Introduction
Subarachnoid haemorrhage (SAH) is caused by aneurysmal rupture in 70–85% of patients [
1,
2]. In a systematic review, Hop et al. found fatality rates ranging from 32% to 67% [
3]. Furthermore, 10–20% of patients remained functionally dependent after SAH. Rapid diagnostic evaluation and treatment are crucial for the patient’s outcome.
Intraarterial digital subtraction angiography (IA-DSA) has been the main technique for detecting and characterizing intracranial aneurysms and remains the gold standard. However, IA-DSA is invasive and time consuming, and carries a risk of neurological complications of 0.5–1.8% with permanent deficit in 0.09–0.5% [
4‐
6]. Serious non-neurological complications, which occur in 0.6% of patients, include groin hematoma, peripheral thromboembolism, transient hypotension and arteriovenous fistulas [
4]. Furthermore, IA-DSA may increase the risk of rebleeding [
7,
8]. It has been demonstrated that three-dimensional CT angiography (3D-CTA) can reliably detect intracranial aneurysms [
9‐
14]. Only after replacement of IA-DSA by CTA can the advantages of CTA be fully realized in the clinical setting. We report here our clinical experience with both 16- and 64-detector row CTA as the first and intended only diagnostic and treatment decision-making study for intracranial aneurysms in patients with acute SAH.
Materials and methods
Subjects
Between April 2003 and January 2006 all patients presenting with a SAH to the University Medical Centre Groningen consecutively underwent CTA as the first diagnostic study. Based on the CTA findings, patients were selected for surgical clipping or endovascular coiling of a ruptured intracranial aneurysm.
SAH was suspected on clinical grounds and confirmed by unenhanced CT or by blood pigments on lumbar puncture.
Imaging protocols
The CT examinations were performed on a 16- or 64-multidetector row spiral CT machine (Somatom Sensation 16 or 64; Siemens Medical Systems, Erlangen, Germany), based on a standard protocol. The 64-multisclice CT was implemented in the Emergency Department in December 2004.
-
Parameters for 16-slice CT for diagnosis of aneurysm: via an intravenous cannula in the antecubital fossa, 80 ml of contrast agent (Visipaque 320) was injected with a power injector at a rate of 4 ml/s. Injection of contrast agent was followed by a flush of 50 ml 0.9% saline (Stellant; NaCl Neck Angio) injected at the same rate. A manual fluoroscopic bolus-triggered system, with the internal carotid arteries as reference point and a delay of 4 s, determined the optimal timing. The CTA protocol parameters were as follows: spiral mode, rotation time 0.5 s, reconstruction interval 0.75 mm at Kernel H20, 120 kV/200 mAs, acquisition time 10 s, scan range from the C1 vertebral body to the vertex parallel to the orbitomeatal line.
-
Parameters for 16-slice CT for diagnosis of SAH: gantry un-angled, spiral mode, rotation time 0.75 s, 16-detector rows at 0.75-mm intervals, table speed 6 mm/rotation, reconstruction interval 3 mm at Kernel H30 and acquisition parameters 120 kV/200 mAs. The actual acquisition time was approximately 15 s.
-
Parameters for 64-slice CTA for diagnosis of aneurysm: rotation time 1 s, table speed 15.4 mm/rotation, reconstruction interval 0.6 mm at Kernel H20, 120 kV/260 mAs, acquisition time 9 s and scan range extending from the C1 vertebral body to the vertex parallel to the orbitomeatal line. The protocol parameters for contrast agent injection remained unchanged.
-
Parameters for 64-slice CT for diagnosis of SAH: gantry un-angled, spiral mode, rotation time 1 s, 64 detector rows at 0.6-mm intervals, table speed 9.6 mm/rotation, reconstruction interval 2 mm at Kernel H30, acquisition parameters 120 kV/260 mAs and acquisition time 14 s.
Postprocessing of CTA
Source images were transferred to a remote computer workstation (Odelft Benelux diagnostic imaging) for viewing. Initial careful review of axial images was considered imperative. During this review any areas of concern could be noted. Two-dimensional maximum intensity projection (MIP) views and three-dimensional (3-D) surface-rendered and volume-rendered reconstructions were reformatted from the raw image date on a Vitrea computer workstation by one of the neuroradiologists.
