A well functioning access is integral to the delivery of adequate hemodialysis (HD). The most preferred vascular access is the arteriovenous fistula (AVF), followed by the arteriovenous graft (AVG), with the tunnelled cuffed catheter (TCC) as the least desirable. Commonly recognized complications of catheter use include infection, thrombosis, and inadequate dialysis delivery. Another common but potentially under recognized complication is central vein stenosis (CVS). This particular complication leads to devastating long-term consequences including poor dialysis delivery caused by recirculation, impaired AVF maturation, decreased long-term patency rates, and superior vena cava (SVC) syndrome. Despite recent evidence supporting increased mortality and morbidity associated with catheters,1 there is an alarming trend of increased catheter use in dialysis. According to USRDS data, the use of permanent catheters in incident patients on hemodialysis has increased by 71% (per 1,000 patient years at risk) from 1996 to 1999.2 In our HD unit, use of catheters is 70% in incident patients3 and 30% in prevalent patients (J. M. MacRae and M. Kiaii, unpublished data, 2002).
The site of catheter placement is an important determinant for the development of CVS. Studies have shown that subclavian vein (SCV) cannulation is associated with higher rates of CVS.4,5 The current K/DOQI recommendations state that the internal jugular (IJ) vein should be preferentially cannulated.6 However, there are no recent studies reevaluating the frequency of CVS using these guidelines.
The purpose of this study is to document the prevalence of CVS in a prevalent HD population in whom a venogram was performed because of AVF related clinical concerns in an era of preferential IJ catheter placement.
We conducted a systematic evaluation of a prevalent cohort of HD patients in a single center, tertiary care hospital. All were receiving regular dialysis three times per week, 4 hours per session, during the study period of January 2002 to March 2003. The unit has a well established vascular access monitoring program, with protocols to investigate access related problems.
The vascular access monitoring program includes 6–8 weekly transonic measurements, regular assessments of venous and arterial pressures, and visual inspection of the fistula and surrounding structures. The clinical criteria for venogram in this unit include the following: decreased access blood flow using the saline dilution technique (HD02 Monitor, Transonic Inc, Ithaca, NY), increased venous pressures, impaired access maturation, and arm or face edema. Increased venous pressure is defined as a venous pressure greater than 150 mm Hg on three consecutive runs at a blood pump speed of 200 ml/min with a 15 gauge needle. Access flow values are obtained within the first 60 minutes of dialysis to ensure stable hemodynamics; if the blood pressure was greater than or equal to 10% lower than previous values, the access flow was deferred until the next dialysis session. Decreased access flow blood is defined as an access flow less than 500 ml/min in an AVF or a decrease in access flow of 20% or more from baseline; all access flows were repeated twice, and the average value was taken.
Patients were selected for inclusion if they were currently dialyzing with an AVF and had undergone venography for investigation of AVF related problems in accordance with our unit protocol (see previous section).
All venograms were performed at a single tertiary radiology department. All studies visualized both the fistula veins and central veins. CVS was defined as a 50% or more narrowing of the central veins of the thorax that included the SVC, brachiocephalic vein (BCV), SCV, and subclavian-cephalic vein junction (SCV-CV).
Data entered into a computerized database included the following: patient demographics, catheter history, and laboratory parameters. Catheter history included all hemodialysis catheters in the patient’s history. The types of HD catheters used at our institution include temporary catheters (Niagara, Vas-Cath Inc.) and tunnelled cuffed catheters (Permcath, Quinton Instruments Co, Seattle, WA). History of pacemaker insertion (a risk factor for CVS) was also documented. Venography results were tabulated as to location of central vein and peripheral vein stenosis.
Statistical analysis included a univariate analysis using chi-square for categorical variables and wilcoxon 2 sample test for continuous variables. All values are described as a mean unless otherwise stated. Logistic regression analysis is adjusted for variables such as duration of catheter use and increased number of catheter insertions, both of which have been associated with CVS in previous studies.
