What a Method Transfer Has to Prove

USP General Chapter <1224> defines the transfer of analytical procedures as the documented process that qualifies a receiving laboratory to use a procedure that originated in another laboratory, so that the receiving unit has both the procedural knowledge and the ability to run it as intended.1 That definition is narrower than it first appears. It is about the laboratory, not the product. USP notes that a single lot can be enough for a transfer study, because the goal is to evaluate how the procedure performs at the new site, not to characterize the manufacturing process.1

The legal basis in the United States is short. Under 21 CFR 211.194(a)(2), laboratory records must show where the data are that establish the method meets proper standards of accuracy and reliability for the product, and the regulation adds: “The suitability of all testing methods used shall be verified under actual conditions of use.”2 A method that was validated in a development lab in one state has not been verified under actual conditions of use at a contract lab on another continent until someone has shown it works there.

In Europe the expectation is written out in more detail. Since the 2014 revision of EU GMP Chapter 6, sections 6.37 to 6.41 cover technical transfer of testing methods. Before the transfer starts, the transferring site should confirm that the method matches the marketing authorization, review the original validation against current ICH requirements, and document a gap analysis that identifies any supplementary validation needed. The transfer is then described in a detailed protocol, deviations are investigated before the transfer is closed, and the report documents the comparative outcome and names any areas that need further revalidation.3

FDA’s 2015 guidance on analytical procedures and methods validation describes transfer studies in similar terms: a preapproved protocol with predetermined acceptance criteria, two or more laboratories testing a sufficient number of representative articles from the same lots, and comparative work focused on accuracy and precision, especially interlaboratory variability. For a stability-indicating method, the guidance says forced degradation samples or samples containing relevant product-related impurities should be analyzed at both sites.4

Comparability Means the Same Reportable Result

It helps to be precise about what is being compared. ICH Q2(R2) defines the reportable result as the result generated by the procedure after calculation or processing and after applying the described sample replication.5 That is the number that goes on the certificate of analysis, into the stability report, and into the annual product review. It is the output of a chain that starts at sample preparation and ends in software: the instrument acquires a signal, the chromatography data system (CDS) integrates it, a calculation turns peak areas into a percentage, the LIMS averages replicates and rounds the value, and the LIMS compares it to a limit.

A transfer that compares only the chemistry, for example by having both labs hand-calculate results from their own printouts, leaves most of that chain untested. Two labs can agree on peak areas and still disagree on the reportable result because their LIMS round differently or average replicates in a different order. This is the central point of this article: comparability is a property of the whole measurement system, and the transfer has to test the whole system.

Who Owns the Outcome

USP <1224> divides the work. The transferring unit supplies the procedure, the reference standards, the validation reports, and training. The receiving unit supplies qualified staff, makes sure instruments are calibrated and qualified, and confirms the work complies with regulations and its own procedures. Both units compare the data and discuss deviations.1 When the receiving lab is a CRO or CDMO, the sponsor keeps responsibility for the result regardless of how the work is split, so the sponsor’s analytical and quality teams need to be able to read and challenge the transfer data, not only sign the report.

4Transfer approaches described in USP <1224>: comparative testing, co-validation, revalidation, and waiver [1]
5Minimum protocol parameters listed in EU GMP Chapter 6, section 6.39 [3]
6Common method transfer findings listed by MHRA inspectors in their 2021 guidance post [6]

Choosing the Transfer Approach

USP <1224> describes four ways to qualify a receiving lab, and says the choice should rest on a risk analysis that considers the receiving unit’s experience, the complexity of the product, and the procedure itself.1 ICH Q14 points the same way. Its table of change examples treats moving a validated procedure to a different site with no change to the procedure as a change that can be supported by partial or full revalidation, by comparative analysis of representative samples and reference materials, or by a justification for not performing additional transfer experiments.7

Each approach produces a different kind of evidence, and that matters for comparability later. The table below sets out what each one gives you and where its blind spots are.

