Mass spectrometry quantitation of immunosuppressive drugs in clinical specimens using online solid-phase extraction and accurate-mass full scan-single ion monitoring

Highlights

• •LC-MS method for immunosuppressive drugs using online SPE and accurate-mass FS-SIM.
• •The FS-SIM LC-MS method correlated well with a reference LC-MS method.
• •The FS-SIM data acquisition mode has advantages in establishing an LDT LC-MS assay.

Abstract

Introduction

Therapeutic drug monitoring (TDM) of immunosuppressants is essential for optimal care of transplant patients. Immunoassays and liquid chromatography-mass spectrometry (LC-MS) are the most commonly used methods for TDM. However, immunoassays can suffer from interference from heterophile antibodies and structurally similar drugs and metabolites. Additionally, nominal-mass LC-MS assays can be difficult to optimize and are limited in the number of detectable compounds.

Objectives

The aim of this study was to implement a mass spectrometry-based test for immunosuppressant TDM using online solid-phase extraction (SPE) and accurate-mass full scan-single ion monitoring (FS-SIM) data acquisition mode.

Methods

LC-MS analysis was performed on a TLX-2 multi-channel HPLC with a Q-Exactive Plus mass spectrometer. TurboFlow online SPE was used for sample clean up. The accurate-mass MS was set to positive electrospray ionization mode with FS-SIM for quantitation of tacrolimus, sirolimus, everolimus, and cyclosporine A. MS² fragmentation pattern was used for compound confirmation.

Results

The method was validated in terms of precision, analytical bias, limit of quantitation, linearity, carryover, sample stability, and interference. Quantitation of tacrolimus, sirolimus, everolimus, and cyclosporine A correlated well with results from an independent reference laboratory (r = 0.926–0.984).

Conclusions

Accurate-mass FS-SIM can be successfully utilized for immunosuppressant TDM with good correlation with results generated by standard methods. TurboFlow online SPE allows for a simple “protein crash and shoot” sample preparation protocol. Compared to traditional MRM, analyte quantitation by FS-SIM facilitates a streamlined assay optimization process.

Introduction

For many drugs with narrow therapeutic ranges and difficult-to-predict pharmacokinetics, such as immunosuppressants, therapeutic drug monitoring (TDM) is an essential component of optimal patient care. In the United States, there are currently over 42,000 solid organ transplants and 22,000 bone marrow transplants performed annually, and these numbers continue to increase [1], [2]. The vast majority of patients will be treated with an immunosuppressive regimen that can benefit from TDM performed in clinical chemistry laboratories [3], [4], [5]. TDM is crucial for maximizing efficacy while minimizing toxicity of immunosuppressants [5]. Under-immunosuppression can lead to allograft rejection, whereas over-immunosuppression can make patients more susceptible to infections and cancers [6], [7], [8]. Moreover, immunosuppressants often have deleterious side effects including nephrotoxicity and neurotoxicity [9]. In addition to drug level titration, knowing the serum concentrations of these drugs can indirectly assist with diagnosis of transplant-related complications associated with the level of immunosuppression [6]. Thus, TDM of immunosuppressants is a part of standard care for transplant patients [10]. Additionally, some immunosuppressants have FDA-approved indications outside the context of solid organ transplantation; in those patient populations, TDM can also be performed. For example, everolimus is approved for the treatment of renal cell carcinoma, neuroendocrine tumors, metastatic breast cancer, and complications related to tuberous sclerosis [11].

Two of the most commonly used analytical methods for TDM of immunosuppressants are immunoassays and liquid-chromatography mass spectrometry (LC-MS). According to the 2021 College of American Pathologists survey, approximately one fifth of the participating clinical laboratories used LC-MS for immunosuppressant TDM, while the rest used immunoassays. Immunoassays often have a quicker turnaround time and fewer personnel or equipment requirements; however, they can suffer from interferences from heterophile antibodies and cross-reactivity with similar compounds such as drugs within the same class or metabolites [12], [13]. Since LC-MS assays are able to circumvent several limitations of immunoassays, particularly to significantly enhance analytical specificity, they are increasingly being used as clinical chemistry LDTs [12], [14].

