Using mass spectrometry for robust identity release testing in a global quality control network

ABSTRACT

Product identity (ID) testing is a fundamental requirement in pharmaceutical quality control. Roche/Genentech has developed and established a mass spectrometry-based platform ID method for routine release testing of biologics in the global clinical and commercial quality control environment. The established method, using peptide mapping combined with liquid chromatography-mass spectrometry, allows automated data evaluation. Previous experience shows that this approach is highly user-friendly in regulated testing environments, significantly reduces validation work for new products, and, due to its high robustness, results in a reduced error rate during test execution and data evaluation. Due to the specific analysis of product peptides, the method also allows for a sensitive and high-resolution evaluation of further product variants that are attributable to chemical and/or post-translational amino acid modifications. The data already recorded for identity testing purposes can therefore be immediately reevaluated during a quantitative product quality impact assessment following unexpected events in the manufacturing process, thus saving additional analyses during root cause analysis and future failure mitigation. In summary, this technology, established and qualified at all Roche/Genentech quality control sites, enables improved product monitoring beyond the routine release and stability testing methods used.

Introduction

At Roche/Genentech locations for technical product development, matrix-assisted laser desorption/ionization mass spectrometry (MALDI-MS) and electrospray ionization mass spectrometry (ESI-MS) are routinely used for the detailed characterization and structural analysis of therapeutic antibodies and other protein formats. ESI-MS is used either for direct analysis of the intact molecule, or in combination with liquid chromatography (LC-ESI-MS) to verify the primary structure of the target protein and to assess molecular integrity (e.g., aggregation and fragmentation), as well as chemical and post-translational amino acid modifications such as N-glycosylation, oxidation, deamidation, isomerization, and glycation events.^1–7 MS is regularly used for analytical extended characterization studies in non-good manufacturing practice (GMP)-regulated laboratory environments to support development studies (e.g., process comparability, formulation and bio-process development studies) and for a wide range of analytical troubleshooting activities. The use of MS in regulated quality control (QC) environments for release and stability testing has recently been debated.^8–14 In principle, the questions are: 1) whether a switch from electrophoretic and chromatographic test methods to an MS-based approach (by multi-attribute monitoring; MAM) for complex protein formats offers more advantages than disadvantages, and 2) whether using these technologies in a regulated GMP QC environment can be implemented. The biggest challenges have been the establishment of quantitative product release and stability criteria and practical system suitability criteria for determining product charge and size variants.¹⁵

Typically, the result of an identity assay in release testing is reported as “positive identity,”; therefore, the validation activities of an identity assay focus on the description of robustness and sufficient specificity, while the assessment of quantitative aspects such as accuracy, precision, linearity, and range are not mandatory.^16–18 As such, the use of MS for identity testing instead of other biochemical and biophysical methods with lower resolution represents a contemporary adaptation of the control strategy for product release testing. It should be mentioned here that, in addition to potency testing, other innovative technologies for identity testing, such as lateral flow immunoassay and Raman spectroscopy, are also being evaluated and used in the biopharmaceutical industry.^19–21

Implementation of MS-based identity testing is the product of gradual shifts in strategy and updates in technology over the past several decades. More than 10 years ago, due to the rapidly growing portfolio of mainly antibody products, specificity-limited identity testing methods such as capillary zone electrophoresis were replaced by higher-resolution peptide mapping-based methods for both clinical and commercial products. To simplify establishment in the global Roche/Genentech QC network, platform peptide mapping with LC-UV detection was initially used for late-phase clinical assets and commercial products. Here, Endoproteinase Lys-C was chosen as the protease to simplify the resulting peptide pattern for visual identity determination (compared to tryptic digestion). In parallel, a method based on MS was developed for early-stage clinical projects and used for release testing. In 2009, identity testing by tryptic peptide mapping combined with MALDI-TOF MS with superior specificity was deployed for release testing of all early-stage clinical development products in the Roche/Genentech network.^22,23 Due to the good experience with LC-UV-Lys-C peptide mapping in the commercial environment and the application of MS in the clinical area, the Roche/Genentech Pharma Technical Operations network recently decided to also use MS for commercial identity release testing of therapeutic proteins. Here, we summarize and discuss the test principle, the validation results, and the status of the establishment of the newly developed platform method in the global Roche/Genentech QC network.

