Increased Excess Pressure After Creation of an Arteriovenous Fistula in End-Stage Renal Disease

ABSTRACT

BACKGROUND

Reservoir-wave analysis (RWA) separates the arterial waveform into reservoir and excess pressure (XSP) components, where XSP is analogous to flow and related to left ventricular workload. RWA provides more detailed information about the arterial tree than traditional blood pressure (BP) parameters. In end-stage renal disease (ESRD), we have previously shown that XSP is associated with increased mortality and is higher in patients with arteriovenous fistula (AVF). In this study, we examined whether XSP increases after creation of an AVF in ESRD.

METHODS

Before and after a mean of 3.9 ± 1.2 months following creation of AVF, carotid pressure waves were recorded using arterial tonometry. XSP and its integral (XSPI) were derived using RWA through pressure wave analysis alone. Aortic stiffness was assessed by carotid–femoral pulse wave velocity (CF-PWV).

RESURLTS

In 38 patients (63% male, age 59 ± 15 years), after AVF creation, brachial diastolic BP decreased (79 ± 10 vs. 72 ± 12 mm Hg, P = 0.002), but the reduction in systolic BP, was not statistically significant (133 ± 20 vs. 127 ± 26 mm Hg, P = 0.137). However, carotid XSP (14 [12–19] to 17 [12–22] mm Hg, P = 0.031) and XSPI increased significantly (275 [212–335] to 334 [241–439] kPa∙s, P = 0.015), despite a reduction in CF-PWV (13 ± 3.6 vs. 12 ± 3.5 m/s, P = 0.025).

CONCLUSIONS

Creation of an AVF resulted in increased XSP in this population, despite improvement in diastolic BP and aortic stiffness. These findings underline the complex hemodynamic impact of AVF on the cardiovascular system.

Graphical Abstract

Graphical Abstract.

Patients with end-stage renal disease (ESRD) require renal replacement therapy through either hemodialysis or peritoneal dialysis. To perform hemodialysis, large quantities of blood are required to achieve sufficient clearance of uremic toxins. Access to this amount of blood is achieved through a central venous catheter or a surgically created arteriovenous fistula (AVF). Creation of an AVF involves connecting an artery to a nearby vein. On average, a mature AVF achieves a blood flow of 500–1,000 ml/min.¹ Creation of an AVF is known to have important hemodynamic consequences, as the creation of a high flow and low resistance environment increases venous return, stroke volume, and myocardial contractility.^2,3 Furthermore, the increase in cardiac output causes increased myocardial oxygen demand, and can result, over time, in left ventricular hypertrophy.^4–7 On the other hand, a recent meta-analysis by Scholz et al. provided estimates of AVF-induced changes in systolic, diastolic, and mean blood pressures (BPs), which were reduced by an average of 8.7, 5.9, and 6.6 mm Hg, respectively.⁸ These observations are in line with reductions of BP achieved by nonpharmacological treatment of resistant hypertension through creation of AVF between the iliac artery and vein.⁹

While brachial systolic and diastolic BPs are convenient, they are relatively simplistic compared with the information that is provided through the analysis of the arterial pressure waveforms. The reservoir-wave analysis (RWA) is an alternative approach to pulse wave analysis, which aims to incorporate arterial reservoir function with wave propagation. The RWA hypothesizes that the measured arterial pressure is the sum of a reservoir pressure (RP) wave and an excess pressure (XSP) wave.¹⁰ The RP accounts for the dynamic storage and release of blood by the compliant arteries (the Windkessel effect) and is related to minimal left ventricular work required to generate blood flow into the aorta. The XSP, which is obtained by subtracting RP from the acquired pressure waveform, is responsible for local changes in the pulse waveform, is analogous to flow, and can provide information about surplus work performed by the left ventricle.^11,12

Using RWA, we and others have shown that higher excess pressure integral (XSPI) is associated with increased cardiovascular and all-cause mortality in ESRD, above and beyond all other available hemodynamic parameters, including aortic stiffness as assessed by carotid–femoral pulse wave velocity (CF-PWV).^13,14 In addition, our previous findings show that XSPI was statistically higher in ESRD patients with an AVF.¹⁵ Indeed, if XSP is analogous to flow, it is conceivable the creation of AVF could increase XSP. In the present study, our aim was to assess the impact of creation of an AVF on reservoir-wave parameters in ESRD patients, in a before and after study design.

