Cerebral Aneurysm Morphology Before and After Rupture
Nationwide Case Series of 29 Aneurysms
This article has a correction. Please see:
Background and Purpose—Using postrupture morphology to predict rupture risk of an intracranial aneurysm may be inaccurate because of possible morphological changes at or around the time of rupture. The present study aims at comparing morphology from angiograms obtained prior to and just after rupture and to evaluate whether postrupture morphology is an adequate surrogate for rupture risk.
Methods—Case series of 29 aneurysms from a nationwide retrospective data collection. Two neuroradiologists who were blinded to pre- versus postrupture images assessed predefined morphological parameters independently and reached consensus regarding all measurements. Prerupture morphology and respective changes after rupture were quantified and linked to risk factors and to the risk of rupture according to the PHASES (population, hypertension, age, size of aneurysm, earlier subarachnoid hemorrhage from another aneurysm, site of aneurysm) and unruptured intracranial aneurysm treatment (UIAT) scores.
Results—All 1-dimensional parameter medians were significantly larger after rupture, except neck diameter. Number of aneurysms with daughter sacs was 9 (31%) before and 17 (59%) after rupture (P=0.005). Aneurysm growth from the images prior to and just after rupture increased with the time elapsed between images. Aneurysms in patients with hypertension were significantly larger at diagnosis. Prerupture morphology did not differ in relation to smoke status. Clinical risk factors were not significantly associated with morphological change.
Conclusions—The changes in aneurysm morphology observed after rupture reflect the compound effect of time with successive growth and formation of irregularities and the impact of rupture per se. Postrupture morphology should not be considered an adequate surrogate for the prerupture morphology in the evaluation of rupture risk.
Subarachnoid hemorrhage (SAH) causes loss of potential life years at a proportion similar to that of ischemic stroke and intracerebral hemorrhage.1,2 The prevalence of intracranial aneurysms (IAs) is estimated to be around 2% to 3.5% in a normal population.3,4 Ruptured aneurysms are the source of 80% of SAH. The SAH incidence is 10 per 100 000 person-years, implying that many IAs never rupture.5 The optimal management of a diagnosed, unruptured IA remains controversial, and the risks of preventive intervention have to be weighed against the unknown risk of rupture for the individual aneurysm.6 Therefore, tools have been developed to establish the risk of IA rupture (such as the PHASES [population, hypertension, age, size of aneurysm, earlier SAH from another aneurysm, site of aneurysm] risk factor score7) and to ease the decision-making in the management of unruptured IAs (such as unruptured IA treatment score [UIATS]8).
Aneurysm size is often applied in the clinical decision-making because it has been shown to be a significant predictor of rupture.7 On the contrary, rupture is a multifactorial end point and cannot be exactly determined by aneurysm size alone.9 Hence, larger aneurysms carry a higher risk of rupture, but, nevertheless, a large amount of SAHs are caused by small aneurysms.10–12 Aneurysm morphology expressed as aspect ratio and bottleneck factor determines the hemodynamics that may affect the risk of rupture. Several studies showed significant morphological differences between ruptured and unruptured aneurysms.9 However, applying results from these studies on unruptured aneurysms to determine their risk of rupture critically relies on the premise that postrupture morphology is not significantly different from that prior to rupture.9 Recent case reports and small studies with 1 to 13 aneurysms indicate that this assumption does not hold.13–16
The aim of the present study is to describe the changes in morphology and morphological indices occurring between images obtained prior to and just after rupture of an aneurysm and to evaluate whether postrupture morphology is an adequate surrogate for risk of rupture.
Patients and Methods
Study Design, Patient Selection, and Data Extraction
The study is a nationwide retrospective data collection from the 4 neurosurgical centers treating all IA and SAH in Norway. We searched electronic health records using codes from the International Classification of Diseases, version 10, to identify patients diagnosed with unruptured IA who later were hospitalized with SAH. We created lists of patients diagnosed with I67.1 (cerebral aneurysm, unruptured) between October 1, 2003, and October 1, 2013. Of these patients, we identified those being subsequently hospitalized with I60.0–I60.9 (nontraumatic SAH).
