What "validated method" means on a peptide certificate: ICH Q2, verification, and the limits of system suitability

A certificate of analysis for a synthetic peptide typically carries two adjacent statements: a number, such as purity by RP-HPLC of 99.2%, and a qualifier, such as “method validated” or “validated HPLC method.” The number attracts all of the attention. The qualifier is doing more work, because it determines what the number is a measurement of — and because “validated” is a defined term in analytical regulation with a specific and surprisingly narrow meaning that a certificate rarely states.

The word is not decoration, and it is not a synonym for careful. It is a claim about scope: that documented evidence exists showing a particular procedure is suitable for a particular purpose, applied to a particular analyte in a particular matrix over a particular concentration range. Strip any of those qualifiers away and the claim does not survive. This post works through what the governing framework actually requires, the three different things “validated” can mean on a certificate, and why the per-run check that most laboratories genuinely do perform answers a different question than the one the word implies.

What validation formally asserts

The governing text is ICH Q2, revised to R2 and adopted by the ICH Assembly in Prague on 31 October and 1 November 2023, alongside the new companion guideline Q14 on analytical procedure development. FDA published both as guidances for industry, with availability announced in the Federal Register in March 2024. The R2 revision broadened the scope beyond the chromatographic and spectrophotometric assumptions of the original — it now explicitly addresses spectroscopic and spectrometric procedures such as NIR, Raman, NMR and MS, including those requiring multivariate statistical treatment, and resolves longstanding ambiguities around range, response, reference materials, replicates and biologics. Q14 sits beside it and covers development, change management and the lifecycle view, distinguishing a minimal from an enhanced development approach.

The structural point in Q2 is that validation is always validation for an intended purpose, and the guideline enumerates the purposes: identification, quantitative testing of impurities, limit testing of impurities, and assay of the active substance. These are not interchangeable. A procedure validated as a limit test — capable of establishing that an impurity is below some threshold — has not thereby been validated to report that impurity as a number. A procedure validated for one peptide in a lyophilized matrix has said nothing about a different peptide, a different counterion, or the same peptide in solution with an excipient present.

This matters immediately for research-grade material, because the economics of a catalogue running to dozens or hundreds of compounds do not support a separate validation exercise per analyte. What such a laboratory realistically operates is a general-purpose gradient method, developed once, applied broadly. That is a defensible way to run a screening laboratory. It is not what Q2 means by a validated analytical procedure, and a certificate that uses the word without naming the analyte and purpose it was validated for is not making a checkable statement.

Which validation characteristics a purity figure actually rests on

Q2 sets out the characteristics to be evaluated: specificity, accuracy, precision (subdivided into repeatability, intermediate precision and, where relevant, reproducibility), detection limit, quantitation limit, linearity, range and robustness. Not all of them bear equally on an area-percent purity figure.

Specificity is the load-bearing one. It asks whether the procedure can discriminate the analyte from everything else plausibly present — and for peptides the plausible alternatives are precisely the species that are hardest to separate. A single-residue epimer, an isoaspartate rearrangement product differing by 0.98 Da, a deletion sequence missing one residue from a long chain, a methionine sulfoxide eluting as a leading shoulder: none of these are guaranteed to resolve on a gradient optimised for the parent. Specificity is demonstrated by challenging the method with known related substances and with deliberately stressed material, which is the reason forced degradation studies exist and the reason a method is called stability-indicating only after it has survived one. Absent that exercise, a specificity claim rests on nothing firmer than the main peak having looked symmetrical.

The quantitation limit sets the reporting threshold, and the threshold is what converts a chromatogram into a percentage. An impurity below the reporting threshold has not been shown to be absent; it has been declined for reporting. A purity figure is therefore always a purity above some threshold, and a certificate that omits the threshold has omitted a term from the equation.

Then there is an assumption that sits underneath area-percent reporting without appearing in the validation characteristics at all. Normalised area percent presumes that all species respond equally per unit mass. At the detection wavelengths used for peptides — commonly 214 nm, sometimes 220 nm — the dominant chromophore is the amide backbone, so response scales roughly with the number of peptide bonds. That approximation is reasonable for an impurity closely related to the parent, such as an oxidation product or a single deletion. It degrades for a substantially truncated fragment, which carries fewer amide bonds than the mass it represents, and it degrades in the other direction for species differing in aromatic residue content, since tryptophan, tyrosine and phenylalanine contribute absorbance the backbone-only model does not account for. Q2’s answer is a relative response factor determined against a characterised reference standard, or direct assay against that standard. Both require a reference material of known content, which for most research peptides simply does not exist. Uncorrected area percent is a normalised comparison, not a validated quantitation, and the distinction is not visible on the certificate.

