Bioanalysis of INCB000928 in Human Saliva: Nonspecific Binding and Inhomogeneous Concentration

Abstract

Aim: To develop a bioanalytical method for quantifying INCB000928 in human saliva. Materials & methods: Human centrifuged saliva and human whole saliva were compared for matrix selection. Protein precipitation extraction and HPLC–MS/MS was used for analysis. Results & conclusion: Nonspecific binding of INCB000928 was reduced in whole versus centrifuged saliva. Whole saliva was a preferred matrix for INCB000928 bioanalytical method validation. Incurred sample reanalysis (ISR) using a successfully validated method failed in a healthy volunteer study because of inhomogeneous INCB000928 concentration across sample tube depths. Individual mixing of sample tubes followed by immediate aliquoting corrected the ISR failure, with 97.2% of repeats passing versus 41.7% for the same ISR samples.

Plain language summary

Fibrodysplasia ossificans progressiva (FOP) is a very rare condition where bone forms outside the skeleton (ossification), leading to restricted movement, decreased quality of life and shortened life span. Mutations in a gene called ALK2 have been identified as causing FOP. INCB000928 is a novel drug (to be taken by mouth) which inhibits ALK2 activity and prevents ossification in a laboratory mouse model of FOP. Because monitoring of the levels and efficacy of a drug often requires blood draws, which can be taxing in patients with FOP, this study aimed to develop a method to measure INCB000928 levels in saliva. The authors propose a unique procedure to process saliva samples to ensure accurate, reproducible quantitation of INCB000928 levels in saliva.

Tweetable abstract

INCB000928 bioanalysis in human saliva resulted in incurred sample reanalysis failure due to inhomogeneous concentration across sample tube depths, which was corrected by individual tube mixing followed by immediate aliquoting.

Graphical abstract

Keywords::

• bioanalysis
• incurred sample reanalysis
• inhomogeneous concentration
• nonspecific binding
• saliva

Figure 1. Clinical saliva collection in the multiple ascending-dose volunteer study and illustration of validation quality control (QC) samples with one-step tube transfer.

Figure 2. Multiple reaction monitoring chromatograms of INCB000928 and D6-INCB000928.

INCB000928 in blank human saliva (top), INCB000928 in saliva spiked at LLOQ (middle), and D6-INCB000928 (bottom) in a LLOQ sample.

LLOQ: Lower limit of quantification.

Figure 3. Experimental design for incurred sample re-analysis failure investigation.

QC: Quality control.

Figure 4. Effects of sample aliquot positions on INCB000928 concentrations at low quality control (LQC), medium quality control (MQC) and high quality control (HQC) levels.

PAR: Peak area ratio.

Figure 5. (A & B) Effects of sample aliquot positions on INCB018424 and INCB050465 saliva concentrations at the low quality control (LQC) level.

(C & D) Nonspecific binding analysis for INCB018424 and INCB050465 in human whole saliva samples over four sequential tube transfers (T0: no tube transfer; T1–T4, one to four sequential tube transfer[s]) at the LQC level.

PAR: Peak area ratio.

Fibrodysplasia ossificans progressiva (FOP) is an ultra-rare and permanently disabling genetic disorder in which skeletal muscles and connective tissues gradually ossify, leading to bone formation outside the skeleton that restricts movement, decreases quality of life and shortens life span [1,2]. Gain-of-function mutations in the activin receptor A type-1 gene, which encodes activin receptor-like kinase 2 (ALK2), involved in bone morphogenetic protein and TGF-β family signaling, have been identified to cause FOP [3]. Although other pathogenic variants have been more recently identified, more than 97% of patients with FOP share the same heterozygous mutation R206H (c.617G>A) in ALK2, and all the pathogenic mutations are located in the glycine-rich and serine-rich domain or in the protein kinase domain, making it an attractive target for therapeutic intervention [2,4].

There are no effective treatments for FOP. Small-molecule inhibitors of ALK2 have been developed and explored as targeted treatments [4–6]. INCB000928 is a novel, potent, selective and orally available small-molecule inhibitor of wild-type and mutant ALK2 (R206H) [7]. INCB000928 significantly reduces the production of hepcidin in human liver cells, primary hepatocytes and rodent models of anemia, and has shown efficacy in preventing injury-induced heterotopic ossification in a human FOP mouse model, offering a promising path to clinical proof-of-efficacy studies and potential benefits to patients with hepcidin-induced anemia and FOP [7].

Blood draws for pharmacokinetic evaluation in patients with FOP are challenging. This is mainly due to poor venous access, contractures and unusual positioning [8]. Although patients with FOP can tolerate peripheral blood collection performed as gently as possible by an experienced phlebotomist, tourniquet time and frequent blood draw attempts should be minimized [8]. Furthermore, minor trauma – such as intramuscular injections, muscle fatigue and blunt muscle trauma from bumps, bruises and falls – can trigger episodic flare-ups and irreversible heterotopic ossification [9,10]. In contrast to plasma, saliva offers a noninvasive alternative and is viable if concordance with plasma pharmacokinetics can be established.

