Inter-observer agreement of color duplex ultrasound of central vein stenosis in hemodialysis patients

Abstract

Purpose

To assess the inter-observer agreement of color duplex ultrasound of central vein stenosis in hemodialysis patients.

Patients and methods

This prospective study was conducted on 35 hemodialysis patients with suspected central vein stenosis. All patients underwent color flow duplex examination of the subclavian, internal jugular and brachiocephalic veins in hemodialysis patients. Image analysis was performed by two reviewers for diameter reduction, peak venous velocity ratio, post-stenotic turbulent flow, waveform changes, and thrombus formation.

Results

There was no significant difference between both observers for diameter reduction (p = 0.105) and for the mean peak systolic velocity ratio (p = 0.515). The overall inter-observer agreement of color duplex ultrasound of central vein stenosis was excellent (k = 0.84, percent agreement = 89.7%, P = 0.001). There was excellent inter-observer agreement of both reviewers for diameter reduction (k = 0.928, percent agreement = 97.14%), peak venous velocity ratio (k = 0.7, percent agreement = 85.7%), waveform changes (k = 0.62, percent agreement = 77.14%), post-stenotic turbulent flow (k = 0.866, percent agreement = 88.6%), thrombus formation (k = 1, percent agreement = 100%).

Conclusion

We concluded that color duplex ultrasound is a reliable and reproducible method for diagnosis of central vein stenosis in hemodialysis patients.

Keywords

Renal dialysis access central vein stenosis hemodialysis fistula color duplex

Introduction

Central venous stenosis frequently occurs in patients with hemodialysis with a history of central venous catheters insertion. Central venous dialysis catheters insertion is the leading cause of central venous stenosis in hemodialysis patients. In addition, as patients live longer, they are subject to interventions as peripheral central line insertions and cardiac devices that increase the incidence of central venous lesions.^1–6 Endovascular intervention is the gold standard treatment of central venous stenosis as color duplex determines the need for angiography and angioplasty. Color duplex ultrasound leads to the reduction in access thrombosis, catheter placement, and access loss.^7–11 The diagnosis of central vein stenosis is based on clinical and imaging findings. Digital subtraction venography is the current gold standard technique for the diagnosis of central vein stenosis but is an invasive procedure with contrast media injection. CT venography is for assessment of fistula but it is associated with contrast medium injection and radiation exposure. MR venography takes long time, needs skill and experience and may be associated with risk of nephrogenic systemic fibrosis in patients with renal impairment.^12–16

Color flow duplex was used for early detection of dysfunctional hemodialysis fistulas, fistula maturation and for diagnosis of central vein stenosis as a cost-effective, non-invasive screening test. Few studies discuss the role of color Duplex in the diagnosis of stenosis of the central veins in patients with hemodialysis.^17–20 However, the previous reports about color flow Duplex in central vein stenosis are overlapping due to operator dependence.^21–26

Patients and methods

Patients

The institutional review board approval and the informed consent of participants were obtained. The prospective study was performed upon 37 patients with failing hemodialysis access. The inclusion criteria were hemodialysis patients with clinical signs suggestive of central vein stenosis. Two patients were excluded from the study as they could not complete the examination. The final patients in this study were 35 patients presented with upper limb swelling (n = 28), dilated veins on the chest wall (n = 24) and prolonged bleeding after dialysis (n = 21).

Technique

All patients underwent duplex ultrasound examination (Logic P5, GE machine, USA) which was done by one radiologist expert in ultrasound for four years (EM). Real-time, color flow map and pulsed waveform were used for all patients. Color duplex ultrasound examination of the subclavian, internal jugular and brachiocephalic veins on the side of the fistula was done. Patients were examined in the supine position with their arms slightly abducted. The internal jugular vein and the subclavian vein were examined through the middle supra-clavicular and distal infra-clavicular windows using 10-MHz phase array probe and small retro-clavicular portion of superior vena cava was evaluated by denoting changes at the proximal vein. The left brachiocephalic vein was quite difficult to be examined because of its longer course, but it was assessed through the supraclavicular space using 5–8 MHz phased array probe and 5–MHz convex probe. Color duplex ultrasound was performed in transverse axis to assess the compressibility of the veins and presence of thrombosis and then in the longitudinal axis. B-mode was used to assess the luminal narrowing, determine diameter reduction and detect thrombus.

