Multiple Aneurysms: Insight through the Mirrors

The overall prevalence of intracranial aneurysms (IAs) is estimated to be 3.2% in the general population [1]. Among carriers, it is estimated that up to 20% have multiple IAs potentially located in a symmetrical paired area at each side of the circle of Willis and named mirror IAs (MIAs) [2‒4]. The prevalence of MIAs is less than 5% among all patients with an IA but rises to 40% in bearers of multiple IAs [4]. A few studies have reported the characteristics of MIAs [2‒5]. Furthermore, it is conceivable that, in a patient with multiple IAs, one IA could be located symmetrically opposite to another by chance, without there being any real segmental arterial vulnerability. Indeed, the very nature of MIAs has never been clearly defined and MIAs have usually been described as a symmetrical occurrence of an IA whatever the location on a paired site [2]. In a large international population of patients with multiple IAs, we analysed the characteristics of MIAs and quantified the prevalence of MIAs for different potentially symmetrical localizations, aimed at gaining further insight into the notion of MIA.

We included IA patients from the ICAN study (France) and the population-based Kuopio saccular IA Database (Finland); detailed information about these cohorts has been previously published [6, 7]. This article follows the STROBE reporting guidelines [8]. We analysed patients with ≥2 definite IAs, IA being defined as a saccular arterial dilatation, whatever the measurement, occurring at a bifurcation of the intracranial vessels. IA locations were categorized into eight paired and unpaired locations (see online suppl. Methods; for all online suppl. material, see https://doi.org/10.1159/000543053). Neuroradiological phenotyping and baseline characteristics of IA carriers were recorded in each centre by experienced interventional neuroradiologists and neurosurgeons (see online suppl. Data). Patients were classified as “mirror patients” if they had at least one pair of IAs at symmetrical locations. Mirror patients were further classified as “pure MIA patients” if they had no additional asymmetrical IAs, “mixed MIA patients” if they had at least one other, asymmetrical IA, and “pure asymmetrical IA patients” if they had multiple and entirely asymmetrical IAs. We measured the largest dimension of the IA (mm) and the size ratio similarity as a ratio of each smaller over larger MIA.

For multivariate analyses, logistic regression models were fitted to compare MIA patients between patients who harboured an IA in each location, with different levels of adjustments, and odds ratios with corresponding 95% CIs were calculated. The covariates of interest were the number of IAs (bivariate model) and the family history of IA (trivariate model). Conditional probabilities were calculated for paired locations to assess the likelihood of having an IA in a location given the presence of an IA in another location (online suppl. Fig. 1). All statistical analyses were performed using R (version 4.0.2, R Foundation for Statistical Computing, Vienna, Austria).

A flowchart of patient selection is available in online supplementary Figure 2. Among 2,124 patients harbouring 5,459 IAs, we found 754 patients (35.5%) with 798 pairs of MIAs (14.6% of all IAs). MIA patients were predominantly female (541; 72%). Among MIA patients, 344 (16.2%) had a pure MIA phenotype and 410 (19.3%) had a mixed MIA phenotype, while 1,370 patients (64.5%) had only pure asymmetrical MIAs.

Patients’ clinical characteristics are shown in online supplementary Table 1. No statistical differences were found between the pure MIA patients and the others regarding most of the baseline characteristics. A history of subarachnoid haemorrhage was more frequent in non-MIA patients (52% vs. 46%, p = 0.006). Non-MIA patients had significantly more IAs on ACom, basilar, ACA, subdural ICArt, ophthalmic ICArt, ICArt/AChA, posterior communicating junction, and PICA locations, whereas patients with MIAs had more IAs with an MCA location.

The anatomical and morphological characteristics of IAs are shown in online supplementary Table 2. MIAs were larger than non-MIAs (mean ± SD: 5.6 ± 4.5 vs. 5.2 ± 4.2 mm, p < 0.001). Non-MIAs were ruptured in 53% of cases (vs. 47% of MIAs, p < 0.001). No differences were found regarding family history of IA. MIAs had a mean size similarity of 0.55 ± 0.25 (max 1).

