Diffusion-Weighted and Gradient Echo Magnetic Resonance Findings of Hemichorea-Hemiballismus Associated With Diabetic Hyperglycemia: A Hyperviscosity Syndrome?

Background  The magnetic resonance (MR) imaging findings of hemichorea-hemiballismus (HCHB) associated with hyperglycemia are characterized by hyperintensities in the striatum on T1-weighted MR images and computed tomographic scans, with a mechanism of petechial hemorrhage considered to be responsible. Diffusion-weighted MR imaging (DWI) has been reported to detect early ischemic damage (cytotoxic edema) as bright areas of high signal intensity and vasogenic edema as areas of heterogeneous signal intensity. We report various DWI findings in 2 patients with hyperglycemic HCHB.

Objectives  To describe the DWI and gradient echo findings and characterize the types of edema in HCHB associated with hyperglycemia.

Setting  A tertiary referral center neurology department.

Design and Methods  Two patients with HCHB associated with hyperglycemia underwent DWI, gradient echo imaging, and conventional MR imaging with gadolinium enhancement. The patients had an elevated serum glucose level on admission and a long history of uncontrolled diabetes, and the symptoms were controlled by dopamine receptor blocking agents. Initial DWIs were obtained 5 to 20 days after symptom onset. Apparent diffusion coefficient (ADC) values were measured in the abnormal lesions with visual inspection of DWI and T2-weighted echo planar images.

Results  T1- and T2-weighted MR images and brain computed tomographic scans showed high signal intensities in the right head of the caudate nucleus and the putamen. Gradient echo images were normal. The DWIs showed bright high signal intensity in the corresponding lesions (patient 1), and the ADC values were decreased. The decrease in ADC and the high signal intensity on DWI persisted despite the disappearance of HCHB, even after 70 days.

Conclusions  Gradient echo MR imaging findings were normal in HCHB with hyperglycemia, whereas DWI and the ADC map showed restricted diffusion, which suggests that hyperviscosity, not petechial hemorrhage, with cytotoxic edema can cause the observed MR abnormalities.

HEMICHOREA-HEMIBALLISMUS (HCHB) is usually continuous, nonpatterned, involuntary, and associated with stroke, and infrequently associated with infections, drug usage, metabolic derangement, neurodegenerative disorder, tumors, and hyperglycemia.¹ A few cases of HCHB with hyperglycemia have been reported.^2-9 Since the significant association between hyperglycemia and striatal hyperintensity on T1-weighted magnetic resonance (MR) images in HCHB was pointed out, much attention has been focused on the mechanism of HCHB, and petechial hemorrhage with blood-brain barrier breakdown in the striatum has been suggested as the most plausible mechanism.^2-4

Two cases of HCHB with hyperglycemia and MR imaging findings, including diffusion-weighted imaging (DWI) and gradient echo imaging (GE), are reported herein. Our findings suggest that mechanisms other than the petechial hemorrhage do in fact exist.

A 69-year-old woman was admitted for sudden onset of abnormal movement on the left side. Three days before admission, hemichorea had developed suddenly, and the intensity had aggravated during these 3 days. She had no hemiparesis, sensory deficits, or any other neurologic deficits. She had a history of diabetes mellitus for 7 years and took an oral hypoglycemic agent irregularly. She had no history of hypertension, headache, parkinsonism, or other neurologic diseases. At admission, the initial blood glucose level was 348 mg/dL (19.3 mmol/L) and hemoglobin A_1c concentration was 9.5%. Neurologic examination showed normal findings except for the hemichorea in the left side. Brain computed tomography was performed on the first hospital day (Figure 1A). Brain MR imaging, including DWI, GE, and T1- and T2-weighted images with gadolinium enhancement, was performed on the second hospital day (Figure 1B-F). The chorea was controlled with risperidone (4 mg/d) after 14 days. Serum glucose level was maintained within the therapeutic ranges by insulin. The patient was discharged with no specific symptoms on the 30th hospital day. Follow-up GE and T1- and T2-weighted MR imaging was performed on the 14th hospital day and 70 days after onset (not shown). On the 120th day after onset, risperidone was discontinued, and chorea had not reappeared by the 190th day.

