Bioanalytical Methods

Two different analytical methods were used for the quantification of PSL in plasma and dried blood Mitra® samples (see supplementary (supp) Table 1 for more details on each method).

Plasma Samples

An analytical method for the determination of PSL in human plasma using solid phase extraction followed by liquid chromatography with tandem mass spectrometry using electrospray ionization in positive ion with multiple reaction monitoring mode was used. A 50 µL aliquot of plasma sample was pipetted into polypropylene tubes and spiked with 50 µL of a deuterated internal standard solution (500 ng/mL in water). After vortexing for 30 s, 400 µL of trifluoroacetic acid (0.5%) were added and the total volume of the mixture was loaded on an Oasis HLB 30-mg adsorbent material in a 96-well plate format for the solid phase extraction. At the end of extraction procedure, the supernatant was transferred (200 µl) to a clean 96-well plate sealed before subsequent LC/MS-MS analysis The method was validated according to FDA and EMA guidelines (17) and covers the concentration range of 0.5 to 500 ng/mL.

Mitra® Samples

Freshly prepared standards and quality control (QC) samples in human whole blood were aliquoted into the wells of a 2-mL 96-well plate (80 µL). Mitra 96-Autorack tips was used to adsorb the standards, QCs and blanks on Mitra tips. Standards and QCs loaded onto Mitra® tips were placed in a plastic 96-well drying rack provided by Neoteryx and allowed to dry overnight (at least 12 h, in closed box with desiccant) at room temperature.

The dried VAMS tips (Mitra® 10 µL) on the plastic holder were removed and placed into wells of a clean 96-well plate for sample extraction. The deuterated internal standard solution for PSL was added to the samples (150 µL, 80.0 ng/mL, formic acid (0.1%) in H20). The 96-well plate was sonicated for 10 min, vortexed (1 h at 600 r.p.m.). The Mitra tips were carefully removed from the 96-well plate and 950 µL of water with trifluoroacetic acid (0.5%) was added to further proceed with solid phase extraction (Thermo SOLA cartridges 10mg-96well plate).

At the end of extraction procedure, the supernatant was transferred (200 µl) to a clean 96-well plate sealed before subsequent LC/MS-MS analysis. The method was validated according to FDA and EMA guidelines (17) and covers the concentration range of 2.00 to 2000 ng/mL.

Collection of Paired Samples in Clinical Studies

Paired PK samples are time-matched PK samples collected in parallel using the two collection methods, i.e, conventional venous blood sampling and blood collection by finger-prick using Mitra® device. The exact date and time of sampling for both methods was recorded in the source document. Actual sampling times were used for calculation of PK parameters and nominal sampling times were used in plots.

Conventional Venous Blood Sampling

Blood obtained from venipuncture was collected in a labeled-vacuum tube containing Li-Heparin. Immediately following collection, the tubes were inverted 8–10 times. The blood was stored for maximum 1 h at room temperature until centrifugation (10 min at 1500 g at room temperature) to collect plasma for PK analysis.

Blood Sampling with Mitra® Device

Blood collection (finger prick) using Mitra was performed by clinical staff. The fingertip of the study participants is first cleaned with alcohol swab and then is allowed to dry. With the help of the lancet, a finger prick is made, and the Mitra® tip is applied on the surface of the blood drop to allow an accurate collection of a 10 µL blood sample (see Fig. 1). Once the tip turned fully red, the tip was allowed to air-dry and was then shipped to bioanalytical lab in a foil bag containing a desiccant.

As a control, in the first clinical study (UP0039) using Mitra®, a Mitra® tip was also dipped into the blood samples collected from venipuncture (before it was centrifuged to collect plasma).

Clinical Studies

Table I summarizes the key characteristics of the clinical studies from which the plasma and blood concentration data were obtained to establish the relationship between PSL plasma and blood exposure.

