TotalLab

How Host Cell Protein ppm Is Calculated From LC-MS Data

Host cell protein (HCP) ppm is nanograms of host cell protein per milligram of product protein, written ng/mg, and one ng/mg is one part per million by weight. That is the whole definition, and almost every argument about HCP numbers is an argument about how the numerator and the denominator were produced rather than about the arithmetic. This page works the calculation through twice with consistent numbers, sets out the three quantitation approaches in USP General Chapter <1132.1> and the label-free algorithms underneath them, shows one dataset reported three legitimate ways, and uses USP’s own study of 18 independent analyses of identical aliquots of one monoclonal antibody to set the floor on how closely two laboratories can be expected to agree. It ends with the nine items that have to travel with a ppm figure for it to mean anything.

Key facts

  • HCP ppm means nanograms of host cell protein (HCP) per milligram of product protein, written ng/mg. One ng/mg equals one ppm by weight.
  • The two USP chapters treat the word ppm differently, and neither bans it. USP General Chapter <1132> states that results are “reported as the ratio of measured HCP (ng/mL) to the product concentration (mg/mL) resulting in units of ng/mg”, and adds that “Although ppm has been used historically, it is not advised because this term is used to reflect mass per unit volume for other types of tests.” USP General Chapter <1132.1> is more relaxed: its Terminology section (Section 2) defines ng/mg as the ratio of nanograms of HCP to milligrams of product protein and lists parts per million among the synonyms in common use for it. Write ng/mg in the report; ppm is not an error, and <1132.1> records it as ordinary usage.
  • USP <1132> sets the practical sensitivity expectation: “The detection limit for many LC-MS/MS methods is currently in the range of about 10 to 100 ng of HCP per mg of product.” Modern data-independent acquisition methods now report limits well below that range.
  • USP General Chapter <1132.1> sets out three LC-MS quantitation strategies and numbers them in Section 5.1: 5.1.1 relative to product protein, 5.1.2 relative to spiked-in proteins, 5.1.3 relative to spiked-in peptides. The Method A, B and C labels come from the qualification literature, not from USP.
  • USP <1132.1> puts current estimates of the accuracy of MS-based HCP quantitation at about plus or minus two-fold, and says actual error can be significantly higher depending on the experiment. That is the number to hold in mind before treating any ppm figure as exact.
  • In the chapter’s own reproducibility case study, 18 independent analyses of identical aliquots of one monoclonal antibody returned 52 to 104 ng/mg for the most abundant HCP, with every result inside a factor of two of the mean.
  • There is no fixed regulatory HCP limit. The 2025 BioPhorum industry review states that “Currently, 100 ng/mg (ppm) is a standard rule of thumb across the industry”, and that “there is no official maximum threshold level of HCP set in the industry.”
  • A total HCP ppm from mass spectrometry is a sum of what was identified and quantified, bounded by the FASTA search space and the limit of quantification, so it is not interchangeable with an ELISA total.

What HCP ppm actually means

HCP ppm is the mass of residual host cell protein divided by the mass of product protein in the same sample, expressed as nanograms per milligram. A result of 45 ppm means 45 nanograms of host cell protein accompany every milligram of drug substance protein. The unit is mass-based and dimensionless: 1 ng over 1 mg is 10⁻⁹ g over 10⁻³ g, which is 10⁻⁶, one part per million by weight.

Two consequences follow. A ppm figure has a numerator and a denominator, and a lab can get either one wrong without the other showing it. And “product” has to mean something specific. USP now says it means product protein, not total protein and not total solids. USP <1132> defines the unit directly: “ng/mg: The numerical quantity (ratio) of HCP per product, where ng represents HCP mass and mg represents the product mass. It is calculated by dividing the HCP concentration (ng/mL) by the product protein concentration (mg/mL).”

On whether you are allowed to say ppm at all, be precise, because the two chapters differ and a lot of web copy overstates this. USP <1132>, the immunoassay chapter, discourages the term: ppm “is not advised because this term is used to reflect mass per unit volume for other types of tests”. USP <1132.1>, the LC-MS chapter, does not repeat that advice. Its Terminology section defines ng/mg and lists parts per million as a “common synonym” (USP <1132.1>, Section 2), which records the usage rather than warning against it. So: USP has not banned ppm, and anyone who tells you it has is quoting one chapter and ignoring the other. The practical rule is unchanged. Write ng/mg in the report, because that is the unambiguous form and it is what both chapters define. Say ppm in conversation, because everyone does. This page uses both.

