Intimal Hyperplasia and Arteriovenous Fistula Failure: Looking Beyond Size Differences

Abstract

The development of venous intimal hyperplasia (IH) has been historically associated with failure of arteriovenous fistulas (AVFs) used for hemodialysis. This long-standing assumption, made on the basis of histologic observations, has been recently challenged by clinical studies indicating that the size of the intima by itself is not enough to explain stenosis or AVF maturation failure. Irrespective of this lack of association, IH is present in most native veins and fistulas, is prominent in many patients, and suggests a role in the vein that may not be reflected by its dimensions. Therefore, the contribution of IH to AVF dysfunction remains controversial. Using only clinical data and avoiding extrapolations from animal models, we critically discuss the biologic significance of IH in vein remodeling, vascular access function, and the response of the venous wall to repeated trauma in patients receiving hemodialysis. We address questions and pose new ones such as the following: What are the factors that contribute to IH in preaccess veins and AVFs? Do cellular phenotypes and composition of the intima influence AVF function? Are there protective roles of the venous intima? This review explores these possibilities, with hopes of rekindling a critical discussion about venous IH that goes beyond thickness and AVF outcomes.

Introduction

Intimal hyperplasia (IH) is a buildup of myofibroblast-like cells (neointimal cells) and extracellular matrix (ECM) within the tunica intima, the innermost layer of the vein (1). In preaccess veins of patients with CKD, IH manifests as an idiopathic and benign histologic feature that does not compromise blood flow (2–6). After arteriovenous fistula (AVF) creation, expansion of the intima may remain innocuous or aggravate inward remodeling, leading to stenosis and access failure (3). The actual contribution of IH in relation to other transformation processes in the wall after AVF creation remains uncertain (7). Importantly, once stenosis becomes a pathology, it is not easily treatable, with frequent recurrences after endovascular procedures (8–14). The lack of more effective therapies against stenosis reflects our simplistic view of the processes of intimal expansion and wall remodeling, and our unawareness of the characteristics that distinguish between benign and occlusive IH.

This review critically discusses what we know about IH in human veins and AVFs, what we are missing, and how this knowledge may influence IH-targeted therapies to improve AVF outcomes. We discuss the historical assumption that IH alone causes stenosis and AVF failure and extend our debate beyond IH size, the focus of published research in this area. Finally, we highlight the necessity for innovation, state-of-the-art omics, and single-cell technology to clarify the actual role of IH in venous remodeling.

Preexisting IH

IH in the Preaccess Vein: More Common than Previously Thought

The cephalic and basilic veins are the preferred choices for AVF creation (15). These are medium-sized veins with the three vascular layers well defined (tunica intima, tunica media, and tunica externa or adventitia) and diameters between 1 and 5 mm (Figures 1 and 2) (16,17). The intima is the innermost layer of the vessel and is demarcated by a thin or discontinuous internal elastic lamina on the medial side, and a continuous endothelial line that separates it from the lumen. Thin folds of collagen-rich connective tissue, covered by endothelium, extend from the intima and form the valves at regular intervals along the vein.

Figure 1.

Minimal to moderate intimal hyperplasia in trauma donors with normal renal function. (A–D) Cross-sections of the (A and B) cephalic and (C and D) basilic veins from a Hispanic 56-year-old male without history of hypertension, diabetes, or coronary artery disease. (E and F) Cephalic and (G and H) basilic veins from a Hispanic 55-year-old male with history of controlled hypertension (<5 years). Sections were stained with Movat pentachrome stain, with cells showing in brown/red, collagen in yellow, and elastin in black. Boxed areas in (A), (C), (E), and (G) are magnified in (B), (D), (F), and (H), respectively. Arrows in (B) and (H) identify the internal elastic lamina. A, adventitia; I, intima; M, media.

Figure 2.

Variability in intimal hyperplasia in preaccess basilic veins from patients with ESKD. (A–C) Cross-sections of basilic veins collected during first-stage surgery of a two-stage brachiobasilic arteriovenous fistula (AVF). Patient in (A) is a 42-year-old Black female with history of hypertension; patient in (B) is a 68-year-old Black female with hypertension and diabetes; and patient in (C) is a 65-year-old Hispanic female positive for hypertension, diabetes, and coronary artery disease. All three veins matured successfully after AVF creation. Sections were stained with Movat pentachrome stain, with cells showing in brown/red, collagen in yellow, and elastin in black. Boxed areas in (A) to (C) are magnified in (D) to (F), respectively. Arrows in (E) and (F) identify the internal elastic lamina. A, adventitia; I, intima; M, media.

