Article, Neurology

Cerebellar stroke presenting with isolated dizziness: Brain MRI in 136 patients

a b s t r a c t

Objective: To evaluate occurrence of cerebellar stroke in Emergency Department (ED) presentations of isolated dizziness (dizziness with a normal exam and negative neurological review of systems).

Methods: A 5-year retrospective study of ED patients presenting with a chief complaint of “dizziness or vertigo”, without other symptoms or signs in narrative history or on exam to suggest a central nervous system lesion, and work-up included a brain MRI within 48 h. Patients with symptoms commonly peripheral in etiology (nystag- mus, tinnitus, gait instability, etc.) were included in the study. Patient demographics, stroke risk factors, and gait assessments were recorded.

Results: One hundred and thirty-six patients, who had a brain MRI for isolated dizziness, were included. There was a low correlation of gait assessment between ED physician and Neurologist (49 patients, Spearman’s corre- lation r² = 0.17). Based on MRI DWI sequence, 3.7% (5/136 patients) had acute cerebellar strokes, limited to or including, the medial posterior inferior cerebellar artery vascular territory. In the 5 cerebellar stroke patients, mean age, body mass index (BMI), hemoglobin A1c, Gender distribution, and prevalence of hypertension were similar to the non-cerebellar stroke patient group. Mean LDL/HDL ratio was 3.63 +- 0.80 and smoking prevalence was 80% in the cerebellar stroke group compared to 2.43 +- 0.79 and 22% (respectively, p values b 0.01) in the non-cerebellar stroke group.

Conclusions: Though there was preselection bias for Stroke risk factors, our study suggests an important propor- tion of cerebellar stroke among ED patients with isolated dizziness, considering how common this complaint is.

Introduction

Dizziness is an ambiguous complaint, commonly seen in the Emer- gency Department (ED) and urgent care clinic. Millions of dizzy patients present each year, and represent about 4% of ED visits [1]. The term diz- ziness, as described by the patient, can refer to a broad range of com- plaints, including gait instability, lightheadedness-presyncope, vertigo or even consciousness clouding. The most frequent dizziness subtype is vertigo, which accounts for about 20-40% of the cases and typically in- volves dysfunction of the central or peripheral vestibular system [1,2].

Although dizziness resolves or ceases to affect quality of life in about three-quarters of patients after three months [3], it is critical that life-

? This study was done without funding of any type.

* Corresponding author at: Department of Neurology, Boston University School of Medicine, 72 E. Concord St, C3, Boston, MA 02118, United States.

E-mail addresses: [email protected] (M.D. Perloff), [email protected] (N.S. Patel), [email protected] (C.S. Kase), [email protected] (B. Voetsch), [email protected] (J.R. Romero).

threatening or morbid diagnosis be ruled out during initial assessment. These diagnoses include stroke, ischemic heart disease, drug reaction, and Acute infection among others. Stroke should certainly be suspected in a patient complaining of dizziness if they have concomitant risk fac- tors and motor or sensory deficits. Clinicians do not commonly consider cerebellar stroke in patients who complain of isolated dizziness (dizzi- ness being the only symptom, with a normal exam and review of sys- tems), although cerebellar stroke signs and symptoms may be few. Approximately 11% of patients with cerebellar infarction present with isolated symptoms mimicking peripheral vestibular disorders [4]. How- ever, what percentage of patients presenting with isolated dizziness harbor a cerebellar stroke, is unknown. As cerebellar stroke recovery can be rapid, patients may present and be discharged without diagnosis, knowledge of stroke etiology, or risk factor correction.

Being able to distinguish peripheral causes of dizziness and stroke has important implications for the patient. Stroke is the fourth leading cause of death and the leading cause of Long-term disability in the United States [5]. Dizziness is the leading symptom of posterior circula- tion strokes, particularly in those involving the cerebellum [6], and

http://dx.doi.org/10.1016/j.ajem.2017.06.034

often, it is the only symptom on presentation [7]. While cerebellar stroke itself does not typically lead to significant morbidity, it can be an initial warning sign for a more devastating brainstem stroke in the future.

