By studying imaging, clinical, cognitive, and blood-based biomarkers in stroke patients, Veena Nair, PhD hopes her research will inform a risk prediction model that could help triage patient care and better target interventions to prevent post-stroke cognitive impairment and dementia.  

“The idea was to use advanced imaging tools to identify MRI biomarkers early on in the timeline after the stroke to see if those markers could help predict which patients are more likely to develop cognitive impairment and dementia,” Dr. Nair said  

Dr. Nair said as their project evolved, they realized it would be helpful to analyze blood samples in addition to comprehensive MRI and cognitive assessments. Blood-based biomarkers, specifically phosphorylated tau 217, or p-tau217, were emerging as promising indicators of Alzheimer’s disease-related pathology. Adding p-tau217 will allow the team to examine whether cognitive outcomes after stroke reflect vascular injury alone or, in some patients, a combination of vascular and Alzheimer’s disease related processes. 

Their project will also involve advanced machine learning methods to build predictive models that can identify changes in the brain associated with post-stroke cognitive outcomes. This work is happening in collaboration with Pallavi Tiwari, PhD’s Integrated Diagnostics and Analytics (IDiA) Laboratory for Precision Medicine.  

The team hopes that the results could be used in targeted interventions to prevent post-stroke cognitive impairments and dementia and for informing future clinical trial designs. Dr. Nair plans to continue her NIH R01 project that could ultimately help ensure better outcomes for stroke patients. 

To support the addition of blood biomarker analyses, Dr. Nair received research and development funding from the department. The R&D program supports new directions in imaging research, educational initiatives, and the development of promising early-stage projects. 

The departmental award was important in two ways. It provided initial resources to expand the scientific scope of the study, and it demonstrated the department’s commitment to the project. Evidence of institutional support can be valuable when investigators seek additional external funding because it shows that the institution is also investing in the success and development of the research. 

The R&D support complemented Dr. Nair’s successful application for an NIH administrative supplement, allowing the team to incorporate blood biomarkers into the larger R01 study. 

“The departmental funding helped us establish the importance and feasibility of adding this new component,” Dr. Nair said. “It also showed that the institution was committed to the direction of the work, which strengthened our ability to seek additional support from the NIH.” 

Learn more about Dr. Nair’s research in this Q&A, which has been edited for clarity and brevity.  

What inspired you to focus on this topic?  

When I came here to UW–Madison as a postdoc, I worked with my mentor Vivek Prabhakaran, MD, PhD, who is a neuroradiologist, and our primary focus was stroke. I came here to do a postdoc with him to study neuroplasticity: How does the brain change after disease, after injury, even with healthy aging? My focus has always been functional MRI. You can use functional MRI to map areas of the brain that are engaged in cognitive or motor or visual functions. It’s also a great tool to study how the brain changes with aging or with disease. Stroke brings many of those questions together. 

What informed your NIH R01 project? 

In 2015 I had published a paper showing that even when stroke patients don’t show any apparent cognitive deficits in routine clinical interactions, they may have subclinical deficits. Somebody post-stroke could appear cognitively intact, but if you administer a neuropsych assessment, they will still perform below average compared to an age-matched healthy individual.  

Using resting-state functional MRI, we also observed changes in the organization of brain networks after stroke compared with healthy individuals. Some of these network changes appeared to evolve as patients recovered. Those findings helped provide the preliminary foundation for the R01. The broader question is compelling to me: How does the brain reorganize after an injury, and why do some people recover more successfully than others? 

How is the NIH R01 project evolving?  

With stroke, one of the challenges is the heterogeneity. Strokes differ in their location, size, severity, vascular territory, and effects on surrounding brain systems. Patients also differ in age, health history, cognitive reserve, and recovery trajectory. 

A cohort of 100 participants is substantial for an intensively characterized imaging study, but it can still be difficult to account for the many sources of variability. As we plan the next phase of the work, we hope to expand the cohort and obtain longer-term follow-ups. A larger sample will allow us to examine clinically meaningful subgroups and develop more reliable models of cognitive outcome. 

What have you learned from the research so far? 

Because we have required that patients be able to consent on their own and be able to participate in the cognitive testing, a fair amount of participants in our sample seem to be doing cognitively OK or have relatively mild impairments. In future work, we hope to develop approaches that will allow broader participation, including appropriate consent procedures and assessment methods for people with more substantial impairments. 

Our preliminary results have also encouraged us to look at the mechanistic aspects, not just why this is happening. We are thinking about it in terms of what Professor Emeritus Howard Rowley, MD’s described as the “Ps” of acute stroke imaging: parenchyma, pipes, perfusion, and penumbra.  I am adapting that framework for our work in chronic stroke patients. I also have another “P” – performance” – to describe how injury to the brain may ultimately affect cognitive function. 

That framework helps us connect what we observe on imaging with the outcomes that matter to patients in their daily lives. 

How do you hope this study will advance the field of radiology and patient care?  

We are getting these different features or markers including blood flow, brain white matter and gray matter integrity, network connectivity, as well as cognitive assessments. Can we combine them into a reliable risk prediction model?  In the future, a patient might undergo a standardized imaging and clinical assessment after a stroke, and the resulting information could help identify whether that person is at relatively higher or lower risk for cognitive decline. Patients at higher risk could then be followed more closely and considered for appropriate preventive strategies, such as management of vascular risk factors, physical activity, cognitive interventions, or other individualized care. 

That’s hopefully where AI-based methods will help us because this modeling work, with large numbers of interacting variables, requires a lot of input data. We do have some data from the UW Hospital and Clinics database that we are starting to work through, so we can develop a model that is clinically interpretable and then test it on our prospective data set as we collect it.