Table I Summary of Key Information on the Clinical Studies Used in the Analysis of the Relationship Between Blood Mitra® and Venous Plasma PSL Concentrations

Four phase 1 clinical trials were used to develop a statistical model to describe and predict PSL plasma concentrations from blood concentrations. Two datasets (one from phase 1 study and one from phase 2b clinical trial) were used to validate the statistical model.

UP0039 (n=40) was a placebo-controlled, phase 1, ethno-bridging clinical trial to evaluate the PK in Japanese and white healthy volunteers after a single (SD) and multiple doses (MD) of PSL (EudraCT no. 2017-000385-30). Three oral PSL dose levels (50, 100, and 200 mg) were studied in Asian participants after a single oral dose and one oral dose (200 mg) to White participants. Ten participants were treated per dose group. Conventional plasma PK samples were collected following single and multiple doses. Mitra® samples were only collected following single dose administration, and paired plasma/Mitra samples were obtained at 0.5, 1, 1.5, 2, 3, 4, 6, 12, 24, 36, and 48 h post-dose.

UP0057 (n=28) was a fixed sequence, phase 1 study in healthy adults, to evaluate the effect of coadministration of erythromycin on the PK of 100 mg Bis In Die (BID) PSL at steady state (EudraCT no. 2017-004694-13). Following single and multiple dose administration, paired PK samples (venous plasma and Mitra®) were collected at pre-dose and at 0.25, 0.5, 0.75, 1, 1.25, 1.5, 2, 3, 4, 6, 8, 12, 24, 36, 48, 72, 96, and 120 h post-dose. Data from both treatment arms were included in the statistical analyses (i.e., samples containing PSL alone or PSL and erythromycin).

UP0070 (n=14/group) was an open label, parallel-group phase 1 trial to evaluate the effect of oxcarbazepine on the PK of PSL in patients with epilepsy (EudraCT no. 2018-001941-16). Control group included patients with epilepsy under stable therapy with levetiracetam (or brivaracetam) and/or lamotrigine. Following attainment of PSL steady state on a multiple oral dose regimen of 400 mg BID (3 days up titration followed by 4.5 days 400 mg BID dosing regimen), PK samples were collected on day 8 at pre-dose and at 0.25, 0.5, 1, 1.5, 2, 3, 4, 6, 8 and 12h post-dose from both control and oxcarbazepine groups following both venipuncture and Mitra® techniques.

UP0075 (n=12) was a randomized, double-blind, placebo controlled, single dose, phase 1 study to evaluate the PK of PSL in healthy Chinese participants following a single oral dose of 200 mg (Investigational New Drug (IND) no. 135622). PK samples were collected using both venipuncture and Mitra® techniques at 0.25, 0.5, 0.75, 1, 1.25, 1.5, 2, 3, 4, 6, 8, 12, 24, 36, and 48 h post-dose. Additional details on this clinical study have been published elsewhere (16).

EP0091 (n=323 (maintenance period)) was a randomized, double-blind, placebo-controlled, dose-finding (50, 100, 200, 400 mg bid), phase 2b study to evaluate the efficacy and safety of PSL as adjunctive treatment of focal onset seizures in adult subjects with drug resistant epilepsy (NCT03373383, EudraCT no. 2017-003200-48). One of the exploratory objectives was evaluation of two blood sampling techniques (venipuncture and Mitra®). At prespecified visits during the trial, a Mitra® sample was collected, and contemporaneous venipuncture samples were collected whenever possible. The time of sampling relative to the dose was recorded, with 269 paired samples (obtained from 220 patients) collected in total. Additional Mitra® samples were collected across different clinical visits as primary source of PK data for population PK modeling and exposure-response analysis.

Data

For global overview and initial analysis of the data (linear regression and Bland-Altman), individual study datasets and a pooled dataset including all paired Mitra® blood and venipuncture plasma samples collected in the four clinical pharmacology studies (UP0039, UP0057, UP0070, and UP0075) were used. For statistical modeling using linear mixed-effect, paired Mitra® and plasma samples collected between 0 to 12 h post-dose (pre-dose and 0.25, 0.5, 1, 1.5, 2, 3, 4, 6, 8, and 12 h) have been included since PSL is used for chronic treatment (BID dosing regimen).

