A Bottom-Up Liquid Chromatography–Tandem Mass Spectrometry Method for Therapeutic Drug Monitoring of Infliximab: Method Development, Comparison With 2 Enzyme-Linked Immunosorbent Assay Methods, and Evaluation of Anti-Drug Antibody Interference Open Access

Ammonium bicarbonate, ammonium sulfate, 2,2,2-trifluoroethanol, formic acid, and tosyl phenylalanyl chloromethyl ketone–treated trypsin were obtained from Sigma-Aldrich (St Louis, Missouri). LC-MS/MS–grade acetonitrile, distilled water, and methanol were purchased from Burdick and Jackson Inc (Muskegon, Michigan). Optima isopropyl alcohol LC-MS grade was obtained from Fisher Scientific. Lyophilized infliximab (Remicade, Janssen Biotech, Inc) was obtained from the drug manufacturer. The light-chain peptide YASESMSGIPSR was selected for quantitation. As an IS to compensate for variability due to extraction recovery and digestion, winged SIL peptides (TNGSPRLLK-YASESMSGIPSR[13C6,15N4]-FSGSG) were synthesized from AnyGen Co, Ltd (Gwangju, Korea). Infliximab-negative pooled human serum was used as the diluent in preparation of controls and calibrators.

Deionized water was added to lyophilized infliximab to obtain a 10 mg/mL stock solution. This stock solution was serially diluted with deionized water to make working solutions of 6 concentrations (0.05, 0.25, 0.50, 1.00, 2.50, and 5.00 mg/mL). Calibrators and controls were prepared by diluting working solutions with drug-free pooled human serum: 6 calibrators (0.5, 2.5, 5.0, 10.0, 25.0, and 50.0 μg/mL) and 3 controls (3, 15, and 30 μg/mL). Thirty percent acetonitrile was added to winged SIL peptide to obtain a 10 mg/mL IS stock solution. This IS stock solution was diluted with 50 mM ammonium bicarbonate to make a 4 μg/mL IS working solution. Stock solutions, working solutions, calibrators, and controls were aliquoted and stored at −80°C.

Sample preparation was based on a bottom-up MS-based proteomic approach, where the target molecule was enzymatically digested and the light-chain peptide YASESMSGIPSR was selected for identification. One hundred microliters of calibrators, quality controls, and patient serum specimens were protein precipitated with 100 μL of saturated ammonium sulfate solution (4.32 M). The vortexed mixtures were then left at room temperature for 30 minutes and centrifuged (3000g) for 20 minutes. After removal of the supernatant, the retained protein pellet was resuspended with 100 μL of 50 mM ammonium bicarbonate. The sample was denatured with 75 μL of 2,2,2-trifluoroethanol, and 10 μL of 4 μg/mL IS was added. Samples were then incubated at 55°C for 30 minutes with mixing (1500 rpm) and cooled to room temperature. Samples were trypsin digested by adding 400 μL of deionized water, 100 μL of 50 mM ammonium bicarbonate, and 50 μL of 2 mg/mL tosyl phenylalanyl chloromethyl ketone. Samples were subsequently vortexed and incubated for 1 hour at room temperature (37°C) with mixing (1500 rpm). After cooling, digestion was arrested by addition of 20 μL formic acid. After brief vortex mixing, samples were centrifuged for 10 minutes (1700g). Finally, 200 μL of samples were loaded on to the autosampler for LS-MS/MS analysis.

LC-MS/MS analysis was performed on a QTRAP 5500+ triple quadrupole mass spectrometer (Sciex) with an Agilent 1290 Infinity II HPLC system (Agilent Technologies). The analytical column was a Kinetex C18 HPLC column (2.1 × 50 mm, 2.6-µm particle size, Phenomenex). The mobile phases were distilled water containing 0.1% formic acid (mobile A) and 25:75 isopropyl alcohol to acetonitrile containing 0.1% formic acid (mobile B). The LC gradients in minutes per percentage of mobile phase B were 0.0 min/5%, 1.5 min/5%, 5.0 min/10%, 6.5 min/80%, 7.5 min/80%, and 7.6 min/5%. The flow rate was 0.35 mL/min and the run time was 5 minutes.

Quantitative analysis was performed in multiple reaction monitoring mode with a jet stream electrospray ionization source operating in positive-ion–detection mode. Multiple reaction monitoring transitions and chromatograms are provided in Table 1 and Figure 1, A through D, respectively. Source parameters were as follows: curtain gas, 30 psi; ion spray voltage, 5500 V; and source temperature, 650°C. Ion source gas 1 was set at 50 psi, and ion source gas 2 was set at 50 psi. Quantitation was performed using peak area ratios of light-chain peptide YASESMSGIPSR and IS with linear 1/x-weighted regression using Sciex OS-MQ program (version 1.7, Sciex).