For the identified patients, we recorded age, sex, date of diagnosis of unruptured IA, date admitted for SAH, and the reason for the aneurysm being conservatively managed. We also retrieved known risk factors, such as hypertension, smoking, connective tissue disease, polycystic kidney disease, family history, and prior aneurysmal SAH. We determined the PHASES7 and UIAT8 scores in all patients. The latest available prerupture and the first available postrupture angiograms were retrieved, from which several quantitative and qualitative features were determined. Patients were excluded if they had previous treatment of the aneurysm of interest, multiple aneurysms of which the ruptured aneurysm was difficult to identify, or image quality precluding reliable aneurysm measurements. Fusiform aneurysms were excluded.
The study was approved by the Northern Norway Regional Committee for Medical Research Ethics, which decided the study to be exempt from patient consent. The study is reported according to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement for observational studies.17
Two neuroradiologists residing in different centers assessed the aneurysms using measuring tools in Siemens syngo.via and syngo InSpace (Siemens Healthcare, Erlangen, Germany). The 2 observers measured all aneurysms independently according to a strictly defined measurement protocol, blinded to each other’s results and with no prior information about aneurysm rupture state.
Initially, the observers interactively evaluated 3-dimensional volume rendering technique images for general morphology, such as smooth/irregular and numbers of daughter sacs (Figure 1A). The aneurysm neck was identified and multiplanar cursors were aligned to define the aneurysm neck. The aneurysm was rotated until the maximum length and diameters were revealed. The resulting volume rendering technique projection was then converted to a thin-slice maximum-intensity picture on which measurements were performed (Figure 1B).
We calculated intraclass correlation coefficient to assess the absolute interrater agreement.18 According to conservative criteria, values >0.81 represent substantial reliability.19 Mean intraclass correlation coefficient for all parameters except 2 was 0.88. Of the 2 with lower intraclass correlation coefficient, 1 parameter (minimal size) was excluded from further analyses, whereas the other (Neck) was redefined to increase precision. The final measurement guide is described below (see Parameter Definitions).
The mean values between observers were chosen when interrater difference was <2 mm. In cases of ≥2 mm differences and for the redefined Neck parameter, values were settled by consensus.
Definitions are illustrated in Figure 1C and 1D. All parameters must be fitted within the aneurysm sac. Maximal size is the maximal distance between any 2 points in the aneurysm sac, including the neck plane. Neck size is the largest observed diameter of the neck plane. Height is the orthogonal distance between the neck plane center and the aneurysm dome. Length is the greatest distance between the neck plane center and any point on the aneurysm dome, not necessarily orthogonal to the neck plane. Width L is the largest diameter that is orthogonal to length. Width H is the largest diameter that is orthogonal to height. When comparing digital subtraction angiography with another image modality, the available digital subtraction angiography projections dictated which projections were used from the other modality.
Aspect ratio was calculated as the ratio between height and neck diameter, and the bottleneck factor was calculated as the ratio between width L and neck diameter.20
Three of the above measured diameters were used to mathematically approximate the aneurysm volume. The conventional volume formula is V=4/3×π(A/2)×(B/2)×(C/2). We replaced A, B, and C with maximal size, length, and width L, respectively. This approximation technique has been shown to underestimate but still correlate with other methods of volume measurements.21
We registered aneurysm location, relation to parent artery (bifurcation or sidewall aneurysm), surface quality (smooth or irregular), and the presence and number of daughter sacs protruding from the aneurysm wall.
The data were analyzed with Stata for Mac (version 14; StataCorp LP, TX) and SPSS for Windows (version 24; IBM, NY). The variables’ distribution was investigated visually with Q-Q plots and numerically with Shapiro–Wilk test and found to be nonparametric. Thus, paired continuous variables were compared using Wilcoxon signed-rank test, and paired nominal variables were compared using McNemar’s test. Independent continuous variables were compared using Mann–Whitney U test or Kruskal–Wallis test in cases of >2 groups. Categorical variables were compared using Chi-squared test. A P value of <0.05 was assumed statistically significant.
The search identified 52 patients with confirmed aneurysmal SAH, originating from aneurysms that were recognized prior to rupture but not repaired. Of these, 23 were excluded (9 were fusiform and 14 because of missing or poor images). The remaining 29 patients were included in the study.
Eight of the 29 patients were men (28%). Mean age at time of SAH was 67 years (SD, 9.3). Median time span between imaging prior to and just after rupture was 12 months (range, 0.3–96 months). The combination of image modalities before and after rupture was computed tomography angiography (CTA) and CTA for 16 patients (56%), CTA and magnetic resonance angiography for 8 patients (28%), CTA and 2D digital subtraction angiography for 2 patients (7%), and magnetic resonance angiography and 2D digital subtraction angiography for 3 patients (10%).
