Con: Vascular access surveillance in mature fistulas: is it worthwhile?

Abstract

The risks and benefits of hemodialysis arteriovenous (AV) access surveillance have been debated since the introduction of AV access surveillance techniques. The debate is fuelled by the lack of robust, high-quality evidence with consistent and comparable patient-important outcomes. Additionally, there is a lack of clarity regarding the diagnostic cut points for AV access stenosis using the various surveillance techniques, questions about the appropriateness of the ‘knee-jerk’ response to intervention on a stenosis >50% regardless of the presence of clinical indicators and whether the intervention results in desired patient-important outcomes. The physiology of the AV access is complex considering the different hemodynamics within an AV fistula, which vary by time of dialysis, location, size of vessels and location of the stenosis. The current evidence suggests that the use of AV access surveillance in an AV fistula does detect more stenosis compared with clinical monitoring alone and leads to an increased number of procedures. It remains uncertain if that leads to improved patient-important outcomes such as prolonged AV fistula patency. Vascular access is an essential component of hemodialysis and further study is needed to clarify this long-standing debate. There needs to be better distinction between the strategies of vascular access surveillance, clinical monitoring with clinical indictors and preemptive intervention and their respective impacts on patient-important outcomes. Randomized controlled studies must be conducted with defined indications for intervention, reproducible methods of intervention and clinically important AV fistula and patient outcomes. The current guidelines need to be challenged and revised to permit these necessary studies to be done.

The benefits and risks of arteriovenous (AV) access surveillance have been debated since the introduction of AV access surveillance [1–3]. In spite of the two meta-analyses [4, 5] and multiple randomized controlled and observational studies, the debate continues. The expansion of new evidence to inform this debate has been limited, in part by the integration of surveillance measures into care, based on previous vascular access guidelines recommendations [6]. The current Kidney Disease Outcomes Quality Initiative guidelines (2006) recommend AV access imaging and preemptive correction of stenosis to >50% when critical AV access flow (Qa) values are present, even in situations when the AV access is still able to provide adequate hemodialysis (preemptive correction of access stenosis) [6]. The guideline recommendations assume that a reduction in Qa identifies a treatable culprit stenosis before the AV access becomes dysfunctional and that preemptive correction of the stenosis will maintain AV access function, prevent thrombosis and prolong the longevity of the AV access as compared with deferred salvage.

There has been controversy over the interpretation of this recommendation based on previous and more current evidence [7]. Over time, the nephrology community has recognized that surveillance for stenosis in isolation of a clinical indicator is not clinically meaningful without also examining the effect of the subsequent intervention on clinically important outcomes, including AV access thrombosis, patency and intervention rates. Many studies have been underpowered to show either benefit or harm from surveillance strategies, and multiple factors (discussed below) contribute to the heterogeneity of the results and interpretations of the studies.

The debate is further confounded in that many of the studies were conducted in AV grafts, which now represent <10% of vascular access use in many countries. The studies in AV fistulas are limited, underpowered and with significant bias limiting the confidence in the reported effects [4, 5]. We now understand the importance of distinguishing the role of AV access surveillance separately in AV fistulas and AV grafts. Importantly, we also now understand the need to distinguish the role of AV access surveillance in AV access ‘with’ versus ‘without’ the presence of any clinical indication (signs or symptoms) that the AV access is at risk of thrombosis (‘clinical indicator’).

The fundamental principle for performing routine vascular access monitoring and surveillance is to detect and correct stenosis to minimize or avoid reduced dialysis clearance and reduce the rate of thrombosis and subsequent AV access loss. Clinical monitoring is used to prospectively detect a clinical indicator of significant stenosis that is at risk of thrombosis or results in inadequate delivery of dialysis. Table 1 lists clinical indicators of AV access stenosis; however, the validity of some of these indicators has also been scrutinized. Clinical monitoring strategies have been studied to detect clinically significant stenosis for both AV grafts [16, 17] and AV fistulas [8]. These strategies include physical examination (inspection, palpation and auscultation) of the vascular access to detect signs that suggest the presence of pathology. The review of routine laboratory studies including those for dialysis adequacy (e.g. urea reduction ratio or K_t/V, documented recirculation), consistent difficulties in cannulation or in achieving hemostasis after needle withdrawal and other clinical signs can supplement this monitoring. Clinical monitoring has been firmly established as a basic principle in the maintenance of the AV access [6].

