Addressing The Challenges of Biomarker Calibration Standards in Ligand-Binding Assays: A European Bioanalysis Forum Perspective

Figures & data

Table 1. Various concepts to normalize concentration of standard.

Normalization concepts	Samples	Experiments	Evaluation	Recommendation
A Test versus reference calibrator sample	Test sample in calibrator matrix, prepared from new lot calibration standard (mid of calibration range) Reference sample in calibrator matrix, prepared from original lot calibration standard or reference/international standard (same level as test sample)	Measurement of test and reference sample in three to five independent replicates (independent dilutions) on 2–3 days/runs (n = 6–15) together with original QC set versus new calibration samples (prepared from new lot)	New concentration (new lot) = mean test/mean reference * label concentration (new lot), If % CV of QC results (n = 6–15) ≤ precision of assay and QC results are parallel to calibration curve and mean test sample is significant different from mean reference sample (e.g., by equivalence test)	Recommended for in-house developed assays to check change of calibration standard lot, only possible approach for normalization versus international standard
B Matrix samples versus calibration curves	Incurred matrix samples as test samples (e.g., QCs original and/or new, monitoring samples/sample controls)	Measurement versus calibration samples prepared from original lot 1 and versus calibration samples prepared from lot 2 (multiple batches and replicates)	New concentration (new lot) = mean result (vs calibration samples from lot 2)/mean results (vs calibration samples from lot 1) * label concentration (new lot)	Recommended if not only lot of calibration standard changes but also kit lot which may also include a change of lots of critical reagents
C New versus original calibration curve	Test calibration samples (whole curve range) from new lot, reference calibration samples (curve) from original lot	Based on performance of calibration samples. Read out with both new and original calibration samples. New calibrations samples measured versus original calibration samples	New concentration (new lot) = label concentration (new lot)/mean recovery new calibration samples, If calibration curves are parallel Confirmed by equivalence test with calibration samples between runs with original standard and runs with new standard	Recommended when no international standard or other external standard is available
D Linear correlation	30 or more incurred samples (e.g., remaining study samples)	Measurement of samples with original and new lot (calibration standard or kit)	Linear correlation of results (new lot vs original lot). 1/slope is used as normalization factor to adjust new calibration standard/kit

QC: Quality Control.

Table 2. Example for normalization concept A.

Nominal value	Code	Concentration	Unit	% deviation from nominal	Mean concentration	% CV	NF
140	WHO_1	187	ng/mL	34
140	WHO_2	179	ng/mL	28
140	WHO_3	187	ng/mL	34
140	WHO_4	190	ng/mL	36
140	WHO_5	178	ng/mL	27	184	2.9%
140	STD_1	136	ng/mL	-3
140	STD_2	129	ng/mL	-8
140	STD_3	145	ng/mL	4
140	STD_4	135	ng/mL	-4
140	STD_5	140	ng/mL	0	137	4.3%	0.744

NF: Normalization factor.

Table 3. Example for normalization concept B.

Nominal value	Code	Concentration	Unit	% deviation from nominal	Mean concentration	NF (new/old)	% CV (NF)	Mean NF
20	LOW-QC_OLD STD_1	19.6	ng/mL	-2
20	LOW-QC_OLD STD_1	16	ng/mL	-20
20	LOW-QC_OLD STD_1	18.3	ng/mL	-9	18.0
20	LOW-QC_NEW STD_1	33.6	ng/mL	68
20	LOW-QC_NEW STD_1	29.3	ng/mL	47
20	LOW-QC_NEW STD_1	30.9	ng/mL	55	31.3	1.74
100	MID-QC_OLD STD_1	98.4	ng/mL	-2
100	MID-QC_OLD STD_1	87.9	ng/mL	-12
100	MID-QC_OLD STD_1	96.6	ng/mL	-3	94.3
100	MID-QC_NEW STD_1	145	ng/mL	45
100	MID-QC_NEW STD_1	133	ng/mL	33
100	MID-QC_NEW STD_1	138	ng/mL	38	138.7	1.47
300	HIGH-QC_OLD STD_1	298	ng/mL	-1
300	HIGH-QC_OLD STD_1	285	ng/mL	-5
300	HIGH-QC_OLD STD_1	275	ng/mL	-8	286.0
300	HIGH-QC_NEW STD_1	423	ng/mL	41
300	HIGH-QC_NEW STD_1	445	ng/mL	48
300	HIGH-QC_NEW STD_1	431	ng/mL	44	433.0	1.51	9.2%	1.575

New concentration = mean NF × new lot nominal standard concentration (1.575 × 1.00 mg/mL = 1.575 mg/mL).