The size of the intima layer in preaccess veins ranges from almost inexistent to thick and rich in intimal cells and ECM, with this latter scenario being the norm rather than the exception (2–6,18–24). Almost 20 years ago, Wali et al. (25,26) observed generalized IH in cephalic veins from 20 patients with renal failure. In more recent and larger patient cohorts (N=57–129), maximal intimal thickness (the longest distance between the media and the lumen) ranged from 1 to 660 µm in forearm and upper-arm preaccess veins (Table 1), with higher values in the latter (2,4,6). The Hemodialysis Fistula Maturation (HFM) Study confirmed the high prevalence of IH in 365 preaccess veins and quantified it as the percentage of luminal occlusion (Table 1) (5). Of the patients, 22% presented moderate IH (21%–40% luminal occlusion), whereas 35% portrayed severe IH (>41% occlusion). In agreement with Allon et al. (4), the percentage of luminal occlusion was lower in cephalic veins (mean 31%) than in upper-arm vessels (mean 40%). Martinez et al. (3) expressed IH as intima/media area ratio (I/M ratio) in 110 upper-arm veins to account for morphometric changes in the media, such as atrophy or hypertrophy (18,20,25,26). The median I/M ratio in this cohort was 0.32, in line with previous reports of cephalic and other preaccess veins (Table 1) (18,19,24).

Table 1.

Intimal morphometry and associations with arteriovenous fistula outcomes in the hemodialysis population

Study	N a	Vein	Parameter(s)	Measurementsb	Association(s)
Preexisting intimal hyperplasia
Feinfeld et al. (23)	15	Brachial (n=15)	Ave. I thickness	6.0±0.9 µm	N/A
Lee et al. (18)	12	Cephalic (n=6), axillary (n=3), antecubital (n=1), basilic (n=1), brachial (n=1)	Ave. IM thickness	0.34±0.12 mm	Significant association with maturation failure (P=0.03, n=7)
			Max. IM thickness	1.16±0.30 mm	N/A
			I/M area ratio	0.24±0.07	N/A
			% Luminal occlusion	47%±9%	Lack of association with maturation failure (P=0.09, n=7)
Wasse et al. (24)	10	Cephalic, basilic	Ave. I thickness	0.066±0.019 mm	N/A
			Max. I thickness	0.166±0.042 mm	N/A
			Mean I/M thickness ratio	0.26±0.07	N/A
			Max. I/M thickness ratio	0.69±0.19	N/A
			Intimal area	0.27±0.08 mm2	N/A
			I/M area ratio	0.24±0.06	N/A
Allon et al. (4)	113	Upper arm (65%), forearm (35%)	Max. I thickness	0.022 (0.013–0.045) mm	Lack of association with postoperative stenosis (P=0.49)
Lee et al. (21)	29	N/A	Mean I/M thickness ratio	0.43±0.07	N/A
			Max. I/M thickness ratio	0.86±0.07	N/A
Lazich et al. (19)	18	Cephalic (n=18)	Max. I thickness	0.052–0.81 mm	N/A
			Intimal area	0.16–7.70 mm2	N/A
			I/M area ratio	0.07–1.80	N/A
			Mean I/M thickness ratio	0.07–1.99	N/A
			Max. I/M thickness ratio	0.14–2.44	N/A
			% Luminal stenosis	45%–96%	N/A
Tabbara et al. (2)	57	Basilic (n=54), brachial (n=3)	Max. I thickness	0.18 (0.10–0.20) mm	Lack of association with primary unassisted patency (P=0.2, n=52)
			I/M area ratio	N/A	Lack of association with primary unassisted patency (P=0.2, n=52)
HFM Study (5,52)c	365	Cephalic (69%), basilic (29%), brachial (2%)	% Luminal occlusion	28%±27% (cephalic), 40%±30% (basilic), 21%±23% (brachial)	Lack of association with postoperative stenosis at 1 day (P=0.49), 2 weeks (P=0.91), or 6 weeks (P=0.07); lack of association with unassisted (P=0.07) or overall maturation failure (P=0.11)
Martinez et al. (3)	110	Basilic (n=104), brachial (n=4), cephalic (n=2)	I/M area ratio	0.32 (0.22–0.52)	Lack of association with maturation failure (P=0.7)
Allon et al. (6)	129	Upper arm (65%), forearm (35%)	Max. I thickness	0.037±0.040 mm	N/A
Postoperative intimal hyperplasia
Roy-Chaudhury et al. (41)	4	Cephalic (n=4), all early failures	Mean I/M thickness ratio	3.12±0.43	N/A
			Max. I/M thickness ratio	7.77±1.49	N/A
			I/M area ratio	1.67±0.10	N/A
			% Luminal stenosis	86%±3%	N/A
Lee et al. (21)	20	Cephalic (n=15), basilic (n=5); all stenotic segments	Mean I/M thickness ratio	3.84±0.55	N/A
			Max. I/M thickness ratio	7.78±0.88	N/A
Tabbara et al. (2)	79	Basilic (n=74), brachial (n=5)	Max. I thickness	0.62 (0.38–0.86) mm	Lack of association with maturation failure (P=0.3); lack of association with primary unassisted patency (P=0.6)
			I/M area ratio	N/A	Lack of association with maturation failure (P=0.4); lack of association with primary unassisted patency (P=0.8)
Duque et al. (71)d	14	Basilic (n=12), brachial (n=2); all AVFs had stenotic and nonstenotic segments	I area	3.33 (1.94–4.86) mm2 in nonstenotic, 3.33 (2.29–5.16) mm2 in stenotic	Lack of association with focal stenosis (P=0.26)
			Min. I thickness	0.09 (0.05–0.31) mm in nonstenotic, 0.11 (0.05–0.43) mm in stenotic	Lack of association with focal stenosis (P=0.18)
			Max. I thickness	0.75 (0.54–1.08) mm in nonstenotic, 0.98 (0.78–1.20) mm in stenotic	Lack of association with focal stenosis (P=0.22)
			Min. IM thickness	0.37 (0.17–0.70) mm in nonstenotic, 0.30 (0.23–0.88) mm in stenotic	Lack of association with focal stenosis (P=0.22)
			Max. IM thickness	1.14 (0.84–1.38) mm in nonstenotic, 1.38 (1.30–1.57) mm in stenotic	Lack of association with focal stenosis (P=0.13)
			I/M area ratio	0.97 (0.63–1.18) in nonstenotic, 1.00 (0.70–1.20) in stenotic	Lack of association with focal stenosis (P=0.73)
Martinez et al. (3)	115	Basilic (n=97), brachial (n=14), cephalic (n=4)	I/M area ratio	0.77 (0.48–1.30)	Lack of association with maturation failure by itself (P=0.09, n=115), but significant association in AVFs with high medial fibrosis (P=0.04, n=58)

