The epidemiology of brain arteriovenous malformations in adults
Arteriovenous malformations (AVMs) of the brain are part of the spectrum of intracranial vascular malformations (IVMs). They are the leading cause of intracerebral haemorrhage in young adults, they account for ~10% of non-traumatic subarachnoid haemorrhage, and they also cause epilepsy. Not only are affected individuals subject to the initial consequences of these events, but there are substantial risks of recurrent haemorrhage and epilepsy, and long-term disability. For a disorder discovered as long ago as the mid-nineteenth century, surprisingly little is known about it. In this thesis, I begin by systematically reviewing the sizeable medical literature about brain AVM frequency, presentation, clinical course and prognosis. I did not find a single prospective, truly population-based study, which is why I set up the Scottish Intracranial Vascular Malformation Study (SIVMS) with the multidisciplinary collaboration of the four clinical neuroscience centres in Scotland. SIVMS aspires to meet the standards of the ideal study of frequency and prognosis, by using multiple, overlapping sources of case ascertainment to prospectively recruit a population based inception cohort of adults, with explicit diagnostic criteria and outcome definitions for events which are validated by independent review. During 1999-2000, 96 adults (of whom 92 were definite) were detected with a first-in-a-lifetime diagnosis of a brain AVM in Scotland. Quality of baseline demographic, clinical and basic morphological data was excellent, although detailed variables about angioarchitecture were less complete, partly because only three-quarters of patients underwent catheter angiography. The cohort was distributed in proportion to the dispersion of the Scottish population, and standardised incidence ratios were not significantly different between healthboards. The sensitivity of ICD-10 coding of brain AVMs in hospital discharge data was 72% (95%CI 61% to 80%), and its positive predictive value was 46% (95%CI 38% to 55%). Reliance on hospital discharge data for case ascertainment or a requirement for catheter angiography to make the diagnosis would have biased the cohort. Furthermore, I found that expert neuroradiologists’ assessment of AVM angioarchitecture on catheter angiography was characterised by greater intra-observer than inter-observer agreement (which ranged from less than chance for e.g. ‘angiogenesis’ to almost perfect for e.g. nidus size). In a survey with multiple, overlapping sources of ascertainment confined to the Lothian healthboard region of Scotland, using capture-recapture analysis, I found the point prevalence of brain AVMs to be 18 (95%CI 16 to 24) per 100,000 adults. In SIVMS, the crude incidence of brain AVMs in Scotland in 1999 and 2000 was 1.1 (95%CI 0.9 to 1.4) per 100,000 adults per year. Of the incident adults, 53% were male and their median age at presentation was 45 years (range 16 to 81); one fifth were incidental discoveries and four fifths were symptomatic (presentation was with intracranial haemorrhage in 59%, one or more seizure(s) in 34%, and focal neurological deficits in 7%). 9% of cases were pure arteriovenous fistulae, 75% were lobar in location, 53% were superficial, and 22% had associated aneurysms. There appeared to be significant differences between SIVMS and well-established hospital-based cohorts. Having established brain AVM prevalence, incidence and the characteristics of presenting adults, the next stage for this work is to describe prognosis for this enlarging population-based cohort. The data are being collected, the hurdles of ethical approval have been negotiated, although the direction in which privacy legislation and confidentiality guidance are heading will make this a challenging task.