Risk Analysis of Unruptured Intracranial Aneurysms
Prospective 10-Year Cohort Study
Background and Purpose—The natural history of unruptured intracranial aneurysms remains unclear, and management strategy is not well defined.
Methods—From January 2003 to December 2012, we enrolled patients with aneurysm in our institution. In total, 2252 patients with 2897 aneurysms were eligible for analysis, and 1960 eligible aneurysms were conservatively managed. Precise 3-dimensional evaluation was conducted using computed tomography angiography, digital subtraction angiography, or magnetic resonance angiography. We then assessed the risk of aneurysm rupture, mortality, and morbidity associated with aneurysm characteristics, demographics, and known health/lifestyle risk factors.
Results—The mean follow-up duration was 7388 aneurysm-years. During observation, 56 aneurysms ruptured, resulting in an overall rupture rate per year of 0.76% (95% confidence interval, 0.58–0.98). The mean initial visit to rupture interval was 547 days. Aneurysm size, location, daughter sac, and history of subarachnoid hemorrhage were significant independent predictors for aneurysm rupture. Aneurysms that were ≥5 mm were associated with a significantly increased risk of rupture when compared with 2- to 4-mm aneurysms (unadjusted hazard ratio, 12.24; 95% confidence interval, 7.15–20.93). Of 56 patients who experienced hemorrhage, 29 (52 %) died or were rendered severely disabled. Of the patients who had large or giant aneurysms, none recovered without deficits, and the mortality rate after rupture was 69%. For aneurysms sized <5 mm, the mortality rate was 18%.
Conclusions—Larger aneurysms are at greater risk for rupture and poor outcome. Ethnic factors may play a role in the risk of rupture.
The natural history of incidental unruptured intracranial aneurysms (UIAs) remains poorly understood,1–5 and whether they should be treated to prevent future rupture is controversial.6,7 The risks associated with treatment have to be balanced with the lifetime risk of rupture, the life expectancy of the patient, and the impact that the patient’s knowledge of the aneurysm has on their quality of life, as patients with aneurysm may experience considerable stress in their daily life because of anxiety over the possibility of rupture or the treatment risk.8
Although recent clinical outcomes after microsurgical clipping and endovascular treatment for incidental UIAs have improved,9 the benefits of treatment compared with a natural course remain unclear. A prospective randomized study of UIAs with a natural course versus endovascular treatment (the Trial on Endovascular Aneurysm Management [TEAM] study) was prematurely stopped by the funding agency because of the inability to recruit patients (only ≈70 patients enrolled >4 years).10 Attempts to obtain a more precise risk analysis of UIAs is an essential first step to evaluate the potential benefits of treatment.7,11 Therefore, the main objective of this study was to analyze the risk factors for eventual rupture of UIAs in a prospective cohort evaluated over a period of ≤10 years, and especially to determine the predictive value provided by accurate sizing of such aneurysms using modern 3-dimensional (3D) imaging technology.
From January 2003 to December 2012, newly diagnosed patients with UIA were enrolled. The prospective database was created, and the study was approved by our institutional review board.
We included all patients presenting an aneurysm that was at least 2 mm in size across its largest dimension. Patients were not enrolled if (1) they had fusiform, traumatic, or mycotic aneurysms or (2) an aneurysm treated before entry into the study; (3) the patient was <18 years old; (4) aneurysmal dilatation was difficult to distinguish from an infundibular dilatation; or (5) they had an internal carotid cavernous sinus aneurysm, which rarely cause subarachnoid hemorrhage (SAH) because of their anatomical location.
All patients were provided with information on general risk factors associated with aneurysm rupture in their particular case (size, location, multiple aneurysms, history of SAH, and family history).
For patients with aneurysms <5 mm, we recommended conservative management with observation. For patients with aneurysms between 5 and 10 mm, we carefully discussed the potential risks and benefits of treatment. For patients with aneurysms of >10 mm, we recommended treatment.
The final decision whether to continue observation or perform treatment was made by the patient. The patients were also informed that they could change their treatment strategy anytime from observation to treatment. Each patient’s initial visit to our institution was designated as day 0 for follow-up. Data collection for each aneurysm case ended when aneurysm rupture occurred, the patient died because of a cause other than rupture, or surgical/endovascular intervention was performed because of either significant aneurysm growth or the patient changing their mind and electing to be treated.
