Chapter 34: Peptide Analysis and Quality Control — How Anyone Knows What's Actually in the Vial

"Not everything that can be counted counts, and not everything that counts can be counted." — William Bruce Cameron, Informal Sociology: A Casual Introduction to Sociological Thinking (1963)

Overview

Every claim in this book eventually runs into the same wall.

You can read a trial carefully. You can weigh the endpoints, notice the population, catch the underpowered subgroup analysis, and arrive at a defensible view of what a molecule does. And then someone hands you a small glass vial with a white powder in it, and none of that analysis transfers, because the trial studied a characterized drug substance and the vial is an unanswered question.

This chapter is about how that question gets answered — and, just as importantly, about the several parts of it that no analysis answers at all.

The machinery is genuinely interesting. Mass spectrometry determines a molecule's mass to a few parts per million and, in tandem mode, reads a peptide's sequence off its fragments. Chromatography can separate species differing by a single missing residue — sometimes. Amino acid analysis gives the composition of something nobody has seen before. A scientifically literate person should understand what each of these measures.

But there is a way of learning this material that goes badly wrong, and it is worth naming before we start. It goes: if I understand the analytical methods, I can evaluate a certificate, and if I can evaluate a certificate, I can source safely. Every step in that chain is false, and the last one is false in a way that has hurt people.

Analysis is not a substitute for a supply chain. A test result is a statement about a sample somebody submitted, at a moment in the past, using a method that looked for certain things and not others. A regulated pharmaceutical is not primarily a tested product; it is a controlled process with documented custody, identified accountability, and a route of recourse. Those are different kinds of object, and no quantity of chromatograms converts one into the other. Nothing here should be read as a route to making an unregulated purchase safe, and §34.9 explains why no such route exists.

What this chapter will give you is the ability to read an analytical claim and know exactly how much it covers — a real skill, uncommon, and one that makes you much harder to impress with a document.

In this chapter, you will learn to:

• Separate the three questions — identity, purity, and content — that get collapsed into the word "quality"
• Explain what a mass spectrum proves, and name two kinds of error it structurally cannot detect
• Read a chromatogram and state why a purity percentage is a floor on ignorance rather than a guarantee
• Explain why "98% pure" and "contains the labeled amount" are different claims, and why the second is the one people actually care about
• Distinguish sterility from endotoxin and say why sterilization does not address the second
• Read a certificate of analysis line by line and identify what each entry establishes
• Explain why no honest failure rate for the gray market exists, and why quoting one is a mistake
• State plainly why testing cannot substitute for oversight

Learning Paths

All five paths should read §34.4 and §34.9. Between them they contain the two ideas from this chapter that change how people think.

💊 GLP-1 — §34.4 and §34.7 are the chapter for you. The compounded and gray-market GLP-1 market runs almost entirely on documents, and the difference between a purity number and a content number is the difference between two very different exposures. Chapter 13 is the payoff. 🏋️ Performance — §34.1, §34.5, and §34.9. The performance market is where "research grade" and "third-party tested" do the most rhetorical work, and §34.5 covers the hazard that no amount of visual inspection detects. 🔬 Science — read straight through. §34.2, §34.3, and §34.6 are the analytical foundation that Chapter 32's synthesis chemistry implies, and §34.10 connects it to how the industry actually operates. 💄 Cosmetic — §34.4 and §34.7. Cosmetic peptide ingredients are supplied with certificates too, and the identity/content distinction determines whether an ingredient percentage on a label means anything. Chapter 30 assumes this chapter. 🏥 Clinical — §34.7, §34.8, and §34.9, and the 💊 callout in §34.8 specifically. Patients bring in vials and documents, and the useful clinical response is neither dismissal nor validation.

34.1 The four vials, revisited

Chapter 6 put four visually identical vials on a table and asked you to tell them apart. You could not, and neither can anyone else, and that was the point. Here they are again, compressed, because this chapter is the answer to the question that scene raised:

• Vial A — a pharmaceutical product dispensed by a licensed pharmacy.
• Vial B — genuine research-grade material, honestly labeled, accompanied by a real certificate, and never released or intended for human use.
• Vial C — an online vendor's product of unverifiable provenance.
• Vial D — mislabeled. Something else entirely, or the right molecule at the wrong strength, or nothing much at all.

Chapter 6's argument was that the four are indistinguishable by inspection. This chapter's argument is narrower and more uncomfortable: they are also not reliably distinguishable by testing. Testing can tell A from D. Testing can often tell C from D. What testing cannot do is tell A from B, because on a purely analytical basis there may be nothing to tell.

That last sentence is the one people collapse, so let us slow down on it.

Vial B is not fake. Readers get this wrong in both directions. One camp hears "research grade" and translates it as "counterfeit," which is simply false — a research chemical supplier producing well-characterized material for laboratory use is doing legitimate work, and its certificate may be entirely honest. The other camp hears "genuine material with a real certificate" and translates it as "the pharmacy product minus the markup," which is the more dangerous error, because it is closer to true and therefore more persuasive.

Here is the actual relationship. The molecule in Vial B may be chemically indistinguishable from the molecule in Vial A. The products are not remotely the same object.

WHAT DISTINGUISHES VIAL A FROM VIAL B
(assume, generously, that the molecule inside each is identical)

                                          VIAL A          VIAL B
                                       (pharmacy)    (research grade)
  Molecule identity confirmed              yes             yes
  Purity characterized                     yes             yes
  Content assayed against a standard       yes           sometimes
  Sterile fill under aseptic control       yes              no
  Endotoxin controlled and tested          yes           rarely
  Excipients selected for injection        yes              no
  Container/closure qualified              yes              no
  Stability program establishing expiry    yes              no
  Batch records retained and auditable     yes           partial
  Named party accountable for the batch    yes         not for human use
  Recall pathway if a defect is found      yes              no
  Adverse-event reporting route            yes              no
  Intended for administration to a person  yes             NO

  The molecule can be the same. Everything that makes a product a medicine is
  in the rows below the molecule.

Notice what the table is not saying. It is not saying that research-grade material is dirty. It is saying that "is this the right molecule and is it pure" occupies the top two rows of a twelve-row problem, and that the analytical techniques in this chapter address those two rows and no others.

Vial C is the ordinary case, and it is the one most people are actually holding. Vial C may contain exactly what it says. It may have started life as material chemically identical to Vial B's contents. What makes it Vial C is not its chemistry but the absence of anything connecting the vial in your hand to any verified history — who synthesized it, under what controls, what it was filled into, by whom, whether the batch it came from is the batch on any document that arrived with it, and what temperature it experienced between there and here.

Vial D is the failure everyone fears and the least interesting case, because it is the one analysis actually catches. Send Vial D to a competent laboratory and a mass spectrum will very likely reveal it. That is precisely why "get it tested" feels like a solution: it does solve the most vivid problem. It solves almost none of the others.

The chapter follows the table. Sections 34.2 through 34.6 work through what the analytical toolkit can establish — identity, purity, content, sterility, endotoxin, composition, sequence. Section 34.7 assembles those into a certificate. Section 34.8 asks what testing programs have found. And §34.9 returns to the ten rows the toolkit never touched.

🔍 Check Your Understanding

1. State in one sentence why testing can distinguish Vial A from Vial D but not Vial A from Vial B.
2. Someone argues: "Research-grade peptide is the same molecule the pharmacy sells, so paying pharmacy prices is just paying for branding." Identify the specific step where this argument substitutes a claim about the molecule for a claim about the product.
3. Which of the twelve rows in the diagram could, in principle, be addressed by testing a submitted sample? Which could not be addressed by any test performed after the fact?

34.2 Identity: mass spectrometry and what an exact mass proves

The first question is the crudest: is this the right molecule?

Mass spectrometry answers it better than anything else, and the reason is arithmetic. A peptide's mass is not an approximate property. It is the sum of the masses of its constituent atoms minus the water lost at each peptide bond, and atomic masses are known to extraordinary precision. If you know the sequence, you can calculate the mass to more decimal places than any instrument can measure. If you measure the mass, you have placed a very tight constraint on what the molecule can be.

What the instrument actually measures

A mass spectrometer does not weigh molecules. It measures the mass-to-charge ratio ($m/z$) of ions — charged particles — by observing how they behave in electric and magnetic fields. Three stages, in every instrument:

1. Ionization. The sample becomes gas-phase ions. For peptides the dominant method is electrospray ionization (ESI): a solution pushed through a fine charged capillary emerges as a spray of droplets that evaporate down to bare ions. ESI is soft — it charges molecules without smashing them, which is why it works on things as fragile as peptides.
2. Mass analysis. Ions are separated by $m/z$ using one of several physical principles — time-of-flight, quadrupole filtering, orbital trapping — which trade resolution against speed against cost.
3. Detection. Ion arrivals are counted, producing a spectrum: intensity plotted against $m/z$.

