Arteriovenous fistula geometry in hybrid recanalisation of post-thrombotic venous obstruction

Abstract

Introduction

Post-thrombotic obstruction can be adequately treated by percutaneous transluminal angioplasty and stenting. When post-thrombotic trabeculations extend below the femoral confluence, proper inflow can be facilitated by endophlebectomy and creation of an arteriovenous fistula. The aim of this study was to investigate whether it is more favourable to place the arteriovenous fistula at the cranial or caudal end of the endophlebectomy to prevent stenosis or occlusion.

Methodology

We retrospectively analysed the clinical data of all patients who underwent a hybrid procedure in our two centres. Demographics, interventional details and post-operative imaging were collected.

Results

Data on 42 limbs with cranially and 23 limbs with caudally placed arteriovenous fistulas were collected. Post-thrombotic disease of the profunda femoral vein alone or in combination with the femoral vein was observed more often in the cranial group. The caudal group more often received a smaller sized and straight polytetrafluoroethylene fistula, while the cranial group comprised a significantly higher amount of stented segments. Logistic regression showed that only reduced femoral inflow (hazard ratio 2.934 (95%CI, 1.148–7.494)) was a significant predictor of stent stenosis and/or occlusion. Logistic regression for risk of occlusion showed a significant influence of stent-related complications (hazard ratio 4.691 (95%CI, 1.205–18.260)) and a tendency towards influence of arteriovenous fistula geometry in favour of the cranially placed fistula.

Conclusion

Placement of the arteriovenous fistula in the cranial part of the endophlebectomy during hybrid recanalisation may result in a more favourable outcome, yet this tendency was not statistically significant. Moreover, femoral inflow is pivotal in maintaining patency and should thus be adequately assessed pre-operatively.

Keywords

Arteriovenous fistula deep venous obstruction stent post-thrombotic syndrome endophlebectomy endovenectomy occlusion

Introduction

Post-thrombotic obstruction of the iliocaval veins can be adequately treated by percutaneous transluminal angioplasty (PTA) and stenting. Several studies with large populations have shown good technical success and patency rates with follow-up as high as 72 months.^1–4 Furthermore, quality of life and venous clinical severity score have been shown to significantly improve.¹ However, successful treatment becomes more difficult if post-thrombotic disease extends below the inguinal ligament, particularly the femoral confluence.⁵ These types of patients can be treated by endophlebectomy of the common femoral vein (CFV) and orifices of all significant veins draining into the CFV, in addition to PTA and stenting of the more proximal tract.^4,6–8 To ensure proper inflow of the recanalised tract and prevent early thrombotic complications, an arteriovenous fistula (AVF) can be placed between the common femoral artery (CFA) and the CFV.^4,9–11

Use of such an AVF can be counterproductive, though, since stenosis or occlusion in the region of the venous anastomosis of the AVF has been observed in our centre. In the past, we placed the AVF caudally from the endophlebectomy to ensure good flow in the complete recanalised, i.e. stented and endophlebectomy, tract, yet due to patency-related problems in the endophlebectomy area, possibly due to intimal hyperplasia, comparable to the problems seen in access surgery, we started to place the AVF cranially in the endophlebectomy area at the distal end of the stented tract. The idea was that, by performing it in this region, the distance between the anastomosis and the opposite vein wall would be greater, and therefore a reduction in the potential intimal hyperplasia, caused by areas of low shear stress due to the created arterial inflow, was expected. (Figure 1) The aim of this study was to evaluate whether this theoretically more favourable position of the AVF at the cranial end of the endophlebectomy would indeed prevent stenosis and/or occlusion and improve the patency of the complete recanalised tract.

Methods and materials

We retrospectively analysed all patients who, during the last five years, underwent a hybrid venous recanalisation of the caval, iliac and/or CFV in the Maastricht University Medical Centre and University Hospital Aachen. Patients who received an AVF for another purpose, such as in addition to bypass surgery or thrombectomy, were excluded from analysis, as were patients who had one or less post-operative follow-up visits with imaging. In patients who received an AVF bilaterally, we decided to solely include the right limb since only independent samples can be used for logistic regression analysis.

