Stenosis is the main cause of arteriovenous fistula (AVF) failure. It is still unclear whether surveillance based on vascular access blood flow (QA) enhances AVF function and longevity.
We conducted a three-year follow-up randomized, controlled, multicenter, open-label trial to compare QA-based surveillance and pre-emptive repair of subclinical stenosis with standard monitoring/surveillance techniques in prevalent mature AVFs. AVFs were randomized to either the control group (surveillance based on classic alarm criteria; n = 104) or to the QA group (QA measured quarterly using Doppler ultrasound [
The criteria for intervention in the QA group were: 25% reduction in QA, QA<500 mL/min or significant stenosis with hemodynamic repercussion (peak systolic velocity [PSV] more than 400 cm/sc or PSV pre-stenosis/stenosis higher than 3).
At the end of follow-up we observed a significant reduction in the thrombosis rate in the QA group (0.025 thrombosis/patient/year in the QA group vs. 0.086 thrombosis/patient/year in the control group [p = 0.007]). There was a significant improvement in the thrombosis-free patency rate (HR, 0.30; 95% CI, 0.11-0.82; p = 0.011) and in the secondary patency rate in the QA group (HR, 0.49; 95% CI, 0.26-0.93; p = 0.030), with no differences in the primary patency rate between the groups (HR, 0.98; 95% CI, 0.57-1.61; p = 0.935).
There was greater need for a central venous catheter and more hospitalizations associated with vascular access in the control group (p = 0.034/p = 0.029).
Total vascular access-related costs were higher in the control group (€227.194 vs. €133.807; p = 0.029).
QA-based surveillance combining Doppler ultrasound and ultrasound dilution reduces the frequency of thrombosis, is cost effective, and improves thrombosis free and secondary patency in autologous AVF.
J Vasc Access 2017; 18(4): 352 - 358
Article Type: ORIGINAL RESEARCH ARTICLE
AuthorsInés Aragoncillo, Soraya Abad, Silvia Caldés, Yésika Amézquita, Almudena Vega, Antonio Cirugeda, Cristina Moratilla, José Ibeas, Ramón Roca-Tey, Cristina Fernández, Nicolás Macías, Borja Quiroga, Ana Blanco, Maite Villaverde, Caridad Ruiz, Belén Martín, Asunción M. Ruiz, Jara Ampuero, Fernando de Alvaro, Juan M. López-Gómez
- • Accepted on 18/01/2017
- • Available online on 20/04/2017
- • Published in print on 14/07/2017
This article is available as full text PDF.
A correctly functioning vascular access (VA) is needed to ensure optimal dialysis. Native arteriovenous fistula (AVF) have proved to be superior to polytetrafluoroethylene (PTFE) grafts and central venous catheters (CVC) (1-2-3).
Stenosis is the main cause of thrombosis and AVF failure (4). When thrombosis occurs, it is usually necessary to place a CVC, thus increasing the cost associated with VA, the risk of infection, and hospital admissions (5). Since VA-associated admissions are one of the leading causes of hospitalization in patients on hemodialysis (HD), any intervention to detect significant stenosis and reduce the possibility of thrombosis will have a high clinical and economic impact (5, 6).
VA guidelines recommend monitoring and surveillance protocols in HD units (7-8-9). Monitoring protocols are based on physical examination; surveillance protocols require the use of specific instruments. Surveillance techniques include first-generation (classic) methods such as increasing dynamic or static venous pressure, decreasing Kt/V, and increasing recirculation. They also include second-generation methods, based mainly on measurement of access blood flow (QA), which could be the earliest indicator of AVF dysfunction (7-8-9-10).
Although numerous observational studies found benefits in protocol-based QA measurement in AVF, few randomized controlled trials (RCTs) have corroborated these benefits (11-12-13). Most RCTs were performed using the ultrasound dilution method (UDM) and had small samples and short follow-up periods. Three meta-analyses analyzing this topic were unable to find enough evidence to recommend the use of QA-based surveillance protocols (14-15-16). Another recent meta-analysis analyzing the impact of pre-emptive repair of subclinical stenosis in VA, found a significant effect on risk of thrombosis but a non-significant effect on risk of access loss (17).
