ELISA validation: how to set up plate? - (Mar/25/2014 )
I am planning/performing ELISA assays for the first time. Using a commercially available kit, that has been (semi-)validated by other researchers for the same matrix as I am testing, but from a different species. The sample matrix is stripped by TCA precipitation and sep-pak C18 column (but I assume some matrix components will still be in the sample after..)
I want to perform a partial validation and have planned out a plate layout which is below. My question is: Is this layout correct? In other words, by performing the test outlined below, will I be able to determine:
Limit of detection (LOD)
Limit of quantization (LOQ)
Another question is: is it generally ok to combine accuracy, precision and spike-recovery in to a single series (as seen below). Also can LOD and LOQ be combined?
Appreciate any help!!
1 2 3 4 5 6 7 8 9 10 11 12 A Blk Std1 Std5 AC/PR/SR25 AC/PR/SR100 AC/PR/SR500 LOD5 LOD10 LOQ5 LOQ20 LIN1:4 LIN1:16 B Blk Std1 Std5 AC/PR/SR25 AC/PR/SR100 AC/PR/SR500 LOD5 LOD20 LOQ5 LOQ20 LIN1:4 Neat C TA Std2 Std6 AC/PR/SR25 AC/PR/SR100 LOD1 LOD5 LOD20 LOQ5 LOQ20 LIN1:4 Neat D TA Std2 Std6 AC/PR/SR50 AC/PR/SR100 LOD1 LOD5 LOD20 LOQ10 LOQ20 LIN1:8 Neat E NSB Std3 Std7 AC/PR/SR50 AC/PR/SR100 LOD1 LOD10 LOD20 LOQ10 LOQ20 LIN1:8 Neat F NSB Std3 Std7 AC/PR/SR50 AC/PR/SR500 LOD1 LOD10 LOD20 LOQ10 LIN1:2 LIN1:8 Neat G B0 Std4 AC/PR/SR25 AC/PR/SR50 AC/PR/SR500 LOD1 LOD10 LOQ5 LOQ10 LIN1:2 LIN1:16 N/A H B0 Std4 AC/PR/SR25 AC/PR/SR50 AC/PR/SR500 LOD5 LOD10 LOQ5 LOQ10 LIN1:2 LIN1:16 N/A
AC: Accuracy: closeness of mean test results to the true concentration. Concentrations tested: 25, 50, 100 and 500pg/mL analyte
PR: Precision (intra-assay): closeness of agreement (degree of scatter) between a series of measurements obtained from multiple sampling of the same homogenous sample. The closeness of individual measures is determined by applying the procedure repeatedly to multiple aliquots of a single homogeneous volume of biological matrix. Concentrations tested: 25, 50, 100 and 500pg/mL analyte.
SR: Spike-recovery: used to determine if the assay is affected by the difference between the diluent used to prepare the standard curve and the sample matrix. Concentrations tested: 25, 50, 100 and 500pg/mL of analyte.
LOD: Limit of detection: LOD1, LOD5, LOD10, LOD20 (1, 5, 10 and 20pg/ml of analyte added to neat sample)
LOQ: Limit of quantization: LOQ5, LOQ10, LOQ15 (5, 10 and 15 pg/ml analyte added to neat sample)
LIN: Linearity: Tests the effects of matrix components on measurement of the analyte, by serially diluting the extract in assay buffer.
Linearity tests determine the extent to which the dose-response of the analyte is linear in particular diluents. Dilutions to use: 1:2, 1:4, 1:8, 1:16
Neat: Neat (non-spiked) milk sample processed same way as all spiked samples (less spike)
(All the above have 5 replicates, except LIN, which has 3 replicates)
TA: Total Activity. Positive control - that ensures that the conjugate still has activity
NSB: Non-Specific Binding (Background, Neg contr). Contains standard dilutant, buffer, conjugate, substrate, stop soln.
B0: “zero standard”. Contains: standard diluent, conjugate, antibody, substrate, stop solution
Std1-7: Diluted standard in buffer: 1000, 500, 250, 125, 62.5, 31.25, 15.6 pg/ml.
Your approach seems reasonable if you have a priori defined acceptance criteria that meet the needs of your assay. You say that matrix components may be left within the sample after TCA stipping and column, so in that case, selectivity may be a good parameter to assess. So, you would want to know whether your assay works in multiple different lots (individuals) of matrix rather than just a single lot. Blank, low QC and high QC spikes in 6 to 10 individual lots of matrix is usual. Also, you might consider inter assay precision and accuracy as you may just get lucky with the first plate.
hope it works!
Thank you for your advice. Yes certainly testing matrix obtained from different individuals would be optimal for this validation and is in accordance with FDA guidelines. However, this is unfortunately not possible for us, only a single matrix can be procured for this validation experiment.
The method has been validated previously in the same matrix, but from a new species.
Performing experiments to assess inter-assay precision and accuracy is also out of the question due to limited funds. I am trying to do as best as I can with little resources.
After assessing (validating) this assay, we will use the method for a study on 130 individuals (should take about 3.5 plates).
Is there some way I can work into the study (of the 130 individuals), a measure of inter-assay accuracy/precision? I am thinking I could reserve a few wells on each of the study plates to include a spiked-sample series for assessing inter-assay parameters. Does this seem like a reasonable approach, in lieu of enough plates?
Regarding selectivity in relation to matrix interference: I have planned on using “linearity” (spiked sample processed same way as other samples, then serially diluted. I understand that this should optimally be done using sample from multiple individuals (which I can’t get). Using sample from only one individual, would such a linearity test be at all useful for assessing potential matrix interference?
Thanks again for all your help, it has been most useful,
What kit do you use for your ELISA?