Ave., average; I, intima; N/A, not reported or studied; IM, intima plus media; max., maximum; I/M, intima/media; HFM, Hemodialysis Fistula Maturation; AVF, arteriovenous fistula; min., minimum.

^aNumber of veins analyzed after study exclusions.

^bValues presented as mean±SEM (SD in the HFM Study [5,52] and Allon et al. [6]), median (interquartile range), or range in Lazich et al. (19).

^cData obtained from Alpers et al. (5) and Cheung et al. (52).

^dPairwise comparison of stenotic and adjacent nonstenotic segments in upper-arm AVFs.

Despite its common occurrence in preaccess veins, whether IH development is influenced by CKD remains unknown because experimental animal models do not develop spontaneous IH. Various studies support an increase in IH in the setting of CKD (6,21,25–27), but the number of non-CKD upper-extremity veins is low (three to 15 individuals), which makes it difficult to draw a definite conclusion on this issue. The I/M ratio of the great saphenous vein was also found to be significantly higher in patients with CKD compared with controls, and in those with ESKD versus CKD stages 1 and 2 (27), but it is not clear whether these groups were matched with respect to age and baseline characteristics. It is tempting to speculate that IH increases during the course of renal dysfunction secondary to volume/flow overload (anemia, sodium, and water retention) and other poorly defined clinical factors. However, frequent IH was observed in cephalic and saphenous veins from elderly patients with normal renal function (28), suggesting uremia is not the only vascular insult causing intimal thickening. Synergistic insults may include endothelial dysfunction in CKD and vascular injury related to venipuncture or catheterization. On the other hand, the presence of a thick intima in basilic veins (2,3), as in the superficial cephalic vein (4–6,19), suggests that mechanisms other than venipuncture-related trauma promote IH. Single-cell sequencing and spatial proteomics may help identify differences in cell and ECM composition (if any) between the CKD and non-CKD preaccess intima. This may, in turn, uncover common and disease-relevant origins of IH.

