Contrast Venography Versus Intravenous Ultrasound in Hemodialysis Arteriovenous Access Dysfunction

Contrast venography is the current standard for arteriovenous dialysis access dysfunction, but this approach has several limitations. Venography images are taken in a single dimension (anteroposterior), which can introduce errors in detecting and assessing the severity of stenosis (a three-dimensional structure), and venography does not allow morphological characterization of the arteriovenous access. Intravascular ultrasound (IVUS) provides a three-dimensional view potentially enhancing the accuracy in detecting stenoses. In contrast to venography, imaging with IVUS does not require intravenous contrast, is not affected by bony structures, and allows circumferential evaluation of the morphology of stenotic lesions1de Korte C.L. van der Steen A.F. Intravascular ultrasound elastography: an overview.Ultrasonics. 2002; 40: 859-865Crossref PubMed Scopus (119) Google Scholar. In this feasibility study, venography and IVUS were compared in 41 participants referred for poor flow, high venous pressure, or persistent bleeding from the hemodialysis vascular access. Five of the 41 participants were excluded from analyses due to insufficient venogram images, and three were excluded because of insufficient IVUS images, resulting in 33 participants for the final analysis (Figure 1a and STROBE statement in supplementary material). Most participants had one or more hemodialysis catheters in the past (Supplementary table S1 for baseline characteristics). Ninty percent of the participants had an AVF; more than half were brachio-cephalic and located on the left upper extremity. Among these 33 participants, 20 were found to have a single stenosis, and 13 had two stenotic areas on venography yielding 46 stenotic lesions for analysis. The IVUS images showed normal and abnormal pathological findings including stent struts, stenosis and calcification (Supplementary figures S1 and S2). For both modalities, the extent of stenosis was measured at the narrowest region and expressed as a ratio to the pre-stenosis reference lumen (Supplementary table S2). Venography detected 27 stenoses (>50% narrowing of lumen), of which 24 (89%) were also detected by IVUS. Venography identified 3 stenoses that were not evident by IVUS. Additionally, IVUS detected a total of 38 stenosis with greater than 50% reduction in the lumen, of which 14 were underestimated in the severity by venography (30.42%). The McNemar ratio of 7.11 (P= 0.0076) and Cohen kappa of 0.163 (SE 0.1281, 97.5% CI of -0.1207 to -0.4534) suggested discordance between the two modalities. This disagreement is depicted in a distribution graph (Figure 1b), where an equal distribution of points on both the side of the line indicates an equipoise between two modalities. An unequal distribution was noted in favor of IVUS. The severity of stenoses was lower by venography (mean + SD 51.7 +14.1) compared to IVUS (64.9 +18.1, p=0.002) (Figure 1c). This difference in the extent of stenosis between two modalities spanned over all grades of stenoses (Figure 1d). These analyses included both peripheral and central venous stenotic lesions. A separate analyses of central vein stenoses was performed because of the unique challenges with venography due to bony interference and effects of respiration on the vein lumen. IVUS detected ten central vein stenoses (>50% reduction in lumen), only three of which were detected by venography. The remaining seven central venous stenoses were either not detected or appeared as <50% by venography. Two morphological features – eccentricity and homogeneity indices – were extracted from the IVUS images. For this analysis, all 50 stenoses detected by IVUS were processed as described in Figure 2a and Supplementary method. A lower eccentricity index implies a uniformly circular lumen, whereas a higher eccentricity index implies a non-circular lumen. Stenotic segments had higher eccentricity indices compared to reference segments (P = 0.0047, Figure 2b). The violin plots (Figure 2c) illustrate a wide variability in the distribution of the eccentricity index in the stenotic and reference segments. Black points correspond to an individual IVUS frame. For example, participants 4 and 7 (P4 and P7, Figure 2c) both had a higher eccentricity index in the stenotic segment compared to the reference segment. There was reversal of this pattern in a few cases. The eccentricity index was lower at the start of stenotic segment and was highest in the center of the stenosis. A pairwise comparison of the distribution of the eccentricity index between stenotic lesions (a) with all other stenoses and (b) with 50 reference segments was performed and a heatmap was generated. A highlighted square in the right side of Figure 2d compares the stenotic segments from participant 47 and participant 41 and depicts a substantial difference in the distribution of eccentricity indices. In a small subset of stenoses the eccentricity indices were similar (white cells) but the eccentricity indices differed in between the majority of stenoses (Figure 2e). The homogeneity index was extracted per workflow depicted in Figure 2f. A lower homogeneity index implies a heterogenous wall arising from differentially backscattered acoustic waves and suggests a mixture of different components of the wall. On the other hand, a higher homogeneity index