Regulations and compliance · Practical guide

Manual HPLC integration under GMP: what reviewers should check

Automatic integration fails to represent a peak correctly. A manual adjustment may be necessary. Record the reason, retain the original and revised evaluations and make the change visible to the reviewer, including when the final result is more favourable.

Table of contents

At a glance

Is manual integration of HPLC data permitted under GMP?

Manual integration is not categorically prohibited. Use it under an appropriate controlled procedure with a scientific rationale. Preserve the original data, relevant processing versions, method, change and reason. Assess the effect on the result and decision. Changing an integration solely to obtain a desired result or hiding unfavourable processing outcomes is not a defensible process.

Manual integration and result selection are different things

The integration method translates the chromatographic signal into an evaluated contribution to the result. In certain signal sequences, the automatic processing cannot adequately represent the technically intended evaluation. This does not justify a free subsequent selection of results. The decision for an intervention is based on the signal, the method and the scientific purpose, not on the question whether the new result is more favourable.

Separate the professional justification from the technical feasibility. The ability of a user to shift a peak limit does not indicate whether such an action is authorized within the regular process. Define the permitted circumstances, responsibilities, and required documentation. Assess when individual handling is appropriate and when a methodological or systematic root cause investigation is necessary. A recurring manual repair may indicate the need for review of a standard process.

Sources: [1]

Obtain the complete evaluation context

For a traceable review, the relevant original data and processing steps are required. This may include signal, injection reference, sequence, method status, integration parameters, and result statuses. The extent required depends on the actual analysis and decision. A final output with a numerical value cannot automatically replace the dynamic evaluation context. Complete electronic records must remain available and verifiable within the appropriate concept.

Define how an original and a subsequent evaluation state can be distinguished. A correction must not inadvertently overwrite historical processing without notice. Maintain the required relationship between data record, method, manual intervention, and review. Also assess export and archiving pathways. If a later review requires a re-evaluation of the signal, a screenshot may be insufficient. The format decision is derived from the required action.

Sources: [2]

Define clear criteria prior to the intervention

An appropriate procedure outlines the rationale and technical criteria for the intervention. What signal or integration scenario is present? Which rule of the approved method must be correctly implemented? What information is required to support the justification? Define when additional technical assessment or investigation is necessary. A general comment such as "integration corrected" is typically less informative than a clear description of the specific finding and the applied rule.

Avoid criteria that merely describe the desired outcome. "Below specification limit" is not a scientific rule of integration. Similarly, an unfavorable initial result should not be deemed invalid simply because a subsequent result is more favorable. The evaluation of out-of-specification (OOS) results, invalid analysis, or methodological errors follows established procedures. Technical assessment and quality-based conclusions are linked but must not be conflated.

Example: Decision changes following processing

In the exercise scenario, the specification limit for a particular finding is at most 0.10%. The initial evaluation yields 0.12%, while a later manually edited value is 0.09%. These figures alone do not prove either the correctness or the invalidity of the correction. However, they indicate that manual editing could enable a different professional judgment. Therefore, the review must adequately assess the signal, rationale, method, change, and impact on results.

The test case includes both evaluation states, the documented rationale, and the actual affected dataset. An auditor assesses the scientific basis independently of whether the result meets the specification limit. If no credible justification is provided, the more favorable value must not simply replace the prior finding. The following work aid is a proposed approach for process and system audits. It does not establish an analytical integration method or a universal specification limit.

