What does an 18 Da, 1 Da or 128 Da mass shift next to the main peak actually tell me?

The deconvolved mass spectrum on a tirzepatide report shows the expected 4813.5 peak plus three satellites I do not understand: one at 4795.5, one at 4814.5 that partly overlaps the main peak, and a much smaller one at 4685.4.

The report calls all three "process-related impurities" and leaves it there, which is not an answer. I can subtract: those are shifts of -18.0, +1.0 and -128.1 relative to the parent. What I want is the interpretation. Is a -18 the same thing whether it appears on a peptide with a Ser in it versus one without? Is +1 always deamidation, and if so does it matter chemically or is it cosmetic? And -128 looks like an entire residue has gone missing, which sounds like a synthesis failure rather than a degradation product.

I am trying to work out which of these I should care about in the sense of "this batch is not what it says it is" versus which are the normal background of solid-phase synthesis that every batch on earth carries at some level.

Each of your three shifts has a short list of chemically plausible causes, and in two of the three cases the list is short enough to name the culprit with reasonable confidence. Here is the working table I use.

Your -18.0 at 4795.5

Tirzepatide contains two aspartic acids (positions 9 and 15). Aspartimide formation is the single most predictable impurity in Fmoc solid-phase synthesis of Asp-containing sequences: the repeated piperidine deprotection steps promote cyclisation of the Asp side chain onto the backbone nitrogen, expelling water. The result is exactly -18.01 Da and it is often the dominant related substance in a crude peptide. It is also the gateway to worse: the succinimide ring can reopen to give iso-aspartate (a beta-linked backbone, same mass as the parent, invisible to MS) or racemise to D-Asp. So a visible -18 peak is a warning that mass-silent isomers are probably present too.

You cannot distinguish aspartimide from a Ser/Thr dehydration by mass alone. You distinguish them by where they sit in a peptide map, which means a digest, not a scan.

Your +1.0 at 4814.5

Check the resolution first. If this was a single quadrupole or a low-resolution TOF, the "+1" is very likely the isotope shoulder of the parent, not a separate species. If the instrument genuinely resolved it, +0.98 is deamidation. Tirzepatide has three Gln (19, 24 and the Gln at 19 in particular) and no Asn; Gln deamidates far more slowly than Asn, so a substantial deamidated fraction in a fresh batch is odd and would make me ask about storage history. Semaglutide has one Gln at position 23 and the same reasoning applies.

Chemically deamidation is not cosmetic — it introduces a negative charge where a neutral amide was — but whether it costs potency depends entirely on position. Ask for the digest if the abundance is over a percent or two.

Your -128.1 at 4685.4

This is a deletion, and it is the one that says something about the manufacturer rather than about the vial. Tirzepatide has two Lys (16 and 20) and three Gln. At high resolution you can tell which: Gln deletion is -128.0586, Lys deletion is -128.0949, and the 0.036 Da gap is resolvable at 4685 Da on any instrument doing better than 10 ppm. If the report only gives you 4685.4, it cannot distinguish them and should say so.

A deletion impurity means a coupling step did not go to completion and was not capped or purified out. Every synthesis produces some. The question is how much, and that is a purity question, not an identity question — which is why your instinct that this one is different from the other two is correct.

What to actually do

Ask for the relative abundance of each satellite in the deconvolved spectrum, and separately ask for the HPLC area percentages, and then notice that the two sets of numbers will not agree. That disagreement is the subject of a different question, but the short version is that ESI response factors differ between the parent and its impurities, so MS abundance is not a purity measurement.

Two shifts worth adding that are specific to the acylated GLP-1 analogues and therefore not in a generic peptide mass-shift table.

Loss of the whole acyl side chain. Both semaglutide and tirzepatide carry a Lys side chain modified with AEEA-AEEA-gamma-Glu-diacid. Semaglutide uses a C18 diacid, tirzepatide a C20. If the acylation step failed entirely you get des-acyl peptide, which is a very large negative shift (several hundred Da) and is trivially obvious. The dangerous version is a partial side-chain failure: one AEEA spacer missing is -145.07 Da, missing gamma-Glu is -129.04 Da. Those are in the same size range as residue deletions and are easy to misattribute.

Why it matters more than a backbone deletion: the diacid side chain is what binds albumin and gives these compounds the multi-day half-life the once-weekly schedules in the published programmes depend on [1]. A truncated spacer alters that binding. So a -145 satellite is arguably a worse finding than a -128 backbone deletion even at the same abundance, because it produces a molecule with plausible receptor affinity and a different duration of action.

Diacid regiochemistry. A C18 diacid attached through the wrong carboxylate, or a mono-acid contaminant in the diacid building block, gives -16 or +14 shifts that look like ordinary oxidation or methylation. Only a digest that isolates the modified Lys-containing fragment will tell them apart.

Both of these are invisible on an intact-mass scan at the abundance levels that matter. If you want them found, the test to order is LC-MS/MS peptide mapping, not a better scan.

Adding the practical filter I use before I take any satellite peak seriously: run the same reasoning on the blank.

A lot of small satellites in real spectra come from the analysis, not the sample. Sodium (+21.98 per charge relative to the protonated ion), potassium (+37.96), TFA (+114.03), and formate (+45.0) adducts are all mobile-phase and glassware artefacts. Water clusters and in-source dehydration produce apparent -18 peaks on compounds that have no chemical route to dehydration at all. Ion-source fragmentation of a labile bond can manufacture an impurity that does not exist in the vial.

The tests that separate sample from artefact:

• Does it co-elute? A genuine chemical impurity has its own retention time. An in-source artefact appears at exactly the parent's retention time, because it is made in the source from the parent. If a report shows you an extracted ion chromatogram, this is a five-second check.
• Does it scale with cone/declustering voltage? Artefacts do; real impurities do not. Labs will not usually have run this, but you can ask.
• Is it in the blank? Adducts and carryover are.

An intact-mass report with no chromatography behind it cannot answer the first question, which is the most useful one. That is why "LC-MS" and "MS" are not interchangeable terms on a COA even though they get used that way.