Task Force Favors Personalized Research, Treatment Over Grouping

Steve Bryson, PhD avatar

by Steve Bryson, PhD |

Share this article:

Share article via email
PD and its subtypes

A review of current literature by Parkinson’s disease researchers worldwide found studies that attempted to classify the disease into a limited number of subtypes fall short in both methods and clinical applications. 

New approaches are needed in re-evaluating Parkinson’s disease subtypes, and should include those recognizing the individual nature of the disease to allow for more personalized treatment, task force scientists recommended. 

The review study, “Parkinson’s Disease Subtypes: Critical Appraisal and Recommendations,” was published in the Journal of Parkinson’s Disease.

A high degree of variability regarding the underlying biology and clinical presentation of Parkinson’s exists, with patients showing considerable variation in disease onset and progression motor and non-motor symptoms. 

To understand this variability, patients are typically classified into different disease subtypes with similar, but not identical, clinical, biological, or genetic features. While this approach may offer a way to identify those likely to respond to specific treatments, placing patients within a group falls short of each individual’s treatment goals. 

The International Parkinson’s Disease and Movement Disorders Society (MDS) convened the Task Force for Parkinson’s Disease Subtypes in 2018 to critically evaluate available subtyping studies and provide guidance for future studies. 

In this review, researchers based at the University of Ottawa Brain and Research Institute in Canada, collaborating with investigators worldwide, evaluated both the quality of methods used and the clinical applicability of studies proposing new subtyping systems.

“Subtyping of PD [Parkinson’s disease] attempts to explain the disease mechanisms, its natural history and, more importantly, to inform therapeutic development, which has justified a large number of studies by different groups over the last 30 years. However, the impact of such efforts remains unclear,” study co-authors Tiago Mestre, MD, PhD, and Connie Marras, MD, PhD, said in a press release

“They have failed to substantially change the understanding of PD or clinical care thus far. Our current review critically appraises the state of the art in PD subtyping,” they added. 

Task force members reviewed 38 studies divided into two publication periods, between 1980 to 2014 and 2015 to 2019, to balance into older or newer study groups and to assess trends over time. Method quality was scored from zero to two with higher scores meaning better quality, and a checklist for clinical applicability was subdivided into unknown/limited, limited/low, moderate, and strong.

Researchers also compared studies with two methodological approaches: those that were data driven (used in 65.8%), in which conclusions are based on data analysis, and those that were hypothesis driven (used in 21.1%), in which research is conducted to address specific questions (hypotheses). 

Most data-driven studies (64%) used at least three disease characteristics (phenotyping) to identify subtypes, while almost all hypothesis-driven studies used a single phenotype. 

Motor symptoms were most frequently used for subtyping in both types of studies, followed by non-motor domains, such as cognitive, emotional, or autonomic (involuntary process) features. Few studies employed non-clinical biomarkers, such as imaging. 

Regarding method quality, most studies did not achieve top ratings except for items such as ‘diagnostic methods’ and ‘variables compared between subtypes.’ Statistical methods were also difficult to evaluate.

Problems identified included recruitment from clinics rather than community sources, the lack of reporting of specific methods of subtype grouping (clustering), missing information on how the number of clusters was determined, and a lack of justification of sample size. Furthermore, studies did not adjust for multiple comparisons when comparing group features or the impact of characteristics such as disease duration.

The four studies with the highest quality ratings were multicenter, longitudinal (conducted over time), data-driven studies published after 2016 due to the availability of large datasets. These studies investigated more than one clinical feature, such as motor and non-motor symptoms, with one using additional biomarker data from imaging scans and assessments of the cerebrospinal fluid (CSF), the fluid that surrounds the brain and spinal cord.

Two of these studies looked at patients with similar disease stages and duration, and validated the identification of Parkinson’s subtypes. Notably, only one study developed specific methods to classify individual participants into a subtype, by incorporating data on the development of clinical or biological features over time into the definition of subtypes. 

Regarding clinical utility — which reflects the importance of differences among subtypes, potential treatment implications, and applicability to the general Parkinson’s population — most studies were rated poorly. Subtype consistency over time and outcome predictability (prognosis) were unknown mainly due to the lack of longitudinal studies. (Longitudinal studies are observational research, which follows patients over short or long periods to collect data on specific variables, without affecting those variables.)

Data-driven subtyping was seen as difficult and time-consuming compared to hypothesis-driven subtyping, although both approaches were usually rated as inexpensive. In terms of clinical applicability, the team found no apparent differences between newer and older studies.

Overall, “subtyping studies undertaken to date have significant methodologic shortcomings and most have questionable clinical applicability and unknown biological relevance,” the researchers wrote.

The task force recommended more extensive use of longitudinal data to better understand the consistency of proposed subtypes and their prognostic value. Incorporating the pre-diagnostic (prodromal) phase of Parkinson’s into studies was also recommended, which may define subtypes earlier to better predict clinical progression and more quickly develop target-specific potential treatments.

Contemporary medicine is increasingly focusing on the individual patient — known as personalized treatment — which extends to Parkinson’s disease, the team noted. Studies should measure a patient’s disease fingerprint rather than the group subtype. However, they acknowledged such an individual approach to clinical trials poses financial and logistical challenges that will have to be overcome. 

“Having reviewed the existing literature on subtyping and explored the methodologic pitfalls and challenges associated with performing the optimal subtyping studies described above, it is time to reevaluate our approach to understanding and describing PD heterogeneity,” Mestre and Marras said.

“We have provided recommendations and formulated questions that, once addressed, will inform new approaches to better explain the variability in PD including emphasis on the variability at an individual level, more aligned with future application of personalized medicine principles,” they concluded. 


Recommended reading