The venous angle as an anatomical landmark to evaluate external ventricular catheter placement

Abstract

Background

External ventricular drain (EVD) placement is a lifesaving procedure that accesses the ventricular system. It is typically performed freehand using external craniometric landmarks. Herein, we study the utility of digital subtraction angiograms (DSA) in determining the EVD location by comparing it to the deep venous system.

Methods

A retrospective review was conducted at a single center from 2021 to 2024 to identify patients with EVD placement for communicating hydrocephalus secondary to a subarachnoid hemorrhage. Two independent reviewers assessed placement using Fargen's grade and a novel grading system. We analyzed inter-reviewer reliability.

Results

We included 48 patients (mean age 60.3 ± 12.9) in this study. Most patients presented with a modified Fisher score of 4 (76.6%) and a Hunt-Hess grade of 4 (31.9%). Based on the computed tomography (CT) scan, 33 patients had good placement, 13 had contralateral placement, and two had suboptimal EVD placement. Inter-reviewer reliability between the Fargen CT and novel DSA grading systems demonstrated 85.4% overall agreement and Cohen's κ = 0.66 (95% CI, 0.41–0.90). The novel DSA grading system demonstrated moderate discrimination among the three CT-defined categories (one-vs-all AUCs: suboptimal 0.668, contralateral 0.847, good 0.841; multiclass AUC 0.700).

Conclusion

This feasibility study suggests that EVD placement can efficiently be determined in the endovascular suite by visualizing the catheter's position in relation to the venous angle on DSA. While this may streamline emergency workflows and reduce reliance on immediate CT scans, subsequent verification with CT or magnetic resonance imaging remains essential. Prospective studies are needed to validate accuracy before wider adoption.

Introduction

External ventricular drain (EVD) insertion is a lifesaving intervention to access the ventricular system and relieve intracranial pressure. It is often performed at the bedside, using consistent craniometric landmarks (i.e. Kocher's point). Training for EVD placement occurs early during neurosurgical residency and thus presenting an opportunity for training simulations and neuronavigational aids.^1–3 Despite this, the use of newer techniques employing ultrasound, frameless neuronavigation, or electromagnetic tracking systems, portable computed tomography (CT) scanners in the intensive care unit, and endoscopic guidance remains limited, with a preference for the familiar freehand technique.^4–6 Possible reasons include the emergent nature of the intervention, cost considerations, and scarcity of resources in the intensive care and emergency units.^7,8

Accurate placement of a frontal approach EVD is characterized by a catheter that traverses the frontal lobe and the frontal horn of the lateral ventricle and terminates with the tip of the catheter at the Foramen of Monro or the third ventricle. Accurate catheter placement is essential for proper cerebrospinal fluid (CSF) drainage. EVD catheter malposition can occur in 40%–80% of cases.^7–9 Cross-sectional imaging, such as a CT scan, confirms catheter position after placement, preventing tract-associated intracerebral hemorrhages. In the context of emergent conditions such as intracerebral or subarachnoid hemorrhage (SAH), the need for cross-sectional imaging may delay the clinical workflow and subsequent care, including a diagnostic cerebral angiography (DSA) to identify the vascular etiology of the hemorrhage. Along with efficient workflow, it provides real-time confirmation of EVD tip location in the angiography suite, potentially allowing on-the-spot repositioning and uninterrupted transition to definitive vascular treatment. If these results are generalized and accepted, it potentially saves patients and payors from the extra cost of additional CTs, in addition to reducing the overall radiation exposure that these patients receive.

The anatomical relationship between the venous angle of the thalamostriate vein (TSV) and internal cerebral vein (ICV), with the Foramen of Monro, can be leveraged to understand catheter placement on cerebral angiography. ¹⁰ The venous angle is formed by the characteristic U-shaped reverse turn where the TSV joins the ICV, at the posterior border of the foramen of Monro. The venous angle can be readily observed in the lateral projection venous phase of cerebral angiography, and the ventricular catheter can be easily visualized in unsubtracted images during cerebral angiography, and can then be localized in relation to these vessels and the foramen. Additionally, no established protocol currently exists for real-time evaluation to confirm catheter tip location in real time.