ApproachWhat it involvesComparability evidence it producesWhere risk hides
Comparative testingBoth labs analyze a predetermined number of samples from the same lot under a preapproved protocol with predefined acceptance criteria [1]Direct, paired evidence that the two labs agree on the same materialAcceptance criteria that are too loose or tested the wrong way; samples that do not stress the method (no impurities near the limit)
Co-validationThe receiving lab joins the validation team and generates reproducibility data during the original validation [1]Interlaboratory precision built into the validation package; ICH Q2(R2) says this can also satisfy transfer at the participating sites [5]Only works if the receiving lab is chosen early; later instrument or software changes at either lab are not covered
RevalidationThe receiving lab performs full or partial validation, addressing the characteristics likely to be affected by the transfer [1]Shows the method performs at the new site, but not necessarily that it gives the same answer as the original siteNo direct comparison unless the protocol adds one; a bias between sites can pass unnoticed
Transfer waiverNo comparative data, justified in writing (for example, an unchanged compendial procedure, a method close to one already in use, or key staff moving with the method) [1]None at the time of transferThe first evidence of agreement comes from routine results, after product decisions have already been made

Comparative Testing Is the Default for a Reason

MHRA inspectors have said that comparative testing is the type of transfer they see most often on inspection.6 It is popular because it produces the most direct answer to the comparability question. Both labs test the same material, and the difference between their results is the thing you want to know. USP allows spiked samples, which is useful when production lots do not contain impurities at levels that make a meaningful comparison possible.1

Revalidation Answers a Different Question

Revalidation at the receiving site shows that the method is accurate, precise, and specific there. It does not, by itself, show that the receiving lab’s results line up with the transferring lab’s results. A method can be accurate within its validated limits at both sites and still carry a consistent offset between them. If you choose revalidation, for example because the receiving lab uses a different instrument platform, consider adding a small comparative arm on a shared lot so that you have at least one direct measure of the difference between sites.

Waivers Need a Plan for the First Real Data

A waiver is legitimate when it is justified and documented. USP’s examples include a product similar in composition to one the lab already tests, a procedure that is unchanged from the USP-NF (where verification applies instead), a procedure close to one already in use, and the movement of staff who developed or routinely ran the method.1 The risk is that a waiver produces no comparative data at all. A sensible control is to define, in the waiver justification, an early check: for example, the first few lots tested at the new site are also tested at the original site, or a retained sample with a known result is run in the first routine sequence.

Four questions to settle before choosing an approach

  • Has the receiving lab run this technique on this product type before, on the same instrument platform and CDS?
  • Is the method stability-indicating, and do the available lots contain the impurities or degradants the method is meant to detect?
  • How close does the product run to its specification limits, and how much bias between sites could the margin absorb?
  • Will the receiving lab release product, test stability, or both? Stability data compared across sites over years is less forgiving of a small offset than a single release test.

Writing Acceptance Criteria That Mean Something

The protocol is where comparability is won or lost. EU GMP section 6.39 lists the minimum content: the tests and methods being transferred, additional training needs, the standards and samples to be tested, any special transport and storage conditions, and acceptance criteria based on the current validation study and ICH requirements.3 USP <1224> adds objective, scope, responsibilities, materials and instruments, experimental design, and how deviations will be handled, and says acceptance criteria should account for variability, for example through a relative standard deviation for each site.1

Before any of that is written, do the gap analysis that EU GMP section 6.37 calls for. Confirm that the method text matches what is filed, check that the original validation still meets current ICH expectations, and list any supplementary validation needed.3 MHRA makes the same recommendation and adds that receiving labs should confirm they have the staff and testing capacity before the transfer begins.6 A transfer built on a weak validation will only confirm that two labs can reproduce a weak method.

Test for Equivalence, Not for a Difference

Many transfer protocols still compare the two labs with a two-sample t-test and call the transfer successful if the test finds no statistically significant difference. That design has two problems. When both labs are very precise, the t-test can flag a difference that is far too small to matter for product quality. When the data are noisy or the sample size is small, the t-test lacks the power to detect a difference that does matter, so a poor transfer passes because the data cannot prove the labs are different.

