Size and location of ruptured and unruptured intracranial aneurysms measured by 3-dimensional rotational angiography

1. Introduction 

Subarachnoid hemorrhage (SAH) from rupture of an intracranial aneurysm is a devastating event with a mortality of up to 50% [2], [5], [19]. The life-threatening and debilitating sequelae of SAH may be prevented if the aneurysm can be occluded before rupture. Recent epidemiological research [22] on patients with aneurysms and family members, as well as laboratory work on the genome [26], may open efficient ways to identify patients with a high risk of having UIAs. To offer aneurysm treatment to patients with UIAs, the periprocedural risks of surgery or endovascular coiling and the cumulative risk of rupture after treatment must be less than the cumulative risk of rupture with the associated mortality and morbidity when managed conservatively. Unfortunately, the natural history of UIAs is not exactly known. The most common estimations for the risk of rupture of a UIA were between 0.5% and 2.5% per year [9], [12], [13], [24], [31], [32]. Data from the retrospective ISUIA study [25] calculated a much lower risk of rupture (0.05%/year) of aneurysm less than 10 mm in patients with no history of SAH. Prospective data from the ISUIA found a risk of rupture of 0.52% for aneurysms 7 to 12 mm in size located at the anterior circulation and even of 2.9% per year for aneurysms of the same size located at the posterior circulation [30]. However, not a single aneurysm less than 7 mm located in the anterior circulation ruptured during an average of 4.1 years of follow-up of those patients, who did not cross over to the treatment group (31.6% were selected for treatment from the observation group during follow-up, and an additional 11.4% were removed because they died). Besides size, there are other known risk factors such as shape [21], and there are changes in the microarchitecture of an aneurysm affecting its risk of rupture [8]. These data still cause considerable confusion on how to manage a patient with a small aneurysm. Our clinical experience is that small aneurysms constitute most of ruptured aneurysms. Hence, knowing the accurate size of a series of ruptured and unruptured aneurysms may help to identify difference in the size distribution and shed light on the growth pattern of these aneurysms. However, the 3-dimensional (3D) anatomy of intracranial aneurysms is complex, and the variable magnification factor of biplanar digital subtraction angiograms make size measurements difficult. The recently introduced 3D-RA is a method to compute 3D images from projection radiograms. 3D-RA recording systems are exactly calibrated and allow for easy quantification of the size of the aneurysm. The objective of the study was to report the location and accurate size of RIAs and UIAs in a consecutive series of patients by 3D-RA taking the complex 3D anatomy and parent vessel morphology into consideration.

2. Patients and methods 

One hundred thirty-four consecutive patients with intracranial aneurysms admitted to our Department of Neurosurgery from January 2002 to May 2003 were included in the study. Overall, 16 patients were excluded because of unavailability of the 3D-RA device during routine checkup (2), a fusiform aneurysm (1), death or critical clinical condition before 3D-RA could be done (5), presence of a large intraparenchymal hemorrhage with mass effect where additional time spent in angiography suite was judged as hazardous (3), or because the patient already had a high-quality angiogram from the referring hospital (5). Thus, 118 patients with 155 saccular aneurysms were included, and high-resolution 3D-RA of the intracranial vessels bearing the aneurysms was performed.

Patient characteristics such as age, sex, and admission status (Hunt and Hess grading system for aneurysmal SAH H and H) were allocated in a prospectively designed database on an SPSS (SPSS Inc, Chicago, IL) spreadsheet. 3D-RA with computerized reconstruction of the aneurysm and parent vessel was performed on a Philips Allura System (Philips Medical Systems, Eindhoven, Netherlands). One hundred unsubstracted angiographic images were obtained after a bolus of 17 mL of nonionic contrast material (Ultravist 300, Schering, Berlin, Germany) injected with a continuous flow of 3 mL/s. The injection started 1 second before the 5.5-second rotational run of the C-arm. All angiograms were analyzed by the same investigator who was blinded for all clinical information. The size of the aneurysm was determined by measuring the maximal height and width of the aneurysmal sac. A cut plane was positioned at the level of the largest extension of the aneurysmal sac (Fig. 1). Height of the aneurysm was defined as the length of the maximum possible line from the base to the dome that could be fitted on the 3D structure. Thereby, the width of the parent vessel was not included in the measurement; that is, only the actual aneurysmal sac was measured. Width was defined as the length of a line approximately perpendicular to the first line.

