Diabetic striatopathy presenting as stroke-mimicking hemiparesis without involuntary movements: A case report

Abstract

Diabetic striatopathy is a rare complication of uncontrolled diabetes, typically presenting with involuntary movements, striatal hyperdensity on computed tomography, and T1-weighted hyperintensity on magnetic resonance imaging. We describe a 52-year-old woman with hyperosmolar hyperglycemic state (glucose 34.0 mmol/L; osmolality 323 mOsm/kg) who developed left-sided weakness without chorea or ballism. Brain computed tomography showed subtle hypodensity of the right caudate nucleus and putamen, closely mimicking acute ischemic stroke; densitometry demonstrated a consistent side-to-side difference (5.89 Hounsfield units, P < .001). T1-weighted imaging showed concordant hyperintensity in the same region (mean per-pair signal-intensity ratio 1.117, P < .001). The weakness resolved within 1 week of glycemic optimization. Follow-up imaging at 12 months showed normalization of the signal abnormality (mean per-pair signal-intensity ratio 0.979, near unity). To our knowledge, the combination of movement-free, stroke-mimicking presentation with nonhemorrhagic striatal hypodensity showing no evidence of acute infarction has not been previously reported. Whether this reflects the timing of imaging relative to metabolic correction rather than a distinct radiologic entity remains to be established.

Graphical abstract

Introduction

Diabetic striatopathy (DS) is a rare neurological complication of poorly controlled diabetes mellitus. Oh et al. [1] described the entity in 2002 in a meta-analysis of 53 cases that identified a seemingly consistent triad: hemichorea or hemiballism, striatal hyperdensity on noncontrast computed tomography (CT), and T1-weighted hyperintensity on magnetic resonance imaging (MRI). Most patients have been elderly women, and although early series came predominantly from East Asian populations, subsequent reports have made clear that no ethnicity is exempt [2,3].

That triad, however, is less reliable than it first appeared. Gómez-Ochoa et al. [2] found that chorea-ballism was present in only about 40% of reported cases; the remainder presented with focal weakness, confusion, or deficits easily mistaken for stroke [[3], [4], [5]]. CT, too, can be entirely normal even when MRI is clearly abnormal; this dissociation is documented in about one-sixth of cases [[1], [2], [3]]. The pathophysiology behind all of this remains unsettled. Leading hypotheses invoke reactive gemistocytic astrocyte swelling, petechial hemorrhage, or protein-rich edema within the striatum, but none of these mechanisms has been definitively proven in living patients [3,6,7].

The patient we describe presented with striatal CT hypodensity, hemiparesis, and no involuntary movements, a clinical and radiological picture essentially indistinguishable from ischemic stroke.

Case report

A 52-year-old woman with a history of poorly controlled type 2 diabetes was admitted to the intensive care unit with hyperosmolar hyperglycemic state (HHS) and coma. Laboratory findings at presentation included serum glucose of 612 mg/dL (34.0 mmol/L), glycated hemoglobin (HbA1c) of 65 mmol/mol (8.1% NGSP), measured serum osmolality of 323 mOsm/kg, serum sodium of 132 mmol/L, blood urea nitrogen (BUN) of 31 mg/dL (11.1 mmol/L), and absent urinary ketone bodies, consistent with HHS.

A noncontrast brain CT obtained on the day of admission was unremarkable. The patient was treated with intravenous fluids and insulin and progressively recovered consciousness over several days.

On hospital day 7, she developed new left-sided weakness affecting both arm and leg, graded 3/5 on the Medical Research Council scale. Examination was otherwise unremarkable: no sensory changes, no cranial nerve deficits, no alteration of consciousness. Hemichorea-hemiballism, the movement disorder most closely associated with DS, was absent at this time and remained so throughout her admission. A repeat noncontrast CT was performed urgently to rule out a cerebrovascular event. The scan revealed subtle hypodensity of the right caudate nucleus and putamen (Fig. 1), interpreted as possibly ischemic in origin.

Fig. 1.

