Cardiac implications of upper-arm arteriovenous fistulas: A case series

Abstract

Background:

Cardiovascular disease is a major cause of morbidity and mortality in patients with end-stage kidney disease. Arterio-venous fistulas (AVF), the gold standard for hemodialysis vascular access, are known to alter cardiac morphology and circulatory hemodynamics. We present a prospective case series of patients after creation of an AVF, explore the timeline for changes in their cardiac morphology, and detail considerations for clinicians.

Methods:

Patients were recruited in 2010 at multiple centers immediately prior to the creation of an upper-arm AVF and the initiation of hemodialysis. Cardiovascular magnetic resonance images were taken at intake before the creation of the AVF, 6-month follow-up, and 12-month follow-up. Image segmentation was used to measure left ventricular volume and mass, left atrial volume, and ejection fraction.

Results:

Eight patients met eligibility criteria. All eight patients had a net increase in left ventricular mass over enrollment, with a mean increase of 9.16 g (+2.96 to +42.66 g). Five participants had a net decrease in ejection fraction, with a mean change in ejection fraction of −5.4% (−21% to +5%). Upon visual inspection the patients with the largest ejection fraction decrease had noticeably hypertrophic and dilated ventricles. Left atrial volume change was varied, decreasing in five participants, while increasing in three participants. Changes in morphology were present at 6-month follow-up, even in patients who did not maintain AVF patency for the entirety of the 6-month period.

Conclusion:

All patients included in this prospective case series had increases in left ventricular mass, with variability in the effects on the ejection fraction and left atrial volume. As left ventricular mass is an independent predictor of morbidity and mortality, further research to determine appropriate vascular access management in both end-stage kidney disease and kidney transplant populations is warranted.

Introduction

Cardiovascular disease contributes to significant morbidity and mortality in patients with end-stage kidney disease (ESKD), with 33% cardiovascular-related mortality 1 year after the initiation of dialysis therapy. ¹ There has been a lasting controversy on the risks and benefits of the arterio-venous fistula (AVF) on circulatory hemodynamics and cardiac function.^2,3 An AVF acts as a strong left to right sided shunt, with a blood flow of 1–2 L/min.^2,3 Upper-arm AVFs have higher flow than lower-arm AVFs, further increasing the magnitude of the hemodynamic implications.^2,3 The AVF shunt is known to alter cardiac morphology, causing morphological changes such as left ventricular hypertrophy (LVH) and ventricular dilation that can lead to high-output cardiac failure, pulmonary hypertension, and congestive heart failure.^1,4 Despite the significant cardiac implications, there is a paucity of literature describing the timeline for cardiac involvement and the implications for patient care.

We present a prospective case series of eight patients after creation of an upper-arm AVF and quantify changes in their cardiac morphology using cardiovascular magnetic resonance (CMR). We also present a brief review of the relevant literature and explore considerations for clinicians and future directions for this research.

Methods

This prospective case series was approved by the ethics board of Western University. Patients were recruited in 2010 from at multiple centers in London, Ontario, immediately prior to the creation of an upper-arm AVF (brachiocephalic or brachial artery–to–transposed basilic vein fistula). All patients were recruited prior to the initiation of hemodialysis. Patients were selected using the following inclusion criteria: (1) age >18; (2) presence of irreversible chronic kidney failure requiring hemodialysis; (3) <5 days of hemodialysis required per week; (4) ineligible for a forearm fistula. Patients were excluded if they met any of the following exclusion criteria: (1) acute kidney failure that was likely to be reversible; (2) involvement in another study related to VA; (3) life expectancy of <6 months; (4) allergies to contrast; (5) >1 failed VA on the ipsilateral side of the proposed AVF; (6) any contraindication to CMR.

CMR (5T MRI clinical scanners) images were taken at intake before the creation of the AVF, 6-month follow-up, and 12-month follow-up. CMR imaging was performed using electrocardiographic gating during repeated breath holds (approximately 7–15 s). Parallel imaging (ASSET, SENSE, or GRAPPA) with an acceleration factor of 2 was used to reduce image acquisition time. Short-axis cardiac images were taken at 8 mm intervals, beginning 1cm above the atrioventricular annulus to beyond the apex. Imaging parameters were 6mm slice thickness, a 2 mm gap between slices, and 30 phases per cardiac cycle. The same imaging protocols were used to take 2, 3, and 4-chamber images.

