Impurities and related substances in GLP-1 peptides: what the lines on a research-use COA represent

Insight · reviewed 2026-10-19 · by Pepti GLP-1 editorial

Vocabulary: what the purity line counts and what it leaves out

A research-use certificate for a synthetic peptide usually carries one purity line, and that line is an area-percent from reversed-phase HPLC. The integrator sums every peak that absorbs in the low-ultraviolet region where the peptide backbone absorbs and reports the main peak as a share of that total. It is a ratio between UV-absorbing species in the sample solution, not the fraction of the vial that is peptide by weight.

Four terms that are not interchangeable

Purity on a research-use COA means HPLC area-percent of the main peak unless the record says otherwise. Related substances are the peptide-like species in the remaining area: sequence variants, modified forms and degradation products that absorb like the parent and elute near it. Process impurities are the subset created during synthesis, cleavage and purification. Degradation products form afterwards and grow with time and temperature.

Two things sit outside all four and are invisible to area-percent: water and counter-ion. A lyophilized peptide holds residual moisture, and its basic residues are paired with trifluoroacetate or acetate from purification. Neither absorbs in the detection region, so a lot at the top of the published range can still contain less peptide by mass than the label strength. That is the purity-versus-content distinction, which the Janoshik report on one Pepti retatrutide sample illustrates by printing both figures.

Where a certificate also prints largest single impurity and total impurities, they are integration results from the same chromatogram: the biggest non-main peak, and the sum of non-main peaks above the reporting threshold the laboratory applied. Where a record does not print them, the complement of the main-peak value is an upper bound on related substances by area, not a measurement of any particular species.

Process-related impurities from solid-phase synthesis

Semaglutide, tirzepatide, retatrutide, cagrilintide and mazdutide are assembled by solid-phase synthesis, one residue per cycle, on a resin. Each cycle has a deprotection step and a coupling step, and each is slightly less than complete. Across a 31-residue chain such as semaglutide or a 39-residue chain such as tirzepatide or retatrutide, those shortfalls accumulate into a predictable family of sequence variants that purification removes only partly.

Backbone variants

Deletion sequences arise when a coupling fails and the chain continues without that residue: one residue lighter, usually eluting just ahead of the parent. Insertion or double-coupling variants are the mirror case, one residue heavier and usually just behind. Truncated sequences come from chains capped or stopped early; much shorter and lighter, they sit well away from the main peak unless they still carry the fatty-diacid arm.

Incomplete deprotection adducts are chains on which a side-chain protecting group survived final cleavage; heavier and more hydrophobic, they elute later. Diastereomers form when a residue racemizes during activation, histidine and cysteine being the usual sources; they have the same mass as the parent and elute very close to it. Aspartate adds the aspartimide route: under repeated base exposure the backbone cyclizes to a succinimide (−18 Da), then reopens to a mixture of aspartate and isoaspartate forms with no mass difference from the parent, which makes them among the hardest process impurities to see by MS.

Acylation-specific variants

Every compound on this site carries a fatty diacid on a lysine side chain through a linker, attached late in the synthesis. A des-acyl variant never received the arm: far lighter, far more polar, eluting a long way ahead of the main peak. A doubly acylated variant received the arm twice, typically on a second amine: heavier by a full arm and later eluting. Linker-related variants include an incomplete linker, a positional isomer with the arm on the wrong amine, and chain-length homologs inherited from the diacid raw material, which appear as small satellites either side of the main peak.

Degradation products that form after release

Process impurities are fixed at release; degradation products keep forming. The classes are the same for any peptide, but the sequence decides which matter most. Semaglutide and tirzepatide both contain tryptophan; sequences derived from glucagon or oxyntomodulin, such as mazdutide, should be checked against the published sequence for methionine; and cagrilintide carries one disulfide.

