Vascularization of the arteriovenous fistula wall and association with maturation outcomes

Abstract

Background and objectives:

The venous vasa vasorum is the mesh of microvessels that provide oxygen and nutrients to the walls of large veins. Whether changes to the vasa vasorum have any effects on human arteriovenous fistula outcomes remains undetermined. In this study, we challenged the hypothesis that inadequate vascularization of the arteriovenous fistula wall is associated with maturation failure.

Design, setting, participants, and measurements:

This case–control pilot study includes pre-access veins and arteriovenous fistula venous samples (i.e. tissue pairs) from 30 patients undergoing two-stage arteriovenous fistula creation (15 matured and 15 failed to mature). Using anti-CD31 immunohistochemistry, we quantified vasa vasorum density and luminal area (vasa vasorum area) in the intima, media, and adventitia of pre-access veins and fistulas. We evaluated the association of pre-existing and postoperative arteriovenous fistula vascularization with maturation failure and with postoperative morphometry.

Results:

Vascularization of veins and arteriovenous fistulas was predominantly observed in the outer media and adventitia. Only the size of the microvasculature (vasa vasorum area), but not the number of vessels (vasa vasorum density), increased after arteriovenous fistula creation in the adventitia (median vasa vasorum area 1366 μm²/mm² (interquartile range 495–2582) in veins versus 3077 μm²/mm² (1812–5323) in arteriovenous fistulas, p < 0.001), while no changes were observed in the intima and media. Postoperative intimal thickness correlated with lower vascularization of the media (r 0.53, p = 0.003 for vasa vasorum density and r 0.37, p = 0.045 for vasa vasorum area). However, there were no significant differences in pre-existing, postoperative, or longitudinal change in vascularization between arteriovenous fistulas with distinct maturation outcomes.

Conclusion:

The lack of change in intimal and medial vascularization after arteriovenous fistula creation argues against higher oxygen demand in the inner walls of the fistula during the vein to arteriovenous fistula transformation. Postoperative intimal hyperplasia in the arteriovenous fistula wall appears to thrive under hypoxic conditions. Vasa vasorum density and area by themselves are not predictive of maturation outcomes.

Introduction

Knowledge of the adaptive response of the venous wall to surgical trauma and supra-arterial circulation remains scarce, which limits our ability to design effective therapies for the early failure of the preferred hemodialysis access. There is a belief, among many, that inadequate pre-existing and post-surgical vascularization of the arteriovenous fistula (AVF) causes stenosis and decreased vascular compliance, due to poor nourishment and oxygenation of the fistula.^1–4 However, the premises of this notion are built upon interpretation of other vascular pathologies and experimental AVF models, while observational clinical studies to support this hypothesis are not available.

The vasa vasorum (Latin: “vessels of the vessels”) is the network of small arteries and veins that irrigates the walls of large vessels.^5,6 These microvessels penetrate the adventitia and provide nourishment to the media when the wall is thicker than 0.5 mm or 29 lamellae.^7,8 Therefore, they have a fundamental function in intramural oxygenation when luminal O₂ diffusion is not sufficient and in the exchange of nutrients and waste across the inner wall.^5,9 The vasa vasorum extends deeper into the wall of muscular veins than in arteries, reflecting a higher demand for intramural irrigation in the former.^10,11

The vasa vasorum plays different roles in physiological vasoactive responses compared to vascular disease. Damage to the vasa vasorum in both arteries and veins leads to vasospasm, loss of distensibility, and decreased patency.^10,12–15 This is why, one of the main surgical considerations during AVF placement is to inflict minimal damage to the adventitia.^16,17 Paradoxically, expansion of the vasa vasorum has been associated with idiopathic pulmonary fibrosis,¹⁸ pulmonary hypertension,^19,20 and atherosclerosis.^21–23 Hemodynamic changes in the lumen of the host vessels also generate a dynamic response in their intramural microvasculature. The vasa vasorum normally dilates when the partial pressure of oxygen (PO₂) in the vessel lumen is low.¹⁰ Therefore, it is not clear how the increased luminal PO₂ after arteriovenous anastomosis in AVFs affects vasa vasorum reactivity and host vessel irrigation.