Noncontrast CT of the brain, axial plane, hospital day 7. Subtle hypodensity of the right caudate nucleus and putamen compared to the contralateral side (arrows). The boxed region is shown magnified in the inset, where the same arrows indicate the hypodense striatum. Quantitative ROI analysis demonstrated a consistent inter-hemispheric density difference (right 29.01 ± 2.31 HU vs left 34.91 ± 1.10 HU; Δ = 5.89 HU, P < .001).

Retrospective quantitative densitometric analysis was performed by placing bilateral mirror-image regions of interest (ROIs) at 6 matched locations within the caudate nucleus and putamen (Table 1). The right striatum measured 29.01 ± 2.31 Hounsfield units (HU), compared to 34.91 ± 1.10 HU on the left (mean difference 5.89 ± 1.40 HU; paired t-test: t = 10.34, df = 5, P < .001), indicating that the density asymmetry was unlikely to reflect measurement variability.

Table 1.

Quantitative CT densitometric analysis (day 7, noncontrast CT).

ROI	Region	Right (HU)	Area (mm2)	Left (HU)	Area (mm2)	Δ (HU)
ROI 1	Caudate	25.76 ± 4.70	14.9	33.85 ± 2.89	14.2	8.09
ROI 2	Putamen	27.39 ± 4.50	33.6	33.56 ± 4.54	33.1	6.17
ROI 3	Striatum	30.37 ± 3.83	64.7	36.17 ± 3.44	61.8	5.80
ROI 4	Striatum	32.26 ± 3.46	78.1	36.01 ± 3.11	75.1	3.75
ROI 5	Striatum	29.86 ± 4.27	112.3	35.31 ± 3.68	96.8	5.45
ROI 6	Putamen	28.43 ± 4.21	28.1	34.53 ± 4.05	27.6	6.10

Bilateral mirror-image ROIs at 6 matched anatomical locations. Values are mean HU ± SD.

Mean right: 29.01 ± 2.31 HU. Mean left: 34.91 ± 1.10 HU. Mean Δ: 5.89 ± 1.40 HU. Paired t-test: t = 10.34, df = 5, P < .001. Δ = left minus right.

A brain MRI was subsequently obtained on a 1.5T scanner (Siemens Healthineers, Erlangen, Germany). T1-weighted MPRAGE images (TR/TE/TI 2250/3/900 ms; 1.0 mm isotropic voxels) demonstrated hyperintensity of the right caudate nucleus and putamen, spatially concordant with the CT hypodensity (Fig. 2). The distribution of the T1 hyperintensity matched that of the CT hypodensity in both the caudate nucleus and the putamen, with sparing of the globus pallidus and thalamus. Quantitative ROI analysis with 19 bilateral pairs (16.6 mm² each) placed across 5 axial slices confirmed significant signal asymmetry: the right striatum measured 314.3 ± 24.4 vs 282.0 ± 17.2 arbitrary units (a.u.) on the left, yielding a signal intensity ratio (SIR) of 1.117 (paired t-test: t [18] = 5.20, P < .001; Cohen’s d = 1.19; Table 2). T2-weighted FLAIR showed only a chronic linear hyperintensity in the left external capsule and minor periventricular white matter disease. Diffusion-weighted imaging (b = 1000 s/mm²) and ADC maps (Fig. 2B and C), obtained approximately 1 week after symptom onset, showed no restricted diffusion in the right basal ganglia, and there was no pathological gadolinium enhancement. Although diffusion restriction can pseudonormalize in the late subacute phase, the absence of restriction, the complete clinical and imaging reversibility, and the concordant T1 hyperintensity, is most consistent with the absence of acute infarction.

Fig. 2.

Initial brain MRI (1.5T, Siemens Healthineers), approximately 1 week after symptom onset (radiological convention; image left = patient right). (A) Axial T1-weighted MPRAGE: hyperintensity of the right caudate nucleus and putamen (arrows), spatially concordant with the CT hypodensity in Fig. 1. (B) Diffusion-weighted imaging (b = 1000 s/mm²) and (C) the corresponding ADC map at the same level show no restricted diffusion in the right striatum (arrows), consistent with the absence of acute infarction. (D) Susceptibility-weighted imaging at the same level shows no blooming compatible with hemorrhage in the right striatum (arrows). Quantitative signal-intensity analysis was performed on the T1-weighted MPRAGE, with a mean per-pair signal-intensity ratio (SIR) of 1.117 (P < .001; Table 2). These findings were not identified in the official radiology report.