Image segmentation

All image segmentation and analysis were performed using ImageJ (ImageJ, Version 1.8.0). End-systolic and end-diastolic phases were determined visually, with the image frame having maximum left ventricular (LV) cavity defined as end-diastole and the image frame having minimum LV cavity area defined as end-systole. Ventricular volumes were determined using the Simpson method, in which manual tracing of the endocardial borders is performed in serial short-axis images from the base of the heart to the apex in both end-systole and end-diastole (Figure 1). The sum of cross-sectional areas in one phase is then multiplied by the slice thickness of 8 mm to yield LV volume. Trabeculation and papillary muscle in endocardial areas were included in LV volumes. If there was a discrepancy in the segment count from one image set to another (e.g. from intake imaging to 6-month imaging), the superior segment in the larger image set was not included in calculations. To determine LV mass (LVM) the difference in the sum of cross sectional areas of the epicardium and endocardium was multiplied by slice thickness of 8 mm to yield myocardial volume. The myocardium was assumed to have an average density of 1.05 g/cm³ and was multiplied by this factor to yield mass. Ejection fraction (EF%) was measured using the formula EF% = $\frac{strokevolume}{end-diastolicvolume}$ × 100.

Figure 1.

Manual segmentation of the left ventricle for summation in the Simpson method.

Left atrial (LA) volumes are taken at end-systole, which is visually identified as the frame prior to the opening of the mitral valve. The cross-sectional area of the LA was measured in both two and four chamber views, excluding the LA appendage and pulmonary veins. The length of the LA was taken in in both two and four chamber views, measured as the perpendicular line from mid-point of the straight line connecting the annulus (Figure 2). LA volume was calculated from these measurements using the Biplane area-length method according to the formula $\frac{8}{3\pi }\left(\frac{A4cxA2c}{L}\right)$ where A4c is the LA area in the four chamber view, A2c is the LA area in the two chamber view, and L is the shortest long-axis length measured in either view.

Figure 2.

Left atrial measurements to be used in the Biplane area-length formula.

Case descriptions and results

Fifty patients with an upper-arm AVF enrolled, but only eight patients met the eligibility criteria (Table 1). The majority of the patients included had significant comorbidities: all eight had hypertension, five had hyperlipidemia, and three had diabetes (Table 1). CMR imaging was done post AVF creation at 6- and 12-months in four patients, three received CMR imaging at 6-months only, and one at 12-months (Table 2). Of the three patients that did not complete the 12-month imaging, two withdrew and one died. All CMR images were taken on non-dialysis days. All eight patients had a net increase in LVM over enrollment, with a mean increase of 9.16 g (+2.96 to +42.66). Five participants had a net decrease in ejection fraction (EF%), with the other three having a net increase in EF (Table 2). Mean change in EF% was −5.4% (−21% to +5%). Notably, the magnitude of EF% reduction was much larger than the magnitude of EF% increase, and the patients with EF% reduction had the largest increase in LVM. Upon visual inspection the patients with the largest EF% decrease had noticeably hypertrophic and dilated ventricles. Finally, LAV results were also variable, decreasing in five participants, while increasing in three participants. There was no observable correlation between LAV and LVM or EF%, with no accompanying visual difference in atrial morphology.

Table 1.

Demographic data and pertinent clinical information of eight patients with planned hemodialysis.

Clinical Data	Patient number
	1	2	3	5	6	9	10	11
Sex	Male	Male	Male	Female	Female	Female	Male	Male
Age	74	84	34	67	61	59	78	51
Weight (kg)	93.3	67.7	105.9	69.2	94.5	56.6	63.0	63.3
Height (m)	1.78	1.73	1.65	1.65	1.78	1.60	1.57	1.75
BMI	39.7	22.6	38.8	25.4	39.8	56.6	63.0	20.7
Race	Caucasian	Black	Caucasian	Caucasian	Caucasian	Caucasian	Caucasian	Caucasian
Comorbidities	Hypertension, hyperlipidemia, diabetes	Hypertension	Depression, obesity, hypertension, peripheral vascular disease, diabetes	Polycystic kidney disease, hypertension, hyperlipidemia, gout, GERD, osteoarthritis	IgA nephropathy, hypertension, hyperlipidemia	Hypertension, diabetes	Congestive heart failure, COPD, asthma, CVA/TIA, coronary artery disease, hypertension, hyperlipidemia, PVD	HIV nephropathy, CAD, hypertension, hyperlipidemia, HIV
Smoking status	Previous	Previous	Active	Previous	Previous	Active	Previous	Active

Table 2.

Cardiac morphology changes at intake, 6 and 12 months imaging sessions. Left ventricular mass (LVM) is measured in grams, ejection fraction (EF%) is measured as a percentage, and left atrial volume (LAV) is measured in milliliters.

	LVM 1	LVM 2	LVM 3	Δ LVM	EF% 1	EF% 2	EF% 3	Δ EF%	LAV 1	LAV 2	LAV 3	Δ LAV
Patient 1	119.13	None	132.44	+13.31 g	55	60	59	+4%	40.04	43.58	50.95	+10.91 mL
Patient 2	104.96	113.04	120.46	+15.50 g	66	53	50	−16%	36.83	30.12	33.43	−3.40 mL
Patient 3	81.79	114.76	115.79	+34.00 g	70	57	57	−13%	50.53	61.47	45.10	−5.43 mL
Patient 4	73.54	80.35	77.42	+3.88 g	56	67	61	+5%	52.26	50.18	45.83	−6.43 mL
Patient 5	136.62	134.78	139.58	+2.96 g	63	57	62	−1%	50.03	45.32	39.88	−10.15 mL
Patient 6	131.13	135.56	None	+4.43 g	63	59	None	−4%	46.56	37.31	None	−9.25 mL
Patient 7	207.35	210.62	None	+3.27 g	76	79	None	+3%	42.86	65.21	None	+22.35 mL
Patient 8	157.50	199.83	None	+42.33 g	61	40	None	−21%	39.04	45.77	None	+6.73 mL
Mean				+9.16 g				−5.4%				+0.67 mL