Historically, TDM by LC-MS has been most often performed on nominal-mass triple quadrupole mass spectrometers using MRM [13], [14], [15], [16], [17]. Since triple quadrupole mass spectrometers have a unit m/z resolution, quantitation of analytes requires matching of both precursor ions and product ions to achieve sufficient analytical specificity. It can take time and effort to optimize these MRM conditions. In this study, we describe a method using accurate-mass full scan-single ion monitoring (FS-SIM) data acquisition mode to simultaneously quantify four commonly used immunosuppressants: tacrolimus, sirolimus, everolimus, and cyclosporine A.

Materials and methods

Chemicals and reagents

Calibrators (6PLUS1 Multilevel Whole Blood Calibrator Set, MassCheck Immunosuppressants Whole Blood Blank Control, and MassCheck Immunosuppressants Whole Blood Control, Four-Level) were purchased from Chromsystems; LC/MS grade methanol, isopropanol, and water were acquired from Sigma-Aldrich; formic acid (LC/MS grade) was obtained from Thermo Fisher Scientific; zinc sulfate heptahydrate and 25% ammonia (HPLC grade) were supplied by Sigma-Aldrich; isotope-labeled internal standards (IS) sirolimus-D₃, everolimus-D₄, and cyclosporine-¹⁵N₁₁ were purchased from Cerilliant; and tacrolimus-¹³C-D₂ was acquired from Cayman Chemical.

Sample preparation

Remnant patient whole blood samples collected at Stanford Health Care were used for assay validation, and processing was performed according to Institutional Review Board (IRB) protocols approved by Stanford Health Care. Additionally, de-identified remnant frozen samples were purchased from Mayo Clinic Laboratories with quantitative results acquired by tandem mass spectrometry in MRM data acquisition mode [18]. Whole blood specimens were collected in EDTA-containing tubes and stored at 2–8 °C for up to 7 days or −20 °C for up to 3 months. Samples were mixed on a rocker prior to analysis to reduce plasma separation. The IS working solution contained 8 ng/mL tacrolimus-¹³C-D₂, sirolimus-D₃, everolimus-D₄, and 160 ng/mL cyclosporine A-¹⁵N₁₁ in methanol. The sample clean up protocol used was modified from previously published methods [15], [17]. Each microtainer included 80 µL of patient whole blood, 120 µL of 0.1 M ZnSO₄, and 200 µL of IS working solution; the samples were then mixed for 5 min at 2000 rpm on a vortexer and centrifuged at 11,000 rpm for 10 min. Finally, 250 µL of supernatant from each sample was transferred to a 96-position deepwell plate for LC-MS processing.

LC-MS method

LC-MS analysis was performed on a TLX-2 multi-channel HPLC (Thermo Fisher Scientific) coupled with a Q-Exactive Plus mass spectrometer (Thermo Fisher Scientific). A TurboFlow Online SPE column (Cyclone-P, 0.5 × 50 mm) (Thermo Fisher Scientific) was used to remove residual large molecules to waste prior to loading onto the C18 analytical column (Acquity UPLC BEH C18, 1.7 μm) (Waters). The HPLC method is described in Table 1.

Table 1.

Liquid chromatography method for immunosuppressant quantitation.