Parameters for IA-DSA and postprocessing
From April 2003 until April 2004 the IA-DSA studies were produced on a digital angiographic unit (Siemens Multiskop with InfiMed image processing) with a 512×512 pixel matrix. From April 2004 onwards the studies were performed on a Siemens Axiom Artis angiographic unit with a 1024×1024 pixel matrix. Selective four- or six-vessel angiography using a standard projection format was performed initially and additional views were obtained if required to identify the parent vessel and aneurysm neck more clearly. The amount of contrast medium (Visipaque 270) used was 8 ml for the internal carotid artery and 6 ml for the external carotid artery, and the injection rate was 6 ml/s when the tip of the catheter was in the internal carotid artery and 3–4 ml/s when the tip of the catheter was in the external carotid artery. The rate of injection into the vertebrobasilar system was 6–8 ml/s to a total amount of 8 ml.
In certain situations, rotational 3-D angiography was performed to better delineate the anatomic details of an aneurysm. Rotational 3-D angiography was performed on a Siemens Axiom Artis angiographic unit. The C-arm rotates in a continuous 200° arc around the patient’s head during a prolonged intraarterial catheter injection of contrast medium (28 ml Visipaque, injection rate 4 ml/s). The raw date images were transferred to a Leonardo workstation (AX Applications) from which 3-D volume-rendered reconstructions were reformatted.
Image review and data analysis
The presence of an aneurysm, its size and morphology, its parent and feeding vessels and the collateral circulation at the circle of Willis were determined by one of the diagnostic or interventional neuroradiologists. If multiple aneurysms were detected, the usual criteria were applied to decide which aneurysm was responsible for the haemorrhage. These criteria included the unenhanced CT findings (distribution of blood) and the size and irregularity of the aneurysm.
All diagnostic findings were discussed with the neurosurgeons. The CTA results were categorized into proven ruptured aneurysm, inconclusive or negative. Patients with a proven ruptured aneurysm were selected subsequently for coiling or clipping. The surgical and endovascular findings were compared to the CTA findings. In general, ruptured aneurysms in the anterior circulation were selected for either coiling or clipping. Ruptured aneurysms located in the posterior circulation were preferably coiled. Giant intracranial aneurysms were preferably treated surgically. A ruptured aneurysm in association with an intraparenchymatous haemorrhage was most often selected for clipping of the aneurysm and surgical evacuation of the haematoma.
Patients categorized as inconclusive or negative underwent IA-DSA. In patients with a perimesencephalic blood distribution, one IA-DSA examination was performed. In patients with a nonperimesencephalic blood distribution a second IA-DSA was performed if the first one was negative. IA-DSA was considered the gold standard. CTA was considered false-negative when IA-DSA revealed a ruptured aneurysm or when rebleeding occurred.
The positive predictive value, negative predictive value, sensitivity, specificity and accuracy of CTA per patient were calculated. The chi-squared test was used to compare the performance of 16-slice CTA and 64-slice CTA for the identification of intracranial aneurysms. Differences with a P value less than 0.05 were considered significant.
The IA-DSA findings in patients in the inconclusive category were compared with the CTA findings to assess whether IA-DSA actually provided any additional information.
Discussion
Our primary aim was to assess whether CTA is useful clinically in planning and performing clipping or coiling, especially in the acute phase in ruptured intracranial aneurysms, without recourse to IA-DSA. We demonstrated that it was possible to treat more than half of all patients with a ruptured intracranial aneurysm using only CTA. By avoiding conventional angiography, it was possible to streamline the management of ruptured aneurysm during the acute phase. Further, 3D-CTA was able to help in deciding whether to clip or to coil; in only two patients was treatment conversion needed due to incorrect treatment selection based on CTA.
We found 3D-CTA to be a simple, reliable, quick and minimally invasive imaging modality that reduces the risk of complications caused by conventional angiography and reduces the delay between the patient’s arrival at the hospital and treatment, leading to diminished rebleeding. Matsumoto et al. analyzed the rate of rebleeding of ruptured aneurysms during CTA and conventional angiography, and found 0% (none of 160 patients) for CTA and 1.5% (5 of 317 patients) for conventional angiography [
17]. In patients with a ruptured aneurysm and intracerebral haemorrhage CTA saves time when aiming for a fast clot removal. Another advantage is that the radiation dosage is low compared to IA-DSA (1.0 mSv at 200 mAs with the CTA Siemens Sensation 16 and 1.8 mSv at 380 mAs with the CTA Siemens Sensation 64 compared with 3.5–6.5 mSv with conventional angiography). Furthermore, the cost of CTA is one-fourth that of conventional angiography.