Herein, we propose using the venous phase of a DSA to visualize the position of the EVD catheter tip in relation to the venous angle as a landmark. The objective of this study is to report the utility of angiography in determining EVD catheter placement in patients with emergent conditions such as SAH to improve workflow efficiency. Subsequent confirmation using cross-sectional imaging remains essential to ensure accurate placement.

Methods

Study design and patient selection

This is a single-center retrospective cohort study at a tertiary care hospital. We obtained the necessary ethical approval to conduct this study, including institutional review board permission from the Johns Hopkins IRB. Patient consent was not required because the study involved a retrospective database search. We included consecutive adult (>18 years) patients admitted with an SAH in 2021–2024 who had EVD placement for hydrocephalus. These patients underwent confirmation cross-sectional imaging, followed by DSA. We collected demographic and clinical data. We reviewed all images of the patients during their inpatient course, including DSA and CT scans performed after EVD placement. We excluded patients with poor-quality images or inadequate imaging, for example, poor venous visualization on DSA. CT angiography images were reviewed if available.

Imaging techniques and review

1. Digital subtracted angiography: DSA was performed using a biplane angiographic system to identify the source of bleeding. We used unsubtracted images featuring the EVD to visualize cerebral vessels and the EVD tip. The venous phase of the DSA ascertained the position of the EVD catheter tip in relation to the venous angle, defined as the intersection of the ICV and the TSV (Figure 1). We used unsubtracted lateral and anteroposterior projection scans exclusively because subtraction eliminates background anatomy and only includes contrast-enhanced cerebral vessels, sometimes subtracting the EVD catheter. The anteroposterior projection was important for assessing the catheter's laterality.

2. CT scan: We studied the position of the EVD catheter tip in relation to the Foramen of Monro and other structures after placement to confirm accurate catheter position. In our institution, axial CT images are taken following EVD insertion to confirm accurate placement.

Figure 1.

DSA, lateral projection, venous phase showing the venous angle. (a) TSV, ICV, a VOR, and the VOL. (b) Magnification over the venous angle (arrow) formed at the juncture between TSV and ICV. Note the characteristic acute reverse posterior turn where TSV joins ICV.

DSA: digital subtraction angiography; TSV: thalamostriate vein; ICV: internal cerebral vein; VOR: vein of Rosenthal; VOL: vein of Labbe.

Grading of the optimal position of the catheter tip

One neurosurgery resident (PGY5) and one postdoctoral research fellow in the Department of Neurosurgery interpreted the images. One reviewer (F.R.) graded the EVD catheter tip position on the CT scans based on the Fargen grading system ¹¹ (Table 1 and Figure 2). The second reviewer (A.K.A.) graded the EVD placement based on a novel grading scale developed by the authors (Table 2).

Table 1.

Fargen system for grading of external ventricular drain (EVD) catheter tip placement in relation to the Foramen of Monro.

Grade I	Catheter tip within 5 mm of the ipsilateral Foramen of Monro
Grade II	Catheter tip in the ipsilateral lateral ventricle but ≥ 6 mm from Foramen of Monro
Grade III	Catheter tip in the third ventricle
Grade IV	Catheter tip in the contralateral lateral ventricle
Grade V	Catheter tip not within the ventricle

Figure 2.

An illustration of an external ventricular drain (EVD) catheter tip in the ventricular system in relation to the deep venous structures. (a) Grades I, II, and III placement in relation to the venous angle and (b) Grades IV and V placement, also in relation to the venous angle according to the Fargen system.

Table 2.

A novel grading scale of external ventricular drain (EVD) placement was developed by the authors.

Grade I	Catheter tip above the level of the venous angle on lateral projection
Grade II	Catheter tip in front and/or below the venous angle on lateral projection
Grade III	Catheter tip between venous angle and vein of Galen, but tip <2 cm from venous angle on lateral projection
Grade IV	Catheter tip crossing the midline in the contralateral ventricle on the anterior projection
Grade V	Catheter tip EVD below venous angle >2 cm or behind the vein of Galen on lateral projection (parenchyma)

On a lateral projection, the DSA image showing both the venous angle and catheter, a line bisecting the venous angle was drawn perpendicular to the catheter to account for any anatomical variations. ¹² If a single image with both structures was unavailable, a subtracted image with the venous angle was overlaid onto a corresponding unsubtracted image to simulate the same view. We took measurements from the point of catheter intersection to the catheter tip and graded placement based on the system shown in Table 2. Grades 1, 2, and 3 were considered good placement; Grade 4 was contralateral placement; and Grade 5 was considered suboptimal placement.