One feature of ESI matters for reading real spectra. Peptides typically pick up multiple charges, so a single peptide appears not as one peak but as a family — singly, doubly, triply charged species, each at a different $m/z$. Software deconvolutes this charge-state envelope back to one neutral mass. That is why one compound can produce five peaks.

READING A PEPTIDE MASS SPECTRUM (schematic)

  RAW SPECTRUM — one peptide, several charge states
  intensity
     │              ▉
     │         ▉    ▉         For a neutral mass M, an ion carrying n protons
     │    ▉    ▉    ▉    ▉    appears at m/z = (M + n × 1.007) / n
     │    ▉    ▉    ▉    ▉
     └────┴────┴────┴────┴──── m/z
        [M+5H]5+ [M+4H]4+ ...

  DECONVOLUTED — the same data, expressed as neutral mass
  intensity
     │                   ▉        ← one species, one mass
     │                   ▉
     └───────────────────┴──────── neutral mass (Da)
                     measured mass

  THE COMPARISON THAT MATTERS
    calculated mass from the claimed sequence :  (arithmetic, exact)
    measured mass from the sample             :  (instrument, with error)
    agreement within instrument error         →  consistent with the claim
    disagreement                              →  something is wrong

What a mass measurement proves

Stated precisely, and the precision is the whole point:

A matching mass proves that something in the sample has the expected mass.

That is a real result and it is not nothing. It rules out a great many substitutions, a great many wrong compounds, and most outright mislabeling. If a vial claims to contain a 3,500-dalton peptide and the spectrum shows a dominant species at 1,200 daltons, the conversation is over.

Tandem mass spectrometry (MS/MS) goes considerably further. The instrument isolates the ion of interest, fragments it — typically by collision with an inert gas — and measures the masses of the fragments. Peptides fragment preferentially at the peptide bond, producing a ladder of fragment ions. Each rung differs from the next by the mass of one residue. Read the differences and you read the sequence.

HOW MS/MS READS A SEQUENCE (schematic, four residues shown)

   sequence:      A  ——  B  ——  C  ——  D
                     ┊      ┊      ┊
   fragmentation at each peptide bond gives two ion series:

   from the N-terminus →   [A]   [A B]   [A B C]
   from the C-terminus ←         [B C D]  [C D]   [D]

   mass difference between consecutive fragments in a series
   = the mass of the residue that was added

   [A B] − [A] = mass of residue B
   [A B C] − [A B] = mass of residue C          ← this is how the sequence is read

   Do this across the whole ladder and the sequence falls out of the arithmetic.

This is genuinely powerful. MS/MS can confirm not just that the mass is right but that the residues are in the claimed order, which is a far stronger identity claim.

What a mass measurement does not prove

Four limitations, and the last two are structural rather than practical.

It does not prove that this is the only thing present. A mass spectrum shows what ionized and got detected. Species that ionize poorly under the chosen conditions may be present and nearly invisible. Mass spectrometry is not naturally quantitative — signal intensity depends on how readily a particular molecule ionizes, which varies enormously between compounds. A dominant peak means a dominant signal, not necessarily a dominant mass fraction.

It does not prove anything about the vial you are holding. It proves something about the material that entered the instrument. Whether that material came from your vial, whether your vial came from the batch on the paperwork, and what happened to your vial afterward are all outside the measurement. This will become the theme of §34.7.

Isomers and some substitutions share a mass. Leucine and isoleucine are structural isomers — the same atoms arranged differently — and their residue masses are identical. No mass measurement of any precision will ever distinguish them. Other pairs are not identical but are close enough to challenge lower-resolution instruments. A substitution that swaps one residue for another of identical or near-identical mass produces a molecule that a mass spectrum may pass without complaint.

Stereochemistry is completely invisible to mass. This is the important one. Every amino acid except glycine exists in two mirror-image forms, L and D, and biology uses L almost exclusively (Chapter 1). A D-amino acid substitution changes the three-dimensional shape of a peptide — and can abolish receptor binding entirely — while changing the mass by exactly zero. Mass spectrometry is structurally blind to it. Detecting stereochemical error requires methods that are sensitive to chirality, and those methods are not part of a routine identity check.

🧬 The Molecule — why "exact mass" has more than one meaning

If you calculate a peptide's mass two different ways you will get two different numbers, and both are correct. The difference confuses people constantly, including people writing certificates.

Monoisotopic mass uses the mass of the most abundant isotope of each element — carbon-12, hydrogen-1, nitrogen-14. It is the mass of the single most common isotopic version of the molecule. High-resolution instruments measure this, because they can actually resolve the individual isotopic peaks.

Average mass uses the natural isotopic abundance-weighted average mass of each element. Carbon in nature is about 1.1% carbon-13, so the average carbon is slightly heavier than 12. For a small peptide the two numbers differ by a fraction of a dalton. For a larger one they can differ by several daltons, because a molecule with hundreds of carbons very likely contains at least one carbon-13.

Why this matters practically: a mass that "doesn't match" sometimes just means the two numbers were computed on different conventions. Before concluding that a spectrum contradicts a claimed sequence, check which mass is being quoted. A certificate that reports a mass without saying which convention it used has left a genuine ambiguity in the record — a small thing, but the kind of small thing that separates a document written by someone doing the work from a document written by someone producing a document.

It also matters conceptually. The isotopic pattern in a high-resolution spectrum is itself information, because different elements have different isotopic signatures — sulfur's is distinctive enough to hint at how many cysteines and methionines a molecule contains. Instruments read more from a spectrum than the position of one peak.

📊 Evidence Rating

Claim: Mass spectrometry, including tandem MS, is a valid method for establishing the identity of a synthetic peptide. Rating: ✅ Strong evidence (as of this writing) Reason: This is a technical rather than a clinical claim, and it is about as well established as analytical chemistry gets — the physics is understood, the method is validated across decades, and it is the accepted identity technique in pharmacopeial practice and regulatory submissions worldwide. Tandem MS additionally establishes sequence, not merely composition. What would change it: Nothing plausible about the method itself. The rating is bounded rather than provisional: it applies to identity of the material analyzed, and it explicitly does not extend to stereochemistry, to isomeric substitutions such as leucine/isoleucine, to quantitation, or to any claim about a container that was not the source of the analyzed sample. A claim that stretches the method past those boundaries is not covered by this rating.

34.3 Purity: HPLC and what a chromatogram actually shows

Identity asks what is it. Purity asks what else is in there, and the workhorse is high-pressure liquid chromatography.

The principle

Chromatography separates a mixture by exploiting differences in how its components distribute themselves between two phases: a stationary phase, packed into a column, and a mobile phase, pumped through it. A component that interacts strongly with the stationary phase moves slowly. One that prefers the mobile phase moves quickly. Inject a mixture at one end, and the components emerge at the other end at different times.

For peptides the standard configuration is reversed-phase HPLC: a nonpolar stationary phase (hydrocarbon chains bonded to silica particles) with a polar aqueous mobile phase, run as a gradient in which the proportion of organic solvent increases over the run. More hydrophobic peptides stick harder to the nonpolar surface and require more organic solvent to release, so they come off later. The time at which a component emerges is its retention time.

A detector at the column outlet records signal against time. That trace is the chromatogram.

A CHROMATOGRAM — what you are actually looking at

  detector
  response
     │                        ▲
     │                       ╱█╲                    ← target peak, large area
     │                      ╱ █ ╲
     │            ▲        ╱  █  ╲         ▲
     │           ╱▒╲      ╱   █   ╲       ╱░╲
     │──────────╱  ▒╲────╱    █    ╲─────╱   ░╲──────
     └──────────┴────────┴──────────────┴────────────→ retention time
                impurity 1          impurity 2

  PURITY, as normally reported:
       area of target peak
    ──────────────────────────  × 100  =  "purity, % by area"
       total area of all peaks

  Read that formula again. It is a ratio of AREAS, of PEAKS the detector SAW,
  under ONE set of conditions. Every one of those three qualifications is a hole.

Purity is conventionally reported as the target peak's area expressed as a percentage of the total peak area. This is the number that ends up on certificates as "98.5%," and it is worth understanding exactly what it is a percentage of.

Four honest limitations

1. Co-elution. Two species that happen to have similar affinity for the stationary phase emerge at the same time and are recorded as one peak. The chromatogram shows a single clean peak; the sample contains two compounds. This is not a rare pathology — it is the fundamental limitation of any separation method, and it is why analytical chemists speak of a method's specificity as something that must be demonstrated rather than assumed. Demonstrating it requires running the method against deliberately stressed or spiked samples to show that known related species do in fact separate. A certificate that reports a purity number without describing the method has not told you whether that demonstration ever happened.

2. The detector only sees what it responds to. Peptide HPLC is usually monitored by ultraviolet absorbance. Two wavelengths dominate: around 214 nanometers, where the peptide bond itself absorbs, and around 280 nanometers, where aromatic side chains — tryptophan, tyrosine, phenylalanine — absorb. The choice matters enormously. At 280 nm, a peptide containing no aromatic residues is nearly invisible; so is a residual solvent, an inorganic salt, or a non-peptide contaminant. At 214 nm you see peptide bonds, which is better for this purpose, but you still do not see species lacking that chromophore. Anything the detector does not respond to contributes zero to both the numerator and the denominator of a purity calculation. It is not counted as an impurity. It is not counted at all. It is, from the arithmetic's point of view, not there.