The arithmetic, worked through

Two routes give the same answer. All numbers here are illustrative.

Route 1, by mass in the digested aliquot:

  1. Digest a 10 microliter aliquot of monoclonal antibody drug substance at 10.0 mg/mL. The aliquot contains 100 micrograms, which is 0.100 mg, of product protein.
  2. LC-MS quantitation returns 4.5 ng of phospholipase B-like 2 (PLBL2) in that aliquot.
  3. ppm = 4.5 ng divided by 0.100 mg = 45 ng/mg = 45 ppm

Route 2, by concentration:

  1. The drug substance is 10.0 mg/mL of product protein.
  2. LC-MS quantitation returns PLBL2 at 450 ng/mL in the drug substance.
  3. ppm = 450 ng/mL divided by 10.0 mg/mL = 45 ng/mg = 45 pp

The routes agree because 10 microliters of a 450 ng/mL solution contains 4.5 ng. If yours disagree, you have a dilution factor error, not a quantitation problem. For a total, sum the individual masses first, then divide once:

Total ppm = (sum of the masses of all reported HCPs, in ng) divided by (mass of product protein, in mg)

Summing individual ppm values gives the same answer only if every value used the identical denominator. That is usually true within one report and almost never true across two.

Diagram showing HCP ppm as nanograms of host cell protein divided by milligrams of product protein

The denominator problem: which protein are you dividing by?

The denominator should be the mass of product protein, and the common ways of measuring it do not agree. This is the largest source of non-comparability between labs, and it is invisible in the final number.

Absorbance at 280 nm (A280) is fast, non-destructive and the default. It converts absorbance to concentration using an extinction coefficient, usually calculated from the product sequence. Two failure modes: the coefficient may be a generic antibody value rather than the product-specific one, which shifts the answer by several percent; and A280 counts everything that absorbs at 280 nm, including the host cell proteins themselves and residual nucleic acid. That is negligible in purified drug substance and not negligible in an in-process sample at several thousand ppm.

Amino acid analysis (AAA) hydrolyzes the sample, quantifies the constituent amino acids against standards, and back-calculates protein mass from the known composition. It is the reference method and the tiebreak when other assays disagree, but it is slow, destructive, and it measures all protein rather than the product specifically.

Mole-to-mass conversion is the failure mode nobody writes into an SOP, and USP <1132.1> devotes Section 7.2 Units of Measurement to it. LC-MS/MS quantitation is mole based and ELISA reports mass, so converting between them requires a molecular weight for each HCP. That molecular weight almost always comes from a database sequence, which reflects the translated gene rather than the protein that was actually in the vial, so it carries no post-translational modification. The chapter points at a sharper version of the same problem: some database entries are protein fragment sequences rather than full-length proteins, and a fragment sequence “will underrepresent the mass of HCP present” (USP <1132.1>, Section 7.2). Two labs using different database releases can therefore divide by different molecular weights for the same protein and disagree without either making a mistake. It is checkable in an afternoon: pull the accessions behind your top ten HCPs and look at how many are fragments.

Colorimetric assays (BCA, Bradford) respond differently to different proteins, so their answer depends on the calibrator. USP <1132> notes that “The most commonly used methods are the bicinchoninic assay (BCA), the Bradford assay, and amino acid analysis (AAA)” and that “Absorbance at 280 nm (A280) may also be used”, and advises that “One should consider using two orthogonal methods to exclude a gross over- or underestimation.”

What USP <1132.1> says, and the 2025 correction

USP General Chapter <1132.1>, “Residual Host Cell Protein Measurement in Biopharmaceuticals by Liquid Chromatography-Mass Spectrometry”, was published in USP-NF 2025 Issue 1 on 1 November 2024, with an official date of 1 May 2025. USP published an erratum on 31 January 2025, also effective 1 May 2025. As summarized by the ECA Academy, it changed “product (or polysorbate)” to “product protein (or polysorbate)” in section 4 (Sample Preparation), changed “product” to “product protein” in three places in section 5 (Quantitation), and changed “Residual HCP ELISA” to “HCP ELISA” in the Introduction.

That reads like housekeeping. It is not. Before the correction a lab could read “per mg of product” as per mg of anything in the vial. The chapter now says product protein. If your SOP divides by total protein and your product is formulated with albumin or another protein excipient, your ppm figures have been systematically low. For how the three methods are run, see running USP <1132.1> Methods A, B and C.

