MDS Blog: Parkinson's Disease or Diseases: Are PD Subtypes Useful in Predicting Disease Progression?

Date: November 2019
Prepared by SIC Member: Alvaro Sanchez-Ferro, MD.
Authors: Eduardo De Pablo-Fernández, MD; Ron Postuma, MD, MSc
Blog Editor: Un Jung Kang, MD

Imagine the following scenario: The first patient of today’s clinic walks into your office and presents with the typical symptoms and signs of Parkinson’s disease. Will this patient progress like the next “textbook” patient exhibiting similar symptoms? 
Go to Discussion


Discussion

Imagine the following scenario: The first patient of today’s clinic walks into your office and presents with the typical symptoms and signs of Parkinson’s disease. Will this patient progress like the next “textbook” patient exhibiting similar symptoms? We know that this is unlikely to be the case. Some of these patients will have mild symptoms over a long period of time, while others might have a more aggressive disease course, showing fluctuations of symptoms or axial disturbances affecting gait just a few years after the diagnosis. It is now clear that Parkinson’s disease is very heterogeneous in presentation and disease course. In this blog entry, we asked two of our experts in MDS, Dr. Postuma and Dr. De Pablo-Fernández their opinions the different PD subtypes and their prognostic implications.

1. Is Parkinson’s a single disease or there are different disease types?


Dr. Postuma: 

That is really an impossible question, since it depends on what one means by ‘disease’.  Clearly there is a profound unifying feature between the large majority of patients with PD, especially the highly-reproducible syndrome of parkinsonism, and abnormal synuclein - we should never forget this as we try to subtype.  Moreover, certain disease states (e.g. dementia with Lewy bodies) might actually need to be unified with PD (i.e. we can ‘lump’ more). On the other hand, there are clearly multiple pathogenic factors at work in this complex disease, and these underlie some profound differences in clinical manifestations and prognosis. 

Dr. De Pablo-Fernández:

This has been a controversial question for many years and has led to a constructive scientific debate. In my opinion, the concept of one disease with a defined clinical syndrome (parkinsonism) associated with specific neuropathological features (Lewy body pathology and substantia nigra degeneration) is certainly applicable (“lumping”) for the great majority of cases with idiopathic PD. There is a minority of cases, particularly those with monogenic forms, who may not fall into this category and future evidence may provide a better understanding of these disorders. Individuals with idiopathic PD can have a very heterogeneous clinical course, and marked variability in the distribution and severity of Lewy pathology. In this scenario, dividing patients into subgroups with similar clinical features (“splitting”) may be useful for care planning and understanding disease pathophysiology. This clinical and neuropathological heterogeneity is likely explained by the complex disease pathogenesis, with multiple interacting individual factors which remain largely unknown. Moreover, the effect of other concomitant neurodegenerative pathologies as a consequence of ageing needs to be added to the equation. All these similarities and differences have resulted in an ongoing “splitting vs. lumping” debate and a better understanding of the PD pathophysiology will allow a better characterization of patients into diseases and subtypes.

2. In your opinion, how will the recent inclusion of non-motor features improve the stratification of different Parkinson’s disease phenotypes?


Dr. Postuma:

Including more non-motor features will certainly improve it dramatically.  In our studies, we find that most of the powerful predictors of prognosis are non-motor.  For example, in one of our clustering studies, the four determinants of subtype were cognition, motor severity (especially gait), REM-Sleep Behavior disorder (RBD), and autonomic dysfunction: three quarters non-motor.  This has led to my personal rule of thumb for prognosis prediction:  if you want to know the prognosis, see how much the degeneration is OUTSIDE the substantia nigra.

Dr. De Pablo-Fernández:

Advances in the understanding of non-motor features in PD over the last several years have made clear that, in addition to their contribution to the clinical burden, they also play a crucial role in PD pathophysiology and disease progression. Some of them have prognostic value and are essential in the definition of subtypes such as cognitive impairment, autonomic symptoms and REM sleep behavior disorder. For example, in one of our studies using data from patients with postmortem confirmation of the diagnosis autonomic dysfunction independently increased the risk of disability and death, and these symptoms were associated with postural instability and dementia, suggesting that some of these symptoms may present together. Having said that, the role of other non-motor features, such as depression, in the progression and pathogenesis of PD is still uncertain and, as our understanding on this area improves, we will be able to determine with more certainty what non-motor features are crucial and determine the minimal amount of motor and non-motor features data needed for accurate subtyping. 

3. Why do you think there are different classifications and what will be in your opinion the best way to define PD subtypes?


Dr. Postuma: 

Data-driven approaches (e.g. cluster analyses) are clearly the best way to define subtypes in most cases, since you are not bound by a pre-existing idea.  Still, these cluster analyses only describe/classify the patients in the specific study, and adding even a few extra cases can change the clustering solution completely. Moreover, the cluster solution is actually just a cloud of characteristics - individual patients might not fit to a cluster.  So probably the key step to improve reproducibility is to examine carefully your clustering solution results, and then create rules to put each patient into a specific phenotype.  Then other groups can directly test what you found.


Dr. De Pablo-Fernández: 

The methods in the definition of PD subtypes have gradually evolved from empirical observation based on motor features of clinical examination to data-driven analysis of multiple motor and NMS, usually involving complex assessment tools, where statistical analysis identifies the features that group together without any a priori hypothesis. While the latter provide a more objective approach, their results are influenced by what NMS are selected for inclusion, the choice of number of different clusters and it is only valid at a group level. These inherent weaknesses are partly responsible for these differences and even studies using the same data have provided different PD subtype classifications. Some cluster analysis results did not show any significant association with what the existing literature describe as key factors in the pathophysiology and clinical progression (such as age of onset or genetics) and were excluded from clinical subtype definition. In my opinion, cluster analysis should not be strictly applied and both methods should be combined, including other additional features in the model if strongly supported by evidence from other areas of research.

