Effect of Possible Osteoporosis on Parenchymal-Type Hemorrhagic Transformation in Patients with Cardioembolic Stroke

Abstract

Background: hemorrhagic transformation (HT) is a frequent complication of ischemic stroke, and parenchymal hematoma (PH)-type HT has been shown to correlate with symptomatic deterioration. Because both bone and vascular smooth muscle cells are composed of type 1 collagen, we hypothesized that the integrity of blood vessels around the infarction area might be more damaged in osteoporotic conditions after a cardioembolic stroke.

Methods: we measured frontal skull Hounsfield unit (HU) values on brain CT images from cardioembolic stroke patients. We conducted a receiver operating characteristic curve analysis in a large sample registry to identify the optimal HU threshold for predicting osteopenia and osteoporosis. Hazard ratios were estimated using a Cox regression analysis to identify whether osteoporotic conditions were an independent predictor of PH-type HT in patients with cardioembolic stroke.

Results: altogether, 600 consecutive patients (>18 years old) with cardioembolic stroke were enrolled over a 12-year period at our hospital. The infarction volume and hypothetical osteoporosis were independent predictive factors for PH-type HT development in patients with cardioembolic stroke. In the male group, hypothetical osteoporosis was an independent predictor for PH-type HT development after cardioembolic stroke (hazard ratio, 4.12; 95% confidence interval, 1.40-12.10; p = 0.010).

Conclusions: our study suggests an association between possible osteoporosis and the development of PH-type HT in patients with cardioembolic stroke. Our findings could help to predict PH-type HT by providing a convenient method for measuring the HU value using brain CT images.

Keywords: Hounsfield unit; bone mineral density; cardioembolic stroke; hemorrhagic transformation; osteoporosis; parenchymal type hematoma.

Figure 1

Scatterplot, linear regression line, and ROC curve from the SHUB registry. (A) Overall distribution of HU values at each of four lines based on T-scores; (B) linear association between the HU value at each of four lines and the T-score; (C) linear regression line showing the association between the T-score and mean skull HU; (D) ROC curve to identify the optimal cut-off skull HU values for predicting osteopenia and osteoporosis. ROC = receiver operating characteristic; SHUB = skull Hounsfield unit and bone mineral density; HU = Hounsfield unit.

Figure 2

Scatterplot with linear regression lines, boxplots, and Kaplan–Meier curves in the study patients. (A) Scatterplot with linear regression lines showing the association between age and mean frontal skull HU values based on the presence of PH-type HT; (B) boxplots with dot plots of the mean frontal skull HU values classified by age group according to the presence of PH-type HT; (C) overall cumulative hazard of PH-type HT development; (D) cumulative hazard of PH-type HT development according to the hypothetical BMD groups. HU = Hounsfield unit; PH = parenchymal hematoma; HT = hemorrhagic transformation; BMD = bone mineral density.

Figure 3

Forest plot of estimates from the multivariate Cox regression analysis to predict PH-type HT occurrence using potential predictive factors (adjusted for sex, age (continuous variable), BMI (continuous variable), NIHSS at admission (continuous variable), cerebral infarct volume (continuous variable), tPA use, thrombocytopenia at admission, classification of mean skull HU, previous stroke history, hypertension, diabetes, current smoking, hyperlipidemia, and prior antithrombotic use) in the study patients. PH = parenchymal hematoma; HT = hemorrhagic transformation; BMI = body mass index; NIHSS = National Institutes of Health Stroke Scale; tPA = tissue plasminogen activator; HU = Hounsfield unit.

Figure 4

Kaplan–Meier curves showing the cumulative hazard and hazard ratio estimated by the multivariate Cox regression analysis for the development of PH-type HT in the study patients according to the hypothetical BMD groups classified by sex. (A) Female; (B) male. PH = parenchymal hematoma; HT = hemorrhagic transformation; BMD = bone mineral density.