I am excited to kick off our Data Community of Practice’s first Ask Me Anything (AMA)!
I am Dr. Michael Alosco (@malosco), Associate Professor of Neurology at Boston University. I work primarily with neurodegenerative diseases and am currently the Director of the Boston University Alzheimer’s Disease Research Center Clinical Core, Co-Director of Clinical Research at the BU CTE Center, as well as the Director of Neuropsychology of the Memory and Aging Clinic.
The topic of this AMA is: The application of neuropsychology to answer questions surrounding cognitive decline in the setting of neurodegenerative diseases.
I’ll be answering your questions on Discourse on Wednesday, October 30th. Please submit any questions you may have below (in this thread) by Wednesday, October 23rd.
I look forward to hearing your questions, and am eager to provide insights pertaining to neuropsychology and its use in assessing, mapping, or treating cognitive decline in those with neurodegenerative diseases!
Hello, it is my first AMA, so I may be on the wrong place, but I will send my question here.
The cognitive decline can be affected by several factors and one is based on the way that we collect the information. Is there any way to address the cognitive decline considering that low level of education? Maybe protocols can give biased results because this, giving over/underestimated measures.
Hi @malosco, I’m interested in understanding the trajectory of cognitive decline over time as well so I’m excited about this opportunity and appreciate your time!
One challenge in assessing this trajectory is the impact of the learning effect. To track cognitive function, we need to use the same test repeatedly. However, for participants, the second, third, or fourth administration of the test is not truly the same as the first. The format and questions are no longer novel, and what participants remember and learn from previous tests introduces another layer of complexity in measuring cognitive change over time.
I don’t yet have a clear strategic approach to address this challenge, so I would love to hear your insights on the topic!
Hi @malosco
Very interesting topic especially for me. I worked on validating the MOCA questionnaire in Ghana before using it to assess cognitive impairment in a cohort of PD patients.
Yet still I found that due to high illiteracy rate,most patient had some cognitive impairment due to scoring system.
Also, cultural and psychosocial factors, including the person’s emotional state can contribute to the underreporting or avoidance of admitting to cognitive impairments in neurodegenerative diseases.
I guess I will be eagerly looking out for you to address how assessment tools used for screening and diagnosing can be structured to address these two challenges.
Hi Mike,
I’m very interested in Parkinson’s disease (PD) progression monitoring. What neuropsychological tests would you recommend for high-functioning individuals and those in the prodromal phase of PD? Current test batteries often have ceiling effects in these populations. Are there any specific tests you’d suggest adding to existing protocols? Also, what are your thoughts on using computerized versions that can be administered remotely? How might these be effectively utilized for progression monitoring?
I would be happy to know what is your take on these points.
How often do you see other diagnoses (depression, anxiety, etc.) with neurodegenerative disease patients and how do you separate that from cognitive decline and other symptoms related to the neurodegenerative disease?
Hi @malosco , how do you assess the clinical relevance of performance on neuropsychological testing when assessing treatment/exposure differences between 2 groups?
This is a great question. As you are alluding to, most neuropsychological tools were developed and normed in English speaking individuals of European ancestry. Therefore, there are quite a number of limitations to many of these tools particularly as it related to the normative data (which are used to determine impairments). Clinically, we use all data points (history, test data, etc) to best understand a persons “baseline level of cognitive function” and then make a judgment to determine if that person has cognitive impairment, e.g., is their test performance a change for them? In large research studies, this type of judgement call is more challenging and not feasible. We often encourage using raw scores as opposed to normative scores. Controlling for factors like education can help. However, even just controlling for education years is insufficient as it doesn’t tap into quality of education. You can also control for estimated level of baseline function using other measures, such as the WRAT-4. There are very wide ranging practices here with no great solutions unfortunately. I think it is critical to be aware of your sample and the potential related limitations of neuropsychological test administration in order to guide interpretation of results.
Awesome question @fbbriggs. I ultimately think it will depend on your research question, as always. History of depression can affect risk for cognitive decline (precedes it) but cognitive impairment can also lead to depression. So, I think depression and fatigue can play multiple roles and how you handle it statistically will depend on your sample and research question. Here is a recent study by our group that examined the relationships between cognition, AD pathology and neuropsychiatric symptoms. Here, we modeled cognition as a mediator of associations between AD pathology and neuropsychiatric symptoms.
Hi @vdardov, thanks for the question! I will point you towards my response to @fbbriggs. Psychiatric symptoms are quite common in the setting of neurodegenerative disease. I conceptualize them to be (1) a separate process if history suggests the symptoms preceded onset of cognitive decline, (2) onset of psychiatric symptoms follows cognitive decline and can thus be in response to cognitive difficulties, and/or (3) psychiatric symptoms are a neuropsychiatric manifestation of neurodegenerative pathology affecting brain regions that modulate mood/behavior. Numbers 2 and 3 are perhaps most challenging to differentiate but, in reality, likely both are occurring and as the disease advances (e.g., dementia) I tend to think it to be more related to the pathology at that point (certainly not always). This is how I tend to think about it clinically.
Another tough question @hirotaka! I’ll answer this question from a neurodegenerative disease framework. There are some things we try to combat practice effects in the study design phase, including alternate versions of the tests, case control designs, and long enough time intervals to limit practice effects (e.g., at least 1 year between assessments). Some of the tests also have reliable change index (see here: Practice effects due to serial cognitive assessment: Implications for preclinical Alzheimer's disease randomized controlled trials - PMC). This index provides provides interpretation of meaningful change on the test over time, and what reflects improvement vs practice effect. It is also comes down to interpretation, particularly at the individual level. As reviewed here Lower practice effects as a marker of cognitive performance and dementia risk: A literature review - PMC, lack of practice effects can reflect the presence of cognitive decline. That is, practice effects might be expected for individuals with normal cognition but if practice effects become diminished that might mean there is pathology present interfering with normal cognitive processes. If you have enough time points and enough time, noise from practice effects should be able to be diminished. Overall, my recommendation is to aim for low frequency assessments separated by at least 1 year and use alternate versions.
Thank @Vidash! These tests, particularly screeners, are so sensitive to sociocultural factors unfortunately. I get particularly concerned about screening tests, as an example, take a look at this study we did: Mini Mental State Examination and Logical Memory scores for entry into Alzheimer's disease trials - PubMed. Briefly, it showed that screening measures might result in inappropriate subject enrollment in clinical trials. Take a look at some of my other responses and let me know if I didn’t address your questions.
I think that there are few ways to do this. First, clinically, we often will convert raw scores to standardized scores that account for demographic variables. We can get T-score distributions, for example, where there is a mean of 50 and a standard deviation of 10. A 1.5 SD below the mean of 50 (T=35) is often interpreted as being impaired. When evaluating change in scores over time, a general rule of thumb is that a 1 SD change in score is clinically meaningful. However, we often will use raw scores of modeling as standardized scores are on a truncated distribution and it limits variability. With evaluation of raw scores, I think you can typically look to your standard ways for assessing effect sizes between groups to determine clinically meaningfulness.