In this work, we evaluated the longitudinal changes in wall vascularization (i.e. vasa vasorum density (VVD) and vasa vasorum area (VVA)) during early remodeling of AVFs that matured and failed, using a paired analysis of vein and AVF cross sections from patients undergoing two-stage AVF creation. We also explored the associations of pre-existing and postoperative vascularization with postoperative morphometry (intimal hyperplasia (IH) and medial fibrosis) of the fistula.

Materials and methods

Study design

This case–control study included 30 chronic kidney disease patients ⩾ 21 years of age who underwent two-stage upper arm AVF creation surgeries at Jackson Memorial Hospital and University of Miami Hospital (Figure 1 and Supplemental Figure 1). The cohort included 28 brachiobasilic AVFs, 1 brachiocephalic AVF, and 1 aberrant radio-ulnar fistula. The AVF matured successfully in 15 of the patients and failed to mature in the other half (Table 1). Failure was secondary to venous stenosis in the juxta-anastomotic area in 14 out of 15 patients (93.3%) and distal to the anastomosis in one individual (6.7%). All patients underwent both first-stage and second-stage procedures and were consented before each of the surgeries. Anatomic AVF non-maturation was defined retrospectively as an AVF that never achieved a cross-sectional luminal diameter of ⩾6 mm, as determined intraoperatively at the time of the second surgery using intravascular probes. AVFs that matured underwent a transposition at the time of the second surgery (Supplemental Figure 1), while AVFs that failed underwent a salvage procedure consisting of a shorter transposition, graft extension, or ligation (Figure 1). Participants were enrolled continuously from August 2013 to May 2017 as part of an actively enrolling biorepository. Patient selection from the biorepository took into consideration: (1) an adventitia thickness of >200 μm in paraffin-embedded cross sections of the native vein and second-stage AVF tissues from each patient, (2) equivalent representation of demographic characteristics (sex, ethnicity, and age range), diabetes, and anti-platelet agent use in each of the outcome groups, and (3) similar time interval from first-stage to second-stage surgery between the groups (Table 1). Twenty-two of the patients analyzed in the study (13 with AVF maturation and 9 with failure) were included in a previous publication.²⁴ The study was performed according to the ethical principles of the Declaration of Helsinki and regulatory requirements at both institutions. The ethics committee and institutional review board at the University of Miami approved the study.

Figure 1.

Flow chart of study design, surgical procedures, and sample collection. Vein and juxta-anastomotic AVF samples from the same individuals were collected at the time of first-stage and second-stage surgeries or salvage procedures, respectively. Patients with successful maturation were compared to those with anatomic maturation failure.

Table 1.

Baseline characteristics of the study cohort.

	All patients	Matured	Failed
Number of patients	30	15	15
Demographics
Age (years, mean ± SD)	58.2 ± 12.1	58.7 ± 11.4	57.7 ± 13.2
Female (%)	15 (50)	7 (47)	8 (53)
Non-Hispanic black (%)	17 (57)	8 (53)	9 (60)
Comorbidities (%)
Hypertension	27 (90)	13 (87)	14 (93)
Diabetes	15 (50)	6 (40)	9 (60)
Drug therapy (%)
Antiplatelet agents	16 (53)	7 (57)	9 (60)
Statin	16 (53)	10 (67)	6 (40)
ACE-I/ARB	8 (27)	4 (27)	4 (27)
Time between surgeriesa (days, median [IQR])	81 [66–112]	70 [56–84]	95 [70–134]

SD: standard deviation; ACE-I: angiotensin-converting enzyme inhibitor; ARB: angiotensin receptor blocker; IQR: interquartile range.

Categorical variables are expressed as number (%), while numerical variables are expressed as mean ± SD or median [IQR]. Patients’ ethnicity is Hispanic or non-Hispanic black.

^aTime interval between first-stage and second-stage surgeries.