Statistical analysis

Categorical data were presented as counts and percentages. Continuous data were reported as means and standard deviations (SDs). Inter-reviewer reliability between the Fargen-based CT grading, decision-based CT grading, and the novel DSA grading system was assessed. We generated a confusion matrix to compare classifications, and we calculated key performance metrics (sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and balanced accuracy) for each grading category (suboptimal, contralateral, and good placement). All statistical analyses were conducted in R (version 3.4.2). Inter-rater agreement was evaluated with unweighted Cohen's κ (95% CI), and discriminative performance assessed via one-versus-all ROC curves per-class AUCs with 95% CIs estimated by 2000 bootstrap replicates and multiclass AUC.

Results

Forty-eight patients met our inclusion criteria. All patients had a CT scan to confirm the accurate positioning of the EVD catheter tip before going to the endovascular suite to find the source of the bleeding. Of the 48, six had CTA scans, including patients who had CTA at any point during the admission after EVD placement. CTA imaging was used to view the relationship between the venous angle and the Foramen of Monro for those patients.

Study population characteristics

The study population had an average age of 60.3 ± 12.9, with 35.4% (n = 17) males and 64.6% (n = 31) females. Racial demographics included 47.9% (n = 23) African American, 27.1% (n = 13) White, 14.6% (n = 7) Asian, 2.1% (n = 1) Hispanic, and 8.3% (n = 4) unidentified. Never smokers constituted 58.3% (n = 28), 25.0% (n = 12) were current smokers, and 16.7% (n = 8) were former smokers. Comorbidities included 68.8% (n = 33) of patients with hypertension, 47.9% (n = 23) with obesity, 18.8% (n = 9) with diabetes, 10.4% (n = 5) with a history of stroke, 4.2% (n = 2) with atrial fibrillation, and 6.2% (n = 3) with cancer. Twenty-six (54.2%) patients were treated by endovascular procedures, 20 (41.7%) by open microsurgery, and 2 (4.2%) patients received no aneurysm treatment. 87.5% (n = 42) of EVD placements were on the left side. The Hunt Hess grade distribution was: 5 in 17.0%, 4 in 31.9%, 3 in 25.5%, 2 in 14.9%, and 1 in 8.5%. The modified Fisher score indicated was 4 in 76.6%, 3 in 17.0%, 2 in 4.3%, and 1 in 2.1%. 87.5% (n = 42) of EVD placements were on the left side. The mortality rate in our cohort was 25.0%. All patient characteristics are summarized in Table 3.

Table 3.

Characteristics of patients included in the study.

Variables	Total (N = 48)
Age mean (SD)	60.3 (12.9)
Sex
Male	17 (35.4%)
Female	31 (64.6%)
Race
White	13 (27.1%)
African American	23 (47.9%)
Asian	7 (14.6%)
Hispanic	1 (2.1%)
unknown	4 (8.3%)
Smoking
Never	28 (58.3%)
Current	12 (25.0%)
Former	8 (16.7%)
Hypertension	33 (68.8%)
Diabetes	9 (18.8%)
Atrial fibrillation	2 (4.2%)
Stroke	5 (10.4%)
Obesity	23 (47.9%)
Cancer	3 (6.2%)
Treatment
None	2 (4.2%)
Open microsurgery	20 (41.7%)
Endovascular	26 (54.2%)
EVD placement side
Left	42 (87.5%)
Right	6 (12.5%)
Hunt Hess grade
Missing	1
0	1 (2.1%)
1	4 (8.5%)
2	7 (14.9%)
3	12 (25.5%)
4	15 (31.9%)
5	8 (17.0%)
Modified Fisher score
Missing	1
1	1 (2.1%)
2	2 (4.3%)
3	8 (17.0%)
4	36 (76.6%)
Death	12 (25.0%)

EVD: external ventricular drain.