Three panels showing how the same HCP sample gives three different ppm values depending on the denominator used

The numerator problem: how an MS response becomes a mass

A mass spectrometer measures ion current, not mass, so turning a signal into nanograms requires a calibration assumption. USP <1132.1> frames the choice as three methods, numbered 5.1.1, 5.1.2 and 5.1.3 in Section 5.1 and widely called A, B and C after the qualification literature that compared them. They are distinguished by what the unknown is measured against, and the label-free algorithms are the arithmetic underneath.

Section 5.1.1 (Method A) calibrates against the product protein itself. Nothing is spiked, which makes it cheap, but the product is typically five or six orders of magnitude more abundant than the HCPs, so it is a long extrapolation. Section 5.1.2 (Method B) spikes purified proteins of known mass before digestion, so the standard experiences the same digestion, cleanup and injection as the HCPs and corrects for digestion efficiency and sample loss. Section 5.1.3 (Method C) spikes stable isotope labeled (SIL) peptides matching specific HCP tryptic peptides. Heavy and light forms co-elute and ionize identically, which makes it the most accurate option for the HCPs you target, but it only quantifies proteins you already chose to watch, and if it is spiked after digestion it cannot correct digestion efficiency, often the largest single error in the workflow. Sourcing the heavy peptides used to be the obstacle. USP now catalogs stable isotope labeled heavy peptide analytical reference materials for nine named CHO host cell proteins, including cathepsin D (1130704), protein disulfide isomerase (1130727), clusterin (1130730, 1130741, 1130752), lipoprotein lipase (1130763, 1130774, 1130785) and PLBL2 (1130796), which gives a Section 5.1.3 number a traceable external anchor. USP classifies these as analytical reference materials and states that they “are different from USP Reference Standards and not required for compendial compliance”.

Hi3, also called Top3, takes the mean MS signal of a protein’s three most intense tryptic peptides as a proxy for molar amount. It rests on the finding by Silva et al. (2006) that “the average MS signal response for the three most intense tryptic peptides per mole of protein is constant within a coefficient of variation of less than ±10%”, with standard proteins quantified to “a relative error of less than ±15%”. Hi3 still needs one protein of known amount in the run to set the response factor, so it pairs with Method A or Method B rather than replacing them. Our Hi3 peptide quantification method resource covers the settings.

iBAQ, intensity-based absolute quantification, divides summed peptide intensity by the number of peptides a protein could theoretically produce, correcting for the fact that large proteins yield more peptides. It was introduced in Schwanhäusser et al. (2011) and is well suited to ranking HCPs by abundance within a sample. Treating an iBAQ value as an absolute mass without external calibration is not defensible for release testing.

Performance figures come from the cited studies and apply to those methods, not to yours.

Total HCP ppm is not one number

A total HCP ppm from mass spectrometry is the sum of the proteins the analysis identified and quantified. It is not a measurement of all protein impurity present. Three boundaries define it, and all three are analyst choices.

The search space. Peptides are matched against a FASTA database, and a host cell protein absent from that database cannot be identified, so its mass never enters the sum. Chinese hamster ovary (CHO) proteome releases differ in entry count, isoform handling and contaminant list, and labs differ on whether they search a full proteome or a curated HCP subset.

The detection limit. USP <1132> puts the working range of the technique at “about 10 to 100 ng of HCP per mg of product”, and a modern DIA method beats that by more than an order of magnitude, which is exactly why totals are not comparable across labs. Proteins below the limit of quantification (LOQ) are excluded from the sum, included at their reported value, or imputed at the LOQ. All three conventions are in use and they give different totals. Published LOQ figures sit in the low single-digit ppm range and below: a 2026 ICH Q2(R2)-aligned validation reported an “abundance-aware LLOQ of 3.6 ppm”, and an Orbitrap Astral benchmarking study reported roughly 1.6 ppm for DDA and 0.6 ppm for DIA.

USP <1132.1> gives a convention for the proteins you did not find, and it is worth adopting because it is defensible and because almost nobody uses it. Section 6 Reporting Results describes reporting an HCP that was not detected as less than the level the system has been shown to be sensitive to, using less than 10 ng/mg as its worked example. The number in that statement is not free: it has to be earned from data, and the chapter describes establishing the lowest level detectable at least 90% of the time as a common approach to defining sensitivity. So the report does not say “not detected”. It says “less than 10 ng/mg”, and the method file behind it says why 10, and shows the spiked system suitability sample that demonstrated it on the day. That is a small change in wording and a large change in what a reviewer can do with the result.