Composition of the Preaccess Intima: Identifying Knowledge Gaps

Three types of cells predominate in the intima of preaccess veins: endothelial cells (ECs), smooth muscle cells (SMCs), and myofibroblasts/fibroblasts (1,2,5,18–20,29). ECs line the luminal side of the intima, whereas SMCs and myofibroblasts, embedded in ECM, populate the core of the layer. ECs play an essential role in preventing thrombosis, but their contribution to controlling IH has not been fully elucidated. Although there is wide support for the inhibitory effect of EC-derived nitric oxide in intimal cell proliferation and migration (30), there is also evidence for other endothelial paracrine factor(s) that stimulate venous IH (31). The overall effect of the endothelium on IH is likely dependent on flow and pathophysiologic conditions. Along these lines, profound changes in EC and SMC morphology (25,26,32) and function (33–39) have been detected in patients with CKD.

Using a combination of contractile (smooth muscle myosin heavy chain [SM-MHC], desmin, h-caldesmon, calponin), synthetic (vimentin), and pan SMC markers (α-smooth muscle actin [αSMA]), various groups have observed a mixture of SMC and/or myofibroblast phenotypes in the intima of preaccess veins (2,18–20,29,40,41), whose functions remain uncertain. Contractile SMCs are typically associated with low proliferation and migration rates, and low secretion of ECM (42). The opposite behavior is characteristic of “synthetic” or “myofibroblastic” SMCs that have lost expression of contractile markers. Serum from patients with dialysis favors the synthetic transformation of cultured human SMCs by promoting epigenetic downregulation of contractile gene expression (27). Interestingly, despite the high number of synthetic SMCs in preaccess veins, they show minimal staining of the proliferation and metabolic markers Ki-67 and phosphoglucomutase 1, respectively, suggesting that intimal cells are relatively quiescent before access creation (5,18). The HFM Study also reported rare apoptotic cells by cleaved caspase 3 expression in <10% of analyzed intimas (four of 48) (5). How expression of contractile markers in CKD veins relates to venoconstriction or dilation is unclear. A thick intima likely serves as a barrier for the diffusion of both circulatory vasoactive factors and EC-derived molecules that regulate medial SMC contraction or dilation. This may be an advantageous adaptation to reduce vasoconstrictive responses. On the other hand, studies in saphenous veins proposed that intimal thickness >120 µm is associated with impaired endothelium-dependent vasodilation (43). If there is a signaling cascade of soluble factors from ECs or the lumen that is amplified by intimal myofibroblasts or SMCs remains to be discovered.

In contrast with the abundance of inflammatory cells in the arterial intima with disease, the number of immune cells in preaccess veins is minimal. Approximately 50% of analyzed veins in the HFM Study (25 of 48) showed only one CD68-expressing macrophage in the intima, with approximately 7.7 cells in the whole section (5). Martinez et al. (44) also reported low numbers of CD68⁺ macrophages (about 40 cells per cross-section) in 45 basilic veins, mostly located at the edge between the media and the adventitia. In terms of T cells, Lee et al. (18) observed minimal CD3⁺ staining in the intima. Despite the low levels of immune infiltration in preaccess veins, a transcriptomic analysis uncovered expression of myeloid-related inflammatory genes in intimal and medial SMCs and myofibroblasts, suggesting a key role of resident cells in vascular inflammation (44). Five genes (CSF3R, FPR1, S100A8, S100A9, and VNN2) were associated with AVF maturation failure, and expression of S100A8 and S100A9 had a weak correlation with postoperative IH (44). Wasse et al. (24) also found expression of TNF-α, TGF-β, and IL-6 in the intima of preaccess veins. Immunohistochemistry analyses demonstrate that not all cells in the intima and media are positive for these proteins (24,44), demonstrating again a heterogeneity of SMC and myofibroblast phenotypes with potential implications for the inflammatory status of the vessel.

The ECM composition of the intima is an important aspect of remodeling, and perhaps the most neglected characteristic of this layer. The HFM Study observed significant interpatient variability by histology in the amount and distribution of collagen and proteoglycans in the intima (5). Intimal expansion in earlier reports of 20 cephalic veins was also characterized by marked deposition of fragmented collagen fibers and dispersed elastin (25,26). Intimal calcification was observed in 2% of the patients in the HFM Study and 15% in Wali et al. (5,25). It is important to note that the proportion and configuration of the ECM and the types of ECM proteins in the intima may play a role in cell proliferation and migration (45–48), vein stiffness (49,50), and/or compressibility of this layer under high flow conditions (51). Accumulation of collagen is associated with fibrosis, whereas high proteoglycan content may confer resistance to compression and act as a reservoir of cytokines and growth factors that influence cell survival and proliferation (49,51). Future proteomic studies are needed for a more accurate characterization of the intimal ECM.