To support clinical pharmacokinetic assessment in patients with FOP, a bioanalytical method for the quantification of INCB000928 in human saliva is warranted for initial investigation of pharmacokinetics and for examination of the correlation between saliva and plasma concentrations in healthy adult volunteers. There are a handful of reports on small-molecule bioanalytical method development and/or validations in saliva using platforms based on LC–MS/MS [11–17]. The aim of the present study is to share with the bioanalytical community the issues and challenges encountered, lessons learned and knowledge gained during method development, validation and in-study conduct for quantification of INCB000928 in human saliva, an alternative specimen to plasma for therapeutic drug monitoring and pharmacokinetics. To the best of our knowledge, this is the first report on: comparison of nonspecific binding (NSB) between human whole saliva (HWS) and human centrifuged saliva (HCS) for small-molecule bioanalytical method validation; incurred sample reanalysis (ISR) failure caused by tube position-dependent concentration variations; and solutions to ensure accurate and reproducible INCB000928 concentration determination in human saliva.

The observation of position-dependent INCB000928 concentration variation in saliva sample tubes was intriguing. To explore whether the position-dependent concentration variation is compound-specific, two other compounds, INCB018424 and INCB050465, were chosen for evaluation, and the results are also presented in this paper.

Materials & methods

Chemicals & reagents

Reference materials INCB000928 and internal standard D₆-INCB000928, INCB018424 (ruxolitinib) and internal standard ¹³C₄-INCB018424, and INCB050465 (parsaclisib) and corresponding analog internal standard INCB050904 were from Incyte Corporation (DE, USA). LC–MS grade acetonitrile (ACN), water and methanol were purchased from EMD Millipore Corporation (MA, USA), and LC–MS grade formic acid was purchased from Thermo Fisher Scientific (MA, USA). HPLC-grade ACN, methanol, water and DMSO were purchased from EMD Millipore Corporation, and American Chemical Society-grade methyl-tert butyl ether (MTBE) was purchased from VWR International (PA, USA). HPLC-grade ammonium acetate and ammonium formate were purchased from Honeywell (NC, USA). Sodium bicarbonate was purchased from VWR International. Pooled human saliva including HWS and HCS were purchased from BioIVT (NY, USA).

Preparation of stocks, calibration standards & quality control solutions

Stock solutions of INCB000928 and internal standard D₆-INCB000928 were prepared separately in DMSO:methanol, 50:50 (v/v) to obtain a concentration of 5 mM. Working solutions of INCB000928 were prepared by diluting appropriate amounts of stock or higher working solutions with ACN:water, 50:50 (v/v). Different solvents were used for the preparation of stock solutions and working solutions, and this was to ensure complete solubilization of the analyte at a stock concentration of 5 mM. Duplicates of the INCB000928 stock solutions were prepared and verified to be equivalent (difference ≤5%) prior to use. A working solution of D₆-INCB000928 (75 nM) was prepared by first diluting the 5-mM internal standard stock to a 50-μM intermediate solution and then diluting the intermediate solution with ACN 100%. The stock solutions and internal standard working solutions were stored at 4°C for further use.

Eight working calibration solutions for INCB000928 were prepared in human saliva by spiking appropriate working solutions to obtain concentrations of 5, 12, 50, 120, 500, 1200, 2500 and 3000 nM. Working quality control (QC) INCB000928 solutions at low, middle and high concentrations of 15, 180 and 2400 nM were prepared in the same manner. Calibration standards and QCs were freshly prepared on the day of analysis, or calibration standards and QCs stored frozen at -70°C were used provided that sufficient stability had been established.

Stock solutions of INCB018424 and internal standard ¹³C₄-INCB018424 were prepared separately in ACN:water, 50:50 (v/v) to obtain a concentration of 5 mM. Working solutions of INCB018424 were prepared by diluting an appropriate volume of stock or higher working solutions with ACN:water, 50:50 (v/v). An internal standard working solution (100 nM) was prepared by first diluting the 5-mM internal standard stock to a 1000-μM intermediate solution and then diluting the intermediate solution with ACN:water, 50:50 (v/v). The low QC (3 nM) solution was freshly prepared on the day of analysis by spiking appropriate working solution into saliva.

Stock solutions of INCB050465 and its analog internal standard INCB050904 were prepared separately in ACN:water, 50:50 (v/v) to obtain a concentration of 5 mM. Working solutions of INCB050465 were prepared by diluting an appropriate volume of stock or higher working solutions with ACN:water, 50:50 (v/v). An internal standard working solution (250 nM) was prepared by first diluting the 5-mM internal standard stock to a 1000-μM intermediate solution and then diluting the intermediate solution with ACN:water, 50:50 (v/v). The low QC (15 nM) solution was freshly prepared on the day of analysis by spiking appropriate working solution into saliva.

Sample preparation

Pooled saliva (HWS and/or HCS) was used as a blank matrix for preparation of saliva calibration standards and QC samples and was stored at -20°C before use.

For INCB000928 sample extraction in the validated method, 50 μl of the saliva sample was placed in tubes in a 96-well rack. The saliva proteins in the sample aliquots were precipitated by the addition of 200 μl of ACN containing internal standard (75 nM D₆-INCB000928) and the capped plate was vortex-mixed. After centrifugation, 100 μl of the supernatant was transferred to a clean 96-well tube plate containing 200 μl of water. The plate was covered and vortexed for approximately 1 min, then it was placed in the autosampler tray and an aliquot of the sample was injected into an LC–MS/MS system for analysis.

