Has anyone performed analysis of PD progression using more than 10 years of PPMI data? I’m particularly interested in identifying PD subtypes and looking at the stability of those subtypes over long periods. It is my hypothesis that subtypes are essentially unstable over long periods. So, for instance, tremor dominant can switch to gait dominant over a few years and vice versa; rapid early progression can slow down and vice versa.
I previously did some analysis using PPMI to look at clinical presentation (primarily MDS-UPDRS scores) over 10 years and was planning to update my analysis with more data given that PPMI now holds almost 14 years of measurements for some participants, but wanted to check if anyone was already working on something similar, or was aware of this already being looked at. Thanks.
Hi! This is a great question. The challenge with the long-term longitudinal data lies in the fact that the longer the follow-up period, the amount of data that are available tends to drop off. This is particularly true once you start to subcategorize individuals according to genetic status. Also, if you are looking for primary outcomes such as MDS-UPDRS, and other measures that may be important (LEDD), there is less homogeneity in the data. Definitely worth looking at though as the dataset continues to mature!
I’m really interested to hear about this… I haven’t looked at this specifically. I’ve worked a lot with the transcriptomics data so I’ve mostly looked at the 4 year window that matches that dataset. I have looked a bit at the MDS-UPDRS beyond that timeframe.
How complete is some of that data? I’d be interested to see how many participants there are on the long-term end - when I have looked at some data points, I’ve seen the number of datapoints drop off over time, so I’d be curious how many participants have which types of data available going out at 7 years, 10 years, 14 years, etc.
That’s a great research question, Jodie, and I wish you luck on that. I haven’t tried dealing with PPMI data that surpassess 5 years of follow-up since even at this stage, there are a lot of missing data yet, so I can only imagine what can happen in longer follow-ups.
Regarding my opinion on this, my hypothesis is the same as yours. I think these hypothesis-driven PD subtyping methods aren’t reliable at all. I think they are created based on a type of data that is rather subjective and prone to different biases, such as the possile influence on the quality of the medical treatment the patient is using, the exact time in which the patient was evaluated in the ON period, and others more.
I also think that another area os research interest is reproducing these analyses using larger numbers, such as, for example, using data from patients from AMP PD. One also could compare how positivity in the alpha-synuclein seed amplification assay correlates with these hypothesis driven subtypes (my hypothesis is that they don’t).
You make a good point about the data tailing off. Just looking today at MDS-UPDRS data I found the following distribution. out of a total population of 3609 participants. I think if I am looking at 12+ years there is enough here to perform some meaningful analysis, though as you say, other tables may not be complete.
|Years of UPDRS data | Number of participants|
|14|5|
|13|50|
|12|122|
|11|93|
|10|85|
|9|77|
|8|118|
|7|187|
|6|164|
|5|196|
|4|86|
|3|133|
|2|392|
|1|647|
|0|1254|
AS PPMI continues into 15+ years, I think the data will be very useful in providing insights into long-term progression.
Thanks for the links - yes, I was familiar with the studies showing the instability of the TD/PIGD classification. I think what I will try to do is analyse the long-term stability of some of the data driven classification schemes that several research teams have developed in recent years. My hypothesis is that these are more to do with individual response to PD rather than fundamentally different underlying pathogenic mechanisms, and over time these clusters will be unstable. Of course, genetic versions of PD may well be different and several of these such as SNCA do seem to have significantly different trajectories compared to idiopathic cases.