Hybrid procedures are performed under general anaesthesia with the patient in the supine position. Access to the femoral vein (FV) of the affected limb is obtained through a 10-French sheath. Subsequently, various types of guiding sheaths, catheters and guidewires are used to recanalise the post-thrombotic tract until a healthy vessel lumen is reached. In cases where recanalisation from the FV is impossible, access is obtained via the contralateral FV or right jugular vein. After successful recanalisation, a groin incision is made and the CFA and CFV with their confluences and all side branches are identified. Then, the CFV is longitudinally opened and intraluminal synechiae and septae are removed from the CFV and orifices of the FV, profunda femoral vein (PFV) and other large branches draining into the CFV to optimise inflow. The CFV is closed either primarily or with the use of a patch, depending on post-venectomy vessel diameter, after which PTA and stenting of the obstructed tract is performed with the most distal part of the stents landing right above or in the top of the endophlebectomy area. In between the closure of the CFV and stenting of the obstructed tract, an AVF is constructed. Creation of the AVF was not always performed in the same fashion though. AVFs have been created in our centre using a ring-enforced polytetrafluoroethylene (PTFE) loop, a straight non-enforced PTFE tube, or a side branch of the great saphenous vein to connect the CFA with the CFV, either cranially (Figures 2 and 3) or caudally (Figure 4) in the endophlebectomy area.

Perioperative data concerning the type of AVF and location of its placement were collected, as were data pertaining to the extent of stenting and closure of the venotomy. Follow-up imaging data, i.e. duplex ultrasonography (DUS) or four-plane abdominal X-ray, were collected to evaluate the occurrence of stenoses, occlusions or stent configuration related problems. Pre-operative data regarding the C-score according to the CEAP-classification, previous deep venous interventions and potential thrombophilia were also obtained. The FV and PFV were scored as either healthy or post-thrombotic based on DUS and magnetic resonance venography (MRV).

Statistics

Statistical analysis was performed using IBM SPSS statistics version 21.0.0.0 (IBM Corporation, Armonk, NY, USA). Continuous variables are expressed as mean values with concomitant standard deviations, while categorical data are expressed using percentages. The χ² test or independent samples t test was used to test for baseline differences between groups. Fisher’s exact test was used in cases where the expected count of one or more cells was less than five and the Mann–Whitney U test was used in cases where variables were not normally distributed. Multiple logistic regression analysis was performed to determine the influence of several variables on the risk of stenosis or occlusion. Kaplan–Meier survival analysis was performed using GraphPad Prism version 5.04 (GraphPad Software, San Diego, CA, USA). Loss of primary patency was defined as occlusion of the treated tract or an additional procedure to prevent occlusion, loss of assisted primary patency as occlusion of the treated tract after additional interventions to prevent occlusion, and loss of secondary patency as occlusion after initially successfully treated re-occlusion. Reduced femoral inflow was defined as post-thrombotic changes in the FV, PFV, or FV and PFV, as diagnosed on DUS and MRV. This was scored as follows: 0 if both veins were not affected, 1 if either the FV or PFV was post-thrombotic and 2 if both veins were post-thrombotic.

Results

A total of 78 limbs in 70 patients were selected. Two limbs were excluded due to insufficient pre-operative data concerning the placement of the AVF, one limb was excluded due to technical failure of the endophlebectomy and eight left limbs of patients with bilateral AVF construction were excluded in order to be able to perform multiple logistic regression analysis. Anastomosis of a native vessel to the AFC was performed in two limbs, which were therefore excluded, leaving 65 limbs in 65 patients for analysis.

Two groups were created: patients with an AVF placed cranially in the endophlebectomy area (group 1, n = 42), and patients with an AVF placed caudally in the endophlebectomy area (group 2, n = 23) (Figures 3 and 4). Patient characteristics were comparable between the groups, except for post-thrombotic involvement of the PFV (54.8% vs. 21.7%, p = 0.010), post-thrombotic involvement of both the FV and PFV (50.0% vs. 21.7%, p = 0.026) and number of C₂-classified patients (CEAP classification, 17.1% vs. 0%, p = 0.010) (Table 1). Intervention details were also comparable, except mean number of stented segments (3.4 vs. 2.8, p = 0.002), PTFE size (88.1% 6 mm vs. 56.5%, p = 0.004) and PTFE shape (97.6% loop vs. 60.9%, p < 0.001) (Table 2). Mean follow-up was 6.9 ± 5.0 months for group 1 compared to 17.8 ± 8.6 months for group 2. During follow-up, problems with stent configuration such as tapering or kinking were found in five patients within the cranial group (11.9%) compared to six in the caudal group (26.1%, p = 0.176).

Table 1.

Patient characteristics.