This is the single most misunderstood feature of a purity number. "98% pure" does not mean "2% of the mass is something else." It means "of the material this detector saw, 98% of the signal was in one peak."

3. Peak area is not mass. Converting an area into an amount requires a response factor — how much signal a given mass of a given compound produces. Response factors differ between compounds, sometimes substantially, because absorbance depends on structure. A related impurity that absorbs more strongly than the target will be over-represented in an area-percent calculation; one that absorbs less will be under-represented. Area percent is a useful, standardized, comparable number precisely because it does not attempt this correction — everyone computes it the same crude way. But it should not be read as a mass fraction, and it constantly is.

4. Deletion sequences may not separate at all. This is where Chapter 32 comes back to collect. Solid-phase synthesis builds a peptide one residue at a time, and the characteristic failure is an incomplete coupling step that leaves a fraction of the growing chains missing one residue (§32.3). The result is a deletion sequence — a molecule differing from the target by a single absent amino acid.

Now ask what such a species does on a reversed-phase column. Its hydrophobicity is nearly the same as the target's, because it differs by one residue out of dozens. Its retention time is therefore nearly the same. It may appear as a shoulder on the main peak, or as a slight asymmetry, or as nothing distinguishable whatsoever — and if it is not resolved, its area is counted inside the target peak, inflating the reported purity.

This is why deletion sequences are the characteristic impurity of synthetic peptides rather than an occasional accident. They are produced by the standard failure mode of the standard method, and they are the impurity class the standard purity assay is least able to see. A peptide impurity profile is not a random assortment of foreign compounds; it is a family of near-relatives of the target molecule, generated by the synthesis itself, differing from it by one residue or one protecting group or one oxidation state — which is exactly the population a separation method has the hardest time resolving.

🔬 Read the Study — how to read an analytical methods section

When a paper, a supplier document, or a regulatory filing reports peptide purity, the number is meaningless without the method. Here is what to look for, and what each absence tells you.

Which technique, and in what mode. "HPLC purity 98%" is incomplete. Reversed-phase HPLC separates by hydrophobicity. Ion-exchange separates by charge. Size-exclusion separates by hydrodynamic size. These see different impurities. A species invisible to one may be obvious to another, which is why serious characterization uses orthogonal methods — techniques whose separation principles are independent, so that co-elution in one is unlikely to be reproduced in the other. A single method, however good, cannot rule out co-elution by itself.

What detector, at what wavelength. Per limitation 2 above, this determines what was counted at all. A purity number monitored at 280 nm on a peptide with one tyrosine is a much weaker statement than the same number at 214 nm.

Was the method shown to be specific. Validated methods are demonstrated to separate the target from known related substances, usually by challenging them with stressed samples. Language like "stability-indicating" signals that this work was done.

Is anything reported besides area percent. Water content, counterion content, residual solvents, and peptide content are separate determinations. Their presence indicates a characterization program. Their absence indicates a single test.

How the sample was obtained. In a paper this is usually stated. In a supplier document it usually is not, and that omission is the subject of §34.7.

The habit to build: when you see a purity percentage, ask "percent of what, measured how, seen by which detector." Most purity claims do not survive those three questions, and the ones that do are making a genuinely useful statement.

34.4 Potency, purity, and content — three different questions

If you retain one section of this chapter, retain this one.

The word "quality" hides three questions that are routinely treated as one. They are answered by different methods, they can come apart dramatically, and the one people actually care about is the one certificates most often omit.

THE THREE QUESTIONS

  ┌──────────────┬──────────────────────────────┬────────────────────────────┐
  │  QUESTION    │  WHAT IT ASKS                │  HOW IT IS ANSWERED        │
  ├──────────────┼──────────────────────────────┼────────────────────────────┤
  │  IDENTITY    │  Is it the right molecule?   │  MS; MS/MS; sequencing;    │
  │              │                              │  amino acid analysis       │
  ├──────────────┼──────────────────────────────┼────────────────────────────┤
  │  PURITY      │  Of the material detected,   │  HPLC, reported as target  │
  │              │  what fraction is target?    │  peak area ÷ total area    │
  ├──────────────┼──────────────────────────────┼────────────────────────────┤
  │  CONTENT     │  How much target peptide is  │  Quantitative assay        │
  │  (potency)   │  in the vial, BY MASS?       │  against a reference       │
  │              │                              │  standard; or AAA;         │
  │              │                              │  or nitrogen determination │
  └──────────────┴──────────────────────────────┴────────────────────────────┘

  IDENTITY says WHAT.  PURITY says WHAT FRACTION OF WHAT WAS SEEN.
  CONTENT says HOW MUCH.  Only the third is a statement about the vial's dose.

Identity is §34.2. Is this the molecule the label names?

Purity is §34.3. Among the species the analysis detected, what proportion of the signal belongs to the target?

Content, also called potency or assay, asks the question a person actually has when they pick up a vial: how many milligrams of the target peptide are physically in here? Answering it requires quantifying the peptide against a reference standard — a well-characterized material of known content — rather than merely comparing peaks to one another.

Why purity and content come apart

Here is the fact that surprises nearly everyone, including some people selling peptides:

A vial can be 98% pure by HPLC and contain substantially less peptide by mass than the label states. Not because anyone cheated. Because purity and content are measuring different things, and the gap between them is filled by material that is not an impurity in the chromatographic sense at all.

Three components account for most of it.

Counterion. Synthetic peptides are typically purified by reversed-phase HPLC using an acidic mobile phase, and the acid most commonly used is trifluoroacetic acid. Peptides with basic residues — lysine, arginine, histidine, and the N-terminus itself — carry positive charges, and those charges require negative counterions. The peptide therefore comes off the purification as a salt, most often the trifluoroacetate salt. That counterion is real mass, physically present in the vial, and it is not the peptide. For a peptide with several basic residues, the counterion can account for a meaningful double-digit percentage of the total weight.

Water. Peptides are hygroscopic. Lyophilized peptide powder absorbs water from the atmosphere, and a lyophilized cake retains residual moisture from the drying process. Water is mass. It weighs the same as anything else on a balance and contains no peptide whatsoever.

Residual salts and solvents. Buffer salts from purification steps, trace organic solvents, and other process residues all contribute weight.

WHAT IS ACTUALLY IN THE VIAL, BY MASS
(schematic — proportions vary widely with sequence and process)

  ┌──────────────────────────────────────────────────────────┬───────┬───────┐
  │              PEPTIDE                                     │counter│ water │
  │                                                          │  ion  │ +salt │
  └──────────────────────────────────────────────────────────┴───────┴───────┘
   ←──────────── this is what "peptide content" measures ────→

  ←────────────────── this is what the vial WEIGHS ──────────────────────────→

  HPLC purity looks only INSIDE the peptide fraction, and asks what proportion
  of THAT is the target rather than a related peptide species. It says nothing
  about how wide the peptide fraction is relative to the whole.

  So: "98% pure" and "the vial contains the labeled milligrams of peptide"
  are not merely different claims. They are claims about different diagrams.

"Purity" and "how much drug is here" are different numbers, and a certificate frequently reports only the first. That is the practical core of this chapter. A supplier weighing out material to fill vials, using the gross powder weight, will deliver less peptide than the label states by exactly the fraction that is counterion, water, and salt — and every analytical number on the accompanying certificate can be accurate.

Note what this is and is not. It is not fraud. It is a specification question: does "5 mg" mean 5 mg of gross powder or 5 mg of peptide? Pharmaceutical labeling resolves this by convention and by regulation — the label refers to the active substance — and the manufacturing process is built around delivering that. Outside that framework, the question is often simply unaddressed, and an unaddressed question in a specification is not a small thing. It is the difference between two exposures that a person would experience as different drugs.

How content is actually determined

Three approaches appear in practice, and their differences are instructive.

Quantitative HPLC against a reference standard. Convert the sample's peak area into an amount using a known quantity of characterized reference material. Precise, and entirely dependent on a standard whose own content is known — which pushes the problem back one step.

Amino acid analysis. Hydrolyze and quantify the liberated amino acids against amino acid standards, which are available in highly pure form. Because the sequence fixes how many of each residue there should be, this yields content directly without needing a peptide reference standard — which is why AAA, a technique that sounds antique, remains a primary content method (§34.6).

Nitrogen determination. Peptides contain nitrogen in a proportion fixed by the sequence, so total nitrogen gives content. Robust, and vulnerable to any other nitrogen-containing material present.

Each has a different failure mode, which is why serious characterization uses more than one.

🧬 The Molecule — the trifluoroacetate that came along for the ride

Trifluoroacetate deserves its own callout, because it is the most consequential molecule in this chapter that nobody is trying to make.

