Arteriovenous Fistula Flow Dysfunction Surveillance: Early Detection Using Pulse Radar Sensor and Machine Learning Classification

Vascular Access (VA) is often referred to as the “Achilles heel” for a Hemodialysis (HD)-dependent patient. Both the patent and sufficient VA provide adequacy for performing dialysis and reducing dialysis-related complications, while on the contrary, insufficient VA is the main reason for recurrent hospitalizations, high morbidity, and high mortality in HD patients. A non-invasive Vascular Wall Motion (VWM) monitoring system, made up of a pulse radar sensor and Support Vector Machine (SVM) classification algorithm, has been developed to detect access flow dysfunction in Arteriovenous Fistula (AVF). The harmonic ratios derived from the Fast Fourier Transform (FFT) spectrum-based signal processing technique were employed as the input features for the SVM classifier. The result of a pilot clinical trial showed that a more accurate prediction of AVF flow dysfunction could be achieved by the VWM monitor as compared with the Ultrasound Dilution (UD) flow monitor. Receiver Operating Characteristic (ROC) curve analysis showed that the SVM classification algorithm achieved a detection specificity of 100% at detection thresholds in the range from 500 to 750 mL/min and a maximum sensitivity of 95.2% at a detection threshold of 750 mL/min.

Analysis of access flow using duplex ultrasonography and the ultrasound dilutional method

Background: This study aimed to evaluate the correlation between duplex ultrasonography (DUS) and ultrasound dilution (UD) measurement’s results and determine the factors affecting the correlation of the measured values among other clinical factors in patients. Methods: This cross-sectional study was conducted from April 2020 to May 2020 and included 60 patients who visited our dialysis centre. The flow of the fistula was measured in the proximal brachial artery using DUS. While dialysis was performed on the same day, the access flow was measured using the UD method. The correlation and agreement between the access flows acquired by each measurement method were analysed. Similarly, the correlation was analysed by classifying the groups based on the predisposing factor, and statistically significant factors were observed through comparison. Results: Both measurements showed a moderate positive correlation ( r = 0.60, p < 0.01). The bias (mean of UD-DUS) between the two measurements was about 230 mL/min. When the measurement site of DUS was near the inflow artery and in the same anatomical section, a strong correlation with the measurement value of UD was observed (brachial based fistula: r = 0.85, radial based fistula: r = 0.87). Similarly, for patients without diabetes and those who regularly underwent access surveillance for the dialysis route, strong correlations were observed between the two measurements ( r = 0.79 and r = 0.88). Conclusions: Several factors can influence the correlation between UD and DUS. The findings showed a high correlation for DUS measurement sites within the same anatomical section as the inflow artery, patients without diabetes, and patients undergoing periodic surveillance.

High-flow arteriovenous fistula is not associated with increased extracellular volume or right ventricular dysfunction in haemodialysis patients

Background High-output congestive heart failure secondary to high-flow arteriovenous fistula (AVF) has been reported in haemodialysis (HD) patients. As high-flow AVF (HFA) would be expected to result in fluid retention, we conducted an observational study to characterize the relationship between AVF flow (Qa) and extracellular water (ECW) in HD patients. Methods We measured Qa by ultrasound dilution in prevalent HD outpatients with an AVF in two dialysis centres. The ECW:total body water (TBW) ratio was measured both pre- and post-dialysis by multifrequency bioimpedance analysis. Transthoracic echocardiograms (TTEs) were performed as part of routine clinical management. Results We included 140 patients, mean age 62.7 ± 15.7 years, 60.7% male, 47.9% diabetic and 22.9% with coronary revascularization. Mean Qa was 1339 ± 761 mL/min and 22 (15.7%) patients had HFA defined as Qa &gt;2.0 L/min. Qa was positively associated with an upper arm AVF (P = 0.005), body mass index (P = 0.012) and N-terminal pro-brain natriuretic peptide (NT-proBNP) (P = 0.047) and negatively associated with diabetes (P &lt; 0.001) and coronary revascularization (P = 0.005). The ECW:TBW ratio was positively associated with age (P &lt; 0.001), Davies comorbidity index (P = 0.034), peripheral vascular disease (P = 0.030) and NT-proBNP (P = 0.002) and negatively associated with serum albumin (P &lt; 0.001). Qa was not associated with the ECW:TBW ratio (P = 0.744). TTE parameters were not associated with Qa. Conclusions In our outpatient HD cohort, high AVF flow was not associated with ECW expansion, either pre- or post-dialysis when accounting for potential confounders. By controlling ECW, high access flow should not necessarily be perceived as a threat to cardiovascular physiology.

