Hello Lea,

I too have far more experience with ELISA than Fluoranything, but will be looking into possibly going the fluorescent route for some upcoming thyroid H assays - my first thought is you've essentially separated out the competition, which, following the theory of the indirect comp assay, shouldn't matter unless there is binding with too high an affinity. In that case I would think you would need to have added the labeled antigen first, and somehow have the label interact with the antibody to produce a higher affinity. What makes the most sense to me is that your labeled antigen concentration is too high. I think I would try backing it off a bit, or if you haven't already, try varying which antigen you add first into the Ab mix, see if it makes a difference.

Hi Stephen E Bartell - I knew we weren't really doing a true competitive assay with the separate steps, but with sensitivity above 2 ng/mL I knew we could do better (and had to or risk needing to concentrate the plasma).

As with ELISAs, the %TB/TC has to be around 30% for TR-FIAs. Thus, the first step was to figure out a combo of [primary] to [label] that produced a ratio of 30%. So, we can't really adjust the amount of label without throwing that off and then affecting the binding. And I know I need the [primary] to be about what I have it at because otherwise the lowest standard will be too noisey around 100%, which leads to some curves being wonky. That being said, I have tried decreasing the [label] at several concentrations of primary Ab (in the general vicinity of what produces the 30% TB/TC ratio) without any good effect on the binding or sensitivity.

Good news is that we at least figured out something. I had been washing the plate between each incubation period. The last try, I added the labeled antigen to the standard/sample wells without washing and voila! Binding went down to 0% on the high end of the standard curve. This was compared to a standard curve where I added the sample/standard and label at the same step (which is what we have been doing for other TR-FIAs and works for binding in the GH assay, but the sensitivity wasn't great), so I'm confident it wasn't a fluke. The sensitivity was also better, though I will repeat everything to confirm and also play around with [primary] and [label] to optimize the assay.

Must be some weird on-off rate binding kinetics between standard and label. But wanted to post my "answer" in case others run into something similar.

Thank you for the follow-up Lea. I did understand you realizing it not being the normal competitive assay, I should have expanded my thought on that. Adding the labeled antigen in a separate step, if it displaces the standard/sample antigen, is competition just the same. Separating the competition into more of a "displacement" scenario could be different depending upon which antigen gets access to the antibody first.

Anyways, it's good to know that modifying wash steps has that impact. Best wishes with the sensitivity end. I've often pondered methods to stop the signal at the low end of the std curve from overwhelming the detection system, essentially blocking the ability to see differences between 80-100%