For INCB018424 QC sample extraction for exploratory purposes, 50 μl of the saliva QC sample was placed in tubes in a 96-well rack. After an aliquot of 50 μl of internal standard (100 nM ¹³C₄-INCB018424, dissolved in ACN:water, 50:50 [v/v]) was added, an aliquot of 100 μl of 0.1 M NaHCO₃ was added, then 800 μl of MTBE was added and the samples were covered and vortexed. After centrifugation, 700 μl of the MTBE layer was transferred to a clean 96-well tube plate. The samples were then dried under nitrogen. An aliquot of 250 μl of reconstitution solution (ACN:water, 50:50 [v/v]) was then added to each sample. The plate was capped, vortexed for approximately 1 min and placed in the autosampler tray and injected into an LC–MS/MS system for analysis. Sample injection volumes of 0.5, 1.0 and 2.0 μl were evaluated, and 0.5 μl was determined to be sufficient for sensitivity and preferred for linearity.

For INCB050465 QC sample extraction for exploratory purposes, 50 μl of the saliva QC sample was placed in tubes in a 96-well rack. After an aliquot of 50 μl of internal standard (50 nM INCB050904 dissolved in 50:50 ACN:water) was added, an aliquot of 100 μl of 0.1 M NaHCO₃ was added, then 800 μl of MTBE was added and the samples were covered and vortexed. After centrifugation, 150 μl of the MTBE layer was transferred to a clean 96-well tube plate. The samples were then dried under nitrogen. An aliquot of 250 μl of reconstitution solution (ACN:water, 50:50 [v/v]) was then added to each sample. The plate was capped, vortexed for approximately 1 min and placed in the autosampler tray and injected into an LC–MS/MS system for analysis.

LC–MS/MS instrumentation & conditions

LC–MS/MS analysis for INCB000928, INCB018424 and INCB050465 was carried out with an Applied Biosystems SCIEX Triple Quad 6500+ or an AB SCIEX QTRAP 6500 MS (Applied Biosystems, MA, USA) coupled with Shimadzu Nexera X2 binary UHPLC pumps, a degasser, an autosampler and a column oven (Shimadzu, MD, USA).

The chromatographic separation for INCB000928 was achieved on a Thermo Hypersil Gold C4 HPLC column (50 × 2.1 mm, 3 μm; Thermo Fisher Scientific), with isocratic elution at 29% mobile phase B for a 2.1-min LC program at a flow rate of 0.5 ml/min. The mobile phase A was 20 mM ammonium acetate (pH ~6, unadjusted) and mobile phase B was ACN, 100%. The column temperature was maintained at 40°C and the autosampler was set to 15°C during the LC–MS/MS analysis. The mass spectrometer was operated in positive ESI mode, and the resolution setting used was ‘unit’ for both Q1 and Q3. The multiple reaction monitoring (MRM) transition was m/z 503.2 → 462.2 for INCB000928, and the MRM transition was m/z 509.3 → 468.2 for the internal standard, D₆-INCB000928. The optimized parameters of the mass spectrometer were curtain gas, 45 psig; ion spray voltage, 5500 V; source temperature, 600°C; nebulizer gas, 60 psig; turbo gas, 60 psig; and collision-activated dissociation gas, 8 psig. The optimized collision energy for both INCB000928 and D6-INCB000928 was 39 V.

The chromatographic separation for INCB018424 was achieved on a Phenomenex Synergi Polar-RP 80 Å HPLC column (30 × 2 mm, 4 μm; Phenomenex, CA, USA), with isocratic elution at 55% mobile phase B for a 2.25-min LC program at a flow rate of 0.3 ml/min. The chromatographic separation for INCB050465 was achieved on a Phenomenex Synergi Polar-RP 80 Å HPLC column (30 × 2 mm, 4 μm), with isocratic elution at 42% mobile phase B for a 1.51-min LC program at a flow rate of 0.35 ml/min. For both INCB018424 and INCB050465, the mobile phase A was 2 mM ammonium acetate and mobile phase B was ACN, 100%. The column temperature was ambient and the autosampler was set to 15°C during the LC–MS/MS analysis. The mass spectrometer was operated in positive ESI mode, and the resolution setting used was ‘unit’ for both Q1 and Q3. The MRM transition was m/z 307.3 → 186.2 for INCB018424, and the MRM transition was m/z 311.3 → 190.2 for the internal standard, ¹³C₄-INCB018424. The MRM transition was m/z 433.2 → 150.2 for INCB050465, and the MRM transition was m/z 440.3 → 150.2 for the internal standard, NCB050904. The optimized parameters of the mass spectrometer were curtain gas, 45 psig; ion spray voltage, 5500 V; source temperature, 600°C; nebulizer gas, 60 psig; turbo gas, 60 psig; and collision-activated dissociation gas, 9 psig.

NSB tests

To test NSB, 1 ml of the low QC samples were prepared in pooled human saliva in 2-ml Eppendorf polypropylene tubes. After mixing, an aliquot (800 μl) of each initial solution (transfer 0, denoted as T0) was transferred to a second tube (transfer 1, denoted as T1). After mixing, 600 μl was then transferred to a third tube (transfer 2, denoted as T2), followed by mixing and two more sequential transfers of 400 and 200 μl, respectively (transfers 3 and 4, denoted as T3 and T4). Each transfer tube (T0–T4) was analyzed in duplicate to evaluate NSB.

Calibration & data evaluation

Concentrations were calculated using eight concentration levels for calibration curves ranging from 5 to 3000 nM, with 1/x² weighted linear regression for INCB000928. No calibration was performed for INCB018424 and INCB050465, in which only peak area ratio was used for the assessment of the QC samples. Peak area integrations were performed using Analyst software (v1.6.3 or v1.7.2; Applied Biosystems), and regressions were carried out in Watson Bioanalytical LIMS (v7.4.1 or v7.6.2) from Thermo Fisher Scientific, Inc. Where Watson Bioanalytical LIMS could not perform a calculation, Excel 2016 or higher (Microsoft Corp., WA, USA) was used.