Comparing Non-invasive Diagnostic Methods for Arteriovenous Fistula Stenosis: A Prospective Study

Background International guidelines recommended screening for AVF stenosis using various non-invasive methods. The previous studies have reported conflicting results. This study aims to evaluate the utility of the different non-invasive AVF flow measurements for detecting AVF stenosis. Methods We evaluated 23 HD patients with the clinical suspicion for AVF stenosis based on physical examination or high venous pressures during. Patients underwent all three non-invasive measurements including ultrasound dilution, urea dilution, and Doppler ultrasonography. Fistulography was performed to confirm the degree of AVF stenosis in all patients. Results Fistulography revealed AVF stenosis in eighteen patients, twelve of whom had severe stenosis (greater than 50% stenosis). About the location of the stenotic lesions, eight were at the inflow site, six at the outflow site, and four at both sites. In those AVFs with severe stenosis, median access flows were 625 mL/min by ultrasound dilution method, 615 mL/min by urea dilution method, and 590 mL/min by Doppler ultrasonography. Receiver operating characteristic (ROC) curve analysis showed that Doppler ultrasonography had a high discriminative ability and the averaged areas under the curves were 0.933 (95% confidence interval [CI], 0.81 to 0.99) for stenosis and 0.929 (95% CI, 0.82 to 0.99) for severe stenosis. The sensitivity of each method for prediction of access stenosis was 73%, 73%, 80% by using the ultrasound dilution, urea dilution, and Doppler ultrasonography methods, respectively. The specificity of each method was 40%, 80%, and 100% respectively. Physical examination showed 80% sensitivity and 80% specificity for detecting AVF stenosis. The combination of non-invasive access flow measurement with physical examination increased the sensitivity for detection of AVF stenosis to 80%, 93%, and 93% respectively. Conclusions Doppler ultrasound combine with physical examination, was more accurate than other non-invasive methods for detection of AVF stenosis.

Predicting arterio-venous fistula failure by urea–method derived access blood flow in chronic hemodialysis patients

Background: Dialysis access is an essential part of hemodialysis. Determining the access blood flow (Qa) can help predict arterio-venous fistula thrombosis. Qa determination by the urea method, which was previously described in the past is simple and is available in most of the dialysis units but was not be able to predict arterio-venous fistula thrombosis. Aim: To compare the efficacy of Qa determination by the urea method and by ultrasound dilution methods in predicting arterio-venous fistula failure. Methods: Qa was measured by urea method and by ultrasound dilution simultaneously, every 3 months for a period of 1 year, in stable chronic hemodialysis patients with arterio-venous fistula. Arterio-venous fistula failure determined by clinical parameters and confirmed by Doppler ultrasound before sending the patient for interventional angioplasty. Results: This study enrolled a total of 16 patients, with 63% being male, 75% with lower-arm arterio-venous fistula and around 43% with previously done angioplasty. 59-Qa measurements were done, and 6 patients underwent angioplasty (one patient for severe upper limb edema, one for access thrombosis, and four for access dysfunction). Qa determination by the urea method had non-significantly lower Qa, (745 mL/min (interquartile range: 509–1143) and 779 mL/min (interquartile range: 530–1160), (p = 0.58)) and high correlation (r = 0.83, p < 0.001) to Qa by ultrasound dilution. The cut-off criteria with its sensitivity and specificity in predicting the access failure were 440 mL/min, 66.67%, and 96.15% for Qa determination by the urea and 400 mL/min, 66.67% and 90.38% for Qa by ultrasound dilution, respectively, with no difference in the area under the receiver operating characteristic curve. Conclusion: Measurement of Qa determination by the urea method is well correlated with Qa by ultrasound dilution and can be used to predict vascular access failure.