Image analysis

Two radiologists (RA, KA), expert in ultrasound with 20, 15 years' experience and who were blinded to the clinical presentation, independently analyzed at a separate session real-time, color flow map and pulsed waveform for all patients. The image analysis was performed for diameter reduction, peak venous velocity ratio, post-stenotic turbulent flow, waveform changes and the presence of thrombus. The vein diameter was measured at and before the stenotic region. The ratio of vein diameter at and before the stenotic region was calculated. The vein velocity was measured before and after the stenotic segment and at the normal vein segment. The thrombosed vein was non-compressible without color flow. The stenosis was significant when diameter reduction ≥50%, velocity ratio ≥2.1, the presence of post-stenotic turbulent color flow and weak continuous flow signals on pulsed wave Doppler study. The stenosis was insignificant when diameter reduction <50%, peak velocity ratio <2, slight post-stenotic turbulent flow and wave changes represented by poor augmentation with wave asymmetry.

Statistical analysis

The statistical analysis of data was done with the Statistical Package for Social Science, version 20 (SPSS Inc., Chicago, Ill, USA). Mann–Whitney U test was used to calculate the median of diameter reduction and the peak systolic velocity ratio. Bland and Altman analysis was used to detect inter-observer agreement for non-parametric continuous variables calculating the mean and difference between two observers with scatter plot for the mean and the difference. Kappa analysis was used to detect the inter-observer agreement for non-continuous variables calculating k coefficient value with 95% confidence intervals (CI). A κ of 1.0 represents perfect agreement, a κ of 0.81 to 1.0 is excellent agreement, and 0.61 to 0.80 is a good agreement. We categorized continuous variables into categorical variables for Kappa analysis. We categorized diameter reduction by grouping more than 50% and peak systolic velocity ratio by grouping more than 2.1.

Results

The final diagnosis was significant stenosis (n = 23) of either subclavian veins (n = 9), internal jugular veins (n = 8), or brachiocephalic veins (n = 6), insignificant stenosis (n = 5) of either internal jugular vein (n = 3) or subclavian vein (n = 2), complete occlusion with thrombosis of internal jugular vein (n = 3) and normal color duplex ultrasound (n = 4). Significant stenosis of the subclavian vein was located above the level of the clavicle and not at an infraclavicular level that confirmed by post-stenotic turbulent flow, accurate demarcation of the vein caliber before and after this segment and velocity measurement. No significant stenosis was detected in the superior vena cava and this was elicited by accurate observation of dilated veins proximal to the obstruction, detection of retrograde flow and velocity measurements.

The median diameter reduction of the central vein reported by the first observer was 53% and by the second observer was 52% with an excellent inter-observer agreement (r = 0.891, P = 0.001). The median peak velocity ratio by the first observer was 2.1 and by the second observer was 2.3 with an excellent inter-observer agreement (r = 0.908, P = 0.001). A Bland and Altman analysis shows a non-statistical significant difference between both observers for diameter reduction (P = 0.105) and for the mean peak systolic velocity ratio (P = 0.515) (Figure 1).

Figure 1.

Scatter diagram for Bland and Altman analysis: (a) scatter diagram for diameter reduction. (b) Scatter diagram for peak venous velocity ratio.

The inter-observer agreement of diameter reduction (Figure 2) was excellent (k = 0.928, percent of agreement = 97.14%). The inter-observer agreement of peak venous velocity ratio (Figure 3) was good (k = 0.7, percent of agreement = 85.7%). The inter-observer agreement of wave changes (Figure 4) was good (k = 0.62, percent of agreement = 77.14%). The inter-observer agreement of venous thrombus (Figure 5) was perfect (k = 1, percent of agreement = 100%). The inter-observer agreement of post-stenotic turbulent mosaic color flow (Figure 6) was good (k = 0.766, percent of agreement = 88.6%). The overall inter-observer agreement of color duplex ultrasound for detection of the central vein stenosis was excellent (k = 0.84, percent agreement = 89.7%). Table 1 shows inter-observer agreement of color duplex ultrasound of central vein stenosis in hemodialysis patients.

