Population-Based Prevalence of Cerebral Cavernous Malformations in Older Adults

This population-based study analyzes the magnetic resonance imaging brain scans of participants in the Mayo Clinic Study of Aging to determine the prevalence of cerebral cavernous malformation in adults aged 50 to 89 years.

Key Points

Question

What is the prevalence of symptomatic and asymptomatic cerebral cavernous malformations in older adults?

Findings

In this population-based study of 2715 participants aged 50 to 89 years who underwent brain magnetic resonance imaging for nonclinical purposes, the overall prevalence of cerebral cavernous malformation was 0.46%, and the observed frequency of symptomatic cerebral cavernous malformation was 0.037%.

Meaning

Because symptomatic cerebral cavernous malformations are rare, multicenter involvement will be necessary for future clinical trials.

Abstract

Importance

The prevalence of cerebral cavernous malformation (CCM) is unknown. Case ascertainment in most previous studies was based on autopsy data or clinical convenience samples, often without detailed clinical or radiologic information.

Objective

To determine the prevalence of CCM in a population-based sample of older adults.

Design, Setting, and Participants

This prospective imaging study included 4721 participants aged 50 to 89 years who were enrolled between January 1, 2004, and December 15, 2015, in the Mayo Clinic Study of Aging, a longitudinal, population-based study of residents of Olmsted County, Minnesota. An age- and sex-stratified sampling strategy was used to randomly select participants from Olmsted County using the medical records linkage system of the Rochester Epidemiology Project. Participants were invited to undergo brain magnetic resonance imaging (MRI). Of the 4721 participants, 2715 had an evaluable MRI. All images were reviewed by a board-certified neuroradiologist, and MRI reports were searched for the terms cavernous malformation, cavernous angioma, and cavernoma. Two vascular neurologists reviewed MRIs, and potential CCMs were classified using Zabramski classification. Medical records of the identified individuals with CCM were reviewed along with their demographic information, medical history, and any symptoms referable to the identified CCM lesion.

Main Outcomes and Measures

Prevalence of CCM and clinical and radiologic characteristics of study participants with CCM.

Results

Of the 2715 participants who underwent MRI scans, 12 (0.44%) had CCM. With the use of inverse probability weights to adjust for participation bias, the overall prevalence was 0.46% (95% CI, 0.05-0.86). The age-adjusted prevalence was found to be 0.61% (95% CI, 0-1.47) for the 50- to 59-year age group, 0.17% (95% CI, 0-0.50) for the 60- to 69-year age group, 0.45% (95% CI, 0.09-0.81) for the 70- to 79-year age group, and 0.58% (95% CI, 0-1.29) for the 80- to 89-year age group. The sex-adjusted prevalence was 0.41% (95% CI, 0-1.00) for women and 0.51% (95% CI, 0-1.07) for men. Observed frequencies were similar in men and women, with a slight male predominance. Of the 12 participants with CCM, 9 (75%) had a single Zabramski type 2 lesion in a supratentorial location. Only 1 participant (0.037%) was symptomatic from the CCM during the study period.

Conclusions and Relevance

The findings and data from this study are important for determining the potential number of patients available for cohort studies and anticipated clinical trials in older patients with CCM.

Introduction

The exact prevalence of cerebral cavernous malformations (CCMs) is unknown because many patients with CCM are asymptomatic. Autopsy studies estimate the prevalence of CCM to be between 0.2% and 0.5% of the population. However, autopsy studies may suffer from selection bias (low rate of autopsy), referral bias (because they are not population based), sampling limitations, and limited or no clinical information. In addition, there may be significant differences in interobserver variation depending on the level of scrutiny for particular lesions between pathologists.

Studies have also attempted to estimate CCM prevalence on the basis of patients undergoing magnetic resonance imaging (MRI) for clinical purposes. Studies using this method for case ascertainment report prevalence to be between 0.39% and 0.9%. In those studies, as many as 40% to 81% of patients may have had symptoms. However, those studies have potential biases. First, studies that are not population based may suffer from referral bias. Second, even population-based studies may be biased if the ascertainment of cases requires clinical symptoms or clinically indicated imaging. Patients with clinical symptoms or undergoing clinical MRI may differ from the general population.

