I am your host Sarah Camargos from the Federal University of Minas Gerais of Brazil. Today I have the pleasure to welcome Dr. Pierre Emmanuel Sugier, a postdoctoral fellow in the laboratory of mathematics and their applications at the University of Pau in Pesa de Ladu and team member of the center of Population in French National Institute of Health and public research Paris.
View complete transcript
He is the senior author of the recent paper published in the Movement Disorders Journal entitled, "Investigation of Shared Genetic Risk Factors between Parkinson's Disease and Cancer."
[00:01:00] Welcome, Dr. Sugier, and many thanks for your time.
Not all the movement disorder specialists are aware of the associations between Parkinson's disease and cancers, or the association of smoking status and Parkinson's disease.
Please give us some background on this epidemiological association.
[00:01:25] Dr. Pierre-Emmanuel Sugier: Well, I can first talk about the associations between the smoking status and Parkinson's disease because it also has many implications in the association between Parkinson's disease and cancer.
So many epidemiological studies have found an inverse association with Parkinson's disease, which means that ever smokers have reduced risk of developing Parkinson's disease. This association is one of the more robust association in Parkinson's disease with a close dose response relationship, this association does not seem [00:02:00] to be explained by cerebral and cerebral bias and genetic confounding and cerebral bias is unlikely because this inverse association is also found in young people well before the risk of death from smoking is significant, reverse causality has been hypothesized as PD patient may stop smoking more easily during the PD prodromal phase than other persons due reduced responsiveness to nicotine.
But this association is found in patients who quit smoking more than 30 years before the onset of these symptoms. So it doesn't seem to be the explanation for the inverse association between smoking and and Parkinson's disease. Also mendalien randomization studies performed in K H P D and in both PDD and E P D GC consortia as both found significant inverse association between smoking and Parkinson's disease and therefore these results are rather in favor of causal association of tobacco of Parkinson's Disease. The [00:03:00] mechanisms are still unknown, but could involve some gene bio smoking interaction and now about the association between Parkinson's disease and cancer. So many epidemiological studies have been conducted to investigate this association and these studies supported general inverse association in in other words patients with Parkinson's disease tend to have a lower risk for cancer in general and cancer patients have a lower risk for Parkinson's disease. But this inverse association is mostly explained by inverse association with smoking related cancers because of the lower prevalence of smoking in patients with Parkinson's disease and conversely, there is a well-established positive association of PD with melanoma and to a lesser extent with breast cancer. And results are foreign and inconsistent for different types of cancers. So the underlying mechanism remain poorly understood.
[00:03:54] Dr. Sarah C Camargos: Yes, some mixture between environmental and [00:04:00] genetics, maybe.
And what were your goals in this study and how did you address them?
[00:04:06] Dr. Pierre-Emmanuel Sugier: Well as I said, epidemiological studies studying the association between Parkinson's disease, and cancer, led to inconsistent results. This study's faced a number of methodological difficulties and possible including confounding diagnostic bias, competing risk or, selective survival, for example.
And alternatively, genetic approaches are less likely to be affected by confounding reverse causation or surveillance bias because genes are randomly assigned at birth and are not influenced by exposures. So genetic approaches can be complementary to epidemiological studies, and may help understand where the genetic pleiotropy could account some of the associations highlighted by epidemiological studies, and that is why we did this work, and it is in this work we investigated the role of the genetic contribution in the relationship between Parkinson's disease and cancer in particular whether it was pleiotropy. [00:05:00] That is the fact that one gene or one genetic locus can affect multiple diseases plays a role in this association. So this study was carried out using individual data from international consortium Courage-PD on Parkinson's disease and a P here on different thyroid cancer. As well as summary statistic from G was results from Consortium and from the I P D GC Consortium. And identifying common genetic risk factor of Parkinson's disease and cancer closely associated with PD, Parkinson's disease, or for which an association is suspected. And we had the opportunity to explore the association melanoma breast cancer, ovarian cancer, prostate cancer, lung cancer, and, and thyroid cancer. And first, we estimated genetic correlations between Parkinson's disease and cancers using results from, this large she was and, and intense Caribbean score regressions. And, and secondly, we analyzed the session of polygenic risk score for Parkinson's [00:06:00] disease. And their individual SNPs, genetic variance, which with each cancer and conversely, the session of the polygenic risk for each cancer with Parkinson's disease.
[00:06:11] Dr. Sarah C Camargos: Thank you. Please tell our listeners a little bit about the courage PD consortium and the iPDGC.