Table 2.

Quantitative T1-weighted MRI signal intensity analysis.

Timepoint	Pairs (n)	Right (a.u.)	Left (a.u.)	SIR	P-value
Initial (1.5T, MPRAGE)	19	314.3 ± 24.4	282.0 ± 17.2	1.117	<.001
Follow-up (3T, TFE)	4	678.9 ± 34.6	701.4 ± 72.8	0.979	.70

SIR = mean of the per-pair right/left signal-intensity ratios. Values are mean signal intensity (a.u.) ± SD.

Initial: 1.5T Siemens Healthineers, T1-weighted MPRAGE (TR/TE/TI 2250/3/900 ms, 1.0 mm isotropic), ROI area 16.6 mm2. Follow-up: 3T Philips Healthcare, T1-weighted TFE (TR/TE 7.4/3.4 ms, 0.56 × 1.1 × 0.56 mm), ROI area 15.1 mm2. Cohen’s d = 1.19 for initial study. Follow-up analysis restricted to 2 of 4 acquired slices (lower slices excluded for partial volume effects).

Taken together, acute HHS with coma, contralateral motor deficit, unilateral striatal T1 hyperintensity concordant with CT hypodensity, no restricted diffusion, no pathological enhancement, the picture was most consistent with DS, following the diagnostic framework of Arecco et al. [3]. The diagnosis was made retrospectively by the authors. Glycemic control was optimized, and the left-sided weakness resolved completely within approximately 1 week.

The CT hypodensity was independently reviewed by a board-certified neuroradiologist (10 years of experience), who assessed and excluded alternative explanations. Early ischemic change was considered unlikely, as the abnormality resolved completely on clinical and imaging follow-up, a course incompatible with an established infarct. Nonspecific edema was excluded because simple edema is T1-hypointense, whereas the lesion was T1-hyperintense, indicating a T1-shortening paramagnetic substrate rather than free-water accumulation. Partial-volume averaging was excluded because the ROIs were placed within the structural core, away from tissue boundaries, and the CT finding colocalized with an independently acquired MRI T1 hyperintensity that subsequently normalized, a cross-modality, longitudinal concordance that an intra-slice acquisition artifact cannot reproduce. Hemorrhage was excluded on susceptibility-weighted imaging from the same 1.5T study, which showed no susceptibility foci compatible with hemorrhage in the right striatum (Fig. 2D).

Twelve months later, the patient was re-imaged for an unrelated clinical indication. By this time her diabetes was well controlled on oral hypoglycemic therapy, with HbA1c and fasting glucose both within normal limits. A follow-up brain MRI was performed on a 3T scanner (Philips Healthcare, Best, The Netherlands). T1-weighted TFE images (TR/TE 7.4/3.4 ms; 0.56 × 1.1 × 0.56 mm voxels) showed complete resolution of the right striatal hyperintensity, with symmetric basal ganglia (Fig. 3). ROI analysis on 2 well-resolved axial slices (4 bilateral pairs, 15.1 mm² each) showed a near-unity signal-intensity ratio (SIR 0.979), consistent with normalization; with only 4 pairs the analysis was not powered for hypothesis testing, so the accompanying P value (.70) is descriptive (Table 2). T2-weighted 3D FLAIR (TR/TE/TI 6000/340/2000 ms) showed only the previously known chronic left external capsular finding and mild periventricular white matter disease (Fig. 4).

Fig. 3.

Follow-up brain MRI at 12 months, T1-weighted TFE sequence (3T Philips Healthcare). (A) Axial and (B) coronal images demonstrating complete resolution of the right striatal hyperintensity, with symmetric basal ganglia. Quantitative analysis showed a near-unity signal-intensity ratio (SIR 0.979), consistent with normalization (Table 2). The patient had achieved sustained glycemic control in the interval.

Fig. 4.