An equivalence test reverses the question. Instead of asking whether the labs are different, it asks whether the difference between them is small enough, inside a margin set in advance. The standard method is the two one-sided tests procedure (TOST). A 2026 Pharmaceutical Technology article on the power of TOST with unequal variances notes that, for method transfers, equivalence testing with TOST has been found more appropriate than the two-sample t-test, and that TOST controls the type I error, which in a transfer is the risk of accepting an unsuccessful transfer, at 5 percent.8 In practice, TOST at the 5 percent level is the same as checking that the 90 percent confidence interval for the difference in means falls entirely inside the equivalence margin.

An illustration, with numbers chosen only to show the logic. Suppose an assay specification of 95.0 to 105.0 percent of label claim, and a protocol that sets the equivalence margin for the difference in site means at plus or minus 2.0 percentage points. If the receiving lab’s mean is 0.5 points lower than the transferring lab’s, and the 90 percent confidence interval for the difference runs from minus 1.3 to plus 0.3, the interval is inside the margin and the transfer passes. Now suppose the receiving lab’s results are much more scattered, so the interval runs from minus 2.8 to plus 1.8. A t-test would likely find no significant difference and pass the transfer. The equivalence test fails it, correctly, because the data cannot show that the labs agree within 2.0 points.

Setting the Margin

The equivalence margin is a judgment that the protocol has to defend, so write down the reasoning. Three inputs usually drive it:

  • Room in the specification. How far is the typical result from the nearest limit? A bias that uses up a large part of that distance raises the chance that the receiving lab will report out-of-specification results on product that is in fact acceptable.
  • Method capability. The intermediate precision from validation tells you how much the method varies within one lab. A margin far tighter than that variation will fail good transfers; a margin far looser will pass poor ones.
  • Use of the data. Release testing and long-term stability trending have different tolerance for offset. Stability data from two sites that will be compared or pooled deserves a tighter margin than a test that is only checked against a limit.

Also plan the sample size so a good transfer has a reasonable chance of passing. The Pharmaceutical Technology article walks through how power depends on the true difference between means, the variability at each site, and the equivalence bounds, and handles the case where the two labs have different precision by using a Welch-type statistic with Satterthwaite degrees of freedom.8 Unequal precision is worth noticing on its own. If the receiving lab’s results scatter twice as much as the transferring lab’s, that is a finding about technique or equipment even if the means agree.

Impurities, Dissolution, and Other Tests

Not every test suits a means comparison. USP <1224> notes that impurity testing may use a simpler descriptive approach and that dissolution comparisons may use the similarity factor f2.1 For impurity methods, the most important comparison is often not the number but the profile: does the receiving lab see the same peaks, at the same relative retention, above the same reporting threshold? A criterion that compares only the total impurities can pass while the receiving lab misses an individual peak entirely. Write criteria for individual specified impurities, for unspecified impurities above the reporting threshold, and for detection itself.

Samples, Lots, and Strengths

MHRA’s list of common findings includes two that are about study design: transfers performed as a single test on one batch (one assay, one dissolution test, one agar plate) without justification, and transfers that used one strength for a method applied to several strengths.6 A single strength may be enough if the sample preparation and concentration range are the same, but that argument needs to be in the protocol. For stability-indicating methods, follow FDA’s recommendation to include forced degradation samples or samples containing relevant impurities at both sites.4 Section 6.39 of EU GMP also asks the protocol to address transport and storage of test items, which matters when samples must travel between sites and degradation in transit would create a false difference.3

When a Criterion Fails

USP <1224> is direct about failure. Changes made after a failed criterion must be approved before additional data are collected, and the procedure cannot be considered transferred until effective remedial steps are in place.1 EU GMP section 6.40 requires deviations from the protocol to be investigated before the transfer is closed.3

A January 2022 FDA warning letter to Aurobindo Pharmaceutical Limited shows what happens when that discipline slips. During transfer of a GC-MS method, the firm failed the acceptance criteria for interlaboratory precision because of peak splitting. FDA found the investigation inadequate because it did not consider all potential equipment sources. The failing result was invalidated without a scientific rationale and was not reported in the problems or corrective action sections of the approved transfer report. The firm attributed the failure to a degraded column, replaced it, obtained a passing result with a fresh sample, and did not put controls in place to make sure only acceptable columns were used afterward. FDA asked the firm to review all of its analytical and microbiological method transfers for discrepancies not described in the approved reports.9

A pass that hides a failure is worse than an honest failure. An honest failure leads to an investigation, a fix, and a method that is better understood. A failure that is invalidated and left out of the report produces a transfer record that does not match what happened, and every result the receiving lab reports afterward depends on that record. Require the transfer report to list every failed injection, aborted sequence, and invalidated result, with the rationale, even when the final outcome is a pass.