Table 1 summarizes patient and aneurysm characteristics at baseline, as well as risk of rupture expressed by PHASES risk score and treatment recommendations according to the UIAT score. The 5-year risk of rupture exceeded 1% (PHASES ≥5) in 79% of the patients and 5% (PHASES ≥10) in 38% of the patients. The UIAT score was indeterminate in 35% of the cases and favored conservative management in 31% of the patients. The UIAT score favoring aneurysm repair tended to be higher for aneurysms that ruptured within 3 months (median, 14; range, 5–16) compared with aneurysms with longer time between images (median, 10; range 5–20; P=0.07). Table in the online-only Data Supplement provides aneurysm location, maximal aneurysm size, PHASES and UIAT scores, and the reason for not performing aneurysm repair for each individual aneurysm.
Table 2 lists the morphological parameters from imaging prior to and just after rupture. All 1-dimensional parameter medians were statistically significantly larger after rupture, except neck diameter. Median aspect ratio before rupture was 1.5 (range, 0.8–4.0) compared with 1.9 (range, 0.8–6.7) after rupture (P=0.008). Median bottleneck factor was 1.5 (range, 0.9–4.0) before and 1.5 (range, 0.7–6.2) after rupture (P=0.068). Number of aneurysms with ≥1 daughter sac was 9 (31%) before and 17 (59%) after rupture (P=0.005). Figure 2 illustrates a typical change from pre- to postrupture image.
The magnitude of change was clearly dependent on the time elapsed between the image prior to and just after rupture, visualized in Figure 3. Seven aneurysms (24%) ruptured within 3 months after the last image. Though median change in aspect ratio for this subset was only 0.10, the range from −0.45 to 0.95 (corresponding to −18% to 98%) demonstrates that we also observed relatively large morphological change within short time spans. We observed a new daughter sac in 1 of these 7 aneurysms after rupture (number of days between images for this particular aneurysm was 18). Change in the morphological indices, however, was not dependent on time elapsed between images.
Morphology and Change in Relation to Clinical Risk Factors
In hypertensive patients, aneurysms were significantly larger at diagnosis (median maximum size, 12.85 mm [range, 3.4–12.3] in hypertensive versus 6.95 mm [2.8–33.4] in nonhypertensive; P=0.041). Indices and presence of daughter sacs were not significantly different. Changes in morphology from before to after rupture were not significant, except for neck diameter, which tended to increase in hypertensive patients (0.3 mm [−1.8 to 3.7]) and decrease (−0.4 mm [−2.9 to 0.5]) in nonhypertensive patients (P=0.047).
Between current and former/never smokers, there was neither a significant difference in morphology prior to rupture nor a significant change in morphology after rupture. These findings were also true when excluding aneurysms that ruptured after a short observation period (<3 months).
The core finding of the present study is that aneurysm morphology had changed between imaging prior to and just after aneurysm rupture. The observed changes increased with the time elapsed between imaging, though gross changes also occurred within short time spans. Change occurred in a nonuniform manner, signified by changes in aspect ratio and fraction of aneurysms with daughter sacs.
Postrupture Morphology as Surrogate for Prerupture Morphology
The present findings do not support the assumption that postrupture morphology is representative for the prerupture morphology.9 The fraction of aneurysms with blebs increased from 31% before to 59% after rupture. Consistent with the present study, a recent literature review reported that 17 of 23 aneurysms increased in size around time of rupture, and a case series in the same work showed presence of new daughter sacs after rupture in 5 out of 6 patients.16 However, aneurysm morphology just after rupture will be subjected to the impact of the rupture per se plus any change that may have occurred along the evolution of the specific aneurysm or even in the short time span between rupture and postrupture imaging. In a meta-analysis including 4972 unruptured aneurysms, 9% of aneurysms enlarged within a mean follow-up time of 2.8 patient-years.22 Accordingly, our data do not reveal what occurs during the exact moment of rupture, but rather support the notion that aneurysms grow over time, with periods with and without growth, and an inconstant risk of rupture over time.23–25
The low rate of rupture of small aneurysms in the International Study of Intracranial Aneurysms led some authors to speculate that aneurysms shrink after rupture.14,26 Aneurysms in the present study that ruptured within 3 months after the prerupture image showed less change in morphology (or even shrunk in some of the parameters) than those that had ruptured after longer time intervals. One could assume that the changes in this subgroup were more subjected to the effect of the rupture per se than those we observed in the other aneurysms. Though the changes are too small to rule out measurement uncertainty, one can speculate that rupture may cause a slight deflation of aneurysms. Three of the aneurysms ruptured after 9, 16, and 22 months after the last prerupture scan, respectively, and also showed a decrease in maximum size; one could speculate that these were stable aneurysms without growth during the time span and that the observed change was caused by the rupture. However, the number of aneurysms is too small to allow for conclusions.