Trifluoroacetic acid is used in peptide synthesis and purification for good reasons: it is a strong acid, it is volatile, it improves peak shape on reversed-phase columns, and it is used in the cleavage step that releases the finished peptide from its solid support (Chapter 32). By the time a peptide has been through synthesis and preparative HPLC, its basic groups are paired with trifluoroacetate counterions.

Three consequences follow.

It is mass. As described above, it contributes weight that is not peptide. A peptide with several basic sites can carry several counterions, and each one weighs more than a hundred daltons.

It is not always inert in biological systems. Trifluoroacetate has documented effects in cell culture, which is why researchers doing cell work often specify a different salt form. Whether a given amount matters depends entirely on the amount and the system, and this book is not going to pretend to a quantitative answer it does not have. The relevant point is structural: the counterion is a real chemical entity with its own properties, not a bookkeeping artifact.

Pharmaceutical products control it. Salt exchange to an acceptable counterion — acetate, chloride — is a defined manufacturing step, and residual trifluoroacetate is a controlled specification with a stated limit. This is a nice example of the difference between a molecule and a product. The molecule is the same either way. Which counterion it arrived with, and how much, and whether anyone measured, are properties of the process.

When you read a certificate that reports "purity 99%" and nothing else, ask yourself what fraction of that vial is trifluoroacetate. The honest answer is that the certificate does not say, and if nobody measured, nobody knows.

📊 Evidence Rating

Claim form — "99% purity" establishes that a vial contains 99% of the labeled peptide by mass. Rating: ❌ Hype outpaces evidence (as of this writing) Reason: Purity by HPLC area percent and peptide content by mass are different quantities measured by different methods. Area percent is a ratio among detected peaks and is blind to counterion, water, and salts, which can together constitute a substantial mass fraction of a lyophilized peptide. A high purity figure is fully compatible with a vial containing considerably less peptide than its label states. What would change it: Nothing about the underlying chemistry — this is a definitional error rather than an empirical one. The specific vial claim would be supported by a quantitative content assay against a reference standard, plus water and counterion determination, performed on the batch in question. Note that this rating targets the inference from purity to content. It does not allege that any given purity number is false.

34.5 Sterility and endotoxin: separate tests for separate hazards

Everything so far has been about the peptide. This section is about the two hazards that have nothing to do with the peptide at all, and it addresses the most consequential confusion in the entire quality-control conversation.

Sterility asks whether viable microorganisms are present in the product. It is assessed by attempting to culture organisms from the material — putting a sample into growth media that support bacteria and fungi, incubating, and observing whether anything grows.

Endotoxin asks something different. Endotoxins are lipopolysaccharides from the outer membrane of Gram-negative bacteria. They are structural molecules, not organisms, and they are released when bacterial cells die and break apart. Injected into a person, endotoxin triggers a powerful innate immune response: fever, chills, inflammatory signaling, and at sufficient exposure a severe systemic reaction.

Chapter 19 coined the formulation and it is worth repeating exactly:

Sterility asks whether anything is alive in the vial. Endotoxin asks whether anything ever was.

That distinction is not a rhetorical flourish. It is a statement about physical chemistry with serious practical consequences, because of one fact:

Endotoxin survives the processes that achieve sterility.

WHY STERILIZATION DOES NOT ADDRESS ENDOTOXIN

  A vial of solution containing live Gram-negative bacteria
              │
              ├─── autoclaved (moist heat) ────→  bacteria killed
              │                                   ENDOTOXIN REMAINS
              │                                   (and is released from the
              │                                    ruptured cells)
              │
              └─── sterile-filtered ──────────→  bacteria removed by the filter
                                                  ENDOTOXIN PASSES THROUGH
                                                  (LPS molecules and fragments are
                                                   far smaller than the pore size
                                                   that excludes a bacterium)

  RESULT IN BOTH CASES:
     sterility test  →  PASSES
     endotoxin       →  present and undetected unless separately tested
     injected        →  fever, chills, systemic inflammatory response

  Endotoxin removal requires DIFFERENT processes than sterilization, and
  endotoxin detection requires a DIFFERENT test. Passing one says nothing
  about the other.

Read that diagram carefully, because it explains a category of harm that surprises people. A preparation can be genuinely, verifiably sterile — no living organism in it — and still cause a serious febrile reaction on injection, because the bacteria that were once there left their cell walls behind. Heat kills them but does not destroy lipopolysaccharide at temperatures an autoclave cycle reaches. Filtration excludes them, but a filter designed to exclude bacteria does not exclude molecules orders of magnitude smaller.

Endotoxin is therefore controlled by preventing bacterial growth in the first place — by controlling water quality, the manufacturing environment, and the time material spends in conditions where organisms could proliferate. It is a process control problem, not a terminal-step problem. Regulated manufacture additionally depyrogenates containers and equipment under conditions well beyond ordinary sterilization. And it is tested separately, by an assay using entirely different chemistry from a sterility test.

Endotoxin testing is routinely not performed outside regulated manufacture. That is the practical bottom line. A research chemical supplier producing material for laboratory use has no reason to perform it, because nobody is supposed to be injecting the material. The absence is not necessarily a failure by the supplier — it is a consequence of what the product is for. It becomes a problem entirely and only when the material's actual use diverges from its intended one.

🩺 Safety and Risk — the hazard with no visual signature

Almost every other quality problem in this chapter has, in principle, some observable correlate. Wrong molecule, wrong amount, degraded material — these are at least the kind of thing that a laboratory can detect from a sample, and some of them occasionally announce themselves as visible particulates or discoloration or a cake that will not dissolve properly.

Endotoxin has none of that. A solution containing enough endotoxin to make someone acutely, memorably ill is visually indistinguishable from one containing none. It is clear. It smells like nothing. It passes a sterility test. It dissolves normally.

The clinical picture, when it happens, is characteristic: fever and rigors beginning within hours of administration, often with headache, malaise, tachycardia, and sometimes hypotension. It is commonly mistaken for a coincidental viral illness, which means it is probably underrecognized as a cause. Someone who develops fever and shaking chills in the hours after injecting an unregulated preparation has a genuine medical situation, and the relevant action is medical evaluation — not waiting to see, and not troubleshooting the vial.

The reason this callout sits in an analysis chapter is that endotoxin is the clearest possible case of a hazard that a purity certificate does not address, cannot address, and is not trying to address. A document reporting HPLC purity of 99% is silent on this hazard in the same way that a car's paint inspection is silent on its brakes. The number is not wrong. It is about something else.

As always, decisions about any of this belong with a clinician who knows your history.

34.6 Amino acid analysis and sequencing

Two older techniques remain in use because they answer questions mass spectrometry answers less directly.

Amino acid analysis: composition without order

Amino acid analysis (AAA) hydrolyzes the peptide completely — breaking every peptide bond, typically under strongly acidic conditions at elevated temperature — and then separates and quantifies the liberated free amino acids. The output is a table: how much alanine, how much glycine, how much lysine, and so on.

From that table you get composition: which amino acids are present and in what molar ratios. If the claimed sequence contains four prolines and one lysine, the analysis should find proline and lysine in a 4:1 ratio. If it finds 3:1, something is wrong.

AAA has three real strengths. It is quantitative in a way mass spectrometry is not, which is why it serves as a primary content method (§34.4). It requires no peptide reference standard, only amino acid standards, which are available in exceptionally pure form. And it is largely independent of the peptide's own properties — it does not care whether the molecule ionizes well or absorbs at a convenient wavelength.

It also has characteristic limitations, and they are chemical rather than instrumental. Acid hydrolysis destroys tryptophan. It converts asparagine to aspartate and glutamine to glutamate, so those pairs cannot be distinguished — an analysis reports the sum. Certain bonds, particularly between adjacent bulky hydrophobic residues, hydrolyze slowly and may be incompletely cleaved. These are known, corrected for where possible, and simply accepted where not.

But the fundamental limitation is not a chemical artifact. It is logical: composition is not sequence.

Chapter 1's case study on Frederick Sanger's determination of the insulin sequence made exactly this point, and it remains the cleanest illustration available. Composition had been known for insulin before Sanger; what Sanger established, at enormous effort, was the order. The reason that took years is that composition does not constrain order very much at all.

COMPOSITION VERSUS SEQUENCE

  Composition:  {Gly, Ala, Ser, Lys}   — one each

  Consistent sequences include:
        Gly-Ala-Ser-Lys
        Gly-Ala-Lys-Ser
        Gly-Ser-Ala-Lys
        Ala-Gly-Ser-Lys
        Ser-Lys-Gly-Ala
        ... and 19 others. For four distinct residues: 4! = 24 arrangements.

  For a 15-residue peptide the number of distinct arrangements consistent with a
  given composition runs into the billions. Composition eliminates a lot. It does
  not identify.

  A molecule with the right composition and the wrong order is a DIFFERENT
  MOLECULE with, in general, NO biological activity at the intended receptor.