Similarly a variety of surveillance methods that require specialized equipment and operators have been proposed for screening the AV access for early detection of stenosis before the AV access becomes associated with signs of dysfunction [16, 18–22]. Given the rationale that stenosis within the AV access reduces Qa and consequently may increase the risk for AV access thrombosis, different noninvasive methods of active surveillance of Qa have been proposed to determine if a functional AV access is at risk for dysfunction. These include Qa measurement [19, 20, 23] by a variety of techniques, including ultrasound dilution methods (UDMs), duplex Doppler ultrasound to measure Qa and visualize anatomic abnormalities [9] and dynamic and static venous pressure measurements [24]. Although these different techniques and methods are available for identifying vascular access dysfunction, the scientific evidence for the optimal methodology is lacking [4, 25].

To meet the criteria of a screening test, AV access surveillance methods should reliably and consistently detect culprit stenosis with reasonable accuracy and reproducibility. Stenosis within the AV access is thought to be reflected by reductions in Qa and alterations in AV access circuit pressures. However, flows and pressures are influenced by factors other than just the presence of stenosis, including the number, location and degree of the stenoses, AV access characteristics, cannulation technique and variations in the hemodynamics over the course of dialysis [e.g. timing, blood pressure, blood flow rates, equipment used (tubing or needle sizes)] [20, 26, 27]. The location of the stenosis can influence the diagnostic characteristics of the surveillance tool and an AV access can have multiple stenosis—not reliably detected with a single surveillance tool—especially if measured in isolation [10]. Additionally, the AV graft and the AV fistula have different processes for the development of stenosis, occurring at different rates, locations and with different hemodynamic consequences based on their configurations [28].

Other issues, such as defining and validating diagnostic thresholds, add to the challenge of establishing surveillance reliability and reproducibility. Diagnostic thresholds for Qa vary from 300 to 900 mL/min depending on the trade-off between low sensitivity but a high positive predictive value at a low Qa and high sensitivity but a low positive predictive value at high AV access Qa [1, 25–27]. The diagnostic cut point, that is, the level of Qa to intervene on, is even more uncertain when also considering if a subsequent intervention results in prolonging AV access patency or not [4, 20]. Finally, the diagnostic threshold must have clinical relevance. This is exemplified by differences in dialysis practice patterns around the world, whereby the standard blood flow rate (Qb) for AV fistula use in countries such as China and Japan is <300 mL/min. In such cases, even if the Qa is <300 mL/min, it would not be clinically relevant, as patients would continue to achieve adequate dialysis without recirculation (given all other factors being nonproblematic). This highlights the need to distinguish between any stenosis and clinically relevant stenosis that threatens the viability of the AV access to provide adequate dialysis.

The need to address whether surveillance methods can detect AV access stenosis and its incremental benefit on AV access patency beyond the clinical exam (monitoring) is important and highly relevant. There are significant costs to both the health care system and the patient with the current trend toward the use of surveillance [29]. Additionally, further understanding of the evidence will assist in focusing efforts for education, training and the attainment of expertise in the most relevant method for detecting AV access problems that threaten Qa, patency and longevity.