Other CCM prevalence studies, including 1 large meta-analysis, assessed the prevalence of a variety of brain conditions in participants undergoing MRI for nonclinical purposes. The estimated prevalence of CCM in those studies ranged from 0.16% to 0.7%. However, many of those studies lacked clinical and radiographic detail on participants identified with CCM.

In this study, we aimed to determine the prevalence of CCM among individuals aged 50 to 89 years participating in the Mayo Clinic Study of Aging (MCSA), a population-based longitudinal study that included MRI. We also aimed to characterize the type, number, and location of lesions and the symptoms experienced by these participants.

Methods

Participants of this study were enrolled in the MCSA, a longitudinal, population-based study of residents of Olmsted County, Minnesota. The MCSA study design and methods have been published elsewhere. Briefly, in 2004, MCSA investigators identified Olmsted County residents aged 70 to 89 years using the medical records linkage system of the Rochester Epidemiology Project. An age- and sex-stratified sampling strategy was used to randomly select participants from Olmsted County. Race/ethnicity were self-reported, and participants were invited to undergo brain MRI starting in 2005. In 2012, the MCSA expanded to enroll a population-based sample of participants aged 50 to 69 years using the same methods as those used for the initial cohort. For the present study, we assessed the prevalence of CCM among MCSA participants aged 50 to 89 years. These participants enrolled between January 1, 2004, and December 15, 2015, and underwent an evaluable MRI for cavernoma. All study protocols were approved by the institutional review boards of the Mayo Clinic and the Olmsted Medical Center. All participants provided written patient informed consent before enrolling in the MCSA.

Medical records of the identified individuals with CCM were reviewed. Demographic information, medical history, and any symptoms referable to the identified CCM lesion were recorded.

Brain MRI was performed on a 3-T system (GE Healthcare). In 2011, T2*-weighted gradient recalled echo (T2*GRE) was added to the MRI protocol. The 2-dimensional straight, axial, gradient-echo pulse sequence was acquired with a repetition time of 200 milliseconds, echo time of 20 milliseconds, section thickness of 3.3 mm with 49 sections, flip angle of 20°, and in-plane matrix of 256 × 224. All MRIs were reviewed by a board-certified neuroradiologist for abnormalities included in the MRI report. Then, the MRI reports for MCSA participants were searched for the terms cavernous malformation, cavernous angioma, and cavernoma. A radiology report noted that 26 individuals had a possible cavernous malformation or cavernous malformation in the differential diagnosis of the lesion identified. These MRIs of potential identified patients with CCM were then reviewed by 2 vascular neurologists (K.D.F. and J.G.-R.).

The potential CCMs were classified using Zabramski classification. Possible type 4 CCMs (n = 7), defined as those identified on GRE sequences only, were excluded from the main outcome because a microbleed resulting from an alternative cause (eg, hypertension, amyloid angiopathy, radiation, or trauma) could not be fully excluded. An additional 7 patients were excluded because, on the basis of standard criteria, they could not be definitely characterized as having CCM. The latter were often participants whose MRIs demonstrated irregular areas of reduced T2 signal suggestive of hemosiderin that were larger than a microbleed but did not definitively have the appearance of a cavernous malformation. Many irregular areas had the appearance suggestive of superficial siderosis.

The numbers of CCM on MRI were counted either on hemosiderin-sensitive sequences when available or on standard T2 sequences when hemosiderin-sensitive sequences were not available (prior to 2011). Notation was made of the location and size of the cavernous malformation on T2 sequences as well as whether an associated developmental venous anomaly was present.

Descriptive statistics, including medians and ranges for continuous variables and frequencies and percentages for categorical variables, were used to summarize the characteristics of patients found to have CCM.

Frequencies of CCM were calculated directly for the 4 age strata following the original sampling scheme of dividing the number of observed CCM by the number of imaged participants (aged 50-59, 60-69, 70-79, or 80-89 years). These frequencies were then adjusted for potential nonparticipation bias using inverse probability weighting. The weights were derived using logistic regression models, adjusting for age, sex, and educational level. The Rochester Epidemiology Project’s medical records linkage system allowed us to use the same source to derive information on nonparticipants and participants alike. Adjusted estimates were similar to unadjusted estimates, but we chose to present the adjusted results in all tables. Finally, to compute prevalence estimates, these adjusted frequencies were standardized by age and sex to the Rochester Epidemiology Project Olmsted County enumeration in 2010 (N = 44 732; aged 50-89 years) using the sampling fractions from our population-based sampling. All statistical analyses were performed using SAS software, version 9.4 (SAS Institute Inc).