Method development and validation

A platform peptide mapping approach with Lys-C digestion combined with LC-UV detection was initially used for the release identity testing of late phases and commercial products. This approach is based on a visual comparison of a reference standard (RS) with a release sample, whereby product-specific peptides (ID markers) must be visually clearly detected in both the chromatogram of the RS and the sample (see Supplementary Material 1 for test principle). Although this test principle has been successfully validated and has been used successfully in the Roche/Genentech QC laboratories for years, performance problems have been identified, which make this assay maintenance-intensive. The obtained UV chromatographic patterns are often complex and show significant variability, such as retention time shifts and differences in resolution (peak splitting/co-elution) by using different analytical conditions (e.g., different column lots or HPLC systems). Moreover, interpretation of the complex peptide pattern requires adequate staff training and profound experience of the lab analysts for consistent data evaluation, but is still open to subjective data interpretation. Thus, health authorities often have questions about the visual comparison approach and request the introduction of additional quantitative method parameters such as retention time windows or specific intensity thresholds. Another disadvantage of the visual LC-UV approach is that the ID markers must be redetermined for each new product, as the specificity compared to the product portfolio must be reverified in validation (by comparing all product peptide profiles within one analysis sequence). Therefore, in 2021 the Roche/Genentech Pharma QC Network decided to discontinue the visual LC-UV approach and switch to an LC-MS detection system (see Figure 1). Briefly, such an MS system should be straightforward to use in the QC environment and easily added to already established HPLC and software systems while meeting all data integrity requirements. After a thorough evaluation, the Waters ACQUITY QDa system, which is equipped with a single quadrupole mass detection, was selected. The suitability of this system for both identity testing and quantitative MAM applications has already been proven in two independent studies.^12,24

Figure 1.

Switching from release identity testing using LC-UV peptide mapping to automated mass spectrometry (MS)-based evaluation of specific product peptides.

Testing strategy and assay setup

The aim was to develop an identity test that is not based on visual comparison, but rather via accurate mass determination of Lys-C product peptides and automated data evaluation using a peptide library database (Table 1). Therefore, a Lys-C peptide library database was initially established with all relevant clinical and commercial protein products based on experimental product data. First, product-specific peptide masses are determined via bioinformatic in silico Lys-C digestion of the target protein and a comparison with the overall product portfolio (currently containing sequences for 120 biologics from clinical and marketed products). The strategy and configuration of in-house developed software for identifying these product specifics in the regulated environment will be further described in the future. Whether these theoretically determined Lys-C digested peptides of the target protein can be detected with sufficient intensity in the experimental data set is then assessed. The peptide masses detected experimentally using the QDa system (MS1 detection only) are subsequently sequenced using a high-resolution LC-MS system (with MS/MS functionality) with the same UHPLC system and chromatographic conditions as the QDa system, thus confirming the identity of the selected peptide masses. To subsequently validate the specificity of the selected Lys-C peptides, it is verified that they are truly product-specific and cannot be detected in the recorded data sets of the overall product portfolio. Currently, a minimum of 2 product-specific peptides (ID peaks) is required to continue with the assay development.

Table 1.

Summary of the most important features of the mass spectrometry-based identity testing method.

Test principle	Targeted identification with a minimum of 2 product-specific marker (ID) peptides using liquid chromatography – mass spectrometry
Digestion Enzyme	Endoproteinase Lys-C (Interchangeable with trypsin)
LC system	Waters ACQUITY UHPLC
Mass detector	Waters QDa single quadrupole
Automated Data Evaluation	Waters EMPOWER® software using pre-defined Custom Fields
Basis for System Suitability and Acceptance Criteria	No visual data assessment only numeric parameters for intensity, retention time, and mass accuracy
Validation status	Generically validated for robustness parameters. Only specificity needs re-validated for new products using a peptide library database
Roche/Genentech Global Quality Control Network	Established at all clinical and commercial QC testing sites (10 in total) and two Roche in-country testing laboratories. Currently, 15 LC-MS (QDa) systems are established in the Roche/Genentech QC network and 2 more in in-country testing laboratories
Filing status (beginning of 2026)	All clinical and 2 approved commercial products. Post-Approval Change Management Protocol submitted in January 2026 to switch all commercial products to new platform method

The method setup and execution are summarized in Figure 2. The protein digestion and subsequent LC-MS analysis are performed according to the procedure described in the Materials and Methods section. To allow for simple and automated data evaluation, the respective extracted ion chromatograms (XICs) are generated from the recorded 3D MS-spectra (based on intensity, retention time, and m/z value) focused on product-specific ID peaks using the respective peptide specific m/z value(s). The XIC of each ID peptide consists of a 2D-chromatogram (based on intensity and retention time at the expected m/z value). The resulting mass spectra for the selected ID peaks are further processed by predefined evaluation criteria, which were established during initial method development. The determined evaluation criteria values for the QDa system are based on the experience gained in method development and lead to the almost complete exclusion of false positive signals in validations and routine measurements. Details of the applied evaluation filters for retention times, minimum integration threshold, minimum intensity requirements, and mass accuracy are summarized in the Material and Methods section.