To inform and simplify this debate, we will discuss the effects of AV access surveillance in the AV fistula with and without a clinical indicator of stenosis. The use of surveillance, with subsequent preemptive intervention, in an AV access without a clinical indicator of stenosis has been referred to in the literature as preemptive surveillance or primary prevention surveillance. The use of surveillance, with subsequent preemptive intervention, in an AV access with a clinical indicator of stenosis has been referred to in the literature as secondary prevention surveillance. The terminology is confusing, as in the first situation, when there are no clinical indicators (i.e. no signs or symptoms of stenosis causing hemodynamic or clinical changes), ‘preemptive’ can refer to ‘before clinical indicators occur or before AV access thrombosis occurs’. In the second situation when there already are clinical indicators, ‘preemptive’ refers to ‘before AV access thrombosis occurs’. The respective primary and secondary prevention refers to the prevention of clinical indictors due to AV access stenosis. Thus, for clarity, we will simply discuss AV access surveillance, highlighting its use in the AV fistula with and without a clinical indicator of hemodynamically significant stenosis.

WHAT IS THE EVIDENCE COMPARING AV ACCESS CLINICAL MONITORING VERSUS SURVEILLANCE TO DETECT STENOSIS >50%?

In AV fistulas, the use of surveillance methods in addition to clinical monitoring appears to increase the rate of detection of AV fistula stenosis and the rate of AV fistula intervention [10, 30]. A single-center randomized controlled trial (RCT) compared AV fistulas that underwent clinical monitoring alone (clinical monitoring group) versus clinical monitoring plus Qa surveillance using UDM (UDM surveillance group) [30]. The clinical monitoring group was referred to angiography if stenosis was clinically suspected. Participants in the UDM surveillance group were referred to angiography if stenosis was clinically suspected or Qa was <500 mL/min at baseline or if Qa fell by >20% once Qa was <1000 mL/min. Participants in the UDM surveillance group were twice as likely to have stenosis detected compared with the clinical monitoring alone group {hazard ratio [HR] 2.27 [95% confidence interval (CI) 0.85–5.98]; P = 0.09}, with a trend for the stenosis to be detected earlier in the UDM surveillance group. The area under the curve demonstrated moderate prediction of >50% stenosis [HR 0.78 (95% CI 0.63–0.0.94); P = 0.006] in the UDM surveillance group. However, in a separate study, Tessitore et al. [10] compared clinical monitoring alone to multiple surveillance measures in AV fistula with identified stenosis >50%. The optimal tests for identifying an inflow stenosis included a combination of a positive finding on physical exam with a surveillance test. The investigators found that omitting the physical exam and relying on a screening program on the basis of Qa and the venous access pressure ratio (VAPR) was not advantageous because combining Qa <650 mL/min or VAPR >0.50 showed a significantly lower sensitivity and specificity than the combination of physical exam or Qa <650 mL/min (P < 0.001) for inflow stenosis. The best tests for identifying outflow stenosis were positive findings on physical exam and VAPR, with no difference between the two (accuracy 91 and 85%, sensitivity 75 and 81%, specificity 93 and 86%, respectively); physical exam was deemed preferable because it was more reproducible, easier to perform and applicable to all AV fistulas.

In summary, the clinical exam can be used to detect stenosis with reasonable accuracy compared with surveillance. In other studies, surveillance detected more stenosis than clinical monitoring but led to an increased number of procedures. Differences may be attributable to the difference in techniques or frequency of the exam. The clinically important question posed is, does surveillance followed by intervention in an AV fistula without a clinical indicator of stenosis result in improved patient-important outcomes such as prolonged AV access patency? Importantly, does this differ from intervention of stenosis detected by clinical exam alone?

DOES AV ACCESS SURVEILLANCE OF AN AV ACCESS WITHOUT A CLINICAL INDICATOR OF STENOSIS IMPROVE THE PATENCY OR SURVIVAL OF THE AV ACCESS?

Below we review the evidence for the routine use of surveillance to detect AV access stenosis in the absence of any clinical indicator of stenosis followed by intervention (angioplasty or surgical revision) if stenosis is >50%.