Method validation was performed according to the Clinical and Laboratory Standards Institute guideline C62-A with respect to linearity, limit of detection (LOD), lower limit of measuring interval (LLMI), precision, accuracy, selectivity, carryover, recovery, and matrix effect. Calibration curves were plotted using a linear regression analysis with a 1/x weighing factor. Linearity was assumed when the measured concentration of each analyte was within 15% of the nominal value, except for the LLMI, for which a 20% deviation was considered acceptable. LOD was assigned to the lowest concentration with a signal to noise ratio greater than 3:1 by analyzing 5 replicates, and LLMI was assigned to the lowest concentration with a signal to noise ratio greater than 10:1, coefficient of variation (CV) less than 20%, and bias less than 20% by analyzing 20 replicates. The intraday precision was determined at each concentration level with 5 runs on the same day, and the interday precision was assessed with 20 independent assay runs. Accuracy was evaluated using samples of 3 different infliximab concentrations (3.0, 15.0, and 30.0 μg/mL) prepared by diluting the World Health Organization international standard for infliximab (NIBSC code 16/170) of 50 μg/mL (500 IU/mL) into drug-free pooled human serum. A relative bias (percentage, calculated by dividing the difference between the measured value and the expected value by the expected value) less than or equal to 15% was considered acceptable. Carryover was estimated by consecutive injection of blank matrix samples immediately after running the highest concentration calibrator. Responses of analytes in blank samples less than 20% of those in the LLMI sample were considered acceptable. Matrix effect was evaluated by postcolumn infusion experiment.^25,26 

Method comparison and evaluation of ADA interference were conducted using residual samples obtained as part of routine TDM of infliximab in IBD patients from November 2017 to August 2022. Blood was collected in plain tubes just before the next infliximab infusion. Clinical information of age, sex, body weight, and blood sampling time was collected from patient medical records. The study and the waiver of informed consent were approved by the Samsung Medical Center Institutional Review Board (IRB; SMC 2021-01-131).

Method comparison was performed against 2 ELISA methods, Remsima Monitor Drug Level (Immunodiagnostik AG, Bensheim, Germany) and IDKmonitor Infliximab Drug Level (Immunodiagnostik AG). Both ELISAs quantitated the free (unbound) form of infliximab and its biosimilar (Remsima or Remaloce). To mitigate the impact of ADA interference on the method comparison analysis, the same analysis was also conducted on ADA-negative samples. The LLMI was 0.003 µg/mL for both the Remsima Monitor Drug Level and the IDKmonitor Infliximab Drug Level.

Within the sample set used for the method comparison, ADA was quantitated using Remsima Monitor Total ADA for patients whose infliximab concentration was determined by Remsima Monitor Drug Level and using IDKmonitor Infliximab Free ADA for patients whose infliximab concentration was determined by IDKmonitor Infliximab Drug Level. As per the manufacturer’s instructions, the same positive threshold of 10 arbitrary units (AU)/mL, determined by linear dilution of sera with high infliximab concentrations, was applied to both assays.²⁷  The potential effect of ADA interferences was evaluated by comparing the serum infliximab concentrations measured by LC-MS/MS and ELISA methods according to the positivity of total ADA. All analyses were performed according to the manufacturers’ instructions.

After checking for normality, data showing a nonnormal distribution were reported as the median and range, and data showing a normal distribution were reported as mean with SD. Samples for which the concentrations measured were below the LLMI were excluded from the statistical analysis. Passing-Bablok regression analysis,²⁸  Spearman correlation analysis, and Bland-Altman mean difference plot analysis²⁹  were used to assess method agreement and the impact of ADA interference. Statistical analyses were performed using MedCalc version 77.5.1.0 (MedCalc Software Ltd, Ostend, Belgium), IBM SPSS Statistics version 25 (IBM, Armonk, New York), and R 4.0.3 (R Foundation for Statistical Computing, Vienna, Austria).

As shown in Figure 1, A through D, the elution time for infliximab was around 5 minutes, and no interfering peaks were observed. The total analytical run time was 5 minutes per sample, and total sample preparation time was 3 hours including a digestion step for a batch size of up to 50 specimens.

The results of precision, accuracy, LOD, and linearity tests are summarized in Table 2. The method exhibited a good linearity from 0.5 to 50.0 µg/mL (R² = 0.998). LOD and LLMI were 0.25 µg/mL and 0.5 µg/mL, respectively. Intraday and interday imprecision values were less than 6.1% CV. The accuracy ranged from 94.2% to 98.7%. No carryover effect was observed. The postcolumn infusion experiment demonstrated no apparent ion suppression or enhancement at the analyte and IS retention time (Figure 2, A and B).

