Abstract
An optimal treatment strategy for subcortical hematomas caused by dural arteriovenous fistulae (dAVF) is important because of the high rebleeding rate. However, it is very difficult to diagnose that on admission. Therefore, an early sensitive predictive marker for subcortical hemorrhage caused by dAVF is necessary, especially during the first contact on admission. S-shaped dilated vessels around the hematoma (bold-S sign) on computed tomography angiography (CTA) performed during admission could be one such marker. Herein, we evaluated the characteristics of these vessels. Among 273 patients with intracerebral hemorrhage between April 2012 and March 2020, 67 patients with subcortical hematomas who underwent CTA on admission without arteriovenous malformations were included. The patients in the dAVF group (n = 7) showed fewer disturbances in consciousness, milder neurological deficits, and more frequent seizures than patients without dAVF (without dAVF group, n = 60). All patients in the dAVF group had dilated S-shaped vessels (2.59 ± 0.27 mm) around the hematomas, and only 20% of the patients in the without dAVF group had these vessels (1.69 ± 0.22 mm). The ratio of the ipsilateral S-shaped/contralateral largest vessels was 1.80 ± 0.29 in the dAVF group and 1.07 ± 0.16 in the group without dAVF. We called the dilated S-shaped vessels the “bold-S sign,” with a cutoff ratio of 1.5. Bold-S sign findings are novel and help in diagnosing subcortical hematomas caused by dAVF on admission.
Keywords: dural arteriovenous fistula, subcortical hemorrhage, computed tomography angiography, bold-S sign, on admission
Introduction
Patients with subcortical hemorrhage are not rare, and physicians consider a variety of causes, including dural arteriovenous fistulae (dAVF).1,2) As the surgical removal of hematomas caused by dAVF is more difficult than that of simple hematomas and dAVF shows frequent rebleeding within the first two weeks after the initial hemorrhage,2,3) physicians should rule out the presence of dAVF even if conservative therapy is chosen. Thus, it is crucial to identify hemorrhagic dAVF as soon as possible, especially on admission, and to provide early surgical/endovascular treatments.
Digital subtraction angiography (DSA) is necessary for diagnosing dAVF. However, it is not recommended for all patients with subcortical hemorrhage in the early phase based on the patients' condition and unnecessary medical care for simple hematoma cases. Although contrast-enhanced computed tomography (CT) or CT angiography (CTA) can contribute to dAVF diagnosis in subcortical hemorrhage,4-6) previous reports have shown 15.4% diagnostic sensitivity of CTA for dAVF.7) The low detection rate is due to the location of the shunt points of dAVF, which are close to the skull. Therefore, it is important to identify a more sensitive detection method based on standard radiological examinations such as contrast-enhanced CT or CTA.
This study aimed to explore a sensitive marker of dAVF using CTA in patients with subcortical hemorrhage on admission. To that end, we investigated the significance of the “bold-S sign” in the brain parenchyma to detect subcortical hemorrhage caused by dAVF using CTA images with high sensitivity and specificity.
Materials and Methods
Study population
We retrospectively reviewed 273 patients with intracerebral hemorrhage (ICH) admitted to our hospital's emergency department between April 2012 and March 2020. All patients with ICH underwent CTA at our neurosurgical emergency department to detect any vascular lesions, including cerebral aneurysms and arteriovenous anomalies, except for patients with high-risk conditions such as kidney dysfunction and restless movement. Of these, 78 patients had subcortical hemorrhage. Subsequently, we excluded patients who did not undergo CTA (n = 8) and those with arteriovenous malformations (n = 3). Finally, there were 7 patients with dural AVF (dAVF group) and 60 patients without dAVF (without dAVF group) (Supplementary Fig. 1).
Immediately after diagnosing ICH, all patients had a controlled blood pressure of <140 mmHg and received adequate treatments according to the Japanese Guidelines for the Management of Stroke. Emergent craniotomy and hematoma evacuation were performed for patients with Glasgow Coma Scale scores of ≤8. For patients with dAVF, DSA and endovascular treatment were performed within two weeks.
Age, sex, and presenting symptoms on admission were recorded for all patients. In addition, the Cognard classification, fistula location, and treatment procedure were recorded for the dAVF group.