		TurboFlow Column					Analytical Column				TurboFlow Configuration	MS
Start Time	Time Length	Flow Rate	Gradient	% MPA	% MPB	% Wash	Flow Rate	Gradient	% MPA	% MPB
0.0	0.5	1.5	Step	70	30	0	0.5	Step	70	30	SPE Loading	Standby
0.5	0.2	0.2	Step	70	30	0	0.5	Step	70	30	SPE Loading	Standby
0.7	0.8	0.2	Step	20	80	0	0.3	Step	70	30	Transferring	Standby
1.5	0.3	1.5	Step	0	0	100	0.5	Step	35	65	Washing	Standby
1.8	0.7	1.5	Step	0	0	100	0.5	Ramp	21	79	Washing	Standby
2.5	0.7	1.0	Step	20	80	0	0.5	Ramp	7	93	Loop loading	Standby
3.2	1.0	1.0	Step	20	80	0	0.5	Step	7	93	Loop loading	On
4.2	0.5	1.0	Step	70	30	0	0.5	Step	0	100	Equilibration	Standby
4.7	0.5	1.5	Step	70	30	0	0.5	Step	70	30	Equilibration	Standby

Note: Time length in minutes; flow rate in mL/min; In gradient, step means immediate mobile phase composition change and ramp means linear gradual mobile phase change over the time course. Mobile Phase A: 10 mM ammonium formate in water and 0.1% formic acid; Mobile Phase B: 100% methanol; Wash solvent: 90% methanol and 10% isopropanol.

The sample injection volume was 50 µL, and the analytical column temperature was set to 60 °C. The accurate-mass spectrometer was set to positive electrospray ionization (ESI) with FS-SIM data acquisition mode. Unlike regular SIM, FS-SIM takes full scan mass spectra and, thus, can monitor all molecular ions in the m/z range for quantitation rather than setting the mass analyzer at a specific m/z value to monitor one analyte at a time. The mass spectrometer settings are shown in Table 2.

Table 2.

Mass spectrometer settings for immunosuppressant quantitation.

General
Runtime	3.2–4.2 (1 min)
Polarity	Positive
Default charge	1
Inclusion	On
Full Scan MS
Resolution	70,000
AGC target	1e6
Maximum IT	100 ms
Scan range	500–1300 m/z
dd-MS2
Resolution	17,500
AGC target	2e4
Maximum IT	50 ms
Loop count	5
Isolation window	2.0 m/z
NCE	20
Heated ESI settings
Sheath gas	40 psi
Auxiliary gas	10 psi
Sweep gas	0 psi
Spray voltage	3.8 kV
Capillary temperature	300 °C
S-lens RF level	50
Auxiliary gas heater	400 °C

Data analysis

The quantitation of tacrolimus, sirolimus, everolimus, and cyclosporine A was performed using LC coupled with accurate-mass MS. Example extracted ion chromatograms of the four immunosuppressants and their corresponding isotope-labeled internal standards are displayed in Fig. 1 (calibrator level 5) and Supplementary Figure S1 (calibrator level 1), and the monoisotopic m/z values of molecular ions and retention times used for FS-SIM quantitation are shown in Table 3. High-energy C-trap dissociation (HCD) was used during the LC-MS analysis to fragment each analyte. The MS² spectrum of each analyte was matched to the MS² spectra stored in the library, requiring a threshold match score of 80 for compound confirmation. The mass tolerance was set to 10 ppm for all four analytes. The auto-integration software Trace Finder 5.1 (Thermo Fisher Scientific) was used to measure the area of each peak with a retention time window of 30 s. All data analysis and statistics were completed using Excel (Microsoft), except for correlation analysis which was performed by cp-R, a graphical interface to the R statistical programming (https://sourceforge.net/projects/cprchempath/).

Fig. 1.

Example extracted ion chromatograms of calibrator level 5 for each of the four immunosuppressants (23.3 ng/mL for tacrolimus, 29.3 ng/mL for sirolimus, 21.3 ng/mL for everolimus, and 777 ng/mL for cyclosporine A) along with isotope-labeled internal standards.

Table 3.

Monoisotopic m/z values of molecular ions and retention times of the four immunosuppressants and internal standards for FS-SIM quantitation. The total elution time refers to the time after the HPLC method starts.