The main purpose of this RCT was to assess whether measurement of QA based on the combination of two techniques, UDM and Doppler ultrasound (DU), could reduce thrombosis, increase thrombosis-free and secondary patency of prevalent AVFs, and reduce VA-associated costs.
Subjects and methods
We conducted an open-label, prospective, multicenter RCT with a three-year follow-up in parallel groups to compare QA-based surveillance techniques added to classic surveillance with the single use of classic surveillance methods in a prevalent HD population with prevalent native AVFs. The patients belonged to five HD units in Madrid, Spain. The RCT started in September 2012 and finished in September 2015.
To be included, patients had to be aged more than 18 years and undergoing HD with a functioning native AVF for at least three months.
The exclusion criteria were coagulopathy or hemoglobinopathy (any cause), hospitalization (any cause) in the previous month, or VA-related complications or dysfunction in the previous three months.
All patients gave their informed consent to participate, and the local ethics committee approved the trial. The clinical trial was performed using standards of Good Clinical Practice (CPMP/ICH/135/95). ClinicalTrials.gov Identifier: NCT02111655.
The sample size estimation was published in
As specified in the protocol, sample size was recalculated after one year of follow-up in an interim analysis after adding 13 patients to the trial.
Once patients gave their written informed consent and the eligibility criteria were verified, they were randomized to the control group (see below) or the QA group using a computerized randomization system based on 10-patient blocks.
Primary endpoint. Differences in thrombosis-free patency rates (or assisted primary patency) in AVF between the groups: the control group in which classic monitoring and surveillance techniques were applied, and the QA group in which DU and UDM were performed every three months in addition to the classic methods.
Secondary endpoints. Differences in primary patency rates (intervention-free access survival) and secondary (cumulative) patency rates (access survival until abandonment of the AVF) (19). A cost-efficacy analysis was performed.
The study design is shown in
Study design. DU = Doppler ultrasound; UDM = ultrasound dilution method.
The complete study design has been published previously in this journal (18).
Classic surveillance and monitoring were applied in the control group. In the QA group, QA was measured quarterly by combining two techniques, DU (Sonosite M-Turbo. Linear array 7.5 MHz) and UDM (Transonic HD-03), in addition to the classic methods.
Fistulography was performed in cases of classic alarm (18). The criteria for angioplasty was based on the anatomic repercussion of the stenotic area (>50% vessel lumen reduction). In the case of a juxta-anastomotic stenosis in radiocephalic fistula, surgical re-anastomosis was performed as first-line treatment after diagnostic fistulography.
DU and UDM were always performed in the same day; DU before the HD session and UDM during the first hour of the HD session.
The alarm criteria considered necessary for intervention of an AVF in the QA group were QA <500 mL/min, ≥25% decrease in QA compared with the previous measurement, or stenosis with anatomical significance (>50% reduction in the vessel lumen) and hemodynamic significance (peak systolic velocity [PSV] higher than 400 cm/s or PSV pre-stenosis/PSV stenosis ratio >3).
In cases of DU and/or UDM alarm, the approaches applied were fistulography (anatomic and hemodynamic significant stenosis and/or QA <500 mL/min and/or ≥25% decrease in QA), surgery (significant juxta-anastomotic stenosis or multiple stenosis not susceptible to angioplasty), or close clinical observation (stenosis with more than 50% reduction in vessel lumen but not hemodynamic significance or reversible cause of QA decrease as transitory hypotension or compressive hematoma in the AVF). In the case of classic alarm between two scheduled visits, QA was measured using DU and UDM in the same day at a nonscheduled visit, and a decision was taken based on QA-related findings.
Cost-efficacy analysis of resources
To calculate VA-related costs, public prices for services and health activity of the autonomous community of Madrid for the year 2014 were taken as a reference.