Inter-reviewer reliability between Fargens grading (with CT scan) and novel DSA grading

The confusion matrix (Supplemental Table S1) demonstrated that one suboptimal placement was correctly identified, while one was misclassified as good. Nine of thirteen contralateral placements were classified correctly; four were misclassified as good. Of 33 good placements, 31 were correctly identified, with one misclassified as suboptimal and one as contralateral.

Sensitivity was 50.0% for suboptimal (n = 2), 90.0% for contralateral, and 86.1% for good placements. Specificity was 97.8%, 89.5%, and 83.3%, respectively. Positive predictive values were 50.0%, 69.2%, and 93.9%, and negative predictive values were 97.8%, 97.1%, and 66.7%. Balanced accuracy was 74.0% for suboptimal, 89.7% for contralateral, and 84.7% for good placements. Overall, DSA agreed with CT in 41 of 48 cases (85.4% agreement), correctly classifying one of two suboptimal, nine of 10 contralateral, and 31 of 36 good placements (Supplemental Table S1).

Inter-rater agreement between DSA and CT grading was good (Cohen's κ = 0.66; 95% CI, 0.41–0.90). Discriminative performance by one-versus-all ROC analysis yielded AUCs of 0.668 (95% CI, 0.304–1.000) for suboptimal, 0.847 (95% CI, 0.718–0.961) for contralateral, and 0.841 (95% CI, 0.704–0.949) for good placements, with a multiclass AUC of 0.700.

Discussion

In this study, we explored the performance of a novel classification scheme that leverages our understanding of the relationship between the venous angle on cerebral angiography with the Foramen of Monro, to determine the accuracy of EVD catheter placement. We assessed inter-reviewer reliability among three independent reviewers and found that DSA can accurately predict EVD placement, with the venous angle serving as a consistent and reliable anatomic landmark. We suspect this classification scheme can serve as a helpful decision aid during cerebral angiography and can safely expedite the clinical workflow, where cross-sectional imaging cannot be obtained.

The relationship between intracranial vasculature and surrounding cerebral anatomy is valuable in orienting and guiding surgeons. In 1929, Walter Dandy described the superior petrosal vein, “Dandy's vein,” as having a consistent posterior and superior relationship to the trigeminal nerve. ¹³ Another common anatomical relationship is the optic strut and the carotid proximal dural ring, which has been described as a landmark to differentiate between intradural and extradural paraclinoid aneurysms. ¹⁴ In this study, we highlight the relationship between the Foramen of Monro and deep cerebral veins, a crucial anatomic iteration explained by Dr Rhoton (Figure 3). The Foramen of Monro is situated on the anterior, superior, and medial aspects of the third ventricle. The crescent-shaped structure becomes more rounded as the ventricles enlarge. ¹⁵ Essential anatomic structures form the walls of the foramen, including the fornix, located medially; the choroid plexus, including distal branches of the medial posterior choroidal artery; the thalamostriate, superior choroidal, and septal veins. ¹⁰ The TSV courses anteriorly between the thalamus and the caudate nucleus at the posterior margin of the Foramen of Monro; TSV joins the medially running septal vein, converging into the ICV at an acute angle, known as the venous angle. For grading in our study, we used a line bisecting the venous angle to calculate the length of the catheter and considered all anatomic variations.

Figure 3.

CTA (left) and CT (middle) images of a patient with an EVD. The co-registered CTA and CT image (right) demonstrates the coincidence of the venous angle and the Foramen of Monro using the Phillips PACS system.

CT: computer tomography; CTA: CT angiography; EVD: external ventricular drain; PACS: Picture Archiving and Communication System

Herein, we present that the DSA scan can help reliably describe the relationship between the venous angle and the placement of ventricular catheters for CSF drainage (Figure 4). We demonstrated that positioning the catheter just inferior and anterior to the venous angle on a DSA scan correlates well with a good EVD placement on a DSA scan (Figure 5). Based on this knowledge, we suggest grading a modification of Fargen's classification system to base it on the relationship between the EVD tip and the venous angle (Table 2). The accuracy of the type of placement using the DSA scan compared to Fagan's grade from the CT scan reinforces the reliability of this technique in assessing placement. This improved workflow visualization is particularly useful in patients with emergent conditions that require both an EVD placement and angiography. One could propose bringing patients directly to the angiography suite, similar to what occurs during mechanical thrombectomy. EVD catheter placement in the angiography suite allows determination of the etiology of the bleed source and proper EVD placement confirmation. If there is a need for additional information, that is, rebleed assessment, subsequent cross-sectional imaging will suffice the need and will confirm accurate EVD catheter positioning.

