Laboratory Considerations

3.1 Assay Optimization

When evaluating immunoassays, it is important to keep in mind that these are governed by the Law of Mass Action. The reagents are thus in an equilibrium condition. The assay then is subject to fluctuations due to temperature (of the reagents and of the laboratory in which the assays are conducted) and length of incubation time. Since reactions are occurring at the surface of the microtiter plate, shaking the plate to mix the contents of the wells may affect the local concentration of reactants. Each of these factors should then be controlled in order to improve the precision of the measurements. Typically assays are conducted with reagents which have been equilibrated to room temperature. If room temperature is not constant (within 3 - 5 degrees of variation), than assays should be conducted using a forced-air incubator. Shaking the plates periodically during incubation periods may also improve precision. For immunoassays utilizing 30 minutes or longer incubation periods, the reactants have likely come nearly to equilibrium and thus conducting assays with precise timing is unnecessary. However, for immunoassays utilizing shorter incubation periods, precise timing will improve precision.

3.2 Protocol Design

The methods used most commonly in the analytical laboratory are based on the 96-well microtiter plate format. There are numerous permutations and combinations of ways that samples and standards can be placed on a 96-well microtiter plate. The number of calibration wells and the known concentrations used for the calibration curve affect the precision of the determinations of the unknowns, as does the choice of the number of replicates of each unknown. Within the framework of a 96-well microtiter plate, how does one maximize the number of samples analyzed while maintaining the best possible accuracy and precision. A statistically based method for determining the weight of these factors has been presented by Rocke et al. (1990). In broad terms there is a tradeoff between efficiency in the number of samples that can be run per plate vs the additional precision obtained by running more replicates of each sample or standard. Samples are generally analyzed at several dilutions. For example, 1:2, 1:4 and 1:8. Values obtained for at least one of the dilutions should fall near to the center of the calibration curve. This approach is taken in the event that a positive response is due to a matrix effect. If multiple dilutions are analyzed then discrepancies among the calculated values may indicate an effect of matrix. If a single dilution is analyzed then a matrix effect may not be revealed until the sample is confirmed by an independent method. However, if the matrix is known to not interfere in the analysis, a single dilution may be analyzed. If the result is too high, then further dilutions can be made. Last, efficiency of analysis may dictate splitting replicates of unknowns between microtiter plates. This allows the achievement of desired accuracy at the lowest cost. A typical layout for a 96-well microtiter plate is shown in Figure 3.

A typical plate format should have a calibration curve with enough replicates as shown in Figure 3. In fact, it is recommended that a calibration curve be run on every plate because the reactants are governed by the Law of Mass Action, they are in a dynamic equilibrium. If a given plate is subject to differences in manipulation time, temperature of incubation or other factors which may effect the equilibrium, the samples on that plate can be compared to a calibration curve subjected to those same variables. See also tutorial 5.10. Guidelines for the Efficient Use of 96-Well Microtiter Plates.

3.3 Sample Preparation

The extraction and cleanup steps used for the more conventional detection methods can be used as a starting point in devising sample preparation schemes for the corresponding immunoassay. Since immunoassays are aqueous based analyses, this may result in a reduction in sample preparation in general. Recommendations are given in each protocol for sample preparation where information is known. General “rules of thumb” are presented in tutorial 5.2. Consideration in Sampling and Sample Preparation for Immunoassay Analysis.

3.4 Matrix Considerations

The utility of any analytical method depends on the absence of interferences derived from reagents and the matrix. The interference question must be addressed by running appropriate blanks as controls prior to analysis. Due to the selectivity of the antibody for the analyte, immunoassays usually do not require the rigorous extraction and cleanup methods often used for other instrumental methods. In addition, the sensitivity of the antibody can be exploited such that interferences may be “diluted out” while still maintaining the desired detection limits. Interferences may vary with reagent batches and matrix sources and thus must be checked frequently by a combination of running appropriate blanks, and confirming positive samples by an alternate analytical method. The latter is crucial to using any assay method, including ELISA, for monitoring samples of unknown origin when corresponding field blanks are not available. Other commonly used methods for identifying and normalizing for matrix effects on an analysis, such as the method of standard addition, can also be used (Miller and Miller, 1984). If these approaches fail to adequately address the problem of interference from the matrix, then some sample preparation may be necessary, for example, solid phase extraction. Once the analyst begins to introduce sample preparation steps prior to immunochemical analysis, than consideration must be given to whether this method is in fact the most time saving and cost effective method available for this specific analytical problem. See also tutorial 5.3. Approaches for Testing for Matrix Effects.

3.5 Data Handling

As with any analytical technique, the generation of a reproducible standard curve with minimal error is critical. The standard curves generally resulting from immunoassays are sigmoidal in shape, suggesting that the best fit curve could be log-logit or 4-parameter. However, other curve fits such as linear, quadratric, semi-log or log-log can be used to find the best fitting standard curve. A good reference pertaining to curve fitting appears in “Data Analysis and Quality Control of Assays: A Practical Primer”, by R. P. Chenning Rogers in Practical Immunoassay, editor Wifrid R. Butt; published by Marcel Dekker, Inc., N.Y. 1984. If the choice of standards provides a complete definition of the shape of the curve, (i.e., the curve has at least 2 to 3 points each defining the upper and lower asymptote and at least 4 points defining the linear region), the 4-parameter fit of Rodbard (1981) is the method of choice for data analysis in the authors’ laboratories. It is important that enough standard concentrations are used to ensure that the curve is well defined and constant for these concentrations. Without this information, the computer could force an improper fit (Gerlach et al., 1993). The equation for the 4-parameter fit is:

=A−D/1+x/C∧B+D

where y is the absorbance, x is the concentration of analyte, A and D are the upper and lower asymptotes respectively, B is the slope and C is the central point of the linear portion of the curve, also known as the IC50 (Figure 4).

The best quantitation of unknowns is carried out when unknown absorbances fall in the central portion of the linear region of the calibration curve. The use of the 4-parameter fit extends the usefulness of the upper and lower concentrations of the calibration curve. However, the values calculated from these upper and lower concentrations have greater error associated with them. To save on reagents, and to keep the error on the estimation of concentrations of unknowns to a minimum, concentrations for standard curves should be performed in the linear range after the complete standard curve has been defined with upper and lower asymptotes. A semi-log curve fit should then be used to fit the data to this truncated calibration curve and the absorbance values for unknowns should fall in the central portion of the linear region of this calibration curve. If a kit is being used, the package insert should indicate the standard curve analysis method to use based on the range of standard concentrations used for the calibration curve. See also tutorial 5.4. Data Analysis Guidelines.

3.6 Quality Control

There are several approaches to quality control for immunoassays. The first is to monitor the parameters of the standard curve to ensure that these remain within the desired coefficient of variation range. Second, it is important to establish relevant quality control standards (i.e. positive and negative controls). These too, should be monitored on a regular basis for variations around a determined mean. This may be evaluated for example, by construction of Shewhart charts (Wernimont & Spendley, 1985). See also tutorial 5.9. Issues in Quality Control and Quality Assurance.

3.7 Pipetting Techniques

Pipetting is an integral part of this immunochemical technology. Assuming the error derived from the specific assay design are fixed, the next largest source of error in analytical data derived from immunoassay is from pipetting errors (Li et al., 1989). Another important...