Saliva sample collection & analysis

Clinical study INCB00928-102 was a phase 1, double-blind, randomized, placebo-controlled, multiple-dose, dose-escalation study to assess the safety, tolerability and pharmacokinetics of INCB000928 when administered orally to healthy adult participants. The study was conducted in accordance with the ICH Guidelines for Good Clinical Practice, including the archiving of essential documents, the principles of the Declaration of Helsinki and other applicable local ethical and legal requirements. Informed consent was obtained from each participant before protocol-specific screening assessments were performed. The Advarra institutional review board approval was obtained for this study (reference number Pro00045716). For saliva collection, 1–3 ml of saliva was collected to a 5-ml collection tube via a saliva collection funnel obtained from Norgen Biotek Corp. (Thorold, Canada). At the end of each collection, the collection funnel was discarded and the saliva tube was tightly capped and placed on crushed ice or ice bath. Within 45 min of the collection, the saliva samples were stored in a -80°C freezer until shipment. Overall, 468 saliva samples from three cohorts were received. In each cohort, nine participants received INCB000928 and three participants received placebo. In this study, placebo samples were not analyzed and, as a result, 117 samples from each cohort for a sum of 351 saliva samples were analyzed, and 10% of the total samples (i.e., 36 samples) were reanalyzed to assess ISR. ISR was considered acceptable if two-thirds (67%) of the ISR samples showed repeat values within ±20% of the mean from the initial concentrations.

Results & discussion

Comparison of INCB000928 NSB between HWS & HCS

HWS contains cell debris, food debris and some high-molecular-weight proteins, which are removed by centrifugation to generate the HCS. Compared with HWS, HCS is substantially easier to handle and less frequently blocks pipette tips, which greatly facilitates accurate and consistent volume transfer and minimizes the effect of cellular components and food debris. NSB is a common and serious issue in bioanalytical assays in which the matrix lacks proteins and lipids to which analytes can bind [18,19]. The NSB of INCB000928 in saliva was investigated with a low QC (15 nM) sample prepared in HWS and HCS by using a sequential transfer approach [18]. The low QC sample prepared in both matrices showed increasing loss over each sequential transfer; however, the loss was greatly reduced in HWS compared with HCS (Table 1). Specifically, after four sequential tube transfers, a bias of -36.2% was observed for the low QC prepared in HCS, whereas it was reduced to -18.7% in HWS; a single-digit bias was observed for the low QC samples prepared in HCS for up to two sequential tube transfers and in HWS for up to three sequential tube transfers (Table 1). Considering the aforementioned primary differences between HCS and HWS, it is hypothesized that high-molecular-weight proteins, cell debris and food debris in HWS help in reducing the NSB of INCB000928 to plastic containers by either preferentially blocking the container walls or favorably binding to the analyte. However, further investigations of the exact mechanisms of HWS in decreasing the NSB of INCB000928 relative to HCS have not yet been performed.

In the multiple ascending-dose healthy volunteer study, saliva was collected into a collection tube via a saliva collection funnel, and the clinical collection scheme is shown in Figure 1A. If saliva spitting through the collection funnel was considered one passive tube transfer, then the clinically collected saliva sample experienced one single-step tube transfer (T1), and therefore the NSB had negligible impact on the quantification of INCB000928. In analogy to the clinically collected saliva, Figure 1B & C show the QC samples undergoing one-step tube transfer during the method validation process including bench-top QC, freeze–thaw QC and dilution QC. Bench-top and freeze–thaw QC samples were transferred to three individual tubes in duplicate for six values to account for potential tube variations for stability assessments, and the above ULOQ QC was independently diluted to six replicates for dilution integrity evaluation. The accuracies of the one-step tube transfer QCs were compared between respective preparations in HWS and HCS (Table 2). For the QC samples prepared in HCS, the accuracy of the high QCs for bench-top (26 h) and freeze–thaw (three cycles) stability assessment showed within but close to -15% bias, whereas that of the low QCs for the same matrix stability tests as well as the tenfold dilution integrity test resulted in exceeding -15% bias, supporting the conventional notion that lower analyte concentrations are more severely impacted by NSB. In contrast, the tenfold dilution QC and the low QCs prepared in HWS for bench-top (47 h) and freeze–thaw (five cycles) stability assessment improved the accuracy to less than -15% bias, and the high QCs prepared in HWS for bench-top (47 h) and freeze–thaw (five cycles) stability assessment further increased the accuracy to within single-digit biases of ±3% (Table 2).

It has been reported that HCS can be used as a substitute for HWS for many in vitro studies, and there is no significant difference in lubricating properties between HCS and HWS [20]. Our data indicated that HWS greatly mitigated the NSB of INCB000928 for a more accurate and reliable concentration determination, despite HCS offering the many aforementioned advantages in sample handling; therefore, HWS was a more desirable matrix for INCB000928 bioanalytical method validation and in-study conduct.