Concentration-Time Data and B/P Ratio Over Time

For each study, blood concentration-time and plasma concentration-time profiles were graphically displayed. For each time point, a B/P was calculated and its change over time was displayed on a linear scale, for each study.

Statistical Analysis

Statistical analysis was performed using SAS Software version 9.4. Drug concentration-time data collected from both methods, Mitra® and venous sampling, were summarized using descriptive statistics and graphical displays. Per individual study and pooled over all studies, blood-to-plasma linear regression plots were made, to reflect the correlation between blood and plasma concentrations. Additionally, Bland‐Altman plots were generated to display the level of agreement between plasma concentrations obtained using Mitra®versus those obtained using conventional venous sampling.

Bland‐Altman plots were developed by obtaining the mean of the two sampling methods (venipuncture and Mitra®) for each participant and plotting the mean concentration of PSL against the percent difference of the Mitra® concentration from the venipuncture concentration for each participant and each sampling time point. Corresponding Bland‐Altman percent agreement estimates were computed, and 95% confidence intervals (CI) were calculated overall. Presenting the differences in the Bland-Altman plot enabled an average discrepancy or bias to be estimated and confirmed if it was constant across the range of values observed.

A statistical model was developed based on the pooled data from the 4 clinical pharmacology studies, excluding the white study participants of UP0039 (n=10 white) who were used as independent dataset to validate the model. A mixed model was fitted to the log-transformed plasma PSL concentration, with (venipuncture sampling), log transformed blood PSL concentration (Mitra® sampling) as a fixed effect, and study participant as a random effect. Due to differences in study designs, parameters such as PSL dose levels (50 mg/100 mg/200 mg/400 mg), dosing schedule (Single/Multiple PSL dosing), sampling time in hour (0, 0.25, 0.5, 1, 1.5, 2, 3, 4, 6, 8, and 12 h), race (white/black or African American/Asian), study population (healthy/patient), baseline hematocrit level (0–100%), gender (female, male), and age (numeric) potentially affected the blood-to-plasma relationship. Variance components, covariance structure and containment degrees of freedom were used to assess the effects of the sampling times, dosing schedule, PSL dose level, race and population. Covariates with backward elimination method at 0.10 significance level were excluded from the model one by one. In this approach, time-matched blood and plasma concentration data pairs were fitted to a mixed-effects model (e.g., log10(blood) = log10 (plasma)+ covariates). Inter- and intra-individual variability terms were evaluated as well as the influence of different covariates. The final model parameters were selected based on the Akaike Information Criterion (AIC) score (smaller is better) and complexity of the model (simpler is better) (Table II).

Table II Summary of the Regression Models and Covariates Evaluated and the AIC Score.

The selected statistical model was validated using data from the white participants from UP0039. With the statistical model, the plasma concentrations of white participants (UP0039) were predicted from the blood Mitra® data. Subsequently, plasma PK parameters (area under the curve (AUC) and Cmax), were calculated by non-compartmental analysis (NCA) (using SAS 9.4) using the predicted plasma concentrations. Least-square (LS) geometric mean ratio of predicted plasma PK parameters over actual plasma PK parameters of UP0039 (White cohort) as well as 90% confidence intervals was calculated. Similar to the approach used for logarithmically transformed data in bioequivalence study, an acceptance criterion for confidence intervals of 80–125% was applied to ensure that systemic exposure determined with the 2 approaches (prediction based on model and observed data) were not different using a range of exposure of ±20%.

Following the evaluation of the statistical model, an external validation of the model was conducted using data from EP0091, the pivotal phase 2b efficacy trial. During the trial, at some pre-selected sites, a limited number of patients were sampled to collect time-matched blood and plasma samples at steady state. Predicted plasma results vs actual plasma results were compared with regression and Bland-Altman analysis.