Where Two Labs Diverge: Instruments, Columns, and Software

When transferred methods give different answers, the cause is rarely the written procedure. It is usually something the procedure did not specify because the originating lab never had to think about it. Two of MHRA’s common findings point straight at this: equipment that was not qualified for the range the transferred method uses, and gap analyses that did not consider differences in equipment, such as different makes and models of HPLC at the two sites.6

Gradient Delay Volume

For gradient liquid chromatography, the best-known source of instrument-driven difference is the gradient delay volume, also called dwell volume: the volume between the point where solvents are mixed and the head of the column. In a 2024 LCGC International column on the effects of gradient delay volume on method transfer, Dwight Stoll describes it as one of the most important, yet least understood, parameters in gradient elution, because it delays the arrival of each programmed change in mobile phase composition at the column inlet.10 Two instruments with different delay volumes running the same gradient table will deliver that gradient to the column at different times, which shifts retention and can change how well closely eluting peaks separate. Stoll illustrates this with a method developed on an instrument with a delay volume of 200 microliters and then run on one with 1,000 microliters: an eight-component mixture that was well resolved on the development system showed a resolution of only about 1.0 for the critical pair on the second instrument. Moving in the opposite direction, from the larger to the smaller delay volume, also produced coelution.10

The practical step is simple and often skipped: measure the delay volume of the instruments at both sites as part of the gap analysis, record it in the protocol, and state in the method whether and how the receiving lab may compensate. Many instruments allow the gradient start to be offset relative to injection. If the method allows that adjustment, it should say so, and the transfer data should show the setting used.

What the Pharmacopeia Allows You to Adjust

The harmonized USP General Chapter <621> on chromatography became official on December 1, 2022. In Tony Taylor’s LCGC summary, the most significant change is that the ratio of column length to particle size can now be altered for gradient separations.11 Taylor also points out that the chapter states changes outside the allowed ranges require full revalidation, that multiple adjustments can have a cumulative effect that users must evaluate, and that adjusted gradient conditions are more critical than isocratic ones because peaks can move to a different step of the gradient and coelute.11 For a method transfer, the important point is that an adjustment within the allowances is still a difference between the two sites. Record it in the protocol, run the transfer samples with the adjusted conditions, and make sure the adjusted parameters are the ones configured in the receiving lab’s instrument method and recorded in its method version. For a non-compendial method, whether those allowances apply at all depends on what your filed procedure and internal procedures say, so settle that before the transfer rather than during an investigation.

Columns, Reference Standards, and Reagents

Column lot differences are a constant source of small shifts. In the Pistoia Alliance pilot that moved HPLC methods between two companies’ chromatography data systems, published in the Journal of Pharmaceutical and Biomedical Analysis in 2025, the authors attributed the retention time differences they saw, which were within system suitability criteria, to different column lots or a difference in dwell volume between the instruments.12 The Aurobindo case shows the other side: once a column is identified as a root cause, the controls on column selection and condition have to follow.9 Agree in the protocol on column brand, chemistry, dimensions, and whether the transfer will deliberately use more than one column lot. Send reference standards from the same lot to both labs where possible, since USP places supply of reference standards with the transferring unit.1

The Processing Method Is Part of the Method

This is the source of divergence most transfer plans miss. A chromatographic method has two halves: the acquisition method that controls the instrument, and the processing method that controls integration, peak identification, thresholds, and calculations in the CDS. The written procedure usually describes the first in detail and the second in a sentence or two. When the receiving lab uses a different CDS, or the same CDS with its own default settings, the processing method is rebuilt by hand.