The protein inference rule. Some tryptic peptides occur in more than one protein. As Nesvizhskii and Aebersold (2005) put it, “The same peptide sequence can be present in multiple different proteins. Therefore, the identification of such shared peptides can lead to ambiguities in the determination of the identities of the sample proteins.” Software may assign a shared peptide to the most likely protein, split its intensity, group the candidates, or discard it, and each rule changes the per-protein mass and therefore the total.

So a mass spectrometry total and an ELISA total measure different things and cannot be reconciled by arithmetic. ELISA reports immunoreactive mass against a polyclonal antibody with its own coverage gaps, calibrated against a null-cell-line standard that is not your HCP population. We cover the mechanics in why ELISA and LC-MS HCP numbers disagree and the antibody side in why you need to know HCP antibody coverage.

Worked example: one dataset, three ppm figures

All numbers below are illustrative and internally consistent. They show how reporting rules move the answer; they do not represent any real product.

A 100 µg aliquot of purified monoclonal antibody drug substance is digested and analyzed. The software reports 62 HCPs identified, 41 of them above the LOQ.

  • Summed mass of the 41 quantifiable HCPs, calibrated against the product protein (Section 5.1.1): 3.80 ng
  • Summed mass of the same 41 HCPs, calibrated against a spiked protein standard (Section 5.1.2): 5.32 ng
  • The 21 sub-LOQ HCPs, imputed at the LOQ of 0.05 ng each: 21 × 0.05 = 1.05 ng
  • Product protein in the aliquot by A280 with the product-specific extinction coefficient: 0.100 mg
  • Product protein in the same aliquot by amino acid analysis: 0.0870 mg (A280 over-reads by about 15% here)

Table 2. The same dataset reported three ways

One vial, one injection, one peak list. Thirty-eight ppm, fifty-three ppm or seventy-three ppm, a spread of 1.9-fold, and all three are arithmetically correct. No instrument problem, no analyst error, no bad chromatogram.

Typical acceptance criteria, and what regulators actually require

There is no fixed regulatory HCP limit in any ICH, FDA or EMA guidance. Expectations are risk based and product specific.

ICH Q6B addresses host cell proteins in section 6.2.1: “For host cell proteins, a sensitive assay, e.g., immunoassay, capable of detecting a wide range of protein impurities is generally utilized.” On limits, section 4.0 states that “Acceptance criteria should be established and justified based on data obtained from lots used in preclinical and/or clinical studies, data from lots used for demonstration of manufacturing consistency, data from stability studies, and relevant development data.” Section 4.1.3 adds that “Individual and/or collective acceptance criteria for impurities should be set, as appropriate.” No number appears. The guideline is adopted in Europe as CPMP/ICH/365/96.

So where does 100 ppm come from? From what has already been approved. USP <1132> observes that “final products may have HCP levels ranging from <1 to 100 ng/mg”. The 2025 BioPhorum Development Group industry review by Coye and colleagues is more direct: “Currently, 100 ng/mg (ppm) is a standard rule of thumb across the industry, based on the level of currently approved products. However, this level does not have to be reached for HCPs present in the final drug substance to become problematic.” The same review states that “there is no official maximum threshold level of HCP set in the industry.”

Read that second sentence carefully. A single immunogenic or enzymatically active HCP at 2 ppm can matter more than 80 ppm of inert background. Polysorbate-degrading lipases are the standard example: lipoprotein lipase (LPL) and phospholipase B-like 2 (PLBL2) cause formulation failures well below any total-HCP criterion, which is why identity matters as much as total. See the named high-risk host cell proteins. Treat 100 ppm as an industry orientation point, not a specification, and never write it into a document as a regulatory requirement. Justify your criterion from your own clearance data, your dose and route, and a documented risk assessment.

Where USP does give numbers, they are method performance numbers rather than product limits, and they are the ones worth writing into an SOP. On spiking studies, USP <1132> states that “Acceptable spike recovery is typically between 70% and 130%”, and allows a wider window, 50% to 200%, for a spike at or near the quantitation limit. The chapter also describes action limits set at a fraction of the reject limit, on the order of 50% to 75% of it, so that a drifting process is caught before a lot fails.