Oxidation

Oxidation adds an oxygen atom (+16 Da). Methionine, where present, is the most sensitive residue and gives the sulfoxide; tryptophan gives several oxidized forms; histidine and tyrosine react more slowly. Oxidized species are more polar than the parent and elute ahead of it, usually as a distinct early peak; the reaction runs faster in solution than in the sealed solid.

Deamidation and isomerization

Deamidation converts asparagine, and more slowly glutamine, to the corresponding acid (+1 Da); asparagine followed by glycine or serine is the fastest site. The product elutes close to the parent, often as a front shoulder that integration may or may not split off. Isomerization of aspartate to isoaspartate goes through the same succinimide as the synthesis-stage aspartimide, carries no mass change, and gives a close-eluting peak that only a shallow gradient separates.

Aggregation and clipping

Acylated peptides self-associate in solution, and amylin analogues can form ordered fibrils. Non-covalent aggregates generally dissociate in the acidic, organic mobile phase of reversed-phase HPLC, so the method under-reports them; size exclusion sees them. Covalent dimers, including disulfide-linked forms for cagrilintide, survive the mobile phase and elute late and broad. Hydrolysis at acid-labile backbone bonds gives clipped fragments that elute early at reduced mass.

How each class appears in reversed-phase HPLC

The release chromatogram of a lipidated peptide has a characteristic shape. The fatty-diacid arm dominates retention: it is the most hydrophobic part of the molecule, and it anchors every variant that carries it to a narrow window of the gradient. Backbone differences that would spread cleanly for an unlipidated peptide are compressed into that window, so small variants often appear as shoulders rather than resolved peaks.

Where each class lands

Front shoulders and early satellites are the signature of oxidation, deamidation, isoaspartate and most deletions. Study this region on an expanded baseline: a shoulder the integrator does not split is counted inside the main peak. Late-eluting peaks are protected adducts, doubly acylated chains, heavier chain-length homologs and covalent dimers. Close-eluting isomers, meaning diastereomers and positional acylation isomers, sit so near the parent that resolution depends on gradient slope, column temperature and column chemistry.

The des-acyl variant is the exception that proves the compression: without the arm it elutes far earlier, so a clean early region is good evidence that the acylation step went to completion.

What a single dominant peak does and does not prove

It proves that, under this column, gradient, detector and integration, most of the UV-absorbing material elutes at one retention time. It does not prove identity, because a wrong or truncated peptide can also give one peak. It does not prove content, because water and counter-ion are not in the picture. It does not exclude isobaric variants co-eluting under the main peak, and it says nothing about salts, residual solvents or aggregates the mobile phase dissociated.

How mass spectrometry complements the chromatogram

Mass spectrometry answers the question HPLC cannot: is the material under the main peak the intended molecule? Electrospray ionization of peptides this size gives multiply charged ions, typically at z = 3, 4 and 5, and the deconvoluted spectrum should return the expected mass, about 4731 g/mol for retatrutide or 4114 g/mol for semaglutide.

Mass shifts that flag a class

The shift of a minor species relative to the parent is often enough to assign its class. A loss or gain of one residue mass points to a deletion or an insertion; a gain equal to a protecting group points to incomplete deprotection; a loss or gain of the whole arm points to des-acyl or double acylation. A +16 Da gain is oxidation, a −18 Da loss is a succinimide, and a species at twice the parent mass is a covalent dimer. With LC-MS rather than direct infusion, each shift is also tied to a retention time.

Isobaric blind spots

MS cannot separate species of the same mass. Diastereomers, aspartate against isoaspartate, positional isomers of the acyl arm, and leucine against isoleucine are invisible to intact-mass analysis. Deamidation is nearly so: a +1 Da shift at z = 4 or z = 5 falls inside the isotope envelope of the parent and needs high resolution and careful deconvolution. Nor is MS quantitative without calibration, because lipidated and unlipidated species ionize differently. HPLC gives the share, MS gives the assignment, and neither alone gives both.