A single surgeon (M.T.) performed all surgical procedures, using preoperative vascular mapping of the upper extremities to plan the AVF.²⁵ We followed the order of AVF preference recommended by the National Kidney Foundation/Kidney Disease Outcomes Quality Initiative.²⁶ Veins that were not visually sclerotic and had a diameter of ⩾3.5 mm were used for AVF creation. The surgeon collected vascular samples at two time points: a sample of the vein used to create the arteriovenous anastomosis during the first-stage surgery and a sample of the juxta-anastomotic area of the AVF during the second-stage procedure. Patients’ demographics, comorbidities, and medications taken within 6 months of AVF creation were obtained from the medical record.

Specimen collection and tissue processing

The cross-sectional vein and AVF specimens consisted of a 1- to 5-mm sample collected at the site of transection during fistula creation (first-stage surgery) or transposition (second-stage surgery). The AVF biopsy was taken at approximately 2 cm from the initial anastomosis, except in one patient with stenosis distal to the anastomosis in whom the stenotic lesion was resected. Samples were submerged in 10% neutral formalin (Sigma-Aldrich, St. Louis, MO, USA) immediately after collection, deidentified with a numerical code in the research laboratory, and fixed for 24 h for paraffin embedding and sectioning.

Immunohistochemistry and morphometric analysis

Cross-sections encompassing the entire circumference of the vein were stained with an anti-CD31 antibody (AbD Serotec, Kidlington, UK; catalog no. MCA1738, 1:500 dilution) for identification of microvessels in vein and AVFs. VVD was calculated in each vascular layer (intima, media, and adventitia) and in the total wall as the number of CD31-positive microvessels per layer (and total) divided by the corresponding wall area. Similarly, VVA was quantified in each layer and in the total wall as the luminal area of the microvessels per layer (and total) divided by the corresponding wall area. The adventitia area was delimited to within 200 μm of the external elastic lamina.

AVF tissue sections were stained with Masson’s trichrome for postoperative morphometric analysis. The intima and media areas were delineated to calculate medial fibrosis (% area of collagen) and the intima/media (I/M) area ratio. The minimal and maximal intimal thickness was defined as the minimal and maximal lineal distance from the internal elastic lamina to the endothelium, respectively.²⁷ Similarly, the intima-to-media thickness was measured from the external elastic lamina to the endothelium. Full digital images were acquired with a VisionTek digital microscope (Sakura Finetek, Torrance, CA, USA). Morphometric quantifications were performed using ImageJ (National Institutes of Health, Bethesda, MD, USA).

Statistical analyses

Statistical analyses were performed using GraphPad Prism 5.00 (La Jolla, CA, USA). Normally distributed data were compared using Student’s t-test and expressed as mean ± standard deviation (SD); otherwise, data were compared using the Mann–Whitney test (or Wilcoxon signed-rank test for paired analyses) and expressed as median and interquartile range (IQR). Comparisons between group frequencies were performed using Fisher’s exact test. Associations between baseline clinical covariates and vascularization were evaluated using multivariate general linear regression models adjusted for age, sex, ethnicity, comorbidities, and medications. Similarly, the association between pre-existing to postoperative change in vascularization and maturation outcomes was assessed using a general linear regression model adjusting for the pre-existing values. Results were considered significant when p < 0.05.

Results

Characteristics of the study cohort

The association of pre-existing and postoperative AVF vascularization with maturation outcomes was evaluated in 30 two-stage AVF patients, in whom half of the AVFs matured successfully while the rest suffered maturation failure (Figure 1). Both the maturation and failure subgroups were similar in terms of demographics, comorbidities, and medications taken at the time of access creation (Table 1). Twenty-eight out of 30 AVFs (93.3%) were created using the brachiobasilic configuration, while the remaining two used the brachiocephalic and aberrant radio-ulnar combinations. All AVFs were created in the left upper extremity. Higher total VVD and VVA in the pre-access veins were significantly associated with angiotensin-converting enzyme inhibitor (ACE-I)/angiotensin receptor blocker (ARB) use, but not in postoperative AVF samples (Supplemental Table I).

