Reprocessing and Manual Integration: When Changing a Chromatogram Is Legitimate

A purity figure is not measured. It is calculated from a peak area, and a peak area depends on where somebody decided the peak started and stopped. Move the baseline and the number moves, with the same sample in the same vial. This is not a flaw in chromatography; it is how integration works. What regulators care about is whether the moving is visible.

Quick answer

Is reprocessing a chromatogram allowed? Yes, under controls. PIC/S states that manual integrations and reprocessing of laboratory results should be performed in an approved and controlled manner, that the quality unit should ensure changes to data are performed only when necessary and by designated individuals, and that original unchanged data should be retained in its original context.

Is it acceptable to keep only the final result? No. The FDA answers that question directly and in one word, then adds the reason: if chromatography is reprocessed, written procedures must be established and followed, and each result retained for review.

And a sentence worth sitting with: for most lab analyses, the FDA says, reprocessing data should not be regularly needed.

What processing actually is

The MHRA defines data processing as a sequence of operations performed on data to extract, present or obtain information in a defined format — and gives as its example the conversion of a raw electronic signal to a chromatogram and subsequently a calculated numerical result.

That chain is the whole subject. The detector produces a signal. The software renders it as a trace. A person or a method defines where peaks begin and end. Arithmetic turns areas into percentages. Every step after the first involves parameters that a human chose, and the MHRA asks for traceability of them: there should be adequate traceability of any user-defined parameters used within data processing activities to the raw data, including attribution to who performed the activity.

Why the signal itself must be kept, rather than the picture, is the subject of static and dynamic records.

The sentence that defines the whole concern

One line in the MHRA guidance states the problem more precisely than any amount of commentary. Audit trails and retained records, it says, should allow reconstruction of all data processing activities regardless of whether the output of that processing is subsequently reported or otherwise used. And then:

If data processing has been repeated with progressive modification of processing parameters, this should be visible — to ensure that the processing parameters are not being manipulated to achieve a more desirable result.

Note what is not being said. It is not that reprocessing is suspicious. It is that a sequence of reprocessings, each one nudging the parameters, has a shape that is recognisable when all the versions survive and invisible when only the last one does. Keeping every run is not bureaucratic completeness. It is the only thing that makes the pattern detectable.

Legitimate reasons to reprocess, and the shape of a defensible record

PIC/S sets the approval requirement at a specific level: any and all changes and modifications to raw data should be fully documented and should be reviewed and approved by at least one appropriately trained and qualified individual. Its inspection prompt asks for evidence demonstrating formal approval of the proposed change, controlled and defined changes, and formal review of the changes made.

Three stages, in order. Approval before, control during, review after. A reintegration done quietly and reported as the result fails all three while producing a document that looks identical.

Save on each injection, not at the end of the set

The FDA makes a point about when data becomes a record that has a direct bearing on what can be quietly dropped. Chromatographic data, it says, should be saved to durable media upon completion of each step or injection — naming peak integration or processing steps, and finished, incomplete, or aborted injections — instead of at the end of an injection set. Changes to the chromatographic data or the injection sequence should be documented in an audit trail.

Then the clause that matters: aborted or incomplete injections should be captured in audit trails and should be investigated and justified.

If a sequence only commits at the end, an injection that ran badly can disappear before anything is written. If each injection commits as it completes, it cannot. The difference between those two configurations is the difference between a complete record and an edited one, and neither looks different on the final report.

The MHRA explains the underlying mechanism in its discussion of transactions: the metadata are not captured in the system audit trail until the user saves the transaction to durable storage, which is why it recommends that the time intervals before saving be minimised and that systems be designed to require saving data to permanent memory before prompting users to make changes. What the trail then shows is covered in what a reviewer is supposed to look at.

Testing into compliance, and the system suitability loophole

The FDA devotes a question to a practice that looks procedural and functions as something else. It prohibits sampling and testing with the goal of achieving a specific result or to overcome an unacceptable result — testing different samples until the desired passing result is obtained — and names the practice: testing into compliance.

The specific mechanism it flags is the use of real samples in preparatory runs. In some situations, it says, use of actual samples to perform system suitability testing has been used as a means of testing into compliance, and 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.

The logic is worth spelling out, because it is not obvious. A system suitability run is not a reportable result. If an analyst injects the actual sample as a suitability or equilibration run, they learn the answer without generating a record of having asked. If the answer is good, the next injection becomes the official one. If it is not, nothing was officially tested.

The FDA's remedy is to anchor suitability to standards rather than samples. It cites the United States Pharmacopeia: system suitability tests must include replicate injections of a standard preparation or other standard solutions to determine if requirements for precision are satisfied, per USP General Chapter <621> on chromatography. Where an actual sample is used, it should be a properly characterised secondary standard, written procedures should be established and followed, and the sample should be from a different batch than the samples being tested.

What system suitability is for in the first place — confirming the method is behaving today as it did during validation — is covered in robustness and system suitability.