implies homogeneous architecture of the wall. The average homogeneity index of stenosis was higher than the reference segments (Figure 2g) (P = 0.013). A wide variation was noted in the distribution of homogeneity index between stenotic and reference segments (Supplementary figures S3a). A pair-wise comparison revealed wide variability in the homogeneity index among stenoses (Supplementary figures S3b) and between a stenotic segment and reference segments (Supplementary figures S3c) was noted. These results suggest that the stenotic walls are more homogenous compared to the reference segments and substantial variability in the wall composition exist among stenoses. This prospective study directly compared IVUS and venography for arteriovenous access dysfunction. IVUS was tolerated well in all participants similar to the previous studies2Higuchi T. Okuda N. Aoki K. et al.Intravascular ultrasound imaging before and after angioplasty for stenosis of arteriovenous fistulae in haemodialysis patients.Nephrol Dial Transplant. 2001; 16: 151-155Crossref PubMed Google Scholar,3Arbab-Zadeh A. Mehta R.L. Ziegler T.W. et al.Hemodialysis access assessment with intravascular ultrasound.Am J Kidney Dis. 2002; 39: 813-823Abstract Full Text Full Text PDF PubMed Google Scholar. Compared to IVUS, venography underestimated the severity of stenosis in 30% of lesions. These findings raise the possibility that some stenotic lesions in patients presenting with arteriovenous access dysfunction might go unaddressed with venography, and that, if the findings are confirmed in larger studies, an interventionist might change the decision based on IVUS findings. The shape of the lumen of arteriovenous access is a function of several factors including remodeling and hemodynamics4Shiu Y.T. Rotmans J.I. Geelhoed W.J. et al.Arteriovenous conduits for hemodialysis: how to better modulate the pathophysiological vascular response to optimize vascular access durability.Am J Physiol Renal Physiol. 2019; 316: F794-F806Crossref PubMed Scopus (28) Google Scholar. A higher stenosis eccentricity index may suggest an abnormal balance of the remodeling factors and hemodynamics and is likely to result from processes such as intimal hyperplasia. The homogeneity index characterizes the texture of the wall. The wall of the outflow vein consists of several cell types such as endothelial cells, vascular smooth muscle cells, collagen, and elastin fibers as well as calcification, hematoma, and vessel wall defects (aneurysm or pseudoaneurysm). Based on the known composition of vessels, the stenotic lesions would be expected to be more heterogenous (low homogeneity index) compared to the normal AVF wall (reference lumen). Unexpectedly the stenotic lesions had higher homogeneity index than the reference lumen, which may suggest homogenous wall and predominance of a small number of cell types or pathological process such as fibrosis in the stenotic lesions compared with the non-stenotic portions of the vascular access. Several imaging techniques are proposed for arteriovenous access dysfunction5Tasaki Y. Sueyoshi E. Takamatsu H. et al.The outcomes of carbon dioxide digital subtraction angiography for percutaneous transluminal balloon angioplasty of access circuits and venous routes in hemodialysis patients.Medicine. 2020; 99e21890Crossref PubMed Scopus (5) Google Scholar, 6He Y. Terry C.M. Nguyen C. et al.Serial analysis of lumen geometry and hemodynamics in human arteriovenous fistula for hemodialysis using magnetic resonance imaging and computational fluid dynamics.J Biomech. 2013; 46: 165-169Crossref PubMed Scopus (60) Google Scholar, 7Zamboli P. Fiorini F. D'Amelio A. et al.Color Doppler ultrasound and arteriovenous fistulas for hemodialysis.J Ultrasound. 2014; 17: 253-263Crossref PubMed Scopus (56) Google Scholar. However, none offer morphological definition other than optical coherence tomography – (OCT)8Araki M. Park S.J. Dauerman H.L. et al.Optical coherence tomography in coronary atherosclerosis assessment and intervention.Nat Rev Cardiol. 2022; 19: 684-703Crossref PubMed Scopus (39) Google Scholar. OCT requires a bloodless area for visualization created by contrast agent, which is feasible in coronary vessels, but not in large AVFs with high blood flow. This study has limitations. Even though IVUS detected a greater number of stenoses (central venous stenosis) and indicated a higher severity of stenosis, further studies are required to rigorously test the clinical significance and implications of these findings on the management of stenoses. While both the indices can provide clues to the mechanism of stenotic lesions, further studies are needed to evaluate their clinical utility in the management of patients such as response to an interventional therapy. Hemodialysis vascular access dysfunction is an area of high unmet need. This study suggests that IVUS may address some limitations of venography and may also provide novel information with potential prognostic and mechanistic value. These findings justify additional investigations of IVUS. We thank the interventional radiology staff members, vascular surgery staff members for their assistance with the study. A part of this study is submitted as an abstract in the 60th European Renal Association Congress in 2023 in Milan and Annual meeting of American Society of Nephrology, 2023 in Philadelphia. Download .pdf (3.77 MB) Help with pdf files