Characteristics of patients included for method comparison are summarized in Table 3. A total of 348 clinical samples from 261 IBD patients (52 with ulcerative colitis, 207 with Crohn disease, and 2 with IBD unclassified) were analyzed using LC-MS/MS, and the results were compared against 2 ELISA methods, Remsima Monitor (217 samples from 130 patients) and IDKmonitor Infliximab (131 samples from 131 patients). All patients whose infliximab concentration was measured by IDKmonitor Infliximab were on maintenance therapy, with a median infliximab dose of 5.8 mg/kg (range, 0.0–11.4 mg/kg), whereas patients whose infliximab concentration was measured by Remsima Monitor were on maintenance therapy following a regimen of either 5 mg/kg or 10 mg/kg and included 11 patients on induction therapy. Patients in this study were treated with Remicade, Remsima, or Remaloce. The median infliximab concentration, as measured by LC-MS/MS, was 3.9 µg/mL, with a range of less than 0.5 to 39.6 µg/mL. The majority of concentrations were below 20 µg/mL, falling within the clinically relevant serum concentration range for patients receiving IV infliximab treatment.^20,30 

After exclusion of values below the LLMI, statistical analysis was performed on 197 samples for LC-MS/MS versus Remsima Monitor and on 120 samples for LC-MS/MS versus IDKmonitor Infliximab. Figure 3, A through D, illustrates Passing-Bablok and Bland-Altman plots for each comparison. A strong correlation was observed between LC-MS/MS and both ELISAs (with Remsima Monitor, correlation coefficient r = 0.94, Figure 3, A; with IDKmonitor Infliximab, r = 0.96, Figure 3, B). However, although a marked negative bias was observed between LC-MS/MS and IDKmonitor Infliximab (mean relative difference, −32.6%; 95% CI, −35.8% to −29.4%; Figure 3, D), no significant bias was observed between LC-MS/MS and Remsima Monitor (mean percentage difference, 5.7%; 95% CI, −1.2% to 12.6%; Figure 3, C). In samples with infliximab concentration measured by ELISA less than 20 µg/mL, a strong correlation was observed (with Remsima Monitor, correlation coefficient r = 0.92, Supplemental Figure 1, A [see Supplemental Digital Content, containing 2 figures, at https://meridian.allenpress.com/aplm in the May 2025 table of contents]; with IDKmonitor Infliximab, r = 0.95, Supplemental Figure 1, B). Although marked negative bias was observed between LC-MS/MS and IDKmonitor (mean relative difference, −32.2%; 95% CI, −35.5% to −28.9%; Supplemental Figure 1, D), a slight positive bias was observed between LC-MS/MS and Remsima Monitor (mean percentage difference, 9.3%; 95% CI, 2.4% to 16.3%; Supplemental Figure 1, C).

To mitigate the impact of ADA interference on the method comparison analysis, Passing-Bablok and Bland-Altman analyses were conducted on ADA-negative samples. and a similar tendency was observed (Remsima Monitor, n = 58; IDKmonitor Infliximab, n = 66; Supplemental Figure 2). A strong correlation was observed in both Remsima Monitor (n = 66, r = 0.95; Supplemental Figure 2, A) and IDKmonitor Infliximab (n = 58, r = 0.94; Supplemental Figure 2, B). However, although IDKmonitor Infliximab showed negative bias (mean relative difference, −27.4%; 95% CI, −32.3% to −22.5%; Supplemental Figure 2, D), Remsima Monitor showed good agreement with LC-MS/MS (mean relative difference, −3.5%; 95% CI, −12.8% to 66.2%; Supplemental Figure 2, C).

Of the total 348 clinical samples, ADA was measured in 237 samples using Remsima Monitor Total ADA in 164 samples and IDKmonitor Infliximab Free ADA in the other 73 samples, and 65% (106 of 164) were positive for total ADA.

In samples in which total ADA was quantitated, the relative difference in trough infliximab concentration between LC-MS/MS and ELISA increased with total ADA concentration, ranging between −93% and +200% (Figure 4, A, n = 146). Similarly, the relative difference increased with free ADA concentration, ranging between −200% and +74% (Figure 4, B, n = 72). Nevertheless, no significant correlation was observed between ADA concentration and relative difference (Pearson correlation coefficient r between relative difference and total ADA, 0.16; between relative difference and free ADA, −0.29).