View Large Image Download to PowerPoint

Fig. 1. The 3D-RA images (A-G) of a basilar tip (A-D), a pericallosal artery (E), an MCA (F), and an anterior communicating (G) artery aneurysm are shown, and the computerized size-measurement procedure is demonstrated. On the computer screen, the images can be deliberately rotated in all directions. For size measurement, it is possible to position a cut plane at the level of the largest extension of the aneurysmal sac (B, blue line; C, E-G, blue plane). Height of the aneurysm was defined as the length of the maximum possible line from the base to the dome that could be fitted on the 3D structure. Thereby, the width of the parent vessel was not included in the measurement; that is, only the actual aneurysmal sac was measured as indicated by the arrows (D-G). Aneurysm width was defined as the maximum length of a line approximately perpendicular to the first line (D, E, F, G). E-G, The computerized size measurement using 3D-RA images is shown for a pericallosal artery (E), an MCA (F), and an anterior communicating artery (G) aneurysm.

Size and location of all 155 aneurysms were recorded. In cases of multiple aneurysms, the determination of the ruptured aneurysm was performed on the basis of the distribution of blood on CT, aneurysm configuration on the angiogram, and finally, intraoperative findings.

Direct measurement of the aneurysm size was technically feasible in all 155 aneurysms that were detected by 3D-RA.

3. Results 

Eighty-six patients were harboring single; 32 had multiple aneurysms (range 2-5). There were 83 ruptured and 72 UIAs. Mean age of patients with RIA was 53.3 years; of patients with UIA, 53.0 years. Of the patients with RIAs, 60 presented with single ruptured, 22 at least with 1 additional unruptured aneurysm, and 1 patient with SAH already had a history of SAH from another RIA. The 72 UIAs presented either as truly incidental (41) or with SAH from another RIA (27), or due to aneurysm mass effect (1), or due to ischemic events (3) with the aneurysm as the most likely cause of small emboli (Table 1). Patients with SAH were graded according to the H and H scale. Patients of all grades were included in the study: 54 (65%) were good-grade (H and H 1-3) and 29 (35%) were poor-grade (H and H 4-5) patients (Table 2). No patient was excluded because of uncertainty of the site of aneurysm rupture in cases of multiple aneurysms.

Table 1. Symptoms of intracranial aneurysms

		Ruptured (n = 83)			Unruptured (n = 72)
		Single RIA	Additional UIA	History of SAH from other RIA	Incidental UIA	SAH from other RIA	Mass effect	TIA/stroke
	No. of aneurysms	60	22	1	41	27	1	3
	Mean height [mm]	6.9	6.2	2.0	6.4	4.5	14	3.8
	Mean width [mm]	6.1	6.1	5.0	6.4	4.4	13	4.7

Numbers of RIA and UIA and their corresponding size. Ruptured aneurysms are categorized as single aneurysms, as the ruptured one causing SAH in multiple aneurysms, or as ruptured with a history of SAH from another previously ruptured aneurysm. Unruptured aneurysms are categorized as truly incidental, as 1 of multiple aneurysms in SAH from another ruptured aneurysm, as symptomatic because of mass effect (cranial nerve deficit), or as symptomatic due to an ischemic syndrome. TIA indicates transient ischemic attack.

Table 2. Clinical grading

		H and H grades of patients with SAH
		°1	°2	°3	°4	°5	all
	Frequency	7	26	21	18	11	83

Frequency of patients with SAH according to the H and H grading scale is shown. Patients of all grades were included in the study: 54 (65%) were good-grade (H and H 1-3) and 29 (35%) were poor-grade (H and H 4-5) patients.

3.1. Measurements 

Mean height of all (UIA and RIA) aneurysms was 6.2 ± 4.2 mm (mean ± SD); mean width was 5.9 ± 4.0 mm. Median height was 5.3 mm; median width was 5.0 mm. Aneurysm size ranged from 1.2 to 29.3 mm.

Mean height of UIAs was 5.7 ± 4.1 mm; mean width was 5.7 ± 3.6 mm. Median height was 4.4 mm; median width was 4.7 mm. The majority of UIAs were smaller than 10 mm (59/72, 81.9%) and 7 mm (49/72, 68.1%), respectively.