Risk of Rupture
An aspect ratio >1.6 has been considered as a predictor for future rupture and is also included in the UIAT score.8 We are not aware of suggested cutoff values for bottleneck factor. Although such thresholds have been criticized and are affected by measurement methodology,20 we note with interest that 15 (52%) of the 29 aneurysms in our material fell below these limits before rupture, whereas 10 (35%) did so after rupture. Because the mean values increased for all parameters except neck diameter, any change would tend to increase aspect ratio and bottleneck factor. Thus, the higher aspect ratio and bottleneck factor seen in ruptured aneurysms in other studies may simply be the effect of change over time or the rupture itself.27
Maximal aneurysm diameter is perhaps the most common denominator for determining the risk of rupture and is incorporated into the PHASES7 and UIAT8 scores. The fact that even small aneurysms rupture is well established.28,29 Close to one third (28%) of our aneurysms were <7 mm and still ruptured. The size of our aneurysms did not predict the time span to rupture. One reason may be an overrepresentation of small aneurysms within the population of unruptured IAs and another may be that aneurysms grow in a nonlinear fashion.23,24
Size is only 1 indicator of the multifactorial causes for aneurysm rupture. PHASES and UIAT scores incorporate a wide array of factors supposedly influencing the risk of aneurysm rupture; still, in merely about one third of our cases, the UIAT score would have favored aneurysm repair. In 20% of cases, the PHASES score indicated a 5-year cumulative rupture risk of <1%. On the contrary, the UIAT score favored repair or suggested special consideration because of indetermination regarding treatment in 70% of patients. According to PHASES, 38% of the patients exceeded a 5-year risk of rupture of 5%. Though our retrospective study aims to compare pre- and postrupture imaging, we note that the majority of the included patients may have required treatment.
Aneurysm growth is a strong risk factor for rupture,22,24,30–32 possibly increasing risk 12- to 24-fold.30,31 Growth rate and risk of growth increases with increased aneurysm size. However, growth can occur at all aneurysm sizes, warranting follow-up imaging of conservatively managed aneurysms, including aneurysms <7 mm.24,25,27,29–31 In a systematic review of 30 unruptured aneurysms <7 mm followed with serial imaging for a median of 6.5 years, 27 (90%) enlarged before rupture.24 Thus, assuming that at least substantial parts of the changes we observe in our study are prerupture changes, our study sample consists of aneurysms of a high rupture risk. Still, aneurysm growth is only one marker of increased risk, and rupture can occur without growth.24,25
Smoking and hypertension are other well-established independent risk factors for aneurysm rupture.8,33 Morphological changes, however, were similar for patients with and without these risk factors. Thus, in our material, the presence or absence of smoking and hypertension did not influence whether postrupture morphology was representative of that prior to rupture.
Studies comparing unruptured aneurysms with aneurysms presenting after rupture have generated important hypotheses about pathophysiology and risk factors for growth and rupture.34 With the addition of the present study to existing data, we argue that the postrupture morphology should not be considered a good surrogate in the evaluation of risk of rupture. Morphological and hemodynamic rupture predictors should be validated in studies of prerupture aneurysms.
The present material is subjected to selection bias because the included patients were selected to conservative management, except those who either refused treatment or experienced SAH while waiting for aneurysm repair. Several factors can affect the rupture risk: the included patients are somewhat older and possibly more comorbid than the expected average of SAH patients. The fraction of smokers in our material is somewhat lower than the country’s average (less than half versus 2 thirds, respectively), and the fraction of patients with multiple aneurysms is higher than what is commonly found in clinical series.