	Cranial group (n = 42)	Caudal group (n = 23)
Sex (M/F)	15/27 (35.7%/64.3%)	10/13 (43.5%/56.5%)	p = 0.538
Mean age (years ±SD)	39.5 ± 16.6	38.8 ± 11.6	p = 0.778
Thrombophilia	12 (70.6%) Out of 17 tested	5 (62.5%) Out of 8 tested	p = 1.000
Previous deep venous intervention	3 (7.1%)	1 (4.3%)	p = 1.000
Post-thrombotic changes below femoral confluence
PFV	23 (54.8%)	5 (21.7%)	p = 0.010^a
FV	37 (88.1%)	19 (82.6%)	p = 0.709
Both PFV and FV	21 (50.0%)	5 (21.7%)	p = 0.026^a
C-class of CEAP^b
C₀	4 (9.8%)	2 (8.7%)	p = 1.000
C₁	5 (12.2%)	–	p = 0.152
C₂	7 (17.1%)	–	p = 0.010^a
C₃	16 (39.0%)	12 (52.2%)	p = 0.273
C_4a	6 (14.6%)	3 (13.0%)	p = 1.000
C_4b	–	2 (8.7%)	p = 0.137
C₅	2 (4.9%)	1 (4.3%)	p = 1.000
C₆	1 (2.4%)	3 (13.0%)	p = 0.123

FV: femoral vein; PFV: profunda femoral vein; CEAP: Clinical-Etiology-Anatomy-Pathophysiology classification.

Statistically significant.

Unknown for one patient in the cranial group.

Table 2.

Intervention details.

	Cranial (n = 42)	Caudal (n = 23)
Intervention type
Unilateral iliofemoral	28 (66.7%)	18 (78.3%)	p = 0.326
Bilateral iliofemoral	2 (4.8%)	–	p = 0.536
Unilateral iliofemoral and IVC	1 (2.4%)	1 (4.3%)	p = 1.000
Bilateral iliofemoral and IVC	11 (26.2%)	4 (17.4%)	p = 0.421
Mean number of stented segments	3.4 ± 0.6 (3–5)	2.8 ± 0.7 (2–4)	p = 0.002^a
Patch	26 (61.9%)	12 (52.2%)	p = 0.447
PTFE size (5 mm/6 mm)	5/37 (11.9%/88.1%)	10/13 (43.5%/56.5%)	p = 0.004^a
PTFE shape (loop/straight)	41/1 (97.6%/2.4%)	14/9 (60.9%/39.1%)	p < 0.001^a

IVC: inferior vena cava; PTFE: polytetrafluoroethylene.

Statistically significant.

At six months, primary patency was 48%, assisted primary patency 63% and secondary patency 78% for the group who underwent the cranial technique. Primary patency was 56%, assisted primary patency 60% and secondary patency 72% in the caudal group, which were not significantly different from group 1 (p = 0.616, p = 0.683, p = 0.894, respectively) (Figure 5).

Multiple logistic regression analysis for the risk of any patency-related event with correction for PTFE size, PTFE shape (loop or straight), reduced femoral inflow, type of venotomy closure, number of stented segments, and stent complications did not yield a significant difference between the two surgical techniques as the hazard ratio (HR) for the caudal technique was 2.560 (95%CI, 0.650–10.087) compared to the cranial technique. Only reduced femoral inflow was found to be predictive of patency-related problems with an HR of 2.934 (95%CI, 1.148–7.494) per affected vein, i.e. PFV or FV (Table 3). PTFE shape showed collinearity with PTFE size in this regression model, thus only the latter variable was used for this analysis.

Table 3.

Hazard ratios for loss of patency of the recanalised tract.

	Any event (occlusion or stenosis)		Occlusion
Caudal compared to cranial AV fistula	2.560	(95%CI, 0.650–10.087)	3.793	(95%CI, 0.799–18.472)
Straight compared to loop shaped AV fistula	–^a		0.895	(95%CI, 0.059–13.620)
6 mm compared to 5 mm PTFE AV fistula	0.311	(95%CI, 0.066–1.459)	0.298	(95%CI, 0.047–1.882)
Reduced femoral inflow^b	2.934^c	(95%CI, 1.148–7.494)	3.391	(95%CI, 0.911–12.617)
Patch compared to primary closure of venotomy	0.900	(95%CI, 0.285–2.838)	0.948	(95%CI, 0.239–3.767)
Number of stented segments^d	1.787	(95%CI, 0.706–4.522)	2.776	(95%CI, 0.941–8.184)
Problems with stent configuration	1.148	(95%CI, 0.227–5.797)	4.691^c	(95%CI, 1.205–18.260)

AV: arteriovenous; PTFE: polytetrafluoroethylene.

Eliminated due to collinearity.

Hazard ratio (HR) per affected femoral vein.

Statistically significant.

HR per stented segment.

