I have found 2 protocols for caspase 1 detection from cells supernatant and both said to avoid fetal calf serum (FSC) or other kinds of serum to the media before the stimulus and to collect the supernatant, because FSC will overload the gel in SDS-Page and mess western blot/ immunobloting to detect Caspase 1 cleaved.
The concentration of protein, mostly serum albumin, in serum is very high. There is a limit to how much protein you can load in the well of a gel before the bands become distorted and the resolution is harmed. With so much albumin in the medium, you would probably have trouble detecting the protein of interest. If you load only as much protein as would be recommended for a Western blot, (1) it would be almost all albumin, reducing the sensitivity for detecting caspase, and (2) the caspase bands would be distorted by the huge albumin bands, even though the molecular weights are quite different.
Serum has no defined composition. There are many things in the serum and it is impossible to be certain of the things in there. Introducing an unknown parameter (more like a mixture of hundreds of unknown parameters) to an assay is generally considered a very bad idea. All assays with cells should be done in buffer. I have used HBSS very effectively when working with HUVEC.
As I tried to think of the ways serum could affect an assay, some possibilities arise. Maybe I am overthinking.
The serum proteins can effectively adsorb most of the dye (if you are using one) and result in poor signal.
The serum proteins may bear ligands that bind to the antibody/antigen-bound probe and give false positive result.
Serum proteins binding to the antibody/antigen-bound probe may also reduce the probes available for binding to the target protein, hence lowering the sensitivity of the assay.