Figure 2.

Overview of the test procedure with automated data analysis. AC; acceptance criteria, ID; identity, RT; retention time, SST; system suitability test criteria.

Automated data evaluation

The aforementioned data evaluation and reporting for product-specific ID peptide peaks are performed automatically by the Waters EMPOWER® software using pre-defined Custom Fields. Custom Fields are a combination of input fields and calculations that allow the user to input minimal data and receive an output of a calculated value utilizing EMPOWER® database field data. The pre-defined Custom Fields allow for automated XIC evaluation of signal intensity, retention time, and mass difference for the product-specific ID marker peaks. The final EMPOWER® report method summarizes all the method and sample information, as well as the obtained results and evaluation of these results, and allows a GMP-compliant review and electronic signature of the analyzed data. The final report summarizes the previously described assessment in a simple form and confirms the identity of the product with confirmation of the product-specific ID marker peptides (“pass”). A summary of the required custom fields and their configuration can be found in Supplementary Material 2.

Summary of validation data and implementation status in the global Roche/Genentech QC network

The described LC-MS test method was generically validated based on the International Conference on Harmonization (ICH) guideline Q2 (R1). For this purpose, the test parameters of specificity and robustness were independently examined in detail for three therapeutic protein products. The specificity was proven if the selected (at least 2) ID markers are only detectable in the sample of the target protein. To date, sufficient ID markers have been theoretically determined and experimentally proven for each product under development. The robustness of the test system was examined in detail for parameters related to sample preparation, chromatography, and mass detection, and the validated ranges are summarized in Supplementary Material 3. It should be mentioned that for all parameters/ranges examined, only a change in the signal intensity of ID marker peptides was observed and the established acceptance criteria for signal intensity, retention time, and mass accuracy were always met. The robustness of the method is expected to extend far beyond the validation range examined. In addition to the demonstrated robustness of the method, its test principle of targeted mass detection makes it far less susceptible to interference in routine operation than the previously used LC-UV method with visual data evaluation (Figure 3). As such, the Waters ACQUITY QDa system is now established and GMP-qualified throughout the Roche/Genentech QC Network in the United States, Europe, Asia, and has been implemented at the Roche in-country testing laboratories in Korea and Argentina (Table 1). It is routinely used for identity release measurements for both clinical and commercial protein products. For regulatory submission, the developed test system is the standard identity assay for clinical protein products and has already been successfully approved for two new marketed products, Ocrevus (Ocrelizumab) and Piasky (Crovalimab), with further submissions already being planned. Currently, a Post-Approval Change Management Protocol (PACMP) was prepared to switch all commercial products to this new platform LC-MS identity method. The PACMP has been submitted globally to the health authorities in January 2026.

Figure 3.

Summary of the advantages of a mass spectrometry (MS)-based peptide mapping identity testing method compared to the previously used UV detection.

Conclusions

The application of LC-MS peptide mapping for product identity testing represents a contemporary adaptation of the previously established test strategy in the highly regulated pharmaceutical environment. This method has made it possible to mitigate application and robustness problems of the established LC-systems with UV detection and significantly increases the specificity for identity testing of protein drugs. The test system described here was relatively easy and inexpensive to establish in the global Roche/Genentech QC network because it could be supplemented with existing UHPLC systems, and the overall concept with automated data evaluation allows user-friendly application without requiring laboratory staff to have extensive mass spectrometric expertise. The experience that is already available suggests that the laboratory error rate in identity testing can be reduced, and time-consuming deviations and error investigations can be minimized. In addition, the data recorded in the release testing can be used for further evaluations, as they allow long-term evaluation of bioprocess key performance parameters that are not fully resolved in routine release testing. For example, amino acid modifications such as oxidation and deamidation events can be tracked and the data can be used to evaluate long-term process trends, as well as current bio-process problems (deviation management including root cause analysis). Moreover, the data we have recorded so far for such peptide modifications are quantitatively comparable to those from high-resolution LC-MS/MS technology, which is in line with previously published studies on the use of LC-MS technology with QDa Detector.^12,24 A recently published study has proven that the LC-MS QDa technology is also suitable for quantitative release testing. It was shown that quantitative assessment of charge variants is possible with this technology and can be used instead of the routine ion-exchange chromatography for release testing at the drug product level.²⁵ Even without further optimization of the described test system, we were able to detect and quantify relevant quality attributes such as N-glycosylation and oxidations in the conserved Fc portion of the heavy antibody chain at the peptide level (refer to Supplementary Material 4). The results obtained are comparable to those from reference methods such as HILIC-UHPLC analysis of 2-AB labeled N-glycans. Therefore, the LC-MS QDa technology, which is now established throughout the Roche/Genentech QC network, represents an alternative option for the specific assessment of product attributes. In principle, these could be included in the evaluation scheme for identity testing (using pre-defined custom fields). Such an approach would have to be successfully validated for quantitative aspects according to ICH guideline Q2(R1), but this also applies to the currently used routine methods with lower resolution (e.g., IEC), as well as to modern high-resolution LC-MS/MS technologies.^26,27