A 62-year-old woman was admitted for sudden onset of choreatic movement on the left side. Twenty days before admission, hemichorea had developed suddenly. The intensity of the symptoms had increased during this period. She had had poorly controlled diabetes mellitus for 2 years. The hemoglobin A_1c concentration was 9.7%, and initial blood glucose level was 370 mg/dL (20.5 mmol/L) at admission. Neurologic examination showed normal findings except for the left hemichorea. Brain MR imaging, including DWI, GE, and T1- and T2-weighted images, was performed on the 22nd day after the onset of symptoms (Figure 2). Serum glucose level was maintained within the therapeutic ranges by insulin. Hemichorea was controlled with oral haloperidol (4 mg/d), and the patient was discharged on the 10th hospital day with no specific symptoms. Haloperidol was stopped on the 70th day after onset, and chorea had not reappeared by the 120th day after onset.

The patients were examined by the same MR imaging protocol used in our previous studies.^10-13 The patients were examined with a 1.5-T MR unit (Signa Horizon, Echospeed; General Electric Medical Systems, Milwaukee, Wis) with echo planar imaging capability. Fast spin-echo, T2-weighted images (repetition time/echo time, 4200/112 milliseconds; field of view, 21×21 cm; matrix, 256×192; and slice thickness, 5 mm with 1.5-mm gap) and T2*-weighted GE sequences (repetition time/echo time, 200-500/15 milliseconds; flip angle, 20°; number of excitations, 2; slice thickness, 1.4 mm; gap width, 0.7 mm) were obtained. The DWIs were obtained in the transverse plane by means of a single-shot echo planar image (repetition time/echo time, 6500/125 milliseconds; field of view, 24×24 cm; matrix, 128×128; slice thickness, 5 mm with 2.5-mm gap; and 2 b values, 0 and 1000 s/mm²). The diffusion gradients were applied along 3 axes (x, y, z) simultaneously. The apparent diffusion coefficient (ADC) was calculated on the basis of the Stejskal-Tanner equation as the negative slope of the linear regression line best fitting the points for b vs ln(SI), where ln represents natural log and SI is the signal intensity from the region of interest within the images acquired at each b value. Performing this calculation on a pixel-by-pixel basis created ADC maps. Normal ADC values of the parenchyma and white matter range from 0.78×10⁻³ to 0.91×10⁻³ mm²/s (K.C., D.-W.K., unpublished data, 2000).^11-13 The ADC values are uniform throughout the brain and not intrinsically lower in specific areas. Regions of interest were carefully drawn in the abnormal areas on calculated average ADC maps, as well as in normal-appearing areas. The region of interest (25 mm²) was selected with the help of T2-weighted echo planar images of the same acquisition as the diffusion images (ie, images generated from the diffusion sequence with diffusion sensitivity b = 0) to avoid errors in selection of region of interest due to spatial distortion problems causing discrepancies between diffusion images and conventional MR images.

The brain MR image of patient 1, performed on the fourth day after onset, showed high signal intensities in the right head of the caudate nucleus, the globus pallidus, and the putamen on T1-weighted images, which spared the anterior limb of the internal capsule, and focal high signal intensity on T2-weighted MR images and DWI (Figure 1B, Figure 1C, Figure 1E). Gadolinium-enhanced T1-weighted images showed no abnormal enhancement, and GE also displayed no abnormal findings (Figure 1D). The ADC map presented low signal intensities and low ADC values (419 × 10⁻⁶ to 432 × 10⁻⁶ mm²/s) in the corresponding lesions (Figure 1F). The ADC values of the surrounding lesions were 519 × 10⁻⁶ to 597 × 10⁻⁶ mm²/s, while those of the contralateral side were normal (812 × 10⁻⁶ to 896 × 10⁻⁶ mm²/s). Brain computed tomography, performed on the third day after onset, showed high-attenuated areas in the same lesions (Figure 1A). Follow-up MR imaging, performed 17 and 70 days after onset, showed similar findings with no significant interval change of ADC values (not shown).