Of the 235 prevalent patients on chronic HD (153 AVF, 70 catheters and 12 AVG), 136 had undergone venography according to our clinical protocol. Three patients were excluded from this analysis because of lack of venogram data regarding central veins. Thus 133 were evaluated for the purposes of this review. Of this study population, 100 patients had a peripheral vein stenosis on venogram, and 41% (55/133) had CVS. Peripheral stenosis was commonly found in both CVS + (42/55, 76%) and CVS − patients (58/78, 74%). No patient had CVS at more than one of the four sites mentioned. The anatomic distribution of CVS is shown in Figure 1. The most frequent site of stenosis was the SCV-BCV junction at 38%, and the SVC stenosis accounted for 24%.
Baseline Characteristics and Central Vein Stenosis
The vascular access distribution in the study patients was as follows: 96 AVF, 33 catheters, and 4 AVG. Table 1 demonstrates patient characteristics between those patients with CVS and those patients without CVS. There were no differences with respect to age, gender, diabetes, or anticoagulation medication. However, those patients with CVS had a significantly longer duration on HD as compared with those without CVS. Furthermore, the history of a hemodialysis catheter insertion is associated with CVS: 95% of the CVS group had an HD catheter insertion compared with 76% of the non-CVS group (p = 0.0039). There was a group of patients in whom there was no history of a hemodialysis catheter insertion (n = 22), and yet (13.6%) three of these patients had CVS.
Among those patients who did have a history of HD catheter insertion, there were no differences in any demographic variable including diabetes and duration of HD between those who did and did not have CVS documented on venography.
Catheter Related Factors
Catheter related factors associated with the development of CVS were determined and shown in Table 2. In those patients with a history of HD catheter insertion (n = 111), increased cumulative days with an indwelling catheter (p < 0.0001) and a history of multiple catheter insertions (p < 0.0001) were associated with CVS. All five patients who had a subclavian catheter as the site of first insertion had evidence of CVS on venogram. Furthermore, in the 18 patients who had a subclavian catheter insertion at any time, 83% (15/18) had CVS; this is in comparison with the 36% (40/111) who never had a subclavian catheter insertion and yet had CVS (p = 0.0001). However, it appears that internal jugular insertions are associated with the same risk of developing CVS because, after excluding all 18 patients who were exposed to a subclavian catheter, the number of catheter days remained longer in the CVS patients (p = 0.0002), and there were more patients with a history of multiple catheters and multiple insertions into the same site (p = 0.0067, p = 0.0095) in these patients.
Of all the patients with a history of catheter insertion, 42 (38%) had a noncuffed catheter (Niagra, Vas-Cath Inc.) inserted. However, exposure to a noncuffed catheter was not associated with an increased risk of CVS (55% of patients with a noncuffed catheter had CVS compared with 45% who did not, p = 0.19). Again the results were similar after excluding the patients with a history of subclavian catheter. Of the two CVS patients with a history of pacemaker insertion, both had a history of HD catheter insertion.
In those patients with a history of HD catheter insertion, multivariate analysis of patient related factors (age, gender, duration on hemodialysis, cumulative months with a catheter, and number of catheters) associated with CVS was performed. The only statistically significant factor associated with CVS was the number of catheters (OR 2.69, p = 0.0004), which remained significant after excluding the subclavian catheters (OR 3.48, 1.70–7.2). In addition, with the subclavian catheters excluded we found that increased duration on HD was also significant (OR 1.54, 1.15–2.1).
Relationship Between Central Vein Stenosis and Access Location
Of the CVS patients who had an AVF (40), there seems to be a significant relationship between side of access and presence of ipsilateral CVS (p = 0.01). Specifically, 16 of 26 patients with a left AVF had left CVS, and 11 of the 14 of patients with a right AVF had a right CVS. Of the 11 AVF patients with a right ipsilateral CVS, 10 had a previous right catheter. In the 16 AVF patients with a left ipsilateral CVS, only 4 had a previous left catheter.
In the CVS patients with a catheter (15), there is a trend for the patients with a right catheter (10) to be more likely to have a right CVS (8/10); however, the 5 patients with a left catheter did not have more frequent ipsilateral CVS (1/5). This may reflect the fact that many of the patients with a left catheter had had previous cannulation of their right sides (4/5).