Qualitative variables are reported as percentages. Quantitative variables are expressed as mean and standard deviation (SD) or median and interquartile range (IQR), as appropriate. The comparison between baseline data was performed according to the CONSORT guidelines, taking into account the clinical relevance of each variable. The models were adjusted for confounders. The association between qualitative variables was evaluated using the chi-square test or Fisher’s exact test (when more than 25% of cells had frequencies of <5). Paired non-parametric data were compared using the Wilcoxon test; the Kruskal-Wallis test was used in the case of several independent samples. The primary outcome of the analysis was the thrombosis-free annual survival rate (or assisted patency rate) in both groups. Kaplan-Meier plots were used to estimate survival functions for the independent variables in the events (thrombosis, thrombosis-free patency rate, primary patency rate and secondary patency rate). Survival functions were compared between the subgroups using the Breslow test. A Cox regression model was fitted. For avoiding bias due to distribution, a bootstrap method was used for survival functions. The hazard ratio (HR) was presented with its 95% confidence interval (95% CI). Quantitative variables were analyzed for each of the independent variables categorized by the
A total of 212 patients were randomized, 107 in the QA group and 105 in the control group. Four patients in the QA group and one patient in the control group withdrew before the first visit because they had a hybrid AVF with a short PTFE segment; therefore, there were 103 patients in the QA group and 104 patients in the control group at baseline (
There were no differences in baseline characteristics between the groups (
Baseline characteristics of study population
|QA group (n = 103)||Control group (n = 104)|
|* Data expressed a mean (SD).|
|** Data expressed as median and interquartile range.|
|AVF = arteriovenous fistula; CVC = central venous catheter; QA = access blood flow; SD = standard deviation.|
|Age (y)*||63.2 (15.4)||66.5 (15.2)|
|Gender (male %)||74.5||68.3|
|Diabetes mellitus (%)||34.3||38.5|
|Antiplatelet therapy (%)||44.1||37.5|
|Double antiplatelet therapy (%)||3.9||6.7|
|Anticoagulation therapy (%)||17.6||13.5|
|Charlson index*||7.0 (2.8)||7.4 (2.8)|
|Body mass index*||31.0 (4.7)||25.8 (4.3)|
|AVF age (mo)**||41.4 (51.6)||38.7 (54.1)|
|Site of anastomosis (wrist/forearm) (%)||49.0||46.1|
|Previous CVC (%)||54.7||47.7|
|Previous AVF (%)||25.5||23.3|
|Previous surgeries with current AVF (%)||18.6||19.0|
|Previous thrombosis with current AVF (%)||2.1||5.8|
At the end of the study, the median follow-up was 25 (15-35) months in the QA group and 27 (11-35) months in the control group (p = 0.577).
More alarm criteria were recorded in the QA group (52 alarm criteria) than in the control group (37 classic alarm criteria) (p = 0.157).
In the QA group, 47% of the alarm criteria were associated with pathologic findings in DU or UDM, with no other abnormalities in the classic parameters. In the case of discrepancies between DU and UDM, the lower QA value was taken as a reference. Dynamic venous pressure was the most frequent classic alarm criterion in both groups, accounting for 44% of cases. All cases with increased venous pressure in the QA group were tested with DU and UDM, and no false positives due to increased venous pressure were found.
Most of the alarms recorded in the QA group occurred during the first year of follow-up (61% of the total), thus increasing the number of interventions in this group. However, over the three years of follow-up, there were similar number of interventions between the two groups (0,138 intervention/patient/year in the QA group compared with 0,141 intervention/patient/year in the control group p = 0.684), with similar primary patency (HR, 0.98; 95% CI, 0.57-1.61; p = 0.935) (
In the QA group, we found a significant reduction in the annual rate of thrombosis (0.025 thrombosis/patient/year compared with 0.086 in the control group; p = 0.007) and in the annual access loss rate (0.050 access loss/patient/year compared with 0.098 access loss/patient/year in the control group) with an increase in both thrombosis-free patency (HR, 0.30; 95% CI, 0.11-0.82; p = 0.011) (
Reasons for access abandonment were irreparable stenosis with new vascular access creation (7 patients in QA group and 4 patients in control group), thrombosis with unsuccessful thrombectomy (1 patient in QA group and 3 patients in control group) and thrombosis not considered susceptible to thrombectomy (3 patients in QA group and 14 patients in control group). Just one patient from QA group had a successful thrombectomy, the rest of the thrombosis required temporary or permanent CVC placement with access abandonment.