METHODS

Patient population and study design

This longitudinal study looks at the central hemodynamic parameters of patients before and after creation of an AVF for hemodialysis. All adult patients with ESRD who were scheduled for an AVF within the following month were invited to participate. Exclusion criteria included a nonfunctional AVF at follow-up time, extreme BP values (systolic BP >190 or <80 mm Hg), severe congestive heart failure, patients having had a previous AVF, those for whom follow-up evaluation was conducted more than 6 months after surgery and those who were transferred to satellite units. The study was conducted at CHU de Québec-Université Laval, L’Hôtel-Dieu de Québec Hospital between November 2004 and December 2007. The study was approved by the institutional ethics committee and all participants gave written informed consent.

All hemodynamic assessments were performed within 1 month prior to AVF creation and within a mean follow-up of 3.9 ± 1.2 months following the surgery. Figure 1 shows the study flowchart. Overall, 57 participants were initially recruited, however, 19 patients were subsequently excluded for the following reasons: severe cardiac failure (N = 1), extremely low BP (N = 1), transfer to another facility (N = 7), death (N = 1), nonfunctional AVF (n = 7), and a follow-up >6 months (N = 2). Twenty-nine of these subjects have been described previously with regard to central BPs and pulse wave velocity.¹⁶ All routine laboratory tests were performed on the mid-week hemodialysis session for patients on hemodialysis and in morning in patients on peritoneal dialysis. Ejection fractions reported in this study were taken from available echocardiographic assessments made prior to creation of AVF for a variety of clinical reasons.

Figure 1.

Study flowchart. The figure shows the number of patients excluded before vascular hemodynamic assessment due to severe illness or death, as well as clinical and methodological reasons. Also shown on the chart are exclusions which occurred following vascular hemodynamic assessment due to no functional AVF or assessment past the predetermined 6 months delay. Abbreviation: AVF, arteriovenous fistula.

Hemodynamic parameters

All hemodynamic measurements were done on the opposite side of the AVF (expected or present). If patients were already on dialysis at baseline (N = 19), all measurements were performed just prior to the mid-week dialysis session. Briefly, after 15 minutes of rest in a supine position, brachial artery BP was recorded 6 times, with a 2-minute interval between each recording using an automatic oscillometric sphygmomanometer BPM-100 (BP-Tru, Coquitlam, Canada). Immediately after BP measurements, radial and carotid pulse wave profiles were sequentially recorded in the same order by aplanation tonometry (SphygmoCor system, AtCor Medical Pty, Sydney, Australia). Three consecutive recordings were performed for each site. Central pressure waveform was obtained from radial artery tonometry through application of a generalized transfer function. Central systolic BP, diastolic BP, pulse pressure, and heart rate adjusted augmentation index (AIx@75) were derived after calibration of central waveform for brachial systolic and diastolic BPs.¹⁷ Carotid pressure waveform was obtained by carotid artery tonometry after calibration for diastolic and mean arterial pressure, or mean BP, which was obtained by integration of the radial artery pressure waveform. Immediately after pulse wave recordings, we determined CF-PWV in triplicate by Complior SP (Artech Medical, Pantin, France), using the maximal upstroke algorithm and direct measurements as previously described.¹⁸

Wave separation analysis was conducted to derive central pressure forward (Pf), pressure backward (Pb), and reflection magnitude (RM = (100 × Pb)/Pf) and reflection index (RI = (Pb × 100)/(Pb + Pf)) were calculated. This was performed on the central pressure waveform after application of a generalized transfer function on the radial artery pressure waveform.^19,20

Reservoir-wave parameters were obtained from carotid pressure waveforms using the pressure wave approach as previously described.^21,22 RP, its integral (RPI), XSPI, diastolic rate constant (DC), and systolic rate constant (SC) were acquired from carotid pressure waveforms. Accordingly, SC is the rate of system filling which is inversely proportional to the product of characteristic impedance (Z₀) and compliance (C), whereas DC is the rate of system emptying, which is inversely proportional to the product of peripheral vascular resistance (R) and compliance (C). RP was derived based on pressure alone and XSP was defined as the difference between total measured pressure and RP. A RP analysis was considered valid with RP > 0, XSPI > 0, a numerical SC and DC, DC > 0 and P_∞ > 0, where P_∞ is the minimal pressure of the system, as modeled by asymptotic decay. Figure 2 summarizes the key parameters of RWA of the carotid artery.