Imaging Technique and Follow-Up
Aneurysm size was mainly measured using either 3D computational tomography angiography or digital subtraction angiography. The risk of radiation and database collection was explained, and informed consent was obtained. The original digital imaging and communications in medicine (DICOM) images were transferred to a 3D workstation, and precise size measurements were made by experienced radiology technologists using a predefined measurement protocol.12 The largest dimension of the aneurysm (regardless of it being width or height) was recorded as the size of aneurysm. Patients were followed up by magnetic resonance angiography if they had a contraindication for the contrast agent or had multiple aneurysms and at least 1 had been treated by endovascular techniques. In these cases, an imaging study was conducted every 6 months, and clinical management included hypertension medication and smoking cessation advice.
Neurological findings were assessed by the modified Rankin Scale score of 30 days after rupture.
Clinical assessments were conducted by board certified physicians in our neurosurgery department. In cases of aneurysm rupture and when the patients were managed at other institutions, detailed information was obtained from local physicians for all endpoints.
Data extraction was performed on December 31, 2013, and the follow-up period was at least 1 year for all patients except in early rupture cases. The incidence of rupture was assessed per aneurysm rather than per patient. For the patients who underwent a surgical or endovascular intervention, data from the period up to the time of the intervention were not included in the analysis of the risk of rupture. Categorical variables were compared using the χ2 test. Continuous variables were compared among groups using the Mann–Whitney U test or the Student t test. In survival analyses, rupture-free rates were calculated by determining the time from the initial visit to rupture. Each patient was followed to the time of SAH, or death caused by causes other than SAH, or to the last possible follow-up contact. The hazard ratio (HR) with 95% confidence interval (95% CI) was computed using a Cox proportional-hazard model. For survival analyses, log-rank/Wilcoxon signed-rank tests were added. The annual incidence of SAH was calculated by determining the number of first-event SAHs divided by the number of aneurysm-years over follow-up. P<0.05 was considered statistically significant. All statistical analyses were performed using STATA 13.1 (STATA Corp, College Station, TX).
Management of Patients
As detailed in Figure 1, a total of 2665 patients with 3434 UIAs were referred to our institution. Table 1 shows baseline characteristics of the patients and aneurysms; 1556 patients with 1960 aneurysms were conservatively observed, and 937 aneurysms were repaired (793 by coiling and 144 by surgical clipping). Of these, 150 patients who originally chose observation changed their mind in favor of therapeutic intervention later on and received treated. The reasons were aneurysm growth in 24 patients and increased anxiety in 126 patients. There were 54 patients (69 aneurysms) who were followed up at other institutions for geographical reasons. However, they continued to also be followed at our institution by clinical consultation and telephone interviews to confirm that their aneurysms remained unruptured.
In 95% of patients, the aneurysm was either completely incidental or patients had only minor headache (group 1 according to International Study of Unruptured Intracranial Aneurysms [ISUIA] criteria). Eighty patients (3.6%) had a previous SAH and a coexisting aneurysm along with the UIA (group 2). Those patients who presented with mass effect or cranial nerve palsy were urgently treated. In the observed patients, there were 1905 (97.2%) group 1 and 55 (2.8%) group 2 aneurysms.
Characteristics of Aneurysm With Conservative Observation
The mean follow-up duration was 7388 aneurysm-years. Age ranged from 18 to 86 years with a mean (±SD) age of 66.0±11.6 years. For monitoring by imaging, 1757 aneurysms were followed by computational tomography angiography, 187 by magnetic resonance angiography, and 16 by digital subtraction angiography.
In total, 68.1% of the patients were women, and 401 aneurysms (20.5%) were located at the internal carotid-posterior communicating artery origin, 525 (26.8%) were internal carotid artery aneurysms not located at the posterior communicating artery origin, 330 (16.8%) were on the anterior cerebral artery, 535 (27.3%) were on the middle cerebral artery, and 169 (8.6%) were vertebral artery-basilar artery aneurysms. In current smokers, 17.6% (38/216 patients) quitted smoking during observation.
Incidence and Predictive Factors of Aneurysm Rupture
Fifty-six aneurysms ruptured during the follow-up period. The overall annual incidence of SAH was 0.76% (95% CI, 0.58–0.98). Figure 2A shows the Cox proportional-hazard curves for the probability of aneurysm rupture. Mean duration to rupture from initial consultation was 547 days; 25%, 50%, and 75% of aneurysms ruptured at 89, 316, and 792 days after the initial visit, respectively. Thereafter, the frequency of rupture declined, but the risk continued even after 2 years.
Risk factors associating with aneurysm rupture were analyzed under single Cox proportional-hazard model (Table 2). Aneurysm size, the specific location, presence of a daughter sac, and history of SAH were significant independent risk factors for aneurysm rupture.