We performed multivariable logistic regression models using different levels of adjustment to find locations associated with MIA patient status. After adjustment for the number of IAs, the only location associated with MIA patient status was MCA (bivariate model, odds ratio 1.36 [95% CIs 1.08–1.72], p = 0.0087). Moreover, when family history of IA (trivariate model 3) was added to the previous model, the result was unchanged (Table 1). Conditional probability revealed patterns of IA co-occurrence among various locations, especially the MCA (online suppl. Fig. 3).

In this large international population with multiple IAs, our analysis found that MIAs were present in 35.5% of patients. Our findings emphasize the specificity of MCA as the location with the highest probability of having an MIA, taking into account the number of aneurysms and their prevalence in our cohort. Among carriers of multiple IAs, the probability of having an MIA if an MCA IA was present was nearly 50%.

We found a higher rate of ruptured IAs among non-MIAs than among MIAs (52% vs. 47%, p = 0.006) although MIAs were larger in size than non-MIAs (5.6 ± 4.5 vs. 5.2 ± 4.2 mm, p < 0.001). This was likely due to the higher proportion of MCA IAs in the MIA patient group. Indeed, IAs at this location are known to be at a low risk of rupture [9]. These results are also in line with the previously published finding that the location of an IA rather than its size is a better indicator of its rupture risk [10].

From a pathophysiological point of view, MCA MIAs might account for a specific vulnerability to the occurrence of a mirror phenotype in patients with multiple IAs. Of note, genetic risk factors have a larger role in the development of aneurysms at the MCA than at other sites [11]. Indeed, contrary to other potential symmetrical locations, the MCA accounted for a large proportion of symmetric IAs. One particular aspect of the MCA is the constancy of its anatomy, with few variations, whether other common IA sites lie on the circle of Willis, which is the site of frequent and asymmetrical variations and could explain the overrepresentation of the MCA within MIA cases [12]. From a clinical point of view, radiologists should pay particular attention to MCA location during diagnosis and follow-up to avoid missing an undiagnosed MIA status or the de novo appearance of an IA.

Our study has several limitations. Firstly, temporal data on IA occurrence were not recorded, thereby limiting the potential extrapolation of predictive ability for de novo IAs. Secondly, we were unable to analyse the geometric aspects and IA patterns among patients and within families.

MIA is a particular condition within multiple IA carriers and has a greater probability of occurring at the MCA than at other locations. This suggests a focal vulnerability for this specific MCA location, which therefore merits particular attention during imaging screening and follow-up.

Ethical approval was waived in accordance with national regulations as retrospective studies involving the analysis of previously collected data do not require ethical approval. Moreover, the ICAN protocol was approved by our local Ethics Committee, GNEDS (opinion issued on February 23, 2016), and by a national Ethics Committee (Approval No. DC-2011-1399). The Kuopio Intracranial Aneurysm Database was approved by the Regional Medical Research Ethics Committee of Wellbeing Services County of North Savo (Approval No. 1095/2021). Written informed consent was obtained from the patients for their participation to ICAN while the need for informed consent was waived (registry study) in the Kuopio Intracranial Aneurysm Database.

The authors have no conflicts of interest to declare.

This study was not supported by any sponsor or funder.

CRediT authorship contribution statement: Pacôme Constant dit Beaufils (P.C.D.B.): conceptualization, methodology, and writing – original draft. Matilde Karakachoff (M.K.): methodology, writing – original draft, and writing – review and editing. Pierre-Antoine Gourraud (P.-A.G.): methodology and writing – review and editing. Pierre Thouant (P.T.) and Antti Lindgren (A.L.): investigation and writing – review and editing. François Zhu (F.Z.): writing – investigation and writing – review and editing. Romain Bourcier (R.B.): conceptualization, investigation, and writing – original draft preparation and review.

The data that support the findings of this study are not openly available for reasons of privacy. Synthetic data are available from the corresponding author upon reasonable request.