Figure 2.

Diameter reduction: Longitudinal ultrasound image shows more than 50% diameter reduction of the left subclavian vein (arrows).

Figure 3.

Peak velocity changes: Peak venous velocity ratio (>2) across the stenotic segment at left brachiocephalic vein (diameter reduction of more than 50%). (a) Spectrum duplex waveform shows pre-stenotic peak venous velocity is 85 cm/s. (b) Spectrum duplex waveform shows post-stenotic peak venous velocity is 187 cm/s.

Figure 4.

Waveform changes: Spectrum duplex waveform shows continuous weak flow wave at the stenotic segment of the left brachiocephalic vein.

Figure 5.

Thrombosis: Echogenic thrombus with luminal attenuation of the internal jugular vein in a hemodialysis patient with a history of central venous catheter insertion.

Figure 6.

Post-stenotic turbulent flow: The mosaic color of the right brachiocephalic vein in a hemodialysis patient with subclavian vein stenosis.

Table 1.

Inter-observer agreement with 95% confidence interval (CI) and percent agreement of color duplex ultrasound of central vein stenosis.

	Observer 1	Observer 2	Percent agreement	K	95% CI
Diameter reduction (by grouping more than 50%)	25	25	94.2%	0.86	0.704–1.0
Peak venous velocity ratio (by grouping more than 2.1)	20	23	85.7%	0.70	0.46–0.94
Post-stenotic turbulence
Present	7	9	88.6%	0.766	0.55–0.98
Absent	28	26
Wave changes
No flow	3	3	77.14%	0.62	0.38–0.85
Asymmetry	8	8
Spontaneous phasic	4	4
Weak continuous	20	20
Thrombosis	3	3	100%	1.0	1.00–1.00
Overall	108	109	86.7%	0.79	0.71–0.89

Discussion

In the current study, the overall inter-observer agreement of color duplex ultrasound for the detection of central vein stenosis was excellent (k = 0.84). One study reported that sensitivity and specificity of color duplex ultrasound, compared with venography, in the assessment of proximal veins of hemodialysis patients, were 80.9%, and 79.3%, respectively, with inter-observer Kappa agreement coefficient of 0.38.²³ Another study added that absence of color signal, direct observation of the lumen stenosis and presence of collateral veins in the upper limb with color-flow duplex ultrasound indicates proximal vein stenosis.²⁵ Previous studies reported that diagnosis of central vein stenosis with color duplex ultrasound is done in the absence of normal respiratory variation in diameter of central veins and polyphasic atrial waveforms.^24,26

In this study, there was no significant difference between both observers for diameter reduction and for peak venous velocity ratio. One study reported that a peak vein velocity ratio of >2.5 across the stenosis is the best criterion to use for the presence of a pressure gradient of =3 mmHg.²⁵

In this study, diameter reduction is significant if there are more than 50% reduction in relation to the normal venous segment. Previous studies reported that direct observation of the lumen stenosis and presence of collateral veins in the upper limb with color duplex ultrasound indicate the presence of proximal vein stenosis. However, each of these symptoms can suggest proximal vein stenosis and their combinations can increase the precision of diagnosis.^23–25

In this study, there is a significant change in peak velocity ratio ≥2.1 in the stenosed venous segment in patients with significant central vein stenosis. Other studies reported that patients with stenosis more than 50% are calculated by diameter reduction and by an area reduction of 75%, and the peak venous velocity ratio is >2.0.²⁵

In this study, color aliasing at the stenotic segment with post-stenotic turbulent flow signals is observed in patients with significant central veins stenosis and the inter-observer agreement of both reviewers is good. One study reported that post-stenotic turbulence in absence of inferior vena cava or superior vena cava obstruction is used for detection of stenosis.²³

This study shows weak slow signals in patients with significant stenosis of the central veins and asymmetry of the venous flow signal in patients with insignificant stenosis. One study reported that changes in cardiac/respiratory phasicity revealed 52.4% sensitivity, 82.8% specificity, 68.8% positive predictive value and 70.6% negative predictive value in the diagnosis of stenosis.²³ Another study added that peak systolic velocity of ≥500 cm/s predicted a 50% or greater stenosis with a sensitivity of 89% and positive predictive value of 99%. The systolic velocity ratio of the fistula is an unreliable parameter in predicting the degree of stenosis in patients with hemodialysis fistula.²⁶

