Introduction
The development of cerebrovascular alterations is an important intracranial complication in acute bacterial meningitis and is associated with poor outcome [
1]. Arterial cerebrovascular complications are reported to occur in approximately one-fifth of patients [
2]. Using digital subtraction angiography (DSA), arterial narrowing is the predominant finding in patients with arterial complications, involving all vessel sizes [
1]. The underlying reason for narrowing of the cerebral arteries during bacterial meningitis is still a matter of debate. Autopsy and animal model studies indicate severe inflammation of the vessel walls (vasculitis) as a key etiology [
3,
4]. Furthermore, patients in whom histopathological correlates in terms of inflammation were not found at the sites of arterial narrowing have been reported [
5,
6]. This suggests vasospasm as a second important etiology. Also, ischemic stroke can occur as the result of endocarditis and septic emboli. Recently, the development of arterial cerebral infarctions was reported to occur up to weeks after initial recovery from pneumococcal meningitis and an association with adjunctive steroid therapy was suggested [
7]. Autopsy studies in such patients showed intraarterial clotting but no signs of vasculitis and intraarterial thrombosis was suggested to be a fourth reason for the development of stroke [
7]. Furthermore, diffuse cerebral vascular coagulation was proposed as the main etiology for stroke in another autopsy series on 15 patients who died from pneumococcal meningitis [
8]. Unfortunately,
in vivo vascular imaging was not performed in these two studies. Given the fact that arterial complications in bacterial meningitis are associated with a poor prognosis, the question how arterial alterations can be detected early and in a feasible way during bacterial meningitis arises. Here, we retrospectively studied the data of 114 patients with acute bacterial meningitis who were treated at our university hospital from 2000 to 2009 with the scope on an increase of cerebral blood flow velocity (CBFv) on transcranial Doppler ultrasound (TCD) and its link with stroke and an adverse outcome.
Discussion
The major findings of our study were that (i) 43.6% of patients showed vascular alterations on TCD and (ii) these alterations were seen more often in patients with a poor GCS on admission. Increased systolic CBFv greater than 150 cm/s evidenced by TCD were associated with (iii) the development of ischemic stroke and (iv) unfavorable outcome. (v) Treatment of an increase of CBFv with nimodipine was not associated with a reduction of stroke.
The current study demonstrates that an increase of CBFv is common in bacterial meningitis. Here, we detected an increase of systolic CBFv greater than 150 cm/s in 41 of 94 patients (43.6%) with acute bacterial meningitis. In earlier smaller case series, similar rates have been reported: for example, an increase of the CBFv was found in 23 of 41 patients (51%) with acute bacterial meningitis [
11] and in 27 of 53 (51%) patients with bacterial meningitis during the first two weeks of admission [
12]. Also, a very high incidence of vascular alterations was reported in a case series of 22 patients with bacterial meningitis, with markedly increased systolic peak velocities (> 210 cm/s) in seven patients [
13]. Taken together with the high incidence of TCD alterations found in the large group of patients of this study, this demonstrates that cerebrovascular complications are frequently found in patients with bacterial meningitis. The high incidence of cerebrovascular complications found on TCD, in comparison with that reported by studies looking at alterations on MR angiography, DSA, and ischemic stroke in selected patients, is likely to only be related to the simplicity of using TCD. This also allowed screening for vascular alterations in critically ill patients without the necessity for invasive, time consuming, and, sometimes, risky investigations that are often only performed if focal neurologic signs appear. In addition, the cost is low. One limitation of TCD is the necessity to obtain bone cranial windows, which might be difficult in some patients. However, the use of bedside TCD seems to be a helpful and feasible tool for clinical practice to screen patients with bacterial meningitis for vascular alterations, aiming at the identification of patients at high risk for ischemic stroke.
Alterations of CBFv on TCD were seen more often in patients that were severely affected clinically on admission. This is in line with the previous observations that a low GCS and a low CSF glucose on admission are risk factors for intracranial complications and poor outcome [
2,
14]. Calculation of odds ratios identified initial adjunctive corticosteroid treatment to be associated with the development of arterial complications. One likely explanation is that corticosteroids were especially given to patients with severe clinical affection on admission. Thus, it cannot be concluded from the results of this study that adjunctive corticosteroids lead to the development of arterial narrowing. However, it has to be kept in mind that case series discussing a possible association of steroid use and stroke in pneumococcal meningitis have recently been published [
7].
An increase of CBFv greater than 150 cm/s on TCD was associated with an increased incidence of ischemic stroke. Nevertheless, negative TCD studies do not rule out the chance of ischemic stroke: 2 of 53 meningitis patients without an increase of CBFv on TCD developed cerebral ischemia (one of them suffering from endocarditis). In the daily routine on our ICU, systolic CBFv greater than 150 cm/s was considered indicative of vasculopathy. Nevertheless, arterial narrowing indicated by an increase of CBFv was only confirmed by MR angiography in 6 of 13 patients, CT angiography in one of five patients, and DSA in two of two patients. Unfortunately, exact timely correlations with imaging studies, especially DSA (gold standard) were not possible from our data. Thus, it remains unclear which threshold of CBFv on TCD should be considered pathologic in patients with bacterial meningitis. Nevertheless, here, a threshold of 150 cm/s helped to identify 14 of 16 patients (87.5%) with stroke.