Previous studies researching the link between dizziness and stroke have shown different results based on the methods of diagnosing and detecting the disease [4,7-10]. The population-based study of in Kerber et al. showed that there was no increased risk of stroke in patients who presented with isolated dizziness. However, this study did not utilize MRI or uniform criteria to make the final diagnosis of stroke [9]. Many studies have suggested that the incidence of cerebellar stroke in pa- tients presenting with isolated vertigo is higher than that diagnosed [8,11,12]. However, no study to date has looked at the incidence of cer- ebellar stroke in patients with isolated dizziness with diffusion weight- ed MRI as the gold standard.

Material and methods

Study design

This was a 5-year retrospective review (01/01/2005-01/01/2010) of patients presenting to the ED with a chief complaint of “dizziness or ver- tigo” with no other symptoms or signs in narrative history or on exam to suggest a central nervous system lesion, and with work-up that includ- ed a brain MRI within 48 h. ED physicians do not typically order brain MRI in the setting of obvious medical illness (infection, metabolic de- rangements, hypotension, hypertensive urgency, etc.) and these pa- tients were eliminated. Patients with nystagmus (lateralizing, unchanging), tinnitus, hearing loss, an abnormal head impulse test or Dix-Hallpike maneuver, or gait instability were included as these

symptoms and signs are commonly peripheral in etiology. Two re- viewers graded the ED physician and Neurology consult gait assessment and rated it as normal, widened/difficulty with stress gait, unsteady/ ataxic but able to walk without assistance, or unable to walk without as- sistance. Any disagreements employed a third reviewer and a consensus rating was made (three occasions). Acute stroke was assessed by diffusion-weighted imaging on MRI. The study design was approved by the Boston Medical Center Institutional Review Board.

Setting

The study was conducted at a 482-bed United States urban medical center.

Outcome measures

Patient demographics, past medical history, smoking habits, medica- tions, and gait assessment by ED physician and Neurology consult were recorded. Lipid panel and hemoglobin A1C were also collected if done at the time of ED presentation or within one year prior. Parameter compar- ison of those with acute cerebellar stroke (based on MRI) to the general (non-cerebellar stroke) population were made.

Statistics

Non-cerebellar stroke patients and cerebellar stroke patients were compared for demographics and stroke risk factors using chi-squared for categorical variables and t-test numerical variable as appropriate. Analyses were conducted using two-sided tests and a significance level of 0.05. All statistical analyses were

Fig. 1. Study design for isolated dizziness population.

conducted using Social Science Statistics web based calculators (http://www.socscistatistics.com/tests/).

Results

One hundred and seventy-four emergency department patients over 5 years with the chief complaint of dizziness or vertigo who were worked up with an acute brain MRI were evaluated. Thirty-eight pa- tients with localized central neurological findings on chief complaint or on the physical exam (assessed by ED physician or Neurology con- sult) were excluded from the study (Fig. 1). These patients were exclud- ed for cranial nerve palsy, dysarthria, aphasia, Extremity weakness, and

sensory deficit. Patients with abnormal gait findings were included in the study. Of the 136 ED patients who had a brain MRI for isolated diz- ziness, 103 (76%) had a Neurology consult placed. A gait exam was doc- umented by the ED in 49% (66/136 patients) of cases and in 87% (90/103 patients) of cases by Neurology consult. Of these, 66% (44/66 patients) were recorded as having a normal gait exam by the ED, and 44% (40/ 90 patients) had a normal gait exam documented by Neurology consult. There was a low correlation of gait assessment between ED and Neurol- ogy consult (49 patients, Spearman’s correlation r² = 0.17).

Based on MRI DWI sequence, 3.7% (5/136 patients) had acute cere- bellar strokes, limited to or including, the medial posterior inferior cer- ebellar artery vascular territory (Fig. 2A). Classic history of present

Fig. 2. Brain MRI diffusion-weighted imaging of cerebellar stroke patients (A) with associated “history of present illness” and official radiology MRI report (B). [4,9,13,14,16,17].

Fig. 2 (continued).

illness (HPI) statements associated with cerebellar stroke were noted in all 5 patients and official Radiology report confirms cerebellar stroke (Fig. 2B). Gait exam was documented as normal for 2 of these cerebellar stroke patients and not documented for 3 of them per the ED evaluator. Gait was documented as abnormal for 3 cerebellar stroke patients (one unable to tandem, two ataxic but able to walk without assistance) and as normal in 2 of the patients per Neurology consult. In the 5 cerebellar stroke patients, mean age, body mass index (BMI), hemoglobin A1c, gender distribution, and prevalence of hypertension were similar to the non-cerebellar stroke patient group (Table 1). Mean LDL/HDL ratio mean was 3.63 +- 0.80 among the cerebellar stroke patients, while

