Ion suppression is a loss of analyte signal caused by co-eluting matrix components competing for charge during electrospray ionisation. It is the dominant source of inaccuracy in quantitative LC-MS, it is invisible in a clean standard curve, and regulated methods must demonstrate it has been assessed and controlled.
The practical problem is that suppression does not announce itself. A calibration curve prepared in neat solvent can be linear, precise and entirely misleading, because the samples it is used to quantify contain phospholipids, salts and dosing vehicle that the standards do not. Recovery looks acceptable. Precision looks acceptable. Accuracy against an authentic incurred sample does not.
What causes ion suppression in LC-MS?
Electrospray produces charged droplets that shrink by solvent evaporation until charge density forces ion release into the gas phase. Everything that co-elutes with the analyte competes for the finite charge available at the droplet surface and for the finite space on that surface.
Three mechanisms dominate:
- Competition for charge. A co-eluting species with higher gas-phase basicity or higher surface activity preferentially takes the available charge. The analyte remains in solution and is never detected.
- Altered droplet properties. Non-volatile solutes raise viscosity and surface tension and slow evaporation, so fewer droplets reach the Rayleigh limit within the source. Involatile buffer salts such as phosphate and ion-pairing reagents such as trifluoroacetic acid suppress heavily through this route.
- Co-precipitation and adduct formation. Analyte trapped in precipitating matrix, or distributed across sodium and potassium adducts rather than the monitored protonated species, is lost from the quantified channel.
Suppression is more severe in electrospray than in atmospheric-pressure chemical ionisation, more severe in negative mode for some matrices, and worse at higher flow rates where droplets are larger and desolvation less complete.
LC-MS PK Peak Integrity Engine
Interrogate carryover, matrix effects, ion suppression, and integration integrity before your PK data goes to QC.
Which matrix components cause the worst ion suppression?
In bioanalysis the offenders are predictable, which is what makes suppression tractable. Knowing which class is responsible determines which sample-preparation change will help.
| Component | Typical source | Where it elutes | Effective countermeasure |
|---|---|---|---|
| Glycerophosphocholines | Plasma, serum, tissue | Late; strongly retained on C18, often eluting into the next injection | Phospholipid-removal plate; extended wash; mixed-mode SPE |
| Lysophospholipids | Plasma, serum | Mid-gradient, frequently overlapping small-molecule analytes | Phospholipid-removal plate; LLE |
| Inorganic salts | Plasma, urine, buffer carry-over | Void volume | Divert the first 0.5-1 min to waste; increase retention |
| Anticoagulant counter-ions | EDTA, citrate, heparin tubes | Void volume | Match anticoagulant between standards and samples; divert |
| Dosing-vehicle polymers | PEG 400, Tween, Cremophor | Broad, spanning much of the gradient | SPE; chromatographic resolution; monitor vehicle blanks |
| Plasticisers and surfactants | Tubes, tips, filter membranes | Mid to late | Change consumables; run a consumables blank |
| Endogenous metabolites | Urine, bile, tissue homogenate | Variable | Chromatographic resolution; selectivity screen across lots |
Phospholipids deserve particular attention because they carry over. A strongly retained glycerophosphocholine from injection 40 can elute during injection 41 and suppress an analyte that was quantified correctly in every validation run. Monitoring the m/z 184 precursor-ion scan in positive mode maps where they land in a given gradient.
How do you detect ion suppression by post-column infusion?
Post-column infusion locates suppression along the chromatogram. It answers where in the run the source is being compromised, which is what determines the fix.
Set up a syringe pump delivering a constant flow of analyte into the column effluent through a tee between column and source. Inject an extracted blank matrix sample. With no analyte in the injection, the detector sees a flat baseline from the infused analyte. Any dip in that baseline marks a retention time where something in the matrix is suppressing the signal.
Read the trace against the analyte’s retention time. A dip at the void volume indicates salts and polar endogenous material, addressed by diverting the early eluent to waste or increasing retention. A dip coinciding with the analyte peak is the serious case and requires either a chromatographic change or a different extraction. A dip late in the run that appears clean now will suppress the following injection.
Post-column infusion is qualitative. It shows where suppression occurs but not how much bias it introduces at the concentrations being reported, which is what the post-extraction spike is for.
How do you quantify a matrix effect after extraction?
The post-extraction spike, described by Matuszewski and colleagues in 2003, separates the matrix effect from extraction losses by comparing three sets of samples prepared at the same nominal concentration.
- Set 1 – analyte in neat mobile phase or reconstitution solvent.