Internal Work Aid: Inspection Questions for Manual Integration
Assessment questionInformation neededWhere an insufficient demonstration becomes apparent
Why was intervention taken? Specific signal finding and approved evaluation criteria Only a desire for a more favorable result is mentioned
What was changed? Visible intervention with dataset and method reference Only the final result remains available
What impact did the change have? Comparison of relevant statuses and decisions The change from 0.12% to 0.09% remains unreported
Who prepared and reviewed? Identity, time point and required review The processing and review cannot be reconstructed
Has the entire relevant population been considered? Sequence, repetitions and further analyses Only selected cost-effective injections appear
Can we reconstruct the process later? Complete required electronic recordkeeping Expression or archive lacks essential connections

An output can mask an intervention

In the April 30, 2024 Warning Letter to Grupo Quimico SRL, the FDA objected, among other issues, that reviewed prints failed to reveal manual integrations. The procedures did not adequately account for relevant electronic information such as reprocessed integrations, aborted runs, and audit trails. This specific historical case is not used as a general error statistic. It raises a critical review question: Does the reviewer actually see the information necessary to support their decision?

Test the actual review view. Can the reviewer see manual edits and reach the correct record and method version? Are relevant changes, further analyses and required reasons available? A report can help when its scope and limits are clear. If it shows only the last selected version, the electronic review must supply the other necessary information. Additional signatures cannot compensate for missing data access.

Sources: [3]

Individual processing and method change differentiation

A manual adjustment to a specific dataset and a change to the general processing method have different scopes of impact. Assess whether the intervention applies only to a justified case or may affect future analyses. When a method is modified, the version, release, and affected data population are evaluated. Testing a single injection does not automatically validate suitability for all intended signal variants.

Investigate the impact on ongoing sequences and reprocessed older data. Which method is employed? Can the original and updated states be distinguished? Are different processing versions of the same signal still discernible? Customer-specific parameters and software versions may influence the technical representation. Therefore, the specific evaluation is performed within the employed chromatography data system and its configuration. A general manufacturer description does not substitute for verification of the actual workflow.

Audit trail and independent review are linked

The audit trail can enable traceability of relevant processing actions, but the specific scope of events captured must be verified within the actual system. Which actions are recorded? What content and reasons are visible? How do technical events relate to substantive data elements? The presence of an audit trail alone is insufficient to demonstrate traceability; responsible reviewers must identify and understand the relevant events from a technical and functional perspective.

An independent review evaluates more than just the technical change justification. It assesses the scientific rationale and completeness of the data under consideration. Roles and workflows are designed such that a required review can actually occur. Test scenarios involving missing justification, incorrect method attribution, or concealed prior work should be evaluated. If the workflow permits such situations without evident mitigation, the existing validation function remains insufficient as evidence of effectiveness.

Sources: [4] [1]

Handle exceptions, out-of-specification results, and trends in organized order

A manual intervention may constitute part of a justified correction but must not circumvent a required investigation. For relevant findings, assess the linkage to the out-of-specification (OOS) or deviation investigation procedure. The original results remain intact in their original context. A recalculation or reanalysis must not solely generate the desired final result at the expense of justifying the underlying rationale and preserving historical uncertainty. The quality-based decision rests with the responsible personnel and the approved procedure.

Consider recurring interventions based on reasonable criteria. Are similar activities accumulating for a specific method, device, signal variant, or after a software change? A higher frequency does not necessarily indicate a defect or erroneous behavior; however, it may serve as a legitimate basis for investigation. Interpret trends in the context of volume and data population. Evaluating only the count without a denominator may lead to conflating changes in workload with changes in process quality.

Build a usable review and validation package

A traceable package links requirements, roles, reference cases, electronic representation, processing documentation, and technical review. Evaluate both a legitimate intervention and the absence or inadequacy of justification. Enhance a method with a new version, a subsequent data analysis, and an archiving or export scenario. The scope aligns with actual, relevant functions; a generic screenshot of every mouse movement would not serve as a substitute for deliberate, purposeful selection.

The conclusion outlines which assertions are supported by evidence and which remain unproven. The team receives clear work guidelines and understands when professional escalation is required. Changes to software, method, or review procedures are traced back to these prerequisites. Manual integration is thus controlled and applicable only where scientifically justified. At the same time, it is ensured that a favorable final outcome does not obscure the complete history of the analysis.