A method comparison analysis was conducted between the infliximab concentration measured by LC-MS/MS and Remsima Monitor according to ADA positivity in samples in which the concentrations of infliximab and total ADA were measured by the Remsima Monitor Drug Level and Total ADA, respectively. Although good agreement was observed for ADA-negative samples (mean relative difference, −3.5%; 95% CI, −12.8% to 5.9%, Figure 5, A), a significant bias was observed for ADA-positive samples (13.6%; 95% CI, 1.7%–25.6%, Figure 5, B).

We developed a simple and cost-effective bottom-up LC-MS/MS method using winged SIL peptide for TDM of infliximab in human serum. The LLMI (0.5 μg/mL), analytical measurement interval (0.5–50 μg/mL), precision, accuracy, carryover, and ion suppression were validated, and its short analytical run time (5 minutes), simple sample preparation step, and cost-effectiveness made this method suitable for clinical practice.

Although both ELISAs correlated well with LC-MS/MS, IDKmonitor Infliximab tended to overestimate infliximab concentration, whereas Remsima Monitor did not show significant bias with LC-MS/MS in both comparisons using all ADA-negative samples (Figure 3; Supplemental Figure 2). This finding was intriguing because the concentration measured by ELISAs should theoretically be lower than that measured by our LC-MS/MS method, because our LC-MS/MS method quantitates total (bound and unbound) infliximab, whereas both ELISAs quantitate only free infliximab. Considering the acceptable accuracy test results using World Health Organization international standards during the validation of our LC-MS/MS method, these findings may indicate that ELISAs for quantitation of infliximab lack not only harmonization but also accuracy. Traditionally, the most commonly used techniques to measure infliximab are immunoassays such as ELISA and electrochemiluminescence-based immunoassays.¹⁶  Being commercially available, immunoassays can be easily implemented in laboratories. However, because of the poor concordance of infliximab concentration among immunoassays, the 2017 American Gastroenterology Association institute guideline on TDM in IBD suggests using the same assay in case of repeated infliximab concentration and ADA measurement in a patient.^4,14  Because immunoassays rely on different antibodies and epitopes, only harmonization is possible. However, LC-MS/MS can be standardized by calibrating to a reference material.¹⁸ 

A monoclonal antibody exists in free (unbound) form or in a complex with either the target antigen or ADAs in serum. The 2 main methods currently applied for infliximab measurement using mass spectrometry are the approach with an immunoaffinity purification step that measures free infliximab (infliximab with unbound paratope) and the approach without immunoaffinity purification that measures total (free and bound) infliximab.^20–24  Immunoaffinity purification using biotinylated TNF-α linked to a streptavidin-coated 96-well plate can selectively capture free infliximab, whereas our method measures total infliximab. For TDM of highly protein-bound small-molecule drugs, free drug concentration may be more reliable than total drug concentration.³¹  However, this concept could be inappropriate for monoclonal antibodies, as the bound form represents not only the inactive ones bound to ADA, but also those bound to target. Moreover, high concentrations of circulating target, often observed in patients with high disease burden, have been shown to interfere with quantitation of free monoclonal antibody quantitation and result in falsely low values.^31,32  Therefore, it may not be appropriate to interpret the free infliximab concentration as drug exposure, as is done with small-molecule drugs. Moreover, ADA concentration should be assessed along with total infliximab concentrations for accurate interpretation. Novel assays that can distinguish among free, target-bound, and ADA-bound infliximab may improve TDM of infliximab.^24,33 

ADA is known to cause unpredictable and variable interference in the measurement of infliximab using immunoassays, resulting in falsely high or low results.^9–11  There are 2 types of assays for the detection of ADAs: an assay that measures free ADA and an assay that measures total (free or bound) ADA with a pretreatment acid dissociation step to separate ADA from infliximab. In our study, 2 separate sample sets were evaluated, one in which free ADA (IDKmonitor Infliximab Free ADA) was measured and the other in which total ADA (Remsima Monitor Total ADA) was measured. Given that the relative difference in infliximab concentration between our LC-MS/MS method (total infliximab) and the 2 ELISAs (free infliximab) represents not only the methodologic variance but also the bound infliximab concentration, a significant correlation between the relative difference and total ADA concentration was expected. However, no significant correlation was observed between the relative difference and total ADA concentration. As per the manufacturer’s instructions, the same positive threshold of 10 AU/mL, determined by linear dilution of sera with high infliximab concentrations, was applied to both assays.²⁷  Although the positivity cutoff for total ADA has been validated in samples obtained from anti-TNF agent–naive healthy adults (positivity cutoff validated as 9 AU/mL), that for free ADA has not yet been validated.