A July 2026 FDA warning letter to Almon Healthcare shows the result. The firm had implemented a method transferred from its contract testing laboratory for routine and stability testing, and the implemented method was unable to detect multiple unknown impurities that the contract lab’s validated method identified. The firm attributed the discrepancy to an incorrectly set processing threshold and provided reprocessed data. FDA found the response inadequate, noting that while the firm tried to explain the threshold differences between the contract lab and its in-house verification study, it did not address whether the methods were in fact equivalent.13

The Pistoia Alliance pilot is candid about how hard this is to fix with technology alone. The project transferred machine-readable HPLC methods and results between two different CDS products without manual keying, but its authors note that it did not attempt to standardize data processing algorithms across CDS, such as integration parameters; it covered only the representation of instrument methods and processed results.12 In other words, even a fully digital method exchange still leaves integration settings for people to align.

Two controls help. First, transfer the processing method as an explicit, version-controlled list of parameters: integration events, peak width and threshold settings, reporting and disregard thresholds, peak naming and relative retention windows, and calculation formulas. Second, cross-process. Have each lab send the other a set of raw data files from the transfer runs, and have each lab process the other’s data with its own processing method. If both labs produce the same results from the same raw data, processing is not the cause of any difference you see. If they do not, you have isolated the problem to the software configuration before it reaches a single routine result.

Acquisition

Instrument Hardware and Settings

Gradient delay volume, extra-column volume, column oven design, detector data rate and response time, injector behavior. Measure and record at both sites during the gap analysis.

Materials

Columns, Standards, and Reagents

Column brand and lot, reference standard lot and assigned value, reagent grades, mobile phase preparation. Agree on them in the protocol and send shared lots to both sites where possible.

Processing

CDS Integration and Thresholds

Integration events, thresholds, peak identification windows, disregard limits. Transfer as a controlled parameter list and confirm by cross-processing each lab’s raw data.

Reporting

Calculation, Rounding, and Limits

Formulas for the reportable result, replicate handling, rounding rule, units, and specification version in each LIMS. Reconcile before go-live and at every change.

Data Integrity at the Receiving Lab

The transfer study is GMP work. Its data are the evidence that every later result from the receiving lab is reliable, so the record of the transfer itself needs the same completeness, attribution, and review as any release test. Two patterns cause most of the trouble: informal runs before the official study, and transfer reports that summarize results the raw records do not fully support.

Treat Transfer Runs as Official From the First Injection

Receiving labs often want to practice before the formal study, and familiarization is reasonable. The risk is how it is done. FDA’s data integrity guidance says it prohibits sampling and testing with the goal of achieving a specific result or overcoming an unacceptable result, and that it considers it a violative practice to use an actual sample in test, prep, or equilibration runs as a means of disguising testing into compliance. It also notes that system suitability tests should use replicate injections of a standard preparation, and that if an actual sample is used for system suitability it should be a properly characterized secondary standard from a different batch than the samples being tested, under written procedures.14

For a transfer, the clean approach is to write familiarization into the protocol: name the material to be used (a standard, a placebo, or a designated familiarization lot that is not part of the comparative set), keep all of that data, and make it reviewable. The comparative samples should not be run until the protocol says so, and every injection of them should appear in the record. If a receiving lab runs the transfer samples informally first and then runs them again “for real,” the transfer result is no longer evidence of anything.

Complete Records, Not Summary Tables

Section 211.194(a) requires laboratory records to include complete data derived from all tests needed to assure compliance with specifications.2 An August 2026 FDA warning letter to Auriga Research, a contract testing laboratory, shows the gap that opens when summaries are trusted over source records. The firm’s summary reports included final microbial plate counts, but the raw data test sheets for some of the samples covered in those reports lacked complete documentation of the counts. FDA stated that data integrity is fundamentally compromised when a firm fails to record or keep complete and accurate records of test results.15 That case involved routine microbiology testing rather than a transfer, but the lesson applies directly: a transfer report built on summary tables can be complete on paper and incomplete at the source.

So define the transfer data package in the protocol, and make it the receiving lab’s obligation to provide it:

  • Raw data files in a form the sponsor can open or reprocess, not only printouts or PDFs.
  • The acquisition and processing methods used, with their version identifiers.
  • The sequence or run log, including any injections that were aborted, repeated, or excluded.
  • System suitability results for every run.
  • Instrument identifiers and their qualification status, column serial numbers, and reference standard lots.
  • The CDS audit trail for the transfer sequences, covering integration changes, reprocessing, and result edits.