So AAA establishes something genuinely useful and structurally weaker than what MS/MS establishes. A certificate reporting amino acid analysis has made a real composition claim. It has not made a sequence claim.

Establishing order: Edman degradation and MS/MS

Edman degradation reads a sequence directly, one residue at a time, from the N-terminal end. The chemistry, developed in the 1950s, labels the free N-terminal amino group, then cleaves that single labeled residue off under conditions leaving the rest of the chain intact. Identify the released residue; the shortened peptide now has a new free N-terminus; repeat. Each cycle yields one letter.

Its limitations define its use. It requires a free N-terminus, so a peptide whose N-terminal end is chemically blocked — a common and deliberate modification in peptide drugs — cannot be sequenced this way at all. Each cycle is slightly less than perfectly efficient, so signal degrades along the chain and readable length is limited.

MS/MS sequencing (§34.2) has largely displaced Edman for routine work: faster, less material, tolerant of blocked termini, able to handle mixtures. Edman retains a niche precisely because it is orthogonal — it reads sequence by an entirely different physical principle, which makes it valuable for confirming an MS/MS assignment rather than merely repeating it.

Together these define a hierarchy of identity claims, and it is worth holding the hierarchy in mind when reading any document:

STRENGTH OF AN IDENTITY CLAIM (weakest to strongest)

  1. "It came in a container labeled X"
        ← not an analytical claim at all

  2. "The correct mass was observed"
        ← constrains identity strongly; blind to isomers and stereochemistry

  3. "The amino acid composition matches the claimed sequence"
        ← rules out many wrong molecules; does not establish order

  4. "The sequence was confirmed by MS/MS"
        ← establishes order; still blind to stereochemistry

  5. "Sequence confirmed by two orthogonal methods, plus purity by orthogonal
      chromatography, plus content by quantitative assay against a standard"
        ← this is what characterization actually looks like

  Most documents in the consumer peptide space live at level 1 or 2 and are
  read by their recipients as though they were at level 5.

34.7 Reading a certificate of analysis

A certificate of analysis — universally abbreviated CoA — is a document reporting the results of tests performed on a material. In regulated pharmaceutical manufacture it is a formal record, part of a batch's documentation, generated under a quality system, signed by a qualified person, and auditable. In the consumer peptide market the same three letters name something that ranges from a serious analytical report to a PDF.

This section is about reading one. It is the practical core of the chapter, and it consists of two lists.

Six things a CoA can establish, when properly done

1. Identity, by mass spectrometry. The material analyzed had the expected mass, and — if MS/MS was performed and reported — the expected sequence.

2. Purity, by HPLC, with the method stated. The proportion of detected signal attributable to the target peak, under stated chromatographic conditions with a stated detector and wavelength. The qualification "with the method stated" is not decorative. A purity number without a method is not a result; it is a number.

3. Peptide content, by quantitative assay. How much peptide the material actually contains, by mass, determined against a reference standard or by amino acid analysis. This is the entry that most often does not appear, and per §34.4 it is the entry that most determines what a person would actually be exposed to.

4. Water content. Typically determined by a moisture-specific method. Real mass, separately measured.

5. Counterion content. How much trifluoroacetate or acetate is present. Again, real mass, separately measured, and — together with water content and peptide content — the entries that make the mass balance of the vial comprehensible.

6. Sterility and endotoxin results, if performed. Two separate results (§34.5), each present or absent independently, and both usually absent outside regulated manufacture.

THE SKELETON OF A COMPLETE CoA
(the structure; a worked line-by-line example is in Case Study 34.1)

  ┌───────────────────────────────────────────────────────────────────────┐
  │  IDENTIFICATION OF THE MATERIAL                                       │
  │    product name · batch or lot number · manufacture date              │
  │    ← every result below is a claim about THIS batch and no other      │
  ├───────────────────────────────────────────────────────────────────────┤
  │  APPEARANCE            visual description                             │
  │  IDENTITY              method · result · specification                │
  │  PURITY                method · conditions · detector · result · spec │
  │  PEPTIDE CONTENT       method · result · specification                │
  │  WATER                 method · result · specification                │
  │  COUNTERION            method · result · specification                │
  │  RESIDUAL SOLVENTS     method · result · specification                │
  │  STERILITY             method · result           (if performed)       │
  │  ENDOTOXIN             method · result           (if performed)       │
  ├───────────────────────────────────────────────────────────────────────┤
  │  AUTHORIZATION         who tested · who approved · when · signature   │
  └───────────────────────────────────────────────────────────────────────┘

  THREE COLUMNS, ALWAYS. A result without a METHOD is uninterpretable.
  A result without a SPECIFICATION has no pass/fail meaning — it is a
  measurement with nothing to be measured against.

That third column deserves emphasis. A specification is a predetermined acceptance criterion: the value the material must meet, decided before the test was run. Results reported without specifications cannot fail. They can only be reported. A document consisting entirely of results with no acceptance criteria is a description, not a release decision, and the distinction is exactly the one §34.10 develops.

Four things a CoA cannot establish — and these matter more

1. That the certificate corresponds to the vial in your hand.

This is the foundational problem and everything else is downstream of it. A CoA describes a sample. Somebody took some material, submitted it, and received results. Between that sample and your vial sit a series of assumptions: that the sample came from the same batch, that the batch was homogeneous, that the material was filled into vials without substitution or contamination, that the vial you received came from that fill, that the document you received was generated for that batch rather than reused from another, and that the entity supplying the vial is the same entity that commissioned the test. Each assumption is ordinary and each is unverifiable from the document.

2. Anything about handling after testing.

Testing happened at a moment. Since then the material has had a history: storage temperature, temperature excursions in transit, light exposure, humidity, freeze-thaw cycles, time. Peptides degrade — by oxidation of methionine and cysteine, by deamidation of asparagine and glutamine, by hydrolysis, by aggregation (Chapter 32). A certificate generated from material in good condition says nothing about material that spent a week in a delivery vehicle in July. The document is a photograph of a past state and is not updated by events.

3. Anything the method did not look for.

A CoA reporting HPLC purity says nothing about endotoxin. It says nothing about sterility. It says nothing about heavy metals, residual solvents, or elemental impurities. It says nothing about stereochemistry. It says nothing about content unless content was assayed. Absence of a finding is not a finding of absence — where no test was run, there is no information, and a document is not required to advertise the tests it omitted. The most consequential part of a certificate is frequently the part that is not on it, and reading for absence is a skill that has to be deliberately practiced, because human attention naturally follows what is present.

4. Its own authenticity.

A document is a file. It can be copied, edited, generated, or reused. A logo is an image. A signature is a picture of a signature. Nothing in a document's own content establishes that the tests described were performed, by whom, or on what. In a regulated environment this problem is addressed not by making documents harder to forge but by embedding them in an auditable system — where records are retained, where an inspector can trace a result back to instrument data and to the analyst who generated it, and where the manufacturer has a legal identity that can be held responsible. Outside such a system, the document is simply an assertion by an interested party.

The general principle, which is worth memorizing in this form:

A certificate of analysis is a claim about a sample. It is not a property of a vial.

Analysis characterizes material that was analyzed. Vials acquire their properties from the entire process that produced and delivered them. The certificate is a window onto one moment of that process, at one location, viewed through the specific methods that were chosen.

⚠️ Hype Check — "third-party tested"

The claim, in its usual form:

"All our products are third-party tested for purity and identity. We publish the CoA for every batch."

What's true in it. Independent analysis is genuinely better than none, and a supplier who commissions testing and publishes the results is doing more than one who does not. The phrase points at something real. Third-party testing exists as a serious practice, laboratories doing it are often technically excellent, and the results they produce are usually accurate descriptions of the samples they received.

Where it fails. Four places, and they compound.

First, "tested" is not one thing. Identity, purity, content, sterility, endotoxin, residual solvents, and heavy metals are separate tests, addressing separate questions, at separate cost. The phrase "third-party tested" specifies none of them. In practice it most commonly refers to identity and HPLC purity — the two cheapest — which are precisely the two that say least about exposure (§34.4) and nothing at all about the hazards in §34.5.

Second, the party is third but the sample is first-party. Independent analysis means the laboratory is independent. It does not mean the sample was independently obtained. The supplier selected the material, packaged it, and sent it. The laboratory analyzed what arrived. Every guarantee of independence attaches to the analysis and none of it attaches to the sampling, which is the step where the correspondence between document and product is actually established or lost.

Third, the document describes a batch and you have a vial. Per limitation 1 above.

Fourth, the phrase is doing rhetorical work that its content does not support. "Third-party tested" is deployed to answer the question is this safe to inject, and it is not an answer to that question. It is a partial answer to a much narrower question about a sample's chemical composition.

Verdict: the phrase describes an activity, not an outcome. Asking "tested for what, on which sample, against what specification, and how does that relate to the container I received" converts it back into a question — and the question does not have a good answer available, which is the actual finding.

