Uptake and radiological findings of screening cerebral magnetic resonance scans in patients with hereditary haemorrhagic telangiectasia

Abstract

Hereditary haemorrhagic telangiectasia (HHT) results in arteriovenous malformations (AVMs), most commonly in the lungs, liver and brain. Discussion of cerebral vascular malformations is an important element of patient management. The current study objectives were to examine uptake and results of screening cerebral magnetic resonance (MR) scans, excluding symptomatic patients requiring neurological investigations. The remaining non-symptomatic individuals received formal pretest counselling that differed according to family history. For the 603 patients with no neurological symptoms of concern, screening scan uptake was higher after publication of the ARUBA trial. Patients with a family history of cerebral haemorrhage were 4 to 14-fold more likely to have a screening scan than patients with no such family history. For patients without neurological symptoms suggesting cerebral AVMs, none of the 59 screening scans performed at our institution demonstrated a cerebral AVM. Four scans (6.8%) demonstrated small aneurysms. The most common abnormality was cerebral infarction (20/59, 33.9%), predominantly identified in patients with pulmonary AVMs. Of 29 pulmonary AVM patients with no previous history of clinical stroke, 16 (55.2%) had between one and five silent infarcts. For HHT patients with pulmonary AVMs, the most frequently affected sites were the cerebellum (40%) and thalamus (14.3%), and the age-adjusted odds ratio for an infarct was 21.6 (95% confidence intervals 3.7, 126), p = 0.001. We concluded that for cerebral screening programmes in HHT, the findings support informed patient choice incorporating understanding that cerebral AVMs are rare in non-symptomatic HHT patients, but that screening scans commonly detect silent cerebral infarction due to pulmonary AVMs.

Keywords: Cerebral infarction; arteriovenous fistulas; cardioembolic stroke; counselling; magnetic resonance imaging; paradoxical emboli; pulmonary arteriovenous malformations; stroke.

Figure 1.

Audit time periods and boundaries. Number of patients refers to number of HHT cases audited across the 4 time periods. Bar charts indicate the % (mean) and 95% confidence intervals of the mean.

Figure 2.

Overall MR scans rates. Percentage of 842 patients audited who had MR scans performed for any indication, across the 4 time periods as in Figure 1: (A) Prior to review by us (external scans), and (B) Following review by us (internal and external scans). Bar charts indicate the % (mean) and 95% confidence intervals of the mean. Displayed p-values calculated by Dunn's test after Kruskal Wallis (overall p-values A: p = 0.023; B: p = 0.0091).

Figure 3.

Proportion of non-symptomatic cases undergoing screening scans in 2015-2017. Patients are categorised by family history of haemorrhage (Group 2), and by whether the scans were performed by other institutions (grey bars, "Pre") or at our institution (black bars, "By us"). Bar charts indicate the % (mean) and 95% confidence intervals of the mean. Overall Kruskal-Wallis p < 0.0001, displayed p-values calculated by Dunn's test.