Figure 4.

Unsubtracted DSA, venous phase showing lateral projection, with the EVD catheter (black box) with its tip inferior to the venous angle (Grade III) located on the floor of the III ventricle, considered a “long EVD,” considering the extra length (17 mm) below the venous angle. The arrow shows the venous angle, and the black box identifies the ventricular catheter. The DSA scan can help reliably describe the relationship between the venous angle and the placement of ventricular catheters for CSF drainage.

DSA: digital subtraction angiography; EVD: external ventricular drain; CSF: cerebrospinal fluid.

Figure 5.

CTA showing an EVD catheter tip (red star) anterior and inferior to the “venous angle” (black arrow), Grade I placement.

CTA: computer tomography angiography; EVD: external ventricular drain.

This finding potentiates an efficient clinical workflow by allowing direct patient transport to the angiography suite after EVD placement, which becomes essential during the critical minutes between clinical presentation and definitive management. This description highlights additional information gathered from the frequently forgotten venous system during routine cerebral angiography. However, this method does not replace regular cross-sectional imaging such as MRI or CT, which can better visualize anatomy, such as enlargement of a parenchymal bleed, ventricular size variation, and so on.

Limitations

This study has several limitations. DSA delineates only cerebral vasculature, and ‘subtracts’ the EVD catheter from the images. One way to prevent this is to have a filming protocol, including subtracted and non-subtracted views. On anteroposterior angiographic views, it is difficult to identify the venous angle, considering the reverse “U” turn between TSV and ICV occurs in the sagittal plane during the venous phase. The EVD catheter may traverse into the contralateral ventricle during insertion. In such scenarios, lateral angiographic views may seem adequate; however, on anteroposterior projections, it becomes obvious that the catheter crosses the interhemispheric fissure. Hence, we also recommend anteroposterior unsubtracted projections for confirmation. While CT is widely accepted as the clinical standard for confirming EVD placement, it is not a perfect anatomical gold standard. Our use of the Fargen grading scale provides a structured and reproducible comparator; however, it is not a gold standard for placement accuracy. The retrospective design inherently introduces selection bias, as only patients with both CT and DSA were included, potentially excluding some failed placements. The number of suboptimal placements was low (n = 2), limiting the statistical power for evaluating this category. While we used bootstrap resampling to derive 95% CIs around our AUC estimates, these results reflect only internal performance on a single 48-case cohort. Lastly, this method is futile in patients with aberrant cerebral venous anatomy. Our novel DSA-based grading system, while showing good inter-reviewer reliability, has not been externally validated and is hence not generalizable. External validation will be required before adopting this grading system in routine practice. Both lateral and anteroposterior views fail to visualize brain parenchyma. Hence, confirmation with CT or MRI is necessary, and DSA does not replace its utility in patients with an EVD catheter.

Conclusion

This feasibility study suggests that the venous phase of a DSA scan may be a useful adjunct for assessing EVD catheter placement directly in the angiography suite. In emergent cases where immediate cerebral angiography is required, this approach could streamline workflow and reduce delays. Our preliminary data show that catheter tips positioned anterior and inferior to the venous angle tend to correspond with accurate placements on CT, but larger prospective studies are required to validate these findings.

Abbreviations

EVD

external ventricular drain

ICP

intracranial pressure

CSF

cerebrospinal fluid

SAH

subarachnoid hemorrhage

CT

computer tomography

MRI

magnetic resonance imaging

CTA

CT angiography

DSA

digital subtraction angiography

ICV

internal cerebral vein

TSV

thalamostriate vein

VOR

vein of Rosenthal

VOL

vein of Labbe