Does Preexisting IH Increase the Risk of AVF Failure?

The initial idea that preexisting IH potentially led to stenosis and poor AVF outcomes has been recently challenged in several independent studies (2–4,52). Allon et al. (4) studied the association between maximal intimal thickness in the preaccess vein and postoperative AVF stenosis. Of the 113 patients included in the analysis, 50% developed a hemodynamically significant stenosis. However, there was no association between IH and the presence of postoperative stenosis (Table 1). This lack of association remained true when analyzed by type of AVF and location of the stenosis (4). The results of this study were confirmed by the HFM Study in 365 individuals (52). The development of stenosis was evaluated by ultrasound at 1 day, 2 weeks, and 6 weeks after AVF creation. Preexisting IH (percentage of luminal occlusion) was not associated with AVF stenosis at any of these time points, nor with the internal diameter of the vessel (52).

The relationship between preexisting IH and maturation failure was analyzed by the HFM Study and Martinez et al. (3,52). The HFM Study found a significant association between the preexisting percentage of luminal occlusion and venous blood flow rate at 6 weeks after access creation (Table 1). However, the association with unassisted or overall maturation failure did not reach statistical significance (52). Preexisting I/M ratio also failed to predict nonmaturation in 110 patients in the study by Martinez et al. (3). Lastly, Tabbara et al. (2) analyzed the association between preexisting IH and primary unassisted patency in 52 upper-arm fistulas. Neither maximal intimal thickness nor I/M area ratio predicted loss of primary patency. Although there seems to be no association between preexisting intimal morphometry and AVF failure, additional studies are needed to assess the effects of intimal cell and ECM composition on postoperative outcomes.

Postoperative IH

IH after AVF Creation: Selective Activation of Preexisting Cells?

The transformation of the vein after AVF creation remains one of the least understood processes in vascular biology. Current knowledge emphasizes the role of ECs in sensing arterial shear stress to release vasodilators that potentially lead to maturation (53,54). However, the endothelium is almost certainly severely damaged by surgical trauma, secondary to the common use of dilators and saline flushing to expand venous size before anastomosis. This suggests that intimal and medial cells likely play a protagonist’s role as mechanosensors of hemodynamic changes and vascular trauma. The best evidence we have about postoperative remodeling of the intima is from two-stage AVFs, which allow us to collect a biopsy of the remodeled vein (now a fistula) during transposition surgery.

In upper-arm fistulas, maximal intimal thickness increased approximately four-fold with respect to the preaccess vein, with values ranging from 0.1 to 2.0 mm in 79 patients who underwent two-stage AVF creation (Table 1) (2). This increase was not associated with the waiting time between AVF creation and transposition surgeries, or with the thickness of the intima in the preaccess vessel. The lack of relationship between preexisting and postoperative IH agrees with a selective activation of cells in the wall and different responses to surgical or hemodynamic injury between patients. Moreover, the absence of correlation between IH and the time between AVF creation and transposition surgeries suggests that most intimal expansion occurs early during maturation. Medial atrophy is frequently seen in AVFs (Figure 3), possibly as a result of cell death or migration of SMCs into the intima. This SMC loss is either replaced by ECM (fibrosis) or results in thinning of the media. A median I/M area ratio of 0.77 was reported in 115 two-stage AVF cross-sections, significantly higher than in native veins (Table 1) (3). In agreement with maximal intimal thickness, I/M ratio also demonstrated a lack of correlation between preexisting and postoperative values in pairwise analyses.

Figure 3.

Heterogeneity of venous remodeling in AVFs from patients with ESKD. (A–C) Cross-sections of juxta-anastomotic AVF segments from two-stage brachiobasilic fistulas collected during second-stage surgery (77–91 days after AVF creation). Patient in (A) is a 38-year-old Hispanic male with history of hypertension, patient in (B) is an 80-year-old Black female with hypertension and diabetes, and patient in (C) is a 40-year-old Black male positive for hypertension and coronary artery disease. All three AVFs failed to mature and underwent a salvage procedure or creation of a new fistula. Sections were stained with Movat pentachrome stain, with cells showing in brown/red, collagen in yellow, proteoglycans in blue, and elastin in black. Boxed areas in (A) to (C) are magnified in (D) to (F), respectively. Arrows in (D) and (F) identify the internal elastic lamina. A, adventitia; I, intima; M, media.

