This was a randomized, three-way crossover, open-label study (ClinicalTrials.gov: NCT03292016) conducted at three sites in the US. The study was designed, conducted, and monitored in accordance with the World Medical Association Declaration of Helsinki (1989) and International Council for Harmonisation guidelines. The study protocol and patient informed consent form were approved by an institutional review board (Copernicus Group, IRB, reference number INC1-17-218 [Cary, NC, USA]).
Eligible patients were males or females ≥ 18 years of age with a clinical diagnosis of idiopathic PD, consistent with UK Brain Bank Criteria and modified Hoehn and Yahr scale stage ≤ 3 in the “ON” state. Patients had a clinically meaningful response to oral CD/LD with well-defined “OFF” episodes, as determined by the investigator. Patients were required to be receiving SC-APO at ≤ 5 mg per dose for ≥ 4 weeks and stable doses of CD/LD (immediate or sustained release) administered ≥ 4 times daily or extended-release CD/LD three times daily for ≥ 4 weeks. Adjunctive PD medications must have been maintained at a stable dose for ≥ 4 weeks except for monoamine oxidase-B inhibitors, which had to be maintained at a stable dose for ≥ 8 weeks. Patients must have had well-defined “OFF” episodes in the morning and must have been willing to delay morning doses on the three study dosing days.
Patients were excluded if they had atypical or secondary parkinsonism; major psychiatric disorders; previous treatment with continuous SC apomorphine infusion or CD/LD infusion; contraindications or hypersensitivity to SC apomorphine; received selective 5-HT3 antagonists, dopamine antagonists (excluding quetiapine and clozapine), or dopamine-depleting agents within the past 30 days; or mouth cankers or sores within the past 30 days.
During the study period, patients received three treatments with a minimum 1-day washout interval between each visit. Patients were randomized to an open-label sequence of single doses of SC-APO (APOKYN®; US WorldMeds, LLC, Louisville, KY, USA ) at their current prescribed dose, SC-APO-GO (APO-go® PEN; Britannia Pharmaceuticals Ltd, Berkshire, UK ) at the same dose as SC-APO, and apomorphine sublingual film (KYNMOBI®; Sunovion Pharmaceuticals Inc., Marlborough, MA, USA)  at doses believed to achieve similar plasma concentrations. The doses of apomorphine sublingual film for this study (15, 20, 25, and 30 mg) were chosen to evaluate the PK and comparative bioavailability of a range of proposed doses of SC-APO and SC-APO-GO and were based on results from an open-label, proof-of-concept, phase 2 study  and a phase 1 bioavailability study . The open-label, proof-of-concept study demonstrated that patients with PD and “OFF” episodes who received apomorphine sublingual film at doses of 10–30 mg achieved a clinically confirmed FULL "ON" with all responders turning FULLY "ON" between 15 and 30 min after administration (mean "ON" duration of 50 min) . The bioavailability study demonstrated that a 15-mg dose of apomorphine sublingual film is approximately equivalent to a 2-mg dose of SC-APO . In the present study, the dosing algorithm included four dose levels of apomorphine sublingual film, SC-APO, and SC-APO-GO (Table 1), and patients were randomized to one of six possible treatment sequences (Table S1 in the electronic supplementary material [ESM]).
Patients were required to not administer SC apomorphine for ≥ 1 day before each study period. Patients arrived in the clinic on each study day after their prescribed morning dose of PD medication but before taking their second dose of medication. After confirmation by the investigator that the patient was “OFF,” study drug was administered according to the randomized treatment sequence. Other PD medications were withheld until 60 min after study drug dosing.
The primary objective was to characterize the PK profile of apomorphine and its inactive metabolites, apomorphine sulfate and norapomorphine [21, 22]. Assessments of apomorphine PK parameters included Cmax, tmax, AUC from time 0 to 24 h (AUC0–24), AUClast, AUC∞, and t1/2. Apomorphine metabolite PK parameters included Cmax, tmax, AUClast, AUC∞, and metabolite-to-parent ratios of Cmax, AUClast, and AUC∞.
Comparative bioavailability among the three study drugs was also evaluated using dose-normalized Cmax, AUC0–24, AUClast, and AUC∞. Blood was collected for PK analyses at time point 0 (right before dosing) and at 0.25, 0.5, 0.75, 1, 1.5, 3, and 6 h postdose.
Safety and tolerability were assessed via evaluation of clinical laboratory tests, 12-lead electrocardiograms, physical examinations, vital signs, and treatment-emergent adverse events (TEAEs).
With a sample size of 12 patients, a two-sided 90% confidence interval for the difference in paired PK parameter means on the log scale would have had an interval that extended no more than 0.221 units from the observed difference with 90% coverage probability. After an unplanned interim analysis, it was determined that sufficient data were obtained from eight patients to meet the primary objective of the study; therefore, additional patients were not recruited.
The PK population included all patients who received ≥ 1 dose of study drug and had ≥ 1 quantifiable PK concentration. The safety population included all patients who received ≥ 1 dose of study drug.
Continuous variables were summarized using the number of observations; arithmetic mean; standard deviation (SD) of the arithmetic mean, geometric mean; and percent coefficient of variation of the geometric mean, median, and range.
PK parameters were derived using noncompartmental methods employing Phoenix WinNonlin version 6.3 (Certara, Princeton, NJ, USA). PK analysis was conducted using actual time elapsed from dosing concentration-time data for apomorphine, apomorphine sulfate, and norapomorphine; PK profiles had to have at least four quantifiable postdose concentrations to be included.
To analyze comparative bioavailability of the three study drugs, a mixed-effects model with fixed effects for treatment and period, with patients nested within sequence as a random effect, was used to analyze the natural log–transformed dose-normalized Cmax, AUC0–24, AUClast, and AUC∞. For each treatment comparison, a point estimate and 90% confidence interval were provided for the geometric mean ratio upon back-transformation.