Dear @DCoP_Innovators :
I wanted to share a topic on a recent experience conducting a systematic review and metaanalysis on Parkinson´s Disease Genetics. We did it with @danieltds, so he can also comment on challenges on this topic.
I know, probably there are no new concepts to share, there are just tips, but If in the future, If I had to do it all over again, or teach someone to do it, I would find very interesting to have this written.
I will summarize the most important and practical features, but remember to follow the PRISMA checklist and statement, for reporting in a transparent way.
A summary of the cjecklist could be found here: https://prisma.shinyapps.io/checklist/
I would summarize in consecutive steps:
- Design
Define the research question the more specific as you can, and try to discuss it with the rest of the team, before doing the next steps. I Will give you an example:
We wanted to describe the frequency of monogenic parkinson´s disease in Latin America. If you look it quickly it seems easy according to PICOT:
P: Latin American people
I: genetic testing for PD
C: genetic testing for controls (if reported)
O: number of pathogenic variants in PD genes
T: time limited to the last day of search.
This apparently simple question, needed further analysis, that I want to share, in case someone does something similar in other populations.
-
P: define if you are going to include multicentric studies that are reporting samples from the country you are interested, or not. In my opinion, unless there are symmetrical proportion of patients contributing to that cohort, it is better to exclude them. But you have to decide this from the beginning,
-
I: For genetic testing: Which are going to be included? It is going to depend on your objective, but if you want to identify patients with genetic variants, GWAS studies probably wont fit.
-
C: If there are controls could help, but not mandatory
-
O: This was the most challenging issues, because monogenic PD has many definitions, the MDS update on nomenclature of genetics of movement disorders is very good, but not widely known, and has some difficulties. On the other hand, many papers report genetic variants, more than patient with genetic variants, so it is challenging to extract information. For recessive genes, you should decide before if you are reporting all genetic variants, or homozygous/compound heterozygous, which I would say is the major confounding factor. Also, for PD genetics, we included GBA1 for its importance, but is a risk factor, not a monogenic form of the disease, so I think this should be discussed from the beginning.
Our text finally got written like this:We conducted a systematic review with a meta-analysis of (1) PD patients reported as having hereditary parkinsonism (2) caused by the genes outlined by the MDS Nomenclature of Genetic Movement Disorders20,21 (summarized in Supplementary Figure 1) and additionally including heterozygous carriers of GBA1 pathogenic variants, (3) reporting results among Latin American patients, and (4) published in English, Spanish or Portuguese. We excluded reports focusing on functional, epigenetic, biomarkers, risk factors, and heterozygous carriers of pathogenic variants in other recessive genes with controversial impact and non-causative pathogenic variants on parkinsonism genes. Studies conducted on animal models or in silico were also excluded. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were followed in the elaboration of this review.
Some other tips:
For the search strategy
It Is important to determine which databases, how many independent searchers, and which MESH or equivalent terms are you using before you perform the search.
for the Selection process
For registering the papers, we used a great tool called Rayyan (https://www.rayyan.ai/), Which allows you to import the results of your search and read and categorize the abstract upon your inclusion an exclusion criterion. The decisions can be recorded blindly and a third reviewer can solve disagreements.
Article assessment
After checking the inclusion and exclusion criteria, it is good to complete the Prisma flow chart, that is available in a simple interface in this link PRISMA Flow Diagram
After the first screening, if there are many articles, they can be divided into different investigators extract the information needed. It was very useful to have a google forms to standardize and summarize the information screened upon the reviewers. Also, after completing this part, reports that deserve further discussion should be screened again by two reviewers to solve discrepancies
Good luck if you are planing to do one of this, and would happy to aswer any questions.