2.43 +- 0.79 in the general population (p b 0.001). Active smoking or N 30 pack years was 80% (4/5 patients) in the cerebellar stroke patients and 22% (25/112 patients) in the general population (p b 0.01). None of the cerebellar stroke patients were taking aspirin daily, while 26% (35/ 136 patients) of the general population were taking aspirin daily. Other anticoagulation or Antiplatelet agents were rare, with no patients taking clopidogrel and only two general population patients taking war- farin. No cerebellar stroke patients reported a previous stroke or tran- sient ischemic attack (TIA), while 12.5% (17/136 patients) of the general population reported previous stroke or TIA (Table 1).

Discussion

Common teaching suggests that in the absence of other signs and symptoms, cerebellar stroke is rare with dizziness as the lone present- ing complaint. So-called “isolated dizziness” implies a chief complaint of dizziness, but with normal exam and neurological review of systems. In the present study, urgent MRI was performed in 136 patients with isolated dizziness and 5 cerebellar strokes were detected. With millions of isolated dizzy patients presenting to the ED each year [1], a 3.7% rate of detection of cerebellar stroke is a very significant number. Of course, this is a biased study. ED attendings saw patients with isolated

dizziness, but wanted a brain MRI none-the-less. They often consulted Neurology, but 25% of the time, an MRI was ordered without Neurology involvement. Why?, stroke risk factors. The cross-section population of 136 patients, where ED attendings wanted a brain MRI, are patients with stroke risk factors. This is obvious selection bias, and likely subtle features of each individual patient also affected ED attendings selection bias to order an MRI (and conversely, medically ill patients with obvious dizziness causes were excluded). However, this selection bias is not nec- essarily a negative study aspect. Cerebellar stroke would be less likely in an 18 year old normotensive non-smoker.

Similar patient demographics and risk factors (Table 1) were seen in 53 patients with isolated dizziness and cerebellar stroke in the large (n = 1613) cross-sectional study by Kerber et al. [9] where cerebellar stroke patients were mean age 69.3 years, 72% hypertensive by medical history, and 23% smokers but had a higher frequency of previous CAD (40%) and TIA or stroke (28%) compared to our cohort (with 13.2% with CAD and 12.5% with previous TIA or stroke). Most importantly, in the Kerber et al. population, cerebellar stroke was only seen in 0.7% of patients with “isolated” dizziness; however, the diagnosis of cerebellar stroke was based on the neurologist’s final impression in the medical re- cord, brain MRI not being ordered in most patients [9]. This suggests that our figure of 3.7% (of patients with documentation of cerebellar in- farction in a cohort presenting to the ED with isolated dizziness) is more likely a reflection of reality, as the diagnosis was based on MRI findings in an acute setting of evaluation.

Most patients in our study had a hypertensive history due to selection bias, but risk factors of obesity or elevated hemoglobin A1c were not more common in cerebellar stroke patients (Table 1). All of the cerebellar stroke patients had a LDL/HDL ratio approximately 1 standard deviation above the mean, or higher (Table 1), and 4 of 5 were smokers. Further- more, none of the cerebellar stroke patients were taking aspirin daily, though 26% of the general population was. This is striking, as it suggests that hyperlipidemia and smoking are important risk factors, and aspirin

Table 1

Demographics and presentation assessment

All patients Cerebellar strokes

patients

Patients, n 136 5/136 (3.7%)

Mean age (years) 59.8 +- 16.7 n = 136 62.2 +- 12.0 n = 5

Sex 80 women (59%)

56 men (41%)

3 women, 2 men

Mean Body Mass Index 29.1 +- 5.6 (n = 113) 31.21 +- 4.8 (n = 5)

LDL (mg/dL) 114.0 +- 37.1 (n = 87) 153.2 +- 37.8 (n = 5)

HDL (mg/dL) 48.8 +- 12.8 43.8 +- 13.4

LDL/HDL ratio 2.43 +- 0.79 3.63 +- 0.80??

Hemoglobin A1c (%) 6.6 +- 1.33 (n = 66) 6.5 +- 1.51 (n = 5)

Hypertension (by history or ED values) 104 patients (76%) n = 131 4/5 patients (80%)

Coronary artery disease (by history) 18 patients (13.2%) 1/5 patients (20%)

Atrial fibrillation (by history) 4 patients (2.9%) 0/5 patients

Smokers (active or N 30 pack years) 25 patients (22%) n = 112 4/5 patients (80%)?