Multiple logistic regression analysis for the risk of occlusion also did not show a statistically significant difference between the two surgical groups (HR of 3.793 (95%CI, 0.799–18.472) for group 2). Based on backward likelihood ratio test analysis, all variables but for stent-related complications like kinking, compression and tapering could be removed from the model without changing it significantly, which was found to influence outcome significantly (HR 4.691 (95%CI, 1.205–18.260)) (Table 3).

Discussion

Both risk of stenosis/occlusion and risk of occlusion alone were not significantly different between patients who received an AVF at the caudal end of the endophlebectomy and those who received one at the cranial end, yet a tendency in favour of the cranial technique is noted. Due to the many variables influencing patency, this study was most likely underpowered to prove a significant effect. The high HR for the caudal method might perhaps be explained by de novo intimal hyperplasia of the vein near the anastomosis of the AVF, which has been described before after vascular access surgery.^12,13 If the AVF is placed caudally in the endophlebectomy area, it is conceivable that such intimal hyperplasia could encumber already poor femoral inflow and gradually lead more easily to a stenosis and/or occlusion, while a cranially placed AVF could, due to the increased distance to the opposite vein wall, potentially lead to less stenosis and/or occlusion. Impaired inflow due to intimal hyperplasia could cause occlusion of the complete stented tract, which would potentially add to the already impaired inflow due to FVs or PFVs with post-thrombotic lesions, as the risk of stenosis or occlusion was almost three times higher in patients with diseased FVs or PFVs compared to patients with healthy FVs/PFVs. This is not surprising, since poor inflow is logically related to a higher risk of occlusion, as has been discussed in literature before.^{9–11,14–16}

Closure with the use of a patch did not significantly affect outcome, which could have been anticipated since the patch was only used when the vein was considered too narrow for primary closure. A patch creates extra space to prevent stricture of the vein and thus does not limit inflow or outflow. Conversely, neither does primary closure necessarily constitute risk of occlusion, since the surgeon deemed the vein to be large enough for primary closure. PTFE size was not a factor influencing patency either. A certain amount of arterial inflow is needed to ensure proper stent inflow, thus it is likely that beyond such point any additional arterial inflow is not relevant anymore to safeguard patency. Both 5-mm and 6-mm AVFs are assumedly sufficient to reach that point, which is consistent with previous research.¹⁷ Since placing the PTFE in a loop or a straight did not influence results, it is apparently not relevant whether flow enters the vein in an antegrade or retrograde direction. Ostensibly, outflow of the lower limb does not seem to suffer under the initial different flow direction of the blood from the AVF.

Similarly, we did not find the number of stented segments to be of influence. However, given the amount of variables tested in relation to the amount of events and the distribution of the confidence interval, it is likely that this study was underpowered to properly test for such effects. Number of stented segments could be indirectly related to outcome, though, since it conveys extent of disease. Stent-related complications were found to be of significant influence, which is not surprising since complications such as kinking, residual compression and tapering impair optimal flow through the stented tract and thus contribute to risk of stent occlusion; a problem which is probably addressed by new dedicated venous stents.¹⁶

Additionally, continuous compression of the endophlebectomy area, due to low venous pressures inside the vein, after closure of the inguinal wound might also form an important problem in maintaining patency, hence measures to prevent collapse of the endophlebectomy area could be considered. An endovenous device preventing such collapse may perhaps improve patency rates in the future. Furthermore, instead of using an AVF, possibilities of developing less invasive techniques that can optimise flow within the treated tract during the first couple of weeks after treatment need also be explored.

Due to the retrospective nature of this study, some limitations have to be taken into account. First of all, we were not able to correct for post-operative anticoagulation since a separate institution in our region monitors and regulates this. However, we have a standard post-operative regimen to anticoagulate with coumarin for at least six months. In the initial phase after surgery, patients also receive therapeutic low molecular weight heparins (LMWH) until the international normalised ratio (INR) has reached its target ratio of 3.0–4.0. Afterwards, if the INR drops below 2.5, a single therapeutic dose of LMWH is again given daily, until the INR returns within its specified range. Furthermore, most patients were not tested for thrombophilia, thus we could not correct for such disease. Second, patient characteristics and specifications of the intervention were not always equally distributed over the two groups. Whilst PTFE size and shape might have had some minor impact, the number of stented segments showed a tendency in increased risk for occlusion and reduced femoral inflow had a significant influence on risk of stenosis. The skewed distribution of the latter two factors might therefore have contributed to the fact that we did not find a statistically significant difference in method of AVF placement. Third, not all variables were properly registered, leading to exclusion of two patients due to lack of perioperative details concerning the type of AVF. For all included patients, only one patient had a missing baseline C-score though. Finally, some bias may be present due to the chronologic nature of the study. When we started performing these hybrid procedures, we typically placed the AVFs caudally, after which a period came when both methods were used, followed by sole placement of the cranial AVFs. Therefore, confounding due to change in intensity of follow-up, improved experience and improvement of stent design could be present. However, no correction is possible for the first two and the different combinations of stent types used were too comprehensive for correction in this population. Conversely, we did correct for stent complications, which likely obviates any significant effects different types of stent might have had.