Project experience and collaboration

Trace manual integration through to the decision on the result

With laboratory owners, we compare a suitable dataset before and after manual integration. A reviewer should be able to identify the reason, affected peak, method version and effect on the result from the records alone. A missing link may need a focused addition. Unclear evaluation rules first require scientific clarification by the laboratory. We support the system tests and review process. Your designated laboratory experts retain responsibility for the scientific justification of the integration method.

Frequently asked questions about Reviewing manual HPLC integration

Is manual HPLC integration generally prohibited?

No. Scientific justification, controlled procedure, complete documentation, and appropriate review are essential. Relying solely on manual intervention to achieve a desired outcome is inappropriate. The technical method and its permitted interventions must be defined for the actual analytical purpose.

Sources: [1] [2]

Is the comment "Peak corrected" sufficient?

Generally not. A general comment provides neither an explanation of the signal finding nor the applicable method rule nor the affected dataset. For subsequent evaluations, the FDA expects a justification for the change (Data Integrity Q&A, Question 1(c)). The required level of detail stems from the intervention and error sequence. Technical change information and scientific rationale must complement each other clearly and understandably.

Must the original analysis status be retained?

Yes. According to the FDA Data Integrity Q&A, each result from a reprocessed chromatography analysis must be retained for review; merely retaining the final result is insufficient (Question 14). A formally justified invalidated result remains part of the record as part of the investigation (Question 2). Original and subsequent processing, including method status, interventions, and justifications, must be maintained in a clearly distinguishable and traceable manner.

Sources: [2]

Is an appropriate result after reprocessing evidence of correct integration?

No. A result within the limit does not explain why the evaluation is scientifically valid. Independently of the desired outcome, assess the signal, method, intervention, and rationale. If a required investigation exists, it is not automatically fulfilled by a subsequent number.

Do we need to review electronic data in addition to the printout?

In dynamic chromatographic recordings, the review must also consider the electronic original. A printed copy does not fully retain its properties. According to the FDA-Q&A, original records including electronic records are the subject of a secondary review (Question 10). Integration parameters, reprocessing, and the rationale for such actions are part of a traceable and justifiable evaluation (Question 1(c)); a clear end curve alone does not address these issues.

Sources: [3]

When does individual processing become a method change?

This depends on which rule or configuration is modified and which analyses it affects. Verify scope, version, and release status. A recurring individual correction may warrant a method review. However, frequency alone does not constitute proven method deviation.

Sources and review date

Status: October 15, 2026. Regulatory classification and manufacturer responsibilities are separately documented. Examples, test matrices, and templates are standalone reference tools; they are customized for actual usage purposes.

  1. FDA: Data Integrity and Compliance With Drug CGMP, Q&A (2018)Final Guidance: Data completeness, electronic source data, review and investigation. Guidance and binding regulations differ.
  2. PIC/S: PI 041-1, Good Practices for Data Management and Integrity (2021)Inspection Guidance on Data Integrity; in particular electronic data, transmission, access, and review.
  3. FDA Warning Letter: Grupo Quimico SRL, 30.04.2024Specific inspection finding: Printouts failed to reveal manual integration issues, and electronic data and reprocessing during review were not adequately considered.
  4. European Commission: EU GMP Annex 11, Computerised Systems (2011)Current version as per EudraLex inventory as of 15.10.2026. Application, risks, operation, and data lifecycle; specific testing proposals are derived independently.

Sources checked: .

Daniel Herrmann, Daniel Herrmann Consulting

Daniel Herrmann Consulting

How we can support you

Daniel Herrmann Consulting combines professional assessment with hands-on work on the specific system. During the first meeting, we clarify your intended use, existing documentation, and the next most appropriate action step.

  • Computer system validation →

    Bring your GxP systems into validated operation, from laboratory software, MES and eQMS to ERP. We define the scope, plan risk-based tests using GAMP 5 and hand the evidence to QA for release.

Send a project enquiry Customer testimonials →