Immunosuppressant	Species	m/z	Total elution time	Retention time in MS acquisition window
Tacrolimus	+NH4	821.51580	3.45	0.25
Tacrolimus-13C-D2	+NH4	824.53171	3.45	0.25
Sirolimus	+NH4	931.58897	3.54	0.34
Sirolimus-D3	+NH4	934.60780	3.54	0.34
Everolimus	+NH4	975.61518	3.56	0.36
Everolimus-D4	+NH4	979.64029	3.56	0.36
Cyclosporine A	+H	1202.84864	3.85	0.65
Cyclosporine A-15N11	+H	1213.81603	3.85	0.65

Method validation

Precision was determined by replicate analysis of QC samples following Clinical and Laboratory Standards Institute (CLSI) EP10 guidelines, presented as a coefficient of variation (CV). For each drug level, three replicates were tested each day over five days (n = 3 × 5). Accuracy was represented by analytical bias, which is determined by comparing average quantitation results of QC samples to their true values provided by Chromsystems, traceable to certified reference materials. The accuracies of tacrolimus, sirolimus and cyclosporine A were tested against one batch of 2021 College of American Pathologists (CAP) proficiency testing materials with six samples for each drug; no CAP proficiency testing material was available for everolimus.

The limit of quantitation (LOQ) was defined as the lowest concentration producing a peak eluting within ±0.1 min of the analyte's retention time, with a signal-to-noise ratio (S/N) ≥20, an analytical bias ≤10%, and a CV of the peak area ≤20% (n = 5). LOQ was determined using low-concentration calibrators prepared separately. Carryover and nonlinearity were examined according to CLSI EP10 guidelines. In brief, 3 samples with low (L), middle (M), and high (H) analyte concentrations were tested in triplicates following the order MHLMMLLHHM (first M was to prime the system), and the experiment was repeated in 5 runs. The parameters could be calculated using multiple regression analysis. The stability of each immunosuppressant was assessed by 5 patient samples stored at 4 °C and at −20 °C over a 20-day period. To evaluate ion suppression by sample matrix, 10 blank whole blood sample matrices were spiked with pure standards at three concentrations (2 ng/mL, 40 ng/mL, and 100 ng/mL) without using an internal standard. The heterogeneity between sample matrices was evaluated by the CV of absolute MS response (peak area). Interference was tested using other drug standards available in the laboratory (e.g. fluconazole, itraconazole, hydroxyitraconazole, isavuconazole, voriconazole, and posaconazole).

Method comparison

Immunosuppressant quantitation using the current FS-SIM LC-MS method was performed at the Stanford Health Care Clinical Chemistry Laboratory. These results were compared to a reference LC-MS method with MRM quantitation from Mayo Clinic Laboratories, and a Roche immunoassay performed at Stanford Health Care Clinical Chemistry Laboratory.

Results

The validation data of the immunosuppressant LC-MS assay, including within-run and between-run imprecision, analytical bias, limit of quantitation, analytical measurement range, and linearity, are reported in Table 4. Comparing the CAP mean and CAP standard deviations (SD) of results obtained by LC-MS to those of the accurate-mass FS-SIM method, the results were within an acceptable cutoff of ±3 SD: −0.09 to 0.91 SD for tacrolimus, −0.94 to 0.17 SD for sirolimus, and 0.16–0.89 SD for cyclosporine A. Carryover and nonlinearity were insignificant (t-statistics < 4.6, P > 0.01). Stability testing showed no significant variations for any of the four analytes over a 20-day time frame: 5 patient samples tested every 3–7 days; at 4 °C the measurement showed CVs of 3.5–8.1% for tacrolimus, 1.3–7.5% for sirolimus, 2.2–4.8% for everolimus, and 5.0–7.6% for cyclosporine A; at −20 °C the measurement showed CVs of 2.7–7.7% for tacrolimus, 3.8–9.8% for sirolimus, 1.8–3.0% for everolimus, and 2.6–5.4% for cyclosporine A. Comparison of ion suppression by 10 sample matrices between patients showed CVs of 6.6–26% for tacrolimus, 13–29% for sirolimus, 18–35% for everolimus, and 11–15% for cyclosporine A; no interference with other drugs available in the laboratory was noted with the FS-SIM LC-MS method.

