So we leveraged the ADNI dataset and we ran 1445 CSF samples on the ELISA panel, Alameda ELISA CNS panel. And to summarize the results, we found out, we did three-fold results analysis. We did cross-sectional study at baseline we saw that the biomarkers which are known in the field, phosphorylated tau 217, 181, 231, NFL, neurogranin, and UCHL1, which is a protein of my interest, they all were showing associations with neurodegeneration at baseline, cross-sectionally...
So we leveraged the ADNI dataset and we ran 1445 CSF samples on the ELISA panel, Alameda ELISA CNS panel. And to summarize the results, we found out, we did three-fold results analysis. We did cross-sectional study at baseline we saw that the biomarkers which are known in the field, phosphorylated tau 217, 181, 231, NFL, neurogranin, and UCHL1, which is a protein of my interest, they all were showing associations with neurodegeneration at baseline, cross-sectionally. Then we did longitudinal analysis to see the associations over time and these proteins survived. However, we noticed that UCHL1 had the maximum beta coefficient. Like in the LME model which we ran, the linear mixed effect model, UCHL1 was standing out. And it’s hard to explain things without having a PowerPoint presentation, but again, words. And finally, I tried to binarize the hippocampal volume which was my outcome and the predictors was from the CNS panel of ELISA. Out of the 1445 participants amyloid beta pathology data was available only for 846 participants so of these 846 we selected the amyloid negative cognitively unimpaired individuals and amyloid positive cognitively impaired individuals. We compared these two groups via ROC analysis and with the Youden index and we found out that they we found a cutoff of less than like in my case in my data set 0.5767 and we applied this cutoff to the entire data set and we thus binarized the continuous outcome hippocampal volume to zero one category that this population which population, which is 1, is undergoing atrophy and whoever is 0, they are stable, they are not undergoing atrophy. And then we ran a logistic regression model and finally we did Cox proportional hazard regression model. And we observed that out of the 130 proteins present in the CNS panel of NULISA, only two survived multiple testing correction and p-value cutoff of less than 0.05. They were UCHL1 and S100A12. So the role of S100A12 has been very briefly explored. I went through the literature and it was not explored very well. But it is a very important protein marker of reactive glial cells. And to hypothesize how these two proteins are contributing to neurodegeneration, the amyloid beta deposits in the Alzheimer’s disease brain. It triggers astrocytes and microglial activity. And these reactive glial cells then release S100A12 protein, which then triggers, which then activates the RAGE and an ASK1 beta pathway, promoting neuroinflammation and contributing to neurodegeneration. And my second hit, which was UCHL1, which is ubiquitin C-terminal hydrolase 1. It’s a protein which maintains the reservoir of ubiquitin and it maintains homeostatically the ubiquitin proteasomal degradation pathway. But under pathological conditions such as reactive oxygen species, nitric oxide and other triggers, they make this UCHL1 less soluble and it’s prone to aggregation, decreasing the UPS degradation pathway, which in turn lowers the clearance pathway of amyloid beta plaques, contributing to neurodegeneration. So yeah, that’s all we found about these two proteins, S100A12 and UCHL1, which might be playing a role or contributing to neurodegeneration. This is just, I would say, a preliminary or pilot data. So we need to replicate this in other cohorts. This was just one ADNI cohort. I would like to replicate this data in other cohorts where we have similar data output, like where we have hippocampal volume, where we have MRI data. And to match up, we have to run it on ELISA. And if the similar hits come up, then it validates, okay, our proteins are really playing a role in neurodegeneration. And it’s a long path to put it into clinical practice unless it’s validated well. But yeah, that might be specific to Alzheimer’s disease since we binarized the hippocampal volume based on amyloid PET data, which is a pathology of AD. So I hope this will be applicable in clinical use later on down the line if it’s validated.
This transcript is AI-generated. While we strive for accuracy, please verify this copy with the video.
