Peptide Reprocessing and Reworking: The Batch History a COA Does Not Record

Peptide reprocessing and reworking are the two routes by which material that failed its specification can be brought back into conformance and shipped with a clean certificate. ICH Q7 permits both, but it requires a reworked batch to have its impurity profile compared against batches made by the established process, and it warns that routine analytical methods may be inadequate to characterize reworked material. A certificate of analysis records the result of that routine method. It does not record whether the batch reached that result on the first pass or the third.

This is a structural gap rather than a testing failure. Every other attribute a buyer scrutinises on a certificate has a measurement behind it. Manufacturing history has none. Two vials of the same peptide, one produced right first time and one recovered from an out-of-specification lot by a second purification pass, generate certificates that are indistinguishable in every field a purchaser can read. The distinction lives in the batch record, and the batch record is not the document that ships.

Reprocessing and reworking are not the same operation

The two words are used interchangeably in casual conversation and mean quite different things in the guideline. ICH Q7, section 14.20, defines reprocessing as introducing an intermediate or active ingredient, including one that does not conform to standards or specifications, back into the process and repeating a step that is already part of the established manufacturing process. The guideline lists crystallization, distillation, filtration, chromatography and milling as examples. Because the operation is already validated and already described in the dossier, reprocessing is, in the guideline’s own words, generally considered acceptable.

Reworking is the harder case. The Q7 glossary defines it as subjecting material that does not conform to standards or specifications to one or more processing steps that are different from the established manufacturing process. The example given is recrystallising with a different solvent. The material is being rescued by a route that was never validated for it, and that is why section 14.30 requires an investigation into the reason for non-conformance before the decision to rework is even taken, and why section 14.31 requires evaluation, testing, stability testing if warranted, and documentation showing the reworked product is of equivalent quality to material from the original process.

Two boundary conditions in Q7 deserve attention because they are where the honest line sits. Section 14.21 states that continuing a process step after an in-process control test has shown the step is incomplete is part of the normal process and is not reprocessing at all. A coupling reaction driven to completion after a Kaiser test comes back positive is simply synthesis. Section 14.22 draws the opposite conclusion for reintroducing unreacted material and repeating a chemical reaction, which is reprocessing unless it was already part of the established process, and which the guideline says should be preceded by careful evaluation because of the potential formation of byproducts and over-reacted materials.

The clause that matters most to a certificate reader

Section 14.32 is the single most useful sentence in the guideline for anyone evaluating a supplier. It requires that procedures provide for comparing the impurity profile of each reworked batch against batches manufactured by the established process, and then adds that where routine analytical methods are inadequate to characterize the reworked batch, additional methods should be used.

Read that carefully. The guideline is stating, as a general expectation, that the standard release assay can fail to characterise reworked material. The routine method for a synthetic peptide is reversed-phase HPLC with ultraviolet detection, and that is the method whose output becomes the purity percentage on the certificate. Q7 is saying the number on the certificate may not be sufficient, and that the manufacturer is expected to know when it is not and to reach for something else. Nothing in that expectation requires the supplementary work to be disclosed downstream.

The reason a second purification pass can shift a profile rather than simply shrink it is chromatographic. Peptide impurities are structurally close to the target sequence, and a repeated separation does not remove all impurities proportionally. It removes the well-resolved ones efficiently and concentrates the ones that travel with the main peak. Deletion sequences differing by a single residue, epimers formed by racemisation at a stereocentre, and deamidation products carrying a mass shift of roughly one unit all elute near the parent. Removing distant impurities raises the area percentage while leaving the co-eluting population intact and proportionally enriched within what remains. We have covered why a single area percentage conceals this in our analysis of peptide impurity thresholds and what a purity number does not tell you.

Key Research Findings

• ICH Q7 section 14.32 requires the impurity profile of each reworked batch to be compared against batches from the established process, and explicitly anticipates that routine analytical methods may be inadequate for that comparison, requiring additional methods.
• ICH Q7 section 14.20 permits reprocessing but sets a frequency trigger: where reprocessing is used for a majority of batches, it must be incorporated into the standard manufacturing process rather than handled as an exception.
• ICH Q7 section 11.21 defines an impurity profile as requiring the identity or a qualitative designation such as retention time, the observed range of each impurity, and a classification of each identified impurity as inorganic, organic or solvent. A single purity percentage satisfies none of these three elements.
• FDA Warning Letter 320-25-106 to Somerset Therapeutics Private Limited, issued 4 September 2025 following a 10 to 21 February 2025 inspection, cited failure to investigate out-of-specification results for any individual unspecified impurity. The firm traced the impurity to a single active ingredient lot that exhibited a different impurity profile across the product shelf life when compared against exhibit batches.
• FDA Warning Letter 320-26-34 to Chemspec Chemicals Private Limited, issued 23 December 2025 following a 28 July to 1 August 2025 inspection, found the firm had manufactured and released active ingredient lots without creating or maintaining batch manufacturing records at all. The firm’s response stated it had recovered approximately half of the original records. FDA noted the deviations repeated those in an untitled letter dated 5 February 2025.
• Streuli, Erckes, Nardone, Bedard, Beland and Steuer, Journal of Peptide Science, 2026, volume 32, issue 3, quantified preparative peptide purification transfer using a defined impurity library and reduced elution-percentage transfer error from approximately 17 percent to under 5 percent, with optimised preparative purifications reaching purities above 90 percent at yields above 30 percent.