And the completeness requirement that closes the door: all data, the FDA writes, including obvious errors and failing, passing, and suspect data, must be in the records that are retained and subject to review and oversight.

Excluding a result is a decision with a paper trail

Sometimes a result genuinely should not count. A vial was cracked, an injection failed, a standard was mis-prepared. Both guidances permit exclusion and both make it expensive in documentation.

The MHRA: data may only be excluded where it can be demonstrated through valid scientific justification that the data are not representative of the quantity measured, sampled or acquired. The justification should be documented and considered during data review and reporting. And — the operative clause — all data, even if excluded, should be retained with the original data set, and be available for review in a format that allows the validity of the decision to exclude the data to be confirmed.

The FDA arrives at the same place: invalidating test results requires a valid, documented, scientifically sound justification, and even where results are legitimately invalidated on the basis of a scientifically sound investigation, the full batch record would include the original invalidated data along with the investigation report that justifies invalidating the result.

So an excluded result is not removed. It is retained, with its reason attached, in a form that lets someone else judge whether the reason was good. That is a meaningfully different thing from deletion, and it is the only form of exclusion either guidance recognises.

Where this lands for a research buyer

None of this is visible in a certificate. A purity figure is a number, and the integration decisions that produced it are two layers behind the page. That is not concealment — a certificate is a summary report by design — but it does shape what the figure can be relied on to mean.

Two practical consequences follow.

The first is that a purity figure quoted to two decimal places implies a precision that integration choices do not support. What a single chromatographic number is actually reporting is set out in what 99% purity actually means, and the spread between methods in how much quantification methods agree.

The second is that agreement between two different measurement principles is worth more than a tighter figure from one. Reintegration can move a chromatographic percentage. It cannot move an independent orthogonal measurement in the same direction by accident.

What to ask, and what to conclude

• Read a purity figure as the output of a calculation, not a reading. Where the baseline sat is part of the result.
• Treat heavy reprocessing as a question about the method, not only about the record. The FDA's own observation is that for most analyses reprocessing should not be regularly needed.
• Do not expect excluded results to appear on a certificate. They should exist in the laboratory's record; the certificate reports the conclusion.
• Prefer two principles over two decimal places. Orthogonal agreement is the only verification available from outside the laboratory.
• Remember that a tidy chromatogram is a processed object. Cleanliness is partly a property of the sample and partly of the processing.

Key takeaways

• The FDA states that it is not acceptable to retain only the final result from reprocessed chromatography; each result must be retained for review.
• PIC/S requires reprocessing and manual integration to be approved, controlled, performed by designated individuals, and reviewed and approved by at least one trained and qualified person.
• Original unchanged data must be retained in its original context.
• The MHRA requires progressive modification of processing parameters to be visible, so parameters cannot be manipulated toward a more desirable result.
• Chromatographic data should be saved on completion of each injection, not at the end of an injection set.
• Aborted or incomplete injections must be captured, investigated and justified.
• Using an actual sample in a test, prep or equilibration run to disguise testing into compliance is a violative practice; USP <621> anchors system suitability to replicate injections of a standard.
• Excluded data must be retained with the original data set alongside a documented scientific justification.

Frequently asked questions

Is manual integration itself a red flag?

No. Automatic integration misreads shoulders, drifting baselines and poorly resolved pairs, and a trained analyst correcting that is doing the job properly. What distinguishes it from the other thing is that the original survives, the reason is written, and somebody independent approved it.

How would anyone outside the laboratory detect bad reintegration?

Generally they would not, which is why both guidances place the control inside the laboratory and place the purchaser's effort on supplier qualification instead — the reasoning behind evaluating a supplier.

Does this apply to mass spectrometry too?

Yes, in principle, since spectra are extracted and recalculated from stored scans. The FDA's named example is FT-IR, whose spectral file it describes as dynamic and reprocessable; the same reasoning covers any instrument whose stored data can be re-rendered.

What happens if someone reports a suspected falsification?

The FDA addresses this specifically: regardless of intent or how, or from whom, the information was received, suspected or known falsification or alteration of records must be fully investigated under the quality system to determine the effect on data reliability, determine the root cause, and ensure corrective actions are taken. It cannot be handled informally outside that system.

Does reprocessing affect a reported detection limit?

It can, since limits depend on how signal and noise are assessed. What a reported limit does and does not establish is covered in not detected is not zero.

Sources

• US Food and Drug Administration. Data Integrity and Compliance With Drug CGMP: Questions and Answers — Guidance for Industry. December 2018. FDA guidance
• Pharmaceutical Inspection Co-operation Scheme. PI 041-1: Good Practices for Data Management and Integrity in Regulated GMP/GDP Environments. Adopted 1 June 2021; entry into force 1 July 2021. PIC/S guidance
• Medicines and Healthcare products Regulatory Agency. GXP Data Integrity Guidance and Definitions. Revision 1, March 2018. MHRA guidance

All quoted text is taken from the published guidance documents linked above. USP General Chapter <621> is cited as quoted within the FDA guidance.

This article describes record-keeping and data processing practice 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.