Review the Audit Trail Before You Approve the Report

FDA’s guidance says that where regulations do not set a review frequency, firms should set audit trail review frequency through risk assessment, considering data criticality, existing controls, and impact on product quality.14 Transfer data are about as critical as laboratory data get, because they qualify everything that follows. A sensible rule is that someone on the sponsor side reviews the audit trails for the comparative runs before approving the transfer report. The review does not need to be long. It needs to answer a few questions: were any results reprocessed, and why; were any integrations changed manually; were any sequences aborted or injections excluded; and does every result in the report trace to a raw data file in the package.

Watch for method changes recorded outside the method. Analysts at a receiving lab sometimes find small practical tweaks that make a method run better: a longer sonication, a different filter, a reequilibration step. Those changes are often sensible. They are also undocumented changes to a validated procedure if they live only in a notebook or in an analyst’s memory. Ask the receiving lab, at the close of the transfer, to list every practical adjustment its analysts made, and route each one through change control or into the procedure text.

Keeping Methods and Specifications in Sync in Both LIMS

Once a method is transferred to a CRO or CDMO, the same test is configured in at least two LIMS and two CDS: the sponsor’s and the receiving lab’s. Each configuration was built by different people, from the same documents, at different times. From the day the transfer closes, the two configurations start to drift unless someone keeps them aligned. We have written separately about keeping LIMS master data under change control inside one organization; the problem here is the cross-organization version of it, where neither side can see the other’s configuration directly.

ICH Q14 is clear about how much detail the procedure has to carry. It says the analytical procedure should describe the steps needed to perform each test, which can include the sample, reference materials and reagents, sample and control preparation, use of the apparatus, generation of the calibration curve, the number of replicates, and the formulas for calculating reportable results, with enough detail for a skilled analyst to perform the analysis and interpret the results.7 Every one of those items becomes master data in a LIMS or CDS. If the procedure is vague about any of them, each lab fills the gap with its own default, and the defaults rarely match.

Rounding and Reportable Values

Rounding is the clearest example. The USP General Notices, as quoted in a 2013 Pharmaceutical Technology column by Chris Burgess, say to consider only one digit to the right of the last place in the limit expression, to drop it and leave the preceding digit unchanged if it is less than 5, and to increase the preceding digit by 1 if it is 5 or more. The Notices also say numbers should not be rounded until the final calculation of the reportable value is complete.16 Burgess points out that the always-round-up-on-5 rule biases data over time and describes an unbiased alternative, rounding a trailing 5 to the nearest even digit, that statisticians have recommended for decades.16

Both rules exist in software. If the sponsor’s LIMS rounds half up and the contract lab’s LIMS rounds half to even, the two systems will disagree on boundary values. Consider an impurity limit of not more than 0.2 percent and a calculated value of 0.25 percent. Rounded half up, it becomes 0.3 percent and fails. Rounded half to even, it becomes 0.2 percent and passes. Same sample, same chromatography, same calculation, opposite disposition. The same problem appears when one system rounds intermediate values (for example each replicate) and the other rounds only the final average, which the General Notices say not to do.

Replicates, Units, and Reporting Conventions

Other common mismatches follow the same pattern:

  • Replicate handling. Is the reportable result the mean of two preparations, each the mean of two injections, or the mean of all four injections? The answer can change the result and always changes how an individual outlier is handled.
  • Units and bases. Percent of label claim versus percent weight by weight, as-is versus anhydrous or solvent-free basis, milligrams versus micrograms.
  • Values below a limit. How results below the quantitation limit or reporting threshold are recorded, whether as a text string, a numerical value, or zero, and how they are counted in a total.
  • Reporting thresholds. The threshold at which an unspecified impurity is reported, which is also a CDS processing setting, as the Almon case showed.13
  • Specification version. Which version of the specification each system holds, and which effective date applies to which lots.
Master data itemWhere it usually livesWhat goes wrong if the two labs differHow to check it
Calculation formula for the reportable resultCDS custom field or LIMS calculationSystematic bias between sites with identical chromatographyRun one shared raw data set through both systems and compare outputs to the full precision
Rounding rule and decimal placesLIMS result or specification configurationBoundary values pass at one lab and fail at the otherTest values ending in exactly 5 at the rounding position
Replicate and averaging schemeLIMS test templateDifferent reportable results and different outlier handlingCompare template structure line by line against the procedure
Reporting and disregard thresholdsCDS processing method and LIMSImpurities reported at one lab and absent at the otherCross-process raw data and compare peak tables
Specification limits and versionLIMS specificationDifferent disposition of the same result; wrong limits applied to a lotReconcile version and effective date at each specification change
Units and basisLIMS result definitionResults that look different by a fixed factor, or that differ by water or solvent contentCompare the result definition and any correction factors used