In the AV fistula, we concede that there remains some uncertainty regarding the role of ‘primary prophylaxis’ or preemptive repair of a stenosis detected by surveillance without a clinical indicator. Ravani et al. [4] updated a systematic review of RCTs examining the use of surveillance in both AV grafts and AV fistulas without a clinical indicator. The effects of any method of measuring Qa (UDM measurements with any dilution technique, Doppler, VAPR; that is, active surveillance) to identify and preemptively correct stenosis (using surgical or imaging-assisted endovascular procedures) in addition to or instead of a strategy of routine physical examination or monitoring of hemodialysis parameters (clinical monitoring) and subsequent interventions prompted by abnormalities found by clinical monitoring (clinical monitoring alone) was compared and evaluated. This study found borderline significant differences in treatment effects on risk of AV access loss by AV access type. In four studies of AV fistulas (310 participants), AV fistula loss was lower in those who had active surveillance compared with clinical monitoring alone [relative risk (RR) 0.50 (95% CI 0.29–0.86)]. The active surveillance group also had reduced thrombosis [RR 0.50 (95% CI 0.35–0.71)]. However, note that three of the four studies reporting AV fistula loss were conducted in the same center and reported by the same author [20, 21, 31]. Furthermore, these studies suffered from methodological issues. For example, in the control group, clinical monitoring was ‘unsystematic’ and the intervention was based on ‘access dysfunction (defined by “inability to sustain adequate dialysis”) or following thrombosis’, thus it was not true standardized clinical monitoring for primary prophylaxis—the undesirable outcomes (e.g. thrombosis) had occurred. The indication for intervention in the control group as ‘inability to sustain adequate dialysis’ could be for several reasons and did not specify the AV fistula as the cause. Furthermore, the open study design urges caution in interpreting the results. These cautionary results are echoed in a previous systematic review [5]; the estimates remain imprecise even with the addition of more recent studies. Ravani et al.'s [4] review did not include a recently published RCT comparing clinical monitoring alone to clinical monitoring and UDM every 3 months (n = 199 patients) [13]. After 1 year of follow-up, there was a significantly higher thrombosis rate in the clinical monitoring alone group compared with the surveillance group (0.099 thrombosis/patient versus 0.022 thrombosis/patient, respectively; P = 0.03). However, there was no significant difference in secondary patency rates with increased rates of interventions in the surveillance group. Reducing thrombosis but not prolonging AV access secondary patency may not be the most clinically relevant outcome for clinicians and patients. Indeed, patients have identified interventions as the most patient-important vascular access outcome [32]. Methodologically the study has a high risk of bias, based on the Grading of Recommendations Assessment, Development and Evaluations classification [33] being open label with an unclear randomization process, with HRs reported without reporting event rates. This study does not change the level of confidence (very low) in the estimates of the effect of intervening with angioplasty in an AV fistula with stenosis but no positive clinical indicator. In other words, there is poor evidence to support preemptive intervention based on surveillance alone without accompanying clinical indication in AV fistulas.

USE OF SURVEILLANCE FOR PREEMPTIVE INTERVENTION WHEN ASSOCIATED WITH A CLINICAL INDICATOR

There are three RCTs that examined this issue in AV fistula [20, 21, 30]. Tessitore et al. [21, 31] reports two similar studies in prevalent forearm AV fistula with angiogram-proven stenosis >50% and K_t/V >1.2 and recirculation as a clinical indication of stenosis (no clinical exam abnormality of the AV fistula). The intervention cohorts were assigned to preemptive correction of the stenosis and the control had deferred intervention until the Qa was reduced to >30% or 40% or recirculation was >5%. In both studies, the thrombosis rate was reduced, although the rate of secondary patency was not reported except that it was similar in preemptively treated and control AV fistulas. In a third study, Tessitore et al. [20] conducted a single open RCT of 58 prevalent AV fistula with K_t/V > 1.2 that had suspected stenosis at screening with a Qa <900 mL/min and/or physical signs or a static pressure ratio >0.5 and later had confirmed stenosis of >50%. The control group had deferred intervention until clinical signs or Qa <400 mL/min (n = 30). The intervention group had preemptive correction of the stenosis, which was repeated if the Qa was <750 mL/min or Qa decreased by >25% (n = 28). The intervention led to a relative risk of 0.47 (95% CI 0.17–1.15) for access failure (P = 0.090), 0.37 (95% CI 0.12–0.97) for thrombosis (P = 0.033) and 0.36 (95% CI 0.09–0.99) for access loss (P = 0.041).