Figure 2.

Relevant parameters and outcomes of reservoir-wave analysis (RWA). The figure shows 2 carotid pressure waveforms sampled from a single subject, the left one obtained before arteriovenous fistula (AVF) creation, the right one after AVF creation. Each waveform has been decomposed into reservoir and excess pressure waves, designated as such on the left waveform. Relevant parameters of RWA are indicated with broken lines on the right waveform. After AVF creation, DBP, diastolic blood pressure decreases; DC, diastolic rate constant decreases; max RP, peak reservoir pressure does not change; max XSP, peak excess pressure increases; RPI, reservoir pressure integral/area under the curve does not change; SC, systolic rate constant decreases; XSPI, excess pressure integral/area under the curve increases.

Statistical analysis

All analyses were done with SPSS, version 25.0 (SPSS, Chicago, IL). Results are reported as the mean ± standard deviation or median [25th–75th percentiles] as appropriate. The variations between pre-AVF and post-AVF creation were evaluated through a paired sample t-test or Wilcoxon tests as appropriate. As part of sensitivity analyses, generalized estimating equation models were used to verify whether the impact of AVF on reservoir-wave parameters was modified by changes in the number of antihypertensive medications. Models considered timing of measurement (pre-AVF vs. post-AVF) as within-subject effect and included number of antihypertensive medication, before and after AVF as covariate. As part of the sensitivity analyses, we examined if the hemodynamic response to AVF was different in subjects according to the dialysis or predialysis status, and baseline clinical and biochemical parameters. For all analyses, a 2-tailed P < 0.05 was considered statistically significant.

RESULTS

Baseline characteristics

Baseline demographic and clinical characteristics are presented in Table 1. Among the 38 subjects, 19 were already treated by hemodialysis, 1 by peritoneal dialysis and 18 were in predialysis care without renal replacement therapy. Of those 18 participants, 9 were still on predialysis care at follow-up, whereas others had begun hemodialysis treatment. The number of subjects with aspirin (N = 17) and statins (N = 25) did not change significantly over the observation period. Overall, the participants can be considered a representative sample of the ESRD population, with high prevalence of hypertension, diabetes, and cardiovascular disease. Accordingly, the participants were taking various medications including antihypertensive medication. Before AVF creation, on average participants were taking 2 [1–3] different antihypertensive drugs and 1.5 [1–3] after creation of AVF (P = 0.267). Given potential differences and confounding factors, we examined baseline characteristics of participants on predialysis care at baseline and compared them to those already on dialysis. Patients on hemodialysis were in general younger (54 ± 17 vs. 65 ± 10, P = 0.014), had a lower serum albumin level (36.1 ± 3.33 vs. 40.1 ± 4.43 g/l, P = 0.008), plasma urea concentration (19.9 [15.3–22.9] vs. 26.1 [22.0–33.0] mmol/l, P < 0.001), but higher creatinine levels (690 [435–869] vs. 425 [346–539] µmol/l, P = 0.026). However, they were similar in terms of cardiovascular disease, diabetes, hypertension, left ventricular mass, and hemoglobin levels.

Table 1.

Baseline demographic

	Pre-AVF (N = 38)
Age (years)	59 ± 15
Men	24 (63)
BMI (kg/m2)	28 ± 6
Hypertension	34 (90)
Diabetes	16 (42)
Dialysis	19 (50)
Dialysis vintage (N = 19, months)	57 [19–338]
History of smoking	21 (55)
History of cardiovascular disease	23 (61)
Hemoglobin (g/l)	110 ± 14
Albumin (g/l)	40 ± 4
Left ventricular mass (g/m2)	118 ± 40
Ejection fraction (%)	63 ± 10

Values are mean ± SD, median (25th–75th percentiles or N (%). Abbreviations: AVF, arteriovenous fistula; BMI, body mass index.