Version Control Across Two Organizations

The structural fix is to treat the receiving lab’s configuration of the method as a controlled copy of the sponsor’s master, with a link between them that both quality systems can see. In practice that means:

  • The method and specification each carry a version identifier that appears in both LIMS and on every certificate of analysis or data package the receiving lab issues.
  • The quality agreement requires the receiving lab to notify the sponsor before changing any configuration item in the table above, and requires the sponsor to send every procedure or specification revision through a defined route with an effective date.
  • Changes on both sides are linked: a sponsor change control for a specification revision is not closed until the receiving lab confirms its configuration was updated and tested.
  • A reconciliation runs at a set interval and after any LIMS or CDS upgrade at either site, comparing the configured items, not just the documents.

A simple go-live test. Before the receiving lab reports its first routine result, have it enter a set of agreed test values into its LIMS, including values that sit exactly on rounding boundaries, values below the quantitation limit, and values on either side of each specification limit. Do the same in the sponsor’s LIMS. The reportable results and pass or fail calls should match exactly. Keep the comparison as part of the transfer record.

After the Transfer: ICH Q14, Q2(R2), and Ongoing Comparability

ICH Q2(R2) and the new ICH Q14, both adopted in late 2023, changed how transfer fits into the larger picture. ICH Q2(R2) says that transfer of a validated analytical procedure should be considered in the context of analytical lifecycle changes in line with ICH Q14, and that transferring a procedure to a different laboratory calls for partial or full revalidation and/or comparative analysis of representative samples, with a justification provided when no additional transfer experiments are done. It also says co-validation can satisfy the transfer requirements at the participating sites.5

Q2(R2) adds a point that is easy to overlook. Reproducibility, the precision between laboratories, is usually not required for a regulatory submission, but the guideline says it should be considered in cases of standardization and in cases where procedures are conducted at multiple sites.5 If you know a method will run at more than one lab, planning a reproducibility study during validation can make every later transfer simpler.

What ICH Q14 Changes

ICH Q14 places transfer inside a lifecycle model. It describes an analytical procedure control strategy, a set of controls derived from development data, risk assessment, robustness, and prior knowledge, that includes the procedure parameters needing control and the system suitability test, and that should be defined before validation and confirmed after it.7 For a transfer, this gives the gap analysis a clear target: the receiving lab has to be able to operate every parameter in the control strategy within its range, and the system suitability test has to be capable of catching the failure modes that matter at the new site.

Q14 also describes established conditions (ECs) for analytical procedures, using the ICH Q12 definition: legally binding information considered necessary to assure product quality, where any change requires a submission to the regulatory authority.7 A transfer that requires adjusting an EC is no longer only a laboratory exercise. Separately, whether adding a new testing site to an approved application needs a regulatory filing depends on the application and the region; FDA’s 2015 guidance notes that postmarketing changes to analytical procedures must be reported in line with 21 CFR 314.70 or 601.12.4 Bring regulatory affairs into the transfer plan early so the filing strategy and the transfer timeline line up.