To compare the CTA findings of dAVF patients without hemorrhage (nonhemorrhagic dAVF group), we analyzed the data of all 37 dAVF patients without hematomas introduced to the outpatient office who were suggested to have dAVF by magnetic resonance imaging (MRI) during the same period between April 2012 and March 2020 (Supplementary Fig. 1).
CTA analysis
CT acquisitions were performed according to standard departmental protocols on eight-section General Electric helical CT scanners (Bright Speed Edge, GE Healthcare, Wisconsin).8) The patients underwent CTA, followed by plain CT.
Unenhanced, contiguous, axial, and 5-mm-thick plain CT images were obtained from the vertex through the skull base at 120 kVp and 320 mAs. For CTA, 70 mL of ioversol (Optiray; Fuji Pharma Co., Ltd., Tokyo, Japan; 320 mg I/mL) was intravenously injected at a rate of 3-3.5 mL/s via a power injector through an intravenous line using the following parameters: 120 kVp; 240 mAs; section thickness, 1.25 mm; section-acquisition interval, 1.25 mm; and pitch, 0.875:1.8) The timing of CTA acquisition was determined using the bolus tracking technique.
In the CTA images of the patients in the dAVF group, we noticed abnormal findings of tortuous and expanded vessels around the hematomas, which were not a branch of the main artery such as the anterior cerebral artery, middle cerebral artery, and posterior cerebral artery.
The abnormal arteries in the dAVF group had an S-like shape and were clearly dilated and found around the hematomas, but they were not always connected to the lesion (Fig. 1). We named the finding a “bold-S sign.” To confirm whether the sign represented specific findings of dAVF, we measured the diameters of the “bold-S sign” vessels and the largest contralateral vessels and calculated the “diameter ratio” (the ratio between the ipsilateral and contralateral vessels) for patients in both the dAVF and without dAVF groups (Figs. 1 and 2). Then, we assessed the diameter ratio only in patients in the without dAVF group who had S-shaped vessels in the ipsilateral hemisphere (12 out of 60 patients).
Fig. 1.
Representative images of the bold-S sign in the dAVF group.
Plain CT (a, d), CT angiography (b, e), and diagnostic angiography (c, f) images of patients no. 2 and 6 specified in Table 2 are shown. Picture (g) is a delayed phase image in angiography. The white arrow and arrowhead indicate the bold-S sign on the ipsilateral side and the largest normal vessel on the contralateral side, respectively. We measured the diameter for both vessels and calculated the diameter ratio (ipsilateral side/contralateral normal side). A ratio <1.5 was defined as a “bold-S sign” (detailed description in the Results section). Patient no. 6 underwent craniotomy for tumor removal 9 months previously. Black arrows (c, g) indicate corresponding abnormal vessels in each 3D-CTA (b, e; white arrows). Abbreviations: CT, computed tomography; dAVF, dural arteriovenous fistula
Fig. 2.
Representative images of the S-shaped vessel (not the bold-S sign) in the without dAVF group.
Plain CT (a, c) and CT angiography (b, d) images are shown. The white arrow and arrowhead indicate the S-shaped vessel on the ipsilateral side and the largest normal vessel on the contralateral side, respectively. The diameters of the S-shaped vessel on the ipsilateral side and the largest normal vessel on the contralateral side of Patient 1 were 1.69 (arrow) and 1.64 (arrowhead), respectively, and the diameter ratio was 1.03. The diameters of the S-shaped vessel on the ipsilateral side and the largest normal vessel on the contralateral side of Patient 2 were 1.61 (arrow) and 2.06 (arrowhead), respectively, and the diameter ratio was 0.78. Abbreviations: CT, computed tomography; dAVF, dural arteriovenous fistula
Statistical analyses
Data were expressed as mean ± standard deviation. Student's t-test and the Mann-Whitney U test were used for parametric and nonparametric evaluations, respectively, of the dAVF and without dAVF groups. Fisher's exact test was used to compare categorical variables between the two groups. A p-value <0.05 was considered statistically significant. All analyses were performed using GraphPad Prism version 6.0 for Windows (GraphPad Software, San Diego, CA, USA). As the main aim of this study was to find a sensitive marker for diagnosing subcortical hemorrhages due to dAVF on emergent admission, we did not include the dAVF patients without subcortical hematomas in the statistical analysis.
