Does aspartimide formation of liraglutide at minus 20 °C show up as a loss of content or of purity?

The case in front of me: aspartimide formation · liraglutide · minus 20 °C.

This is one of those things that everyone repeats and nobody derives.

This matters practically, not just academically, because it changes what I would do next.

Why does this happen, and what would falsify the usual explanation?

The relevant detail is that the lyophilised solid is far more robust than anything anyone says about it, and the solution is far less robust. Most of the confusion in this area comes from advice about one being applied to the other.

Freeze-concentration is the mechanism people miss. As ice forms, everything that is not water is excluded into a shrinking unfrozen fraction, so the local concentration of peptide, buffer salts and preservative rises sharply. If the buffer components crystallise at different rates, local pH can shift by more than a unit. That is why a phosphate-buffered solution can behave badly on freezing while an unbuffered one is fine.

Adsorption to the container is a real loss at low concentration. For a peptide at 0.1 mg/mL in an untreated glass vial, single-digit percentage losses to the wall are plausible; at 5 mg/mL it is negligible. This is one of several reasons not to reconstitute to a very dilute working solution and store it.

One qualification: none of this addresses sterility. A vial can be chemically pristine and microbiologically compromised, and a chromatogram will not tell you which.

Minimise transitions rather than minimising temperature. One freeze and one thaw is fine; five is a different question.

Concretely, freeze-thaw damage happens at the moving ice front, not at the storage temperature. Once the sample is frozen solid and cold, very little is happening. The damage is done during freezing and thawing, which is why the number of cycles matters and the duration of the hold mostly does not.

Light matters for specific residues rather than in general. Tryptophan and to a lesser extent tyrosine and methionine are photo-labile; a sequence without them is largely indifferent to ambient light over the timescales in question. Amber glass is cheap insurance rather than a requirement.

On the detail: practical thermal arithmetic for a shipment: a single 250 g phase-change pack in a thin-walled polystyrene box holds sub-ten-degrees for roughly 24 to 48 hours in a 25 °C ambient, and considerably less at 35 °C. Any lane taking eight to fourteen days is therefore not temperature-controlled for most of its duration regardless of what was in the box, which is the argument for shipping the material lyophilised.

Worth stating: research-use-only material has no stability programme behind it at all, so any beyond-use date you apply is your own construct.

Store solid, store cold, store dry, and reconstitute what you will use rather than what fits in the vial.

The relevant detail is that degradation is not one process, and which one dominates depends on the condition you are asking about. In solution at refrigerated temperature the rate-limiting pathway is usually deamidation and hydrolysis; at room temperature aggregation overtakes them; frozen, the damage happens during the transitions rather than during the hold.

For the solid state, residual moisture is the dominant variable. A cake at two per cent water is considerably more stable than the same cake at six per cent, because water is both a reactant in hydrolysis and a plasticiser that lowers the glass transition temperature. This is why a desiccant in the outer packaging is not theatre, and why opening a cold vial in a humid room is a genuine error — you condense water onto the cake.

Concretely, on re-freezing something that thawed in transit: if it arrived as a lyophilised solid that warmed but never got wet, re-freezing costs you nothing except the thermal cycle. If it arrived as a solution that thawed, re-freezing adds a second transition and therefore a second dose of ice-front shear. The asymmetry is worth internalising.

The Arrhenius relationship underpinning accelerated stability testing is the basis of ICH Q1A, which is why accelerated studies at 40 °C and 75 per cent relative humidity are used to predict shelf life at 25 °C. The same relationship lets you reason about a warm transit lane, with the same caveats about extrapolation.

The practical rule is that time and temperature multiply, so shorten whichever one you control.

edited 28 Apr 2025 by tri_gly_ala — expanded the table to cover the lower concentration

Start by separating chemical degradation from physical degradation, because they fail differently and they are detected differently. Chemical degradation changes the molecule and shows up as new peaks on a chromatogram. Physical degradation aggregates the molecule and often shows up as nothing at all on reverse-phase HPLC, because the aggregate never makes it onto the column.

The 28-day figure for a reconstituted preserved vial is microbiological, not chemical. Chemically, a well-behaved peptide at 5 mg/mL at 4 °C will typically lose well under a per cent of content per month. The reason to respect the date is bioburden, and bioburden is a function of how many times you have opened it, not of the calendar.

Deamidation kinetics for asparagine in peptides are well characterised and strongly sequence-dependent: the residue following the asparagine dominates the rate, with glycine and serine at the n+1 position accelerating it by an order of magnitude relative to bulkier residues. That is why two peptides in the same buffer at the same temperature can have quite different shelf lives.

I would be careful about generalising across sequences. Stability is sequence-specific, and a rule derived from semaglutide will not transfer cleanly to a tri-agonist with different residues in different local environments.

If the material arrived warm and it was lyophilised, test it and proceed on the result. If it arrived warm and it was in solution, the result is more likely to be interesting than reassuring.

The honest answer is that published stability data for these specific molecules in a research-grade presentation essentially does not exist, so what you get is extrapolation from the licensed formulations and from general peptide chemistry. That extrapolation is reasonable. It is still extrapolation.

The temperature dependence is roughly Arrhenius over the range that matters, which in practice means every ten degrees of increase roughly doubles to triples the rate. Ten days at thirty degrees is therefore comparable to something on the order of a month or two at four degrees — bad, but not the catastrophe it feels like when you open a warm parcel.

The licensed semaglutide and tirzepatide presentations carry in-use periods of several weeks at room temperature in their labelling, which is the closest thing to real stability data in this space — and it applies to a buffered, surfactant-containing, preservative-containing formulation, not to a reconstituted research vial.

The limitation is that you cannot detect slow aggregation by eye until it is well advanced, so a clear vial is weak evidence of an intact one.

The single highest-value change most people can make is buying a cheap logging thermometer, because it converts an assumption about their storage into a record.