- Set 2 – blank matrix taken through the full extraction, then spiked with analyte.
- Set 3 – blank matrix spiked with analyte before extraction, then extracted.
| Parameter | Calculation | What it isolates | Interpretation |
|---|---|---|---|
| Matrix effect (ME) | (Set 2 / Set 1) × 100% | Ionisation only | <100% suppression; >100% enhancement; 100% no effect |
| Recovery (RE) | (Set 3 / Set 2) × 100% | Extraction efficiency only | Losses during sample preparation |
| Process efficiency (PE) | (Set 3 / Set 1) × 100% | Combined effect | PE = (ME × RE) / 100 |
Separating the two matters because they call for opposite responses. Poor recovery with a clean matrix effect is an extraction-chemistry problem: change solvent, pH or sorbent. Good recovery with severe suppression is a chromatography or source problem: change the gradient, the column, or divert more eluent to waste. Optimising extraction recovery when the fault is suppression wastes weeks.
Report the variability across lots, not just the mean. A method with a mean ME of 96% that ranges from 71% to 118% across six donors is not controlled; the mean simply conceals it.
What does ICH M10 require for matrix effect assessment?
This is where a great deal of published guidance is now out of date. The EMA guideline in force from February 2012 required a specific matrix-factor calculation. ICH M10, adopted by the CHMP in July 2022 and effective from 21 January 2023, does not.
| EMA 2011 guideline (from Feb 2012) | ICH M10 (effective Jan 2023) | |
|---|---|---|
| Required calculation | Matrix factor: peak area with matrix (post-extraction spike) ÷ peak area without matrix (neat solution) | No matrix-factor calculation prescribed |
| Internal standard handling | IS-normalised MF = MF(analyte) ÷ MF(IS) | Not prescribed |
| Matrix sources | At least 6 lots from individual donors | At least 6 different sources or lots |
| Replicates | Two concentration levels: low (max 3× LLOQ) and high (near ULOQ) | At least 3 replicates of low and high QC from each source |
| Acceptance criterion | CV of the IS-normalised MF across the 6 lots ≤ 15% | Accuracy within ±15% of nominal and precision ≤ 15% CV for each individual source |
| Special matrices | Addressed separately | Haemolysed and lipaemic matrix, and relevant patient populations, case by case where expected in the study |
The shift is from a ratio metric to a performance test. ICH M10 asks whether the method still measures accurately and precisely in every individual lot, rather than whether a computed factor is consistent across them. In practice the M10 test is harder to pass, because a single problematic donor lot fails on its own terms instead of being averaged into a coefficient of variation.
The matrix factor has not become useless. It remains the most efficient diagnostic during development, and it is still required by some regional and non-ICH programmes. But a validation report submitted against ICH M10 that presents only IS-normalised matrix factors has not demonstrated what the guideline asks for.
Note also that M10 requires this assessment for each analyte and its internal standard, and that the internal standard’s own suppression is assessed rather than assumed to cancel.
Which sample preparation reduces matrix effects most?
Sample preparation is the highest-leverage variable. The ranking below is consistent across published comparisons for small molecules in plasma, though the margin varies with analyte polarity.
| Technique | Typical matrix effect | Phospholipid removal | Cost and throughput | Best suited to |
|---|---|---|---|---|
| Protein precipitation (PPT) | Severe; commonly 30-60% suppression | Minimal – phospholipids remain | Lowest cost, highest throughput | Early discovery; high-concentration analytes |
| PPT with phospholipid-removal plate | Moderate to good | Typically >90% of glycerophosphocholines | Moderate; same workflow as PPT | Regulated methods where PPT selectivity suffices |
| Liquid-liquid extraction (LLE) | Good for neutral and lipophilic analytes | Good, solvent-dependent | Moderate; harder to automate | Neutral analytes; low-level quantification |
| Solid-phase extraction (SPE) | Best; often within 15% of neat | Excellent with mixed-mode sorbents | Highest cost; automatable | Regulated bioanalysis; low LLOQ |
| Supported liquid extraction (SLE) | Comparable to LLE | Good | Moderate; plate format | LLE chemistry without emulsions |
| Dilute and shoot | Variable; depends entirely on dilution factor | None | Lowest | Urine; high-concentration analytes only |
Protein precipitation is the default in early work and the most common reason a method fails matrix-effect assessment later. If a method must survive validation, the extraction decision is worth making before the assay is locked, not after.
How does chromatography reduce ion suppression?
Where extraction cannot remove an interference, chromatography can move the analyte away from it.