Validation, verification and transfer are three different claims

The pharmacopoeial framework separates cases that the single word “validated” collapses. USP General Chapter ⟨1225⟩ covers validation of procedures being established for the first time. Chapter ⟨1226⟩ covers verification: a laboratory implementing an existing compendial procedure is not expected to repeat the validation, but to assess selected performance characteristics sufficient to show the procedure is suitable under its own actual conditions of use — a deliberately narrower, risk-based exercise. Method transfer is a third case again, in which a sending and a receiving laboratory demonstrate equivalence, usually by comparative testing on shared samples.

Each produces a defensible statement, and each supports a different inference. A certificate reading “validated method” may reflect a full development and validation exercise for that analyte; a ⟨1226⟩ verification of a compendial procedure; a transfer from a sponsor; or none of the above. For most research peptides the second option is unavailable in principle, because there is no compendial monograph for the compound — nothing exists to verify against. That absence pushes the default toward a generic in-house method, which is exactly the case where the word carries least information.

System suitability answers a different question

Nearly every laboratory running a chromatographic purity method does perform a per-run check, and that check is real. System suitability testing, as described in USP General Chapter ⟨621⟩ and its equivalents, evaluates retention time reproducibility, peak tailing, theoretical plate count, replicate injection precision, and resolution between a specified critical pair. A failed system suitability run invalidates the data from that sequence. This is meaningful quality control, and it is more than many suppliers document.

But it is calibrated to a different question. System suitability asks whether the instrument, column and mobile phase are behaving today as they behaved when the method was established. It does not ask whether the method can see what matters. The resolution criterion is the sharp illustration: resolution is assessed between a named critical pair, typically the analyte and one known related substance included in the suitability solution. A run that passes has demonstrated exactly that separation. An impurity that co-elutes with the main peak and was never represented in the suitability mixture is invisible to the criterion and invisible to the integration, and the purity figure will be higher than the material warrants with every system suitability parameter comfortably in specification.

The two claims — the run was in control, and the method is fit to characterise this analyte — are independent. Certificates routinely present evidence for the first while using vocabulary that implies the second.

What a research-grade certificate can legitimately support

None of this makes an unvalidated purity figure useless; it makes it a comparative rather than an absolute measurement. Two lots of the same peptide run on the same method, same column chemistry, same gradient and same integration parameters can be compared with reasonable confidence, because whatever systematic bias the method carries is shared between them. That is the single most useful thing a supplier’s purity figure does, and it survives the absence of formal validation intact.

What does not survive is comparison across laboratories. Gradient slope, column chemistry and pore size, detection wavelength, injection load and integration threshold each move the reported number, in some cases by more than the difference between the two figures being compared. Ranking suppliers by decimal places on certificates generated by undisclosed methods is not a measurement.

Where formal validation is unavailable, orthogonality is the practical substitute, and in one respect a better one. A purity claim corroborated by a separation operating on a different mechanism, an independent quantitation anchor such as amino acid analysis for peptide content, and an identity confirmation by mass spectrometry is stronger than a single fully validated method, because the failure modes do not overlap — a species that co-elutes under reversed-phase conditions has no particular reason to co-elute under ion exchange, and no reason at all to be isobaric.

The questions worth putting to a supplier follow directly, and they are all answerable: what column and gradient, what detection wavelength, what reporting threshold, is the purity figure uncorrected area percent or corrected against a standard, and was specificity ever demonstrated against deliberately stressed material. Silence in response is itself a data point.

Q2 and Q14 were written for material heading toward a marketing application, where an analytical procedure is part of a regulatory commitment and validation has a defined audience. Research-use-only material has no such application and no such file, so the compliance status of these guidelines does not transfer. What transfers is the vocabulary and the underlying logic about what counts as evidence — that a method is fit for a stated purpose and not in general, that a purity figure is bounded below by a reporting threshold and conditioned on an equal-response assumption, and that a well-behaved instrument and a method capable of seeing the relevant impurity are two separate things. A certificate compresses all of that into one word. Reading it well means decompressing it.