Platform requirements and methods

Reagents and materials

Biopharmaceutical proteins for assay development were supplied by Roche (Basel, Switzerland) and Genentech (South San Francisco, USA). Guanidine hydrochloride (Gua-HCl), Tris-(hydroxymethyl)aminomethane (Tris base), Tris-hydrochloride, Dithiothreitol (DTT), Iodoacetic acid (IAA) and Urea were purchased from Sigma Aldrich. Endoproteinase Lys-C (from lysobacter sequencing grade) and recombinant Endoproteinase Lys-C were purchased from Roche. Trifluoroacetic acid (TFA) and Formic Acid (FA) were obtained from Thermo Fisher Scientific. Purified water was used from a Milli-Q Advantage A10. PD-10 columns (Sephadex® G-25 medium) and NAP-5 desalting column (Sephadex® G-25 medium) were purchased from Cytiva (Wilmington DE, USA).

Sample preparation

For sample preparation, the biopharmaceutical solutions were diluted with purified water. One-milligram diluted protein was denatured by adding denaturation buffer containing 6 M Guanidine hydrochloride, 360 mM Tris and 2 mM Ethylenediamine tetraacetic acid (EDTA). The following subsequent reduction was performed by adding 220 mM DTT and incubating for 10 min at 37°C. Afterward, alkylation was achieved by the addition of 300 mM Iodoacetic Acid (IAA) and incubation for 10 min at 37°C. The alkylated samples were desalted by previously equilibrated PD-10 or NAP-5 desalting columns using a 25 mM Tris, 1 M Urea and 1 mM EDTA buffer.

The enzymatic digestion was performed by the addition of an Endoproteinase Lys-C solution (0.1 mg/mL in water) in a 1/100 enzyme–substrate ratio (mass/mass), followed by incubation for 2 h at 45°C. The digestion was stopped by adding a 10% TFA solution.

LC-MS measurement

RP-UHPLC was carried out using a Waters ACQUITY UHPLC system (H-Class or I-Class) equipped with a Waters QDa I detector. The temperature of the analytical column (Waters ACQUITY UHPLC Peptide CSH C18 Column, 130Å, 1.7 μm, 2.1 X 150 mm) was set to 80°C and a 0.3 mL/min flow rate was used. Depending on the UHPLC system used, different gradients were used to obtain the same separation result.

Using the Waters ACQUITY UHPLC systems (high pressure mixing binary pump), the enzymatic digested protein solutions were analyzed with a binary gradient of mobile phase A (0.1% TFA in water) and mobile phase B (0.08% TFA in acetonitrile). The starting conditions were 0.5% mobile phase B for 0.4 minutes followed by a gradient to 8% mobile phase B within 4.4 minutes, then a gradient to 32% mobile phase B within 18.4 minutes and finally, an increase to 100% mobile phase B within 3.2 minutes to elute the most hydrophobic peptides. The total peptide separation time was 26.4 minutes. A column washing is used to clean the column by injecting an Isopropanol/Acetonitrile mixture (80/20 v/v) and a gradient corresponding to 2 successive increases of mobile phase B from 0.5 to 100% within 1 minute and back to initial conditions within 0.1 minutes. The column is then re-equilibrated prior to the subsequent injection.

Using a Waters ACQUITY H-Class (low pressure mixing quaternary pump), a ternary gradient system was used, using a third mobile phase which is a premix of 90% mobile phase A and 10% mobile phase B. The separation and washing gradients were adapted accordingly.