Table 4.

Validation data for quantitation of four immunosuppressants by accurate-mass single ion monitoring. Three replicates were tested each day over five days for each quality control level (n = 3 × 5).

Analyte		Within-Run CV	Total CV	Analytical Bias	LOQ (ng/mL)	AMR (ng/mL)	Linearity
Tacrolimus (ng/mL)	L1 (2.69)	3.6%	4.7%	0.9%	1.0	1.0–40	Slope = 1.010
	L2 (7.55)	4.4%	5.3%	3.0%			Intercept = 0.177
	L3 (15.6)	2.4%	3.6%	4.3%			R = 1.000
	L4 (33.3)	2.6%	2.8%	1.1%			P = 1.5 × 10-4
Sirolimus (ng/mL)	L1 (2.87)	3.5%	3.9%	1.8%	2.0	2.0–40	Slope = 1.038
	L2 (9.80)	3.7%	5.4%	4.2%			Intercept = -0.012
	L3 (19.3)	2.4%	3.2%	3.9%			R = 1.000
	L4 (37.8)	3.1%	3.4%	3.8%			P = 3.4 × 10-6
Everolimus (ng/mL)	L1 (2.52)	3.1%	4.3%	0.7%	2.0	2.0–40	Slope = 1.018
	L2 (4.57)	4.1%	4.5%	0.4%			Intercept = -0.118
	L3 (9.02)	2.6%	2.7%	−1.6%			R = 1.000
	L4 (30.3)	1.6%	2.7%	1.5%			P = 4.5 × 10-5
Cyclosporine A (ng/mL)	L1 (48.3)	3.8%	4.7%	4.6%	20	20–1700	Slope = 1.014
	L2 (244)	5.3%	6.3%	2.8%			Intercept = 0.981
	L3 (491)	3.5%	4.1%	0.8%			R = 1.000
	L4 (1158)	3.6%	4.5%	1.6%			P = 1.6 × 10-5

Results generated by the immunosuppressant LC-MS assay using FS-SIM data quantitation in this study were compared with LC-MS results from a clinical reference laboratory. Linear regressions were calculated using the Passing-Bablok method; correlation with an immunoassay method was also assessed for tacrolimus. Tacrolimus levels measured by the FS-SIM LC-MS method correlated with those determined by both the reference LC-MS method (n = 310) and the immunoassay (n = 318) with r = 0.954 and r = 0.984 respectively (Fig. 2). Sirolimus (n = 160), everolimus (n = 182), and cyclosporine A (n = 121) also had correlation with the LC-MS results from the reference laboratory with r values ranging from 0.926 to 0.979 (Fig. 3). For all four immunosuppressants, slopes of the linear regression of correlation with the reference LC-MS results were between 1.01 and 1.12. Furthermore, hypothesized major metabolites of the four immunosuppressants identified by intact molecular masses could be observed in the accurate-mass FS-SIM LC-MS method (Supplementary Figure S2, Supplementary Table S1).

Fig. 2.

Correlation of tacrolimus quantitation by the FS-SIM LC-MS method with that of the reference laboratory LC-MS method and immunoassay. Passing-Bablok regression analysis revealed residual standard error = 1.3 ng/mL and 0.63 ng/mL for the reference LC-MS method and Roche immunoassay respectively. The shaded region indicates 95% confidence interval. Bland-Altman plots showed a mean difference = 11.5% with SD = 17.5% for comparison with the reference LC-MS method and mean difference = -4.1% with SD = 12.8% for comparison with the immunoassay. The shaded region indicates ±4 SD.

Fig. 3.

