This study is a retrospective case-control study conducted at a single comprehensive stroke center (Tan Tock Seng Hospital, Singapore), on consecutive hospitalized patients between 1 October 2015 to 31 March 2016. Inclusion criteria were the following: (1) isolated complete or incomplete hemisensory syndrome, regardless of the duration of symptoms; (2) brain MRI (magnetic resonance imaging) performed within the same admission. Exclusion criteria were the following: (1) bilateral sensory symptoms; (2) concomitant focal neurological symptoms; (3) signs of diplopia, dysarthria, dysphagia, focal weakness or ataxia; (4) contraindication to brain MRI.
All patients had a physical examination performed by a trained neurologist. This included an Available clinical information was retrieved through the hospital electronic medical records. Information collected included patient demographics, medical history, and details of the symptom presentation. The presence of traditional ischemic stroke risk factors, such as diabetes mellitus, hypertension, hyperlipidemia, smoking and atrial fibrillation, were also recorded. A previous study had also described migraine and psychiatric disorders (depression, anxiety disorder, schizophrenia) [2] as concurrent medical history in patient’s presenting with hemisensory syndrome, thus these were also included in our data collection.
Abnormal sensory symptoms were characterised into two categories: positive and negative [14]. Positive symptoms were characterised as presence of increased sensory symptoms, described as tingling (or pins and needles), pricking, tightening or burning. This may also be described as paraesthesia or dysesthesia [6]. Negative sensory symptoms were characterized by diminished or absent sensation, with abnormal findings on sensory examination. We also characterized the patients’ descriptions of their onset of symptoms and divided them into those with acute (≤24 hours) versus subacute (>24 hours) symptom onset. This was calculated from the time of symptom onset to the time of presentation to the emergency department. All clinical information was retrieved blinded to the neuroimaging findings.
Cases of confirmed acute ischemic stroke were defined as subjects whose MRI brain performed during the hospitalization showed a lesion demonstrating restricted diffusion. The infarct locations in the patients with restricted diffusion were recorded. Patients who had no restricted diffusion on MRI brain were classified as “non-stroke” (controls). The final diagnoses made by the attending physicians were recorded.
Imaging
All MRI studies were performed on one of the three MRI scanners in our center: a 3T (Tesla) MR scanner (Achieva; Philips Healthcare, Netherlands) using an 8 channel SENSE head coil, a 1.5T MR scanner (Ingenia; Philips Healthcare, Netherlands) using a 15 channel SENSE head coil, or a 1.5T MR scanner (Signa HDxt, General Electric Healthcare, Milwaukee, WI, USA) using an 8 channel brain coil. Diffusion weighted images (DWI) magnetic resonance images were acquired with the following parameters: repetition time (TR), 3155 milliseconds (ms); echo time (TE), 58.9 ms; flip angle, 90°; matrix, 124 × 124; section thickness, 4 millimetres (mm); 28 slices; axial acquisition; scan time, 1 minute (min) 50 seconds (s); b value, 1000s/mm2 on the Achieva; TR, 3455 ms; TE, 90.7 ms; flip angle, 90°; matrix, 112 × 109; section thickness, 5 mm; 28 slices; axial acquisition, scan time, 1 min 10 s; b value, 1000s/mm2 on the Ingenia; TR, 8000 ms; TE, 73.6 ms; flip angle, 90°; matrix, 128 × 128; section thickness, 5 mm; 28 slices; axial acquisition, scan time, 1 min 40 s; b value, 1000s/mm2 on the Signa HDxt. An independent neuroradiologist analyzed the images obtained according to our hospital protocol. Presence of restriction of the DWI with corresponding attenuation of apparent diffusion co-efficient (ADC) were considered as neuroimaging evidence of ischemic stroke.
