Developing predictive biomarkers for dementia of Parkinson’s disease

Keywords::

• biomarker
• dementia
• EEG
• mild cognitive impairment
• MRI
• Parkinson’s disease
• PET

In 2001, the NIH Biomarkers Definition Working Group published what has become the standard definition of a biomarker: “…a characteristic that is objectively measured and evaluated as an indicator of normal biological processes, pathogenic processes or pharmacologic responses to a therapeutic intervention” [1]. Attention to biomarker discovery is accelerating in both funding applications and the scientific literature. Now, biomarkers are being vigorously pursued in Parkinson’s disease (PD), including for the prediction and progression of dementia in PD [2,3]. Although the idea of a reliable objective measurement for predicting PD dementia and cognitive decline is attractive, numerous questions remain for such biomarkers.

Why do we need to predict dementia in PD?

At first glance, predicting dementia in PD does not seem so difficult. Studies show that the accumulative prevalence of dementia (and certainly cognitive decline) is quite high. Therefore, merely having PD seems to confer high risk of dementia [4–8]. If a treatment for preventing or slowing down cognitive decline in PD was available, it may be useful to give such treatment to all PD patients once the PD diagnosis is made. However, we must realize that there is a critical need to develop predictive biomarkers for PD dementia. First, the pathophysiology of PD cognitive decline is poorly understood. At autopsy, the brain that has suffered PD dementia during life shows Lewy bodies and Alzheimer’s disease pathology changes in the cerebral cortex and in subcortical regions, in addition to significant depletion of acetylcholine and monoamine neurotransmitters [2]. In such an instance, it is not possible to attribute the beginning of cognitive decline to a single or even small group of pathological and/or biochemical changes. A predictive biomarker for PD dementia will be useful for the study of the early and more specific biochemical changes of PD cognitive decline. Second, the time from onset of PD to onset of dementia varies. Typically, there is a period with no or very slow cognitive decline, followed by an inflection point after which there is a much more rapid decline. The time to this point varies considerably from very early to late in the disease course, and biomarkers may help to predict the time to the inflection point. Third, evaluating treatments for preventing or slowing PD dementia development will be a challenge because of the large number of subjects and long follow-up duration required to prove an effect. A predictive biomarker for PD dementia could be used to define an enriched population for study over a defined period of time. Such a biomarker could reduce the cost of such treatment studies, as well as offer an opportunity to test the effect of the treatment on the biomarker itself.

How is a biomarker for predicting cognitive decline in PD validated?

The literature for PD biomarkers has steadily increased in recent years [2,3]. Numerous articles and proposals seek to ‘validate’ their biomarker(s). However, it is uncommon for authors to state a priori criteria for biomarker validation or the lack thereof. Indeed, the evaluation of a biomarker critically depends on the application needed and the relative value of biomarkers already available. Moreover, multiple predictive biomarkers for PD dementia decline may be needed to play different roles, regardless of whether one type of biomarker may seem more ‘valid’ than another.

PD patients who experience dementia subjectively complain about poor memory, a change in mental sharpness and disability when performing tasks of everyday life. Their spouse/caregiver expresses concerns about the intense care and supervision that such patients need. Historically, the gold standard for measuring cognitive ability to aid in the diagnosis of dementia has been a battery of neuropsychological tests. In addition, early psychometric abnormalities have been studied for their predictive value for PD dementia under the entity ‘mild cognitive impairment in PD’ (PD-MCI) [9,10]. However, neuropsychological testing in PD patients present challenges. Motor impairment, decreased attention and PD medication, among other factors, may confound neuropsychological measurement. As a result, it seems prudent to study other predictive methods that do not have the same confounders (although they will have others). In this way, multiple predictive biomarker types can complement each other for the overall purpose of studying and evaluating treatments of PD dementia decline. Currently the possible predictive biomarkers of PD dementia being pursued for this purpose include electrophysiology, radiology imaging techniques, proteins in bodily fluids and genomic/proteomic markers, among others. For the time being, there is some evidence for imaging, such as structural MRI and FDG-PET, fluid markers, such as low CSF Aβ-42, and quantitative EEG [11–16]. However, longitudinal studies performed in different populations should be a requirement to assess a predictive biomarker’s usefulness.

Determining the threshold to declare ‘validation’ of a biomarker will depend on many factors. Moreover, the data transformation and statistical methods that are most appropriate to use are in flux, making it perhaps premature to state a widely applicable validation threshold for predictive biomarkers of dementia in PD. Odds ratios, hazard ratios, receiver operating characteristic curves, synergistic models and risk reclassification measures all have merit as ways to analyze predictive biomarker data [17]. However, it needs to be emphasized that the investigators should not use any of these methods indiscriminately and without considering their advantages, assumptions, and disadvantages. Of note, there is still much work to be carried out regarding standardization of methods to enable comparisons between centers, and very few longitudinal studies exist.

PD dementia versus cognitive decline?

The recent Movement Disorder Society criteria for PD dementia have helped to clarify the dementia syndrome and its criteria for patients with PD. The diagnosis of dementia is often declared precipitously despite the fact that cognitive decline occurs for years before dementia criteria are met. Thus, in order to understand PD dementia, it would seem that the process and prediction of PD cognitive decline is worthy of study. Early stages of cognitive decline may reveal secrets on how the path to dementia is traveled, as well as provide an easier therapeutic window to intervene so that dementia does not develop.

PD-MCI has been studied as an intermediate state of PD cognitive decline that is believed to carry an increased risk for PD dementia within 3–5 years [9,10]. The attractiveness of PD-MCI is that it may be conceptualized as a defined cognitive state intermediate between relatively normal cognition and PD dementia that carries a high risk of progressive cognitive decline. Biomarkers that are continuous measures across the spectrum of PD cognitive decline may afford additional insight if they can be shown to predict further cognitive decline and dementia development. Such biomarkers could mimic the natural continuous progression of cognitive decline from relatively normal cognition to PD dementia.

What are the future predictive biomarker research needs for PD cognitive decline/dementia?

We are truly only at the beginning stages of evaluating predictive biomarkers for PD cognitive decline/dementia. Much more research is needed to realize the positive impact that such biomarkers could have on the lives of our PD patients. Some recommendations and suggested perspectives are as follows:

• • At this point in time, the focus should be on studying the evaluation of these predictive biomarkers rather than their ‘validation’;

• • Predictive biomarkers for PD cognitive decline/dementia should be investigated for their: first, ability to understand cognitive decline pathophysiology; second, use in studies evaluating treatment directed at the prevention and/or symptomatic improvement of PD cognitive decline/dementia; and third, eventual potential usefulness in the clinic;

• • Multiple biomarker types may be complementary and can provide evidence that a single biomarker change is not owing to confounders;

• • All longitudinal studies of PD, including treatment trials, should include biomarker evaluation and analysis for the prediction of cognitive decline/dementia.

Conclusion

The mechanisms that trigger PD dementia development are poorly understood. PD dementia adds a tremendous burden to the patient who already suffers from motor disability. Thus, it is critical that the pathophysiology of PD dementia is understood so that effective treatments may be developed. The development of predictive biomarkers for PD dementia will play an integral role towards attaining this goal. The field of predictive biomarkers for PD dementia is in its infancy. Evaluation of such biomarkers should be a priority for the PD scientific community.

Financial & competing interests disclosure

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.