Increasing retention shifts the analyte off the void volume where salts and polar endogenous material elute; this alone resolves a large fraction of suppression seen with short, fast gradients. Diverting the first minute and the final high-organic wash to waste keeps both the salt front and the phospholipid tail out of the source entirely, and costs nothing but a valve program.
Extending the gradient or changing selectivity, for instance moving from C18 to phenyl-hexyl or pentafluorophenyl, can separate an analyte from a co-eluting suppressor that no extraction removes. Reducing flow rate improves desolvation and reduces suppression, which is part of why microflow and nanoflow LC-MS show markedly less matrix effect than analytical-flow methods.
HILIC deserves a caution. It resolves polar analytes that reversed-phase cannot retain, but it retains salts strongly, and phospholipids elute early rather than late. HILIC changes where suppression occurs rather than removing it, and a post-column infusion trace should always be re-run after switching modes. The same applies after any change to gradient, column chemistry or flow rate, which is covered in the systematic checklist for diagnosing sensitivity loss.
Can an internal standard correct for ion suppression?
A stable isotope-labelled internal standard is the strongest available compensation, because it co-elutes with the analyte and experiences the same suppression at the same instant. The ratio survives what the raw response does not.
Three limits apply. Deuterated standards can separate chromatographically from the analyte through the deuterium isotope effect, particularly with several deuterium atoms on a reversed-phase gradient; even a few seconds of offset places the standard in a different suppression environment. Carbon-13 and nitrogen-15 labels avoid this and are preferred where available. Second, at severe suppression the analyte response can fall towards the noise, and no ratio recovers a signal that is no longer measurable — the practical floor is set by the same considerations that govern the limit of detection. Third, a structural analogue internal standard does not co-elute exactly and will not track suppression reliably across lots.
An internal standard compensates. It does not fix. A method relying on a SIL-IS to mask 60% suppression is one matrix change away from failing, and ICH M10 requires the internal standard’s own matrix effect to be assessed rather than presumed.
Does APCI avoid ion suppression?
Atmospheric-pressure chemical ionisation is substantially less susceptible than electrospray, because ionisation occurs in the gas phase after desolvation rather than in competition at a droplet surface. Where an analyte is amenable, switching source can resolve a suppression problem outright.
The constraint is analyte scope. APCI requires thermal stability and reasonable volatility, which excludes many peptides, phospholipid-like species and thermally labile compounds. Sensitivity for polar and basic small molecules is frequently lower than electrospray even accounting for the suppression avoided. Atmospheric-pressure photoionisation occupies similar ground for non-polar analytes.
Treat source change as a genuine option to evaluate rather than a fallback, but confirm sensitivity meets the required lower limit before committing — the net gain must be measured against the suppressed electrospray response, not the neat-standard response.
Frequently asked questions
Is ion suppression the same as a matrix effect?
Matrix effect is the broader term, covering both suppression and enhancement of analyte response by co-eluting components. Ion suppression is the more common and more damaging direction. Enhancement is less frequent but equally a validation failure, and both are captured by the same assessment.
What level of matrix effect is acceptable under ICH M10?
ICH M10 does not set a threshold on the matrix effect itself. It requires that accuracy stay within ±15% of nominal and precision within 15% CV for low and high QC prepared in each of at least six matrix sources. A method with 40% suppression that still meets those criteria in every lot is acceptable; a method with 10% suppression that varies unpredictably between lots is not.
Can ion suppression be corrected mathematically?
Not reliably. Suppression varies between matrix lots, between subjects, and with sample age and handling, so a correction factor derived during validation does not transfer to study samples. Calibrating in matrix rather than neat solvent compensates for the average effect but not for lot-to-lot variability, which is the part that causes failures.
How often should matrix effects be re-assessed?
After any change to extraction chemistry, chromatography, source conditions or consumables, and whenever the method is applied to a new matrix, species or patient population. Haemolysed and lipaemic samples warrant specific assessment where they are expected in the study.
References
- ICH M10 Bioanalytical Method Validation and Study Sample Analysis. Adopted by CHMP July 2022; effective 21 January 2023. European Medicines Agency.
- Guideline on bioanalytical method validation, EMEA/CHMP/EWP/192217/2009. In effect 1 February 2012. European Medicines Agency.
- Matuszewski BK, Constanzer ML, Chavez-Eng CM. Strategies for the assessment of matrix effect in quantitative bioanalytical methods based on HPLC-MS/MS. Analytical Chemistry 2003;75(13):3019-3030. doi:10.1021/ac020361s.
Related: how to calculate the limit of detection and signal-to-noise ratio determination.