Method validation

An LC–MS/MS-based method was fully validated with respect to selectivity, specificity, precision, accuracy, carryover, recovery, matrix effect, dilution integrity and stability, according to the US FDA and European Medicines Agency guidelines [21,22]. Typical blank saliva and LLOQ sample chromatograms with an injection volume of 0.5 μl are presented in Figure 2. INCB000928 showed a linear response over the concentration range of 5–3000 nM, with correlation coefficients invariably ≥0.9938. The assay performance results are summarized in Table 3. The elements of the calibration curve included a blank sample (no analyte, no internal standard), a zero sample (blank plus internal standard) and eight non-zero calibrators (5, 12, 50, 120, 500, 1200, 2500 and 3000 nM). The acceptance criteria for the calibration curve were that the non-zero calibrators should be ±15% of nominal concentrations except at LLOQ, where it should be ±20% of the nominal concentration, and at least 75% and a minimum of six non-zero calibrators should meet the aforementioned criteria. For all the back-calculated non-zero calibration standards obtained during the validation study at the time of manuscript preparation, the inter-assay accuracy (% bias) ranged from -2.5 to 4.4% and the inter-assay precision (% CV) was ≤3.6%. For accuracy and precision, intra-assay accuracy (% bias) ranged from -8.7 to 5.6% and precision (% CV) ≤3.9%, and inter-assay accuracy (% bias) ranged from -3.9 to 0.4% and precision (% CV) ≤4.8%. The extraction recovery of INCB000928 was high and consistent, between 91.9 and 97.3%, with % CV ranging from 1.2 to 3.0% for the three concentrations at low (15 nM), middle (180 nM) and high (2400 nM) QC. The matrix factor for INCB000928 evaluated with six lots of blank saliva was between 0.951 and 0.965, and the internal standard-normalized matrix factor was very close to 1.0 with % CV ≤2.0%, indicating negligible matrix effect. The selectivity and specificity were demonstrated with six lots of blank saliva, in which the response was between 0.1 and 0.2% of the LLOQ for INCB000928 and 0.0% for the internal standard D6-INCB000928. In addition, INCB000928 was found to be stable in human saliva after 47.5 h at ambient temperature (benchtop stability), through five freeze–thaw cycles. The long-term frozen saliva sample storage stability was verified to be at least 175 days after the samples were stored frozen at nominal -70°C and at least 62 days after the samples were stored frozen at nominal -20°C. It was also verified that INCB000928 in the extracts was stable for at least 102 h when kept in the autosampler set to 15°C (processed sample stability) and upon reinjection of extracts after 139 h in the autosampler kept at 15°C (reinjection reproducibility).

Failure of ISR in a clinical study

A bioanalytical method for the quantification of INCB000928 was fully and successfully validated, and the validated method was applied to analyze saliva study samples from a multiple ascending-dose healthy volunteer study. The performance of the calibration standards and QCs from all four runs met the acceptance criteria: the % bias ranged from -5.0 to 4.6% and % CV ranged from 1.3 to 6.1% (Table 4). ISR was conducted to verify the reliability of the method used to determine the study sample INCB000928 concentrations. A total of 36 samples representing 10% of the study samples from three dose-escalation cohorts (i.e., 12 samples per cohort) were selected for ISR. Surprisingly, only 15 of the 36 repeats, representing 41.7% of the ISR samples, showed repeat values within ±20% of the mean from the initial concentrations (Table 5), far below the acceptance criteria of at least 67% of the repeats showing concentrations within ±20% of the mean from the initial values. Furthermore, the majority of the repeated samples resulted in >50% difference from the mean of the ISR and the original concentrations. The ISR failure triggered a thorough investigation.

Investigation of ISR failure revealed tube position-dependent concentration variations & corrective measures

ISR is a necessary component of bioanalytical method validation and is conducted to critically support the precision and accuracy measurements established with the QCs during method validation [21,22]. It was puzzling that the method for the quantification of INCB000928 was successfully validated, whereas the ISR in a clinical study failed. When the clinical saliva samples were directly compared with the validation samples, there were noticeable differences in tube sizes and sample volumes. For validation samples, 2-ml plastic tubes were used with ≤1-ml sample volumes, while 5-ml plastic tubes were used with 1- to 3-ml sample volumes for clinical saliva study samples. Additionally, for validation samples, owing to the small number of standards and QC samples, the samples were individually vortexed and immediately transferred to the extraction tubes, whereas the large number of saliva study samples were batch-vortexed and then an individual sample was manually transferred to the extraction tube. It was suspected that the saliva study samples had settled down in the collection tubes when they were transferred to the corresponding extraction tubes. Furthermore, the positions where the saliva were taken out of the 1- to 3-ml collection tubes were arbitrary during initial and ISR sample analyses. It was plausible that the analyte concentrations were not homogeneous at different depths of the tube.

To test the hypothesis of inhomogeneous concentration in the saliva sample collection tubes, low, medium and high QCs were prepared by using 5-ml plastic tubes in the volume of 3 ml to mimic the study samples. The experimental design is depicted in Figure 3. After preparation of the QC samples, the tubes were vortexed and allowed to settle for 10 min. Triplicate saliva samples were then taken out of the tubes at three positions including upper (>2-ml mark), middle (~1-ml mark) and bottom (where the saliva appeared cloudy) for extraction. The data at the three tube positions for the QC concentration levels are shown in Figure 4. For all three QC levels, the concentration denoted in peak area ratio was similar between the upper position and the middle position; however, it was significantly higher at the bottom of the tube, at approximately 1.5- to twofold relative to the upper and middle positions (Figure 4). The data confirmed the hypothesis of inhomogeneous INCB000928 concentration across the sample tube depths and explained the >50% discrepancies observed between the original and the ISR results. Comparing the saliva at the bottom versus that at the upper and middle positions, it is fairly equivalent to the comparison between HWS and HCS. Specifically, the saliva at the upper and middle positions after settling for 10 min in the experiment highly resembled HCS and appeared clearer than the saliva at the bottom, which was comparable with HWS in which debris and some high-molecular-weight proteins can be present and appeared cloudy. Therefore one possible explanation for the drastically higher analyte concentration at the bottom relative to the upper and middle positions was that INCB000928 preferentially bound to some high-molecular-weight proteins, cell and food debris.