Vascularization of pre-access veins and maturation outcomes

Microvessels were predominantly found in the outer media and the adventitia layers of pre-access veins, with only 5 out of 30 veins (16.7%) showing 1–3 microvessels in the intima (Table 2). The median VVD in the overall wall was 5.3 microvessels/mm² (IQR 1.1–9.6) in veins that matured, compared to 7.4 (IQR 1.8–18.3) in veins that failed, with no significant differences between these groups (Table 2 and Supplemental Figure 2). There were also no significant differences in the VVD of individual wall layers between veins with distinct maturation outcomes (Table 2).

Table 2.

Pre–existing and postoperative VVD in fistulas that matured and failed.

	VVD (count/mm2)		p-value
	Matured	Failed
Pre-existing—veins
Intima	0.0 [0.0–0.0]	0.0 [0.0–0.0]	0.52
Media	4.8 [0.3–9.4]	6.8 [0.0–17.5]	0.85
Adventitia	6.4 [1.7–11.4]	8.4 [2.0–23.5]	0.62
Total	5.3 [1.1–9.6]	7.4 [1.8–18.3]	0.65
Postoperative—AVFs
Intima	0.0 [0.0–0.2]	0.0 [0.0–0.0]	0.54
Media	3.0 [0.8–10.7]	6.3 [1.7–11.8]	0.41
Adventitia	6.4 [5.3–14.1]	8.0 [5.7–11.4]	0.74
Total	3.5 [1.8–7.9]	5.0 [2.3–10.3]	0.25

VVD: vasa vasorum density; AVF: arteriovenous fistula.

VVD was calculated as the number of microvessels per layer (and total) over the area of the corresponding wall layer or of the entire wall (total). Values are presented as median [interquartile range].

Similarly, the median VVA in the overall wall was 807 μm²/mm² (IQR 229–1409) in veins that matured versus 1034 (506–1466) in veins that failed. There were no significant differences between these values (Table 3 and Supplemental Figure 2), nor between the VVA of individual wall layers in veins from the two outcome groups (Table 3).

Table 3.

Pre-existing and postoperative VVA in fistulas that matured and failed.

	VVA (μm2/mm2)		p-value
	Matured	Failed
Pre-existing—veins
Intima	0 [0–0]	0 [0–0]	0.52
Media	287 [6–1283]	649 [0–1574]	0.92
Adventitia	1046 [492–2982]	1756 [628–2449]	0.46
Total	807 [229–1409]	1034 [506–1466]	0.65
Postoperative—AVFs
Intima	0 [0–29]	0 [0–0]	0.42
Media	1067 [218–1927]	666 [417–1602]	0.82
Adventitia	3065 [2347–6628]	3089 [1669–4869]	0.85
Total	1403 [816–2512]	1229 [766–2148]	0.77

VVA: vasa vasorum area; AVF: arteriovenous fistula.

VVA was calculated as the luminal area of the microvessels per layer (and total) over the area of the corresponding wall layer or of the entire wall (total). Values are presented as median [interquartile range].

Postoperative vascularization and maturation outcomes.

Similar to our observations in native veins, microvessels were predominantly observed in the outer media and the adventitia of AVFs (Table 2). The intima was vascularized in 8 out of 30 AVFs (27%), with 1–5 microvessels observed in 6 of these samples and 21–39 in the remaining 2. Figure 2 shows representative CD31 stainings of the three vascular layers in AVFs. The VVD did not change significantly between pre-access veins and AVFs in paired analyses (Figure 3). In addition, neither the absolute VVD values in AVFs (Table 2 and Supplemental Figure 2) nor the change from pre-existing to postoperative VVD in paired tissues (Supplemental Table II) was associated with maturation outcomes.

Figure 2.

Intramural vascularization of the three vascular layers in AVFs. Representative CD31 stainings of the intima (a), media (b), and adventitia (c) layers in AVFs. Arrows indicate the location of vasa vasorum microvessels.

Figure 3.

Longitudinal change in vascularization during the vein to AVF transformation. Pairwise comparisons of vasa vasorum density (a) and area (b) in the three vascular layers of veins and AVFs from the same individuals.