Follow-up brain MRI, T2-weighted 3D FLAIR (TR/TE/TI 6000/340/2000 ms). (A) Axial and (B) coronal images showing only chronic changes: a linear hyperintensity in the left external capsule and minor periventricular white matter disease. No acute pathology in the right basal ganglia.

All ROI measurements were performed by a single observer under neuroradiologist supervision and verified in a joint reading session; ROIs were placed within the structural core of each nucleus to avoid partial-volume contamination. Their number reflected the in-plane resolution and the dimensions of the target nuclei, representing the maximum set of nonoverlapping samples obtainable without resampling identical voxels; on follow-up, only 4 pairs across 2 slices satisfied these criteria. Because the clinical syndrome was strictly lateralized, the contralateral basal ganglia served as an internal, within-patient reference, and affected-side values were normalized to the homologous contralateral region (mirror ROIs) to control for inter-individual and acquisition-related variability.

For the quantitative analysis, affected-side ROIs were compared with their homologous contralateral (mirror) ROIs using a 2-tailed paired t-test; values are reported as mean ± standard deviation and a 2-sided P < .05 was considered statistically significant. The standardized effect size (Cohen’s d) was computed for the initial MRI comparison. Six mirror pairs were analyzed on CT and 19 on the initial MRI; on the 12-month MRI, 4 pairs across 2 slices met the placement criteria. Because the CT ROIs sampled a single contiguous lesion, they are not fully spatially independent; the interhemispheric difference therefore rests on the magnitude and consistency of the asymmetry rather than on the P-value alone. Statistical analyses were performed with GraphPad Prism version 11.0.0 (GraphPad Software, San Diego, CA).

Discussion

The central finding is that the affected striatum was hypodense on CT, not hyperdense. In cases of DS without a coexisting vascular or hemorrhagic cause, abnormal CT has been described as hyperdense [[1], [2], [3], [4],6]. Hyperdensity, although unusual, can prompt consideration of entities beyond ischemia. Hypodensity does not: it resembles stroke and is likely to be interpreted as such (right striatum 29.01 vs 34.91 HU). The absence of chorea or ballism, with isolated hemiparesis, reproduces the presentation of a lacunar infarct. Stroke-mimicking symptoms and stroke-mimicking imaging together constitute a diagnostic pitfall that requires a high index of suspicion to recognize.

Pathophysiology of CT hypodensity in DS

Why would DS produce hypodensity rather than the expected hyperdensity? One possibility is timing. Our CT was obtained on day 7, after the acute metabolic crisis had already been partially treated. If the classic hyperdensity depends on ongoing hyperglycemia-driven processes (gemistocytic swelling and proteinaceous edema), imaging after partial correction may capture a transitional window in which tissue damage persists but the protein or blood-product concentration is not yet high enough to raise CT attenuation. MRI a week later already showed T1 hyperintensity (SIR 1.117) and is more sensitive than CT to such T1-shortening agents, which may account for the intermodality discrepancy. In our patient, susceptibility-weighted imaging from the index study showed no susceptibility foci compatible with hemorrhage in the affected striatum (Fig. 2D), providing direct evidence of a nonhemorrhagic substrate. This is concordant with the only DS autopsy study, which reported gemistocytic astrocyte proliferation without hemorrhage or calcification [7], and with prior susceptibility-weighted imaging in the hyperglycemic setting [8]. Complete normalization of the T1 signal at 12 months under glycemic control is consistent with a reversible metabolic process rather than permanent structural damage.

Striatal low attenuation on CT has occasionally been reported in the hyperglycemic setting, but under circumstances distinct from the present case. Two reports documented putaminal hypodensity in patients with a demonstrable vascular cause, Moyamoya disease and concurrent acute ischemic stroke, so that the attenuation change could not be ascribed to striatopathy alone [9,10]. A third described putaminal hypodensity in a hyperosmolar hyperglycemic state, the closest precedent to our case, but with hemichorea and an associated putaminal microhemorrhage [11]. In our patient, the absence of restricted diffusion argued against acute infarction, no hemorrhagic component was identified, and the presentation was an isolated hemiparesis without involuntary movements, a combination that distinguishes the present case from these reports.