Two senses of the word reprocessing, and only one is about material

A terminology collision causes real confusion here. In manufacturing, reprocessing means putting physical material back through a step. In the analytical laboratory, reprocessing means reintegrating an existing chromatogram, adjusting baseline placement or peak boundaries and recalculating the result from data already acquired. Both practices are governed, both leave records, and they are entirely separate exposures.

Data reprocessing is the one regulators have pursued hardest, because integration parameters can be altered until an impurity peak falls below a reporting threshold without any physical change to the material. That failure mode is visible only in the audit trail of the chromatography data system, which is why it belongs to the discussion of certificate data integrity and ALCOA+ audit trails rather than to this one. Material reprocessing leaves no analytical fingerprint whatsoever. The chromatogram of a correctly reprocessed batch is a legitimate chromatogram of genuinely conforming material. There is nothing in the data to find.

Why the research peptide market has no mechanism here

Every control described above depends on a quality system that does not exist for research-grade material. Section 14.30’s investigation requirement assumes a deviation system. Section 14.31’s equivalent-quality demonstration assumes a validation function with the authority to reject. Section 14.32’s profile comparison assumes historical batch data from an established process to compare against, which in turn assumes the process is established and that someone has been keeping the records.

The Chemspec findings show what the floor looks like when that assumption fails at a facility that was at least subject to inspection. Lots were released with no batch manufacturing records created or maintained. Where no batch record exists, the question of whether material was reprocessed is not merely undisclosed, it is unanswerable, because the document that would carry the answer was never written. A certificate issued against that material is not false. It reports a real measurement on a real sample. It simply sits on top of nothing.

The Somerset finding illustrates the detection problem from the buyer’s side. The impurity that triggered the investigation was classed as an unspecified impurity, meaning it had no entry in the specification and no acceptance criterion of its own. It surfaced because a downstream manufacturer ran related-compound analysis and compared results against exhibit batches over time. A purchaser holding one certificate for one lot has neither the comparator batches nor the longitudinal data to run that comparison. Profile divergence is only visible across lots, which is the same reason a single certificate cannot substitute for the trend data discussed in our examination of out-of-specification results and the limits of retesting.

What a purchaser can actually establish

No question put to a supplier will produce verified manufacturing history, and it would be dishonest to suggest otherwise. What questions can do is establish whether the supplier possesses the records at all, which is a lower bar and a more informative one than it first appears.

Asking whether a specific lot was reprocessed or reworked is worth doing, not because the answer can be audited but because a supplier who cannot answer has told you the batch record does not exist or is not accessible to whoever handles enquiries. Asking whether the certificate reflects a first-pass release or a repeat determination separates a measurement from a retest, and the retesting rules that govern the second case are strict. Asking whether the laboratory holds prior lots of the same peptide and whether their impurity profiles are comparable tests for the historical data that section 14.32 assumes exists. A supplier who tests each lot in isolation with no reference to previous lots has no comparator, and therefore no way to notice a profile shift even if one occurred.

The most practical check requires no cooperation at all. Compare certificates for successive lots of the same peptide from the same supplier and look at the impurity peaks rather than the headline purity figure. Retention times and relative areas of the named impurities should be broadly stable across lots produced by a stable process. A lot whose purity matches its predecessors while its impurity pattern looks different is the observable signature of a process change, a reprocessing event, or a different manufacturing source. This is the same reasoning that makes a numeric result more informative than a conformance statement, a distinction we set out in our discussion of skip-lot and periodic testing. Most research peptide certificates report a single purity figure and no impurity table, which forecloses the comparison entirely and is itself a finding.

Limitations and open questions

The evidence assembled here is regulatory and inferential rather than experimental. ICH Q7 states what manufacturers are expected to do; it does not measure how often reprocessing or reworking occurs, and no public dataset reports the frequency of either practice in peptide manufacturing specifically. The two warning letters are documented enforcement findings at named facilities on stated inspection dates, but neither concerns a research peptide supplier and neither cites reworking directly. They are cited for what they establish about record-keeping and impurity profile divergence at inspected sites, which sets a reasonable upper bound on what to assume about uninspected ones.

The chromatographic argument that a second purification pass enriches co-eluting impurities follows from the separation mechanism and is consistent with the Streuli group’s finding that repurification workflows required deliberate optimisation to reach purities above 90 percent. It is not, however, a measured claim about any particular compound or supplier. A lot-to-lot study comparing full impurity profiles across research peptide vendors would settle several of these questions, and to our knowledge none has been published.

The structural point

A certificate of analysis is a snapshot of a sample at a moment. It answers what this material measured. It is silent on how the material got there, and reprocessing and reworking are precisely the events that make those two questions come apart. Q7 addresses the gap by requiring the manufacturer to hold the comparison data, not by requiring the certificate to disclose it. That allocation is deliberate and works when a quality system and an inspectorate stand behind it. For research peptides, neither is reliably present, so the buyer inherits a document engineered on the assumption that someone else was checking. Reading it well means knowing which questions it was never built to answer. Our certificate of analysis library publishes the third-party reports we hold so the reports themselves, rather than a summary claim, are what a researcher evaluates.

For research purposes only. Not for human consumption. Not for diagnostic or therapeutic use.