NF: Normalization factor.

Table 4. Example for normalization concept C.

Nominal value	Code	Concentration	Unit	% deviation from nominal	NF (NEW/OLD)	% CV (NF)	Mean NF
10	NEW STD_1	<LLOQ	ng/mL	–	–
20	NEW STD_2	15.2	ng/mL	-24	0.760
40	NEW STD_3	30.9	ng/mL	-23	0.773
80	NEW STD_4	64.8	ng/mL	-19	0.810
160	NEW STD_5	128	ng/mL	-20	0.800
320	NEW STD_6	251	ng/mL	-22	0.784
640	NEW STD_7	540	ng/mL	-16	0.844	3.8%	0.795

New concentration = Mean NF × New lot nominal standard concentration (0.795 × 1.00 mg/mL = 0.795 mg/mL).

NF: Normalization factor.

Figure 1. Example for normalization concept D.

Comparing incurred matrix samples measured against new and old calibration curves by linear regression analysis. The slope of 0.862 was obtained from the linear regression analysis. In this case, the original label concentration of the new lot would be multiplied by a factor of 1.16 (1/0.862).

Table 5. Experimental approaches to guarantee long-term comparability of calibration standards.

No.	Tool/approach	Advantages	Challenges, limitations	Suggested for validation levels	Requirements	Normalization/trending
1	Retrospective risk analysis and management Use several lots of calibration standard without any a priori comparison or correction strategy	No additional work required	Not science based Risk of noncomparable results Introduction of bias	Exploratory biomarker Basic validation level	None	Trending of QC results may help support this approach
2	Use of one lot only within expiry date Use of one lot of calibration standard within a particular project or study within expiry date of vendor	No risk of lot-to-lot variability No additional work required	Only suitable for short studies May be difficult to compare results across different studies where different lot numbers have been used Risk of study extension resulting in calibration standard falling outside of expiry	Exploratory biomarker Basic or advanced validation level	Purchase sufficient material at right time Careful timing of order and sample measurement	Not applicable
3	Use of one lot only ignoring expiry date Banking (at vendor or in-house) and use of one lot of calibration standard across numerous studies Storage of standards at ≤-80°C or in lyophilized form ignoring the expiry date from the vendor	No risk of lot-to-lot variability	High frontloading costs For commercial immunoassay kits it may only be possible if a separate calibration standard is available Requires assumption of long-term stability	Exploratory biomarker, advanced validation level	Purchase sufficient material, store in aliquots Monitoring of assay performance (sample or QCs – analyte concentration or assay response)	May be beneficial to check results against reference population range Trend analysis of banked monitoring samples
4	Use of multiple lots with normalization to the previous lot	No banking of reference standard required Commercially supplied kit material can be used	Long-term trends may be ignored Requires multiple bridging experiments and normalization strategy with available standard from previous lot Considerable work should there be several lot changes over course of study/project	Exploratory or confirmatory biomarker, advanced or full validation level	Performance of several batches using both new and original lots of calibration standard Possible inclusion of independent quality controls or incurred samples	Normalization concept required – suggested B (Table 1) Trending of QC data subsequent to normalization is recommended
5	Use of multiple lots with normalization to a banked reference standard lot or international standard	Kit material can be used	Possible long-term stability issues with banked reference standard Requires multiple bridging experiments and normalization strategy which takes into account whether it is possible to prepare a full calibration curve from an international standard	Confirmatory biomarker, full validation level	Performance of several batches using both new and original lots of calibration standard and international standard Possible inclusion of independent quality controls or incurred samples	Normalization concept required – suggested A (Table 1) Trending of QC data subsequent to normalization is recommended

QC: Quality control.

Figure 2. Decision tree for choice of an appropriate approach regarding handling lot-to-lot variability of calibration standard.

Figure 3. Case study: monitoring of quality control performance (trending).

Quality controls failed after the switch to a new calibration standard lot B as they exceeded acceptance limits. Using a normalization concept runs measured versus lot B would have passed.