With the new knowledge of tube position-dependent INCB000928 concentration variation (i.e., inhomogeneous concentration) in the saliva matrix, the corrective measure to obtain accurate and reliable concentrations for saliva study sample was to individually vortex each sample tube followed by immediate aliquot transfer to the extraction tube. With appropriate justifications and approval from the management and clinical pharmacokineticist, the study samples were reassayed with revised sample process procedures. ISR was repeated by selecting the identical 36 samples from three cohorts analyzed in the previous ISR run. As expected, the new revised sampling procedure corrected the initial ISR failure, with 97.2% of samples (35/36) showing repeated concentrations within ±20% of the mean from the reassayed values, adequately passing the ISR acceptance criteria of >67% of the repeats (Table 6).

Tube position-dependent substantial analyte concentration variation is compound-specific

The observation of position-dependent substantial INCB000928 concentration variation in saliva sample tubes was intriguing. To the best of our knowledge, similar observations have not been reported for other compounds in this laboratory or in the literature. It was interesting to explore whether the position-dependent substantial concentration variation was compound-specific. To answer this question, two other compounds, INCB018424 and INCB050465, were chosen for evaluation, with the understanding that the clinical relevance of their presence in saliva was unknown at the time of manuscript preparation. Using the same experimental design depicted in Figure 3, the concentration denoted in peak area ratio of the low QC sample in HWS was determined at the upper, middle and bottom positions in the tube for INCB018424 and INCB050465. Similar to what was observed for INCB000928, the concentrations were close between the upper and middle positions for both INCB018424 and INCB050465. Different from INCB000928, both INCB018424 and INCB050465 showed only slightly higher concentrations at the bottom of the tube compared with the upper and middle positions (Figure 5A & B). Meanwhile, the NSB properties were investigated for INCB018424 and INCB050465 in HWS using the same sequential transfer approach as for INCB000928. Interestingly, both INCB018424 and INCB050465 showed negligible NSB over each sequential tube transfer, with approximately 10% loss after four sequential transfers (Figure 5C & D). It was conceivable that the mildly elevated concentrations at the bottom were associated with the insignificant NSB of INCB018424 and INCB050465 in HWS. The correlation between scales of tube positional concentration variations and severity of NSB in HWS is worth further investigation and will be reported in the future.

Conclusion

This study demonstrated that HWS reduced the NSB of INCB000928 to polypropylene sample storage containers for a more accurate and reproducible concentration determination and was a preferred matrix for INCB000928 bioanalytical method validation and sample analysis. It was determined that the substantial tube position-dependent concentration variation and the batch-mixing practice led to initial ISR failure in a multiple ascending-dose healthy volunteer study. Individual mixing of each saliva tube followed by immediate sample transfer corrected the ISR failure. The extent of position-dependent analyte concentration variation was compound-specific. The experience and knowledge gained during the method development, validation and sample analysis for a healthy volunteer study ensures the reliability of the method for successful pharmacokinetic support in patients with FOP and are useful for future method validation and sample analyses. This work demonstrated significant tube position-dependent INCB000928 concentration variation in human saliva and proposed a unique sample process procedure to ensure accurate and reproducible concentration determinations in small-molecule bioanalysis in saliva matrix.

Future perspective

Saliva is an attractive and alternative specimen to plasma and is gaining interest in therapeutic drug monitoring and pharmacokinetics. The experience and knowledge acquired in comparing HWS and HCS for appropriate matrix selection for bioanalytical method validation and in investigating and solving the ISR failure are useful for the establishment of robust saliva bioanalytical methods and successful in-study conduct in the future.

Table 1. Sequential tube transfer results from INCB000928 in human centrifuged saliva and human whole saliva.

Sequential transfer numbers (T#)^†	T0	T1	T2	T3	T4
Low QC nominal concentration, nM	15	15	15	15	15
Human centrifuged saliva
Replicate 1, nM	14.2	14.3	13.6	11.6	9.54
Replicate 2, nM	14.2	13.3	13.3	11.2	9.59
Mean, nM	14.2	13.8	13.5	11.4	9.57
% bias	-5.3	-8.0	-10.0	-24.0	-36.2
n	2	2	2	2	2
Human whole saliva
Replicate 1, nM	14.8	15.0	13.6	13.1	12.5
Replicate 2, nM	15.7	13.7	14.3	13.4	11.8
Mean, nM	15.3	14.4	14.0	13.3	12.2
% bias	2.0	-4.0	-6.7	-11.3	-18.7
n	2	2	2	2	2

† Transfer # is defined as follows: T0 indicates no transfer (i.e., initial prepared solution in the first tube); T1 indicates the initial prepared solution was transferred to a second tube (transfer 1); T2 indicates the solution from the second tube (transfer 1) was transferred to a third tube (transfer 2); T3 indicates the solution from the third tube (transfer 2) was transferred to a fourth tube (transfer 3); T4 indicates the solution from the fourth tube (transfer 3) was transferred to a fifth tube (transfer 4).

QC: Quality control.

Table 2. Performance of INCB000928 quality control samples in human centrifuged saliva and human whole saliva.