Therapeutic monoclonal antibodies represent an emerging field in the pharmaceutical industry, evolving into a core therapeutic option for autoimmune disorders and malignancies.³⁴  However, the variability in clinical responses and the emergence of resistance or loss of response underscore the need for personalized treatment guided by TDM based on serum concentration measurement of monoclonal antibodies.^3,14,34  Although ELISA is commonly used, its limitations, such as lack of comparability across assays and susceptibility to interference, have led to the development of LC-MS/MS methods for monoclonal antibody quantitation in clinical laboratories. Mass spectrometry–based techniques to quantify serum concentration of monoclonal antibodies have been widely used by pharmaceutical companies during the development phases of therapeutic antibodies and are now increasingly adopted by clinical laboratories.^35,36  Unlike ELISAs, mass spectrometry–based assays do not rely on monoclonal antibody–specific reagents such as recombinant antigens or anti-idiotypic antibodies.³⁵  Moreover, the use of molecular mass as a measurement tool enhances specificity, facilitates accurate quantification, and enables multiplexing.³⁵  To date, MS-based methods either are available as clinical tests or are documented in the literature for therapeutic monoclonal antibodies such as infliximab, adalimumab, rituximab, vedolizumab, eculizumab, tocilizumab, risankizumab, and daratumumab.^37–45 

In protein analysis, it is important to control the variability of digestion.⁴⁶  A full-length SIL infliximab will theoretically be the best option because it undergoes exactly the same sample preparation; processing, including trypsin digestion; and detection processes as the analyte. However, the high price of full-length SIL monoclonal antibodies impedes its use in clinical laboratories, making SIL peptides the most widely used ISs. However, SIL peptides can correct only for detection process and not for sample pretreatment inconsistencies.⁴⁷  We used winged SIL peptide for IS. Winged SIL peptide, in principle, cannot compensate for stages other than the trypsin digestion and detection process. In previous studies, Faria et al⁴⁸  showed that extended SIL peptide better accounted for variability in trypsin activity than SIL peptide, and Barnidge et al⁴⁹  reported that the cleavable IS peptides did not effectively reflect the proteolysis condition of the target protein. The validation results of our study demonstrated excellent precision (%CV 2.6–6.0) and accuracy (94.2%–98.7%).

Our study has several limitations. First, the majority of samples had an infliximab concentration less than 20 µg/mL, with a limited number of samples having higher concentrations. Although this distribution effectively reflects the clinically relevant concentration, further studies including samples with higher infliximab concentration would be anticipated. Second, direct comparison between the 2 ELISAs and our LC-MS/MS method was not obtainable because the ELISAs quantitated free infliximab concentration whereas our LC-MS/MS method quantitated total infliximab concentration. Third, an optimization study to evaluate the digestion kinetics of the winged SIL peptide to validate it as a substitute for SIL infliximab was not conducted.

Previous studies have reported interassay variability between LC-MS/MS and immunoassays for measurement of infliximab.^21–23,39  Our study included the largest number of samples to date, strengthening the confidence of the results. In line with previous reports, the infliximab concentration measured by LC-MS/MS and the ELISA methods showed strong correlation (correlation coefficient r > 0.9) but significant difference. Although both ELISAs quantitated only free infliximab whereas our LC-MS/MS method quantitated total (free or bound) infliximab, Remsima Monitor showed good agreement with LC-MS/MS, whereas IDKmonitor Infliximab overestimated infliximab concentration compared with LC-MS/MS, implying a significant difference between the 2 ELISAs. This result supports the current knowledge that the results from different immunoassays should not be interpreted interchangeably because of interassay variability. Furthermore, noticeable interference by ADA on infliximab concentration was observed.

Current therapeutic range for infliximab is 3 to 8 μg/mL, which is established based on the results measured by immunoassays.¹  Nemoz et al⁵⁰  reported a therapeutic threshold of 6.2 μg/mL for biological remission using LC-MS/MS in 55 IBD patients. Considering the interassay variability between immunoassay and LC-MS/MS, the pharmacokinetic relationship and therapeutic range should be redefined using infliximab concentration measured by LC-MS/MS methods.

We successfully developed a simple, fast, and affordable LC-MS/MS method for quantitation of infliximab in human serum suitable for clinical use. Infliximab quantitation using ELISAs showed considerable interassay variability and noticeable interference by ADA. Biologic agents are an expensive treatment option; therefore, accurate TDM using LC-MS/MS can improve patient care while reducing health-related costs. Further studies to determine the therapeutic ranges associated with clinical remission using the LC-MS/MS method would be needed to maximize the clinical benefit of TDM.