I've been discussing this with a few people, but I'm seeking more definitive arguments. If one is looking to find whether a difference exists in Cytokine levels between cases and controls and also wants to profile the response (i.e. not just interested in one or two cytokines) multiplex is a good tool, but there are issues like somewhat lower sensitivity.
Questions are: What is the best way to deal with out of range values (below threshold) as opinions vary on this one? Are there implications for picking up small changes from baseline in cytokines that are present in extremely low concentrations in normals? Are there any other issues that would suggest that the use of ELISA vs multiplex might be a better option. Any input is appreciated as current opinions are quite diverse.
Dear Luke,
My areas of expertise are sensitization and early events in carcinogenesis. I work a lot with cellular responses upon exposre with chemicals and proteins. My response is based upon that experience.
I general I would advise against single parameter appraoches and in favor of multiparameter description of responses to a challenge. Important is that you can describe changes (e.g cytokines, genes) that are never seen with your test compounds. These changes are obviously related to background and vehicle induced, but it can also include changes induced by compounds your are not interested in. In the case of sensitization for instance, one wants to discriminate compounds that are irritants from compounds that are sensitizing and irritants. In that case you take the changes induced by sensitizers and subtract the changes induced by non-sensitizing irritants and vehicle, as well as background changes. This approach can be applied on proteomics data, cytokine data, genomics data etc. For sensitization, a multiparameter protein profile (N = 100), gene profile (N = 200) and cytokine profile (N = 25) was established. Only this multi parameter approach made it possible to confidently label a compound as a sensitizer and not an irritant in most of the cell models. So far I have seen only one exception to that rule. The as yet best characterized profile is the gene profile, which revealed a sensitivity, specificity and accuracy of > 95%.
The benefit of multiplexing is that it is possible to obtain a profile describing ''all senstizers'' in a generic way. That adds to the flexibility of the test approach, and I avoid that when a new set/class of compounds has to be tested I have to start from scratch.
I hope that helps.
Dear Luke,
I fully agree with Erwin about multiplexing. Especially for the screening purposes, the more the merrier. Changes in gene expression profiles/patterns are great for that purpose, but sometimes changes in protein levels are must for the confirmation.
In case of multiplexing proteins with ELISA or Luminex, we saw little variability between the lab techs who were running the samples and in my lab got new platform. It is more expensive than ELISA, speaking about machine, with comparable price per sample. We switched to Mesoscale Discovery platform and have very nice results. You can take a look at that if you have long term project and you are willing to spend some money. Protocol is very straight forward and lasts about 5-6 hours with sample preparation and have a good sensitivity (comparable to ELISA) with available multyplexing in same well.
Good luck with project
Erwin and Vlado. Thank you both. I am certainly drawn in by the ability to visualise the "whole" response rather than focusing on a couple of factors. That does make sense to me and thanks for adding to my convictions. I have always been drawn in by the "profiling" idea for this project (rather than trying to pick very limited targets) but have previously heard some negative appraisals with regards to multiplex limitations. We're using the Bio-Rad platform for testing.
Thank you both again!
- Badges
- Science method
- Similar topics
- Medicine
- Immunology
- Immunology Techniques
- Immunoassay
- ELISA