Why two laboratories get different numbers on the same lot

An HPLC purity figure is a method-dependent number. Where the differences between two reports on one lot come from, what measurement uncertainty is, and how to tell an expected gap from a real discrepancy.

By Touchstone PeptidesSeptember 22, 20269 minute read

Two competent laboratories measuring the same lot of the same peptide will usually report purity figures that differ by a small amount, because a reversed-phase HPLC purity is not a property of the molecule alone. It is the fraction of detector signal that fell under one peak, under one set of conditions, integrated one way. ICH Q2(R2) calls the lab-to-lab spread reproducibility, and a gap inside it is expected. A different observed mass, or a large peak on one trace with no match on the other, is not a method difference.

What an HPLC purity number actually is

Bachem's technical FAQ gives the working definition: "The purity determined by (analytical) HPLC (%) using UV detection corresponds to the percentage of requested peptide in relation to the total amount of material absorbing at 220 nm."

It is an area percent. The instrument integrates the area of every peak, and purity is the main peak's share of the total. Nothing is weighed and nothing is compared to a standard; the number is relative to whatever else the detector saw.

It assumes equal response. Area percent assumes an impurity absorbs light exactly as the main peptide does, mass for mass. At 214 to 220 nm the detector is reading the peptide bond, so peptide-like impurities (deletion sequences, oxidised or deamidated variants) respond similarly and the assumption is reasonable. Anything without peptide bonds, and anything with an extra chromophore, responds differently.

It is blind to what does not elute or does not absorb. Salt, water and counter-ion do not appear. That is why purity and net peptide content are different numbers, and why a 99% purity can sit on a vial that is one fifth trifluoroacetate by mass.

Reading a chromatogram covers how to read the trace itself.

Seven places the number can move

Column. Two C18 columns from different makers, or the same column at different ages, separate a peptide from its closest impurities differently. An impurity that is resolved on one column and co-elutes with the main peak on another is counted as impurity by the first lab and as product by the second. The second lab's purity is higher and neither lab has made an error.

Gradient and run time. A faster gradient compresses peaks. Two impurities that stand apart in a forty-minute run can merge into a shoulder in a fifteen-minute one, and a shoulder may or may not be split off by the integrator.

Wavelength. Purity at 214 nm and purity at 220 nm are different measurements. The lower wavelength sees more of everything, including baseline drift from the acid in the mobile phase; the higher one is quieter but less sensitive to small peptide impurities. Some laboratories also report at 280 nm, which sees only tryptophan, tyrosine and cysteine-containing species and gives a purity that is not comparable to either.

Integration. Where the baseline is drawn, the minimum peak height counted, whether the main peak's tail is split from a trailing impurity: these are analyst or software settings. Q2(R2) notes that for impurity tests the basis must be stated, "e.g., weight/weight or area percent with respect to the major analyte", which is a reminder that the basis is a choice.

Reference standard. A content assay, as opposed to an area-percent purity, compares the sample to a standard of known content. FDA's guidance lists pharmacopoeial and NIST sources for standards, but for most research peptides no compendial standard exists, so each laboratory qualifies its own. Two labs with two in-house standards will produce two content figures that differ by the difference between the standards.

Sample preparation. The solvent the peptide is dissolved in, the concentration, how long it stands before it is loaded onto the column and at what temperature, all change what the detector sees. Peptides with methionine, cysteine or tryptophan can oxidise in solution; peptides with asparagine can deamidate. A sample prepared and run the same hour is a different sample from one that sat overnight.

Moisture and weighing. For any figure that depends on a weighed mass (content, net peptide content by nitrogen or amino-acid analysis) the water in the solid matters. A lyophilized peptide takes up water from air while it is being weighed, and hygroscopic sequences do so quickly. Two laboratories weighing the same lot on days of different humidity are weighing different amounts of water.

Purity versus content: which number is more likely to differ

Area-percent purity is a ratio within one run. Most of the seven factors above shift it by a small amount, because the main peak dominates the total whichever way the small peaks are counted.

Content figures are absolute. They depend on a balance, a standard, a water figure and a counter-ion figure, and each adds its own uncertainty. Bachem's FAQ describes two routes to net peptide content, nitrogen determination and amino-acid analysis, and each has its own assumptions: nitrogen counts every nitrogen-containing species as peptide; amino-acid analysis depends on complete hydrolysis and on the yield of each residue after hydrolysis. Sikora and colleagues note in their review of counter-ions that trifluoroacetate "may occur in two forms, either directly bonded with basic residues of peptides or adsorbed in lyophilizate as a contaminant", and that ion chromatography "is a technique of the first choice for counter-ion determination". A laboratory that measures the counter-ion and one that assumes it will reach different net contents on identical material.