Clinical Indications Prompting Venography and Central Vein Stenosis Diagnosis
To delineate the clinical features that may be associated with CVS, we evaluated the clinical criteria that prompted venography. The criteria for venogram differed between those patients who had CVS compared with those who did not (p = 0.0024). As expected, arm edema as an indication for venogram was most likely to be associated with CVS; of the 13 patients with arm edema, 10 had CVS. The remaining indications for venogram included the following: increased venous pressure (n = 60, 30CVS+, 30CVS−), decreased access flow (n = 28, 5 CVS+, 23 CVS−) impaired access maturation (n = 16, 7CVS+, 9 CVS−), and “other” indications (n = 16, 3CVS+, 13 CVS−). Of the 23 CVS, 15 patients had a peripheral stenosis to account for their decreased access flow.
Six month follow up data on the CVS patients (n = 55) yield that 10 died and 20 continued to have CVS related problems such as repeated AVF failure (13) (including nonmaturation and thrombosis), inadequate dialysis with a Kt/V less than 1.2 (3), and severe arm edema (4). Of the previously mentioned 13 patients, 2 eventually required a thigh graft because of the inability to create a functioning access in the upper limbs, and 7 are reliant upon permanent catheters. In this study, a significant number of patients (13, 24%) developed SVC stenosis. Eight of these patients had clinical evidence of SVC syndrome involving (facial edema and dyspnea), and two required intubation and subsequently died on a ventilator.
CVS in the current dialysis era with new catheter materials, and in the context of preferential IJ insertion, requires systematic evaluation. Our study demonstrates that CVS may be a common occurrence in patients on HD with access problems, more frequently than previously recognized. Although this study is not able to determine the true incidence of this complication, it does demonstrate the prevalence of the disorder in patients with problematic vascular access. Our finding of CVS in the absence of classic SVC obstruction symptoms and in patients both with and without catheter insertion is important and novel.
Treatment strategies for established CVS have been described,7,8 although usually in the context of SVC syndromes. Thus it is not yet established that all of the lesions that we describe here are amenable to repair. However, the recognition and long-term follow up of these patients may lend further insights over time. The identification and repair of CVS may, therefore, improve the longevity of an AVF and maintain its use as the preferred access. This study helps to increase awareness of the prevalence of CVS in a contemporary population of patients on HD.
This study confirms, as others have demonstrated, that the number of catheters inserted and increased duration of catheter days are associated with the development of CVS.4,9–11 In fact, our analysis demonstrates that exposure to multiple catheters produces an almost threefold increase in risk in the development of CVS and highlights the need to reduce catheter use in the dialysis population. However, unlike previous studies, we have demonstrated that, in an era of preferential cannulation of the IJV, CVS remains pervasive. Literature suggests that cannulation of the SCV is associated with up to 50% stenosis rates.4,5 These observations led to the idea that cannulation of the IJV would not be associated with CVS and that it should be the preferred site of cannulation.5,12 However, as our study shows, cannulation of the central veins, regardless of the site, is associated with a high frequency of CVS.
Given the strong association of any hemodialysis catheter insertion with CVS, irrespective of location, it may be prudent to incorporate the systematic screening of patients with a history of catheter insertion for assessment of central vein anatomy before permanent access creation.13
The etiology of CVS remains complex and is likely related to a number of factors, including the following: 1) mechanical injury from either repeated catheter insertion or continuous catheter movement inside the vein invoking endothelial damage, subsequent inflammation, intimal hyperplasia, and fibrosis; 2) catheter or AVF related changes in the flow dynamics leading to increased shear stress, platelet aggregation, and intimal hyperplasia; or 3) a combination of the previous two factors in the presence of patient specific factors. The possibility that the changes in flow dynamics mitigated by the presence of the catheter or AVF leads to stenosis in the appropriate milieu (of endothelial cell dysfunction or cytokine activation, for example) may explain our finding that the majority (38%) of CVS occurs at the SCV-CV junction, despite the lack of direct cannulation at this site. It is interesting to note that 5% of patients that developed CVS did not have any history of hemodialysis catheter or pacemaker insertion. Furthermore, our finding that the presence of an AVF access is associated with ipsilateral central vein stenosis is consistent with the theory that increased shear stress caused by high flows created by the fistula itself produces endothelial damage and subsequent stenosis in at risk patients. Future studies will be needed to confirm these observations and to explore the hypotheses more fully.