The brain MR images of patient 2, performed on the 21st day after onset of symptoms, showed high signal intensities in the right striatum on T1-weighted MR images (Figure 2A). The T2-weighted MR images showed low to mild high signal intensities in the corresponding lesions (not shown). The DWI and GE presented normal findings (Figure 2B and Figure 2C), and the ADC map showed decreased ADC values in the corresponding lesions (650 × 10⁻⁶ to 695 × 10⁻⁶ mm²/s) compared with the contralateral side (854 × 10⁻⁶ to 912 × 10⁻⁶ mm²/s).

Our 2 patients had hemichorea associated with hyperglycemia. The DWI findings showed restricted diffusion in both patients and bright high signal intensity on DWI in patient 1. There was no evidence of petechial hemorrhage on GE images. The DWI and ADC map of patient 1 displayed low ADC values in surrounding areas as well as T1 hyperintensity lesions. The ADC values were as low as in acute arterial ischemia, being the lowest (419 × 10⁻⁶ to 432 × 10⁻⁶ mm²/s) in the focal high-signal-intensity lesion on T1- and T2-weighted MR images and DWI. The surrounding areas showed T1–high density, T2-isodense, and DWI-isodense lesions with low ADC values (519 × 10⁻⁶ to 597 × 10⁻⁶ mm²/s). The brain MR image of patient 2 displayed high signal intensities on T1-weighted images with normal DWI and GE findings. The ADC values of the corresponding area were slightly decreased. The time from onset to MR imaging was variable, ranging from 4 to 21 days, which implied that different MR imaging stages could occur.

The critical pathophysiology of HCHB associated with hyperglycemia is unknown. Hyperglycemia can disrupt the blood-brain barrier and produce a global decrease in regional cerebral blood flow, intracellular acidosis, accumulation of extracellular glutamate, brain edema formation, and decreased activity of γ-aminobutyric acid–enkephalin inhibitory neurons.^4,14,15 The decreased γ-aminobutyric acid activity due to its metabolism as an alternate energy substrate during hyperglycemic crisis has been proposed. The fact that the chorea may persist well beyond the episode of hyperglycemia argues against this mechanism.⁹ In addition, the MR imaging abnormalities, including high signal intensities on computed tomography and T1-weighted MR imaging, cannot be explained. The nature of the characteristic MR signal changes associated with HCHB has been the subject of considerable controversy. Petechial hemorrhage,^2-4 myelinolysis,⁵ and calcifications⁶ have been suggested as possible mechanisms. Petechial hemorrhage with blood-brain barrier breakdown in the striatum has been suggested as the most plausible mechanism.

However, the GE findings presented here suggest that petechial hemorrhage cannot be responsible for the lesions. Recently, a salient autopsy report of HCHB was published.¹⁶ The main findings included multiple infarcts associated with reactive astrocytic and interneuronal response, not with petechial hemorrhage and calcification.¹⁶ Shan et al⁷ hypothesized that the possible cause of the MR imaging abnormalities might be the mild ischemia with gemistocyte accumulation, and they described one patient whose biopsy specimen showed gliotic brain tissue with abundant gemistocytes.⁷ Gemistocytes are swollen reactive astrocytes, containing a rich protein content, that usually appear during acute injury but later gradually shrink. Shortening of T1 relaxation time can result from the protein hydration layer inside the cytoplasm of swollen gemistocytes, as in a reported case of gemistocytic astrocytoma.^17,18 The T1 relaxation time depends on the movement of molecules, and the rich protein content causes electrostatic forces that restrict the motion of water molecules, which may explain the DWI abnormalities of patient 1 in the surrounding lesions with low ADC values. The core lesion on DWI might result from transient, mild ischemia due to high viscosity and decreased perfusion. The MR image of patient 2 showed the subacute findings of HCHB. The DWI and T2-weighted MR image began to be normalized, but the abnormalities on T1-weighted MR images remained. The ADC values of patient 2 were slightly lower despite the normal DWI, and T2-weighted MR images showed low to normal signal intensities in the corresponding lesions. In DWI, high-amplitude bipolar gradients sensitize a T2-weighted MR image to the brownian motion of water molecules. Despite the small magnitude of motion during diffusion gradient, it results in intravoxel dephasing and signal loss.¹⁹ Signal intensity on DWI is thus due to 2 intrinsic competing components: degree of T2 signal intensity and degree of diffusion. Provenzale et al¹⁹ have shown normal DWI findings despite significant vasogenic edema in posterior reversible encephalopathy syndrome. They term this phenomenon T2 washout, since the intravoxel dephasing related to the increased water diffusion in the vasogenic edema washes out the inherent increased T2 signal intensity in the lesions.^19,20 A T2 shortening in HCHB has been well known with T1 shortening.^8,9 The T2 washout effect can be applied in the normal DWI findings of patient 2, since decreased water diffusion in the hyperviscous striatum may wash out the decreased T2 signal intensity in the corresponding lesions.