The MS detection was performed using a connected Waters ACQUITY QDa I Performance mass spectrometer in positive ion mode. The cone voltage was set to 10 V, capillary voltage to 1.5 kV, probe temperature to 600°C, mass range to 150–1250 m/z and the sample rate to ≥2 points per second. The 3D MS-spectra were acquired using the EMPOWER® chromatographic data system in run-only modus. Post-run data processing was performed using product-specific processing method sets.

Data evaluation

For simple and automatized data evaluation, respective XICs were generated from the 3D MS-spectra (Intensity vs Time vs m/z) focused on each ID peptide using the respective peptide specific m/z value(s). The XIC of each ID peptide consists of a 2D-chromatogram (Intensity vs Time at the expected m/z values) for all ID peptides of the analyzed protein. To generate the XICs using the Waters EMPOWER® software, the m/z values used for the extraction were defined as “Derived Channels” in the Method Set with a “Peak Separation” value set to 1.2000 Da, providing a mass tolerance of 0.6 Da for data collection. A smoothing by 19 points using Savitzky-Golay filter of the extracted chromatograms was applied to obtain a consistent peak shape. Each ID peak is integrated in the respective MS channel, based on peak intensity. In each XIC, the peak with the highest intensity is integrated and identified as the respective ID peak. To avoid integration of low abundant and nonspecific peaks, the peak needs to be higher than the predefined minimum height value, usually 500,000 ion counts, to be automatically integrated by the EMPOWER® software using peak integration threshold in ion counts. Once integrated, the retention time of the highest peak in the XIC is automatically measured and displayed by the software. If the expected ID peak is absent, no integration occurs and no retention time is displayed, resulting automatically in a negative identity result.

By using Waters ACQUITY QDa, the EMPOWER® processing method returns to the 2D-MS spectra of the integrated ID peak in the XIC and gets the highest m/z value found within the tolerance window of ±0.6 Da for each theoretical average m/z value. Peak Separation parameter is set of 1.20 Da in the EMPOWER® processing method using “MS Expected Mass” window. Once integrated, the 2D-MS spectrum (m/z vs Intensity at a given retention time) corresponding to each ID peak is checked for the presence of the expected m/z. The mass error represents the absolute value of the difference between the expected m/z and the measured m/z for each respective ID peak. The Mass Error is calculated for each expected m/z value using the following formula:

Mass error = Measured m/z – Expected m/z.

To avoid considering noise signal in the 2D-MS spectra, the “expected intensity” (%intensity) of the found m/z value must be greater than or equal to 0.2% of the most intense signal (base peak) present in the spectrum. The “expected intensity” is calculated according to the following formula:

Expected Intensity = Intensity of measured m/z/Intensity of most intense signal x100 ≥ 0.2%.

The “Expected Intensity” value of 0.2% is set in the EMPOWER® processing method for each “Expected Mass” using the “MS Expected Mass” window.

Characterization of ID peak candidates by high-resolution mass spectrometry

For the implementation of the generic peptide mapping method using MS-detection by QDa as an identity method, the selected ID peak candidates were sequenced by high-resolution LC-MS/MS. For the characterization of eluting product-specific ID peaks, the LysC digested antibody was separated under the same conditions as described in LC-MS Measurement section above. Mass spectrometric detection and CID fragmentation of ID marker peptides were performed on a coupled Xevo G2-S mass spectrometer from Waters in the positive ion mode. The source was set for ESI + and a capillary voltage of 3000 V was applied. Data was collected using a scan range from 100 to 2000 m/z in continuum mode. The lock spray reference for accurate mass was phosphoric acid using the masses in the range 100–2000 m/z for calibration. Parameters for MS and MS/MS detection were adjusted according to existing knowledge gained from experience with peptide analysis of recombinant antibodies. The data evaluation was performed by using the vendor-specific software (Waters).

List of abbreviation

2-AB

2-aminobenzamide

AC

acceptance criteria

ESI-MS

electrospray ionization mass spectrometry

GMP

good manufacturing practice

HILIC

hydrophilic interaction liquid chromatography

ICH

international conference on harmonization

ID

identity

IEC

ion exchange chromatography

LC

liquid chromatography

MALDI-MS

matrix-assisted laser desorption/ionization mass spectrometry

MS

mass spectrometry

QC

quality control

RS

reference standard

RT

Retention time

SST

System suitability test

UHPLC

ultra high performance liquid chromatography

UV

ultraviolet

XIC

extracted ion chromatogram

Funding Statement

The author(s) reported there is no funding associated with the work featured in this article.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/19420862.2026.2643964

Disclosure statement

No potential conflict of interest was reported by the author(s).