In conclusion, placement of an AVF at the cranial end of the endophlebectomy during hybrid venous recanalisation may result in a more favourable outcome than placement at the caudal end of the endophlebectomy area, though this tendency was not yet shown to be statistically significant. Moreover, femoral inflow is found to be pivotal in maintaining patency. Hence, it is crucial to adequately assess the condition of inflow providing vessels before attempting recanalisation of the obstructed tract.

Footnotes

Acknowledgements

The authors would like to thank Patricia Joan “Patty” Nelemans, MD, PhD for her help in clearing up any statistical issues.

Conflict of interest

None declared.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

References

Neglen

Hollis

Olivier

. Stenting of the venous outflow in chronic venous disease: long-term stent-related outcome, clinical, and hemodynamic result. J Vasc Surg 2007; 46: 979–990.

Lou

Jalaie Gu

. Endovascular treatment for iliac vein compression syndrome: a comparison between the presence and absence of secondary thrombosis. Korean J Radiol 2009; 10: 135–143.

Hartung

Loundou

Barthelemy

. Endovascular management of chronic disabling ilio-caval obstructive lesions: long-term results. Eur J Vasc Endovasc Surg 2009; 38: 118–124.

de Wolf

Arnoldussen

Grommes

. Minimally invasive treatment of chronic iliofemoral venous occlusive disease. J Vasc Surg: Venous Lym Dis 2013; 1: 146–153.

Neglen

Tackett

Jr. Raju

. Venous stenting across the inguinal ligament. J Vasc Surg 2008; 48: 1255–1261.

Comerota

Grewal

Thakur

. Endovenectomy of the common femoral vein and intraoperative iliac vein recanalization for chronic iliofemoral venous occlusion. J Vasc Surg 2010; 52: 243–247.

Puggioni

Kistner

Eklof

. Surgical disobliteration of postthrombotic deep veins—endophlebectomy—is feasible. J Vasc Surg 2004; 39: 1048–1052.

Garg

Gloviczki

Karimi

. Factors affecting outcome of open and hybrid reconstructions for nonmalignant obstruction of iliofemoral veins and inferior vena cava. J Vasc Surg 2011; 53: 383–393.

Vogel

Comerota

Al-Jabouri

. Common femoral endovenectomy with iliocaval endoluminal recanalization improves symptoms and quality of life in patients with postthrombotic iliofemoral obstruction. J Vasc Surg 2012; 55: 129–135.

10.

Jost

Gloviczki

Cherry

Jr. . Surgical reconstruction of iliofemoral veins and the inferior vena cava for nonmalignant occlusive disease. J Vasc Surg 2001; 33: 320–327.

11.

Nazarian

Austin

Wegryn

. Venous recanalization by metallic stents after failure of balloon angioplasty or surgery: four-year experience. Cardiovasc Interventional Radiol 1996; 19: 227–233.

12.

Allon

Litovsky

Young

. Medial fibrosis, vascular calcification, intimal hyperplasia, and arteriovenous fistula maturation. Am J Kidney Dis 2011; 58: 437–443.

13.

Roy-Chaudhury

Arend

Zhang

. Neointimal hyperplasia in early arteriovenous fistula failure. Am J Kidney Dis 2007; 50: 782–790.

14.

de Wolf

Arnoldussen

Wittens

. Indications for endophlebectomy and/or arteriovenous fistula after stenting. Phlebology 2013; 28 (Suppl 1): 123–128.

15.

Rosales

Sandbaek

Jorgensen

. Stenting for chronic post-thrombotic vena cava and iliofemoral venous occlusions: mid-term patency and clinical outcome. Eur J Vasc Endovas Surg 2010; 40: 234–40.

16.

Jalaie H, Arnoldussen C, Barbati M, et al. What predicts outcome after recanalization of chronic venous obstruction: hemodynamic factors, stent geometry, patient selection, anticoagulation or other factors? Phlebology 2014; 19; 29(1 suppl): 97–103.

17.

Menawat

Gloviczki

Mozes

. Effect of a femoral arteriovenous fistula on lower extremity venous hemodynamics after femorocaval reconstruction. J Vasc Surg 1996; 24: 793–799.