Results
There were 7 patients in the hemorrhagic dAVF group and 60 patients in the without dAVF group. Age was similar between groups (66.1 ± 11.1y.o. in the hemorrhagic dAVF group vs. 71.7 ± 13.2y.o. in the without dAVF group). The proportions of men in the hemorrhagic dAVF group and without dAVF group were 57.1% and 46.7%, respectively. The patients in the hemorrhagic dAVF group and without dAVF group showed disturbances of consciousness (n = 4 vs. n = 55; p = 0.03), headache (n = 1 vs. n = 15; p = 1), seizure (n = 4 vs. n = 5; p < 0.01), and focal neurological deficits (n = 3 vs. n = 48; p = 0.051; Table 1).
Table 1.
Characteristics of the subcortical hematoma in the dAVF and without dAVF groups
| Hemorrhage cases | Nonhemorrhage
cases |
|||
|---|---|---|---|---|
| dAVF group
(n = 7) |
Without dAVF group
(n = 60) |
p-value | dAVF (n = 6) | |
| Age | 66.1 ± 11.1 | 71.7 ± 13.2 | 0.29 | 63.8 ± 10.1 |
| Sex (Male:Female) | 4:3 | 28:32 | 0.70 | 5:1 |
| Presenting symptoms | ||||
| ∙ DOC | 4 (57.1%) | 55 (91.7%) | 0.03 | 0 (0.0%) |
| ∙ Headache | 1 (14.3%) | 15 (25.0%) | 1 | 3 (50.0%) |
| ∙ Seizure | 4 (57.1%) | 5 (8.3%) | <0.01 | 0 (0.0%) |
| ∙ Focal neurological deficit | 3 (42.9%) | 48 (80.0%) | 0.051 | 5 (83.3%) |
Abbreviations: dAVF, dural arteriovenous fistula; DOC, disturbance of consciousness
Furthermore, there were 37 patients in the nonhemorrhagic dAVF group. Among them, only six patients underwent 3D-CTA to discriminate dAVF from other cerebrovascular diseases such as cerebral aneurysm and cervical internal cerebral artery stenosis (Supplementary Fig. 1). The parameters for the patients in the nonhemorrhagic dAVF group were age, 63.8 ± 10.1; males, 83.3%; 50% of patients with headache; and 83.3% of patients with focal neurological deficits such as aphasia, narrowing of vision, and cognitive impairment due to brain edema (Table 1).
As shown in Table 2, the shunt points in the hemorrhagic dAVF group were as follows: one at the cavernous sinus, three at the superficial sagittal sinus, and three at the transverse sinus. All patients were classified with Cognard type IIb or higher dAVF, and the hemorrhage locations varied. All patients underwent transvenous or transarterial embolization with or without open surgery.
Table 2.
Clinical and radiological characteristics of the seven patients in dAVF group
| No | Age | Sex | Shunt
location |
Cognard
class |
Bold-S
sign |
Maximum
diameter (mm) |
Ipsi/cont
ratio |
Hematoma
location |
Presenting
symptom |
Treatment | mRS at
discharge |
Diagnosis by
3D-CTA |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 81 | F | CS | II b | Yes | 2.15 | 1.63 | Lt. frontal | Aphasia | TVE | 2 | Impossible |
| 2 | 71 | F | TS | IV | Yes | 2.98 | 1.74 | Lt. temporal | DOC | TVE | 1 | Impossible |
| 3 | 62 | M | SSS | II b | Yes | 2.50 | 1.93 | Lt. frontal | Seizure | TAE | 0 | Impossible |
| 4 | 70 | M | TS | II a+b | Yes | 2.69 | 1.59 | Lt. occipital | Aphasia Hemiparesis |
TVE | 2 | Suspectable |
| 5 | 49 | M | SSS | IV | Yes | 2.76 | 2.40 | Rt. occipital | Headache | TAE + open surgery |
1 | Possible |
| 6 | 74 | F | TS | III | Yes | 2.66 | 1.60 | Lt. temporal | DOC Aphasia Hemiparesis |
TVE | 3 | Impossible |
| 7 | 56 | M | SSS | II b | Yes | 2.41 | 1.71 | Lt. temporal | DOC Seizure | TAE | 3 | Suspectable |
Abbreviations: 3D-CTA, three-dimensional computed tomography angiography; Class, classification; cont. contralateral; CS, cavernous sinus; DOC, disturbance of consciousness; F, female; ipsi, ipsilateral; lt, left; M, male; mRS, modified Rankin Scale; rt, right; SSS, superior sagittal sinus; TAE, transarterial embolization; TS, transverse sinus; TVE, transvenous embolization
Figures 1 and 2 show representative images of the hemorrhagic dAVF and without dAVF groups, respectively. All patients in the hemorrhagic dAVF group showed tortuous and expanded vessels around the hematomas, which appeared as S-shaped vessels and were clearly more dilated than the contralateral normal branches (Fig. 1). Some cases showed dilated abnormal vessels in the area apart from the hematoma, and their coronal (and/or sagittal) recontraction images were sometimes useful to detect those vessels (Supplementary Fig. 2). By contrast, in patients in the without dAVF group, the S-shaped vessels were not frequently seen (12 of 60 patients), and the diameter of the S-shaped vessels on the ipsilateral side was not relatively large compared with those on the contralateral side (Fig. 2).