In contrast to the similar VVD values between veins and AVFs, VVA increased significantly in the adventitia after access creation but not in the intima or media (Figure 3). This postoperative increase in vascularization occurred to a similar degree in AVFs that matured compared to those that failed, since neither the absolute postoperative VVA values (Table 3 and Supplemental Figure 2) nor the pairwise change in VVA from the vein to the fistula (Supplemental Table II) was associated with maturation outcomes.

Association of vascularization with postoperative morphometry.

Postoperative IH, expressed as maximal intimal thickness, ranged from 204 to 1463 μm in AVFs that matured (mean ± SD: 799.7 ± 349.0 μm), compared to 57–1541 μm in those that failed (724.7 ± 421.7 μm), with no significant differences between the groups (p = 0.60). Similarly, there were no significant differences in I/M ratio or intima-to-media thickness between the two maturation outcomes (Supplemental Table III). Postoperative medial fibrosis ranged from 28.55% to 78.40% in fistulas that matured (mean ± SD: 44.22% ± 13.21%), compared to 39.72%–72.53% in AVFs that failed (54.82% ± 9.80%) and was significantly higher in the latter (p = 0.02, Supplemental Table III).

The maximal intimal thickness in AVFs demonstrated a negative correlation with postoperative vascularization of the media (both VVD and VVA), but not the adventitia (Figure 4 and Supplemental Table IV). In contrast, we found no significant correlations between postoperative medial fibrosis and pre-existing, postoperative, or pairwise change in vascularization (Supplemental Table IV).

Figure 4.

Correlation between postoperative vascularization of the media layer and intimal hyperplasia. Correlation analyses between postoperative medial vasa vasorum density (a) and area (b) and maximal intimal thickness in AVF cross sections.

Discussion

Successful maturation of the AVF involves a significant increase in wall thickness and area,^28,29 presumably increasing the initial requirements of the vein for intramural oxygenation and nourishment. In this study, we challenged the hypothesis that inadequate vascularization of the AVF wall is associated with maturation failure. We found that (1) the size of microvessels increases after AVF creation without an evident angiogenic process, (2) postoperative IH correlates with lower vascularization of the media, and (3) neither the pre-access, postoperative nor longitudinal change in the density or size of the microvessels was associated with maturation outcomes.

There is a critical wall thickness beyond which vasa vasorum irrigation is required, which differs widely between vessels.^7,8 This critical value is thought to depend on intraluminal PO₂ (which determines oxygen diffusion) and on the structure and function of the vessels (which establishes oxygen demand).^11,30 The critical thickness of the aorta is 0.5 mm, but much lower (0.35 mm) in coronary arteries.¹¹ The critical wall thickness of veins, including those used for AVF creation, has not been systematically investigated, but it is well accepted that it is lower than in arteries.^10,11 A recent analysis of 48 pre-access veins (mostly cephalic) reported the presence of intimal microvessels in 52% of the patients, with a mean of 20 microvessels per intima in those positive samples.³¹ This is significantly higher than what we observed in basilic veins and suggests differences in patient populations or anatomy. Comparisons of both AVF patient cohorts would require normalization of microvessels density by area in Alpers et al.’s study. In this study, the intima-to-media thickness of a remodeled AVF ranged from 0.05 to 2.24 mm (including the minimal and maximal values recorded for all AVFs), and intima thickness ranged from 0.01 to 1.54 mm. However, the intraluminal oxygen content in AVFs is high secondary to the close arterial feeding,³² which can significantly decrease the requirements for irrigation.

