For most of the last decade, protein quantification in biological matrices has been the territory of the ligand binding assay. That has not changed, but the boundaries have. A new perspective from the European Bioanalysis Forum (EBF) Hybrid MS team, published in Bioanalysis, gathers seven case studies from pharmaceutical companies and CROs showing where liquid chromatography-mass spectrometry answers questions an immunoassay struggles to reach.
The framing matters. The authors are explicit that the goal is not to position LC-MS as a replacement for ligand binding assays, but to demonstrate its role as a complementary tool. What the case studies map out is the when: where molecular-level selectivity, structural insight or multiplexing capacity justify the extra method development effort.
Why hybrid LC-MS earns its place
The core argument rests on two selectivities working together. An immunoaffinity step gives selective extraction and enrichment of the analyte, often reaching sensitivity that approaches a ligand binding assay. LC-MS detection then adds molecular-level selectivity capable of distinguishing closely related protein forms.
That pairing enables protein quantification in exactly the situations that trouble an immunoassay: isoform-specific measurement, post-translational modifications, and endogenous proteins that differ only slightly. The paper adds a pragmatic driver: LC-MS is increasingly used where commercial reagents do not exist, or where cross-reactivity undermines an antibody-based method.
Telling near-identical proteins apart
The first case study is close to a worst-case selectivity problem. A therapeutic mutein protein differs from its circulating endogenous counterpart by four amino acids. The endogenous form sits at roughly 400 times the drug’s anticipated peak plasma concentration, and about 100,000 times the requested lower limit of quantification. Raising a capture reagent with that discrimination is, as the authors put it, very challenging.
The team worked the problem at the peptide level instead. An in-silico assessment simulated cleavage by four proteases. Trypsin was dropped early: the specific tryptic peptide had a ragged C-terminal end with adjacent cleavage sites, inviting missed cleavages and variable response. Peptide mapping then eliminated Glu-C and Lys-N, which produced only larger, missed-cleaved products. The Glu-C peptide also carried two methionines that oxidised rapidly even with antioxidants. Lys-C delivered a fully cleaved surrogate peptide with better characteristics.
Notably, the capture reagent was not specific to the drug; it pulled down both isoforms. That was still fit for purpose, because the mass spectrometer supplied the selectivity, though it cost sensitivity: peptides from the abundant endogenous isoform raised chemical background on a triple quadrupole, capping the LLOQ at 100 ng/mL. A Q-TOF high-resolution instrument cut that background and delivered a five-fold improvement, to 20 ng/mL. The validated method then quantified the therapeutic protein in clinical trial plasma.
The second case study applies the same logic to vaccine work. Monitoring IgG1, IgG2, IgG3 and IgG4 responses separately required an LLOQ of 100 ng/mL per subclass against endogenous immunoglobulin at mg/mL, a background that can exceed the vaccine-induced signal by more than 70,000-fold.
The workflow used a biotinylated recombinant antigen on streptavidin magnetic beads, blocking conditions screened until a SuperBlock/BSA/Tween-20 combination suppressed non-specific IgG binding, thermostable trypsin digestion at 70 °C, and stable isotope-labelled winged internal standards per subclass.
The profiles showed IgG1 and IgG3 dominating early, IgG2 and IgG4 emerging later, distinctions the authors note are harder to resolve by immunoassay given cross-reactivity and limited true multiplexing.
Reading a modification the immunoassay cannot see
The tau case study is the clearest example of LC-MS doing something structural rather than simply counting molecules. Phosphorylation at serine-396 has been linked to aggregate formation in Alzheimer’s disease, but tau exists as six proteoforms with many phosphorylation combinations, and finding an immunological reagent that cleanly separates phosphorylated from non-phosphorylated S396 across all of them is a challenge.
The method turns trypsin’s own behaviour into the discriminator. Trypsin cleaves after the lysine preceding S396 only when S396 is unphosphorylated, so digestion of cerebrospinal fluid yields a short signature peptide from non-phosphorylated tau and a long peptide from the phosphorylated form.
Both were then enzymatically dephosphorylated; the short peptide was quantified first, the long peptide re-digested, and the increase in short-peptide concentration read as the original phosphorylated fraction. Residual trypsin activity during dephosphorylation had to be blocked with excess inhibitor.