VA-related costs were recorded throughout the whole study period, taking into account the interventions and hospital admissions before and after the VA abandonment, until the end of follow-up (
Vascular access (VA)-related costs
|VA related costs||QA group||Control group||p|
|aFindings in fistulographies were distributed as follows:|
|bCosts in Euros (€).|
|CVC = central venous catheter.|
|Fistulography||n = 47a||n = 37a||0.313|
|€243 per fistulography||11,421b||8,991b|
|Angioplasty||n = 28||n = 22||0.662|
|€984 per angioplasty||27,552 b||21,648b|
|Surgery||n = 22||n = 21||0.977|
|€1429 per procedure||31,438b||30,009b|
|Days with CVC||n = 1432||n = 4367||0.044|
|€5.38 in urokinase per day with CVC||7704b||23,494b|
|Days of hospitalization associated with VA||n = 51||n = 131||0.027|
|VA-related hospitalization costs (€1092/day)||55,692b||143,052b|
|Total VA-related costs||€133,807||€227,194||0.029|
This RCT shows that QA-based surveillance reduces the frequency of thrombosis and increases thrombosis-free and secondary patency in prevalent autologous AVF. This approach is also cost effective.
Findings for the rate of thrombosis and thrombosis-free AVF survival are consistent with the RCT conducted by Tessitore et al (11-12-13). To our knowledge, this is the first RCT showing an improvement in secondary patency using these techniques.
All meta-analyses published to date have concluded that there is insufficient evidence for the generalized use of protocol-based QA measurement in VA. However, they agree that results are promising for autologous AVF and that new RCTs are needed to determine whether QA-based surveillance can reduce thrombosis and improve the secondary patency of this type of VA. Our results provide part of the clinical evidence needed to clarify the real effectiveness of QA-based surveillance.
Review articles written to interpret the results of meta-analyses stress that the insufficient evidence for surveillance of QA can be due to the heterogeneity of the RCTs included, which have different designs, objectives, and hypotheses (10, 20-21-22-23). In addition, sample size was small and follow-up limited in most cases. Furthermore, most RCTs analyzed the impact of QA surveillance on PTFE grafts, in which the development of neo-intimal hyperplasia occurred faster than in AVF and the decrease in QA acted as a late marker of dysfunction (20-21-22-23-24).
Another key aspect in most studies is the use of the UDM, which took QA <500-750 mL/min as a reference alarm, without using DU at the same time (17). Despite being observer-dependent, DU enables us to study the hemodynamic impact of stenosis (10, 25, 26). In addition, we now know that not all stenoses lead to thrombosis and that not all thromboses are caused by stenosis; therefore, we should not perform angioplasty in all cases of stenosis where vessel lumen is reduced by more than 50%, but only in those which combine a high risk of thrombosis with a significant drop in QA or significant hemodynamic consequences. In recent years, it seems more advisable to check the risk of thrombosis in autologous AVF, with QA <500 mL/min or a ≥25% decrease in QA and using a PSV pre-stenosis/stenosis ratio of >2-3 (10, 24-25-26-27). An unnecessary angioplasty in a stenosis with a low risk of thrombosis can cause significant intimal damage, thus favoring restenosis and worse prognosis of the AVF.