Brachial and central pressure changes after AVF creation

Table 2 summarizes hemodynamic parameters, namely brachial and central BP obtained through generalized transfer function and CF-PWV. After AVF creation, diastolic BP and mean BP decreased significantly at the brachial level but pulse pressure did not change significantly because of a small but nonsignificant reduction in brachial systolic BP. A similar response was observed for central and carotid diastolic BPs, systolic BPs, and pulse pressures (Table 3). After creation of an AVF, the average heart rate did not change, the ejection duration increased slightly without reaching statistical significance (P = 0.084), but the subendocardial viability ratio decreased significantly. Subendocardial viability ratio is a marker of subendocardial perfusion expressed as the ratio of the diastolic pressure time index and systolic pressure time index.¹⁶ Despite a reduction in CF-PWV there was a statistically nonsignificant increase in AIx@75 and a significant increase in RM.

Table 2.

Hemodynamic parameters and wave separation analysis before–after AVF

	Before AVF (N = 38)	After AVF (N = 38)	P
HR (bpm)	71.6 ± 13.7	71.0 ± 13.4	0.766
Brachial BP
SBP (mm Hg)	132.7 ± 19.6	127.3 ± 25.7	0.137
DBP (mm Hg)	78.5 ± 10.3	71.7 ± 12.2	0.002
PP (mm Hg)	54.1 ± 17.2	55.6 ± 19.7	0.572
MBP (mm Hg)	96.7 ± 12.5	91.3 ± 16.9	0.040
Central BP (GTF)
SBP (mm Hg)	120.9 ± 19.1	116.8 ± 25.2	0.251
DBP (mm Hg)	79.5 ± 10.5	73.0 ± 12.6	0.004
PP (mm Hg)	41.3 ± 17.1	43.7 ± 19.3	0.350
AP (mm Hg)	11.5 ± 9.5	13.5 ± 11.0	0.171
AIx (%)	24.9 ± 13.3	27.6 ± 13.3	0.169
AIx@75 (%)	23.2 ± 12.0	25.6 ± 12.5	0.131
SEVR	149.0 ± 33.8	139.2 ± 32.3	0.030
Tr (ms)	135.1 ± 10.1	136.9 ± 11.1	0.217
ED (ms)	309.2 ± 37.6	320.3 ± 28.0	0.084
Pf (mm Hg)	27.5 ± 9.6	28.4 ± 11.2	0.568
Pb (mm Hg)	16.4 ± 7.5	18.2 ± 9.3	0.135
RM	0.59 ± 0.12	0.63 ± 0.12	0.044
PWV
CF-PWV (m/s)	13.2 ± 3.6	12.0 ± 3.5	0.025

Values are mean ± SD with a corresponding P values obtained by paired t-test, or median [25th–75th percentiles] with corresponding P values obtained by Wilcoxon Signed Rank Test. Abbreviations: AIx@75, heart rate adjusted augmentation index; AP, augmentation pressure; AVF, arteriovenous fistula; BP, blood pressure; CF-PWV, carotid–femoral pulse wave velocity; DBP, diastolic blood pressure; ED, ejection duration; GTF, generalized transfer function; HR, heart rate; MBP, mean blood pressure; Pb, pressure backward; Pf, pressure forward; PP, pulse pressure; RM, reflection magnitude; SBP, systolic blood pressure; SEVR, subendocardial viability ratio; Tr, time of return of reflected wave.

Table 3.

Carotid reservoir-wave analysis before and after arteriovenous fistulae creation

	Before AVF (N = 38)	After AVF (N = 38)	P
Carotid BP
SBP (mm Hg)	122.8 ± 20.0	119.4 ± 27.0	0.380
DBP (mm Hg)	79.4 ± 10.4	72.6 ± 12.4	0.003
PP (mm Hg)	43.4 ± 17.6	46.9 ± 21.5	0.212
Carotid artery RWA
RP carotid (mm Hg)	34.9 [24.5–42.4]	35.8 [23.7–44.8]	0.455
RPI carotid (kPa s)	1,515 [1,088–1,952]	1,672 [1,096–2,179]	0.482
Time to max RP (cs)	30.4 ± 4.4	31.8 ± 3.7	0.083
XSP carotid (mm Hg)	14.1 [11.5–18.9]	17.2 [12.4–22.2]	0.031
XSPI (kPa sa)	275.3 [211.7–335.0]	333.75 [240.8–439.1]	0.015
Time to max XSP (cs)	9.0 [8.6–10.7]	10.3 [9.0–11.5]	0.034
SC × 10−2	20.2 ± 7.1	17.3 ± 7.7	0.025
DC × 10−2	3.4 ± 1.4	2.9 ± 1.3	0.019
P∞	75.1 ± 11.0	68.7 ± 17.9	0.009