Several other studies assessing whether CTA may serve as the sole imaging method for the preoperative work-up of patients with ruptured intracranial aneurysms have been published [
17‐
26]. An overview of these previous studies is presented in Table
9. There is a wide variation in the percentage of patients who have had their symptomatic aneurysms treated based on CTA. This may be influenced positively by the very high aneurysm prevalence and the subsequent very low negative rates of CTA in some studies [
22,
23,
25]. In other studies patients with a negative CTA were not enrolled at all [
17,
20]. In general, a mean of 15–20% negative angiographies after SAH is accepted [
27]. The present study showed a high negative rate for CTA. This may reflect the good awareness of the diagnosis SAH in first-line and second-line health-care and the good access to CTA when the diagnosis SAH is considered. Furthermore, the wide variation in CTA-based treatment may be partially explained by differences in hardware and software used by each group, the rate of technical failures in performing CTA, scanning parameters set for screening the circle of Willis and more peripheral vessels, the experience and scrutiny of the neuroradiologist evaluating each CTA and the willingness of the neurosurgeon and neurointerventional radiologist to rely on CTA alone in each individual case.
Table 9
Presentation of previous studies and present study
| 87 | 46 (55%) | 44 (96%) | 44 (100%) | 15 (17%) | 6 (60%) | 26 (30%) |
| 109 | 88 (81%) | 87 (99%) | 87 (100%) | 5 (5%) | 5 (100%) | 16 (15%) |
| 84 | 62 (74%) | 62 (100%) | 62 (100%) | 7 (8%) | 0 (0%) | 15 (18%) |
| 90 | 45 (100%) | 45 (100%) | 45 (100%) | – | – | 45 (50%) |
| 150 | 61 (41%) | 61 (100%) | 60 (98%)a
| 24 (16%) | 24 (100%) | 65 (43%) |
| 120 | 40 (27%) | 40 (100%) | 40 (100%) | 13 (11%) | 13 (100%) | 67 (56%) |
| 78 | 27 (35%) | 27 (100%) | 27 (100%) | 20 (26%) | 20 (100%)b
| 31 (40%) |
| 100 | 93 (93%) | 93 (100%) | 93 (100%) | – | – | 7 (7%) |
| 96 | 87 (91%) | 87 (100%) | 86 (99%)2
| – | – | 9 (9%) |
| 61 | 44 (72%) | 44 (100%) | 44 (100%) | 15 (25%) | 14 (93%) | 2 (3%) |
Present study | 224 | 133 (59%) | 133 (100%) | 132 (99%)2
| 60 (27%) | 55 (92%) | 31 (14%) |
In the present study CTA was false-negative in 8% of patients. The risk of rebleeding after a negative initial CTA was 7%. All false-negatives were in patients with a nonperimesencephalic blood distribution, giving a false-negative rate of 29% and a risk of rebleeding of 24%. It seems unlikely that the false-negative rate of initial CTA and the risk of rebleeding despite a negative initial CTA in patients with a nonperimesencephalic SAH might be influenced negatively by the use of CTA as the first diagnostic tool. Firstly, in all patients with a rebleeding, repeat IA-DSA was also false-negative. Secondly, repeat angiography with CTA performed after a rebleeding still demonstrated an aneurysm.
Furthermore, the findings of other studies using IA-DSA as the first diagnostic tool were similar. In the study by Urbach et al. in 67 patients with a negative initial angiogram after SAH, four ruptured aneurysms were revealed by repeat angiography [
28]. Three patients presented with a nonperimesencephalic SAH and one presented with a perimesencephalic SAH. In the study by Bradac et al., 60 of the 440 patients presenting with spontaneous SAH had a negative angiogram [
29]. A second angiogram performed 1–4 weeks later revealed a ruptured aneurysm in 5 of the 40 patients. Of these patients, 3 had a second SAH. In all patients a nonperimesencephalic blood distribution was seen on CT.