USP’s own reproducibility study: 52 to 104 ng/mg for the same HCP in the same sample

The best answer to “why do two labs get different numbers” is not a vendor benchmark. It is in Section 6 of USP General Chapter <1132.1>, and it is a case study USP ran on one monoclonal antibody.

The design is what makes it useful. The same monoclonal antibody sample was tested 18 times over a period of weeks. Each analysis was complete and independent: a separate aliquot of the same sample, denatured and digested on its own, then acquired and processed. The aliquots were identical. The instruments were similar LC-MS/MS systems but not the same ones, and the work was spread across different laboratories and different analysts. In other words, everything that varies between two contract labs was varied deliberately, and everything about the sample was held constant.

The results:

  • For the most abundant HCP in the sample, reported values ranged from 52 to 104 ng/mg, a two-fold spread.
  • Lower abundance HCPs were not detected at all in some of the runs.
  • Every result sat within a factor of two of the mean across the 18 analyses.

USP’s reading is that this is the expected performance of the technique rather than a failure of it, and that batch-to-batch comparisons should be made with that variability in mind. The chapter states the general case alongside the study: current estimates of the accuracy of MS-based quantitation are often reported as plus or minus two-fold, and actual error can be significantly higher depending on the experiment.

Three things follow, and they are worth writing into a comparison protocol.

  1. A ratio of less than two between two labs is not a discrepancy. It is the measurement. Investigating it as an out-of-trend event wastes analyst time and produces a report that cannot conclude anything.
  2. A missing HCP is not the same as an absent HCP. Lower abundance proteins dropped out of some of USP’s own runs on identical material. Treat an HCP that appears in three of four batches as a detection-limit observation, not as a process excursion, unless the quantified batches also moved.
  3. Trend against yourself. The variability above is dominated by instrument, laboratory and analyst. Holding all three fixed, and holding the database, the quantitation method and the sub-limit convention fixed with them, is what makes a trend line mean something. TotalLab’s TrendLab software can directly interface with LC-MS data to automatically plot long term trends and identify QC failures in an easy to use dashboard and highlight variances.

None of this excuses sloppy work, and it does not make the seven variables below unimportant. It sets the floor. Even after you have eliminated every one of them, two competent labs measuring identical aliquots of one sample can report 52 and 104 ng/mg for the same protein.

Why two labs disagree about the same vial

Seven independent variables, any one of which moves the number. They sit on top of the floor described above, not underneath it.

  1. Different denominators. A280 with a generic extinction coefficient, A280 with the product-specific coefficient, BCA and amino acid analysis will not agree, and a 15% difference in denominator is a 15% difference in ppm.
  2. Different quantitation methods. Chrone et al. (2025) compared USP Sections 5.1.1, 5.1.2 and 5.1.3 on the same samples and dilutions and concluded that “Quantitative results are highly dependent on standard and method used.”
  3. Different digestion efficiency. Walmsley et al. (2013) found that “significant differences were observed depending on the origin of the trypsin (i.e., bovine vs porcine)”, with bovine trypsins producing more missed cleavages and porcine trypsins more semitryptic peptides. Two labs on different trypsin lots are running different assays.
  4. Different FASTA search space. A protein absent from the database contributes zero to the total, and proteome release, isoform policy and contaminant file all differ between labs.
  5. Different detection limits. A lab at 0.6 ppm LOQ finds and sums HCPs that a lab at 3 ppm never sees. The more sensitive lab reports the higher total, which reads like a worse result and is a better measurement.
  6. Different reporting rules for shared peptides. Razor-peptide assignment, intensity splitting and protein grouping each produce a different per-protein mass from identical spectra.
  7. Different sub-LOQ conventions. Excluded, included as reported, or imputed at LOQ. Three totals from one peak list.

Acquisition mode sits underneath several of these. Data-independent acquisition (DIA) fragments every precursor in a series of wide isolation windows, so quantitation is more complete and more reproducible across runs than data-dependent acquisition (DDA), where the instrument picks precursors on the fly and low-abundance HCPs drop out of some injections. An Orbitrap Astral benchmarking study reported that DDA showed “substantial missingness, with over 50% of peptides dropping out at lower spike levels”, while DIA gave “lower and tighter CV% distributions”. Expect a DIA lab and a DDA lab to report different totals and different HCP lists. See DIA vs DDA for HCP analysis.

