Method Transfer: Why the Same Procedure Gives a Different Number in Your Hands

You run the supplier's method, on the supplier's material, and get a different number. The instinctive readings are that someone made a mistake or that the material is not what it claims. There is a third reading, and it is usually the right one: a validated procedure moved between laboratories is not the same measurement until somebody has shown that it is.

Quick answer

What is analytical procedure transfer? The act of implementing an already validated procedure at a different location. ICH Q2(R2) states that when transferring analytical procedures to a different laboratory, a partial or full revalidation of the performance characteristics and/or comparative analysis of representative samples should be performed — and that justification for not performing additional transfer experiments should be provided if appropriate.

Note the structure: the default is that evidence is generated. Not generating it is the exception that requires a justification.

The three kinds of change, and what each demands

ICH Q14 sets out a table of change scenarios with the bridging strategy each one calls for. Transfer is the mildest of the three, which is itself informative.

Even the mildest case — nothing about the written method changed, only the building it runs in — is treated as something that needs either evidence or an argument. That is the point worth carrying across to research work.

Why a procedure does not travel intact

The written method is not the measurement. The measurement is the written method plus everything the written method did not think to specify. Transfer exposes exactly those unstated things:

• The analyst. Timing, technique in transferring viscous or small volumes, how long a sample sits before injection.
• The instrument. Detector response, gradient delay volume, injector precision — all within specification, all different.
• The consumables. A different column batch is a real variable, which is why ICH Q14 explicitly allows categorical variables such as different instruments in robustness design.
• The reference. If the two laboratories calibrate against different materials, the numbers are not comparable even when both procedures are performed perfectly — the chain described in metrological traceability.
• The environment. Temperature and humidity are among the variations the guideline names for intermediate precision.

None of these is a mistake. They are the ordinary variation that intermediate precision is designed to capture and that reproducibility — the inter-laboratory level described in precision — is designed to measure.

How much disagreement is normal

The best public answer comes from a United States Pharmacopeia multi-laboratory collaborative study on the nonapeptide oxytocin, which compared an HPLC assay, quantitative NMR and amino acid analysis across national metrology and regulatory institutes in several countries. The study exists because inter-laboratory agreement was worth establishing rather than assuming — and the detail of what it found, including the uncertainty term that was excluded from the headline figure, is covered in how much quantification methods agree.

If institutes of that calibre produce enough spread on a well-characterised peptide to be worth publishing, a difference between your bench and a supplier's certificate is not evidence of anything by itself.

Co-validation: the efficient version

ICH Q2(R2) describes an alternative that avoids doing the work twice. Co-validation can be used to demonstrate that the procedure meets predefined performance criteria by using data generated at multiple sites, and could also satisfy the requirements of analytical procedure transfer at the participating sites.

The glossary definition is worth having: co-validation is demonstration that the procedure meets its predefined performance criteria when used at different laboratories for the same intended purpose, and can involve all or a subset of the performance characteristics potentially impacted by the change in laboratories.

In other words, validate once across both sites rather than validating in one and then transferring to the other.

What a research buyer can actually do

You will not run a formal transfer study. But the structure tells you what a disagreement means and what resolves it.

• Treat a supplier-versus-bench difference as a transfer problem first, not as a material problem. It is the more likely explanation and the cheaper one to test.
• Compare on the same sample, not on the same lot. The guideline's own instrument is comparative analysis of representative samples. Two measurements of one vial are informative; two measurements of two vials confound material with method.
• Fix your reference before investigating anything else. If you and the supplier calibrate against different things, no amount of care in technique will make the numbers converge.
• Know your own intermediate precision before calling a difference real. A gap smaller than your own day-to-day scatter is not a finding.
• Record which method produced which number. A value without its procedure cannot be re-examined later, by you or anyone else.

Key takeaways

• ICH Q2(R2) treats transfer to a different laboratory as requiring partial or full revalidation and/or comparative analysis of representative samples.
• Not performing transfer experiments is permitted but requires a justification — the default is evidence.
• ICH Q14 grades three change scenarios: change of principle, change within a principle, and transfer with no change to the procedure.
• Even transfer with no procedural change is treated as an event needing evidence or an argument.
• A written method plus unstated local conditions is what actually produces a number; transfer exposes the unstated part.
• Co-validation generates data at multiple sites at once and can satisfy the transfer requirement directly.
• A difference between a supplier's figure and your own is more likely a transfer effect than a material defect.

Frequently asked questions

My result differs from the certificate by 8%. Is the material wrong?

Not demonstrably. Before that conclusion is available you would need to know your own measurement's scatter, whether both measurements used the same reference, and whether the methods share a measurement principle. Any one of those can account for a gap that size.

What is comparative analysis of representative samples?

Both laboratories measuring the same material and comparing results. It is the simplest form of transfer evidence and the one most available outside a regulated setting.

Does a supplier have to do transfer studies?

Research-grade supply operates outside the regulatory framework these guidelines govern, so no. The guidelines are useful here as a description of what rigorous practice looks like, not as a standard anyone is obliged to meet.

Is this the same as lot-to-lot bridging?

Related but distinct. Bridging as described in lot-to-lot variability connects two materials. Transfer connects two laboratories running one method. Confusing them is how a method difference gets recorded as a material difference.

Why would anyone skip transfer experiments?

Where prior knowledge is strong and the risk is low — a simple, well-understood procedure moving to a site already running similar work. The guideline permits it with justification rather than silently.

Sources

• International Council for Harmonisation. ICH Harmonised Guideline Q2(R2): Validation of Analytical Procedures. Final version adopted 1 November 2023. ICH guideline
• International Council for Harmonisation. ICH Harmonised Guideline Q14: Analytical Procedure Development. Final version adopted 1 November 2023. ICH guideline
• Li C, Bhavaraju S, Thibeault MP, Melanson J, Blomgren A, Rundlöf T, et al. Survey of peptide quantification methods and comparison of their reproducibility: a case study using oxytocin. J Pharm Biomed Anal. 2019;166:105–112. DOI

Journal citation retrieved via PubMed. Guideline text is quoted from the adopted ICH documents.

This article describes analytical methodology only. It contains no usage instructions, no dosing information and no protocols of any kind.

For laboratory and in-vitro research use only. Not for human or veterinary use.