📊 Evidence Rating

Claim form — "third-party tested" establishes that a product is what its label says. Rating: ❌ Hype outpaces evidence (as of this writing) Reason: The phrase does not specify scope, and identity, purity, content, sterility, and endotoxin are separate determinations that are rarely all performed. More fundamentally, a certificate describes a supplier-provided sample at a past moment rather than the container in a buyer's possession, so even a complete and accurate report leaves the correspondence between document and vial unestablished. What would change it: For a specific product, the claim would be supported by independently drawn samples rather than supplier-submitted ones, a stated and complete test panel including content and endotoxin, stated specifications rather than bare results, verifiable batch traceability from the tested lot to the individual container, and an auditable quality system standing behind the document. That combination is a description of regulated manufacture, which is the point of §34.9.

34.8 What independent testing programs have found

Various independent efforts have analyzed peptide products obtained from unregulated sources — academic groups, journalists, professional societies, and regulators sampling the market. Their findings have been reported in the categories this book has been naming since Chapter 6:

• Wrong identity. Material that is not the labeled compound, or is a related but different compound, or is a mixture including compounds not on the label.
• Content not matching the label. Substantially less peptide than stated, and sometimes substantially more. Both directions are documented, and the second is not a bonus — an unexpectedly concentrated product is an unexpected exposure.
• Unexpected species present. Related peptide impurities, degradation products, process residues, and occasionally entirely unrelated substances.
• Inadequate documentation. Missing certificates, certificates that do not match the product, certificates lacking methods or specifications, and documents that could not be traced to any laboratory.

Those are the categories. Now the harder point.

Why this book will not give you a failure rate

You will encounter percentages. "X% of tested peptide products failed." Those numbers circulate, they are memorable, they are quoted confidently, and no defensible figure exists. Here is why, in four parts:

Samples are not randomly drawn. Testing programs do not sample the market at random, because there is no enumerable market to sample from. Products get tested because someone was suspicious of them, because they were seized, because they were cheap enough to buy in bulk, or because they were popular enough to be worth investigating. Every one of those selection mechanisms is correlated with the outcome being measured. A sample selected for suspicion will show more problems than the market average; a sample drawn from the largest and most established vendors will show fewer. Neither estimates the population.

The market changes constantly. Suppliers appear and disappear, sources shift, and the interval between a testing effort and its publication is often long enough for the specific products tested to no longer exist. A snapshot from one period does not describe another.

Methods and thresholds vary between programs. One program's "failure" is content outside ±10% of label; another's is outside ±20%; a third counts only identity failures. One tests for endotoxin; most do not. Aggregating these into a single percentage combines quantities that are not the same quantity, which is a category error dressed as arithmetic.

Negative results are less likely to be published. A program that tests twenty products and finds twenty acceptable ones has produced a less publishable result than one finding five failures. This is ordinary publication bias, and it means the observed literature is enriched for problems relative to whatever the truth is.

WHY THE PERCENTAGE HAS NO DENOMINATOR

  A percentage requires:        Do we have it?

    a defined population        NO — the set of unregulated peptide products
                                     in existence is unknown and unstable

    a random or otherwise       NO — samples are selected by suspicion,
    characterized sample             availability, popularity, or seizure

    a consistent definition     NO — "failure" means different things in
    of the outcome                   different programs

    unbiased publication        NO — clean results are less likely to appear

  Four missing ingredients. Any number produced anyway is a figure with the
  shape of a statistic and none of the properties of one.

So the honest statement is categorical rather than quantitative: independent testing has repeatedly found products that were not what their labels claimed, across all four categories above, in enough instances and from enough independent sources that the phenomenon is not in doubt. What is in doubt — genuinely, not rhetorically — is how common it is.

And notice that this is not a hedge in the market's favor. "We do not know the rate" is not reassuring. It means you cannot calculate your exposure, which is worse than a high known rate, because a high known rate is at least a thing you can reason about. Uncertainty is not the same as safety, and treating an absent number as a low number is one of the most common errors in this entire domain.

This book applies the same standard to itself that it applies to marketing. A confident percentage would make this chapter more persuasive and less honest. Chapter 5's rules are not suspended when the conclusion is one we like.

💊 In the Clinic — when someone brings in a vial and a document

Clinicians encounter this. A patient arrives with a vial obtained outside the regulated supply chain, sometimes already using it, often with a printed certificate they have read carefully and found reassuring. The certificate is frequently the reason they feel comfortable, and they may have put real effort into evaluating it.

Two responses are common and both are poor.

Dismissal — "that's not a real drug, throw it away" — ends the conversation and, more importantly, ends disclosure. A patient told the thing they are using is not worth discussing generally keeps using it and stops mentioning it, which removes the clinician's ability to interpret symptoms, laboratory results, or interactions. The information loss is the harm.

Validation — "well, at least it's been tested" — accepts a document as evidence of a product's properties and, worse, communicates that the patient's evaluation framework was sound. It was not. The framework is the subject of this chapter.

A more useful response works from what the document is: a description of a sample, at a past moment, using methods chosen by whoever commissioned it. The questions it leaves open are the ones that determine what happens in a person — how much peptide is present, whether it is sterile, whether it contains endotoxin, whether any of it describes this container. Those are not answerable from the document and not answerable by further testing either, and saying so plainly is more respectful of a patient's intelligence than either dismissing or endorsing.

What remains available to a clinician is the part that does not depend on the vial's contents: knowing what is being used, monitoring for effects and adverse effects, evaluating symptoms with the exposure in mind — particularly the fever-and-rigors picture in §34.5 — and keeping the conversation open. Chapter 39 is about how to have it from the patient's side. Chapter 40 addresses the clinician's.

34.9 Why analysis is not a substitute for oversight

This is the section the chapter exists for.

Suppose everything goes right. Suppose a person submits material for analysis, and suppose the analysis is impeccable: correct identity by MS/MS, purity by two orthogonal chromatographic methods, content by quantitative assay against a characterized reference standard, water and counterion determined, sterility tested, endotoxin tested. Suppose every result meets a sensible specification.

That is far more analysis than anyone in the gray market performs. Grant it entirely. What is still unknown?

Batch-to-batch variation. You tested that batch. Synthesis is a process, processes vary, and the impurity profile of a peptide depends on how well each of dozens of coupling and deprotection steps proceeded on that particular run. A subsequent batch from the same supplier using the same route is a different batch. Without process control, a good result on one batch is a fact about that batch and a hope about the next.

Chain of custody. The vial you received is not the vial that was tested — testing consumes the sample. So there is a chain: material was made somewhere, filled somewhere, shipped, perhaps relabeled, perhaps repackaged, before arriving. Every link is an opportunity for substitution, mix-up, or contamination, and in an unregulated chain none of the links are documented. Chain of custody is not a paranoid concern; it is the ordinary basis on which analytical results are connected to physical objects, and without it the connection simply does not exist.

Handling and storage after testing. Covered in §34.7 and worth restating as a matter of physics: a peptide's condition is a function of its history. Temperature excursions, freeze-thaw, moisture ingress, and time all act on the molecule after the last measurement was made. No test result is updated by subsequent events.

Aseptic filling. This one is different in kind, and it is the deepest idea in the chapter.

A sterile injectable product is not made sterile by testing it. It is made sterile by a process — sterilizing or sterile-filtering the solution, and then filling it into sterilized containers within a controlled environment under conditions designed so that contamination cannot occur, with continuous environmental monitoring, qualified personnel, validated equipment, and simulated production runs that demonstrate the process itself reliably yields sterile units.

Why can't you just test the finished product instead? Because a sterility test destroys the units it tests, so you can only test a sample — and sampling statistics are brutal here. Imagine a batch in which a small fraction of units are contaminated. Testing a modest number of units has a low probability of encountering a contaminated one, and the test passes. The batch is not sterile. The test result is correct. The inference is wrong.

WHY YOU CANNOT TEST STERILITY INTO A PRODUCT

  A batch of many units, of which a SMALL FRACTION are contaminated.

    ████████████████████████████████████████████████░░
                                                    ↑
                                        contaminated units

  A sterility test destroys what it tests → you can only sample.
  Sample a modest number at random → you probably miss them entirely.

     TEST RESULT:  pass
     BATCH:        not sterile
     BOTH STATEMENTS ARE TRUE AT THE SAME TIME

  The end-product test has low power to detect low-level contamination.
  It is a CONFIRMATION that a controlled process worked — not a means of
  establishing sterility in a process that was not controlled.

  THE GENERAL FORM OF THIS ARGUMENT:
    Some properties are properties of a PROCESS, and can only be
    weakly sampled in a PRODUCT. Testing the product cannot create them.

This argument generalizes far beyond sterility, which is why it is the deepest idea here. Consistency is a process property. Traceability is a process property. Freedom from cross-contamination between products made on shared equipment is a process property. Each can be sampled, none can be manufactured by sampling, and each is invisible in a certificate that reports only results.