SpotMap MS, TotalLab’s LC-MS HCP analysis software, works from DIA or DDA data against a user-supplied FASTA file, which makes items 4, 6 and 7 recorded software settings rather than per-run analyst decisions.

Flow diagram showing five analytical decisions that make two labs report different HCP ppm from the same vial

A reporting checklist

A ppm number on its own is not interpretable. These nine items must travel with it, in the report.

  1. Quantitation method. The USP <1132.1> section used, 5.1.1, 5.1.2 or 5.1.3, named by number rather than only by the A, B and C shorthand, plus the label-free algorithm (Hi3/Top3, iBAQ, other).
  2. The standard. Identity, supplier, lot, purity and spiked amount of any protein or SIL peptide standard, and whether it went in before or after digestion.
  3. The denominator. Which protein assay produced the product protein mass, which extinction coefficient or calibrator was used, and whether the value is product protein or total protein.
  4. Limit of quantification and the sensitivity claim. The LOQ in ng/mg, how it was determined, and the convention applied to sub-LOQ identifications. State the level the system was shown sensitive to and report undetected HCPs as less than that level, the convention in Section 6 of USP <1132.1>, rather than as “not detected”. Establishing the lowest level detectable at least 90% of the time is the usual basis for that number.
  5. Database. FASTA source, release or version, entry count, contaminant file, and whether a full proteome or a curated subset was searched.
  6. Search and inference settings. Enzyme, missed cleavages, modifications, false discovery rate threshold and the protein grouping rule. Our primer on p-values, FDR and q-values covers the statistics.
  7. Software and version. Exact version numbers, including any spectral library version.
  8. Sample preparation. Digestion protocol, protease supplier and lot, denaturant, reduction and alkylation, cleanup, and any depletion step.
  9. Acquisition. Instrument, acquisition mode (DIA or DDA), window scheme or dynamic exclusion settings, gradient and load.

If a contract lab cannot supply all nine, you cannot compare their number to anyone else’s, including your own from last quarter. That burden is a large part of the case for bringing HCP analysis in-house. For the control strategy this sits inside, start with what host cell protein is and HCP analysis in biologics across ELISA, 2D and LC-MS.

Run the calculation yourself

SpotMap MS is TotalLab’s software for LC-MS host cell protein analysis. It processes DIA or DDA data against your own FASTA file, automates the identification and quantitation steps described above, and pairs with AuditSafe for 21 CFR Part 11 audit trails and electronic signatures. See the SpotMap MS product page, read about USP <1132.1> alignment, or request a free trial.

Frequently Asked Questions

Q: What is a good HCP ppm level?
A: For a purified drug substance, most approved products sit below 100 ng/mg, and USP <1132> notes that final products typically range from less than 1 to 100 ng/mg. “Good” depends on dose, route and patient population, not on a single threshold. A low total with a polysorbate-degrading lipase present is worse than a higher total of inert proteins. Justify your criterion from your own clearance data.

Q: Is 100 ppm a regulatory limit?
A: No. No ICH, FDA or EMA guidance sets a numeric HCP limit. ICH Q6B says acceptance criteria should be justified from preclinical, clinical, consistency and stability lot data. The 2025 BioPhorum industry review calls 100 ng/mg “a standard rule of thumb across the industry, based on the level of currently approved products” and states that there is no official maximum threshold. Treat it as an orientation point, never as a specification you can cite.

Q: Why is my LC-MS HCP total higher than my ELISA result?
A: Because the two assays measure different quantities. ELISA reports immunoreactive mass against a polyclonal antibody that covers only part of your HCP population, calibrated against a null-cell-line standard. LC-MS reports the summed mass of every protein it identified and quantified. A more sensitive LC-MS method finds more proteins and therefore reports a larger total. The direction of the difference is not predictable in either sense.

Q: Does HCP ppm mean ng per mg or ng per mL? And has USP banned the word ppm?
A: Nanograms per milligram, and no. The denominator is a mass of product protein, not a volume, and the concentration form, ng/mL of HCP divided by mg/mL of product, gives the same answer because the volumes cancel. On the word itself the two chapters differ. USP <1132> says ppm is not advised, because ppm means mass per unit volume in other pharmacopeial tests. USP <1132.1> lists parts per million in its Terminology section as a common synonym for ng/mg. Write ng/mg; ppm is not an error.

