Can radiological characteristics of preoperative cerebral lesions predict postoperative intracranial haemorrhage in endocarditis patients?

Objectives: Infective endocarditis (IE) is associated with high mortality (20-40%) and neurological complications (20-50%). Postoperative intracranial haemorrhage (ICH) is a feared complication especially in patients with preoperative cerebral infarcts. The aim of this study was to determine the radiological characteristics of cerebral lesions that could predict the occurrence of postoperative ICH in IE patients.

Methods: We retrospectively reviewed all charts, brain imaging and follow-up data from patients operated for left-sided endocarditis between January 2007 and April 2013.

Results: A total of 308 patients (age 62.0 ± 13.9) underwent surgery for IE. Preoperative cerebrovascular complications were present in 122 patients (39.6%), representing stroke in 87, silent cerebral infarctions in 31 patients and transient ischaemic attacks in 4 patients. Among 118 patients with cerebral lesions, the aetiological classification of the lesions was ischaemic in 63.6%, ischaemic with haemorrhagic transformation (HT) in 17.8%, ischaemic with concomitant microbleeds in 16.1% and intracerebral bleeding in 2.5%. Postoperative ICH occurred in 17 patients and its incidence was slightly higher in patients with preoperative cerebral infarcts compared with those without preoperative cerebral infarcts [7.6 vs 4.2%, respectively, odds ratio (OR) 1.88, 95% confidence interval (CI) 0.70-5.02, P = 0.21]. However, the difference was not statistically significant. Similarly, the incidence of postoperative ICH was higher in cases of HT of ischaemic infarcts than in cases of ischaemic infarcts not complicated with HT (19.0 vs 5.3%). However, the difference was not statistically significant (P = 0.24). The radiological pattern of preoperative cerebral lesions was single in 35.6% and multiple in 60.0% of cases. Multiple cerebral lesions were associated with a non-significantly lower incidence of postoperative ICH than single lesions (5.6 vs 11.9%, respectively, OR: 0.44, CI: 0.11-1.73, P = 0.29).

Conclusions: The results suggest that the incidence of postoperative ICH in IE patients was slightly higher in the presence of preoperative cerebral infarcts. In addition, preoperative cerebral ischaemic infarcts complicated with HT tended to have a higher incidence of postoperative ICH than those not complicated with HT. However, the difference was not statistically significant. Multiple preoperative cerebral infarcts were not associated with higher incidence of postoperative ICH compared with single cerebral infarcts.