When two reports on one lot disagree, expect the gap to be larger on content than on purity, and larger still if the two labs used different routes to content.

What measurement uncertainty is

The international Guide to the Expression of Uncertainty in Measurement (the GUM) defines the term:

uncertainty (of measurement): parameter, associated with the result of a measurement, that characterizes the dispersion of the values that could reasonably be attributed to the measurandJCGM 100:2008, definition 2.2.3

Every measured figure has one. A purity printed as 98.7% is shorthand for an interval, and the interval's width comes from the factors above.

ICH Q2(R2) breaks precision into three levels, which map onto the two-report problem directly:

Term (ICH Q2)	Definition as given	What it means for two reports
Repeatability	"the precision under the same operating conditions over a short interval of time"	The spread one lab sees running the same sample twice today
Intermediate precision	"intralaboratory variations", for example "different days, different environmental conditions, different analysts, and different equipment"	The spread one lab sees across weeks
Reproducibility	"the precision between laboratories (e.g., interlaboratory studies, usually applied to standardization of methodology)"	The spread between the two reports on your desk

Reproducibility is always the widest of the three. Two laboratories differing by less than the method's reproducibility have not disagreed; they have reported the same value twice with the expected scatter. A laboratory that states its uncertainty on the report has told you how wide that scatter is. Most research-market reports do not, which is a gap to ask about rather than a reason to distrust the figure.

FDA's guidance on analytical procedures adds a practical piece: every run should pass a system suitability test, "to ensure that the system (equipment, electronics, and analytical operations and controls to be analyzed) will function correctly as an integrated system at the time of use", with criteria such as peak tailing, precision and resolution. Two laboratories with different system-suitability criteria are holding themselves to different standards before the sample is even loaded onto the column.

What a reasonable disagreement looks like, and what a real discrepancy looks like

The existing piece on why we pay for a second laboratory says that when the two reports here differ it is usually by a few tenths of a percent, and that a real divergence stops the lot. In general:

Expected, and not a finding:

Purity figures that differ by a fraction of a percent, especially where the two methods used different columns, gradients or wavelengths.
Content or net peptide content figures that differ by a few percent where the two labs used different routes (nitrogen versus amino-acid analysis, measured versus assumed counter-ion, different water figures).
Slightly different retention times, which reflect the method rather than the molecule.

A finding, and not a method difference:

A different observed mass. Mass spectrometry answers whether the molecule is the one ordered, and a monoisotopic or average mass that does not match the calculated value within the instrument's resolution is not explained by any column or gradient. Two laboratories should agree on identity even when they disagree on purity. That question is expanded in what a mass spectrometry match proves.
A purity gap large enough that one report would pass a specification and the other would fail it by a wide margin, where the two chromatograms cannot be reconciled by integration alone.
An extra peak of meaningful size on one trace with no counterpart on the other, which suggests different material, degradation between the two tests, or a sample-preparation artefact that needs explaining.
A content figure far below the label on one report and near it on the other, which points to a weighing, water or sample-identity problem rather than to the peptide.

The honest reading of a discrepancy is not "one lab is wrong". It is "these two measurements need a third piece of information", which is usually the method details or a re-test on a fresh sample from the same lot.

How to read two reports side by side

Field	Compare	If they differ
Lot number	Must be identical	You are not reading two reports on one lot
Sample description and date	Same material, dates in order	A retained release sample and a shipped vial are different samples
Identity (observed mass)	Both within instrument tolerance of the calculated mass	Stop; this is not a method difference
Purity method	Column, wavelength, gradient, integration basis	Expect small gaps between different methods
Purity figure	Within a fraction of a percent	Larger gaps need the chromatograms, not the numbers
Content route	Nitrogen, amino-acid analysis or assay against a named standard	Different routes give different figures on the same solid
Water (Karl Fischer)	Present on both	A missing water figure explains part of any content gap
Counter-ion	Measured or assumed, and which ion	An assumed counter-ion is a source of content error
Stated uncertainty	Present on either	If absent, ask for it

Read the methods before the numbers. Two purity figures from different wavelengths are not the same measurement and should not be averaged.

Every lot here, from BPC-157 to semaglutide on the GLP-1 and metabolic shelf, carries the manufacturer's certificate and an independent second assay under the same lot number, so anyone can make this comparison.

Sources

For laboratory research use only. Not a drug, not a supplement, and nothing here is a claim about what any of this material does in a person or an animal.

Two things before you come in.