Q: Should I use A280 or amino acid analysis for the denominator?
A: Use A280 with the product-specific extinction coefficient for routine work, and qualify it once against amino acid analysis. USP <1132> recommends using two orthogonal methods to exclude a gross over- or underestimation. Whatever you choose, state it in the report. A 15% difference in the denominator is a 15% difference in the reported ppm, and it is invisible in the final number.

Q: How do I report HCPs that are detected but below the limit of quantification, or not detected at all?
A: Pick one convention, write it into the SOP, and state it in every report. For sub-LOQ identifications the three in use are: exclude them from the total, include them at their reported value, or impute them at the LOQ. Excluding them gives the lowest total and the cleanest defense, and reporting them separately as “identified, below LOQ” with a count is better than burying the choice. For an HCP that was not detected at all, USP <1132.1> Section 6 describes reporting it as less than the level the system has been shown sensitive to, less than 10 ng/mg in its example, with that level established as the lowest detectable at least 90% of the time.

Q: Which USP <1132.1> quantitation method should I choose?
A: Section 5.1.2, relative to spiked-in proteins and often called Method B, is the usual default for total HCP because the standard experiences the same digestion as the analytes. Section 5.1.3, using stable isotope labeled peptides, is the most accurate for named high-risk HCPs and is the right choice for a release specification. Section 5.1.1 needs no spike and suits early development. Chrone et al. (2025) found all three linear and precise. The A, B and C labels are the qualification literature’s shorthand, not USP’s.

Q: Why did two contract labs give different HCP ppm for the same vial?
A: Partly method, and partly the technique’s floor. USP <1132.1> reports a case study of 18 independent analyses of identical aliquots of one monoclonal antibody, across different instruments, labs and analysts, in which the most abundant HCP ranged from 52 to 104 ng/mg and every result sat within a factor of two of the mean. Below a two-fold ratio you are looking at normal variability. Above it, check seven things: the denominator assay, the quantitation method and standard, the trypsin lot, the FASTA version, the limit of quantification, the shared-peptide rule and the sub-LOQ convention.

Run the calculation yourself

SpotMap MS is TotalLab’s software for LC-MS host cell protein analysis. It processes DIA data against your own FASTA file, automates the identification and quantitation steps described above, and pairs with AuditSafe for 21 CFR Part 11 audit trails and electronic signatures. Read about USP <1132.1> alignment, or start a free trial and produce the number yourself.