The increase in vasa vasorum size but not density after fistula creation argues against the importance of angiogenesis during maturation. Instead, the larger microvessels in the adventitia of AVFs compared to their vein pairs are likely due to microvasculature remodeling and/or chronic vasa vasorum vasodilation. Interestingly, the location of vasa vasorum in AVF cross sections is similar to that in veins, that is, predominantly found in the outer media and adventitia, and with little penetration to the inner media and the intima. The fact that the latter two regions remained mostly avascular during early remodeling and that VVA did not increase in the outer media of AVFs (only in the adventitia) suggests that the increase in vasa vasorum caliber in the adventitia was not in response to hypoxic stimuli. To illustrate the opposite scenario, significant irrigation of the hyperplastic intima, entire media, and thrombi is observed in patients with recurrent varices or vein thrombophlebitis as a pathophysiological reaction to hypoxia.³³ In the case of AVFs, we propose that the increased luminal area of the adventitial microvasculature occurred as a result of the higher flow and intraluminal pressure in the artery where these microvessels originate.³⁴

Importantly, we found a significant correlation between lower vascularization of the media in AVFs (both VVD and VVA) and maximal intimal thickness. These results suggest that IH thrives under hypoxic conditions, as observed in femoral arteries³⁵ and contrasts the proposed association between AVF vascularization and IH in experimental arteriovenous grafts and fistulas.^1–3,36 In experimental AVFs, inadequate oxygenation was associated with IH and activation of hypoxia-associated transcription factors (e.g. hypoxia-inducible factor 1-alpha (HIF-1α)),^1–3 which in turn is thought to trigger vascular endothelial growth factor (VEGF)-mediated neoangiogenesis.⁴ In agreement with our results, only a minority of veins in the Hemodialysis Fistula Maturation Study presented foci of neoangiogenesis in the intima,³¹ indicating that the mechanisms of neointima formation in human pre-access veins and AVFs differ from those in experimental models. The known differences between experimental and human AVFs in terms of anatomical location, vessel size, surgical trauma, hemodynamics, and fistula configuration may account for these discrepancies. Of note, our results do not rule out a potential role of HIF-1α and downstream mediators such as VEGF in AVF remodeling. Instead, they suggest that the effects that such mediators may have on fistula outcomes are based on local or circulatory differences in their concentration and signaling, but not triggered by differences in tissue ischemia or vascularization. Given that veins are normally exposed to low PO₂, it is possible that the stimulus for HIF-1α activation is not hypoxia-related, but due to the oxidative stress³⁷ to which the vein is exposed after anastomosis. One can even postulate that the vein is not well adapted to this high oxygen content, and a thick intima may be protecting medial smooth muscle cells (SMCs) from the oxidative stress associated with high PO₂. Interestingly, we did not find any associations between diabetes and pre-access vein or AVF vascularization. However, given the known relationship between diabetes and deficient microcirculation, this is a finding that would need to be confirmed with a larger cohort of patients.

Postoperative medial fibrosis was recently associated with AVF failure.²⁴ Differences in intramural vascularization can contribute to fibrosis is a number of ways: (1) through the infiltration of pro-fibrotic factors and inflammatory cells and (2) through deficient or excessive oxygenation leading to vascular cell apoptosis or necrosis. However, the fact that wall vascularization does not influence maturation outcomes nor medial fibrosis in this study suggests that the main mediators of AVF fibrosis are local. In support of local factors associated with maturation failure, a recent transcriptomics analysis of pre-access veins that failed versus those that matured demonstrated that the inflammatory phenotype of venous SMCs contributes to AVF failure.³⁸ Therefore, the answer to maladaptive remodeling of the fistula after access creation may lie in the local cellular response to peri- or postoperative stimuli.

In conclusion, we found no associations between pre-existing, postoperative, or longitudinal change in vascularization and maturation outcomes. This further suggests that, in the absence of other aggravating factors, intramural vascularization of the pre-access vein or AVF is not a crucial element to target to optimize the maturation process. The main limitations of the study are the small number of patients and the analysis of only upper arm two-stage AVFs, which decreases the generalizability of our findings. In addition, our findings may not accurately reflect all vascular changes occurring in the vein distal to the anastomosis or in the arterial component of the vascular access. The tissue biopsy may not include the full stenotic lesion responsible for failure due to limitations imposed by the length of the fistula. Despite these shortcomings, the study increases our understanding of the nature and role of intramural vascularization in early AVF remodeling and provides clinically relevant human-based data that can further support our search of preventive therapies for AVF complications.