Gene expression profiling of blood for the prediction of ischemic stroke

Abstract

Background and purpose: A blood-based biomarker of acute ischemic stroke would be of significant value in clinical practice. This study aimed to (1) replicate in a larger cohort our previous study using gene expression profiling to predict ischemic stroke; and (2) refine prediction of ischemic stroke by including control groups relevant to ischemic stroke.

Methods: Patients with ischemic stroke (n=70, 199 samples) were compared with control subjects who were healthy (n=38), had vascular risk factors (n=52), and who had myocardial infarction (n=17). Whole blood was drawn ≤3 hours, 5 hours, and 24 hours after stroke onset and from control subjects. RNA was processed on whole genome microarrays. Genes differentially expressed in ischemic stroke were identified and analyzed for predictive ability to discriminate stroke from control subjects.

Results: The 29 probe sets previously reported predicted a new set of ischemic strokes with 93.5% sensitivity and 89.5% specificity. Sixty- and 46-probe sets differentiated control groups from 3-hour and 24-hour ischemic stroke samples, respectively. A 97-probe set correctly classified 86% of ischemic strokes (3 hour+24 hour), 84% of healthy subjects, 96% of vascular risk factor subjects, and 75% with myocardial infarction.

Conclusions: This study replicated our previously reported gene expression profile in a larger cohort and identified additional genes that discriminate ischemic stroke from relevant control groups. This multigene approach shows potential for a point-of-care test in acute ischemic stroke.

Figures

Figure 1. PAM prediction accuracy of IS and Healthy controls using the set of 29 gene predictors of IS from Tang et al, 2006

The Prediction Analysis of Microarrays (PAM) algorithm (K-NN, number of neighbors n=10) was trained on the expression values of a first random half of IS (n=35, 100 samples) and healthy (n=19) subjects from the current study using the 29 IS predictors from Tang et al, 2006. Then, these 29 IS predictors were used to predict the class of the second half of the samples (IS n=35, 99 samples; and healthy n=19, Test Set) and calculate the prediction accuracy. The X-axis represents the patient sample number and the Y-axis represents the Test Set probability of diagnosis. A sample is considered misclassified if the predicted class does not match the known class with a probability greater than 0.5.

Figure 2. PAM prediction accuracy of IS predictors in the current study

Prediction accuracy of the Test Set using PAM. Prediction Analysis of Microarrays (PAM) was used to perform the predictions (K-NN, neighbors n=10; threshold =0). For panels A, B and C the X-axis represents the patient sample number and the Y -axis represents Test Set probabilities. A sample is considered miss-classified if its correct class predicted probability is less than 0.5. The numbers of subjects in the Training Set were: 3h IS n=34; 24h IS n =33; SAVVY vascular controls n=26; and MI n=9. The numbers of subjects in the Test Set were: 3h IS n=33; 24h IS n=33; SAVVY n=26; and MI n=8. A. 3h IS predictors. The 60-probe set predictors for 3h IS (combined from comparisons of 3h IS samples to healthy, MI and SAVVY samples from the Training Set) were put into PAM to predict the class of the Test Set subject samples by calculating the probability that they were in a given class. B. 24h IS predictors. The 46-probe set predictors for 24h IS (combined from comparisons of 24h IS samples to healthy, MI and SAVVY samples from the Training Set) were put into PAM to predict the class of the Test Set subject samples by calculating the probability that they were in a given class. C. Combined 3h and 24h IS predictors. The 97-probe set predictors for 3h IS and 24h IS (combined from comparisons of 3h IS and 24h IS samples to healthy, MI and SAVVY samples from the Training Set) were put into PAM to predict the class of the Test Set subject samples by calculating the probability that they were in a given class.