In contrast to native veins (5), IH in most upper-arm AVFs is eccentric (Figure 3). Unequal hemodynamic forces along the length of the vein are thought to explain this morphometry. This has been recently imaged in mice (55), although confirmatory studies in large animal models and humans are needed. It has been proposed that low wall shear stress, pulsatile stretch, and flow turbulence causes injury and elicits proliferation of intimal cells (56–60). However, it is possible that such eccentric appearance may be explained by pockets of increased cell proliferation, migration, and/or ECM deposition; and that these pockets are determined, in turn, by the phenotypes of the preexisting intimal cells.

At the cellular level, the postoperative remodeling process seems to favor the intimal expansion and/or survival of myofibroblasts and synthetic SMCs in the AVF wall. Tabbara et al. (2) observed that the intima of upper-arm AVFs collected at the time of transposition was mostly made up of synthetic SMCs (positive for αSMA, negative for SM-MHC). Other studies also reported a majority of myofibroblasts (αSMA⁺, vimentin⁺, desmin⁻) (40,41,61) and reduced expression of SM-MHC and calponin (29) in the intima of stenotic samples collected at the time of AVF revision. In all studies, medial SMCs retained contractile protein expression (2,29,40,41). The proportions and phenotypes of AVF intimal cells that are actively proliferating remain unknown. High intimal levels of the proliferation marker proliferating cell nuclear antigen (PCNA) were observed in stenotic areas of resected AVFs (41,62), but were contradicted by the more accurate marker Ki-67 (63). High PCNA and cyclin dependent kinase 2 (CDK2) levels in the study by de Graaf et al. (62) were also accompanied by significantly lower expression of the cell cycle regulator p21^Waf1. Future single-cell tracing experiments will help define whether the increase in synthetic SMCs and myofibroblasts in the AVF intima is due to the postoperative dedifferentiation and expansion of contractile SMCs in the wall, or proliferation of a preexisting synthetic population. This information will be instrumental for the design of targeted therapies.

The role of immune cell infiltration in postoperative IH and AVF dysfunction is not clear at the moment. Increased macrophage and T-cell infiltration was seen in 15 stenotic AVF sections compared with preaccess veins (64). In contrast, a comparison of 13 nonthrombosed stenotic samples and 23 thrombosed specimens revealed that immune cell infiltration was in fact characteristic of the latter (65). Similar to the localization of proinflammatory proteins in intimal SMCs and myofibroblasts in preaccess veins (44), various studies have also demonstrated elevated inflammatory and oxidative markers in resident intimal cells of resected AVF specimens (64,66). It is important to note that most of the information about immune cell infiltration after AVF creation comes from extrapolation from animal models (67–70), where it is possible to obtain AVF samples early after surgery. Whether inflammation from infiltrated or resident cells plays a role in human AVF maturation or dysfunction will require the analysis of early human AVF samples (within 2 weeks of surgery), including nonstenotic segments. This may be possible through a multicenter collection of veins from steal syndrome and stenotic accesses that require early surgical revision.

As in preaccess veins, the ECM composition of the AVF intima has been barely studied. Martinez et al. (3) observed various levels and patterns of ECM deposition in the intima of two-stage transposition fistulas, although a comparative analysis of the samples was not presented. These patterns included intimas that were mostly cellular (low in ECM), with widespread ECM distribution, or with separate areas for cells and ECM deposition. Such interpatient variability in composition is likely relevant to the occlusive character of the intima, the response of the vein to cannulation injury, and the efficacy of endovascular treatments.

Postoperative IH and AVF Outcomes: Let Us Update the Theory

The most important question about IH is whether growth of this layer after AVF creation underlies access failure. Various case reports of AVFs that failed observed the presence of moderate to severe IH, but lacked a comparative group of functional fistulas (21,29,41,61). Despite this limitation, these observations reinforced the assumption that a thicker intima was responsible for AVF failure. The analysis of postoperative samples from two-stage upper-arm AVFs has challenged this idea (2,3,71).

Tabbara et al. (2) found a lack of association between postoperative IH (measured as maximal intimal thickness and I/M ratio) and maturation failure in a cohort of 79 individuals (Table 1). These analyses were not adjusted for any other clinical characteristics. Martinez et al. (3) also failed to find an association between I/M ratio and maturation failure in 115 individuals after adjusting for sex effects. Interestingly, the same study demonstrated that postoperative medial fibrosis was significantly associated with failure. Furthermore, IH was associated with failure only in those AVFs with medial fibrosis over the median value, and not in the other half of the accesses (Table 1) (3). Given that high medial fibrosis can adversely influence the biomechanical properties and distensibility of AVFs, this study proposed that, under highly fibrotic wall conditions, high IH is occlusive, but not when the vessel is able to compensate through other biomechanical mechanisms. This underscores the importance of understanding AVF remodeling as a whole and the mechanistic relationships between IH and other wall remodeling processes.