Ongoing Performance Verification Across Both Sites

A transfer shows that two labs agreed during the study. It says nothing about next year. USP General Chapter <1220>, the analytical procedure life cycle, became official in May 2022 and organizes the work into three stages, with the third being ongoing performance verification during routine use.17 A 2023 Pharmaceutical Technology article by Phil Borman and coauthors explains that Stage 3 uses risk to decide how much monitoring a procedure needs. For high-risk procedures, control-chart monitoring of selected performance data, such as results for a quality control sample or indirect measures like chromatographic resolution, signal to noise, and peak tailing, can give early warning of changes. For low-risk procedures, monitoring may be limited to atypical results and system suitability failures. The authors also note that verification acceptance criteria can be derived from validation or system suitability criteria when no analytical target profile was defined.17

For a method run at two sites, the most useful version of this is a shared view. Run the same control sample at both labs, plot both labs’ results on the same chart, and trend stability results by testing site. A slow drift at one site, caused by a new instrument, a CDS upgrade, or a change in analysts, shows up as separation between the two sets of points long before it causes an out-of-specification result.

When to Compare Again

Agree in the quality agreement on the events that trigger a fresh comparison between sites. Typical triggers include replacement of the main instrument or a change of instrument platform at either lab, a major CDS or LIMS upgrade, a change of column supplier, a revision to the procedure at either site, and a sustained offset in the shared trend. A small comparative study after one of these events is far less work than a retrospective investigation across months of stability data.

A Working Sequence for Sponsors

The guidance documents describe what a transfer must contain. The sequence below is how we suggest pharma and biotech sponsors put those pieces in order so that comparability is designed in rather than checked at the end.

1

Inventory and Gap Analysis

Confirm the method matches the filed procedure, review the validation against current ICH Q2(R2) expectations, and compare the two sites’ instruments (including gradient delay volume), CDS products and versions, LIMS calculation and rounding behavior, columns, and reference standards. Document the gaps and what will be done about each.

2

Choose the Approach and Write Equivalence Criteria

Select comparative testing, co-validation, revalidation, or a justified waiver based on risk. For comparative work, set equivalence margins from the specification, method precision, and data use, and size the study so a good transfer can pass.

3

Agree the Data Package and Master Data First

Specify the raw data, methods, audit trails, and metadata the receiving lab will provide. Transfer the processing method as a parameter list. Agree on the reportable result formula, replicate scheme, rounding rule, units, and specification version before any sample moves.

4

Familiarize Under the Protocol

Let the receiving lab practice on designated material that is not part of the comparative set, with all data kept. Do not run comparative samples until the protocol says the study has started.

5

Execute and Cross-Process

Run the comparative study. Exchange raw data files and have each lab process the other’s data, so that software differences are separated from chemistry differences.

6

Review the Source, Then Approve

Review raw data and audit trails for the comparative runs. Investigate every deviation and failed criterion before closing, and make sure the report lists them all with rationale. Collect the list of practical adjustments analysts made and route them through change control.

7

Go Live With Ongoing Checks

Run the LIMS boundary-value test at both sites, link change control between the two organizations, set the reconciliation interval, and start shared trending of a control sample and stability results by site.

Signs a transfer was done well. A year later, the sponsor can open the receiving lab’s raw data for any lot and reprocess it. The two sites’ control sample results overlap on one chart. Every procedure and specification revision since the transfer shows a matching configuration change at the receiving lab, with a date. And when someone asks why the receiving lab’s method has a gradient offset or a different column lot, the answer is in the transfer record rather than in someone’s memory.

Conclusion

Analytical method transfer is usually treated as a gate: write the protocol, run the samples, sign the report, move on. The regulatory texts support a fuller view. USP <1224>, EU GMP Chapter 6, FDA’s guidance, and 21 CFR 211.194 all ask the same underlying question, whether the method works under actual conditions of use at the new site, and ICH Q14 and Q2(R2) place the answer inside a lifecycle that does not end when the report is approved. The data that make two labs comparable are not only the transfer results. They are the instrument settings, the processing method, the calculation and rounding rules in each LIMS, and the completeness of the record, all kept aligned for as long as both labs test the product. A transfer that tests the whole measurement system, uses equivalence criteria with a defended margin, and sets up shared trending afterward gives a sponsor data it can rely on years later.

Sakara Digital works with pharma and biotech organizations on the data side of laboratory quality: method and specification master data, CDS and LIMS configuration across sites and partners, and the records that make outsourced testing defensible. If you are planning a method transfer to a CRO or CDMO, or trying to explain why two labs no longer agree, we are happy to talk it through and share an independent perspective on where to start.

For Further Reading