The incidence of thrombosis in AVF is low, especially according to the monitoring and surveillance used in an RCT. Some authors have considered a sample size as high as 850-1000 patients necessary to show an improvement in secondary patency (14, 17). Nevertheless, thanks to the combination of both techniques, DU and UDM, and to the use of very strict criteria for angioplasty, we were able to identify AVFs with a high risk of thrombosis and to show major benefits with a much lower sample size.
More AVFs were identified at high risk of thrombosis in the QA group with a higher pre-emptive interventionism in these patients, although the difference in angioplasties or surgeries did not reach statistical significance. The small difference between groups in interventionism can be explained because the use of DU gave us the opportunity to detect as well some false positive classic criteria alarms (for example higher venous pressure conditioned for a compressive hematoma or higher recirculation caused by a needle inside a pseudoaneurysm), so we were able to avoid some unnecessary procedures in the QA group.
Based on the results of this study, the ideal QA-based surveillance approach seems to be that combining UDM and DU quarterly, although, depending on the characteristics and individual risk of the AVF, UDM could probably be performed every three months and the study could be completed with DU when an alarm criterion is detected. In addition, when a classic alarm criterion appears, a nonscheduled UDM and DU should be performed. Routine DU should be performed at least once a year in HD patients with an AVF. We intend to perform a
In addition to the major clinical benefit for patients, the use of these techniques in HD units is completely justified, not only for their clinical benefits, but also for the significant reductions in the cost of VA-associated hospital admissions as well as in the need for CVCs and their complications.
Although new RCTs are needed to consolidate these results, our data provide sufficient evidence to asseverate that QA-based surveillance is a useful, powerful, and inexpensive tool that should form part of monitoring and surveillance protocols in HD units.
In conclusion, after three years of follow-up, the results of this RCT show that quarterly QA measurement using UDM and DU reduces the frequency of thrombosis, is cost effective, and improves thrombosis-free and secondary patency in autologous AVF.
- Aragoncillo, Inés [PubMed] [Google Scholar] 1, 2, * Corresponding Author (firstname.lastname@example.org)
- Abad, Soraya [PubMed] [Google Scholar] 1
- Caldés, Silvia [PubMed] [Google Scholar] 3
- Amézquita, Yésika [PubMed] [Google Scholar] 3
- Vega, Almudena [PubMed] [Google Scholar] 1
- Cirugeda, Antonio [PubMed] [Google Scholar] 3
- Moratilla, Cristina [PubMed] [Google Scholar] 2
- Ibeas, José [PubMed] [Google Scholar] 4
- Roca-Tey, Ramón [PubMed] [Google Scholar] 5
- Fernández, Cristina [PubMed] [Google Scholar] 6
- Macías, Nicolás [PubMed] [Google Scholar] 1
- Quiroga, Borja [PubMed] [Google Scholar] 7
- Blanco, Ana [PubMed] [Google Scholar] 8
- Villaverde, Maite [PubMed] [Google Scholar] 8
- Ruiz, Caridad [PubMed] [Google Scholar] 8
- Martín, Belén [PubMed] [Google Scholar] 9
- Ruiz, Asunción M. [PubMed] [Google Scholar] 9
- Ampuero, Jara [PubMed] [Google Scholar] 9
- de Alvaro, Fernando [PubMed] [Google Scholar] 3
- López-Gómez, Juan M. [PubMed] [Google Scholar] 1
Nephrology Unit, Hospital Gregorio Marañón, Madrid - Spain
Clínica Fuensanta, Hemodialysis Unit, Madrid - Spain
Nephrology Unit, Hospital Infanta Sofía, Madrid - Spain
Corporació Sanitària i Universitària Parc Taulí, Hospital de Sabadell, Barcelona - Spain
Nephrology Unit, Hospital de Mollet, Mollet del Vallès, Barcelona - Spain
Preventive Medicine Unit, Hospital Clínico, Madrid - Spain
Hospital Universitario La Princesa, Madrid - Spain
Clínica Dialcentro, Hemodialysis Unit, Madrid - Spain
Clínica Los Enebros, Hemodialysis Unit, Madrid - Spain