Values are mean ± SD with a corresponding P values obtained by paired t-test, or median [25th–75th percentiles] with corresponding P values obtained by Wilcoxon Signed Rank Test. Abbreviations: BP, blood pressure; DBP, diastolic blood pressure; DC, diastolic rate constant; P_∞, pressure infinity; PP, pulse pressure; RP, reservoir pressure; RPI, reservoir pressure integral; RWA, reservoir-wave analysis; SBP, systolic blood pressure; SC, systolic rate constant; XSP, excess pressure; XSPI, excess pressure integral.

^aTo convert from kPa s to mm Hg s, multiply by 7.5.

Carotid reservoir-wave analysis

Table 3 summarizes the carotid BP and parameters of reservoir-wave analysis obtained before and after AVF creation. Figure 2 shows a representative carotid reservoir-wave analysis in a subject before and after creation of AVF. We observed a significant increase in XSP, XSPI, and time to maximum XSP. However, there were no changes in RP, RP integral or time to maximum RP. There were also significant reductions in SC and DC. These observed changes in XSP, XSPI, SC, and DC remained statistically significant even after adjustment for the number of antihypertensive medications. Also, despite differences in baseline characteristics of patients already on dialysis and those on predialysis care, there were no statistically significant differences in hemodynamic response to AVF creation between subgroups.

Discussion

In ESRD patients, this study shows an increase in XSP following AVF creation, despite a reduction in diastolic BP and aortic stiffness. Given that increased XSP has a prognostic value above and beyond aortic stiffness and other potential confounders in this population,¹³ our observations provide important information on the complex hemodynamic effects of AVF that cannot be captured through traditional systolic and diastolic BPs alone.

The RWA, an alternative approach to pulse wave analysis, aims to integrate arterial reservoir function with wave propagation. In the RWA model, it is proposed that the measured arterial pressure is the sum of a RP and XSP waves¹⁰ where the RP wave accounts for the dynamic storage and release of blood by the compliant arteries (the Windkessel effect) and is related to minimal left ventricular work required to generate blood flow into the aorta. In this model, the XSP wave is responsible for local changes in the pulse waveform, is related to flow, and can provide information about the work performed by left ventricle. As such, the increased XSP after creation of an AVF that we observed in this study is in keeping with increased flow and increased cardiac workload. Indeed, all parameters related to XSP, such as peak XSP, timing of the peak XSP and XSPI increased significantly after creation of AVF.

There were also changes in other parameters of RWA such as the SC and the DC. However, the interpretation of these parameters is more complex, and their clinical relevance are less well established. In fact, SC represents the rate of system filling and it is inversely proportional to the product of characteristic impedance (Z₀) and compliance (C). Our findings show a decrease in SC after AVF (i.e., lower rate of system filling), which could be explained by an increase in C and possibly through increased Z₀. According to water hammer equation, Z₀ is proportional to the aortic PWV divided by aortic cross-sectional area in diastole. In our study, CF-PWV decreased, but since there was a reduction in the diastolic BP, aortic diameter in diastole may also have decreased after AVF creation. Therefore, the amplitude and directionality of change in Z₀, as well as its contribution to SC cannot be quantified. While, the ability of SC to predict clinical outcomes has yielded conflicting results across various populations,^23,24 a lower SC has been associated with increased cardiovascular and overall mortality after adjustment for confounding factors (including aortic stiffness) in ESRD patients.¹³ On the other hand, DC represents the rate of system emptying and is inversely proportional to the product of peripheral resistance (R) and compliance (C). While an increase in compliance may contribute to the decrease in DC, as with SC, the contribution of R to the decline in DC remains unclear, due to the interplay with P_∞.

