Moving Beyond the Assumed: Improving Fistula Success Rates

Vascular access is the lifeline for patients undergoing hemodialysis, and yet, our understanding of the basic biology of access success is, at best, rudimentary. Observational data show that arteriovenous fistulas (AVFs) are associated with the lowest hospitalization rates, mortality, and costs compared with arteriovenous grafts and central venous catheters. In 2003, the Centers for Medicare and Medicaid Services (CMS) launched the National Vascular Access Improvement Initiative to address a prevalent AVF rate that, at around 30%, lagged behind most industrialized countries in the world. This project subsequently evolved into the Fistula First Breakthrough Initiative and most recently, Fistula First/Catheter Last. Unfortunately, the AVF prevalence rate has plateaued around 68% compared with 62% in 2010,¹ despite the CMS’s quality incentive program for dialysis providers that financially rewards higher AVF rates.

A major barrier to increasing the number of AVFs is a primary failure rate of 30%–70% and a 1-year patency rate of 40%–70%. Many patients will undergo multiple surgical procedures in an attempt to establish vascular access with an AVF, which may not succeed.² Formulas to predict risk for AVF maturation failure, such as that proposed by Lok et al.,³ have not been effective in patient selection for AVF surgery, because even those patients with the lowest risk score have a 35% AVF failure to mature rate (compared with 71% in patients with the highest risk score). Preoperative venous mapping by ultrasound remains the standard by which vessels are selected for vascular access surgery. However, a large proportion of AVFs fails to mature even when venous mapping is used, leading to prolonged catheter dependence and risk for infection. Many of these patients require multiple AVF interventions that further decrease cumulative patency. The ability to save such patients the morbidity, infection risk, and ultimate disappointment of AVF nonmaturation by histologic risk scoring at the time of AVF creation surgery would be welcome, because patients with high histologic risk for nonmaturation could be converted to arteriovenous graft without having to endure multiple surgical attempts that do not ultimately salvage the AVF. Thus, understanding the biology of AVF success and failure is important to improving outcomes for patients undergoing hemodialysis.

Patients with CKD are prone to vascular pathology due to underlying risk factors, such as diabetes and hypertension, and uremia-induced abnormalities that result in both endothelial cell abnormalities and vascular smooth muscle cell proliferation. Given this, it had been assumed that a major factor in the etiology of AVF failure was preexisting venous intimal hyperplasia or calcification. In this issue of the Journal of the American Society of Nephrology (JASN), Alpers et al.⁴ and Cheung et al.⁵ report the results of the Hemodialysis Fistula Maturation Study and show that this hypothesis is of little consequence in AVF success. These studies describe the histologic changes in vein segments obtained at the time of AVF creation and the correlates of those changes with subsequent AVF maturation. Vein segments were obtained from 554 patients and examined with a variety of histologic techniques. In 57% of fully circumferential vein samples, neointima occupied >20% of the lumen; this was positively associated with patient age and inversely associated with black race. Expected consequences of the uremic state, such as markers of inflammation, cell proliferation, cell death, calcification, and neoangiogenesis, were found in a minority of veins examined.⁴ However, correlation of these histologic findings with subsequent AVF maturation was inconsistent. Preexisting intimal hyperplasia, expressed per 10% increase in hyperplasia index (range of 0%–100%) was modestly associated with lower AVF blood flow rate at 6 weeks, but there was no correlation with stenosis on 6-week ultrasound or with maturation failure without procedural assistance. AVF stenosis on 6-week ultrasound was associated with maturation failure but only after controlling for case mix factors, dialysis status, and AVF location.⁵ Unfortunately and predictably, overall stenosis prevalence did increase over time (14% at 1 day, 28% at 2 weeks, and 30% at 6 weeks).⁵

Possible explanations for this counterintuitive lack of correlation between hyperplasia index and subsequent maturation include unrepresentative vein sampling and the lack of sensitivity of postoperative ultrasound to detect clinically meaningful changes in peak flow velocity ratio. Another more likely explanation of these findings is that preexisting venous hyperplasia is not as important in the pathogenesis of access failure as previously thought. These results, although somewhat surprising, should be interpreted as promising news, because there is little hope for interventions to fix preexisting lesions. Instead, this opens the opportunity for prospective interventions at the time of AVF creation to improve success.

