Neuroimaging markers of early neurological deterioration in acute isolated pontine infarction.

Abstract

Imaging indicators of early neurological deterioration (END) in patients with acute isolated pontine infarctions (AIPI) remained ambiguous. We aimed to find more specific neuroimaging markers for the development of END in patients with AIPI. Patients with AIPI within 72 h of stroke onset were screened from a stroke database from January 2018 to July 2021 in the First Affiliated Hospital of Zhengzhou University. Clinical characteristics, laboratory tests, and imaging parameters were collected. The layers having the largest infarct area on diffusion-weighted imaging (DWI) and T2 sequences were chosen. On the transverse plane of DWI and sagittal plane of T2-Flair images, the maximum length (a, m) and maximum width (b, n) vertical to the length of the infarcted lesions were measured respectively. On the sagittal plane of T2-Flair image, the maximum ventrodorsal length (f) and rostrocaudal thickness (h) were measured. On the sagittal plane, lesions were evenly split into upper, middle, and lower types based on the lesion's location in the pons. The ventral and dorsal types of location were separated based on whether the ventral borders of the pons were involved on transvers plane. END was defined as a ≥2 point increase in the National Institutes of Health Stroke Scale (NIHSS) total score or a ≥1 point increase in the motor items within 72 h after admission. Multivariate logistic regression analyses were used to explore risk factors associated with END. The receiver operating characteristic (ROC) curve analysis and the area under the curve (AUC) was performed to estimate the discriminative power and determine the optimal cut-off points of imaging parameters on the prediction of END. A total of 218 patients with AIPI were included in the final analysis. END occurred in 61 cases (28.0%). Multivariate logistic regression analysis showed that the ventral type of lesion location was associated with END in all models adjusted. In addition, in Model 1, b (odds ratio (OR) 1.145, 95% confidence interval (95% CI), 1.007-1.301) and n (OR 1.163, 95% CI 1.012-1.336); in Model 2, b*n (OR 1.010, 95% CI 1.002-1.018); in Model 3, n (OR 1.179, 95% CI, 1.028-1.353); and in Model 4, b (OR 1.143, 95% CI 1.006-1.298) and n (OR 1.167, 95% CI 1.016-1.341) were found to be associated with END respectively after different adjustments. ROC curve analysis with END showed that the AUC, the optimal cut-off value, and its sensitivity and specificity were 0.743 (0.671-0.815), 9.850 mm, and 68.9% and 79.0% for b; 0.724 (0.648-0.801), 10.800 mm, and 57.4% and 80.9% for n; and 0.772 (0.701-0.842), 108.274 mm2, and 62.3% and 85.4% for b*n, respectively (b*n vs b: P =0.213; b*n vs n: P =0.037; b vs n: P =0.645). Our study revealed that besides the ventral type of lesion location, the maximum width of lesion on the transverse plane of DWI and sagittal plane of T2 image (b, n) may be imaging markers for the development of END in AIPI patients, and the product of the two (b*n) showed a better prediction value on the risks of END.