Vascular alterations occur mainly during the first days after admission but can be found even later. If alterations were detected on TCD in this study, they were present more than one week after admission in 21%. This is in line with the observation that stroke can occur later after the onset of bacterial meningitis [
1,
3,
7,
15]. As this was a retrospective study without a predetermined time point of TCD and/or other imaging evaluation of the patient, the exact onset of vascular alterations and/or stroke might have been missed. Nevertheless, the fact that an increase in CBFv was closely associated with stroke in such a high percentage even though TCD was performed not on a predetermined timely basis underlines its value in clinical practice. Altogether, this demonstrates the necessity to monitor patients early, closely, and long enough in order to identify patients at risk of stroke early.
What can be done if vascular alterations such as arterial narrowing are detected? An increase of CBFv on TCD can be the result of narrowing of the arterial lumen or hyperperfusion. Possible underlying reasons for narrowing of cerebral arteries in meningitis were shown to be vasospasm and/or vasculitis, as evidenced by histopathology in patients who died during acute bacterial meningitis [
1,
3]. Clearly, randomized controlled trials on possible treatment strategies of arterial narrowing in bacterial meningitis are non-existent. Another disease with prolonged narrowing of cerebral arteries is subarachnoid hemorrhage (SAH) [
16], TCD having become a valuable tool to identify patients with SAH-associated vasospasms [
17,
18]. Such patients with SAH that develop cerebral vasospasms are often treated with an elevation of blood pressure, hypervolemia, and hyperperfusion (triple H therapy) in combination with intravenous or oral nimodipine [
19]. Whether such therapeutic strategies can be transferred to practice in bacterial meningitis is unclear at present. Due to the risk of cerebral bleeding in infections of the central nervous system, extensive hypertension might even be harmful to the patient. Nimodipine would potentially also act in the case of meningitis-associated vasospasm (but not vasculitis) as the underlying etiology for an increase of CBFv on TCD. Here, therapy with intravenous or oral nimodipine was not of benefit. One could argue that a possible effect of nimodipine might have been missed due to the very small number of patients not receiving nimodipine, but the proportion of patients with stroke despite nimodipine treatment was high. Administration of nimodipine is often associated with a drop of the arterial blood pressure. Keeping in mind that the cerebral autoregulation can be affected during bacterial meningitis [
20], such a decrease of the blood pressure could easily affect the cerebral perfusion pressure, potentially being of risk to the patient (especially in the case of vasculitis of large and small blood vessels). Combined, nimodipine cannot be recommended for routine therapy of meningitis-related TCD alterations.
Intraarterial clotting might be another mechanism for the development of stroke. Thus, anticoagulation with heparin or antiplatelet strategies could be discussed as potential treatment strategies. However, given the fact that intracranial bleeding is a frequent complication in pneumococcal meningitis, administration of drugs that interfere with coagulation seem dangerous. In a small study on the use of heparin which included 15 patients with bacterial meningitis, outcome not improved in patients receiving heparin vs. those not receiving heparin (mortality 57% vs. 25%, respectively) [
21]. Thus, the use of heparin cannot be recommended without clinical studies. If strategies that aim at the inhibition of the coagulation cascade or stimulation of fibrinolysis could be of benefit in bacterial meningitis yet need to be evaluated in experimental studies.
Due to its retrospective nature, this study has several limitations. Only 94 of 114 patients with bacterial meningitis received TCD studies. However, it has to be noted that the outcome of the remaining 20 patients without TCD studies was worse than in the group of patients with TCD studies (mortality in patients without TCD studies = 55% vs. mortality in patients with TCD studies = 5.3%). The major reason for this selection bias was that in many of the non-TCD-investigated patients, clinical problems such as renal failure with the need of dialysis, airway control, and severe sepsis and extensive blood pressure management as well as control of intravascular disseminated coagulopathy dominated the treatment of these patients, giving highest priority to high urgent directly life saving measures. Thus, the actual number of patients with an increase of CBFv on TCD might have been underestimated. Unfortunately, continuous TCD studies were not available in this patient series, therefore we cannot comment on the course of vascular alterations and it is unclear how long patients with vascular alterations remain at risk for ischemia. However, the fact that patients with bacterial meningitis develop stroke even weeks after meningitis shows that a high level of suspicion for vascular complications should be sustained even after the acute stage of the disease is overcome.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
MK developed the study design and carried out the data collection, data analysis, manuscript draft, and revision. UK contributed with critical manuscript revisions. TP contributed with critical manuscript revisions. GZ carried out data collection and manuscript revision. BW contributed with manuscript revision. HWP brought up the study idea and carried out critical manuscript revision. All authors have read and approved the manuscript for publication.