Comparing Non-invasive Diagnostic Methods for Arteriovenous Fistula Stenosis: A Prospective Study

Background International guidelines recommended screening for AVF stenosis using various non-invasive methods. The previous studies have reported conflicting results. This study aims to evaluate the utility of the different non-invasive AVF flow measurements for detecting AVF stenosis. Methods We evaluated 23 HD patients with the clinical suspicion for AVF stenosis based on physical examination or high venous pressures during. Patients underwent all three non-invasive measurements including ultrasound dilution, urea dilution, and Doppler ultrasonography. Fistulography was performed to confirm the degree of AVF stenosis in all patients. Results Fistulography revealed AVF stenosis in eighteen patients, twelve of whom had severe stenosis (greater than 50% stenosis). About the location of the stenotic lesions, eight were at the inflow site, six at the outflow site, and four at both sites. In those AVFs with severe stenosis, median access flows were 625 mL/min by ultrasound dilution method, 615 mL/min by urea dilution method, and 590 mL/min by Doppler ultrasonography. Receiver operating characteristic (ROC) curve analysis showed that Doppler ultrasonography had a high discriminative ability and the averaged areas under the curves were 0.933 (95% confidence interval [CI], 0.81 to 0.99) for stenosis and 0.929 (95% CI, 0.82 to 0.99) for severe stenosis. The sensitivity of each method for prediction of access stenosis was 73%, 73%, 80% by using the ultrasound dilution, urea dilution, and Doppler ultrasonography methods, respectively. The specificity of each method was 40%, 80%, and 100% respectively. Physical examination showed 80% sensitivity and 80% specificity for detecting AVF stenosis. The combination of non-invasive access flow measurement with physical examination increased the sensitivity for detection of AVF stenosis to 80%, 93%, and 93% respectively. Conclusions Doppler ultrasound combine with physical examination, was more accurate than other non-invasive methods for detection of AVF stenosis.

Screening for hemodialysis graft stenosis and short-term thrombosis risk: A comparison of the available tools

Introduction: Guidelines recommend hemodialysis graft screening to identify and repair significant (>50%) stenosis at high risk of thrombosis, but there is insufficient evidence to prefer one or other screening tool due to the lack of studies comparing all available options. Methods: Seeking an optimal screening approach, we compared the performance of all currently used tools (duplex ultrasound to detect significant stenosis (StD) and measure access blood flow (QaD), ultrasound dilution access blood flow (QaU), static venous intra-access pressure ratio (VAPR), dynamic arterial and venous pressures measurement, and monitoring) for diagnosing significant angiography-proven stenosis (StA) and predicting incipient thrombosis (occurring within 4 months) in 62 grafts. All thrombotic and symptomatic acute hypotension episodes were recorded during follow-up. Results: VAPR > 0.70 and QaU < 1600 mL/min were the best indicators to angiography for those aiming to identify the majority of StA (91% sensitivity) and QaU < 1000 mL/min or StD for those aiming to avoid unnecessary angiograms (95%–93% positive predictive value). At Cox’s analysis, the only significant thrombosis predictors were acute hypotension episodes (relative risk = 4.4 (95% confidence interval = 2.2–8.8), p < 0.0001) and QaU or QaD (14% (95% confidence interval = 8–21) or 16% (95% confidence interval = 6–25) increased risk per 100 mL/min drop in Qa, p < 0.003). Thrombosis risk (adjusted for acute hypotension) became significantly higher at QaU = 1000–700 mL/min (relative risk = 3.6 (95% confidence interval = 1.6–8.2), p < 0.001) and QaD = 1300–1000 mL/min (relative risk = 3.1 (95% confidence interval = 1.1–12.8), p = 0.031). The proportion of thromboses attributable to acute hypotension was 40% (95% confidence interval = 24–57). Conclusions: Our comparative study showed that an effective screening for graft stenosis and short-term thrombosis risk can rely on Qa surveillance alone, and suggested that avoiding acute hypotension and correcting stenosis at QaU < 1000 mL/min or QaD < 1300 mL/min can contain thrombosis risk.