QC	Human centrifuged saliva
	Dilution QC (5000 nM)	LQC (15 nM) Bench-top (26 h)	LQC (15 nM) Freeze–thaw (three cycles)	HQC (2400 nM) Bench-top (26 h)	HQC (2400 nM) Freeze–thaw (three cycles)
Replicate 1	4220	12.8	12.7	2110	2160
Replicate 2	3930	12.8	12.5	2110	2100
Replicate 3	4070	12.9	12.3	2070	2110
Replicate 4	4200	12.6	12.3	2110	2100
Replicate 5	3580	12.8	12.9	2100	2140
Replicate 6	3540	12.3	12	2160	2060
Mean	3920	12.7	12.5	2110	2110
SD	300	0.219	0.321	29	34.9
% CV	7.7	1.7	2.6	1.4	1.7
% bias	-21.6	-15.3	-16.7	-12.1	-12.1
n	6	6	6	6	6
QC	Human whole saliva
	Dilution QC (5000 nM)	LQC (15 nM) Bench-top (47 h)	LQC (15 nM) Freeze–thaw (five cycles)	HQC (2400 nM) Bench-top (47 h)	HQC (2400 nM) Freeze–thaw (five cycles)
Replicate 1	4420	14.3	14	2380	2350
Replicate 2	4170	14.5	13	2320	2500
Replicate 3	4310	13.9	13.1	2330	2310
Replicate 4	4250	13.1	12.4	2300	2450
Replicate 5	4450	13.3	13.9	2360	2360
Replicate 6	4270	13.8	13.8	2290	2500
Mean	4310	13.8	13.4	2330	2410
SD	106	0.546	0.635	34.6	82.3
% CV	2.5	4	4.7	1.5	3.4
% bias	-13.8	-8	-10.7	-2.9	0.4
n	6	6	6	6	6

HQC: High quality control; LQC: Low quality control; QC: Quality control.

Table 3. Summary of validation results.

Validation parameters		Validation results
Back-calculated calibration standards	Inter-assay accuracy (% bias)^†	-2.5 to 4.4
	Inter-assay precision (% CV)^†	≤3.6
Accuracy (% bias)	Intra-assay^‡	-8.7 to 5.6
	Inter-assay^‡	-3.9 to 0.4
Precision (% CV)	Intra-assay^‡	≤3.9
	Inter-assay^‡	≤4.8
Recovery (%)		91.9–97.3 (% CV 1.2–3.0)
Matrix factor		0.95–0.965
Internal standard-normalized matrix factor		0.992–0.999 (% CV 1.4–2.0)
Blank selectivity: INCB000928 (% response, six individual lots)^§		0.1–0.2
Blank selectivity: D6-INCB000928 (% response, six individual lots)^§		0.0
Bench-top stability (47.5 h)	Accuracy (% bias)	-8.0 to -2.9
	Precision (% CV)	≤4.0
Freeze–thaw stability (five cycles)	Accuracy (% bias)	-10.7 to 0.4
	Precision (% CV)	≤4.7
Long-term stability (-20°C, 62 days)	Accuracy (% bias)	-6.7 to -5.8
	Precision (% CV)	≤2.8
Long-term stability (-70°C, 62 days)	Accuracy (% bias)	-5.3 to -2.9
	Precision (% CV)	≤2.4
Long-term stability (-70°C, 175 days)	Accuracy (% bias)	-6.0 to -2.1
	Precision (% CV)	≤1.9
Processed sample stability (102 h at 15°C)	Accuracy (% bias)	-2.7 to -2.1
	Precision (% CV)	≤2.4
Reinjection reproducibility (139 h at 15°C)	Accuracy (% bias)	-4.4 to -1.3
	Precision (% CV)	≤3.0

† From all calibration standards obtained during the validation study at the time of manuscript preparation.

‡ From three independent accuracy and precision runs.

§ % response = (peak area of INCB000928 or D6-INCB000928 in control blank plasma/peak area of INCB000928 or D6-INCB000928 in lower limit of quantification sample) × 100.

Table 4. Back-calculated INCB000928 concentrations of calibration standards and performance of quality control samples in a clinical study.

	Calibration standards (5–3000 nM)								QCs (15, 180 and 2400 nM)
Concentration (nM)	5	12	50	120	500	1200	2500	3000	15	180	2400
Run 1	5.46	12.4	51	117	543	1190	2510	2940	15.3	186	2360
	4.67	10.6	53.7	113	525	1170	2510	2840	13.5	182	2320
Run 2	5.28	11.8	51.8	111	541	1140	2550	2900	14.2	183	2330
	4.79	11.7	51.2	116	552	1150	2540	2950	13.6	172	2280
Run 3	7.82^†	12.7	49.5	117	495	1210	2620	2960	14.3	166	2380
	7.47^†	11.5	49.7	112	490	1170	2710	3080	14.4	170	2310
Run 4	4.96	12	50.7	112	498	1190	2460	3030	15.4	177	2330
	4.89	12.9	50.1	113	539	1170	2520	3030	15.3	174	2320
n	6	8	8	8	8	8	8	8	8	8	8
Overall mean	5.01	12	51	114	523	1170	2550	2970	14.5	176	2330
SD	0.302	0.735	1.35	2.42	24.9	22.6	78.1	77.8	0.76	6.98	30.4
% CV	6	6.1	2.6	2.1	4.8	1.9	3.1	2.6	5.2	4	1.3
% bias	0.2	0	2.0	-5.0	4.6	-2.5	2.0	-1.0	-3.3	-2.2	-2.9

† >15% difference from nominal and excluded from statistic calculations.