4. Most of these classifications are mainly used in the research setting. How do you think that they can be translated into clinical practice? In other words, how will a movement disorder specialist care differently a new patient with one subtype vs another?


Dr. Postuma:

The key, again, is to boil down the cluster solution into a simpler way to classify patients.  We have had some luck here; for example, just checking for disease-duration-adjusted UPDRS, RBD, Mild Cognitive Impairment, and autonomic loss (especially cardiovascular) can classify patients as ‘diffuse malignant’ PD.  In a separate autopsy study, patients with this ‘diffuse-malignant’ phenotype had a 10-fold increase in major milestones of decline (plus a 3-fold increased risk of death).  We have also managed to make a simple 8-point office-based checklist for dementia (the MoPaRDS); over the subsequent 4 years, those scoring 0-2 had less than a 1% annual risk of dementia, compared to 15% per year in those with scores >6.  That’s pretty strong predictive value for such a simple scale.

Dr. De Pablo-Fernández:

As I briefly mentioned before, recent subtyping solutions have become more complex and the clinical assessments often involved extensive questionnaires and evaluations that are not usually available or may not be practical in clinical practice. Researchers should find ways to simplify classification systems and assessments in order to translate these advances into clinical practice. We recently demonstrated that an adaptation of one of the subtyping solutions into a simple classification using information from routine clinical assessments was able to predict long-term disability and prognosis in motor and non-motor areas. This has important implications for the individual care of PD patients as clinicians would be able to predict the disease course and plan accordingly any future interventions at the time of diagnosis.  

5. Do you foresee that biological markers or other novel technologies such as artificial intelligence will be used to better stratify disease subtypes in the near future?


Dr. Postuma:

Biomarkers will almost certainly help, once we get better ones!  I would guess that the key biomarkers for overall prognosis will be those that assess non-substantia nigra/diffuse degeneration.  But the final thing to remember is that there will never be a perfectly clean subtyping solution.  Humans are vastly complex organisms, and as we age, multiple pathologies are the rule rather than the exception. There will always be boundary issues, and people who do not fit into our boxes!

Dr. De Pablo-Fernández:​

Inverting the answers to your question, artificial intelligence may be the key for detailed extensive assessments applicable to clinical practice. Although still on its early days, the rapid development of wearable devices able to monitor motor performance and other non-motor functions has opened the avenue of systematic collection of big clinical data outside the hospital environment. With future development of these devices, artificial intelligence is expected to become cost-effective and widely available in clinical practice, and it will certainly have an impact in the way individuals are assessed for clinical subtyping.

Development of accurate biomarkers would provide a biological basis for PD subtyping and this will be crucial once disease modifying treatments are available. Individualized medicine is certainly in the horizon but development of more accurate biomarkers (or more likely a combination of them) is needed first.   

Summary

In summary, recent advances have led to more systematic classifications of clinical subtypes of patients with PD with important clinical implications on prognosis and survival. Consideration of non-motor features had dramatically improved our ability to prognosticate the disease progression. However, challenges remain as the biological basis for these subtypes and the optimization of data-driven cluster analyses needed for subtyping are still to be established. Future advances in the understanding of PD pathogenesis will refine current classification systems and biomarkers and artificial intelligence may play an important role in the assessment of individuals for subtype classification. Fortunately, these more refined, yet simple classification systems are to be translated into clinical practices to help us better predict the disease course of individual patients who come to our clinic.
 



References

 

  1. De Pablo-Fernandez, E., Lees, A. J., Holton, J. L. & Warner, T. T. Prognosis and neuropathologic correlation of clinical subtypes of Parkinson disease. JAMA Neurol. 2019; 76: 470–479.
     
  2. Fereshtehnejad SM, Zeighami Y, Dagher A, Postuma RB. Clinical criteria for subtyping Parkinson's disease: biomarkers and longitudinal progression. Brain. 2017; 140: 1959-1976.
     
  3. De Pablo-Fernandez E, Tur C, Revesz T, Lees AJ, Holton JL, Warner TT. Association of Autonomic Dysfunction With Disease Progression and Survival in Parkinson Disease. JAMA neurology. 2017;74(8):970-976.
     
  4. Greenland JC, Williams-Gray CH, Barker RA. The clinical heterogeneity of Parkinson’s disease and its therapeutic implications. Eur J Neurosci. 2019; 49: 328-338.

 

 

Submit Your Comment

The following required items were not provided or are in the wrong format. Please provide the required responses and submit again:

Name
Comment Title
Comment: 1000 characters
  [[put error message here]]

 

About the Authors
 

Eduardo De Pablo-Fernández, MD.
Clinical Research Associate and Honorary Consultant Neurologist, UCL Queen Square Institute of Neurology

Ron Postuma, MD, MSc.
Professor, Montreal General Hospital, Montreal, Canada

About the Contributor
 

Blog Prepared by SIC member:
Alvaro Sanchez-Ferro, MD.
Neurology and Senior Researcher, HM-CINAC, Mostoles, Spain

Top

We use cookies to give you the best possible experience with our website. These cookies are also used to ensure we show you content that is relevant to you. If you continue without changing your settings, you are agreeing to our use of cookies to improve your user experience. You can click the cookie settings link on our website to change your cookie settings at any time. Note: The MDS site uses related multiple domains, including mds.movementdisorders.org and mds.execinc.com. This cookie policy only covers the primary movementdisorders.org and mdscongress.org domain. Please refer to the MDS Privacy Policy for information on how to configure cookies for all other domains on the MDS site.
Cookie PolicyPrivacy Notice