Previous stroke or TIA 17 patients (12.5%) n = 136 0/5 patients

Taking aspirin 35 patients (26%) n = 136 0/5 patients

Taking clopidogrel 0 patients (0%) n = 136 0/5 patients

Taking warfarin 2 patients (1.5%) n = 136 0/5 patients

Gait assessment by ED 44 normal, 23 abnormal, 69 not documented 2 normal, 3 not documented

Gait assessment by Neurology consult 40 normal, 50 abnormal, 13 not documented 2 normal, 3 abnormal

Body Mass Index, lipid panel, hemoglobin A1c were included if documented at presentation, or within 1 year prior. Smoking habits were included when documented at presentation.

t-test was used for continuous variable, Chi-squared test was applied on categorical variables.

Abbreviations: low density lipoprotein:high density lipoprotein ratio (HDL:LDL); Transient ischemic attack ; Emergency Department (ED).

* p value b 0.01.

?? p value b 0.001.

may be protective, for cerebellar stroke in isolated dizziness. However with only 5 cerebellar strokes, the subpopulation is a small sample.

Neurology’s HPI did not parse out cerebellar strokes any more clearly than the ED’s history. However, noting of key aspects of the HPI can make cerebellar stroke more likely (Fig. 2B). Difficulties with the pa- tient’s report and assessing the etiology of dizziness have been reported previously [13]. Abnormal gait was not considered an exclusion criteri- on in our study, as opinions differ on gait as a reliable indicator of poten- tial cerebellar stroke versus a peripheral mechanism of dizziness [8-10, 14]. More specifically, gait assessment was done in 49% of patients eval- uated by the ED physician, while Neurology consultants did 87%. In the patients where both the ED and Neurology did a gait exam, there was no correlation of their assessments (gait graded on a 4-point scale, see Methods), with a Spearman’s correlation of r² = 0.17. In the 5 patients that had cerebellar strokes, Neurology consultants documented other exam findings (a negative head impulse test or subtle nystagmus find- ings for example) that made cerebellar stroke more likely [10-12,14], but these were not consistently applied to all patients seen by Neurolo- gy. Seminal cerebellar stroke studies about vestibular head impulse test- ing and nystagmus were introduced in 2008-2009 (when our study data were being collected) [11,12], and these are now part of the stan- dard Neurological exam when cerebellar stroke is questioned. All 5 cer- ebellar stroke patients had strokes including, or limited to, the medial posterior inferior cerebellar artery (PICA) territory. Stroke to this vascu- lar territory have previously been associated with vertigo or dizziness in isolation [14]. Considering the above, with clinical presentation of iso- lated dizziness and significant stroke risk factors, acute brain MRI ap- pears prudent. Though most patients will rapidly recover from small cerebellar strokes [15], confirmation of a posterior circulation stroke changes workup and management, potentially preventing a future stroke event that could easily result in catastrophic brainstem infarct.

Conclusions

Isolated dizziness is a common presentation and patients with sig- nificant stroke factors (possibly focusing on hyperlipidemia and smoking habits) have a significant cerebellar stroke incidence. Gait as- sessment does not consistently stratify patients with greater stroke risk, so a low threshold for brain MRI seems reasonable even with a nor- mal exam.

Author contributions

Study concept and design: Michael Perloff, Nimesh Patel, Carlos Kase, Barbara Voetsch, Jose R. Romero.

Acquisition of data: Michael Perloff, Nimesh Patel, Anuja Oza. Analysis and interpretation of data: Michael Perloff, Nimesh Patel,

Carlos Kase, Anuja Oza, Barbara Voetsch, Jose R. Romero.

Study supervision: Michael Perloff.

Critical revision of manuscript for intellectual content: Michael Perloff, Nimesh Patel, Carlos Kase, Barbara Voetsch, Jose R. Romero.

Disclosures

Michael Perloff has provided expert testimony for legal cases on neurologic pain and opioid toxicity.

Carlos Kase has been on the Advisory Board at Boehringer-Ingelheim and had research support from Pfizer Astra-Zeneca Acorda Genentech and NIH.

Nimesh Patel, Barbara Voetsch, Anuja Oza and Jose R. Romero have no disclosures.

References

AJEM