Mean height of RIAs was 6.7 ± 4.2 mm, mean width was 6.1 ± 4.2 mm. Median height was 5.6 mm, median width was 5.0 mm. The majority of RIAs were smaller than 10 mm (68/83, 81.9%) and 7 mm (49/83, 59%), respectively.

There was only a small difference of 1.0 mm in height and of 0.4 mm in width between RIA and UIA. Statistical analysis revealed that this small difference did not reach significance, neither between height (P = .7) nor width (P = .9) of ruptured and unruptured aneurysms (Fig. 2).

View Large Image Download to PowerPoint

Fig. 2. Results of size measurements with 3D-RA are shown. The bars represent the mean height (black) and width (hatched) of all, that is, ruptured and unruptured aneurysms (All), of RIA and of UIA. There was no statistical difference, neither between height (P = .7) nor between width (P = .9) of ruptured and unruptured aneurysms (t test). Error bars indicate the SEM.

The pattern of size distribution of RIAs and UIAs was not different when aneurysms were further subdivided into 2-mm subgroups (Fig. 3). There was no size threshold at which the frequency of ruptured aneurysms increased abruptly or steadily. Starting from the smallest aneurysms, the frequency of RIAs increases, reaching a maximum in the 4- to 6-mm height group—this is the size category at which most aneurysms ruptured. Beyond that size group, the frequency of RIAs is decreasing. Width of ruptured aneurysms was even smaller: the size group with the maximum frequency of ruptured aneurysms was the 2- to 4-mm category, with the frequency declining with larger width.

View Large Image Download to PowerPoint

Fig. 3. A, B: Aneurysms were split into 2-mm size categories. The frequencies of UIAs and RIAs in each of the 2-mm size categories are shown for aneurysm height (A) and width (B). The size categories with the highest frequency for RIAs are between 4 and 6 mm in height and 2 and 4 mm in width.

There was no statistically significant difference in the proportion of small aneurysms (<10/<7 mm) between the group of RIAs and UIAs (P = 1.0 [<10 mm] and P = .32 [<7 mm]).

3.2. Aneurysm location 

Analysis of distribution of UIAs and RIAs by location showed that, in this consecutive group of patients, aneurysms at the anterior cerebral or anterior communicating arteries were significantly more often ruptured than unruptured (P = .02), and aneurysms at the MCA were significantly more often unruptured than ruptured (P = .04, Table 3). Aneurysms were further subdivided by size into 2 groups at the 7-mm threshold. Evaluation of distribution of UIAs and RIAs by size and location revealed that the majority of small (<7 mm) aneurysms, that is, 71.4% and 69.4% of UIAs and RIAs, respectively (not significant, P = .83), were located in the anterior circulation (Table 3).

Table 3. Aneurysm location and size

		UIA (n = 72)			RIA (n = 83)
		<7 mm (n = 49)		>7 mm (n = 23)	<7 mm (n = 49)		>7 mm (n = 34)
	Cavernous part of ICA	2 (4.1%)	71.4%	0 (0%)	0 (0%)	69.4%	0 (0%)
	ICA	6 (12.2%)		4 (17.4%)	2 (4.1%)		5 (14.7%)
	ACA or AcommA	10 (20.4%)		6 (26.1%)	22 (44.9%)		11 (32.4%)
	MCA	17 (34.7%)		7 (30.4%)	10 (20.4%)		5 (14.7%)
	PcommA	10 (20.4%)	28.6%	2 (8.7%)	6 (12.2%)	30.6%	2 (5.9%)
	Vertebrobasilar	2 (4.1%)		1 (4.3%)	4 (8.2%)		4 (11.8%)
	Basilar tip	2 (4.1%)		3 (13.0%)	5 (10.2%)		7 (20.6%)

The frequency of aneurysm according to location and size is shown. Aneurysms are divided in 2 size categories: small (<7 mm) and large (>7 mm). The majority of both UIAs and RIAs that are smaller than 7 mm lie in the anterior circulation (71.4% and 69.4%, respectively; P = 1.0, Fisher exact test). ACA indicates anterior cerebral artery; AcommA, anterior communicating artery; ICA, internal carotid artery; PcommA, posterior communicating artery.

4. Discussion 

In this study, 3D-RA as an advanced device was used to measure the size of RIAs and UIAs with the currently highest technical accuracy available. Although biplanar 2-dimensional (2D) DSA is still considered the gold standard for depiction of intracranial aneurysms, recent publications have shown that 3D-RA provides significantly more detailed and accurate information for the evaluation and measurement of cerebral aneurysms [10], [23]. Even more important than the possible technical accuracy of this method is the image accuracy that can be achieved in the clinical setting. With the computerized reconstruction of the 3D aneurysm morphology, it is possible to visualize and rotate the complete aneurysmal sac morphology on a computer screen and therefore to easily acknowledge the largest extension of an aneurysm (Fig. 1). On conventional DSA, only if the axis of radiation is perpendicular to the largest extension of the aneurysm that the measured size on the 2D films will be correct. Unfortunately, due to the limited number of projection views, it is not always possible to reproduce the appropriate one, which may result in incorrect measurements.

The variable magnification of DSA causes geometric distortion and introduces a considerable uncertainty factor [6]. Even small changes of the distance from the aneurysm to the image intensifier or to the radiation source will lead to false results of aneurysm diameter. Because there are no standardized scales, radiation opaque markers (eg, coins) are placed on the patient scalp to be able to gauge measurements. Some studies even estimated aneurysm size in comparison to the diameter of a certain portion of an intracranial or extracranial artery—the diameters of the respective arteries were regarded as known from mean values [16].

In 3D-RA, however, an automated calibration procedure applied to all images corrects for geometric distortion. This results in a technical accuracy of approximately 0.2 mm [4] and a negligible visual distortion [1].

In aneurysms with broad necks, size may be difficult to determine on plain 2D angiograms. If the neck of the aneurysmal sac is not well defined, then aneurysm size may be overestimated by measuring from the base of the parent vessel instead of an imaginary line along the neck that can be easily drawn with the aid of the computer after choosing the best view on the screen (Fig. 1). These procedures add an error of unknown magnitude when measuring aneurysm size on biplanar 2D-DSA. Especially with small aneurysms, biplanar DSA may lead to overestimation of the real size. Using the technical advances of the 3D-RA, however, these sources of mismeasurement of aneurysm size, as may occur on plain 2D images with projection artifacts, can easily be avoided.

In this series, 2 very small aneurysms were not detected at all by DSA but by 3D-RA, demonstrating the superiority of 3D-RA. Projectional obscuration has been accused to be responsible for false-negative results of biplanar DSA [10]. Small aneurysms can also be missed by current CTA as was shown by Kangasniemi et al [14]. In summary, aneurysm size measurement with 3D-RA may result in a more precise estimation of real aneurysm size in patients harboring such lesions.

The mean (median) height of 6.7 (5.5) mm and width of 6.1 (5.0) mm of RIAs, as well as of 5.7 (4.5) mm and 5.7 (4.9) mm of UIAs as measured with the advanced 3D-RA in our study, were smaller than in previous clinical studies: as compared with the mean (median) maximum diameter of 8.2 ± 4 (7) mm of RIAs from the Cooperative Aneurysm Study from 1970 to 1977 [27], to the mean size of 11.4 ± 1.2 mm of RIAs and to 7.6 ± 1.5 mm of UIAs from the Johns Hopkins series [3], to the mean diameter of UIAs in the retrospective group of the ISUIA without a history of SAH of 10.9 mm [25], to the mean size of 10.8 mm of RIAs and to 7.8 mm of UIAs as reported by Weir et al [29], or to the mean size of 8.0 mm of RIAs in a recent study by Weir [28]. Although comparing distinct patient cohorts, the consistency of these findings suggests that aneurysm measurement by 3D-RA yields smaller sizes than measurement by 2D-DSA.

Other 3D techniques such as CTA are emerging with the advantage to depict the vascular anatomy as well as surrounding structures. In a study of 45 patients, 43 presenting with SAH, mean aneurysm size as assessed by 3D-CTA was 6.78 mm which is almost exactly the size of the RIAs (6.7 mm) depicted in our study [15]. This underscores the need and superiority of a 3D visualization of aneurysms. An in vitro study comparing aneurysm volume between CTA and 3D-RA also showed advantage of 3D-RA [20].

Not only that the mean size of RIAs is smaller than in previous studies, but also the size categories with most RIAs, that is, the group with the highest frequency of RIAs (4-6 mm in height and 2-4 mm in width), are at a substantially lower range than in previous reports. In the present study, we did not measure aneurysm size by biplanar angiography and thus cannot quantitatively compare our 3D-RA results with biplanar measurements of the same aneurysms. Due to this limitation, further studies directly comparing biplanar (2D) DSA with 3D-RA are necessary to define a 3D-RA measurement that corresponds to the 7-mm biplanar (2D) measurement. It would be of considerable clinical significance if the potential measurement errors in other studies are confirmed in a more extensive group of patients and from different centers, because all size thresholds in the literature postulating an increased risk of rupture of large aneurysms are based on biplanar (2D) DSA.

In comparing the results from other studies with our findings, one has to assume that, especially in small aneurysms (Fig. 4), the many listed drawbacks and sources of error in size determination by conventional DSA may account for these differences.

View Large Image Download to PowerPoint

Fig. 4. The 3D-RA images (A-C) illustrate very small intracranial aneurysms of the anterior communicating artery (A: ruptured, 2.6 × 1.9 mm), the MCA bifurcation (B: unruptured, 1.7 × 1.7 mm), and the M1 segment of the MCA (C: unruptured, 1.3 × 1.7 mm) of 3 different patients.

Our 3D-RA data do not confirm that ruptured aneurysms are larger than unruptured ones [3], [25], [28], [29]. The data show that there is only a small difference between RIAs and UIAs of 1.0 mm in height and 0.4 mm in width, not reaching a statistically significant level.

The risk of rupture of UIAs cannot be extrapolated from evaluation of RIAs alone without knowing the prevalence of UIAs. There is, however, a striking mismatch between our data and the natural history of UIAs as found by ISUIA. Ruptured aneurysms less than 7 mm located in the anterior circulation constitute the majority of all RIAs found in our consecutive patients (Table 3), but unruptured aneurysms less than 7 mm located at the anterior circulation bear a risk of rupture of 0% according to the prospective ISUIA data. These numbers are at odds to such an extent that we believe that aneurysm size, at least in aneurysms of 10 mm or less, may not be the most important determinant for the risk of rupture, and small aneurysms do rupture as well. Because the latter is unequivocally proven by reality, there must be another factor in our model of when UIAs rupture.

One possible explanation would be that aneurysms shrink after rupture, and therefore, the postrupture measurement of aneurysm size yields too small diameters for the size at which aneurysms actually rupture. Forget et al [7] discussed histological and angiographic evidence that this is not the case. In our own cases of intra-angiographic aneurysm rupture, a decrease in size after rupture has never been observed. Moreover, when considering the surgical anatomy and morphology of ruptured aneurysms during clipping, there is no intraoperative finding that would support such a theory. This makes the other possible explanation more likely: aneurysms that rupture form acutely and rupture very shortly after their formation. These aneurysms escape from being diagnosed as unruptured aneurysms—regardless of their size. According to our data, these ruptured aneurysms would be small. The aneurysms that do not rupture may remain stable with a lower risk of rupture—depending on their size. These considerations lead to the conclusion that size may only be the main determinant for the risk of rupture in stable aneurysms. Currently, however, there is no test or morphological criteria to discern between the 2 groups of aneurysms. Due to the short interval, there is only a very little chance for diagnosing an unruptured aneurysm angiographically.

To provide better means to identify high-risk aneurysms or patients at a high risk for aneurysm formation, many factors such as sex, smoking habits, drug abuse, hypertension, aneurysm shape, and the emerging evidence for a genetic disposition [11], [17], [18] have to be considered as well.

5. Conclusion 

The 3D-RA data demonstrate that intracranial aneurysms may be smaller than previously thought. The majority of aneurysms had ruptured before reaching 10 mm (81.9%) or 7 mm (59%), and there was no size threshold beyond which the incidence of RIAs is increasing. The majority of RIAs less than 7 mm were located in the anterior circulation. The size distribution among ruptured and unruptured aneurysms is equal with only a 1.0-mm difference in height and 0.4 mm in width (not significant) between ruptured and unruptured aneurysms. The size category with the highest incidence of RIAs is only 4 to 6 mm in height and 2 to 4 mm in width. From these data and epidemiological considerations, we hypothesize that RIAs form acutely and rupture very shortly after their formation.