These 4 factors remain significant even under multivariate Cox proportional-hazard model (Table I in the online-only Data Supplement). On the contrary, age, female sex, hypertension, diabetes mellitus, hyperlipidemia, smoking history (current and past), and family history of SAH were not associated with rupture.
Then, we evaluated aneurysm size as a risk factor of rupture in detail. The average size of ruptured aneurysms was 7.5±5.74 mm; 39 ruptures (69.6% of total rupture) occurred in aneurysms <7 mm. The annual rupture rate by aneurysm size was 0.33% (95% CI, 0.22–0.51) for <5 mm (2–4 mm), 3.1% (95% CI, 1.91–4.93) for 5 to 6 mm, 2.9% (95% CI, 1.09–7.72) for 7 to 9 mm, 10.2% (95% CI, 5.67–18.49) for 10 to 24 mm, and 33.1% (95% CI, 8.27–132.17) for ≥25 mm, respectively. Cox proportional-hazard curves were compared across sizes (Figure 2B). The probability of rupture increased with size (size 2–4 mm, reference; 5–6 mm: HR, 8.9; 95% CI, 4.70–16.68; 7–9 mm: HR, 9.1; 95% CI, 3.13–26.33; 10–24 mm: HR, 27.96; 95% CI, 13.54–57.73; ≥25 mm: HR, 90.73; 95% CI, 21.5–387.29). Next, we have set cutoff point of size at 5, 6, 7, 8, 9, and 10 mm and compared between <5 and ≤5 mm, and so on, using Cox proportional-hazards model. Both unadjusted HR and adjusted HR with the presence of a daughter sac, location of aneurysm, and history of SAH were computed (Table II in the online-only Data Supplement). As results, statistical significant differences were observed in all cutoff point of the aneurysm size with or without multivariate adjustment, not only at 7 mm but also at 5 mm.
When compared with other locations, vertebrobasilar artery aneurysms demonstrated a significantly higher risk of rupture (HR, 8.64; 95% CI, 2.75–27.14; P<0.001). Internal carotid-posterior communicating artery aneurysms were also associated with higher risk (HR, 5.19; 95% CI, 1.74–15.33; P=0.003). Middle cerebral artery (HR, 3.89; 95% CI, 1.30–11.63; P=0.015) and anterior cerebral artery (HR, 3.54; 95% CI, 1.09–11.63; P=0.035) aneurysms were associated with a modest risk of rupture, whereas non–posterior communicating internal carotid artery aneurysms showed the lowest rupture risk in this study. Cox proportional-hazard curves were compared for different aneurysm locations (Figure 2C). Aneurysms in vertebrobasilar artery or internal carotid-posterior communicating artery had significantly higher HRs than others.
Table 3 shows 5-year cumulative hemorrhage rates by aneurysm location, size, and each group. Group 2 showed higher rupture rate in small size (2–6 mm). Table 4 shows more detailed size and location rupture rate.
Of the ruptured aneurysms in 56 patients, 15 (26.8%) resulted in death and 16 (28.6%) in moderate-severe disabled status (modified Rankin Scale score of 3–5). Only 46.4% patients returned to normal life. Of the patients who had large or giant aneurysms, none recovered without deficits and the mortality rate after rupture was 69% (9/13). In aneurysms <5 mm, the mortality rate was 18% (4/22) in rupture cases. Twenty-five patients died for reasons other than aneurysmal rupture.
In this study, we evaluated the risk of eventual rupture of UIAs based on their size, location, and a spectrum of demographic and health risk factors. Although larger aneurysms are associated with higher risk, the size at which treatment should be recommended remains unclear. For the patients in this study, we recommended observation in small aneurysms (<5 mm) based on previous natural history studies and national guidelines.1,5,13 Although 22 patients with aneurysms that were <5 mm had a rupture, but the incidence of rupture was still low (0.3% per year).
In clinical practice, it is always difficult to decide between a recommendation of treatment and observation for aneurysm sizes between >5 and <7 mm. This decision has to be determined by balancing the risks of a natural course, the life expectancy of the patient versus the risks of treatment. The largest nationwide cohort study, the Unruptured Cerebral Aneurysm Study (UCAS) Japan,4 reported that aneurysms with a size of 5 to 6 mm were not associated with an increased risk of rupture when compared with 3- to 4-mm aneurysms.
By contrast, in our series, aneurysms that were 5 to 6 mm or larger were associated with a significantly increased risk of rupture when compared with those that were 2 to 4 mm. In 5- to 6-mm size range, rupture rate was 3.1% per year. Although therapeutic value for aneurysms ≥5 mm is unclear, this rupture rate was higher than in other reported large cohorts.
