dominated by quiescent, nondividing contractile vSMCs, the biological efficacy of paclitaxel is expected to be substantially attenuated.36, 37, 38, 39 This may help explain the limited success, restenosis, or heterogeneous response observed with paclitaxel-based endovascular therapies in mature AVFs10,40, 41, 42, 43, 44 where stenosis occurs in a vessel wall that has already undergone arterialization and cellular quiescence rather than active hyperplasia. The low number of patients, the retrospective nature of this investigation, and the use of a single marker for the secretory phenotype are limitations of this work. Our objective was to investigate positive remodeling associated with stenosis. In contrast, aneurysmal and pseudoaneurysmal segments are defined by wall thinning and structural dilation, processes that are mechanistically distinct and are generally associated with loss or reduction of matrix components, including proteoglycans (negative regulation), rather than their accumulation. Future studies are required to examine the pseudoaneurysmal formation because of repeated trauma and bleeding. This was a retrospective study with a limited number of patients, and all AVF explants were formalin-fixed and paraffin-embedded at the time of clinical acquisition, precluding AVF remodeling follow-ups. However, we hope that this study paves the way for a single-cell sequencing experiment to define the cell state and associated differentially expressed multiple genes, as was done for the maturing AVF.41 The current findings support the detailed characterization of mature AVF walls and deeper insight into alterations in cellular and noncellular components of stenosis in mature AVF. Also, the observations provide potential insights into the heterogeneous response of paclitaxel-coated balloons and stents in AVF stenosis lesions and support studies using mechanistically derived therapies targeting specific components of stenotic lesions in mature AVFs. Disclosure LMD received consulting fees from AstraZeneca and Merck, and compensation for serving on Data and Safety Monitoring Boards for the National Institute of Diabetes and Digestive and Kidney Diseases, and Data Monitoring Committees for US Renal Care, CSL Behring, and Vertex Pharmaceuticals. AF received consulting fees from Le Maitre, BioGenecell, and VARMx, Sanifit, and grants from Novo Nordisk Foundation, and participated on a Data Safety Monitoring Board of iThera and Dialysis X. All other authors have declared no competing interests.