Pairwise comparisons of adjacent stenotic and nonstenotic segments from 14 two-stage upper-arm AVFs further confirmed that IH does not define the true luminal area of the access (71). In this report, there were no significant differences in intima size between both segments (Table 1). Lastly, no significant association has been found between maximal intimal thickness or I/M ratio and primary unassisted patency (2). The above postoperative data on maturation failure, focal stenosis, and primary patency are limited to upper-arm AVFs due to practical limitations. It is possible that postoperative IH has a larger effect on the outcomes of forearm fistulas.

Research Models, Current Challenges, and Pending Questions

Role of Animal Models in the Study of IH

Animals are essential to address basic science questions such as the origin and differentiation of intimal cells, temporal remodeling of the wall, and the effects of local and circulatory stimuli or treatments (29,67,68,72–79). Research mice and rats have the added advantage of allowing genetic manipulation (gene knockins and knockouts, cell labeling, etc.) and inclusion of high numbers of animals. Swine and sheep are often used as translational models to test pharmacologic and endovascular interventions, primarily in arteriovenous grafts (80–83). Arguably, the best animal model for AVF functional studies is the one in sheep, due to the superficial location of peripheral veins, which allows for not only AVF creation but also for potential cannulation (73).

Most animal models develop a certain form of venous IH within 2–6 weeks after AVF creation (29,67,68,72,74–79). However, they have important limitations for the study of the occlusive role of IH in AVF remodeling and failure. In the case of small animals, there are profound differences in preexisting vein morphology (very thin walls and subendothelial space) and hemodynamic characteristics (low blood flow) with respect to humans (29,67,68, 76,78,79). In addition, most models lack an underlying long-term CKD component and a human-equivalent definition of failure. These limitations underscore the necessity of expanding tissue biobanks of human AVFs to all possible forms to promote retro-translational studies, where human findings could be further dissected at the mechanistic level in animal models. Excellent reviews of AVF and CKD animal models have been included in the References (84–87).

Current Challenges in Treating AVF Stenosis: A Call for Mechanism-based Approaches

The idea that IH was the main cause of stenosis in AVFs motivated the use of therapies that treat restenosis in coronary circulation to salvage dysfunctional accesses. As a result, percutaneous transluminal angioplasty (PTA) became the first-line treatment for postoperative stenoses (15). Angioplasty mechanically stretches the vein and compresses the intima, but may cause significant injury to the vessel. Although efficacious in the short term, PTA frequently requires reintervention within 1 or 2 years after the first angioplasty procedure, either due to regrowing of occlusive IH and/or fibrotic scarring of the AVF wall (8–14). Stent placement is the last line of treatment for recurrent and high recoil stenoses due to concerns of vein depletion, stent migration or fracture, and intrastent thrombosis (88,89). Stent grafts are favored for in-stent restenosis (15,89), but are prone to “edge stenosis,” which occurs close to both ends of the stent and migrates toward the center (90,91).

In an attempt to improve postprocedure patency, antiproliferative drugs (mainly paclitaxel) are delivered to the AVF wall by means of drug-eluting balloons (DEB) or stents. Multiple individual studies have shown patency and/or reintervention benefits of DEB versus conventional angioplasty (12–14,92–95). However, meta-analysis studies have yielded variable conclusions, and patency rates >6 months still have much room for improvement (96–99). The observed variability in efficacy with antiproliferative agents and failure to significantly extend long-term patency may indicate insufficient delivery or retention of the drug (100–102), low sensitivity of cells to treatment (103), or a mismatch between the therapeutic effects of the drug and mechanisms of restenosis in AVFs. In vitro data suggest that paclitaxel targets all three presumed processes of restenosis (proliferation, migration, and ECM production) (104,105). However, it is not clear which of these cellular mechanisms are actually targeted in AVFs in vivo. Of note, the effect of paclitaxel in SMCs is cytostatic and not cytotoxic (106,107). Therefore, any stenotic mechanisms that remain unaffected may continue happening or possibly worsen after DEB treatment.