Using classical wave separation analysis, we observed a nearly significant increase in ejection duration as well as an increase in augmentation index and increased RM. Given the concurrent decrease in carotid–femoral PWV, these changes most likely indicate an increase in ventricular filling and/or contractility, rather than increased wave reflection.^25,26

While our findings suggest that AVF could have a negative impact on the RWA parameters that are associated with increased cardiovascular and overall mortality in ESRD, AVF may also some beneficial hemodynamic effects. The aforementioned meta-analysis by Scholz et al. provided estimates of AVF-induced changes in systolic, diastolic, and mean BPs, which were reduced by an average of 8.7, 5.9, and 6.6 mm Hg, respectively.⁸ This reduction in traditional BP measurement has been proposed to explain a slightly slower loss of renal function, following AVF creation, in predialysis ESRD patient.²⁷ Also, epidemiological observational studies using administrative databases show that hemodialysis with an AVF gives a better survival advantage over the use of central venous catheters and remains the vascular access of choice for hemodialysis.^28,29 However, this view has recently been challenged, as newer approaches associate these findings rather to a better general health of patients referred for AVF creation and who develop a functioning AVF.^30,31 Taken together, these findings suggest that although AVF creation might be beneficial in preserving residual renal function, it may also increase the risk of cardiovascular and all-cause mortality in ESRD patients. Further studies are needed to determine whether certain subgroups of ESRD patients are more likely to benefit from the creation of an AVF and whether other groups are at higher risk of its consequences.

The study has several strengths that need to be mentioned. In a previous cross-sectional study, we showed an association between the presence of an AVF and higher XSP in dialysis patients. With the current study, we have provided a direct comparison of each individual before and after creation of a functioning AVF, supporting a causal link between AVF and increased XSP. The relative short-term follow-up of a maximum of 6 months after AVF was specifically chosen to ensure that AVFs had undergone maturation, providing the desired blood flow, which usually occurs from 2 to 6 months after the procedure. We limited follow-up to no later than 6 months to avoid confounding effect due to significant changes in BP medication and potential chronic adaptation of cardiovascular system to the creation of a high flow AVF. There are also some limitations that need to be underlined. First, the AVF blood flow was not measured at the time of hemodynamic assessment during follow-up. Second, aortic flow and cardiac output were not assessed during the study, thus RP was calculated using a pressure-only approach. Third, there was some heterogeneity regarding changes in medication and dialysis treatment from baseline. However, using a generalized estimating equation model adjusting for the number of antihypertensive medication and dialysis status at baseline, there was there was no statistically significant difference in hemodynamic response to AVF creation between subgroups. Forth, the pressure-only approach for calculation of RP waves builds on the assumption that resultant XSP is proportional to volume flow rate out of the left ventricle, but this assumption has been recently validated in humans.³² Finally, as for all models, there are conceptual limitations to the RWA that have been reviewed previously,^33–35 but still many experts agree on the utility of XSPI as a prognostic marker.^{13,14,36–38}

In conclusion, the creation of an AVF has a complex impact on the cardiovascular system, that cannot solely be assessed through traditional brachial systolic and diastolic BPs. Indeed, our study shows that despite a reduction in diastolic BP and aortic stiffness, the creation of an AVF resulted in increased XSP, a parameter that has been shown to be independently associated with increased adverse clinical outcomes in this population. Besides patients with ESRD, our findings may have clinical implications in treatment of resistant hypertension by AVF, an experimental nonpharmacological approach, for which long-term effectiveness on clinical outcomes remain unknown.^9,39

FUNDING

This project was supported by the Canadian Institute of Health Research, New Emerging Team Grant (NET-54008), the Heart & Stroke Foundation of Canada, the Kidney Foundation of Canada, and the Canadian Diabetes Association. C. Fortier is supported by a postdoctoral exchange program between Fonds de Recherche du Québec—Santé (FRQ-S) and the Institut National de la Santé et de la Recherche Médicale (INSERM). R. Goupil is supported by a FRQ-S clinician-scientist scholarship.

DISCLOSURE

The authors declared no conflict of interest.