One Peak, Two Compounds, One Purity Figure That Is Wrong

Hydrophobicity is a coarse sorting rule

A reversed-phase column ranks molecules by how strongly they prefer the stationary phase to the mobile one. Nothing finer than that is being asked of it. Two species with similar hydrophobicity therefore descend the column at similar speed, and once the difference between their travel times shrinks below the width of a peak, they reach the detector as one event.

Now consider which species have similar hydrophobicity to the target. They are its nearest structural relatives, which is to say precisely the population the assay was commissioned to find. A chain short by one residue. A deamidated variant. A diastereomer thrown off by racemization somewhere in the synthesis. Any of these may land a hair from the parent in retention time. Deletion, truncation and oxidation impurities explains what each of them is.

The error has a preferred direction

Suppose the vial is honestly 95% target and carries an impurity accounting for 3% of the area that leaves the column at the same instant as the parent. The report says 98%. Nothing on the trace announces the substitution; there is one peak standing where two should be, and integration software has no means of suspecting it.

That outcome is structural, not accidental. Purity off a chromatogram is main peak area over total area, a calculation resting on a premise it has no way to test, namely that one peak corresponds to one compound. When the premise fails it always fails the same way. Area concealed under the parent is area awarded to the parent, and the number only ever moves up.

Which suggests something about what is worth requesting. A percentage cannot be interrogated; no one has ever inspected a number for a shoulder. A trace can be interrogated. So a purity claim offered without the chromatogram behind it is a weaker claim than its two significant figures imply.

Detection, in ascending order of certainty

• Peak shape. Cheapest of all, and the easiest to scroll straight past. Overlapping species often betray themselves through a shoulder, a base that has spread wider than it should, or an asymmetry the method offers no other account of.
• Diode-array peak purity. A photodiode array detector collects a whole spectrum over and over as the band passes it, instead of absorbance at a single wavelength. Should the spectrum recorded at the leading edge fail to match the one recorded at the trailing edge, more than one species is present. This is the conventional check and it is conclusive when the two components carry different chromophores.
• Orthogonal separation. Rerun the sample by a method that sorts on a different principle: another column chemistry, another pH, ion exchange rather than reversed phase. Two things that migrate together under one mechanism rarely migrate together under a second.
• Mass spectrometry through the peak. An LC-MS run gives the mass of whatever is eluting at each moment. Two masses under one peak permit exactly one interpretation.

The second item carries a blind spot worth naming. Where two species produce identical ultraviolet spectra, diode-array purity cannot tell them apart, and a diastereomer or an isoaspartate rearrangement absorbs exactly as the parent absorbs. Those cases leave only orthogonal separation or mass spectrometry.

Questions for the laboratory, and the limits of the answer

Request the chromatogram rather than the percentage. Ask what detection wavelength was used. Ask whether peak purity was assessed at all. Anyone running a validated method will already have fought this during method development, because keeping overlap under control is part of the point of resolution criteria and system suitability.

Two analyses of one vial that return different numbers have several ordinary explanations, and overlap occurring under one method but not the other is one of them. The others are collected in why certificates of analysis disagree.

Detecting the overlap, finally, establishes one thing only: a peak is shared. It does not identify the other occupant, quantify it, or say at what stage it appeared. And in common with every identity and purity measurement, it speaks to composition and to nothing beyond composition, which leaves out sterility, endotoxin and suitability for any use in a person or an animal. Endotoxin, sterility and the limits of research grade marks out where that line runs.

All of the above concerns laboratory research material and the published literature. Nothing supplied is a drug or a supplement, and none of it is intended for administration to people or to animals.