Our data have demonstrated that of patients with multiple catheter insertions, 33% did not have CVS. This finding is equally interesting and also warrants further study. There may well be patient related protective factors regarding the propensity to stenose. Other possibilities include spontaneous recanalization9,14 and the underdiagnosis of significant stenosis on venogram.
The presence of CVS has major long-term consequences, the most important of which is lack of AVF maturation and its attendant morbidities.4 Although our 6 month follow up certainly demonstrated many adverse outcomes, the retrospective nature of this study precludes imparting CVS as a causative factor in the patient outcomes.
Limitations of this study include that the study design is retrospective and there is a possible selection bias. However, given that we included all patients in whom a venogram was ordered for any access related concerns, our conclusions can be generalized to some extent. In patients undergoing venography for a variety of reasons, the diagnosis of CVS is evident even in the absence of classical signs and symptoms. Although this study does not identify the true incidence of CVS, it increases awareness as to the high unexpected incidence of CVS in patients with access dysfunction. Only prospective longitudinal studies can truly assess CVS incidence and prevalence. The logistics and ethics of prospective venography in all patients on HD are potentially problematic. Thus, this study serves as a baseline from which to consider venography in patients with catheter insertion who are undergoing AVF creation.
This study serves as a reminder for clinicians and researchers alike that the problem of CVS has not been solved by advocating for a change in catheter location. Obviously, the pathophysiology of central vein stenosis, identification of those at highest risk, and proactive strategies need to be developed for this vulnerable population.
1. Dhingra RK, Young EW, Hulbert-Shearon TE, Leavey SF, Port FK: Type of vascular access and mortality in U. S. hemodialysis patients. Kidney Int
60: 1443–1451, 2001.
2. United States Renal Data System: USRDS Assessment of Clinical Care of Dialysis Patients. The National Institute of Health, National Institute of Diabetes and Digestive and Kidney Diseases, Bethesda, 2001.
3. MacRae JM, Stevens L, Chan Yan S, Karim M, Djurdjev O, Kiaii M: Vascular access distribution in prevalent hemodialysis patients in British Columbia, Canada. J Am Soc Nephrol
13: 408A, 2002.
4. Barrett N, Spencer S, McIvor J, Brown EA: Subclavian stenosis: A major complication of subclavian dialysis catheters. Nephrol Dial Transplant
5. Schillinger F, Schillinger D, Montagnac R, Milcent T: Post catheterisation vein stenosis in hemodialysis: Comparitive angiographic study of 50 subclavian and 50 internal jugular accesses. Nephrol Dial Transplant
6: 722–724, 1991.
6. NKF/DOQI clinical practice guidelines for vascular access: Update 2000. Am J Kidney Dis
37: S137–S181, 2001.
7. Aytekin C, Boyvat F, Yagmurdur MC, Moray G, Haberal M: Endovascular stent placement in the treatment of upper extremity central venous obstruction in hemodialysis patients. Eur J Radiol
49: 81–85, 2004.
8. Kalra M, Bloviczki P, Andrews JC, et al
: Open surgical and endovascular treatment of superior vena cava syndrome caused by non-malignant disease. J Vasc Surg
38: 215–223, 2003.
9. Hernandez D, Diaz F, Rufino M, et al
: Subclavian vascular access stenosis in dialysis patients: Natural history and risk factors. J Am Soc Nephrol
9: 1507–1510, 1998.
10. Schwab SJ, Quarles LD, Middleton JP, Cohan RH, Saeed M, Dennis VW: Hemdialysis-associated subclavian vein stenosis. Kidney Int
33: 1156–1159, 1988.
11. Davis D, Petersen J, Feldman R, Cho C, Stevick CA: Subclavian venous stenosis: A complication of subclavian dialysis. JAMA
252: 3404–3406, 1984.
12. Cimochowski GE, Worley E, Rutherford WE, Sartain J, Blondin J, Harter H: Superiority of the internal jugular over the subclavian access for temporary dialysis. Nephron
54: 154–161, 1990.
13. Huber TS, Seeger JM: Approach to patients with “complex” hemodialysis access problems. Semin Dial
16: 22–29, 2003.
14. Manca O, Murgia AM, Loi L, Pili GF, Murgia MG, Barracca A: Internal jugular vein thrombosis after positioning CVC in dialysis patients: The most common ultrasound patterns. J Vasc Access
3: 127–134, 2002.