Quantitative imaging analysis

Most published cases of DS rely on qualitative imaging assessment; signal changes are described visually but not measured. In this case the quantitative measurements served to corroborate the qualitative imaging pattern rather than as a stand-alone diagnostic criterion. The SIR approach, normalizing each measurement to the contralateral hemisphere within the same scan, has a practical advantage: it is robust to differences in scanner hardware, field strength, and acquisition parameters, which allowed us to compare the initial 1.5T study with the 3T follow-up despite the change in equipment. The drop from SIR 1.117 (P < .001) to a near-unity value of 0.979 documents the resolution of the signal abnormality; on the underpowered 4-pair follow-up the associated P value is descriptive only.

Differential diagnosis

Unilateral T1 hyperintensity in the basal ganglia narrows the differential considerably. Acquired hepatocerebral degeneration affects the globus pallidus bilaterally and symmetrically, reflecting manganese deposition in the context of liver failure, none of which applied here [4]. Ischemic infarction can show T1 shortening in the subacute phase, but diffusion restriction was absent at imaging performed in the late subacute window, and it resolved completely, features difficult to reconcile with infarction. Osmotic demyelination seemed unlikely: her hyponatremia was mild (132 mmol/L) and was corrected carefully. Wernicke encephalopathy and methanol intoxication were excluded on clinical grounds and by the lesion distribution, which did not match the characteristic bilateral symmetric patterns of either condition [4,12,13]. Calcium deposition (e.g., hypoparathyroidism, Fahr’s disease, and tuberous sclerosis) can appear T1-hyperintense, but no calcification was detected on CT [12,13].

Limitations

This is a single case. The CT hypodensity pattern we describe needs to be confirmed in other patients before it can be considered a recognized variant of DS, and it remains entirely possible that it reflects something specific to this patient’s metabolic trajectory rather than a generalizable phenomenon. Histopathological proof is unavailable (biopsy is contraindicated when the clinical course is favorable [4]), and so our pathophysiological interpretation remains speculative. The 2 MRI studies were acquired on different scanners and field strengths (1.5T Siemens vs 3T Philips) with different T1 sequences (MPRAGE vs TFE), which precludes direct comparison of absolute signal values, though the SIR approach was specifically chosen to mitigate this limitation. Finally, the follow-up quantitative analysis was restricted to 4 ROI pairs from 2 axial slices because lower slices did not permit selective placement on the basal ganglia; this limits statistical power, although the result is consistent with both the visual impression and the clinical outcome. ROI measurements were performed by a single observer under neuroradiologist supervision without a formal interobserver agreement statistic; the within-scan mirror-ROI design and the concordance with the qualitative read mitigate but do not replace such an analysis.

A literature search (PubMed and Scopus, last updated June 6, 2026) combining diabetic striatopathy and non-ketotic hyperglycemic hemichorea–hemiballism with computed-tomography and basal-ganglia terms did not identify a prior case of non-hemorrhagic striatal CT hypodensity occurring in the absence of involuntary movements. The 3 comparable reports of acute striatal CT low attenuation in the hyperglycemic setting [[9], [10], [11]] all involved involuntary movements and, where a substrate was identifiable, a vascular or hemorrhagic process. We therefore describe this combination as previously unreported to our knowledge, while acknowledging that movement-free presentations may be under-recognized.

Conclusion

We describe a case of DS presenting with striatal CT hypodensity, rather than the expected hyperdensity, in the absence of involuntary movements, without a hemorrhagic substrate and with no evidence of acute infarction, a combination we did not find previously reported. Quantitative analysis on both CT and MRI confirmed significant asymmetries that resolved completely under glycemic control. The practical message is that DS can look like ischemic stroke on CT, and radiologists should keep this entity in mind when evaluating diabetic patients with unexplained basal ganglia changes, even, or perhaps especially, when the density is low rather than high.

Declaration of generative AI and AI-assisted technologies in the manuscript preparation process

During the preparation of this work the author(s) used Claude (Anthropic) to assist with statistical analysis verification and language editing. After using this tool, the author(s) reviewed and edited the content as needed and take(s) full responsibility for the content of the published article.

Patient consent

Written informed consent was obtained from the patient for the publication of this case report and any accompanying radiological images.