Correlation of sirolimus, everolimus, and cyclosporine A quantitation by the FS-SIM LC-MS method with the reference laboratory LC-MS method. Passing-Bablok regression analysis revealed residual standard errors = 1.7 ng/mL, 0.98 ng/mL, and 45.1 ng/mL for sirolimus, everolimus, and cyclosporine A, respectively. The shaded region indicates 95% confidence interval. Bland-Altman plots showed a mean difference = −11.0% with SD = 26.6% for sirolimus, mean difference = 4.5% with SD = 16.2% for everolimus, and mean difference = -4.8% with SD = 15.0% for cyclosporine A. The shaded region indicates ±4 SD.

Discussion

The FS-SIM LC-MS method demonstrated excellent analytical performance, as evaluated by precision, accuracy, and linearity. The correlation with the reference LC-MS method showed consistent results with other clinical laboratories running MS-based immunosuppressant TDM assays. Additionally, the FS-SIM LC-MS method was closely correlated with the immunoassay used in this study, indicating that improved immunoassays are less prone to interferences. However, multiple previous publications have highlighted that the risk of interference in immunoassays should still be taken into consideration when immunosuppressant TDM is implemented in clinical laboratories [12], [13], [16].

This LC-MS assay improved analytical specificity by combining accurate-mass FS-SIM (with ≤10 ppm mass error criteria) for analyte quantitation and MS² fragmentation pattern for compound confirmation. The high accuracy and resolution of accurate-mass MS allow for precise identification of compounds by full scan (mass error within 5–10 ppm). Matching the MS² fragmentation pattern of an analyte to a spectral library confirms compound identity with greater confidence than matching one or two fragments, which is typically implemented in MRM. This advantage helps minimize the risk of interferences from similar compounds.

Because FS-SIM quantifies analytes only by molecular ions, not requiring product ions from fragmentation, it substantially simplifies the assay optimization process. In addition, since every full scan cycle measures all molecular ions in the m/z range, it allows for the simultaneous monitoring and quantification of all detected compounds. This cannot be easily achieved using MRM due to the lack of pure standards to optimize transitions and the demand for increased data acquisition cycle time to include metabolites, which may result in insufficient data points in a chromatographic peak for analyte quantitation. Although not currently well-studied, knowledge about the concentration of metabolites relative to intact immunosuppressants may become informative in understanding the pharmacokinetics of drugs in individual patients in the future.

In this LC-MS assay, samples were prepared with protein crash down with ZnSO₄ followed by cleanup by TurboFlow online SPE prior to loading into reverse-phase LC separation, as performed in previous studies [15], [17]. This “protein crash and shoot” sample preparation protocol minimizes manual labor and shortens overall assay time to fit into one staff shift. Moreover, it facilitates rapid staff training and can be further enhanced with automation of laboratory workflow through an automatic liquid handler. These advantages are particularly important for LDTs that require a short turnaround time.

Compared to immunoassay, a known limitation of using MS-based methods for immunosuppressant TDM is the longer turnaround time. Although the total run time of the FS-SIM LC-MS method was 5.2 min, the time needed per sample with the current duplex (TLX-2) HPLC system was 2.6 min as samples were injected in a staggered manner. The MS data acquisition window was only 1-minute wide, so the run time per sample could be further shortened to 1.3 min with a quadruplex (TLX-4) HPLC system. However, the throughput of MS-based sequential sample processing is not comparable to that of automated immunoassays with parallel sample processing; thus, in laboratories with high test volumes, immunoassay remains the most time-efficient method for TDM testing. In addition, although online SPE simplifies the sample preparation protocol required, automation using a liquid handler is highly recommended to prevent repetitive strain injuries for laboratory staff, which might add cost to the assay implementation. Finally, since triple quadrupole mass spectrometers are more commonly available in clinical laboratories, it may take some time for the FS-SIM LC-MS method to be widely adopted, as it requires an accurate-mass mass spectrometer.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following are the Supplementary data to this article: