Clinical Scales in Aneurysm Rupture Prediction

Abstract

The rate of incidentally discovered unruptured intracranial aneurysms has increased with the broad availability of neuroimaging. The determination of the risk of rupture of brain aneurysms is challenging. Several clinical scales for aneurysm rupture prediction have been developed. The most common scales are PHASES, ELAPSS, and UIATS. These scales are not routinely used in clinical practice due to inherent shortcomings. In this review, we analyze the risk factors used in generating these scales and the performance of these scales in clinical studies. We also discuss new potential biomarkers and tools to predict aneurysm rupture.

Nonstandard Abbreviations and Acronyms

AComm

anterior communicating artery

aSAH

aneurysmal subarachnoid hemorrhage

AWE

aneurysmal wall enhancement

ISUIA

International Study of Unruptured Intracranial Aneurysms

RIA

ruptured intracranial aneurysm

SUAVe

Small Unruptured Aneurysm Verification Study

UIA

unruptured intracranial aneurysm

UCAS

Unruptured Cerebral Aneurysms Study of Japan

Clinical Perspective

• What Is New?

• This review provides a comprehensive analysis of the most common scales used in assessing risk of rupture of brain aneurysms. It also discussed potential new biomarkers and tools of aneurysm instability.

• What Are the Clinical Implications?

• A comprehensive scale to determine the risk of aneurysm rupture should include known demographic risk factors and aneurysm morphological metrics.

The wide availability of neuroimaging has increased the diagnosis of incidentally found unruptured intracranial aneurysms (UIAs). ¹ It is estimated that the prevalence of UIAs in the general population is about 3%. ² Treatment of UIAs can lead to significant complications. Surgical clipping can cause brain tissue injury, infections, and cerebrospinal fistula formation. ³ , ⁴ Furthermore, intraoperative rupture of the aneurysm may occur. ³ Although endovascular treatment of aneurysms is less invasive, it can be complicated by the perforation of the aneurysm or by thromboembolic events. ⁵ , ⁶ Therefore, it is important to assess the risk of aneurysm rupture and triage aneurysms for observation versus treatment. However, this assessment can be challenging because aneurysm characteristics vary considerably. While some aneurysms might remain stable, others might grow, change in morphology over time, and rupture. ⁷ To address this issue, systematic analyses of large population cohorts led to the development of scoring systems to assess the risk of aneurysm rupture and growth at follow‐up. ⁸ This study reviews the most frequently used scoring systems in aneurysm rupture and growth prediction. We analyze the advantages and limitations of the Practical Risk Score (PHASES), the Earlier Subarachnoid Hemorrhage, Location of the Aneurysm, Age >60 years, Population, Size of the Aneurysm, and Shape of the Aneurysm (ELAPSS) score, and the UIA Treatment Score (UIATS) consensus. Furthermore, we describe emerging imaging tools that may aid in risk stratification of UIAs beyond clinical scales.

Identification of Risk Factors of Aneurysmal Subarachnoid Hemorrhage

High‐Risk Demographics

Demographics play an important role in the stratification of aneurysm rupture. Factors, such as population, ⁹ hypertension, ¹⁰ biological sex, ⁹ age, ¹¹ smoking status, ¹² and history of aneurysmal subarachnoid hemorrhage (aSAH) from another aneurysm, ¹³ contribute to the risk of aneurysm rupture (Table 1).

Table 1.

Demographic Risk Factor for aSAH

Risk factor	Author	Number of aneurysms	Comments
Population	Vlak et al 2	1450 UIAs	Finnish and Japanese populations have a higher incidence of aSAH, which is not explained by a higher prevalence of UIAs
Hypertension	Karhunen et al 14	GWAS	SBP OR=2.17 and DBP OR=3.21 per 10 mmHg increase. P<0.001
	Sandvei et al 15	74 845 participants	SBP>130 and DBP>75 increased the risk for aneurysm rupture in a dose dependent manner (P<0.001)
	Korja et al 16	118 patients	SBP>140 was not a significant risk factor for aneurysm rupture (P>0.05)
Age	Juvela et al 17	181 UIAs	Average age of patients with an RIA was 38.5 years and with an UIA was 42.8 years. Annual rupture rate (%) was 1.5 for patients ≤30 years of age and decreased as age increased
	Sonobe et al 11	448 UIAs	For patients <50 years of age HR=5.23 (95% CI 1.03–26.52, P=0.046) for aneurysm rupture
	Korja et al 12	118 patients	Aneurysm risk of rupture OR=3.36 for patients <50 years of age compared to patients ≥50 years of age
Earlier SAH	Ishibashi et al 13	529 UIAs	History of aSAH was associated with aneurysm rupture HR=7.3 (95% CI 2.5–21.2)
Gender	Juvela et al 17	181 UIAs	The univariate HR for rupture was 1.6 (95% CI 0.78–3.29) for women compared to men
	Rooij et al 9	51 studies	The incidence for subarachnoid hemorrhage in women was 1.24 (1.09–1.42) times higher compared to men
Size	Sonobe et al 11	448 UIAs	Aneurysms that are ≥4mm in size have an HR=5.86 compared to those that are <4 mm (P=0.023)
	Morita et al 18	6697 UIAs	Aneurysm size ≥7 had a significant increase in risk for rupture compared to smaller aneurysms
	Wiebers et al 19	6221 UIAs	UIAs <7 mm in size in patients without history of aSAH had a low risk of rupture (0.1% per year)
	Waqas et al 20	12 609 RIAs	Mean RIA size was 7 mm (95% CI [6.2–7.4])
	Sanchez et al 21	13 025 RIA and UIAs	Mean RIA size was 6.1 (95% CI [5.7–6.6]) and mean UIA size was 4.9 (95% CI [4.5–5.3])
Location	Ishibashi et al 13	529 UIAs	Posterior circulation location was an independent predictor (HR=2.9) for aneurysm rupture
	Juvela et al 17	181 UIAs	AComm aneurysms had an HR=3.73 (95% CI [1.23–11.36]) for rupture compared to aneurysms in other locations (HR=1)
	Morita et al 18	6697 UIAs	AComm and PComm aneurysms were more likely to rupture than MCA aneurysms (HR=2.02 and 1.90, respectively)
	Wiebers et al 19	6221 UIAs	Aneurysms located at the tip of basilar and PComm were more likely to rupture compared to those located in the ICA (relative risk=2.3 and 2.1, respectively)
Irregular shape	Kleinloog et al 22	28 812 RIAs and UIAs	Irregular shape had strong evidence for increasing risk of rupture OR 4.8 (95% CI [2.7–8.7])
	Morita et al 18	6697 UIAs	Aneurysms with blebs were more likely to rupture HR=1.63 (95% CI [1.08–2.48])
	Wang et al 23	394 UIAs	Irregular aneurysm shape significantly increased the risk of rupture OR=31.5 [3.02–328.34]

AComm indicates anterior communicating artery; aSAH, aneurysmal subarachnoid hemorrhage; DBP, diastolic blood pressure; HR, hazard ratio; OR, odds ratio; PComm, posterior communicating artery; RIA, ruptured intracranial aneurysm; SBP, systolic blood pressure; and UIA, unruptured intracranial aneurysm.

Certain populations, such as those from Finland or Japan, may have an increased risk of aneurysm rupture. ⁹ This increased risk of aneurysm rupture is not explained by an increase in aneurysm prevalence. ² Population has been included in aneurysm assessment clinical scales, such as PHASES and ELAPSS. ¹⁴ , ¹⁵ However, the risk within a specific population may be biased due to a wide variety of demographic factors and aneurysm morphology variables. These scales have been critized for not including other large cohorts with possible inherent risks for aneurysm rupture.

Hypertension may increase the risk of aneurysm rupture. A large mendelian randomization study found that the risk of aneurysm rupture increases per each 10 mmHg increase in systolic blood pressure or diastolic blood pressure. ¹⁰ Sandvei et al (2009) performed a prospective analysis of 74 977 residents in Norway. The annual incidence of aSAH was 9.9 per 100 000 people. Compared with patients with systolic blood pressure <130, patients with a systolic blood pressure >130 had a hazard ratio (HR)=2.3 (95% CI [1.4–3.8]) of having aneurysm rupture, and those with an systolic blood pressure >170 mmHg had an HR=3.3 (95% CI [1.7–6.3]). ¹⁶ However, the definition and documentation of hypertension varies among studies, and fair comparisons among studies could not be performed without access to granular data.

Biological sex is another well‐described risk factor for aneurysm rupture. ¹² Most patients with ruptured aneurysms are female. ¹⁷ , ²⁴ Being female has also been linked to aneurysm growth and high‐risk morphology. A systematic review that included 51 studies observed that the incidence of subarachnoid hemorrhage was 1.24 (1.09–1.42) times higher in women. ⁹ The changes in hormone levels that happen throughout pregnancy and menopause may play a role in the endothelial disfunction that increases the risk of aneurysm formation and rupture. ²⁵

Currently, there is conflicting evidence about whether age influences the risk of aSAH. The SUAVe (Small Unruptured Aneurysm Verification) study found that patients younger than 50 years of age were at higher risk of rupture, ¹¹ whereas the pooled analysis that generated the PHASES aneurysm risk assessment score reported that patients older than 70 years are at higher risk of aneurysm rupture. ⁸ The UCAS Japan (Unruptured Cerebral Aneurysms Study of Japan) found no association between age and risk of rupture. ¹⁸

Smoking may also lead to an increased risk of aneurysm rupture. Exposure to the nicotine found in cigarettes promotes vessel wall inflammation through receptors in vascular smooth muscle. ²⁶ A prospective study of patients from Finland found that smoking was significantly associated with aneurysm rupture. ¹² Moreover, smoking might alter aneurysm morphology, leading to a higher risk of aneurysm rupture. Ho et al (2015) reported that patients who smoked had more aneurysms, and their aneurysms had a higher risk morphological profile, compared with patients who had never smoked. ²⁷ Also, the risk of aneurysm rupture may decrease in patients with UIAs who quit smoking. ²⁸

Aneurysm Size Does Not Always Matter

Aneurysm size was identified as a significant risk factor in various prospective and retrospective studies, including ISUIA (International Study of Unruptured Intracranial Aneurysms), ²⁹ UCAS Japan, ¹⁸ SUAVe, ¹¹ and others (Table 1). ¹² , ¹³ Aneurysm size is usually quantified as the maximal diameter visualized in the best 2‐dimensional projections. The ISUIA identified that aneurysms bigger than 7 mm may be considered high‐risk. ²⁹ The UCAS study in Japan also identified aneurysms ≥7 mm as having a higher risk of rupture, and risk increased for aneurysms with larger sizes. ¹⁸ The SUAVe study focused primarily on analyzing small aneurysms and concluded that aneurysms ≥4 mm were at a significant increased risk of rupture compared with aneurysms <4 mm. ¹¹ These studies included heterogeneous populations with different risk factors, which may explain why they report different thresholds of aneurysm size in determining an increased risk of rupture. Aneurysm measurements can also vary based on the imaging modality used for the assessment, the projection used to determine the measurements, and the identification of the aneurysm boundaries.

Aneurysms <7 mm accounted for 75%, ³⁰ 66%, ³¹ and 78% ³² of cases in different cohorts of ruptured aneurysms. A recent meta‐analysis (which included 13 025 aneurysms) performed by our group found that the mean size of ruptured aneurysms was <7 mm. ²¹ However, significant heterogeneity among studies was observed in determining aneurysm location and how aneurysm size was defined. It is also known that certain aneurysms may reach a large size and remain unruptured. Hence, it is unclear whether aneurysm size defines the risk of rupture. Moreover, aneurysm size is a limited morphological parameter, as it only accounts for one geometric measurement. Another morphological parameter that has been described in identifying aneurysms with a high risk of rupture is size ratio. Size ratio is defined as the ratio of the maximal aneurysm diameter divided by the average vessel diameter. ³³ Kashiwazaki and Kuroda (2013) found that size ratio was a marker for high risk of rupture in small (<5 mm) aneurysms. Their multivariate logistic regression analysis showed that size ratio was the only significant predictor of aneurysm rupture (P=0.008; odds ratio=9.1). ³⁴ Aneurysm aspect ratio is another alternative metric that is calculated by dividing aneurysm height by aneurysm neck width. ³⁵ An aspect ratio of >1.5 differentiated ruptured intracranial aneurysms (RIAs) from UIAs, even in small aneurysms. ²¹ Hence, other morphological characteristics, such as size ratio, shape, and aspect ratio, may also be evaluated in determining risk of rupture. These metrics are especially relevant in small aneurysms. A 7‐mm size threshold would miss most ruptured aneurysms in the anterior communicating artery (AComm) and posterior inferior cerebellar artery, as the mean size of RIAs in these locations is 5.5 and 4.2 mm, respectively. ²⁰ In the meta‐analysis performed by our group, the size of ruptured and unruptured aneurysms was not statistically different for aneurysms located in the anterior cerebral artery, AComm, and basilar artery. ²¹ In this study, the best predictor of rupture presentation among all locations was size ratio; the mean size ratio was 2.3 (2.1–2.5 range) for ruptured aneurysms and 1.6 for unruptured aneurysms (1.4–1.8 range).

Aneurysm Shape: Aneurysm Morphology Beyond Size

The shape of an aneurysm is a pivotal factor in determining the risk of rupture. Shape is an evolving morphological characteristic that reflects various factors, such as hemodynamic stress along the aneurysm wall, aneurysm wall thickness, and the distribution of tension among different aneurysmal compartments. ³⁶ , ³⁷ Shape is independent of size; aneurysms can reach large sizes and still have a regular contour. Conversely, some aneurysms may be irregular in shape despite having a small size. ³⁸ It has been observed that irregular aneurysms with multiple lobes and blebs have a higher risk of rupture. ³¹ A population‐based registry of 5814 saccular aneurysms showed that most RIAs (87%) that were <7 mm in size had an irregular shape. ³⁸ The authors concluded that irregular or multilobular shape is associated with rupture in saccular aneurysms regardless of size, aneurysm location, and the patient's background. These epidemiological studies have been confirmed with histological data showing that daughter sacs are thin sections of the vessel wall where rupture may occur. ³⁹ Most irregular aneurysms have daughter sacs. In the UCAS Japan prospective study, aneurysms with daughter sacs had a higher risk of rupture. ¹⁸ Furthermore, a meta‐analysis that analyzed 102 studies and 144 different risk factors reported that shape is an important predictor of risk of rupture and should be used in clinical practice and risk stratification. ²² Hence, shape is a key feature that can increase the risk of rupture regardless of aneurysmal size. Aneurysm shape should be included in predictive scales that aim at determining risk of rupture.

Aneurysm Location Matters

The location of the aneurysm was a high‐risk factor for rupture in the UCAS Japan and ISUIA trials. ¹⁸ , ²⁹ UCAS Japan showed that aneurysms located in the posterior communicating artery or AComm were at a higher risk of rupture when compared with aneurysms in other locations. ¹⁸ The ISUIA trial found that aneurysms located in basilar or posterior communicating artery locations had a higher risk of rupture compared with aneurysms located in the internal carotid artery. ²⁹ Conversely, aneurysms <12 mm located in the cavernous segment of the internal carotid artery had almost no risk of rupture (Table 1). ²⁹ A prospective study by Ishibashi et al (2009) found that posterior circulation aneurysms have a higher risk of rupture compared with anterior circulation aneurysms. ¹³ The SUAVe study found that aneurysms in the AComm have a high risk of rupture compared with those located in all other locations. ¹⁷ Carter et al (2006) described that the decreasing caliber of the wall thickness of the distal vessels may facilitate rupture, as less tension is required to weaken the parent vessel wall. ⁴⁰ Other studies have also found that most aneurysms located in the distal anterior cerebral artery can rupture despite being <5 mm. ⁴¹ Therefore, location is a key risk factor at the time of determining the risk of rupture. Most scales include location in determining risk of rupture.

Phases and Risk of Rupture Prediction

The PHASES scoring system is a clinical scale aimed at assessing the 5‐year‐risk of rupture of UIAs. The PHASES scoring system was generated from a pooled analysis of 8382 aneurysms. ⁸ Data were acquired from 6 prospective studies conducted in diverse geographical locations and focused on aneurysm risk factors that have been associated with risk of rupture. ¹¹ , ¹³ , ¹⁷ , ¹⁸ , ²⁹ , ⁴² These include the following data: population, presence of hypertension, age, aneurysm size, history of an earlier aSAH from another aneurysm, and aneurysm location (Table 2). ¹⁴ The ultimate result of the scoring system is a score that assesses the 5‐year risk of rupture of an UIA. The scoring system uses a scale in which the 5‐year risk of rupture increases with every point. Aneurysms with a score of ≤2 have a 5‐year risk of rupture of 0.4 (0.1–1.5) %, while aneurysms with a score of ≥12 points have a 5‐year risk of rupture of 17.8 (15.2–20.7) %. The variables that do not add points to the score are aneurysm size of <7 mm, age <70 years, being of North American or European descent (other than Finnish), absence of hypertension, no history of subarachnoid hemorrhage, and having the internal carotid artery as the location of the aneurysm. Conversely, an aneurysm size of ≥20 mm adds the greatest number of points to the score, adding 10 points by itself (Table 2).

Table 2.

Summary of PHASES and ELAPSS Scores

	PHASES score 14	ELAPSS score 15
Earlier aSAH
Yes	1	0
No	0	1
Location of aneurysm
ICA	0	0
ACA/AComm	4	0
MCA	2	3
PComm/posterior circulation	4	5
Age (y)*
≤60	NA	0
>60 (per 5 y)	NA	1
<70	0	NA
≥70	1	NA
Population
Reference 1	0	0
Japan	3	1
Finland	5	7
Aneurysm size (mm)
1.0–2.9	0	0
3.0–4.9	0	4
5.0–6.9	0	10
7.0–9.9	3	13
10.0–19.9	6	22
≥20.0	10	22
Aneurysm shape
Regular	NA	0
Irregular	NA	4
Hypertension
Yes	1	NA
No	0	NA

NA: some risk factors are not evaluated by one of the scales. ACA indicates anterior cerebral artery; AComm, anterior communicating artery; aSAH, aneurysmal subarachnoid hemorrhage; ELAPSS, Earlier Subarachnoid Hemorrhage, Location of the Aneurysm, Age >60 Years, Population, Size of the Aaneurysm, and Shape of the Aneurysm; ICA, internal cerebral artery; MCA, middle cerebral artery; PComm, posterior communicating artery; and PHASES, Practical Risk Score.

^* ELAPSS used: Canada, Netherlands, and China; PHASES used: North America and Europe (other than Finland).

One advantage of the PHASES scoring system is that it can be widely used. Most variables are part of the clinical history and are easy to collect. Moreover, the scoring does not require a complex assessment of aneurysm morphology and it can be calculated in the outpatient setting. However, the question of clinical validation emerged shortly after the PHASES score was published. Multiple studies have tried to validate the PHASES score in cohorts of RIAs. A significant number of ruptured aneurysms in these studies have a low PHASES score. Neyazi et al (2019) retrospectively studied a cohort of 100 RIAs and reported that 70% of patients had a PHASES score <7, which corresponds to an estimated 5‐year risk of rupture of <2%. Only 11% of patients had a score of ≥10 points, corresponding to a risk of rupture ≥5.3%. ⁴³ Pagiola et al (2020) studied 155 RIAs and observed that 71% had a PHASES score of ≤5, which represents a 5‐year risk of rupture of 1.3%. ⁴⁴ Similarly, Foreman et al (2018) studied 149 RIAs and determined that more than half of patients (61.7%) had a PHASES score of ≤5. ⁴⁵ Moreover, Rutledge et al (2020) retrospectively reviewed 628 RIAs and found that most aneurysms had a low risk of rupture, with a median PHASES score of 5 (interquartile range=4–6). ⁴⁶ Webb et al (2022) studied a cohort of 246 RIAs and calculated the annual risk of rupture of 0.32%±0.004%, according to the PHASES score. ³⁰ Bijlenga et al (2017) analyzed a cohort of 841 patients who were either treated or observed. Patients with a PHASES score of >3, which represents a 5‐year risk or rupture of 0.9% or more, were more likely to be treated. ⁴⁷ Hilditch et al (2021) found that, out of 700 retrospectively studied RIAs, 17% had a PHASES score of ≤3, corresponding to a <1% 5‐year risk of rupture. ⁴⁸ Hollands et al (2021) performed a multicenter study of 623 patients with 803 UIAs to assess whether management decisions for UIAs changed after implementation of the PHASES scoring system. Approximately 34% of patients were treated before the PHASES score was implemented, whereas 26% of patients were treated after implementation. Age had a relevant role, as this difference was most pronounced in patients <50 years of age. ⁴⁹

In these studies, most of the RIAs with a low‐risk PHASES score were <7 mm in size, suggesting that even small UIAs are at a high risk of rupturing. ⁴⁴ , ⁴⁵ , ⁴⁶ , ⁵⁰ Therefore, using an aneurysm size threshold of 7 mm to determine a higher risk of rupture might not be optimal. Additionally, in these studies, most RIAs were present in patients younger than 70 years of age, ³⁰ , ⁴³ , ⁴⁵ and these studies included very few patients from Finland (Table 3). Therefore, size, age, and population may be important confounders affecting the sensitivity and specificity of the PHASES score.

Table 3.

Assessment of PHASES from Retrospective Studies of Ruptured Intracranial Aneurysms

Author	Neyazi et al 43	Pagiola et al 44	* Bijlenga et al 47	Foreman et al 45	Webb et al 30
Number of patients	100	155	243	149	246
North American or European (%)	99.0	99.3	100.0	100.0	99
Hypertension (%)	59.0	52.9	34.0	60.4	53.8
≥70 years of age (%)	15.0	11.0	54.0 ± 13.0 †	11.4	19.1
Size <7 mm (%)	66.0	74.8	56.7	62.4	75.6
Earlier aSAH (%)	1	0	0	0	2
ICA (%)	10.0	7.74	10.0	6.0	15.9
MCA (%)	14.0	20.64	22.0	13.4	17.9
ACA/AComm/PComm/posterior circulation (%)	76.0	71.6	68.0	80.5	66.3
PHASES score ≤5 ‡ (%)	65.0%	70.9%	median=5 §	61.7%	82.0%≤6

ACA indicates anterior cerebral artery; AComm, anterior communicating artery; aSAH, aneurysmal subarachnoid hemorrhage; ICA, internal carotid artery; MCA, middle cerebral artery; PComm, posterior communicating artery; and PHASES, Practical Risk Score.

^* Numbers for the study by Bijlenga et al⁴⁵ are only reported for the ruptured aneurysms in the cohort.

^† Age of the cohort is expressed as mean ± SD.

^‡ PHASES score of ≤5 has a risk of rupture of 1.3% per year.

^§ The exact number of patients with a PHASES ≤5 is not reported.

The accuracy of the PHASES score in special scenarios, such as the presence of multiple intracranial aneurysms, has also been questioned. The prospective SUAVe study found that the presence of multiple aneurysms increases the risk of rupture. ¹¹ Another study reported that, among 40 patients with multiple aneurysms, the PHASES score estimated a low 5‐year rupture risk in a relatively large proportion of RIAs (≤1.7% for 82.5%; and ≤0.7% for 22.5%). ⁵¹

Limitations of the PHASES Score

The PHASES score has many limitations. (1) It evaluates 5‐year risk based on only one specific point in time. Aneurysms are dynamic structures that undergo different stages of remodeling. ⁵² Skodvin et al (2017) evaluated 29 UIAs before and after rupture. They observed that aneurysms were significantly larger immediately after rupturing, when compared with the size before rupture. Imaging showed that RIAs also exhibited more blebs than before rupturing. ⁵³ The extent of change depended upon how much time passed between the images taken before and just after rupture. ⁵³ (2) The only morphological parameter accounted for in the PHASES score is size. The PHASES score does not factor in aneurysm shape, which correlates with risk of rupture (Table 1). ²³ Aneurysms with irregular morphology may have a high risk of rupture despite being assigned a low PHASES score (Figure 1). (3) Patient age is a significant factor in determining the PHASES score. A study of 100 RIAs, in which approximately 70% of patients had a low PHASES score, determined that the low score was mostly driven by aneurysm size (<7 mm) and patient age (<75 years). ⁴³ This study suggests that an estimated low risk of rupture by the PHASES score in patients younger than 70 years old may not be fully accurate. (4) There is intrinsic bias in the prospective studies that generated the PHASES score, and this bias likely influenced the scoring system. For example, there is selection bias in most of these studies because patients who were enrolled had UIA that were not considered dangerous enough to require immediate treatment. ⁴⁵ Also, the studies included in PHASES were biased by selection of patients, because younger patients, smokers, and those with larger UIAs were more likely to be treated at baseline. Therefore, these patients were excluded from prospective natural history studies. Moreover, most of the cases in the PHASES study were provided by the ISUIA and UCAS Japan studies, and the follow‐up time among studies was highly variable. ⁴⁵ Another important source of bias in the PHASES study is that high‐risk populations (Finland and Japan) included in this study may compromise its external validity. ³⁰ Moreover, the predictive value of the PHASES score in other populations, such as in China, has not been determined. ²³ (5) The PHASES score does not include smoking status and biological sex as part of its assessment. Both factors have been linked to an increased risk of rupture (Table 1). ¹² Juvela (2021) modified the PHASES score to include cigarette smoking; their modified PHASES score that included smoking was more accurate at determining risk of rupture compared with the standard PHASES. ⁵⁴

Figure 1.

Digital subtraction angiograms showing examples of ruptured aneurysms with different PHASES. A and B, Some aneurysms may have a high PHASES score due to demographic factors, size, and/or high‐risk location. However, the PHASES scoring system does not account for aneurysm morphology. C, Irregular aneurysms with various lobes and daughter sacs may rupture despite a low PHASES score. This ACOM aneurysm has a PHASES score of 4 (5‐year risk of rupture of 0.9%), but it is very irregular and has several daughter sacs. The inlet shows the 3‐dimensional reconstruction of the aneurysms.

Elapss in Determining Aneurysm Growth

Aneurysms that are managed conservatively with frequent imaging are routinely assessed for evidence of growth. Approximately 15%–35% of UIAs display growth during the first 9 years of follow‐up, and up to 45% of aneurysms may grow over a 19‐year period. ⁵⁵ The increased risk of growth over time underlines the need for standardized and long‐term follow‐up assessments, especially in patients with an untreated UIA with life expectancy longer than 5–10 years. Villablanca et al (2013) determined that aneurysms that grow over time may have a 12‐fold higher risk of rupture. ⁷ Hence, the prediction of aneurysm growth may be as important as the assessment of risk of rupture.

ELAPSS is a scoring system that was generated to evaluate the risk of UIA growth. It was created from an international multicenter analysis of 1909 aneurysms from 10 international cohorts. The data included at least 6 months of follow‐up imaging, with growth as the primary outcome. The ELAPSS score incorporates earlier aSAH, aneurysm location, age, population, aneurysm size, and shape. ¹⁵ In addition to including variables similar to those used in the PHASES score, the ELAPSS score included aneurysm shape (Table 2). Irregular aneurysm shape was defined as the presence of blebs or daughter sacs, aneurysm wall protrusions, or multiple lobes. ¹⁵ Sex, hypertension, and aspect ratio >1.6 were excluded from the multivariate model of ELAPSS due to their limited predictive value in the scoring system. ¹⁵ Data on smoking status and family history of intracranial aneurysms were also not included in ELAPSS. ¹⁵ The ELAPSS scoring system adjudicates aneurysm risk of growth at 3 and 5 years. The scoring system utilizes a scale where values are grouped into 6 categories of risk. The lowest category includes scores of <5, corresponding to a 3‐year risk of growth of 5% (95% CI: 3.1%–7.4%) and a 5‐year risk of growth of 8.4% (95% CI: 6%–11.5 %). The highest category includes scores ≥25 and corresponds to a 3‐year risk of growth of 42.7% (95% CI: 33.5%–53.3%) and a 5‐year risk of growth of 60.8% (95% CI: 51%–70.5 %). The rest of the categories include scores between 5 and 25 (Table 4). Among all variables, the presence of a prior subarachnoid hemorrhage, an age of ≤60, an aneurysm size of 1–2.9 mm, a regular shape, and having descent other rather than Japanese or Finnish are the lowest weighted variables in the scale. None of these variables add points to the score. Conversely, size is one of the highest weighted variables in the scale; an aneurysm with a maximal size of >10 mm adds 22 points to the score.

Table 4.

Risk Stratified by Scores in the PHASES and ELAPSS Studies

Points	5‐year risk of rupture ‐ PHASES 14
≤2	0.4%
3	0.7%
4	0.9%
5	1.3%
6	1.7%
7	2.4%
8	3.2%
9	4.3%
10	5.3%
11	7.2%
≥12	17.8%

Points	3 (and 5) year risk of growth ‐ ELAPSS 15
<5	5% (8.4%)
5–9	7.8% (13%)
10–14	11.7% (19.3%)
15–19	17.5% (28.1%)
20–24	25.8% (39.9%)
≥25	42.7% (60.8%)

ELAPSS indicates Earlier Subarachnoid Hemorrhage, Location of the Aneurysm, Age >60 Years, Population, Size of the Aneurysm, and Shape of the Aneurysm; and PHASES, Practical Risk Score.

Although aneurysm rupture was not the primary outcome, UIAs ruptured during follow‐up in 18 patients. These patients had a median PHASES score of 8 (range 4–15), which corresponds to a median 5‐year rupture risk of 3.2% (range 0.9%–17.8%). Among these 18 patients, aneurysm growth before rupture was only observed in 8 aneurysms (44%). ¹⁵ Finally, an external validation of the ELAPSS scoring system in a cohort of 1452 aneurysms reported an accurate calibration for 3‐ and 5‐year risks of aneurysm growth. ⁵⁶

The ELAPSS Score Performance in the Real World

As in the PHASES study, the ELAPSS study found an increased risk of aSAH in patient populations from Finland and Japan, and an increased risk of rupture associated with larger aneurysm size. ¹⁵ A large baseline aneurysm size may be a pivotal predictor of aneurysm growth (Figure 2). In a study of 20 aneurysms that grew over a follow‐up period of 47 months, the risk of growth increased with larger sizes. Specifically, an aneurysm size of >8 mm was correlated with a significant increase in risk of growth. ⁵⁷ Most of the risk factors included in ELAPSS and PHASES scores are risk factors for both aneurysm growth and aneurysm rupture, suggesting that both processes share a common pathophysiology. Backes et al (2015) analyzed a cohort of 734 UIAs and observed that patients with a higher PHASES score had a higher hazards ratio for aneurysm growth per PHASES point (1.32, 95% CI [1.22–1.43). ⁵⁸

Figure 2.

ELAPSS score and aneurysm growth. A, A large left internal carotid artery (ICA) aneurysm with a high ELAPSS score at baseline. B, At 20‐month follow‐up, the volumetric reconstruction of this aneurysm shows enlargement. C, A smaller left ICA aneurysm (arrowhead) with a low ELAPSS score. D, At 7‐month follow‐up, the volumetric reconstruction of this aneurysm is virtually unchanged.

The implementation of ELAPSS in clinical practice has been challenging. A retrospective analysis of data from 245 RIAs found that almost half (45.5%) of ruptured aneurysms had a low‐to‐intermediate (≤13%–≤19% 5‐year‐risk) risk of growth. ⁵⁹ Expanded and modified versions of ELAPSS have been used by others to predict aneurysm growth. Juvela (2020) evaluated aneurysm growth in 87 patients with 111 UIAs by using ELAPSS. ⁶⁰ The area under the curve (AUC) for ELAPSS was weak (0.47, 95% CI [0.34–0.60]) in the receiver operator curve analysis. They recommended a new, simpler score consisting of only female sex, cigarette smoking, and age (<40 years) to predict long‐term aneurysm growth. ⁶⁰ Wang et al (2021) studied 491 UIAs and generated a purely morphological model that incorporated aspect ratio, irregular shape, and bifurcation location. This model had a slightly higher sensitivity for aneurysm growth compared with ELAPSS (AUC=0.76 versus 0.71, respectively). ⁶¹ Van der Kamp et al (2021) proposed a triple‐S risk prediction model, after an analysis of 312 patients with 329 aneurysms. ⁶² This triple‐S model included aneurysm size, shape, and site (location). In multivariable analysis, predictors of rupture were aneurysm size (≥7 mm; HR=3.1; 95% CI 1.4–7.2), irregular shape (HR=2.9; 95% CI 1.3–6.5), and site (middle cerebral artery: HR=3.6, 95% CI 0.8–16.3; anterior cerebral artery, posterior communicating artery, or posterior circulation: HR=2.8, 95% CI 0.6–13.0). ⁶²

Limitations of ELAPSS

Whether aneurysm growth is the best parameter to determine risk of rupture remains controversial. There is no consensus on how to determine aneurysm growth, and some imaging modalities are more accurate than others in documenting growth. ⁶³ Magnetic resonance angiography is the preferred imaging method for follow‐up surveillance. ⁶⁴ However, it might not capture the development of daughter sacs. ⁶³ Moreover, aneurysm growth might not be linear but a stochastic process. ⁶² Some aneurysms might rupture during an initial period of formation and rapid growth, ⁵² and other aneurysms might experience growth spurts. Once an aneurysm stops growing, it might stabilize again before experiencing another period of growth. ⁶² Van der Kamp et al (2021) reported that aneurysms can remain stable without rupturing for up to 12 years after the detection of growth. ⁶⁵ In light of these limitations, predictive scales that rely on aneurysm growth might not be accurate in determining risk of rupture.

The UIATS Consensus: The Decision to Pursue Treatment

UIATS is an international multidisciplinary consensus that used the Delphi methodology and incorporated key factors in determining the management of UIAs. UIATS is not a prognostic nor a predictive model for UIA rupture, but rather it reflects contemporary practice in UIA management and guidance for clinicians treating patients with UIA. UIATS is also the most comprehensive scale, incorporating data from 29 variables. ⁶⁶ The variables that generate UIATS originate from 3 different domains: patient, aneurysm, and treatment. The patient domain includes variables, such as age, medical history, clinical symptoms, life expectancy, and comorbidities. The aneurysm domain includes size, morphology, location, and other (growth, de novo formation, and contralateral steno‐occlusive disease). The treatment domain includes age‐related risk, aneurysm size‐related risk, aneurysm complexity‐related risk, and intervention‐related risk. Each variable has subcategories that are scored. The variables are divided into subcategories that favor treatment and subcategories that favor conservative management. Finally, the scores of the subcategories that favor treatment and the subcategories that favor conservative management are summed. A difference of 3 points between the treatment or conservative subcategories generates a recommendation to observe or to treat. If the difference is equal or lower than 2 points, either decision is acceptable and the result is classified as inconclusive (Supplementary Table S1). ⁶⁶ UIATS included the consensus of 69 multidisciplinary specialists. The goal of UIATS is to generate a score that will guide treatment decisions. ⁶⁶

UIATS and Treatment: A Challenging Decision

The applicability of UIATS has been questioned because it might not represent the treatment decisions made in actual clinical practice; various reasons may explain this. First, as in PHASES, UIATS is heavily weighted by aneurysm size. Molenberg et al (2021) followed 277 UIAs and observed that UIATS had a specificity of only 44% for identifying aneurysm growth or rupture; the mean aneurysm size in their study was 4.6±2.7. ⁶⁷ Second, in addition to aneurysm size, age might also be a confounding factor in UIATS. Rutledge et al (2021) reported that UIATS had a sensitivity of 12% for recommending treatment in patients older than 65 years of age, compared with 63% in patients younger than 65 years. ⁶⁸ The authors concluded that the UIATS score may lead to undertreatment of elderly patients. Older patients get more points that favor conservative treatment. Finally, UIATS may lead to an inconclusive score in cases when the pros and cons of treatment recommendation balance out. Hernández‐Durán et al (2021) evaluated UIATS in a retrospective cohort of 212 RIAs. In approximately 44% of cases, UIATS was inconclusive, in 25% of cases it recommended treatment, and in 32% of cases it recommended conservative management. ⁶⁹

The Puzzle of Risk Assessment

Aneurysm Size is Only Part of the Story

Additional metrics besides size have been described to evaluate aneurysm morphology. Aspect ratio is a surrogate for determining the degree of elongation of an aneurysm. ³⁵ Elongated aneurysms tend to have more irregularities because the shape of the aneurysm influences the direction and pattern of intrasaccular blood flow. ³⁶ Thus, aspect ratio does not discriminate among aneurysms of different sizes, as it represents a ratio of aneurysmal elongation. Aspect ratio can potentially identify aneurysms with high‐risk geometry despite having a small size. Another morphological metric that has been identified as a potential biomarker of increased risk of aneurysm rupture is size ratio. As described earlier, size ratio is calculated by dividing the aneurysm maximal height by the diameter of the parent vessel. ³³ Size ratio thresholds for determining high risk aneurysms vary between different vascular territories. ²¹ Size ratio has also been shown to be a good metric to identify small aneurysms that may pose a high risk of rupture. ³⁴ Additionally, novel 3‐dimensional volumetric parameters, such as ellipticity index, nonsphericity index, and undulation index, may provide a comprehensive evaluation of aneurysmal geometry. ⁷⁰ Aneurysms that exhibit large volumes and an aspect ratio greater than 1.4 are more likely to develop irregular shapes. ⁷¹ In addition to known demographic risk factors for UIA rupture, morphological metrics should be included in the creation of more accurate aneurysm rupture prediction models.

High‐Resolution Vessel Wall Imaging: The New Frontier

High‐resolution vessel wall imaging has emerged as a promising technique in identifying aneurysm's risk of rupture. ⁷² , ⁷³ , ⁷⁴ High‐resolution vessel wall imaging can be used to determine the thickness of the aneurysm wall and to detect aneurysm wall enhancement (AWE) after the administration of contrast gadolinium. Several observational data suggest that AWE may be the result of active inflammation within the aneurysm wall. ³⁹ Recently, sophisticated 3D analysis of AWE has provided a comprehensive map of the aneurysm wall. ⁷⁵ , ⁷⁶ AWE may distinguish between unstable and stable aneurysms, ⁷⁷ and the absence of AWE correlates with a small risk of rupture. ⁷⁸ Larger areas of AWE might be seen in growing and unstable aneurysms, compared with stable aneurysms that display uniform enhancement or almost no AWE. ⁷⁹ Moreover, AWE has been correlated with symptomatic presentation. ⁸⁰ Symptomatic presentation has been defined as aneurysm rupture, cranial nerve symptoms due to local compression of the aneurysm in the cranial nerve, sentinel headache, and/or documented aneurysm growth.

Retrospective studies have analyzed the possible relation of AWE with different prognostic scales. We analyzed 178 UIAs and determined that aneurysms with high AWE also had significantly higher PHASES (5.6±3.9 versus 4.4±2.6; P=0.04) and ELAPSS scores (19.4±8.9 versus 15.4±7.3; P=0.006) compared with nonenhancing aneurysms (Figure 3). ⁷² We then analyzed these same UIAs using UIATS and compared the presence of AWE with the treatment recommendation. Of the 46 UIAs with AWE, observation was recommended for 17 (37%), treatment was recommended for 21 (45%), and a nondefinitive recommendation was determined for 8 UIAs (17%). In this analysis, ELAPSS had the best correlation with AWE.

Figure 3.

PHASES, ELAPSS, and AWE. Aneurysms with high PHASES and ELAPSS scores may also have high AWE. A, An anterior communicating artery (AComm) aneurysm (arrowhead) with high PHASES and ELAPSS scores. B, High‐resolution vessel wall imaging (HR‐VWI) shows enhancement of the aneurysm wall (arrow). C, 3D‐AWE reconstruction shows an area of AWE in the aneurysm dome (arrow) that correlated with the enhancement seen on HR‐VWI. D, Conversely, a terminal internal carotid artery (ICA) aneurysm with low PHASES and ELAPSS scores (arrowhead). E, HR‐VWI shows the aneurysm with minimal AWE (arrow). F, 3D‐AWE reconstruction confirms the lack of AWE (arrow).

AWE can potentially identify high‐risk aneurysms. Size is an important factor in determining the risk of rupture in PHASES, the risk of growth in ELAPSS, and an indication for treatment in UIATS. Larger aneurysms tend to have more AWE, and increased AWE has been correlated with aneurysm growth at follow‐up. ⁷⁴ However, AWE also has the potential of identifying small aneurysm that have a high risk of rupture. This may be the case for AComm and posterior inferior cerebellar artery aneurysms.

Artificial Intelligence in the Analysis of UIAs

AI may be a promising tool for the clinical assessment of risk of rupture. A recent study by Zhu et al (2020) analyzed a cohort of 1897 aneurysms using machine learning to discriminate between stable and unstable aneurysms. The authors generated machine learning models to analyze the performance of the PHASES scoring system. The machine learning models achieved a higher AUC when discriminating stable and unstable aneurysms compared with PHASES (AUC=0.83 versus 0.59, respectively). ⁸¹ Artificial intelligence (AI) has also been used to evaluate morphological characteristics involved in aneurysmal instability. Liu et al (2019) analyzed 719 aneurysms by generating a machine learning model that utilized PyRadiomics‐derived morphological features. ⁸² The authors defined unstable aneurysms as aneurysms that rupture within 1 month, aneurysms that grew in follow‐up imaging, and aneurysms causing compression in adjacent structures. The model achieved an AUC of 0.85 (95% CI: 0.767–0.94) for detecting unstable aneurysms. ⁸²

AI volume‐based measurement tools may detect small changes in UIA that are not detected by manual measurements. Sahlein et al (2022) analyzed a cohort of 5 aneurysms that were being followed conservatively and ruptured. An AI‐based volumetric measurement determined volumetric aneurysm growth in 2 cases where manual radiological measures were not sensitive enough to detect aneurysm growth. ⁸³ This is especially relevant when deciding when to treat an UIA that is being followed conservatively.

AI may also be used as an adjuvant tool for clinicians in detecting UIAs. A meta‐analysis of 20 studies determined that AI increased the sensitivity (P<0.001) of aneurysm detection, without major changes in specificity (P=0.53). ⁸⁴ Another meta‐analysis determined that the overall sensitivity of AI in detecting intracranial aneurysms using MRI could reach values above 90%; however, the false‐positive rate was 16.5%. ⁸⁵

Although AI is a promising tool, there are various caveats that limit the wide use of AI in clinical practice. First, AI requires a large set for training and validation to be accurate and widely used. The heterogeneity in aneurysm morphology and patient characteristics makes this endeavor challenging. Second, the high false positive rate of AI algorithms may require an experienced clinician to discriminate between a normal structure and an UIA. Finally, AI is a relatively new tool that requires standardization.

Current Risk Assessment

Clinical scales are not routinely used in clinical practice and the determination of risk of rupture is largely subjective and not well documented. In our practice, the main factors in determining risk of rupture are age, location, size, presence of irregular morphology, and AWE (Figure 4). Age and life expectancy are weighed in at the time of triaging patients for treatment versus observation. A younger patient with a medium sized aneurysm in a high‐risk location is more likely to be treated than an older patient with multiple co‐morbidities. Aneurysms in high‐risk locations, such as AComm, posterior communicating artery, and top of the basilar artery, are more likely to rupture than aneurysms located in the para‐ophthalmic internal carotid artery, or in the cavernous segment. We almost never treat extra‐dural aneurysms in the cavernous segment, as the risk of rupture of these aneurysms is minimal. Size is also considered at the time of decision making. However, the threshold of size varies based on location. A 4 mm aneurysm located in the middle cerebral artery bifurcation has a lower risk of rupture than a 4 mm aneurysm located in the AComm. We also consider morphology, as it is well documented that aneurysms with irregular morphology are more likely to rupture than aneurysms with a regular shape. Finally, we routinely obtain 3‐tesla high‐resolution vessel wall imaging in the presurgical evaluation of aneurysms. Aneurysms that have high subjective enhancement after the administration of contrast are more likely to be treated than aneurysms without enhancement. AWE is also used routinely in patients with aSAH who have multiple aneurysms and an even distribution of blood in the subarachnoid space. In this scenario, the aneurysm that has more enhancement is believed to have ruptured and undergoes treatment.

Figure 4.

The risk assessment of aneurysm rupture is multifactorial. Clinical scales are used for risk stratification. PHASES is an assessment tool for aneurysm risk of rupture, ELAPSS is used to assess the risk of aneurysm growth, and UIATS is an expert consensus used to quantify the risk of treatment versus the risk of rupture if no treatment is performed. Some aneurysms may still pose a high risk of rupture or growth despite being classified as low risk based on PHASES, ELAPSS, and UIATS. A recent meta‐analysis found that aspect ratio (A) and size ratio (B) are significantly different between UIAs and RIAs. (C) The average size of RIAs in this study was 6.1 mm. ²¹ (D) The presence of multiple aneurysms in one patient (arrowheads) also increases the risk of rupture. An irregular shape has been extensively described as a high‐risk morphological feature. Aneurysms may have an irregular shape due to the presence of daughter sacs (arrowheads, E) or the presence of multiple lobes (arrowheads, F). Finally, the 3D analysis of AWE using high resolution vessel wall imaging is a promising emerging tool that identifies high‐risk areas within the aneurysm wall. G, Areas of increased AWE (arrowhead) are depicted in the 3D‐AWE heatmap of this para‐ophthalmic aneurysm.

Conclusion

Assessing the risk of rupture of UIAs is challenging. A comprehensive assessment should account for demographic information, patient‐specific information, aneurysm‐specific data, and potential risks of treatment. To date, no clinical scale for predicting aneurysm rupture has been validated in prospective studies and adopted as a standard of care. As more morphological information is acquired from high‐resolution vessel wall imaging studies, more risk stratification imaging‐based tools will be available for clinicians to assess the risk of rupture. The decision to observe or to treat an UIA should be supported  by all the available data, including advanced imaging in addition to clinical scales.

Sources of Funding

This work was conducted on an MRI instrument funded by 1S10RR028821‐01.

Disclosures

None.

Supporting information

Table S1.