A conversation with Dr. Eva Mistry, MBBS, MSCI, Assistant Professor of Clinical Neurology and Rehabilitation, University of Cincinnati, on predicting 90-day outcome following thrombectomy.
Interviewed by Dr. Yasmin N. Aziz, MD, T32 Stroke Fellow, University of Cincinnati.
They will be discussing the paper “Predicting 90-Day Outcome After Thrombectomy: Baseline-Adjusted 24-Hour NIHSS Is More Powerful Than NIHSS Score Change,” published in the August 2021 issue of Stroke.
Dr. Aziz: How did the idea behind this paper come to fruition?
Dr. Mistry: To help emphasize the importance of choosing ideal surrogate markers and their correct definitions. When we are developing tools to identify most efficacious acute stroke therapies that improve patient outcomes, and the surrogate outcome measures that we use to predict these patient-centered outcomes are flawed, then all that follows will be flawed as well. Ultimately, there is a real need for solid surrogate markers in acute stroke studies. NIHSS-based surrogate outcome measures are widely used in acute stoke trials. However, they are defined heterogeneously. With the paper, we wanted to establish the pros and cons of using various definitions NIHSS-based surrogate markers, and especially to show how each of them predict the patient-centered outcomes of 90-day modified Rankin Scores.
Dr. Aziz: What mistakes have we made in the past in this domain?
Dr. Mistry: The NINDS tPA trial and other recent acute stroke trials have used change from baseline NIHSS at 24 hours in a dichotomous fashion. Change from baseline is less optimal for various reasons compared to an absolute follow-up NIHSS adjusted for the baseline. Further, when dichotomizing NIHSS-based outcomes, critical clinical information is lost.
Dr. Aziz: What are the advantages and disadvantages of using this kind of approach?
Dr. Mistry: The advantage of using absolute follow-up NIHSS as a surrogate outcome is that you preserve the power of the continuous scale and retain clinically and statistically meaningful information. The downside is that non-dichotomous or continuous definitions are not clinically intuitive. Dichotomizing outcome is an easier approach, but the concern is that it’s leading to a lot of oversimplification. To address this, we provided an automated calculator in our paper that was developed, based on the BEST data and validated using the IMS3 data, that shows the operator the probability of a favorable 90-day outcome for a given 24-hour and baseline NIHSS set of values.
Dr. Aziz: Let’s talk about Figure 2 in your article.
Dr. Mistry: I’m glad you asked about this figure because I think it really hits home about the focal point of this paper. If you look at the beginning of the figure, you see a really tight, almost linear relationship between 24-hour NIHSS and predicted 90-day mRS, meaning that at lower NIHSS scores, we have a reasonable ability to predict outcome. However, in the middle of the graph, around NIHSS of 12-13, you see a definitive change, which highlights that 24-hour NIHSS has a nonlinear relationship with 90-day mRS. This means it really doesn’t make sense to use a dichotomized 24-hour NIHSS as a surrogate marker for outcome at 90 days, as the score does not linearly predict mRS.
Dr. Aziz: Thanks for walking me through that. I can’t help but wonder if we are oversimplifying more than just short-term surrogate markers for outcome. Do you foresee a trend in our field to start steering away from dichotomized outcomes in general?
Dr. Mistry: Yes, I do, at least in research. Practically speaking, this means collecting information and analyzing outcomes such as NIHSS and stroke volume, for example, as a continuous variable rather than dichotomizing results at an arbitrary cut-off. I think the main takeaway behind this paper is that we need to use markers across the board that afford more power and precision in stroke research.
Dr. Aziz: Thanks so much for your time. I really enjoyed speaking with you today.
Dr. Mistry: Absolutely, thanks for allowing me to highlight this work.