In this study, the inter-observer agreement for detection of complete occlusion with thrombosis is perfect. The previous study reported that the compression ultrasound is the highly accurate test for detection of deep vein thrombosis.²⁷ Another study added that thrombosis and/or complete obstruction of the subclavian vein occur in 35% (15%–50%) and of the internal jugular vein in 3% of patients with hemodialysis.²⁸

There are a few limitations of this study. First, this study used color duplex ultrasound for diagnosis of central vein stenosis. Further studies using advanced ultrasound tools as ultrasound elastography and intravascular ultrasound or combined color duplex with contrast MR angiography^29,30 and CT angiography^31–33 will improve the results. Second, this study included a small number of patients with no follow-up after treatment. Further studies on a large number of patients before and after therapy^34–37 will improve the results.

Conclusion

We concluded that color duplex ultrasound is a reliable and reproducible method for diagnosis of central vein stenosis in hemodialysis patients.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Ethical Approval

Obtained

Guarantor

Abdel Razek A

Contributorship

All authors shared equal in manuscript writing and editing.

Acknowledgement

None.

References

Expert Panel on Interventional Radiology, Shaw

Shah

, et al. ACR appropriateness criteria radiologic management of central venous access. J Am Coll Radiol 2017; 14: S506–S509.

Krishna

Eason

Allon

Central venous occlusion in the hemodialysis patient.

Am J Kidney Dis 2016; 68: 803–807.

Modabber

Kundu

Central venous disease in hemodialysis patients: an update.

Cardiovasc Intervent Radiol 2013; 36: 898–903.

ToomayS Rectenwald

Vazquez

MA.

Central venous stenosis in hemodialysis patients.

Sem Dial 2016; 29: 201–3.

Trerotola

Kothari

Sammarco

, et al. Central venous stenosis is more often symptomatic in hemodialysis patients with grafts compared with fistulas. J Vasc Interv Radiol 2015; 26: 240–246.

Agarwal

AK.

Central vein stenosis.

Am J Kidney Dis 2013; 61:1001–1015.

Kundu

Review of central venous disease in hemodialysis patients.

J Vasc Interv Radiol 2010; 21: 963–968.

Kundu

Central venous disease in hemodialysis patients: prevalence, etiology and treatment.

J Vasc Access 2010; 11: 1–7.

Horikawa

Quencer

KB.

Central venous interventions.

Tech Vasc Interv Radiol 2017; 20: 48–57.

10.

Agarwal

Haddad

Khabiri

How should symptomatic central vein stenosis be managed in hemodialysis patients?

Sem Dial 2014; 27: 278–281.

11.

Cho

Lee

Kim

Value of Doppler evaluation of physically abnormal fistula: hemodynamic guidelines and access outcomes.

Kor J Inter Med 2019; 34: 137–145.

12.

Okur

Kantarci

Karaca

, et al. Non-contrast-enhanced imaging of haemodialysis fistulas using quiescent-interval single-shot (QISS) MRA: a feasibility study. Clin Radiol 2016; 71: 244–249.

13.

Razek

Saad

Soliman

, et al. Assessment of vascular disorders of the upper extremity with contrast-enhanced magnetic resonance angiography: pictorial review. Jpn J Radiol 2010; 28: 87–94.

14.

Meyer

Geiger

Benck

, et al. Imaging of patients with complex hemodialysis arterio-venous fistulas using time-resolved dynamic ct angiography: comparison with duplex ultrasound. Sci Rep 2017; 7: 12563.

15.

Kim

Mirza

Bryant

, et al. Central veins of the chest: evaluation with time-resolved MR angiography. Radiology 2008; 247: 558–566.

16.

Bakhshoude

Ravari

Kazemzadeh

G.H

, et al. Diagnostic value of computerized tomography venography in detecting stenosis and occlusion of subclavian vein and superior vena in chronic renal failure patients. Electron Phys 2016; 8: 2781–2786.