All patients in both the hemorrhagic dAVF and nonhemorrhagic dAVF groups showed a clear “bold-S sign.” To confirm this finding, we analyzed the vessel diameters (Table 2). As shown in Table 3, the maximum diameter of the S-shaped vessels was 2.59 ± 0.27 mm in the hemorrhagic dAVF group and 1.69 ± 0.22 mm in the without dAVF group. The diameter ratio (diameter of ipsilateral S-shaped/contralateral largest vessels) was also significantly different: 1.80 ± 0.29 (1.59-2.40) in the hemorrhagic dAVF group and 1.07 ± 0.16 (0.78-1.38) in the without dAVF group (p < 0.01; Table 3). Thus, the cutoff point of the bold-S sign was chosen as >1.50. Accordingly, the bold-S sign was not found in the without dAVF group even in the presence of S-shaped vessels, and all patients in the hemorrhagic dAVF group showed a bold-S sign with 100% sensitivity and specificity (Table 3). The abovementioned phenotypes in the hemorrhagic dAVF group were seen in the nonhemorrhagic dAVF group. In addition, we retrospectively reviewed our conference records whether 3D-CTA alone could detect subcortical hemorrhage due to dAVF on admission. As shown in Table 2, only one case was “possible,” four cases were “impossible,” and two cases were “suspectable.” Interestingly, three patients who were judged as “suspectable” were finally diagnosed as AVM by DSA (Supplementary Fig. 1). Furthermore, all vessels with a “bold-S sign” corresponded to angiographical dilated veins (Fig. 1 and Supplemental Fig. 2).
Table 3.
Features of the S-shaped vessels in the dAVF and without dAVF groups
| Hemorrhage cases | Nonhemorrhage
cases |
|||
|---|---|---|---|---|
| dAVF | Without dAVF | p-value | dAVF | |
| Total number | 7 | 60 | 6 | |
| S-shaped vessel | ||||
| Positive cases | 7 (100%) | 12 (20%) | <0.01 | 6 (100%) |
| Max diameter (mm) | 2.15–2.98 | 1.2–1.92 | 3.70–2.01 | |
| Mean ± SD (mm) | 2.59 ± 0.27 | 1.69 ± 0.22 | <0.01 | 3.04 ± 0.76 |
| Ipsi/contra ratio | 1.59–2.40 | 0.78–1.38 | 1.63–3.25 | |
| Mean ± SD | 1.80 ± 0.29 | 1.07 ± 0.16 | <0.01 | 2.16 ± 0.62 |
| Bold-S sign | ||||
| Positive cases | 7 (100%) | 0 (0.0%) | 6 (100%) | |
Abbreviations: contra, contralateral; dAVF, dural arteriovenous fistula; ipsi, ipsilateral; max, maximum; SD, standard deviation
Discussion
The strategy for the treatment of subcortical hematomas caused by dAVF was different from that for another simple hematoma case where 1) DSA in the early phase after admission is recommended; 2) endovascular embolization is sometimes needed before surgical removal for safer operation; and 3) even if the subcortical hematoma is not large and can be treated conservatively, the physician should manage the dAVF surgically because the vascular anomaly shows frequent rebleeding within the first two weeks after the initial hemorrhage.2,3) Therefore, our present study focused on patients with a subcortical hemorrhage caused by dAVF and tried to find specific findings in their CTA images on admission. In particular, an increase in the exact diagnosis is helpful in avoiding rebleeding, which was observed within the first two weeks. We found dilated S-shaped vessels around the hematomas in patients with dAVF.5,9) The “bold-S sign” was observed in all patients with dAVF with or without hematomas in our hospital. Indeed, there were S-shaped bent and meandering vessels in the without dAVF group, and it was initially difficult to discriminate them from the bold-S sign. However, the two vessel types could be clearly distinguished based on the diameter ratio at an axial view, which is an important characteristic of the bold-S sign. We thought that the ratio, which ranged from 1.59 to 2.40 in the hemorrhagic dAVF group, was critical. The maximum and mean values of the ratio in the without dAVF group were 1.38 and 1.06, respectively. Therefore, we set the cutoff point at 1.5 to detect the subcortical hemorrhage caused by dAVF.
