I'm trying to measure serum IgG levels in bovine calves using the "Bovine IgG Quantitation Kit" from Bethyl Labs. My standard curves are consistently reproducible and have very high r-squared values. I run each sample in triplicate and the values of each replicate are generally quite close (w/i 10% or better).
However, when I run the same serum samples on different days I often get very different values. Not only are the values different, but there is no pattern to the differences - for example, sample 1 might be higher than sample 2 in one assay, and the opposite case in the next. Sometimes there are wildly different results, an order of magnitude or more. It's as if the serum was non-homogeneous, despite my sincere attempts at mixing (inversion and vortexing).
A few notes: the samples are collected in Corvac serum-separator tubes, centrifuged at 3220 x g, the serum transferred to cryovials and frozen at -20 until the time of assay. The samples I have re-tested were refrozen and rethawed. In order to get the samples in the "linear range" of the assay, I am required to do a 1:800,000 dilution. However, the reproducible standard curve is a dilution range of 1:48,000 to 1:3,072,000.