<1132.1> alignment

Request a free SpotMap MS trial

SpotMap MS product page

References

  1. USP General Chapter <1132> “Residual Host Cell Protein Measurement in Biopharmaceuticals”, United States Pharmacopeial Convention. USP 39 and NF 34, official 1 May 2016. Free full-text PDF posted by USP. Note that this free copy is the 2016 version and that no Notice of Intent to Revise has been published for <1132> since it became official. https://www.usp.org/sites/default/files/usp/document/our-work/biologics/USPNF810G-GC-1132-2017-01.pdf
  2. “USP Announces Corrections to General Chapter <1132.1> on Residual Host Cell Protein Measurement”, ECA Academy / GMP-Compliance.org, 2025. https://www.gmp-compliance.org/gmp-news/usp-announces-corrections-to-general-chapter-1132-1-on-residual-host-cell-protein-measurement
  3. “General Chapter <1132.1> Residual Host Cell Protein Measurement in Biopharmaceuticals by Liquid Chromatography-Mass Spectrometry”, Notice of Intent to Revise, USP-NF, 2024. https://www.uspnf.com/notices/gc-1132-1-nitr-20241101
  4. USP General Chapter <1132.1> “Residual Host Cell Protein Measurement in Biopharmaceuticals by Liquid Chromatography-Mass Spectrometry”, United States Pharmacopeial Convention. USP-NF, official 1 May 2025. DOI 10.31003/USPNF_M17756_03_01. Subscription access. Sections used on this page: 2 (Terminology, for ng/mg and its synonyms), 5.1 (quantitation methods 5.1.1 to 5.1.3), 6 (Reporting Results, for the accuracy statement, the 18-analysis case study and the reporting convention for undetected HCPs) and 7.2 (Units of Measurement, for the molecular weight caveat). https://doi.usp.org/USPNF/USPNF_M17756_02_01.html
  5. “Guidance for Industry: Q6B Specifications: Test Procedures and Acceptance Criteria for Biotechnological/Biological Products”, FDA / ICH, 1999. https://www.fda.gov/media/71510/download
  6. “ICH Q6B Specifications: test procedures and acceptance criteria for biotechnological/biological products” (CPMP/ICH/365/96), European Medicines Agency, 1999. https://www.ema.europa.eu/en/ich-q6b-specifications-test-procedures-acceptance-criteria-biotechnological-biological-products-scientific-guideline
  7. Silva JC, Gorenstein MV, Li GZ, Vissers JPC, Geromanos SJ. “Absolute Quantification of Proteins by LCMSE: A Virtue of Parallel MS Acquisition.” Molecular & Cellular Proteomics, 2006, 5(1):144-156. DOI 10.1074/mcp.M500230-MCP200, PMID 16219938. https://doi.org/10.1074/mcp.M500230-MCP200
  8. Schwanhäusser B, Busse D, Li N, Dittmar G, Schuchhardt J, Wolf J, Chen W, Selbach M. “Global quantification of mammalian gene expression control.” Nature, 2011, 473:337-342. DOI 10.1038/nature10098, PMID 21593866. https://www.nature.com/articles/nature10098
  9. Nesvizhskii AI, Aebersold R. “Interpretation of Shotgun Proteomic Data: The Protein Inference Problem.” Molecular & Cellular Proteomics, 2005, 4(10):1419-1440. DOI 10.1074/mcp.R500012-MCP200, PMID 16009968. https://doi.org/10.1074/mcp.R500012-MCP200
  10. Chrone VG, Blaszczyk AJ, Zhang DH, Nielsen SB, Crawford J, Schwämmle V, Højrup P, Kofoed T, Peckham N, Mørtz E. “Host cell protein quantitation by LC-MS. Experimental demonstration, qualification, and comparison of methods in USP 1132.1.” Journal of Pharmaceutical and Biomedical Analysis, 2025, 265:117051. DOI 10.1016/j.jpba.2025.117051, PMID 40639094. https://doi.org/10.1016/j.jpba.2025.117051
  11. Coye L, Jones MA, Gaza-Bulseco G, et al. “Host Cell Protein Clinical Safety Risk Assessment: An Updated Industry Review.” Biotechnology and Bioengineering, 2025. DOI 10.1002/bit.70029, PMID 40757641. https://doi.org/10.1002/bit.70029 . Quoted passages retrieved from the summary at https://www.cygnustechnologies.com/industry-review-and-recommendations-for-host-cell-protein-risk-assessments
  12. Walmsley SJ, Rudnick PA, Liang Y, Dong Q, Stein SE, Nesvizhskii AI. “Comprehensive Analysis of Protein Digestion Using Six Trypsins Reveals the Origin of Trypsin As a Significant Source of Variability in Proteomics.” Journal of Proteome Research, 2013, 12:5666-5680. DOI 10.1021/pr400611h. https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=914042
  13. Khalil S, Dierick JF, Bourguignon P, Plisnier M. “Prospective ICH Q2(R2)-Aligned Total-Error Validation of Label-Free Untargeted Proteomics for Host Cell Protein Quantification in Biotherapeutics.” Proteomes, 2026, 14(2):21. DOI 10.3390/proteomes14020021. https://www.mdpi.com/2227-7382/14/2/21
  14. “Comprehensive Benchmarking of Data-Independent Acquisition for Host Cell Protein Quantification on the Orbitrap Astral Mass Spectrometer”, bioRxiv preprint, 2025. DOI 10.1101/2025.07.31.667876. https://www.biorxiv.org/content/10.1101/2025.07.31.667876v1.full
  15. Yao Y, Wen X, Pan H, Chen Z. “Residual host cell proteins: sources, properties, detection methods and data acquisition modes.” Frontiers in Microbiology, 2025, 16:1658366. DOI 10.3389/fmicb.2025.1658366. https://www.frontiersin.org/journals/microbiology/articles/10.3389/fmicb.2025.1658366/full
  16. “Monoclonal Antibody (mAb)-related USP-NF Standards and Analytical Reference Materials”, United States Pharmacopeial Convention, 7 October 2025. https://www.usp.org/sites/default/files/usp-bio_monoclonalantibodies-standards-materials_07oct2025.pdf
  17. “USP-NF Publication Schedule”, United States Pharmacopeial Convention. Confirms USP-NF 2025 Issue 1 publication date 1 November 2024 and official date 1 May 2025. https://www.uspnf.com/print/pdf/node/16296