MRI cervical spine studies (T2 sagittal, T1 sagittal, short tau inversion recovery (STIR) sagittal, T2 axial, gradient echo (GRE) axial) were also performed on one of the three MRI scanners: the 3T MR scanner (Achieva; Philips Healthcare, Netherlands) using an 18 channel neurovascular (NV) coil, the 1.5T MR scanner (Ingenia; Philips Healthcare, Netherlands) using a 20 channel NV coil, or the 1.5T MR scanner (Signa HDxt, General Electric Healthcare, Milwaukee, WI, USA) using an 8 channel NV array coil. T2 sagittal images were acquired with the following parameters: TR, 3000 ms; TE, 100 ms; flip angle, 90°; matrix, 308 × 315; section thickness, 3 mm; 12 slices; scan time, 2 min on the Achieva; TR, 3000 ms; TE, 110 ms; flip angle, 90°; matrix, 416 × 254; section thickness, 3 mm; 12 slices; scan time, 2 min 30 s on the Ingenia; TR, 3000 ms; TE, 100 ms; flip angle, 90°; matrix, 320 × 260; section thickness, 3 mm; 12 slices; scan time, 2 min 42 s on the Signa HDxt. T1 sagittal images were acquired with the following parameters: TR, 715 ms; TE, 10 ms; flip angle, 90°; matrix, 300 × 300; section thickness, 3 mm; 12 slices; scan time, 2 min 45 s on the Achieva; TR, 488 ms; TE, 10 ms; flip angle, 90°; matrix, 356 × 251; section thickness, 3 mm; 12 slices; scan time, 3 min 11 s on the Ingenia; TR, 400 ms; TE, 9 ms; flip angle, 90°; matrix, 320 × 224; section thickness, 3 mm; 12 slices; scan time, 2 min 7 s on the Signa HDxt. STIR sagittal images were acquired with the following parameters: TR, 3554 ms, inversion time (TI), 200 ms; TE, 70 ms; flip angle, 90°; matrix, 252 × 200; section thickness, 3 mm; 12 slices; scan time, 3 min 33 s on the Achieva; TR, 2500 ms, inversion time (TI), 150 ms; TE, 50 ms; flip angle, 90°; matrix, 256 × 211; section thickness, 3 mm; 12 slices; scan time, 2 min 45 s on the Ingenia; TR, 3000 ms, inversion time (TI), 150 ms; TE, 28 ms; flip angle, 90°; matrix, 256 × 192; section thickness, 3 mm; 12 slices; scan time, 3 min 42 s on the Signa HDxt. T2 axial images were acquired with the following parameters: TR, 3000 ms; TE, 100 ms; flip angle, 90°; matrix, 248 × 200; section thickness, 4 mm; 30 slices; scan time, 4 min 33 s on the Achieva; TR, 3587 ms; TE, 100 ms; flip angle, 90°; matrix, 228 × 195; section thickness, 4 mm; 30 slices; scan time, 3 min 57 s on the Ingenia; TR, 6720 ms; TE, 93 ms; flip angle, 90°; matrix, 320 × 192; section thickness, 4 mm; 30 slices; scan time, 4 min 36 s on the Signa HDxt. GRE axial images were acquired with the following parameters: TR, 29 ms; TE, 7 ms; flip angle, 7°; matrix, 208 × 208; section thickness, 3 mm; 35 slices; scan time, 3 min on the Achieva; TR, 594 ms; TE, 6 ms; flip angle, 25°; matrix, 252 × 250; section thickness, 4 mm; 30 slices; scan time, 4 min 8 s on the Ingenia; TR, 1070 ms; TE, 16 ms; flip angle, 25°; matrix, 256 × 192; section thickness, 4 mm; 30 slices; scan time, 3 min 34 s on the Signa HDxt.
Statistical analysis
Demographics and baseline clinical features were reported as mean ± standard deviation and frequency (percentage) for continuous and categorical variables, and compared between the two groups (stroke vs non-stroke) using independent two-sample t-test and Fisher’s exact test, respectively. Univariate logistic regression analysis was conducted to investigate the association between demographic and clinical parameters with stroke outcome, and odds ratio (OR) along with 95% confidence interval (95% CI) were reported. Multivariable logistic regression analysis was performed to adjust the association results for the selected variables via stepwise variable selection approach (including age, smoking, onset ≤24 hours, presence of positive symptoms). Statistical significance was set at p < 0.05. Data analysis was performed in SAS version 9.4 for Windows (SAS Institute Inc., Cary, NC, USA). The study was approved by the hospital institutional review board, and waiver of informed consent was obtained in view of the retrospective nature of the study.