Against a fully ¹⁵N-labelled tau internal standard, the calibration range ran from 2 to 100 pM in 200 µL of CSF, with precision and accuracy inside 20% and stability to 437 days at −70 °C. Samples from healthy volunteers and Alzheimer’s patients gave tau concentrations of 5–25 pM with up to 35% phosphorylation. What made LC-MS essential here was that the method actively modifies the protein, by digestion and dephosphorylation, to create the measurement.
When the critical reagents are not available
Two case studies deal with reagent scarcity rather than selectivity. After the proprietary ELISA for Pegasys, a PEGylated interferon alfa-2a, was discontinued along with its critical reagents, a hybrid LC-MS/MS assay for free drug in non-human primate serum was built on a commercially available polyclonal anti-interferon antibody, with a range of 10–1000 ng/mL.
The data then raised a second question. Free drug exceeded 300 ng/mL early but fell below the quantification limit from Day 29 onward, suggesting progressive binding to serum components.
A total-drug assay was developed without immunoaffinity capture, using deoxycholate and heat pretreatment before digestion. A standard addition test on pooled incurred samples confirmed it: recovery of spiked drug in the free assay fell from 96% at Day 8 to around 51% at Day 29 and roughly 3% at Day 91.
The transgenic mouse case study hits the same constraint from another direction. Tg32 mice do not produce human IgG, so pooled human immunoglobulin is dosed before the test compound, leaving no reagent able to separate Fc-containing candidates from that background, at a stage where raising one was not affordable and with only 10 µL of plasma.
A multiplexed LC-MS/MS method using protein precipitation, pellet digestion and MRM on carefully chosen surrogate peptides quantified both in a single run.
Integrity and stability, not just concentration
Multispecific and fusion formats bring the risk of in vivo truncation, and the fifth case study shows LC-MS quantifying a molecule and assessing its integrity in one assay.
Signature peptides were positioned across the Fab, Fc and fused-protein regions of a mAb fusion protein, and three capture strategies compared. With generic anti-Fc and anti-Fab capture, mAb peptides tracked together while fused-protein peptides sat lower with shorter half-lives, a divergence indicating instability.
Capture by the fusion protein’s own target, which retains only intact constructs, brought all peptides back onto superimposed profiles, confirming the reading.
The final case study addresses antibody-drug conjugates, where Total Antibody and Total ADC endpoints together describe payload deconjugation in vivo. Producing both by immunoassay needs at least two critical reagents unlikely to exist in early discovery.
Using 10 µL of mouse plasma, the workflow combined automated immunoprecipitation with on-bead tryptic digestion at 70 °C in one hour, then split the sample: one aliquot for signature peptides as Total Antibody, the other treated with papain to release payload as the Total ADC surrogate. In one example the profiles tracked each other, indicating stable conjugation; in another the Total ADC trailed the Total Antibody, revealing minor in vivo instability relevant to lead selection. More on these endpoints in our resource on the bioanalytical characterization of ADCs.
What it means for protein bioanalysis
Read together, the seven studies describe a toolkit rather than a hierarchy. Each turns on a specific strength: molecular specificity against an abundant near-identical isoform, multiplexed subclass resolution, access to a post-translational modification, integrity assessment across a construct, or independence from unavailable reagents. Each is matched to a scientific question, a modality and a development stage.
The authors expect the role of LC-MS in protein bioanalysis to grow as modalities become more complex, supported by advances in high-resolution instrumentation, automation and sample preparation, and by maturing hybrid workflows in regulated settings.
The practical takeaway is the phrase in the paper’s own title: fit for purpose. In protein quantification the right question is not which platform is better, but which one answers the question in front of you. That assessment is where bioanalytical services start, and the same reasoning underpins platform choice in large molecule bioanalysis across discovery and regulated study phases.
Source: Wilson A. et al. Fit-for-purpose application of LC-MS in protein bioanalysis, case studies shared within the European Bioanalysis Forum Hybrid MS team. Bioanalysis, 2026;18(7):561–569. Read the full open-access article: https://www.tandfonline.com/doi/full/10.1080/17576180.2026.2683269
