LBA Cross Validation: A Detailed Case Study

Validation vs Cross Validation

Monoclonal antibody-based drugs have revolutionized the treatment of cancer and autoimmune diseases. Before a new antibody drug can enter human clinical trial, a method must be developed to enable the detection and quantitation of the drug in serum for pharmacokinetics (PK) studies. The main quantitation method is ligand binding assays (LBAs), with ELISAs and the MSD platform being some of the most common.

Many pharmaceutical companies develop and validate in-house LBA methods for their antibody drugs during preclinical development [1, 2]. However, due to the large number of clinical samples to be tested, scheduling demands, or capacity limitations, the method may need to be transferred from the original laboratory to a contract research organization (CRO). If no major changes occur in the assay platform, format, or detection system, a cross validation (also called partial validation) can be performed by the receiving lab instead of a full validation. Table 1 summarizes the circumstances and parameters required in a full versus a cross validation.

Table 1. Comparison of full validation versus cross validation. *Parameter may be tested if applicable.

In this case study we will examine the cross validation for a two-component antibody drug substance. The antibody drugs, denoted as FN and KG, are formulated in a 1:1 ratio as drug substance. Our goal in this blog is to provide a real-life example of a cross validation study, both from a regulatory perspective, but also highlight practical insights and recommendations for laboratory testing.

Getting started

The first step in cross validation for LBAs is the transfer of existing documentation from the sponsor’s lab to the CRO, followed by initial R&D assay runs with QCs and calibration curves. These preliminary test runs confirm that no unexpected results occur during the method transfer. While some may consider skipping this for straightforward ELISA assays, early R&D testing ensures assay robustness and helps identify gaps in method instructions or instrument setup between labs.

In the FN and KG methods, the blocking step incubation time in the original protocol ranged from 1–4 hours. R&D runs revealed high variability depending on the chosen incubation time. After review with the sponsor’s lead scientist, it was discovered that the original analyst consistently used ~90 minutes. This discrepancy between written instructions and actual practice significantly impacted reproducibility. The revised cross-validation protocol therefore specified 90 ±10 minutes for blocking, illustrating how small procedural details can strongly affect assay robustness.

Initial R&D runs should ideally use the same critical reagents as cross validation testing, including drug substance, pooled human serum, coating antigen, or detection antibodies. Per ICH M10 bioanalytical validation guidance [2], critical reagents must be clearly defined in the method. For FN and KG, antibody reference standards were received, and concentrated calibration standards and QCs were prepared, then diluted for cross validation. Table 2 summarizes the calibration standards and QC concentrations.

The full list of QCs and calibration standards, including their names and concentration for each drug component, are listed below in Table 2.

* = non-calibrator anchor points for both FN and KG ELISAs, ** = non-calibrator anchor point for FN ELISA only, 1 = LLOQ for FN ELISA, 2 = LLOQ for KG ELISA

Table 2. Calibration Standards and Quality Controls for FN and KG ELISAs.

You may have noticed that there are two different LLOQ QCs: one at 100 ng/mL for Antibody FN, and one at 60 ng/mL for Antibody KG. Fortunately, besides the differences in quantitation range and antigen binding target the two antibodies utilize the same ELISA workflow and many assay reagents. This simplified the cross-validation process, both in the lab and in terms of quality documentation, as a single test method, cross validation protocol, and report were generated for this study.

Below we will go over in detail the cross-validation parameters tested:

Sensitivity, Calibration Curve, Accuracy & Precision (A&P), and Incurred Sample Analysis (ISR).

Cross validation parameters

The specifications for this validation were adopted from the Ligand Binding Assay portions of the FDA’s “Guidance for Industry: Bioanalytical Method Validation” and ICH’s “Guideline M10 on Bioanalytical Method Validation” [1, 2]. The instrument used for cross validation was a qualified BioTek (now a part of Agilent) Synergy H1 microplate reader. System suitability test was performed prior to each assay run. A total of seven assay runs were performed. Of these, six were for Sensitivity, Calibration Curve, and Accuracy & Precision (A&P), and one was for Incurred Sample Reanalysis. The six assay runs were performed by two analysts. Although FDA and ICH guidance do not specify the number of analysts for bioanalytical validation, having multiple analysts perform the testing allow for the assessment of robustness and possible bias [3].

The “sample” tested in this cross validation have two different definitions: actual clinical samples (for the ISR) and quality control (QC) samples at different levels prepared from the reference standard.  Table 3 below shows a sample plate map for the six Sensitivity, Calibration Curve, and A&P runs.

*=LLOQ concentration value is different for FN and KG.

Table 3. Sample plate map of FN and KG ELISA runs.

Sensitivity

Test parameters:

The Sensitivity of the method was determined by preparing 3 independent LLOQ quality control (QC) samples in technical duplicates (see Table 3). Sensitivity assessment was performed 6 times: 3 assays for each of two analysts on different days. The mean of the technical duplicates was calculated, and the mean became the single value for each LLOQ QC sample. % Bias, % CV, and Total Error were calculated based on the 18 independent LLOQ QC samples. These values were then compared to the acceptance criteria for Sensitivity: % Bias within ± 25%, % CV ≤ 25%, and Total Error ≤ 40%.