17.

Itoga

Ullery

Tran

, et al. Use of a proactive duplex ultrasound protocol for hemodialysis access. J Vasc Surg 2016; 64: 1042–1049.e1.

18.

Bozoghlanian

Lall

Houshyar

, et al. Duplex Doppler imaging of dialysis fistulae and grafts. Curr Probl Diagn Radiol 2016; 45: 420–28.

19.

Lomonte

Meola

Petrucci

, et al. The key role of color Doppler ultrasound in the work-up of hemodialysis vascular access. Semin Dial 2015; 28: 211–215.

20.

Teodorescu

Gustavson

Schanzer

Duplex ultrasound evaluation of hemodialysis access: a detailed protocol.

Int J Nephrol 2012; 2012: 508956.

21.

Mousa

Morkous

Broce

, et al. Validation of subclavian duplex velocity criteria to grade severity of subclavian artery stenosis. J Vasc Surg 2017; 65: 1779–1785.

22.

Jones

Cassada

Heidel

, et al. Maximal venous outflow velocity: an index for iliac vein obstruction. Ann Vasc Surg 2012; 26: 1106–1113.

23.

Rad

Kazemzadeh

Ziaee

, et al. Diagnostic value of color Doppler ultrasonography in detecting stenosis and occlusion of central veins in patients with chronic kidney disease. Saudi J Kidney Dis Transpl 2015; 26: 279–284.

24.

Osman

El-Magzoub

Elamin

Prevalence and risk factors of central venous stenosis among prevalent hemodialysis patients, a single center experience. Arab J Nephrol Transplant 2014; 7: 45–47.

25.

Labropoulos

Borge

Pierce

, et al. Criteria for defining significant central vein stenosis with duplex ultrasound. J Vasc Surg 2007; 46: 101–107.

26.

Kellie

Morrison

Harada

, et al. Developing duplex ultrasound criteria for diagnosis of arteriovenous fistula stenosis. Ann Vasc Surg 2017; 38: 99–104.

27.

Garcia

Labropoulos

Duplex ultrasound for the diagnosis of acute and chronic venous diseases.

Surg Clin North Am 2018; 98: 201–218.

28.

Babadjanov

Bernstein

Kirksey

Surgical reconstruction of central venous obstruction in salvaging upper extremity dialysis accesses.

J Vasc Access 2017; 18: e39–e41.

29.

Abdel Razek

Ashmalla

Samir

Clinical value of classification of venous malformations with contrast enhanced MR angiography.

Phlebology 2017; 32: 628–633.

30.

Razek

Gaballa

Megahed

, et al. Time resolved imaging of contrast kinetics (TRICKS) MR angiography of arteriovenous malformations of head and neck. Eur J Radiol 2013; 82: 1885–1891.

31.

Abdel Razek

AAK

Elrakhawy

Yossof

, et al. Inter-observer agreement of the Coronary Artery Disease Reporting and Data System (CAD-RADS(TM)) in patients with stable chest pain. Pol J Radiol 2018; 83: e151–e159.

32.

Abdel Razek

Alvarez

Bagg

, et al. Imaging spectrum of CNS vasculitis. Radiographics 2014; 34: 873–894.

33.

Abdel Razek

Denewer

Hegazy

, et al. Role of computed tomography angiography in the diagnosis of vascular stenosis in head and neck microvascular free flap reconstruction. Int J Oral Maxillofac Surg 2014; 43: 811–815.

34.

Abdel Razek

Elkammary

Elmorsy

, et al. Characterization of mediastinal lymphadenopathy with diffusion-weighted imaging. Magn Reson Imaging 2011; 29: 167–172.

35.

Razek

Fouda

Elmetwaley

, et al. Sonography of the knee joint. J Ultrasound 2009; 12: 53–60.

36.

Abdel Razek

Mossad

Ghonim

Role of diffusion-weighted MR imaging in assessing malignant versus benign skull-base lesions.

Radiol Med 2011; 116: 125–132.

37.

Abdel Razek AA, Kamal E. Nasopharyngeal carcinoma: correlation of apparent diffusion coefficient value with prognostic parameters. Radiol Med 2013; 118: 534–9.