Measuring drug-to-antibody ratio
DAR is an average over a distribution, and the distribution matters more than the average. A DAR 4 that is a clean single species and a DAR 4 that is half unconjugated and half DAR 8 behave nothing alike in an animal. Any method reporting only the mean is discarding the information you need.
Hydrophobic interaction chromatography
HIC is the reference method for cysteine-linked conjugates. Payload adds hydrophobicity, so species elute in order of load, and a well developed method resolves DAR 0, 2, 4, 6 and 8 as baseline-separated peaks under non-denaturing conditions.
Weighted average DAR comes straight from the peak areas. The distribution comes free with it, which is the reason to run this first.
Two cautions. Peak area assumes equal response, and payload chromophores that absorb at the detection wavelength break that assumption, so quantify at 280 with a correction or use a wavelength where only protein absorbs. Method development is real work, because the salt gradient and the stationary phase both need tuning per conjugate, and a method transferred from a different payload will not resolve.
Reversed-phase after reduction
Reducing the conjugate separates light and heavy chains, and RP-HPLC then resolves L0, L1, H0, H1, H2 and H3. Summing the loaded fractions gives DAR and also tells you where the payload sits, which HIC cannot.
This is the method for positional information on a cysteine conjugate. It is denaturing, so it says nothing about the intact assembly.
Native mass spectrometry
Native MS under non-denaturing conditions gives the intact conjugate mass and resolves each DAR species directly, with the payload mass as confirmation of identity rather than inference from retention.
It is the best method for site-specific and lysine conjugates, where HIC often fails to resolve because the hydrophobicity difference between species is small or because the positional isomers smear the peaks. For lysine conjugates the distribution is broad and roughly binomial across dozens of possible sites, and native MS is close to the only way to see it cleanly.
Desalting and careful ammonium acetate buffer exchange are prerequisites, and response factors still vary between species, so treat quantitation as semi-quantitative unless it has been calibrated.
UV spectroscopy
If the payload has a distinct absorbance maximum, deconvoluting the spectrum at two wavelengths gives average DAR in minutes with a spectrophotometer. It is fast, it is cheap, and it gives no distribution at all.
Useful as an in-process check during conjugation, where you want to know whether the reaction is progressing. Not sufficient for release.
Free payload, which is a separate question
Unconjugated payload and payload-linker in the drug product is a safety attribute, not a DAR attribute, and it needs its own method. Reversed-phase LC-MS on the filtrate after protein precipitation or solid-phase extraction is the usual approach, with a limit set in the low percent or below.
Measure it after the final formulation step and again on stability, because free payload generated by linker degradation during storage is one of the things a stability program exists to catch.
What to run
Two orthogonal methods, always. HIC plus native MS for cysteine conjugates, native MS plus reduced RP for site-specific. When two methods disagree by more than a few tenths of a DAR unit, one of them has a response factor problem, and finding out which before anyone doses anything is time well spent.
Report the distribution alongside the average. A certificate quoting DAR 3.8 with no distribution has told you almost nothing about what is in the vial.