Results:

• The % Bias, % CV, and Total Error of FN ELISA LLOQ QC samples was 0%, 6%, and 6%, respectively.
• The % Bias, % CV, and Total Error of KG ELISA LLOQ QC samples was 4%, 11%, and 15%, respectively.

Conclusion:

The LLOQ QC for both FN and KG ELISAs meet the acceptance criteria for Sensitivity.

Calibration Curve

Test parameters:

One sample of each 13-calibration standard—including the Blank, ULOQ, LLOQ, and anchor points—in technical duplicates were prepared (see Table 3). The Calibration Curve assessment was performed 6 times: 3 assays for each of two analysts on different days. The concentration-response relationship using four-parameter logistic model with 1/y2 weighting. The mean of the technical duplicates was calculated, and the mean became the single value for each calibration level. % Bias and % CV were calculated for each non-anchor point calibrator.  These values were compared to the accuracy criteria (% Bias ± 20% for calibrator and ± 25% for LLOQ and ULOQ) and the precision criteria (i.e., % CV ≤ 20%, and ≤ 25% at LLOQ and ULOQ) for calibrators.

Results:

• For FN ELISA, the highest non-anchor point % Bias value was 8% and the highest non-anchor point % CV was 4%.
• For KG ELISA, the highest non-anchor point % Bias value was 9% and the highest non-anchor point % CV was 4%.

Conclusion:

All non-anchor point calibrators for the FN and KG ELISAs fulfilled the accuracy and precision criteria.

Accuracy & Precision (A&E)

Test Parameters:

A&E (for within- and between-runs) was determined by preparing 3 independent samples each of LLOQ, LQC, MQC, HQC, and ULOQ, in technical duplicates (see Table 3). A&E was performed 6 times: 3 assays for each of two analysts on different days. The mean of the technical duplicates was calculated, and the mean became the single value for each QC level. For within-run A&E, % Bias, % CV, and Total Error were calculated based on the 3 independent samples for each QC level within a run. For between-run A&E, % Bias, % CV, and Total Error were calculated based on the 18 independent samples for each QC level. Within-run and between-run performance were assessed using the following % Bias, % CV, and Total Error acceptance criteria:

Accuracy: % Bias within ±20% for LQC, MQC, and HQC (±25% at LLOQ and ULOQ)

Precision: %CV ≤20% for LQC, MQC, and HQC (≤25% at LLOQ and ULOQ)

Total Error: ≤30% for LQC, MQC, and HQC (≤40% at LLOQ and ULOQ)

Results:

• FN ELISA: Across all QC levels, both within- and between-run data met acceptance criteria. A single outlier was noted at the ULOQ in one of the six assay runs, which initially caused higher % Bias and Total Error. After excluding this anomalous value, results fell well within criteria.
• KG ELISA: All QC levels also met acceptance requirements for both within- and between-run evaluations. Variability was slightly higher at the LLOQ but remained within protocol limits.

Conclusion:

Both FN and KG ELISAs demonstrated strong accuracy & precision across the full calibration range, from LLOQ to ULOQ. The assays were robust in both within- and between-run analyses, confirming their reliability for quantitative measurements.

Incurred Sample Reanalysis (ISR)

Test Parameters:

For Incurred Sample Reanalysis (ISR), 15 serum samples were received from the sponsor and tested to determine concentration of FN and KG antibodies. Each sample was prepared in technical duplicate. The value obtained were compared to previously obtained data from the sponsor and the % Difference was calculated. The acceptance criteria are that the % Difference cannot exceed 30%, and that at least 67% of samples must pass.

Results:

• In the FN ELISA run, the % Difference of 12 out of 15 samples (i.e., 80%) met the acceptance criteria, therefore the FN ELISA passes ISR.
• In the KG ELISA run, the % Difference of all 15 samples (i.e., 100%) met the acceptance criteria, therefore the KG ELISA passes ISR.

Conclusion:

Both FN and KG ELISAs passed ISR analysis.

Cross validation report

Once all cross validation and ISR runs have been performed and reviewed by the sponsor and QA, the results were coalesced into a cross validation report. This report contained not only the data and common elements of scientific reports (background, data, results & conclusion, references) but also lists the critical reagents, equipment used, and excursions that occurred during the course of the study. For FN and KG, the cross validation was successful and subsequently 300+ clinical serum samples were analyzed by the Emery Pharma team.

If you are looking for a qualified lab to transfer your bioanalytical method to, look no further than Emery Pharma. With our team of experienced scientists and rigorous quality system, we are your ideal partner in ELISA method validation. Contact us online or call us at +1 (510) 899-8814!

References

1) Guidance for Industry: Bioanalytical Method Validation, U.S. Department of Health and Human Services, Food and Drug Administration, May 2018.

2) ICH Guideline M10 on Bioanalytical Method Validation, International Council for Harmonization of Technical Requirements for Pharmaceuticals for Human Use, May 2022.

3) Large Molecule Specific Assay Operation: Recommendation for Best Practices and Harmonization from the Global Bioanalysis Consortium Harmonization Team. AAPS Journal. 2013 Nov 16;16(1):83–88. doi: 10.1208/s12248-013-9542-y