Results

The Rochester Epidemiology Project 2010 enumeration included the 44 732 people aged 50 to 89 years living in Olmsted County; 4721 people who participated in the MCSA between January 1, 2004, and December 15, 2015; 2715 MCSA participants who underwent brain MRI; and 12 MCSA participants who were identified with CCM. The overall unadjusted frequency of CCM was 0.44% (12 of 2715). The sex-adjusted prevalence was 0.41% (95% CI, 0-1.00) for women and 0.51% (95% CI, 0-1.07) for men. The population prevalence by 10-year age strata and the overall prevalence are shown in Table 1.

Table 1. Adjusted Prevalence of Cerebral Cavernous Malformation by Age Group.

Age Group, y	Prevalence, % (95% CI)
50-59	0.61 (0-1.47)
60-69	0.17 (0-0.50)
70-79	0.45 (0.09-0.81)
80-89	0.58 (0-1.29)
50-89	0.46 (0.05-0.86)

Of the 2715 participants who had an MRI scan, 12 (0.44%) were found to have CCM. Cervical cord cavernous malformation was detected on the sagittal brain MRI of 1 patient but was not included in the analysis. The median (range) age of participants with confirmed CCM was 76.8 (51.3-83.6) years, and most participants (8 [66.7%]) were male. No participant had symptoms relevant to CCM at the time of the research MRI. However, 1 patient had a symptomatic hemorrhage from CCM 17 years earlier (prior to the study’s inception date), and 1 other patient had an acute CCM-associated hemorrhage 1 year after the research MRI. Thus, between January 1, 2004, and December 15, 2015, the observed frequency of symptomatic CCM was 0.037% (1 of 2715). These 12 cases had 59 total patient-years of subsequent follow-up after the MRI; there were no additional CCM-associated symptoms, seizures, or hemorrhage. Three participants died in 2015, but none of them died because of brain bleeding or CCM symptoms.

Radiologically, most patients (9 [75%]) had single, supratentorial CCM (Table 2). The unadjusted prevalence of multiple CCMs without developmental venous anomaly, suggestive of a familial form, was 0.07% (2 of 2715). Eight patients (66.7%) had type 2 lesions, and 4 patients (33.3%) had type 3 lesions. The median (range) size of the lesion was 7.5 (3-17.6) mm.

Table 2. Demographic Information of 12 Study Participants With Identified CCM.

Characteristic	No. (%)
Age, median (range), y	76.8 (51.3-83.6)
Male	8 (66.7)
Race
White	10 (83.3)
Asian	1 (8.3)
Unknown	1 (8.3)
Ethnicity
Non-Hispanic	11 (91.7)
Unknown	1 (8.3)
Symptoms related to CCMa	2 (16.7)
Location of CCM
Supratentorial	9 (75)
Infratentorial	3 (25)
Size of CCM, median (IQR), mm	7.5 (3-17.6)
Multiplicity	2 (16.7)
Associated DVA	3 (25)
CCM type
2	8 (66.7)
3	4 (33.3)

Abbreviations: CCM, cerebral cavernous malformation; DVA, developmental venous anomaly; IQR, interquartile range; MRI, magnetic resonance imaging.

^aNo patient had symptoms referable to CCM at the time of the research MRI. One patient had symptoms before the research MRI, and 1 patient had symptoms after the research MRI that were referable to the CCM.

For participants in the MCSA (n = 2006) but not the research MRI portion, we searched the radiologic records for any clinical brain MRI performed during the study period. We found 2 additional patients who had symptomatic CCM. In total, among the 4721 MCSA participants, there were 4 symptomatic patients (0.08%) with CCM.