We consider that the bold-S sign reflects a cortical or intramedullary venous reflux, as reflected by a dilated vessel. All patients in the hemorrhagic dAVF group were graded as having Cognard type IIb dAVF. A cortical/intramedullary venous reflux on cerebral angiography might match the bold-S sign. The presence of a venous reflux is a sign of the hemorrhagic risk associated with dAVF.10-12) However, the bold-S sign was not found near the hematoma in any of the cases, and the findings were not related to the point of hemorrhage. Duffau et al.3) reported that 7 (35%) of 20 patients with dAVF presenting with ICH and patients with cortical venous reflux experienced rebleeding in the first two weeks after the initial hemorrhage and that a hemorrhage caused by a dAVF with cortical venous reflux requires early therapeutic intervention. All patients in the hemorrhagic dAVF group were provided conservative treatment because their hematomas were not large. Thus, if the dAVF is misdiagnosed in the early period and only conservative treatment is planned, patients might experience poor outcomes due to rebleeding. Therefore, the bold-S sign could be a useful predictor of the need for early treatment of dAVF to avoid poor patient outcomes.
This study has several limitations. This was a single-center, retrospective study with a small sample size. Another limitation is the existence of several confounding factors for the measurement of the blood vessel diameter in CTA. The appearance of the blood vessels may differ based on several factors, such as the machine model, contrast medium, patient movement, and intracranial pressure. Another important limitation is the selection bias based on our institutional system, which is an emergency and critical care medical center. Therefore, some patients were possibly transferred for suspected subcortical hemorrhages caused by the dAVF. Moreover, it would be better to evaluate the S-shaped vessels using the coronal and sagittal section on CT in addition to the axial findings. However, we would like to provide an easy diagnosis method on admission for emergency physicians. In the future, we are planning to conduct prospective studies with a larger sample size to validate our results.
Conclusion
We propose the “bold-S sign” as a specific finding for predicting subcortical hemorrhage caused by dAVF on admission. This sign might be associated with the cortical venous reflex, in the absence of a strong relationship with the bleeding point. The bold-S sign can help emergency physicians in the rapid identification of subcortical hemorrhage caused by dAVF and allows early surgical treatment before rebleeding can occur.
Ethics Approval
The study was approved by the Ethics Committee of Kurume University Hospital (No. 1614; date of registration: 17.10.2016) and performed in accordance with the principles of the Declaration of Helsinki.
Informed Consent
Written informed consent was obtained from the main relatives of all patients before enrolment in this study.
Conflicts of Interest Disclosure
The authors have no relevant financial or nonfinancial interests to disclose.
Supplementary Material
Flowchart outlining the inclusion and exclusion criteria
Abbreviations: AVM, arteriovenous malformation; CTA, computed tomography angiography; dAVF, dural arteriovenous fistula; without dAVF, without dural arteriovenous fistula; DSA, digital subtraction angiography; MRA, magnetic resonance angiography; MRI, magnetic resonance imaging.