Because in the present study some aneurysms could be correctly identified retrospectively, we suggest that if, under strong clinical suspicion of a ruptured aneurysm, the CTA is reported as normal, the study should be reviewed by a second neuroradiologist before proceeding to repeat angiography. It is essential to perform a review of axial raw source images. Next, we recommend repeat CTA or IA-DSA when the initial CTA is negative in patients with a nonperimesencephalic SAH. There is no consensus about the time interval for repeat angiography. In practice, the guideline is to repeat angiography after several days to months. The substantial risk of rebleeding in patients with an aneurysmal pattern of haemorrhage in the present study indicates that some cerebral aneurysms are occult on initial CTA. Several factors may explain this finding. Most importantly, there is a learning curve in assessing aneurysms on CTA. Pedersen et al. reported an increase in sensitivity from 88% to 94% after 1 year’s experience [
30]. Small aneurysms can be missed when using CTA. CTA had a sensitivity of 50% for aneurysms <2 mm in the study of Wintermark et al. [
13]. Distal pericallosal and PICA aneurysms can be missed when restricting the area of coverage to the proximal circle of Willis [
31‐
34]. Thrombosis of the neck of the aneurysm or of the entire sac is another possible reason [
23]. Perianeurysmal blood or haematoma may reduce lesion conspicuity [
34]. Aneurysms may be mistaken for vascular infundibula (persistent fetal nonaneurysmal dilatation of the proximal vessel) of the posterior communicating or anterior choroidal artery origins if a vessel cannot be identified arising from them [
35]. Aneurysms may masquerade as tight vascular loops if the MIP thickness is wide (>3 mm) [
34]. In patients with multiple intracranial aneurysms large aneurysms may obscure smaller ones on the CT reconstruction [
33]. Aneurysms close to bone (e.g. carotid siphon, ophthalmic and posterior communicating artery) may be overlooked when relying on surface-rendering and volume-rendering techniques or using MIP with bone editing [
32,
34,
36‐
38]. Aneurysms located within or close to the cavernous sinuses are easy to overlook unless thin-section axial and coronal MIP images are reviewed on a slightly wider window width [
9].
In patients with a perimesencephalic SAH the chance of finding a posterior fossa aneurysm is low: 2.5–5% [
39,
40]. Nonaneurysmal perimesencephalic haemorrhage carries no risk of vasospasm and rebleeding and has been shown to follow a benign course with an excellent prognosis [
41]. The chance of finding an aneurysm in 5% of patients has to be weighed against the risk of complications from angiography imposed upon the remaining 95% of patients. CTA has a high accuracy for diagnosis of vertebrobasilar aneurysms and of intracranial aneurysms in general [
9‐
13,
42]. In the present study, in patients with a perimesencephalic SAH and a negative initial CTA, no rebleedings occurred and CTA was true-negative in all. Similarly, in the prospective study of Huttner et al., 69 patients with a perimesencephalic SAH had a negative initial CTA and IA-DSA [
43]. A repeat IA-DSA was performed in 38 patients (55%). None of the repeat IA-DSAs showed any additional distinctive features with respect to the first IA-DSA. It therefore seems practical and safe to perform CTA as the first diagnostic tool and to omit repeat angiography if CTA is negative. A formal decision analysis based on these observations confirmed that a strategy where CTA is performed and not followed by conventional angiography, if negative, results in a better utility than a strategy of CTA followed by conventional angiography or of conventional angiography as primary investigation [
44].
According to the results of the present study, it seems important to distinguish the two patterns of SAH on CT. The CT criteria of perimesencephalic bleeding have been defined [
40]. Different data show that experienced radiologists can accurately discriminate between a perimesencephalic and nonperimesencephalic SAH [
12,
40,
45]. Early CT within 3 days is necessary for reliable assessment of the pattern of haemorrhage [
12,
40,
46].
A criticism of this study might be that patients treated with endovascular coiling underwent IA-DSA as part of the endovascular procedure and thus should not be counted in the analysis of efficacy of the prospective protocol. However, a shift in management of ruptured intracranial aneurysm from surgery to endovascular treatment has appeared [
47]. Endovascular treatment is replacing clipping. The use of CTA as the initial investigation for cerebral aneurysms may offset some of this increased workload whilst also improving workflow.
In conclusion, in this evaluation of the use of 16-row and 64-row multislice CTA in the management of ruptured intracranial aneurysms, we demonstrated that CTA can be used as the first-line diagnostic modality for the management of SAH patients. In CTA-negative patients IA-DSA provided no or marginal added value. IA-DSA is not needed in patients with negative CTA and classic perimesencephalic SAH. Repeat IA-DSA or CTA should still be performed in patients with a nonperimesencephalic SAH, due to false-negative CTAs and IA-DSAs in this patient group. The remaining true indication for IA-DSA was in patients with an inconclusive CTA result. In more than half of those IA-DSA provided relevant new diagnostic information.