Problems with this group of small studies include the use of Qa cut points that have not been validated and the control cohort had a nonstandardized approach to monitoring and intervention. The clinical indicators were rather nonspecific (recirculation, ‘cannulation difficulties’, etc.) and thus it was difficult to interpret whether a clinical indicator of AV access pathology was truly present and if the overall findings might be more supportive of surveillance and preemptive intervention in the presence of more clearly defined clinical indicators. Although results were statistically significant, the small numbers, potential biases and wide confidence intervals in all three studies lead one to interpret these unreliable estimates with caution.

In summary, there are several points to consider when comparing the outcomes of clinical monitoring to routine surveillance and preemptive intervention:

• The hemodynamics of the AV access are challenging. The use of a screening measure such as Qa has not been shown to reliably detect all stenosis or the culprit stenosis within the AV access.

• The indicator cut points for Qa have not been reliably determined for the best accuracy of clinically significant stenosis within the AV access.

• It is difficult to determine which stenotic lesion will progress to cause clinically problematic AV access function and/or loss. Not all stenoses >50% lead to clinical manifestations or loss and measurement of Qa is unable to discern this.

• The correlation between the presence of a clinical sign or symptom, the pathophysiological time frame of lesion development (e.g. early or late) and the degree of underlying stenosis is unclear and requires further study.

• Angioplasty as the form of intervention has been associated with a controlled injury model and is thought to lead to further injury, proliferation and stenosis.

• The need for angiographic intervention is associated with excess cancer risk attributable to radiation exposure, and recurrent interventions should be considered carefully, particularly in younger patients and in patients who will undergo kidney transplantation [34].

• Surveillance requires increased use of resources and increased patient and health system costs.

• Surveillance leads to an increased rate of intervention without confidence that it extends the life of the fistula as measured by secondary patency rates.

• Although AV access patency is a patient-important outcome, when it comes to surveillance, we have not brought patients into the conversation about uncertainty and decision making regarding AV access intervention.

• AV access intervention, for any reason, has been independently identified by patients worldwide as the most significant patient-important outcome of all AV access outcomes.

Moving forward to address the remaining uncertainties, we recommend RCTs with rigorous methodology and adequate power to delineate the role, if any, of current and future surveillance and intervention methods within a strategy to improve AV access patency particularly focused on AV fistulas. It appears there is little role for surveillance and preemptive intervention in AV grafts, and it is unlikely that these studies will be done, given their low use. There need to be studies validating the indicators of clinical monitoring and their associations with clinically significant stenosis, identification of subgroups of patients who may benefit from earlier endovascular or surgical interventions, the most appropriate timing of detection and interventions for various types of AV access lesions and standardized processes for both endovascular and surgical interventions. Importantly, we need to evaluate whether our strategies result in patient-important, clinically relevant and cost-effective vascular access and patient outcomes.

The current evidence does not give us the confidence to recommend the use of surveillance to improve AV access patency, although we welcome and look forward to further rigorous research in this area.

CONFLICT OF INTEREST STATEMENT

None declared.

(See related article by Tessitore and Poli. Pro: Vascular access surveillance in mature fistulas: is it worthwhile? Nephrol Dial Transplant 2019; 34: 1102--1106)

REFERENCES