[00:06:20] Dr. Pierre-Emmanuel Sugier: The Q H P D consortium combined data from studies carried out in many countries around the world as this consortium is founded by the GP and D, the European program on neurodegenerative disorders. And these studies has been genotyped by using the same neuro chip array in order to perform GWAS, genomewide association studies, and to then be able to also perform ization analysis and play analysis. And in our team, we were responsible for this part. And the iPDGC Consortium is the largest consortium on Parkinson disease. I am not involved in this consortium, but some statistic from GWAS [00:07:00] data of E P D GC are publicly available online.
[00:07:03] Dr. Sarah C Camargos: So there is a lot of shared data in this studies, right?
This is amazing.
[00:07:08] Dr. Pierre-Emmanuel Sugier: Yes, that's right.
[00:07:10] Dr. Sarah C Camargos: What were your exclusion criteria for your study?
[00:07:15] Dr. Pierre-Emmanuel Sugier: The summary statistics from GWAS data, which we had access to for several consortium were carried out on participant of your European origin. And therefore, in order to have more financing our analysis, we choose to exclude participants who were not of European origin in the, H, PD and data.
And we also choose to exclude case only studies and those with less than 50 cases and control. And there were no other exclusion criteria than those performed by each consortium.
[00:07:46] Dr. Sarah C Camargos: Right. Perfect. Most of us clinicians are not familiar with the methodology of G Y G W Y S and polygenic risk scores. Can you brief us on how the [00:08:00] team worked to find a shared risk for both diseases?
[00:08:03] Dr. Pierre-Emmanuel Sugier: So the genome wide as genome-wide association study or GWAS analysis analyzed millions of genetic variants, mostly SNPs for single single nucleotide polymorphism, distribute across the genome to, detect association between traits of a disease with these genetic markers testing each SNPs independently to others.
For their association with the disease adjusted for some cobis to in particular, avoid some population stratification bias for example. Then we can get from this GWAS beta estimates of effect of each SNPs on the disease and corresponding not error, and p values. And that is what we call the summary statistic from GWAS results in fact. And this is what we add as type of data for most of the consortium we access for. We then used the LD score regression methods to [00:09:00] estimate the genetic correlation between Parkinson's disease and each cancer. This is a statistical approach based on GWAS data that do not require individual level genotype data. It leverages the linkage ion or LD information that is a measure of the correlation between genetic variance along the genome to estimate the contribution of common genetic variance to the trait irritability or the genetic correlation between two different diseases. And this methods uses the hypothesis that each genetic variance capture a part of the effect of the variance on the disease for which it is in ld. And there is a proportional relation between the total score of LD for SNPs and the estimates of the project score for the two disease, so then we can get the genetic correlation disease from it. And for polygenic risk score this is a numerical called score that estimates an individual genetic risk for a particular disease based on multiple genetic variants across the [00:10:00] genome.
It is calculated by summing the weighted contribution of each SNPs where the weights are derived from, the GWAS results, in fact. So then in each concert data, we rebuild the previously calculated PD, Parkinson's disease, PRS, PRS for polygenic risk score as a predictive genetic variable for Parkinson's disease to test if it is associated with the cancer.
And in Parkinson's disease data, we build previously calculated cancer PRS as a predictive variable for each cancer to then test if it is associated with Parkinson's disease. And finally, we look at in detail the association of each SNPs with both disease for, polygenic risk score, we found an association.
[00:10:46] Dr. Sarah C Camargos: Tell us about your major findings in your effort.
[00:10:51] Dr. Pierre-Emmanuel Sugier: Yes. by using genetic correlation analysis, we first confi confirmed previously reported positive genetic correlation of Parkinson's disease [00:11:00] with melanoma. And we also identified an additional significant positive correlation of Parkinson's disease with prostate cancer. And by using cross phenotype, polygenic risk score analysis, we also identified significant association between Parkinson's disease and ovarian and breast cancer. And this association was positive with breast cancer and negative or inverse for ovarian cancer. Analysis of the association between SNPs of the PRS highlighted in cross phenotype analysis, showed that the inverse association between Parkinson's disease and ovarian cancer was mostly driven by one SNP in the 17 Q 21.31 region and even if there is association between PD and breast cancer seems more diffuse on many SNPs. We also found SNPs positively associated with breast cancer in this genomic region. In particular, the gene map team in this region tagged by SNPs in the PD PRS was already known as a [00:12:00] potential predictive marker in epithelial patients treated with axel platinum first line chemotherapy, and as a mass marker of the paclitaxel sensitivity in breast cancer, and also the gene NSF of the ovarian cancer.
PRS has already been reported to be associated with Parkinson's Disease through the same SNIP This gene has also been recently reported as associated with cancer pleiotropy. So overall this evidence in favor of contribution of pleiotropic genes to the association between Parkinson's disease and specific cancers, and suggests the interest of studying the, TRO between Parkinson's disease and cancer to better understand the shared biologic mechanisms.
