In sandwich ELISA experiments, the construction of a standard curve is the core foundation for accurately quantifying unknown samples, directly influencing the reliability and accuracy of detection results. The following key considerations must be emphasized:
Standards with high homology to the target antigen (e.g., recombinant proteins, purified natural antigens) and a purity of ≥95% should be used to avoid signal deviations caused by cross-reactivity or impurity interference. Priority should be given to standards
certified by authoritative institutions (e.g., international standards, national reference materials) to ensure batch-to-batch stability.
Dilute the standards in strict accordance with the instructions using a dedicated diluent (e.g., PBS containing BSA to prevent antigen adsorption). Adopt "serial dilution" or "continuous dilution" during the process to ensure accurate concentration gradients (usually 5–8 concentration points covering the expected sample concentration range).
Avoid repeated freeze-thaw cycles of standards. It is recommended to aliquot the standards and store them at -80°C;
discard unused portions after a single use to prevent antigen degradation or concentration changes.
The standard concentrations must include a "0 concentration" (serving as a blank calibration point) and at least 4 increasing concentration points. The span between the highest and lowest concentrations should cover the expected concentration of unknown samples (typically ranging from near the limit of detection to just before the signal plateaus). This avoids extrapolation errors caused by the OD value of unknown samples falling outside the curve range.
Concentration intervals should be uniform (e.g., 1, 5, 10, 50, 100 ng/mL) to prevent curve fitting distortion due to excessively large intervals. If preliminary experiments show non-linear signals (e.g., "S"-shaped curves), additional concentration points should be
added in the region where the signal changes sharply (e.g., EC30–EC70) to improve fitting accuracy.
Each standard concentration requires 2–3 replicate wells to reduce random errors. Calculate the mean and standard deviation (SD) of the OD values of replicate wells. If the OD value of a well deviates from the mean by more than ±10%–15% (adjusted according to experimental requirements),
it should be identified as an outlier and excluded to ensure data stability.
A "blank well" (containing only diluent and detection reagents, no antigen) must be set up. The OD values of all standards and samples should be subtracted by the OD value of the blank well to eliminate non-specific signal interference such as reagent
background and inter-plate variation.
Linear Regression: Only applicable when the standard concentration range is narrow and the OD value shows a significant linear relationship with concentration (R² ≥ 0.98). Deviations are likely to occur at high/low concentrations, leading to large errors.
4-Parameter Logistic Regression (4PL):
The preferred model for sandwich ELISA, especially when the concentration range is wide and the signal exhibits an "S"-shaped curve (rising rapidly in the first half and leveling off in the second half).
It can more accurately fit non-linear relationships (R² ≥ 0.99 is required).
Avoid arbitrary model selection; residual analysis (deviation between measured values and fitted values) should be used to determine whether the model aligns with the actual signal trend.
Correlation Coefficient (R²): R² ≥ 0.98 is required for linear regression, and R² ≥ 0.99 for 4PL. A lower R² indicates poor fitting.
Lower Limit of Quantitation (LLOQ) and Upper Limit of Quantitation (ULOQ): Defined as the lowest/highest concentrations in the standard curve that can be accurately quantified (usually the lowest concentration with CV% < 20% is taken as LLOQ). If the OD value of an unknown sample falls outside this range, extrapolation is not allowed, and the sample must be re-diluted for detection.
Standards and samples must be tested on the same microplate to ensure identical conditions such as incubation time, temperature, number of washing cycles, and chromogenic solution volume, avoiding curve drift caused by inter-plate variation.
Microplate readers should be calibrated regularly to ensure accurate OD value detection. Chromogenic substrates (e.g., TMB) must be freshly prepared to avoid activity loss due to light or temperature, which could cause abnormal signals and interfere with curve shape.
The same standard concentration gradient should be tested repeatedly in different experimental batches (at least 3 times), and the inter-batch coefficient of variation (CV%) should be calculated.
A CV% < 15% is required to ensure the stability and reliability of the standard curve. If inter-batch variation is excessive,
issues such as reagent inactivation or operational errors should be investigated.
In summary, the construction of a standard curve requires balancing "accuracy, reproducibility, and coverage." Only by strictly controlling standard quality, optimizing concentration gradients, selecting appropriate models, and standardizing operations can a reliable basis be provided for the accurate quantification of unknown samples.