Reading the Pepti records: ranges and lot-to-lot spread

Every lot in the public library is manufactured by Pepti, Guangzhou, and released with a batch certificate reporting the HPLC main-peak result and the MS identity check. The library held 76 entries when checked on 2026-10-19, and the main-peak values cluster per compound: semaglutide 98.7%–99.6% across 7 records, tirzepatide 98.4%–99.7% across 9, retatrutide 98.72%–99.55% across 9, cagrilintide 99.2% on both of its 2 records, and mazdutide 99.12%–99.37% on 2. Survodutide has no public record yet. The quality page links the full set.

Two records show what an entry looks like. PEP260819-RT20 is a retatrutide 20 mg lot manufactured 2026-08-19 and tested 2026-08-22, with a main peak of 99.26%. PEP260709-MZ10 is a mazdutide 10 mg lot manufactured 2026-07-09 and tested 2026-07-15, at 99.12%. Both sit inside their compound ranges, and neither value is a specification: the ranges describe what the process has delivered, not a limit for the buyer to apply.

Why spread inside the range is normal

The spread has mundane causes. Coupling efficiency varies from run to run; the purification cut can be taken wider or narrower; column lot, the baseline the integrator draws and the reporting threshold each move the area-percent by small amounts. Reported precision also differs between records, so two cagrilintide records at the same 99.2% show that the lots agreed at that precision, not that they were identical. What a buyer should query is a change in the shape of the profile rather than in the headline: a new late-eluting peak where earlier lots had none, or a front shoulder grown from a trace to the largest single impurity.

What to ask for beyond the headline number

The headline summarizes a chromatogram. For a lot a buyer intends to accept into stock, the following items turn the number back into evidence; each should refer to the lot alias on the certificate, not to a representative lot.

1. The chromatogram at full scale and with the baseline expanded, so the front region can be checked for shoulders and the late region for adducts and dimers.
2. The integration table: retention time, area and area-percent for every reported peak, plus the reporting threshold. This is where largest single impurity and total impurities come from, whether or not the certificate prints them.
3. The MS spectrum, ideally the raw multiply charged spectrum and the deconvoluted mass, and where LC-MS was run, the mass under the largest impurity peak as well as under the main peak.
4. The method summary in words: column chemistry, mobile-phase modifiers, the gradient described as shallow or steep, column temperature as ambient or elevated, and the detection region. Without it the area-percent cannot be compared across laboratories.
5. Lot specificity: the alias on the certificate should match the alias on the chromatogram header, the integration table and the vial label.

Storage-driven change and re-testing near the retest date

Of the classes in the table, four keep moving after release: oxidation, deamidation, isomerization and aggregation, plus moisture-driven hydrolysis once a container has been opened. Lyophilized GLP-1-family peptides are commonly held at −20 °C or 2–8 °C, dry and dark; the batch certificate states the condition for the lot, and the storage and cold-chain guide covers the handling.

A re-test near the retest date should compare the profile, not just the headline. New or grown peaks in the front-shoulder region are the usual signature of age; a new late peak suggests dimer formation. Run the release method where possible, file the new integration table beside the release table, and treat a shift in the profile as the trigger for a conversation with Pepti, with the lot alias and both tables attached.

Sources and verification

• Pepti COA record PEP260819-RT20 (retatrutide 20 mg) — Lot alias, strength, HPLC result and dates checked against the record
• Pepti COA record PEP260709-MZ10 (mazdutide 10 mg) — Lot alias, strength, HPLC result and dates checked against the record
• Pepti public COA library — 76 entries as checked 2026-10-19; per-compound main-peak ranges taken from the records
• Janoshik Analytical report #210105 — Content and purity figures for one retatrutide sample, analyzed 2026-10-07
• PubChem CID 56843331 (semaglutide) — Formula and molecular weight
• PubChem CID 171934787 (retatrutide) — Formula and molecular weight

Related pages

• Quality and batch records
• Retatrutide raw material
• Mazdutide raw material
• Semaglutide raw material
• Storage, cold chain and retest dates

This article explains documents and procedures; it is not regulatory or medical advice.