The restricted diffusion of water (low ADC values) has been reported in various diseases, such as acute stroke,¹⁰ Wernicke encephalopathy,¹³ epidermoid mass,¹⁸ brain abscess,²¹ and status epilepticus.¹¹ The proposed mechanisms of ADC decrease, which are still being discussed, address changes in the diffusion characteristics of the intracellular and extracellular water compartments, including restricted diffusion, water exchange across permeable boundaries, the concept of extracellular tortuosity, and the intracellular and extracellular volume fraction.²² In addition, while the initial triggering factors leading to restricted diffusion may vary according to the main conditions, the subsequent results may be similar, leading to cell death. In arterial ischemia,¹⁰ the cessation of blood flow can cause initiation of the ischemic cascade of cell death. In status epilepticus,¹¹ the primary mechanisms are neuronal hyperexcitability and the excessive release of excitatory amino acids, such as glutamate. In brain abscess and high-cellularity tumor, hyperviscosity can be the probable mechanism of restricted diffusion.^18,20 Shan and coworkers' results⁷ and our DWI findings suggest that hyperviscosity can be responsible for the restricted diffusion in HCHB caused by hyperglycemia, which can cause the partial neuronal death and dysfunction on the vulnerable striatum, similar to the partial ischemic injury model,²³ in predisposed individuals.

The MR imaging abnormalities in HCHB are often reversible, but the mechanisms of such abnormalities are unknown. We suggest that the ADC threshold values might play a role in determining the "tissue fate." Dardzinski et al²⁴ reported the ADC changes over time after permanent occlusion of the middle cerebral artery and suggested the following ranges of ADC values: (1) less than 450 × 10⁻⁶ mm²/s: severe ischemia and irreversible damage occur; (2) greater than 550 × 10⁻⁶ mm²/s: infarction will not occur; (3) 450 × 10⁻⁶ to 550 × 10⁻⁶ mm²/s: potentially reversible. Our results (ADC values of patient 1: core, 419 × 10⁻⁶ to 432 × 10⁻⁶ mm²/s; surrounding areas, 519 × 10⁻⁶ to 597 × 10⁻⁶ mm²/s; patient 2: 650 × 10⁻⁶ to 695 × 10⁻⁶ mm²/s) corresponded well to the previous reports and hence suggest that, with adequate treatment, the DWI abnormalities (cytotoxic edema) might be reversible, as with the cells in the ischemic penumbra.¹³

The DWI and GE MR imaging findings in HCHB associated with hyperglycemia suggest that gemistocyte accumulation, hyperviscosity, neuronal dysfunction, and possible cytotoxic edema can cause the observed MR imaging abnormalities in possible susceptible individuals. Diffusion-weighted imaging can be used in determining the tissue fate with the analysis of ADC values.

Accepted for publication November 6, 2001.

Author contributions:Study concept and design (Drs Chu, Kang, Kim, Park, and Roh); acquisition of data (Drs Chu, Kang, and Roh); analysis and interpretation of data (Drs Chu, Kang, Kim, Park, and Roh); drafting of the manuscript (Drs Chu, Kang, and Roh); critical revision of the manuscript for important intellectual content (Drs Chu, Kang, Kim, Park, and Roh); administrative, technical, and material support (Dr Chu); study supervision (Drs Kang, Kim, Park, and Roh).

Corresponding author and reprints: Jae-Kyu Roh, MD, PhD, Department of Neurology, Seoul National University Hospital, 28, Yongon-Dong, Chongro-Gu, Seoul 110-744, South Korea (e-mail: rohjk@snu.ac.kr).