Antistenotic treatment modalities to improve venous remodeling during maturation include perivascular delivery of sirolimus (108), allogeneic ECs (109,110), or pancreatic elastase (111,112), and devices (VasQ, Optiflow) that support the ideal angle of arteriovenous anastomosis (113–117). The latest results on the sirolimus implants are pending (108). VasQ resulted in high maturation rates in single-arm and retrospective studies (117–119), but mixed results in short-term primary or secondary patency compared with the control arm (114,118,119). The rest are currently not recommended by the Kidney Disease Outcomes Quality Initiative guidelines due to lack of phase 3 studies (Optiflow) or significant benefits in AVF outcomes (ECs, elastase) (15). Far infrared radiation and external pneumatic compression (Fist Assist) have shown promising results in AVF maturation parameters (120–123) and secondary patency after PTA in specific patient demographics (124), but require further validation in a broader hemodialysis population. These clinical trials illustrate the desire to innovate in the search for preventive and postoperative AVF treatments. However, until we understand how human AVF cells respond to flow disturbances, repeated cannulations, and endovascular trauma at the molecular level, it is likely that any successes will come after significant trial and error.

What to Look at Beyond Intimal Thickness

Why have we failed to find an association between intima size measurements and AVF outcomes? The answer may be methodologic and/or biologic in nature. From the methodologic point of view, there are many limitations to the way we measure intima size. Two-dimensional and static histologic assessment of IH misses the actual size of the lumen under circulation. In addition, none of the measurements considers the potential compressibility of the intima or distensibility of the vessel. At least one clinical study has reported a lack of association between maximal intimal thickness by histology and internal diameter of the vein (4), illustrating the limitations of two-dimensional morphometry measurements in determining luminal area.

From the biologic point of view, intima size only represents a partial measurement of inward remodeling. Looking for associations between intima size and AVF outcomes ignores other macro processes of the wall, such as inward remodeling of the media, outward remodeling of the wall, and changes in the ECM. Importantly, we still do not understand what drives any of these processes. Is it cell death, changes on SMC phenotypes, or SMC- or immune cell-derived inflammation? Are IH and medial fibrosis mechanistically related? From the biomechanical point of view, how compressible is the intima? What characteristics make it more or less compressible? Does vein distensibility change after AVF creation? A better understanding of the role of SMCs and myofibroblasts in intimal expansion and wall remodeling, and where in the range of their phenotypic transformation they become problematic, will require detailed phenotypic analyses and single-cell omics techniques in clinically relevant human samples.

Lastly, the complexity of AVF remodeling lies in identifying an optimal level of IH and fibrosis that maintains vein integrity under extreme hemodynamic conditions and frequent cannulations, but without causing stenosis. Thus, can we envision protective roles for the intima? Does it prevent excessive immune cell infiltration? Does it protect medial SMCs from the oxidative stress of high oxygen pressures? Do synthetic cells confer regenerative capacity for wall healing after cannulation? We must consider all of these possibilities if we truly want to optimize maturation and prevent restenosis after endovascular treatments.

Disclosures

L. H. Salman reports receiving research funding from Albany Medical Center, Roche funds, and Transonics Inc.; having other interests in/relationships with the American Society of Diagnostic and Interventional Nephrology, American Society of Nephrology, Phraxis (data safety monitoring board), and Renal Physician Association; and having a patent application for the use of 4-methylumbelliferone in diabetic kidney disease (the application is pending review). R. I. Vazquez-Padron reports receiving research funding from the National Institutes of Health, and serving as a scientific advisor for, or member of, Scientific Reports. All remaining authors have nothing to disclose.

Funding

This study was supported by the National Institute of Diabetes and Digestive and Kidney Diseases grants R01-DK098511 (to L. H. Salman and R. I. Vazquez-Padron) and R01-DK121227 (to R. I. Vazquez-Padron); National Heart, Lung, and Blood Institute grant K08-HL151747 (to L. Martinez); and the US Department of Veterans Affairs Merit Award IBX004658 (to R. I. Vazquez-Padron).

Author Contributions

J. C. Duque was responsible for data curation; J. C. Duque, L. Martinez, M. Tabbara, and R. I. Vazquez-Padron were responsible for methodology; J. C. Duque, L. Martinez, and R. I. Vazquez-Padron were responsible for formal analysis; L. Martinez was responsible for visualization; L. Martinez, L. H. Salman, M. Tabbara, and R. I. Vazquez-Padron conceptualized the study; L. Martinez and R. I. Vazquez-Padron were responsible for funding acquisition, provided supervision, and wrote the original draft; and all authors were responsible for investigation and reviewed and edited the manuscript.