Additive representative images of the bold-S sign in the dAVF group
CT angiography (a, b, d, e), and diagnostic angiography (c, f) images of patients no. 7 and 5 specified in Table 2 are shown. Black arrows (c, f) indicate corresponding abnormal vessels that are seen as a dilated S-shaped ones in each 3D-CTA (b, e; white arrows).
Acknowledgments
This study was partially funded by JSPS KAKENHI, Grant Number 21K09193.
References
- 1). Gandhi D, Chen J, Pearl M, Huang J, Gemmete JJ, Kathuria S: Intracranial dural arteriovenous fistulas: classification, imaging findings, and treatment. AJNR Am J Neuroradiol 33: 1007-1013, 2012 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2). Choi JH, Jo KI, Kim KH, et al. : Early rebleeding of intracranial dural arteriovenous fistulas after an intracranial hemorrhage. Acta Neurochir 159: 1479-1487, 2017 [DOI] [PubMed] [Google Scholar]
- 3). Duffau H, Lopes M, Janosevic V, et al. : Early rebleeding from intracranial dural arteriovenous fistulas: report of 20 cases and review of the literature. J Neurosurg 90: 78-84, 1999 [DOI] [PubMed] [Google Scholar]
- 4). Lee CW, Huang A, Wang YH, Yang CY, Chen YF, Liu HM: Intracranial dural arteriovenous fistulas: diagnosis and evaluation with 64-detector row CT angiography. Radiology 256: 219-228, 2010 [DOI] [PubMed] [Google Scholar]
- 5). Narvid J, Do HM, Blevins NH, Fischbein NJ: CT angiography as a screening tool for dural arteriovenous fistula in patients with pulsatile tinnitus: feasibility and test characteristics. AJNR Am J Neuroradiol 32: 446-453, 2011 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6). Lin YH, Wang YF, Liu HM, Lee CW, Chen YF, Hsieh HJ: Diagnostic accuracy of CTA and MRI/MRA in the evaluation of the cortical venous reflux in the intracranial dural arteriovenous fistula DAVF. Neuroradiology 60: 7-15, 2018 [DOI] [PubMed] [Google Scholar]
- 7). Cohen SD, Goins JL, Butler SG, Morris PP, Browne JD: Dural arteriovenous fistula: diagnosis, treatment, and outcomes. Laryngoscope 119: 293-297, 2009 [DOI] [PubMed] [Google Scholar]
- 8). Orito K, Hirohata M, Nakamura Y, et al. : Leakage sign for primary intracerebral hemorrhage: a novel predictor of hematoma growth. Stroke 47: 958-963, 2016 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9). Willinsky R, Goyal M, terBrugge K, Montanera W: Tortuous, engorged pial veins in intracranial dural arteriovenous fistulas: correlations with presentation, location, and MR findings in 122 patients. AJNR Am J Neuroradiol 20: 1031-1036, 1999 [PMC free article] [PubMed] [Google Scholar]
- 10). Cognard C, Gobin YP, Pierot L, et al. : Cerebral dural arteriovenous fistulas: clinical and angiographic correlation with a revised classification of venous drainage. Radiology 194: 671-680, 1995 [DOI] [PubMed] [Google Scholar]
- 11). van Dijk JM, terBrugge KG, Willinsky RA, Wallace MC: Clinical course of cranial dural arteriovenous fistulas with long-term persistent cortical venous reflux. Stroke 33: 1233-1236, 2002 [DOI] [PubMed] [Google Scholar]
- 12). Zhao LB, Suh DC, Lee DG, et al. : Association of pial venous reflux with hemorrhage or edema in dural arteriovenous fistula. Neurology 82: 1897-1904, 2014 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Flowchart outlining the inclusion and exclusion criteria
Abbreviations: AVM, arteriovenous malformation; CTA, computed tomography angiography; dAVF, dural arteriovenous fistula; without dAVF, without dural arteriovenous fistula; DSA, digital subtraction angiography; MRA, magnetic resonance angiography; MRI, magnetic resonance imaging.
Additive representative images of the bold-S sign in the dAVF group
CT angiography (a, b, d, e), and diagnostic angiography (c, f) images of patients no. 7 and 5 specified in Table 2 are shown. Black arrows (c, f) indicate corresponding abnormal vessels that are seen as a dilated S-shaped ones in each 3D-CTA (b, e; white arrows).