And everything that follows administration. The final gap is not analytical at all, and it is the largest. In the regulated setting, a medicine arrives embedded in a system: a prescriber who assessed whether it was appropriate, a baseline against which effects can be judged, monitoring for known adverse effects, an interaction check against everything else being taken, a pharmacovigilance pathway through which a bad outcome enters a database that changes future practice, a recall mechanism if a defect is found in a batch, and identifiable parties who bear responsibility. Outside it, none of those exist. No baseline. No monitoring plan. No adverse-event route. No recall. Nobody accountable.

The regulated supply chain is not primarily a testing regime. That framing is the error at the root of this entire section. Testing is one component, and not the largest. What a regulated supply chain principally consists of is documented process control, accountability, and recourse — a system in which how the material was made is written down and auditable, in which a named party is responsible for each batch, and in which there is a route by which harm gets detected, reported, and acted upon.

None of those can be established retrospectively by analyzing a sample. They are not properties that a measurement can reveal, because they are not properties of the material at all. They are properties of an institution.

So, plainly, and this is the sentence to carry out of the chapter:

Testing a sample cannot make an unregulated product equivalent to a regulated one, and nothing in this chapter should be read as suggesting otherwise.

If reading §34.2 through §34.7 has produced the feeling that you now know how to evaluate a purchase, that feeling is the predictable effect of learning a technical vocabulary and it is worth distrusting. Understanding what a chromatogram shows makes you better at reading chromatograms. It does not give you access to the batch records, the environmental monitoring data, the chain of custody, or the accountable party — because those were never in the chromatogram to begin with.

📊 Evidence Rating

Claim form — independent testing can make a gray-market product equivalent to a pharmaceutical one. Rating: ❌ Hype outpaces evidence (as of this writing) Reason: The claim rests on a category error. Pharmaceutical equivalence is constituted by documented process control, batch traceability, chain of custody, aseptic process validation, stability-established expiry, accountability for defects, and a pharmacovigilance and recall pathway. None of these are properties of a material that a measurement can detect, and none can be established after the fact by analyzing a submitted sample. End-product testing has low statistical power to detect low-level defects and is designed as confirmation of a controlled process rather than as a substitute for one. What would change it: Nothing that testing can supply. The gap closes only if the product is made under a quality system with the process controls, documentation, and accountability described — at which point it is no longer a gray-market product, which is the argument's conclusion rather than a workaround for it.

34.10 Pharmacopeial standards and GMP

Two words appear constantly in this territory and are frequently used as though they meant the same thing. They do not, and the difference is the cleanest possible summary of the chapter.

Pharmacopeias and monographs

A pharmacopeia is an official compendium of standards for medicines. The major ones — the United States Pharmacopeia (USP), the European Pharmacopoeia (Ph. Eur.), the Japanese Pharmacopoeia, and others — are legally recognized in their jurisdictions, and they publish monographs.

A monograph is a specification document for a particular substance. It states what the substance is, what tests must be performed to establish that a given material is that substance and meets requirements, what methods to use, and what the acceptance criteria are. A peptide monograph will typically address identity, assay (content), related substances, water, residual solvents, counterion where relevant, and — for injectable materials — sterility and endotoxin.

The monograph's function is to make a claim meaningful. When a material is described as meeting a pharmacopeial standard, that phrase points to a specific published document specifying specific tests with specific limits, rather than to a general impression of quality. It converts "high quality" from an adjective into a set of numbers someone can check.

Monographs also deliver harmonization. Two laboratories running "an HPLC purity method" may get different numbers because they chose different columns, gradients, and detection. A monograph specifies the method, making results comparable across laboratories and across time — and without comparability, quality data cannot accumulate into knowledge.

GMP

Good Manufacturing Practice is something categorically different. GMP is not a specification; it is a system of controls over how manufacturing is conducted. Its components include:

• Documented procedures for every operation, written before the operation and followed during it
• Batch records capturing what was actually done to each batch, by whom, with what materials, on what equipment — retained and auditable
• Personnel qualification and training records
• Facility and equipment qualification — demonstrating that equipment does what it is supposed to do, reproducibly
• Environmental monitoring, particularly for sterile operations
• Change control — a formal process ensuring that no change to a process, material, or supplier happens without assessment of its impact
• Deviation and investigation procedures — a required response when something does not go according to procedure, including determining what other batches might be affected
• Qualified oversight and a defined release decision by a responsible person
• Supplier qualification — extending control backward to raw materials

Notice that almost none of this is testing. GMP is about the process, its documentation, and the accountability structure around it.

The distinction that matters

MONOGRAPH  vs  GMP

  MONOGRAPH                              GMP
  ─────────────────────────              ────────────────────────────────────
  Defines the SPECIFICATION              Defines HOW MANUFACTURING IS DONE
  "what the material must meet"          "how you know the process reliably
                                          produces material that meets it"

  A destination                          A road, with maintenance records

  Verified by TESTING a sample           Verified by INSPECTING a system
                                          — records, procedures, facilities,
                                            training, investigations

  Can be checked after the fact          CANNOT be reconstructed after the fact

  ─────────────────────────────────────────────────────────────────────────
  Material can meet a monograph on the batch tested and still come from an
  uncontrolled process. The monograph tells you about that batch. GMP is what
  tells you about the next one, and the one after that.

A monograph defines the specification. GMP is how you know the process reliably produces material meeting it.

That single distinction contains the whole chapter. Sections 34.2 through 34.7 were about specifications and how to check them — the monograph half. Section 34.9 was about everything the monograph half cannot reach — the GMP half. A certificate of analysis lives entirely on the left side of that diagram. A pharmaceutical product exists because of the right side.

One clarification worth making, because the phrase gets used loosely in marketing. "GMP facility" and "GMP grade" are not self-certifying. GMP compliance is established by inspection by a regulatory authority, and a facility may be GMP-certified for one category of product and not another, or inspected for one market and not another. A supplier's assertion about its own compliance status is an assertion. This is the §34.7 problem again, one level up: a claim about a quality system faces exactly the same authenticity question as a claim about a sample, and is answered the same way — by audit and enforcement, not by document.

🔍 Check Your Understanding

1. Explain the difference between a monograph and GMP in one sentence each, without using the words "quality" or "standard."
2. A material meets every requirement of a pharmacopeial monograph on the batch tested. Name two things you still do not know about the manufacturer.
3. Why is it significant that GMP compliance is verified by inspecting a system rather than by testing a product? Connect your answer to the sterility argument in §34.9.

📋 Your Evidence Dossier

This chapter fills Field 10 — Preparation and Verification — and it is the field that does not resolve.

Every prior field converged on something. Field 1 established what a molecule is. Fields 4 through 7 weighed what the evidence supports. Field 10 is structurally different: it is an inventory of questions that cannot be answered, and its value lies precisely in seeing how long the list is and how few entries any document addresses.

A necessary framing, and it is not boilerplate. This is not a purchasing checklist. It is not a set of criteria that, once satisfied, indicate that a preparation is acceptable to use. There is no combination of entries below that produces that conclusion, because — per §34.9 — the properties that would produce it are not properties of the material. If completing this exercise leaves you feeling that you now know how to shop, re-read §34.9. The exercise is designed to demonstrate the opposite: that the list of unanswered questions is long, that it stays long no matter what documentation arrives, and that the standard is not met by better paperwork.

Step 1 — For each dossier compound, record what would have to be established

For each of your five to ten compounds, write down what would have to be true to know what a given preparation contains. Not what a seller says. What would have to be established.

FIELD 10 — PREPARATION AND VERIFICATION

  A. WHAT WOULD HAVE TO BE ESTABLISHED
     Identity          right molecule, right sequence, right stereochemistry
     Purity            what fraction of detected material is target;
                       by which method; with what specificity demonstrated
     Content           how many milligrams of PEPTIDE, by mass, in this container
     Water             how much of the mass is moisture
     Counterion        which counterion, how much of the mass
     Related species   which impurities, at what levels, against what limits
     Sterility         no viable organisms, established by a controlled process
     Endotoxin         below a defined limit, separately tested
     Container         closure integrity; material compatibility
     Stability         what expiry, established by what study
     Custody           this container traceable to that tested batch
     Handling          storage history from fill to hand

  B. WHICH OF THESE A TYPICAL CERTIFICATE ADDRESSES
     (mark each: addressed / partially addressed / not addressed / cannot be addressed)

  C. WHICH CANNOT BE ADDRESSED BY ANY TEST PERFORMED ON A SUBMITTED SAMPLE
     (this is the important column, and it is the longest one)

Worked demonstration — two entries, deliberately opposite

FIELD 10 — SEMAGLUTIDE, as a pharmacy-dispensed product   [worked demonstration]

  Identity        established — pharmacopeial/regulatory identity testing
  Purity          established — validated, specified methods
  Content         established — assayed against a reference standard; the label
                  refers to the active substance
  Water/counterion  controlled specifications
  Sterility       established by validated aseptic process, not by end-product
                  testing alone
  Endotoxin       tested against a defined limit
  Container       qualified closure system
  Stability       expiry established by a formal stability program
  Custody         traceable batch, licensed distribution, recall pathway exists
  Handling        cold-chain requirements defined and, in the licensed chain,
                  monitored

  UNANSWERED QUESTIONS: essentially none in this field. Note that this is NOT a
  statement that the drug is right for any given person — that is Fields 4–9 and
  a clinician. It is a statement that the CONTENTS question is closed.

