Spike-and-recovery and linearity-of-dilution experiments are important methods for validating and assessing the accuracy of ELISA. Spike and recovery is used to determine whether analyte detection is affected by differences in the standard curve diluent and biological sample matrix. Sample matrix is either a neat (undiluted) biological sample or a mixture of the biological sample with sample diluent.

Linearity of dilution refers to the extent in which a spike or natural sample’s (in a particular diluent) dose response is linear and in the desired assay range. Spike and recovery and linearity of dilution are related. Experiments can be designed to test both simultaneously. The following discussion considers each one separately.


Learn more

View products


Spike and recovery

The meaning and purpose of spike-and-recovery assessment

In spike and recovery, a known amount of analyte is added (spiked) into the natural test sample matrix. Then the assay (here assumed to be ELISA) is run to measure the response (recovery) of the spiked sample matrix compared to an identical spike in the standard diluent.

The ELISA method involves comparison of test samples to a standard curve prepared using known concentrations of the analyte (e.g., purified recombinant protein). The goal in assay development is to maximize signal-to-noise ratio while achieving identical responses for a given amount of analyte in standard diluent (the standard curve) and sample matrix (biological sample + sample diluent). The sample matrix may contain components that affect assay response to the analyte differently than the standard diluent. A spike-and-recovery experiment is designed to assess this difference in assay response.

Performing a spike-and-recovery experiment

  1. A known amount of analyte is added to the sample matrix and standard diluent.
  2. The two sets of responses are compared based on values calculated from a standard curve.
  3. If the recovery observed for the spike is identical to the analyte prepared in standard diluent, the sample matrix is considered valid for the assay procedure. If the recovery differs, then components in the sample matrix are causing the difference, and adjustments must be made to the method to minimize the discrepancy.

Correcting for poor spike-and-recovery results

Two kinds of adjustments can be made to re-optimize an ELISA when a spike-and-recovery experiment detects a discrepancy.

  1. Alter the standard diluent. Use a standard diluent whose composition more closely matches the final sample matrix. For example, culture medium could be used as the standard diluent if the samples will be culture supernatants. If the standard diluent had previously been optimized for signal-to-noise performance, changing the diluent to match the sample matrix performance will result in decreased assay range, sensitivity or signal-to-noise ratio. In some cases, a compromise may be necessary.
  2. Alter the sample matrix. If neat biological samples had been used, retest it upon dilution in standard diluent or other logical “sample diluent”. For example, if an undiluted serum sample produces poor spike and recovery, perhaps one that is diluted 1:1 in standard diluent will work better. If the level of analyte in the diluted sample is sufficient to be detected by the assay, this method will correct many recovery problems.

Better results for a sample matrix may be obtained by altering its pH (to match the optimized standard diluent) or by adding BSA or other purified protein as a carrier/stabilizer. Be aware that the best sample diluent will not necessarily be the same as the best standard diluent. For example, serum samples contain considerable background protein (e.g., albumin and immunoglobulins). But, the purified recombinant protein used as a standard may not contain any carrier protein. In this case, the best standard diluent may be phosphate-buffered saline (PBS) that contains 1% BSA and the best sample diluent for the serum may be PBS without any additional protein.


Linearity of dilution

The meaning and purpose of linearity-of-dilution assessment

A linearity-of-dilution experiment provides information about the precision of results for samples tested at different levels of dilution in a chosen sample diluent. Linearity is defined relative to the calculated amount of analyte based on the standard curve, not relative to the raw absorbance measurements (the best fit standard curve usually is not linear).

If the linearity is good over a wide range of dilutions, then the assay method provides flexibility to assay samples with different levels of analyte (i.e., a sample with high levels of analyte can be diluted several-fold to ensure that its values fall within the standard curve range and compared to a low-level sample that is assayed without dilution).

Performing a linearity-of-dilution experiment

There are two ways to perform a linearity-of-dilution experiment.

  1. Traditional method. The traditional method involves using a low-level sample containing a known spike of analyte, and then testing several different dilutions of that sample in the chosen sample diluent.
  2. Alternative method. An alternative method involves first preparing several dilutions of a low-level sample and then spiking the same known amount of analyte into each one before testing. Assay recovery is assessed by comparing observed vs. expected values based on non-spiked and/or neat (undiluted) samples.

Interpreting linearity-of-dilution results

Poor linearity of dilution indicates that the natural sample matrix, the sample diluent and/or standard diluent affect analyte detectability differently. This difference may be caused by dilution of components in one solution that inhibits or enhances detection in the assay method compared to the other solutions. The causes of poor linearity of dilution are, therefore, related to the same causes of poor spike and recovery.

The goal in either case is to have equality between the standard diluent and sample diluent. If desired, a single experiment can be performed using a checkerboard matrix of spike levels, sample types, sample diluents and dilution factors to simultaneously assess both spike and recovery and linearity of dilution.


Example data

Table 1. ELISA spike and recovery of recombinant human IL-1 beta in nine human urine samples using the Novex™ IL-1 beta ELISA Kit, Human. Samples were assayed by adding 50µL of sample and 10µL of spike stock solution calculated to yield the intended 0, 15, 40 or 80pg/mL spike concentration. Values reported for spiked samples reflect subtraction of the endogenous (no-spike) value. Recoveries for spiked test samples were calculated by comparison to the measured recovery of spiked diluent control. Diluent for the diluent control, spike stock solutions and standard was the same. All values represent the average of three replicates.

SampleNo spike
(0pg/mL)
Low spike
(15pg/mL)
Medium spike
(40pg/mL)
High spike
(80pg/mL)
Diluent control0.017.044.181.6
Donor 10.714.639.669.6
Donor 20.017.841.674.8
Donor 30.615.037.668.9
Donor 40.015.136.967.8
Donor 5
0.512.533.563.6
Donor 6
0.014.033.568.7
Donor 7
0.014.438.569.6
Donor 87.116.341.469.5
Donor 9
0.712.437.668.2
Mean recovery
(+/- S.D.)
NA86.3%
+/- 9.9%
85.8%
+/- 6.7%
84.6%
+/- 3.5%


Table 2. Typical presentation for summarizing spike-and-recovery results.
(Data from Table 1.)

Sample (n)Spike levelExpectedObservedRecovery %
Urine (9)Low (15pg/mL)17.014.786.3
Med (40pg/mL)44.137.885.8
High (80pg/mL)81.669.084.6


Table 3. ELISA linearity-of-dilution results for three human IL-1 beta samples using the IL-1 beta ELISA Kit, Human.
The best sample diluent was chosen from the previous spike-and-recovery experiment (Table 1 and 2) and four dilutions were made for each sample. Observed values were assessed relative to the assay standard curve produced by an IL-1 beta standard.

SampleDilution factor (DF)Observed (pg/mL)
× DF
Expected
pg/mL
(neat value)
Recovery %
ConA-stimulated cell culture supernatantNeat131.5131.5100
1:2149.9114
1:4162.2123
1:8165.4126
High-level serum sampleNeat128.7128.7100
1:2142.6111
1:4139.2108
1:8171.5133
Low-level serum sample spiked with recombinant IL-1 betaNeat39.339.3100
1:247.9122
1:450.5128
1:854.6139

For Research Use Only. Not for use in diagnostic procedures.