A length time bias may pertain to the included aneurysms because other more rupture-prone aneurysms might have ruptured early on in their pathogenesis, never being diagnosed before rupture. Also, patients with a recognized IA that succumbed to their aneurysm without reaching a hospital are not part of this study. Still, we regard the external validity as high because none of the clinical risk factors were statistically significantly associated with morphological change. However, the retrospective nature of the study reduces the accuracy of the patient risk factors recorded. The study sample is small and does not allow for definite conclusions, but pre- and postrupture angiograms of IAs are exceedingly rare, making adequately powered enquiries into this matter difficult. This study, thus, contributes to shed light on an area that is rarely available for investigation.
The neck diameter definition used in this study is the maximal neck size. This definition provided the highest interrater reliability, but reduces comparability with studies using average or minimal neck diameter definitions. To answer our study question, the interrater reliability was paramount. Manual measurements can introduce interrater discrepancies. Strict parameter definitions guided the measuring process to counter error. The intraclass correlation coefficient demonstrated substantial agreement between the 2 raters.
Finally, 45% of the cases were evaluated with different image modalities, introducing technical limitations in measurement precision. However, other studies have shown that different modalities can be reliably compared.35 Moreover, image quality, CTA and magnetic resonance angiography slice thickness, as well as radiocontrast filling effects introduce variability. This variability may be assumed to be of the same magnitude like the one meeting the clinician in every day practice when evaluating serial imaging in a patient. The observed changes in the present study are of such a magnitude that they still would be consistent even after considering a margin for measurement errors.
Aneurysm morphology was significantly different after rupture as compared with that before rupture. To an extent, changes had occurred in a nonuniform manner. The changes observed after rupture reflect the compound effect of time with successive growth and formation of irregularities and the impact of rupture per se. Postrupture morphology should not be considered a good surrogate in the evaluation of risk of rupture.
We thank Rune Grov Eilertsen (University Hospital of Northern Norway, Tromsø, Norway), Christian A. Helland (Haukeland University Hospital, Bergen, Norway), Ole Solheim (St. Olav Hospital, Trondheim, Norway), and Wilhelm Sorteberg (Oslo University Hospital Rikshospitalet, Norway) for participating in the data collection.
Sources of Funding
UiT The Arctic University of Norway funded the study. The funder had no role in the study design, data collection, data analysis, data interpretation, writing of the article or decision to submit for publication.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.116.015288/-/DC1.
- Received September 7, 2016.
- Revision received January 13, 2017.
- Accepted January 20, 2017.
- © 2017 American Heart Association, Inc.
- Müller TB,
- Sandvei MS,
- Kvistad KA,
- Rydland J,
- Håberg A,
- Vik A,
- et al
- Greving JP,
- Wermer MJ,
- Brown RD Jr,
- Morita A,
- Juvela S,
- Yonekura M,
- et al
- Schneiders JJ,
- Marquering HA,
- van den Berg R,
- VanBavel E,
- Velthuis B,
- Rinkel GJ,
- et al
- Yi J,
- Zielinski D,
- Chen M
- Shrout PE
- Sadato A,
- Hayakawa M,
- Tanaka T,
- Hirose Y
- Backes D,
- Rinkel GJ,
- Laban KG,
- Algra A,
- Vergouwen MD
- Bor AS,
- Tiel Groenestege AT,
- terBrugge KG,
- Agid R,
- Velthuis BK,
- Rinkel GJ,
- et al
- Sonobe M,
- Yamazaki T,
- Yonekura M,
- Kikuchi H
- Froelich JJ,
- Neilson S,
- Peters-Wilke J,
- Dubey A,
- Thani N,
- Erasmus A,
- et al
- Serrone JC,
- Tackla RD,
- Gozal YM,
- Hanseman DJ,
- Gogela SL,
- Vuong SM,
- et al
- Villablanca JP,
- Duckwiler GR,
- Jahan R,
- Tateshima S,
- Martin NA,
- Frazee J,
- et al
- Teo M,
- St George EJ
- Brinjikji W,
- Zhu YQ,
- Lanzino G,
- Cloft HJ,
- Murad MH,
- Wang Z,
- et al
- Sandvei MS,
- Romundstad PR,
- Müller TB,
- Vatten L,
- Vik A
- Xiang J,
- Natarajan SK,
- Tremmel M,
- Ma D,
- Mocco J,
- Hopkins LN,
- et al
- Goubergrits L,
- Schaller J,
- Kertzscher U,
- Petz Ch,
- Hege HC,
- Spuler A