FIELD 10 — BPC-157, as an unregulated preparation          [worked demonstration]

  Identity        UNKNOWN. A published sequence exists (Chapter 1, Field 1).
                  Whether this container holds it is a separate question,
                  answerable only about a sample, not about the container.
  Purity          UNKNOWN. Even a genuine chromatogram would be blind to
                  deletion sequences co-eluting with the target (§34.3).
  Content         UNKNOWN, and unaddressed by any purity figure (§34.4).
                  This is the number that determines exposure.
  Water/counterion  UNKNOWN. Unmeasured mass of unknown proportion.
  Sterility       UNKNOWN — and unestablishable, because sterility is a
                  property of a process (§34.9), not a test result.
  Endotoxin       UNKNOWN, and routinely not tested at all (§34.5). No visual
                  signature. Passing a sterility test is not informative here.
  Container       UNKNOWN. Closure integrity unaddressed.
  Stability       UNKNOWN. No stability program means no meaningful expiry;
                  a printed date without a study behind it is a printed date.
  Custody         NOT ESTABLISHED, and not establishable retrospectively.
  Handling        NOT ESTABLISHED. History from fill to hand is unrecorded.

  UNANSWERED QUESTIONS: ten of twelve, and the last four cannot be closed by
  any analysis of any sample. Note that this entry would look identical if a
  certificate of analysis had arrived with the container. That is the finding.

Step 2 — Mark the certificate column honestly

Where a certificate is available to you — including cosmetic ingredient documentation, which follows the same logic — mark what the document addresses. Then, separately, mark what it cannot address regardless of how thorough it is. The second column is the one carrying the lesson: a supplier could improve the first indefinitely and the second would not move at all.

Step 3 — Write one sentence

For each compound: what is the single most important thing about this preparation that no document could tell me?

Most people write some version of "whether this specific container matches the paperwork." That is the right answer, and it is worth writing yourself rather than reading here, because writing it makes it available later — when a well-designed document arrives and produces the feeling of reassurance that documents exist to produce.

Conclusion

Analytical chemistry can establish a great deal about a peptide. Mass spectrometry determines mass to a few parts per million and, in tandem mode, reads a sequence off a fragment ladder. Chromatography separates a target from many of its close relatives. Amino acid analysis quantifies composition and, through it, content. Sterility and endotoxin can be tested. Each is a real capability developed over decades, and understanding what each measures is part of scientific literacy about this field.

The chapter's central distinction is that these techniques answer three different questions — identity, purity, and content — that ordinary usage collapses into one. Identity asks what it is. Purity asks what fraction of the detected material is the target. Content asks how many milligrams are actually present. A vial can be 98% pure and contain considerably less peptide than its label states, because counterion, water, and salts are mass that a purity calculation never sees. That single distinction is the most practically useful thing here, and it is the one most consistently absent from the documents people rely on.

The second distinction is between sterility and endotoxin — between whether anything is alive in the vial and whether anything ever was — and the fact that sterilization addresses the first without touching the second.

And then the limits, which are the point. A certificate of analysis is a claim about a sample, not a property of a vial. It describes material submitted by an interested party, at a past moment, using methods that looked for some things and not others, and nothing within the document establishes that it corresponds to any particular container, or what happened to that container afterward, or that the document is what it appears to be.

Which is why the frame that matters is not "how do I get this tested." Sterility cannot be tested into a product; it is produced by a validated process and confirmed, weakly, by sampling. The same is true of consistency, traceability, and freedom from cross-contamination. These are properties of a process and of an institution, and they cannot be reconstructed after the fact by measuring a material. The regulated supply chain is not primarily a testing regime — it is documented process control, accountability, and recourse, and no quantity of analysis substitutes for it.

Chapter 35 turns to how peptide drugs are manufactured at scale, where these controls stop being abstractions and become specific operations on specific equipment. If this chapter has done its job, you will read that one knowing why the boring parts — records, qualification, change control — are the parts doing the work.

Key Terms

Mass spectrometry (MS) — an analytical technique that measures the mass-to-charge ratio of ions, used to establish the mass and hence constrain the identity of a molecule.

Mass-to-charge ratio (m/z) — the quantity a mass spectrometer actually measures: an ion's mass divided by its charge. Peptides commonly carry multiple charges, producing several peaks for one molecule.

Electrospray ionization (ESI) — a soft ionization method that converts molecules in solution into gas-phase ions without fragmenting them, making mass spectrometry of peptides possible.

Monoisotopic mass — a molecule's mass calculated using the most abundant isotope of each element. Distinct from average mass, which uses natural abundance-weighted isotopic averages.

Tandem mass spectrometry (MS/MS) — isolating an ion, fragmenting it, and measuring the fragment masses. Because peptides fragment at the peptide bond, the mass differences between fragments reveal the sequence.

HPLC — high-pressure liquid chromatography; separation of a mixture by differential partitioning between a stationary phase in a column and a mobile phase pumped through it.

Chromatogram — the output of a chromatographic run: detector response plotted against time, with each component ideally appearing as a distinct peak.

Retention time — when a component emerges from a chromatography column. A property of the compound and the method, not of the compound alone.

Co-elution — two or more species emerging together and appearing as a single peak. The fundamental limitation of any separation method.

Area percent purity — the target peak's area divided by the total area of all detected peaks. Blind to anything the detector does not respond to.

Response factor — the relationship between a compound's mass and the signal it produces. Because response factors differ between compounds, peak area is not proportional to mass across species.

Peptide content (assay, potency) — the mass of target peptide actually present in a preparation, determined against a reference standard or by amino acid analysis. Distinct from purity.

Counterion — an ion of opposite charge accompanying a charged molecule. Synthetic peptides commonly exist as trifluoroacetate salts, and the counterion contributes real mass that is not peptide.

Sterility — the absence of viable microorganisms. Assessed by attempting to culture organisms from a sample, and produced by a validated process rather than by testing.

Endotoxin — lipopolysaccharide from the outer membrane of Gram-negative bacteria. Heat-stable, passes sterilizing filters, and produces fever and inflammatory responses when injected. Tested separately from sterility.

Amino acid analysis (AAA) — complete hydrolysis of a peptide followed by quantification of the liberated amino acids. Establishes composition and, because it is quantitative, serves as a primary method for peptide content.

Edman degradation — sequential chemical removal and identification of the N-terminal residue, reading a sequence one residue at a time. Requires a free N-terminus.

Certificate of analysis (CoA) — a document reporting test results for a material. A claim about a sample, not a property of a vial.

Specification — a predetermined acceptance criterion a material must meet, established before testing. Results reported without specifications cannot pass or fail.

Chain of custody — documented control of a material from origin to point of use. What connects an analytical result to a physical object; absent, the connection does not exist.

Orthogonal methods — analytical methods whose separation or detection principles are independent, so that a species hidden from one is unlikely to be hidden from the other.

Monograph — a pharmacopeial document specifying what a substance must be and what it must meet, including tests, methods, and acceptance criteria.

Pharmacopeia — an officially recognized compendium of drug standards, such as the United States Pharmacopeia or the European Pharmacopoeia.

GMP (Good Manufacturing Practice) — a system of documented process control, personnel and facility qualification, environmental monitoring, batch records, and change control. Verified by inspection of a system, not by testing a product.

Aseptic processing — manufacturing a sterile product by filling sterilized material into sterilized containers under controlled conditions. A process guarantee that end-product testing cannot replace.

Spaced Review

1. A document reports "HPLC purity: 99.1%" for a lyophilized peptide and nothing else. Using §34.4, state the two separate reasons this figure does not establish how much peptide is in the vial. Then state, using §34.3, one reason it may also overstate purity itself.

2. Chapter 19 introduced the formulation that sterility asks whether anything is alive in the vial and endotoxin asks whether anything ever was. Explain, in terms of the physical properties of lipopolysaccharide, why a preparation can pass a sterility test and still cause fever on injection. Why does this hazard have no visual signature?

3. Chapter 32 identified deletion sequences as the characteristic impurity of solid-phase peptide synthesis. Explain why they are also the impurity class that area-percent HPLC purity is least able to detect, and what that implies about the relationship between a synthesis method's typical failure and the assay used to police it.

4. Return to Chapter 6's four vials. A person obtains Vial C and has a sample analyzed by a competent laboratory, which reports correct identity by MS/MS, 98.7% purity by two orthogonal methods, and content within 5% of label. List four things that remain unknown, and identify which of them could in principle be resolved by further testing and which could not.

5. A supplier states that its material "meets USP monograph requirements." Explain what that claim does establish, what it does not establish, and why the distinction between a monograph and GMP is the reason a person cannot infer the second from the first.