SPOTLIGHT: Q&A with Dr. Bivard
Stroke Progress and Innovation Awards 2017
Andrew Bivard, PhD
Andrew Bivard, Christopher Levi, Longting Lin, Xin Cheng, Richard Aviv, Neil J. Spratt, Min Lou, Tim Kleinig, Billy O’Brien, Kenneth Butcher, Jingfen Zhang, Jim Jannes, Qiang Dong, Mark Parsons
SPOTLIGHT: Q&A with Dr. Bivard
What is the key take-away message from your article?
In this paper, we present a statistical model to predict patient outcome incorporating the output from multi modal imaging, including perfusion and angiography. Overall, we found that the model was highly accurate (AUC >0.9) and that the baseline ischemic core volume was, by far, the most accurate predictor of patient outcome. Next, collateral status and penumbra volume were equally weighted, which is not surprising considering that these are often treatment targets in trials. While these results may seem obvious to some clinicians, we present data on the quantification and relative importance of these variables using a very large database, which is invaluable information.
What prompted you and your co-authors to study this topic or perform this study?
This study used data from the INSPIRE database. This database is the result of a multi-centre international collaboration between centres in Australia, China and Canada. The original intention of INSPIRE was to undertake an analysis similar to what we have presented in this paper. Originally, the intention was to compare the accuracy of multi modal imaging to clinical predictors of patient outcome, with a primary focus of reducing the perceived importance of the time to treatment effect because the authors had noticed that a comprehensive diagnostic assessment of stroke patients was often overlooked for speedy time to treatment metrics. However, with the advent of endovascular therapy, the need for multi modal imaging is now certain as angiography is required to assess an occlusion prior to therapy. Therefore, we modified the original project to what we have now.
What is innovative about this work? And what are its applications?
We present what is really the first large and comprehensive derivation and validation of an advanced imaging model in ischemic stroke for the prediction of patient outcome. The model was shown to be very accurate (AUC >0.9, which was higher than we were expecting). With this data, it is now possible to implement this data into clinical practice and predict patient outcome accurately or to build decision support systems using the variables that we have shown to be the most important. Accurate prediction of patient outcome is important because it can help in the communication with patient families, as well as be used to guide the design of new clinical trials using multi modal imaging in the future.
Tell us about the biggest challenge you came across while conducting this study.
Gathering the data. In order to perform the statistical analysis that we had in mind, it was clear that we would need over 2,000 acute stroke cases with baseline and 24 hours imaging, with complete clinical data including a 3 month mRS. Collecting this data through the Australian NHMRC-funded INSPIRE database took 4 years. In addition, we also had to manage the data quality and ensure that no silly errors were made and that the data was authentic; this, as you can image, takes a considerable effort. Luckily, we had formed a tight international collaborative network of like-minded, science-focused clinicians who were committed to INSPIRE.
Is there anything more you would like to add about your work?
This project took the effort of many individuals across many different countries, not to mention the patients who were required to be consented and examined. While this one paper is important, the idea that there is one first author does not adequately reflect the large team effort required to gather the data.
Click the article title above to access the manuscript online for free.