QC: Quality control.

Table 5. Initial incurred sample reanalysis results.

Participant	Day	Hour	Original concentration (nM)	ISR concentration (nM)	% Diff
100103	10	1	320	762	81.7
100103	10	2	317	531	50.5
100104	10	4	468	1660	112.0
100106	10	8	787	2700	109.7
100107	10	0.5	394	906	78.8
100111	10	6	1010	1470	37.1
100102	13	72	433	589	30.5
100102	14	96	35.9	77.4	73.3
100105	10	16	965	2110	74.5
100105	11	24	696	817	16.0
100110	10	12	835	1980	81.3
100111	10	16	1660	AQL-3230>(3000)	64.2^†
100203	10	0	162	312	63.3
100209	10	0.5	1340	1790	28.8
100210	10	1	1540	1770	13.9
100210	10	2	1480	1710	14.4
100208	10	4	1390	1210	-13.8
100212	10	6	1240	1100	-12.0
100204	10	12	1770	1070	-49.3
100201	10	16	1070	1240	14.7
100205	11	24	703	AQL-4510>(3000)	146.1^†
100205	12	48	369	1550	123.1
100209	13	72	249	289	14.9
100206	14	96	192	555	97.2
100309	14	96	329	728	75.5
100301	13	72	481	425	-12.4
100303	12	48	1090	1330	19.8
100302	10	0.5	1660	2270	31.0
100305	10	2	1410	1120	-22.9
100306	10	4	2910	2480	-16.0
100309	10	1	1950	2410	21.1
100310	10	0.5	2820	2620	-7.4
100301	11	24	1850	1920	3.7
100306	10	8	3350	3760	11.5
100310	10	6	3490	3570	2.3
100312	10	16	2890	2580	-11.3

† Calculated using the extrapolated concentration.

% Diff = 100 × (ISR – original)/mean.

% Diff: Percentage difference; AQL: Above upper quantification limit; ISR: Incurred sample reanalysis.

Table 6. Repeated incurred sample reanalysis results.

Participant	Day	Hour	Reassayed original concentration, nM	Repeated ISR concentration, nM	% Diff
100103	10	1	280	272	-2.9
100103	10	2	278	283	1.8
100104	10	4	479	454	-5.4
100106	10	8	670	582	-14.1
100107	10	0.5	391	378	-3.4
100111	10	6	593	559	-5.9
100102	13	72	289	280	-3.2
100102	14	96	38.4	46.2	18.4
100105	10	16	348	328	-5.9
100105	11	24	268	257	-4.2
100110	10	12	292	290	-0.7
100111	10	16	906	894	-1.3
100203	10	0	152	159	4.5
100209	10	0.5	1020	841	-19.2
100210	10	1	1460	1450	-0.7
100210	10	2	1240	1250	0.8
100208	10	4	1300	1290	-0.8
100212	10	6	1310	1180	-10.4
100204	10	12	1260	1360	7.6
100201	10	16	1000	1010	1.0
100205	11	24	775	742	-4.4
100205	12	48	365	392	7.1
100209	13	72	252	250	-0.8
100206	14	96	136	140	2.9
100309	14	96	239	213	-11.5
100301	13	72	359	449	22.3
100303	12	48	253	287	12.6
100302	10	0.5	949	923	-2.8
100305	10	2	3470	3350	-3.5
100306	10	4	3280	3450	5.1
100309	10	1	1580	1610	1.9
100310	10	0.5	2820	2900	2.8
100301	11	24	2000	1900	-5.1
100306	10	8	2780	2670	-4.0
100310	10	6	3250	3350	3.0
100312	1	16	2250	2460	8.9

% Diff = 100 × (ISR-Original)/Mean.

% Diff: Percentage difference; ISR: Incurred sample reanalysis.

Executive summary

• Human whole saliva reduced INCB000928 nonspecific binding versus centrifuged saliva.

• Human whole saliva is a preferred matrix for INCB000928 bioanalytical method validation.

• Incurred sample reanalysis (ISR) failed in a healthy volunteer study using a successfully validated method.

• ISR failure was due to inhomogeneous INCB000928 concentration in saliva sample tubes.

• Individual mixing of each sample tube followed by immediate sample transfer corrected the inhomogeneity and ISR failure.

• This is the first report on comparison of human whole saliva and human centrifuged saliva for small molecule bioanalysis.

• This is the first report on ISR failure caused by inhomogeneous concentration distribution across saliva sample tube depths.

• Experience and knowledge gained in the study are useful for the establishment of robust saliva bioanalytical methods and successful in-study conduct in the future.

Acknowledgments

The authors are thankful to site personnel and all volunteers who participated in the clinical trial and contributed samples for analysis. The authors are thankful to K Rockich for being responsible for authoring the protocol and providing clinical pharmacology support for clinical study INCB 00928-102.

Financial & competing interests disclosure

All the authors are employees and shareholders of Incyte Corporation. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Editorial assistance was provided by Envision Pharma Group (PA, USA) and funded by Incyte Corp. (DE, USA).

Data sharing statement

No individual participant data that underlies the results reported in the article will be available, and nor will information regarding the study protocol or statistical analysis, beyond what is reported in the text.

Ethical conduct of research

The authors state that they have obtained appropriate institutional review board approval or have followed the principles outlined in the Declaration of Helsinki for all human or animal experimental investigations. In addition, for investigations involving human subjects, informed consent has been obtained from the participants involved.