Week 8 — Nov 2: Mass Spectrometry

(Lecture 7.) The midterm is behind us. Separations pulled a mixture apart; a UV detector told you how much, but not what. Mass spectrometry weighs each component — and, with fragmentation, weighs its pieces — which makes it the primary tool for identity at trace level and for quantitation where nothing else is sensitive or specific enough. It is also the last pure-instrument week before the course turns to the data layer.

The one idea

A mass spectrometer measures mass-to-charge, nothing more. Everything useful — a formula, a structure, a concentration at parts-per-billion — is inference built on that one measurement, and the quality of the inference depends entirely on how well the ionization and the calibration are controlled.

The pieces

Low-resolution (unit-mass, e.g. QqQ) vs high-resolution (HRMS, e.g. Q-TOF / Orbitrap) is the distinction that matters most:

• QqQ excels at targeted quantitation — selected/multiple reaction monitoring (SRM/MRM): pick a precursor m/z, fragment it, monitor a specific product ion. Very selective, very sensitive, wide dynamic range.
• HRMS measures accurate mass to a few ppm, which (with isotope pattern) gives an elemental formula — the starting point for identifying an unknown impurity, degradant, or modification. It also records everything at once, so you can go back to the data for a peak you didn’t know to look for.

What MS is used for

The quantitation problem — ion suppression

ESI response is not a fixed property of an analyte. Co-eluting matrix components compete for charge and change the analyte’s signal, often suppressing it, sometimes by more than half — and the effect drifts across a batch. This is the central validation challenge for LC–MS quantitation, and the standard fixes:

• A stable-isotope-labeled internal standard (SIL-IS) — chemically identical, co-elutes exactly, experiences the same suppression, so the analyte/IS ratio is preserved.
• Matrix-matched calibration and a measured matrix factor; post-column infusion experiments to map where suppression occurs.
• Better chromatography to move the analyte away from the suppression zone.

Validation follows Q2 for impurity/assay work and ICH M10 for bioanalytical methods (calibration model, QCs, matrix effect, carryover, stability, incurred-sample reanalysis).

The multi-attribute method

MAM is a single LC–HRMS peptide-mapping assay that monitors a predefined list of product quality attributes — specific oxidations, deamidations, glycation, glycoforms, sequence variants, clips, C-terminal lysine — each quantified from its peptide’s extracted-ion chromatogram. One method, run under GMP, can replace several conventional assays (icIEF, released-glycan HILIC, parts of peptide mapping) that each measured one attribute indirectly.

Its second half is new peak detection (NPD): the software compares each sample map against a reference and flags any peak that is new or changed — the “purity” safety net the targeted attribute list would otherwise miss. NPD is also MAM’s hardest problem: too sensitive and every run throws false positives that need investigation; too lax and it stops being a safety net. Platform-to-platform transfer and the data-system burden are the other adoption barriers.

MAM previews the modalities weeks: mass spectrometry absorbing a panel of separations into one information-rich measurement — with the analyst now responsible for a detection threshold (NPD) instead of a set of pass/fail assays.

Risk-assessment assignment

Build the method FMEA (Week 2) for an LC–MS/MS nitrosamine method at a 30 ng/day acceptable-intake limit. Give particular weight to the MS-specific failure modes: ion-suppression drift across the batch, a SIL-IS with isotopic impurity, in-source fragmentation creating an interfering ion, mass-calibration drift, and carryover. Score the detectability of each — which would the run’s own system-suitability and QC samples actually catch?

Where the analyst sits

Accurate mass gives you a formula, not a structure — C₉H₁₀N₂O₃ is dozens of molecules, and fragmentation narrows it but rarely to one. Deciding when an identification is confirmed (matching a synthesised standard? orthogonal NMR? a defensible mechanistic argument?) is judgment, and it has regulatory weight — an “identified” impurity is controlled differently from an “unspecified” one. On the quantitation side, the analyst decides whether the ion-suppression correction is trustworthy for this batch. That is the STEAM “A”. The refrain: science → evidence → reduced uncertainty → control → regulatory confidence → patient trust.

For discussion

• HRMS gives you an unknown degradant’s formula to 2 ppm. Walk through what you do next to get to a structure, and where you would stop and call it “sufficiently identified.”
• A QqQ SRM method and an HRMS method both quantify a nitrosamine at the limit. What are the arguments for each in a regulatory filing?
• Your LC–MS/MS assay for a drug in plasma reads 15% low on incurred samples versus spiked standards, even with a SIL-IS. What could still cause that?
• MAM’s new-peak-detection flags a 0.08% peak that turns out to be a known, previously-uncontrolled sequence variant. Should it have been on the targeted attribute list? Who decides?
• One MAM assay replaces icIEF, released glycans, and part of the peptide map. What is lost, if anything, by consolidating?
• When is “the isotope pattern matches” strong evidence, and when is it nearly worthless?

Source note. MS fundamentals follow standard texts (Gross, Mass Spectrometry; de Hoffmann & Stroobant, Mass Spectrometry: Principles and Applications). Bioanalytical validation: ICH M10; impurity work connects to ICH M7 and Q3; residual solvents to USP ⟨467⟩ / Q3C. MAM follows the published inter-company work (the MAM consortium papers; Rogers et al.) and the evolving regulatory feedback on new-peak detection; native / charge-detection / HDX-MS follow the biopharmaceutical-characterisation literature. (Instructor: confirm ICH M10 status and current thinking on MAM/NPD in submissions; keep the nitrosamine risk assignment aligned with the Week 2 example rather than duplicating it.)

• ←7 · Oct 26
• 9 · Nov 9→