Discussion

In this population-based study, the adjusted prevalence of CCMs was 0.46% (95% CI, 0.05-0.86) among Olmsted County residents aged 50 to 89 years. This overall prevalence is similar to the prevalence found in other studies (range, 0.16%-0.9%) but significantly higher than that found in the original population-based study of vascular malformation incidence and prevalence in Olmsted County between 1965 and 1992. This difference is likely because ascertainment of cases in the previous study was based on clinical and radiographic diagnosis, and the study period (1965-1992) preceded the widespread availability of MRI. Thus, we consider it unlikely that prevalence has truly increased. In the Rotterdam study, also a population-based longitudinal study in which ascertainment of cases was based on research rather than clinical MRI, CCMs were among the top 5 incidental findings, with a prevalence of 0.63%. Only meningioma, aneurysm, arachnoid cyst, and pituitary abnormality were more frequent.

The observed frequency of symptomatic CCM during the study period was only 0.037% in the MCSA population, which differs from the frequency in previous studies in which as many as 40% to 81% of patients had CCM-related clinical symptoms. Thus, previous studies, which based case ascertainment on clinical MRIs only, overestimated the prevalence of symptomatic disease. Findings from our population-based study suggest that symptomatic CCM could qualify as a rare disease (ie, a condition affecting fewer than 250 000 people).

There were slightly more men than women with CCM in our study, but the difference was not significant. Similarly, the Rotterdam study showed a nearly equal prevalence of CCM in men and women. Many large clinical series have reinforced this finding as well. The prevalence was similar across the 4 age strata except the 60- to 69-year group, which may represent a spurious estimate.

Strengths and Limitations

A strength of this study is the clinical and radiologic characterization of the participants as a result of the availability of clinical records, which were lacking in some published studies. Ten participants (83%) with CCMs and between 50 and 89 years had a single lesion suggestive of the sporadic form of CCM, and 2 (17%) had multiple lesions without associated developmental venous anomalies suggestive of the familial form of CCM. This finding is consistent with findings in other large cohort studies, which showed that approximately 10% to 30% of patients with CCM had the familial form. Similar to investigators of non–population-based studies, we found that most CCMs are supratentorial. The median size of CCM as measured on T2 MRI was 7.5 mm, slightly smaller than the mean sizes of 1.4 to 1.7 cm found in other large clinical studies.

Although this study, to our knowledge, is one of the few population-based studies and the first to characterize clinical and radiologic findings of participants with CCM, it has limitations. Population-based studies are preferred to estimate true incidence and prevalence of a disease, but the total number of participants in the imaging portion of this study remains small. While the MCSA is population based, neuroimaging participation is voluntary among those without contraindications to MRI. Thus, the imaging subset may have some bias. We tried to overcome potential bias by using propensity analysis, which attempts to control for differences in the groups in observational studies such as this.

The MCSA, which samples from Olmsted County, is composed of predominantly white participants. Because the CCM1 (OMIM 116860) familial form occurs more commonly in the Hispanic population, the prevalence presented here may differ from the prevalence in the Hispanic population. However, the familial form is present in approximately 15% to 20% of patients with CCM, with CCM1 representing approximately half the familial form. The overall affect, we believe, would be minimal.

The T2*GRE sequences were only available after 2011. Because we did not include solitary Zabramski type 4 lesions in our prevalence calculation (see the Methods section for the rationale), the lack of T2*GRE sequences prior to 2011 should not affect our overall prevalence. However, we could have underestimated the percentage of the familial form (ie, those with multiple lesions without developmental venous anomalies). Data from the MCSA were limited to prevalence in people older than 50 years. Most data suggest that symptomatic disease occurs between the ages of 20 and 40 years. Thus, patients older than 50 years may have more quiescent disease, and we may underestimate symptomatic prevalence. However, concomitant comorbidities in the over-50 age group that require antithrombotics and other medical conditions might adversely influence the behavior of CCM, but we did not find that to be true.

Conclusions

We found an overall CCM population prevalence of 0.46% (95% CI, 0.09-0.81) in participants between 50 and 89 years of age. The observed frequency of symptomatic disease, however, was much lower. Radiologic characteristics were similar to those found in large cohort studies. This finding provides clinicians, researchers, and patients with CCM a more accurate measure of how frequently CCMs are detected in the population.