The results above may be influenced by the patency outcomes of the AVFs. Five patients maintained the AVF patency. Patients 3 and 6 lost AVF patency 3-months after enrollment, with a new AVF created 7-months after enrollment (1-month after the second imaging date). Patient 4 lost AVF patency 1-month after enrollment, but a second AVF was created 2-months after enrollment (4-months before the second imaging date).

Discussion and review of the literature

This case series represents the first in literature to prospectively measure cardiac morphology changes immediately following upper-arm AVF creation. Left ventricular hypertrophy was observed in all of our cases on CMR segmentation, accompanied by visual changes to LV morphology. As LVM is a strong and sensitive predictor of cardiovascular mortality, these findings have relevant implications for patient outcomes. Although it is plausible that both systolic and diastolic cardiac functions could be affected by the hemodynamic changes that accompany AVF creation, we could not confirm this with the variability in morphological changes in EF% and LAV.

The cardiac implications of AVF creation have been previously identified, including changes such as LVH, high-output cardiac failure, central vein stenosis, congestive heart failure, pulmonary hypertension, and pulmonary arterial hypertension.^1,3,4 We observed significant LVH in our cases at our 6-month imaging, even in patients who had lost AVF patency 3-months prior to imaging. This timeline for cardiac remodeling is rapid, and given the prevalence of cardiovascular mortality in the dialysis population, it could directly influence morbidity and mortality. Despite the rapid changes in cardiac morphology attributed to AVF creation, AVFs remain the first choice for vascular access among those who are suitable for an AVF creation. ⁵ The benefits of the AVF including reduced rates of infections and durable patency rates must be balanced with possible cardiac implications. ⁶

The management of the AVF in a patient post kidney transplant has provided insight into these cardiovascular risks. A 2019 randomized control trial by Rao et al. ³ examined the effects of AVF ligation after kidney transplantation in 54 patients, finding a statistically and clinically significant reduction of left ventricular myocardial mass in the ligation group. As LVM is a strong predictor of adverse cardiovascular events and has often been used as a surrogate endpoint for cardiovascular mortality, this team predicted this change could reduce mortality in the transplant population. ³ Hetz et al. ⁶ studied AVF ligation in a transplant population in 2020, but instead looked at incidence of high-output heart failure. Five of 13 control patients had high-output heart failure during the follow-up period, while none of the 13 ligation group patients had high-output heart failure. ⁶

Despite these demonstrated instances of ligation improving cardiovascular outcomes, there is currently no physician consensus on vascular access management after kidney transplantation—vascular access is often maintained as it may be required should a resumption of hemodialysis be indicated. ⁷ There are also currently no studies examining the holistic effects of ligation in the transplant population, but given the timeline of cardiovascular change presented in this case series and the known benefits of AVF ligation, further investigation is warranted.

Beyond the implications for cardiovascular health and for resumption of dialysis, other complex physiological processes should factor into clinical decision making regarding AVF management. There is growing literature to suggest AVFs may have nephroprotective function independent of dialysis. A 2016 cohort study of 3026 veterans with advanced chronic kidney disease found a significant deceleration of estimated glomerular filtration rate (eGFR) decline after vascular access creation in those with AVF or arterio-venous grafts (AVG) when compared to catheters. ⁸ These findings were analogous to studies by Golper et al. ⁹ and Dupuis et al. ¹⁰ who also found that AVF creation was associated with a deceleration of kidney function decline. Given the significant mortality risk in the ESKD patients from both renal and cardiovascular causes the benefit of AVF ligation and maintenance presents a complex clinical challenge.

This study has three major limitations. The first is a small sample size of eight. This is exacerbated by the medical needs of the target population, as most patients who are at hospital multiple days per week for dialysis are not eager to enroll in research. The second is three of the participants lost AVF patency with new AVFs creation at variable times. While the AVFs of patients could plausibly have had cardiac implications, the quantification of the changes is less accurate. Finally, the patients here had upper-arm AVFs. Upper-arm AVFs have significantly higher flow than lower-arm AVFs, and could subsequently have higher magnitude effects on cardiac morphology.

Conclusion

In summary, we present a prospective case series of eight patients cardiac imaging 12-months following upper-arm AVF creation. All patients included had increases in left ventricular mass, with variability in the effects on the ejection fraction and left atrial volume. Given the cardiovascular implications of the changes, further research to determine appropriate vascular access management in both ESKD and kidney transplant populations is warranted.