In a study by Greving et al11 that introduced PHASES (population, hypertension, age, size of aneurysm, earlier SAH from another aneurysm, and site of aneurysm) score to predict rupture based on pooled data analysis, age, hypertension, history of SAH, aneurysm size, aneurysm location, and geographical region were independent predictors of aneurysm rupture. By comparison with populations from North America and European countries other than Finland, Finnish people had a 3.6× increased risk of aneurysm rupture and Japanese people had a 2.8× increased risk.
Our results indicated that only size, location, daughter sac, and history of SAH were independent predictors for rupture.
Multiple aneurysms, family history, smoking, and hypertension did not emerge as predictors in this study. It is known that current smoking is risk factor for rupture. Our study and UCAS Japan did not show this as risk factor. In ISUIA, 41% observed patients were current smoker and only 13.4% were current Japanese cohort. In addition, the frequency of current smoker was low in our cohort when compared with general Japanese population (19.8%). Ethnic factors or the frequency of smoking in the population may be reason of this paradoxical result. However, despite this result, the importance of smoking cessation cannot be denied based on this result.
The main limitation of this study is that as a single-center cohort study, there is inevitably some selection and referral bias and our findings are entirely based on a Japanese population, and it is known that the incidence of aneurysmal SAH is higher in the Japanese and Finnish populations than in the general North American and European population and unlike the ISUIA study, few patients with a history of SAH (group 2) were included. This result may be only useful for incidental UIAs. One third of aneurysms discovered in this study received surgical or endovascular treatment, and among these, 150 patients who originally chose observation subsequently requested aneurysm treatment. Irregular-shaped aneurysms were also aggressively treated, probably leading to some bias in the observed population. Eliminating selection bias is simply not feasible in the modern medical environment, and ethical considerations preclude a purely observational study. Thus, the true natural history of incidental aneurysm rupture may be higher than reported in our analysis; however, it is unlikely that this dilemma can ever be fully resolved.
Size, history of previous SAH, daughter sac, vertebrobasilar artery, and internal carotid-posterior communicating artery location were all significant predictive risk factors for aneurysm rupture. We observed that medium size (5–6 mm)/single aneurysms were at modest risk for rupture in this study of a Japanese population, complicating the optimal treatment recommendations for small UIAs. We cannot rule out that ethnic factors may play a role in the risk of rupture.
We thank Naoya Kunigane, RT, and Hiroto Narita, RT, PhD, for 3-dimensional computational tomography angiography measurement. We also thank Ms Rie Sakima for clinical data preparation.
Dr Murayama reports grants from Stryker, Siemens, NTT docomo, and personal fees from Stryker, and Asahi Intecc, outside the submitted work. Dr Ishibashi reports grants Stryker, Siemens, NTT docomo and personal fees from Stryker, outside the submitted work. Dr Takao reports grants from Stryker, Siemens, NTT docomo, outside the submitted work. Dr Molyneux is a Consultant to Sequent Medical Inc. He provides Clinical Adjudication and Clinical study advice. He provides Expert Witness evidence in Court Proceedings in respect of subarachnoid hemorrhage and cerebral aneurysms.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.115.010698/-/DC1.
- Received July 15, 2015.
- Revision received December 3, 2015.
- Accepted December 8, 2015.
- © 2016 American Heart Association, Inc.
- Ishibashi T,
- Murayama Y,
- Urashima M,
- Saguchi T,
- Ebara M,
- Arakawa H,
- et al
- Juvela S,
- Poussa K,
- Lehto H,
- Porras M.
- Wiebers DO,
- Whisnant JP,
- Huston J III,
- Meissner I,
- Brown RD Jr,
- Piepgras DG,
- et al
- Etminan N,
- Beseoglu K,
- Barrow DL,
- Bederson J,
- Brown RD Jr,
- Connolly ES Jr,
- et al
- Ishibashi T,
- Murayama Y,
- Saguchi T,
- Ebara M,
- Arakawa H,
- Irie K,
- et al
- McDonald JS,
- McDonald RJ,
- Fan J,
- Kallmes DF,
- Lanzino G,
- Cloft HJ.
- Greving JP,
- Wermer MJ,
- Brown RD Jr,
- Morita A,
- Juvela S,
- Yonekura M,
- et al
- Takao H,
- Murayama Y,
- Ishibashi T,
- Saguchi T,
- Ebara M,
- Arakawa H,
- et al
- 13.↵Japanese Society for Detection of Asymptomatic Brain Diseases. Guidelines for UIA. http://www.jbds.jp/guideline.html. Accessed May 4, 2012.