Evaluating Stroke Biomarkers

Early identification and treatment of acute stroke continues to be challenging despite numerous advances in treatment protocols and diagnostic technologies, primarily more sophisticated imaging systems. However, intriguing new research showed that a panel of biomarkers was complementary to clinical variables and discriminatory for all acute stroke, hemorrhagic stroke, and ischemic stroke. This issue of Strategies explores the findings.

Despite a decade of solid improvement in treatments for and public education about stroke, the condition remains a vexing challenge to clinicians, as well as a leading cause of death and long-term disability in the U.S. Timely identification of stroke is critical to effective treatment, yet only an estimated 2% to 6% of stroke patients are definitively diagnosed and able to be treated within the golden 3-hour window to administer tissue plasminogen activator (tPA), the only FDA-approved drug for acute treatment of ischemic stroke. Even without tPA, early management decisions still influence clinical outcomes, reinforcing the need for rapid diagnosis. Designated stroke centers have sophisticated multimodal imaging capabilities that have significantly enhanced sensitivity in diagnosing stroke. Unfortunately these methods are not available widely, particularly in community hospitals where most stroke patients present initially.

In addition, the brain itself presents a challenge in identifying biomarkers of importance in ischemic stroke. Aside from the blood-brain barrier and the complexity of the ischemic cascade, the brain has a heterogenous cell population, distributed differently throughout the central nervous system and with different tolerances to ischemia. Previous studies evaluating individual biomarkers have not shown potential as standalone diagnostic tools.

Now, a seminal study evaluating a panel of four biomarkers has demonstrated the potential of a new diagnostic paradigm for stroke (Stroke 2009; 40:77-85). "This takes the theoretical discussion of biomarkers of stroke out of the theoretical realm and puts it in an area where it's likely to reach prime time, at least at some point in the foreseeable future," observed Robert Adams, MD, professor of neuroscience, university eminent scholar, and director of the MUSC Stroke Center at the Medical University of South Carolina in Charleston. Adams was not affiliated with the study.
The researchers evaluated the performance of MMP-9, D-dimer, S100β, and BNP using the Triage Stroke Panel, which is a rapid POC fluorescence immunoassay being developed by Biosite. The study involved 1,146 patients with neurological symptoms consistent with stroke who were enrolled prospectively at 17 sites. A separate cohort of 343 patients subsequently was enrolled to validate the biomarker panel's performance. The biomarker results were blinded both to treating clinicians and the researchers, who subsequently reviewed medical records of all subjects, also in a blinded fashion. Both the treating clinicians and researchers assigned final diagnoses to each case. Of 946 patients included in the final analysis, 31% had ischemic stroke, 10%, intracranial hemorrhage, 21%, TIA, and 38% stroke mimics, i.e., underlying nonvascular medical conditions that resulted in neurological deficits.

The logistic model incorporating the four biomarkers provided good discrimination between stroke and mimic, with an area under the ROC curve of 0.69, but showed higher sensitivity in discriminating intracranial hemorrhage from mimic, with an area under the ROC curve of 0.76, and slightly less in ischemic stroke, with an area under the ROC curve of 0.67. A test result in the highest quartile indicated a 6-times increased probability of acute stroke, 15 times increased probability of hemorrhagic stroke, and a 5 times increased probability of ischemic stroke (all with p<0.0001). When the threshold was set at the first quartile, the resulting sensitivity was 86% for detecting all stroke, 94% for hemorrhagic stroke, and 85% for ischemic stroke. The results were "virtually identical" in the cohort of subjects used to validate the study, according to the researchers.

Different Aspects of the Ischemic Cascade

Of the four biomarkers, BNP made the greatest contribution in discriminating stroke from mimic, and S100β made the least. The latter finding is interesting given that S100β, a marker of astrocyte activation, is the only one of the four that is specific to central nervous system tissues. However, the other three represent different components of the ischemic cascade, and when used together, provide complementary information in the diagnosis of stroke, the researchers found. The four biomarkers contributed the majority of discriminatory power in this model, but the addition of gender, age, and presence of atrial fibrillation slightly improved performance for both ischemic stroke and intracranial hemorrhage.

Principal investigator Daniel Laskowitz, MD, MHS, characterized the results as an "important first step", but cautioned that the study did not demonstrate a definitive biomarker-based approach to stroke diagnosis. "It suggests that this approach may be viable. One of our concerns was whether the biomarkers would be elevated soon enough after acute ischemia to be within a framework where someone could intervene," he explained. "I think the answer to that is yes." Laskowitz is associate professor of medicine, anesthesiology and neurobiology, and director of the neurovascular laboratories at Duke University Medical Center. "We also showed that a panel of biomarkers can be used in conjunction with the clinical evaluation to provide complementary information," he noted.

Upon further validation, Laskowitz believes the biomarker panel holds the most potential in helping community hospitals rapidly evaluate suspected stroke patients and transfer them to stroke centers for advanced treatment. "The idea is if we had a POC platform with a rapid turnaround, the test would be like an EKG of the brain, and within 10 to 15 minutes you'd be able to say this person's at high risk and needs to be transferred to a stroke center," he explained.

Adams also believes the biomarkers could assist in the rapid diagnosis of certain subgroups of stroke patients. "The test may provide additional confirmatory data in cases where you have an ambiguous or unclear clinical story and don't have CT-scan changes or don't have access to diffusion-weighted imaging," he said. Eventually the test might prove useful in stratifying risk, along with other measures such as the NIH Stroke Scale, he predicted.

Laskowitz expects the study will lead to investigation of other candidate biomarkers with further refinement of diagnostic capabilities.