Progressive stenosis of de novo lesions may result from mechanical forces on the venous wall after AVF creation. The generation of such forces is likely due to surgical technique rather than underlying abnormalities in vein histology. The importance of mechanical forces may also explain the higher rate of distal stenosis at 6 weeks of upper arm AVFs (19.5%) versus lower arm AVFs (12.3%), because mechanical forces in the cephalic arch are greater in patients with upper arm AVFs. However, juxta-anastomotic stenoses at 6 weeks were more common in patients with forearm AVFs (20.8%) than upper arm AVFs (14.2%).⁵ This is likely related to the smaller caliber of the inflow artery in forearm AVFs but may again be related to surgical technique. Unfortunately, although the importance of hemodynamic factors in fistula (and graft) failures is appreciated, these factors are poorly understood, and they are hard to predict and correct in vivo.

Advances in computational fluid dynamics can help predict anastomotic angles that maximize flow but minimize shear stress distribution, the latter thought to be the driving force behind reactive intimal hyperplasia. For example, Lee et al.⁶ recently used fluid dynamics using three-dimensional computational techniques to compare wrist AVF for end to side radiocephalic anastomosis. Three AVF conditions (anastomosis angles of 45°, 90°, and 135°; vein diameter; and anastomosis area) were simulated by examining blood flow characteristics, cardiac cycle, and vessel distention. The results predicted that angles >90° would generate less shear stress. Retrospective review of patients supported these simulations, although the results did not reach significance. Such an analytic approach to optimize surgical techniques with subsequent prospective longitudinal trials in patients is needed if there are to be improvements in primary success rates. Unfortunately, a recent systematic review by Cunnane et al.⁷ identified inconsistencies in cross-sectional studies, with both low shear (presumably from recirculating flow) and increase in wall shear stress leading to intimal hyperplasia and inward remodeling. The review also highlighted the paucity of longitudinal studies. Given the cost of AVF failure to both patients and the health care system, the lack of such studies on the effect of surgical techniques is alarming.

Another limitation to AVF research is the difficulty in assessing patency with ultrasound. In this issue of the JASN, the study by Cheung et al.⁵ found that 37% of participants with stenosis on the day 1 ultrasound did not have stenosis at week 2 and 36% with stenosis at week 2 did not have stenosis at week 6. The authors suggested that this might be due to technical variability in ultrasound measurements, expansion of the lumen resulting in decreased ratios, or regression of hyperplasia. The latter seems unlikely, and taken together, the results indicate the need for optimizing assessment of AVFs to assess hyperplastic changes. This will be critical both for enhancing our ability to clinically predict who will have a successful AVF outcome and as a validated end point for use in interventional trials.

Although data support that abnormal sheer stress is a cause of intimal hyperplasia, the mechanisms involved are poorly understood. Although many mechanisms have been postulated and recently reviewed,^6,8 relatively few have been tested. Recent experimental studies have shown that toll-like receptor 4 gene silencing,⁹ administration of microRNA221,¹⁰ vitamin K2,¹¹ and upregulation of heme oxygenase-1¹² all inhibit hyperplasia, offering a glimpse of potential future treatments. However, review of studies listed on Clinicaltrials.gov shows that the majority of human research is investigating novel devices for grafts and catheters as opposed to biologic approaches. Understanding the biology of changes at the time of anastomosis and those that occur due to cannulation offers opportunities for future interventions that can be applied intraoperatively or via needles at the end of a dialysis treatment. A new therapy vonapanitase (type 1 pancreatic elastase, Proteon Therapeutics) applied at the time of fistula creation and now in phase 3 trials (NCT02414841) shows some promise⁹ and is an example of this approach. The direct application of biologics at the time of AVF creation and periodically thereafter is not only feasible but offers an opportunity to improve other less accessible sites, such as cardiac bypass vein grafting and kidney transplant vessel anastomoses. Importantly, the studies in this issue of the JASN allow us to move beyond blaming preexisting vascular changes as a cause of AVF failure to research that can facilitate biologic therapies to ensure AVF success.

Disclosures

None.