This month's question from the Molecular Biology Forums (online at molecularbiology.forums.biotechniques.com) comes from the “Immunology and Immunochemistry Methods” section. Entries have been edited for concision and clarity. Mentions of specific products and manufacturers have been retained from the original posts, but do not represent endorsements by, or the opinions of, BioTechniques.
Molecular Biology Techniques Q&A
How do I determine the sensitivity of an ELISA? (Thread 22460)
Q I am running simple ELISAs to detect the presence of antibodies against an enzyme in serum. The assay works well, with positive controls showing strong signals, clean negative controls, and clear detection of antibodies in the test samples. However, I have some questions regarding interpretation of the results.
I am not certain of the proper method for determining whether a sample is positive or negative for the antibody. I understand that it must be based on the cutoff or endpoint of the varying dilutions of the sample. As I dilute the sample in the wells across the plate, at some point, I lose the colored signal. Is that the endpoint used to calculate the titer? If so, is there a standard way of calculating the titer by comparing the dilution factor of the sample to the negative controls? Will someone please explain to me exactly how to determine an antibody titer in an ELISA?
Also, when discerning between positive and negative wells, is a well considered positive simply because a color is produced in that well? Would that also be true for barely diluted samples? Or is there a standard ratio of antibody titers that forms a threshold between the positive control and test samples?
A You are talking about the sensitivity of the ELISA. In general, as you dilute the standard with a known amount of antigen, at some point, it will reach an absorbance that is twice higher than that of the negative control. This is considered the minimal amount of antigen the ELISA can detect and indicates the sensitivity limit of your assay. You may consider a sample as positive if its absorbance is twice as much as the absorbance of the negative control.
A You cannot say that a sample is automatically positive just because the absorbance is twice as high as the negative control. A sample might be positive even if it is only 1.5×higher. There are also ELISA tests where samples are defined as positive when they are three times higher than the negative controls. It all depends on the manufacturer and the test system you are using; the cutoff has to be calculated with defined samples, standards, and programs.
A Yes, but a sample is positive if its absorbance is repeatedly twice that of the control. This is the common practice. If you want to be sure, you can run statistics. If you are using commercial kits, they should provide suggestions for determining the cutoff.
Q One of the things I am worried about is that my positive control, which is pooled from reacting patients, has a very high titer and needs to be diluted a great deal to reach the endpoint. Signals from my positive samples are far lower than the signal from the positive control. I know that pooled positive sera will give a very strong signal, but because the signals from my test samples are so far from the positive control signal, it makes me question whether I should really consider those samples positive. When compared exclusively against the negative control, the signals are clearly positive. Is it correct to use only the negative controls to determine whether or not the sample is positive, and use the positive control only to determine if the ELISA worked properly?
A I usually use the positive control to check if the ELISA is working well, just in case all of my test samples turn out negative. When a sample is only weakly positive, you should repeat the test to ensure that it is indeed positive.
Q I have been following this conversation on calculating the antibody titer in an ELISA, but I still have a few questions. (They are listed below). I am working on a previously established protocol, but have some concerns about the experimental design that I'd like to resolve before proceeding with my experiment. I am new to ELISA, so I'm not yet familiar with the best practices for these assays.
I found a method for determining endpoint cutoff values (Frey et al. 1998. A statistically defined endpoint titer determination method for immunoassays. J. Immunol. Methods 221:35–41) and wondered if this is widely accepted or if there is a more straightforward method. Specifically, is twice the negative control signal a generally accepted threshold?
Is the negative control referred to in the previous answers the blank or background value? Or is this a measure of the baseline reactivity?
To determine the percentage over baseline, should I compare mean endpoint OD values of the baseline versus samples?
A The cutoff for an ELISA is usually calculated using a sufficient number (~50–100) of negative specimens from the population. Normally, a cutoff is equal to the mean OD of the negative samples plus 3 standard deviations. The cutoff can be made more robust by adding a higher factor to the mean. This will make the ELISA more specific, but with a slight loss of sensitivity. That is more desirable in most cases.
To determine antibody titer, a positive specimen is serially diluted 5-fold or more and then tested on the ELISA. The endpoint titer is determined by the last diluted specimen that gives positive results on the ELISA.
A You should look in “The ELISA Guidebook ” by John R. Crowther to find answers to all of these questions. It is common to use the average of negative samples + 3 sd to determine the cutoff. All the ELISA tests used in serology for antibody detection should come with details regarding their sensitivity and specificity. You can determine the cutoff by whatever means works best for you; just be certain to reference it when presenting your results. You will need to mention how you calculated it, how many negative samples you ran, how you knew they were negative, and if you compared ELISA results with other tests.
Q2 I am running a sandwich ELISA for TNF-α in 96-well ELISA plates. I put my standards in rows A and H with the samples in between. Recently, one row of standards develops normally, but the other row does not develop at all. I load from the same reagent reservoir using a multichannel pipet, so everything should be the same for the samples. I did the capture overnight in a refrigerator with a glass door, but I don't think that the capture antibodies are light-sensitive. What could be causing this?
A Is your plate reader working properly? Does it work if you move the standard from row H to row B?
Q For some of the ELISAs, row A does not develop and in others, the problem was in row H. I also could not see any color in those wells. I tried using rows A and B and didn't have a problem, but then I am concerned about how much I can trust the data from samples in row H. This only happens with the TNF-α ELISA.
A Check the multichannel pipet to see if it takes up the reagent equally in all tips. If one tip is loose, it might not deliver the same amount of reagent as the others.
Q I usually check the multichannel by eye while adding reagent and will check the volume in all of the tips on my next attempt.
A Row A and H are the leftmost and rightmost pipetting channels. Will your plate reader accept plates that are rotated 180°?
A If this only happens in row A or H, it might be caused by how you hold the pipettor or how the plate is oriented. The plate reader is probably innocent since it cannot read if there is no color.
A This looks like a case of one loose tip or a pipettor with one clogged channel. When did you last have it serviced and calibrated? Make sure all of your tips are tight and keep an eye on the liquid level as you pipet.