[00:12:40] Dr. Sarah C Camargos: Great. One question of curiosity. Did you solve the question about smoke PD and lung cancer?
[00:12:48] Dr. Pierre-Emmanuel Sugier: Well, as ASCO consortium, on lung cancer performed GWAS is, stratified by smoking statue, however, never were we able to examine the session of Parkinson's disease [00:13:00] PRS with lung cancer risk according to smoking statues. First, cross phenotype analysis offline cancer stratified by smoking, that showed an inverse association in ever smokers, whereas the association was positive in never smokers. And we also found a trend toward a gene environment interaction, even if it was not statistically significant possibly because of the small number of never smokers compared with the number of ever smokers and small effect sizes. We also had access to histological subtypes for lung cancer and we found an inverse association between the Parkinson's disease, PRS, and squamous carcinoma of the lung. And these results are consistent with the fact that squamous carcinomas are known to be the lung cancer surgical type with the strongest association with strabeco.
[00:13:48] Dr. Sarah C Camargos: Very interesting. What do you think are the strengths and limitations of your study?
[00:13:56] Dr. Pierre-Emmanuel Sugier: The main strengths of our study is that we use the largest GWAS [00:14:00] available as it's present date for several cancer together and with two independent large P Parkinson's disease data sets to replicate our findings. While correcting for multiple testing, For the limitation, I would say, the size of the GWAS data sets, that was different for different phenotypes and for some of them we did not have access to individual data, but for the GWAS summary statistics, and that can affect slightly the power of our analysis. The panels of SNPs available in each GWAS were also different which could affect the results, especially for genetic correlation analysis. As I said already our analysis were restricted to participants of European descent and then additional study are needed in other populations. And finally, this is very likely that specific genetic FICO may, make some people more susceptible to the harmful effect of a lot of different exposures. Then it would be interesting to have a focus on possible [00:15:00] environmental factors that can interact with pleiotropic gene, as I said, with both Parkinson's disease and cancer. Unfortunately, as analysis stratified by environmental factors were not available for different consort GWASs except for the lung consort GWAS that performed analysis is stratified by smoking. The only effect of environmental factor we could not consider is the smoking statues, in fact, and in the long consortium GWAS.
[00:15:27] Dr. Sarah C Camargos: I'm curious about other population as well. Maybe you can consider using GP two global Parkinson's geneticist program the near future.
What do you believe to be the next steps?
[00:15:43] Dr. Pierre-Emmanuel Sugier: Our overall results suggest the importance of shared genetic variance between Parkinson's disease and some cancers. So, one interesting thing we found is a disparity in the results, depending on the genetic scale at which we look at this relationships. [00:16:00] For example, by looking at the genetic correlation between Parkinson's disease and ovarian cancer, we find no significant associations.
But if we look at the scale of the polygenic risk score, which is a predictive construct from certain SNPs in particular, we found an inverse association between these two diseases. We can see something similar happening with breast cancer, which had a trend for positive correlation, but not significant genetic correlation with Parkinson's disease.
But then we found a positive association by using cross phenotypes, PRS analysis. We can even go further by looking, in particular at the SNPs use for the construction of cancer PRS. We find many SNPs associated with Parkinson's disease, even when we look for cancers for which we did not find a significant cross phenotype association with Parkinson's disease. What seems to happen be that there are many regions of the genome that seems to be associated with both diseases at the same time, regardless of the type of cancer.[00:17:00] However, the effect of each disease can vary greatly between these regions and what is interesting to understand in particular is that the direction of these effects can vary from one locus to another, and the effects can thus compensate each other and match an overall effect, particularly for the genetic correlation at the genome scale, but also inside an usable, viable, such as the PRS.
Then the absence of a discovery at the scale of the genome does not mean absence of local correlations in specific genomic regions. So this phenomenon shows the importance of looking more closely at the genetic relationships between Parkinson's disease and cancer and by using appropriate methodologies.
Hence, there is also a need to develop well-designed methods for pleiotropy analysis. In fact, we have recently developed several methods which make it possible to analyze [00:18:00] Pleiotropy by taking into account the heterogeneity in the direction and also the magnitude of the genetic effects between the phenotypes and this at the scale of the SNPs but also at the scale of the gene and all the biological pathways, and in particular the GCP based methods is adapt to this context and we intend to use this method to explore in more detail the relationship between Parkinson's disease and cancer, for which we found an association exploring pleiotropy at SNP and gene level to better understand the shared biological mechanisms between Parkinson's disease and cancer.
[00:18:37] Dr. Sarah C Camargos: That's great. Well, congratulations on your terrific research. It is indeed great work. We've shared data, a lot of collaboration, huge work, and many thanks for your time. We are waiting for your next studies.
Thank you, Pierre.