A randomized, single‐dose, crossover study of the effects of ulotaront on electrocardiogram intervals in subjects with schizophrenia

Abstract This study (NCT04369391) evaluated the effects of ulotaront (SEP‐363856), a novel trace amine‐associated receptor 1 (TAAR1) agonist in development for schizophrenia, on electrocardiogram parameters. Study design was a randomized, single‐dose, three‐period crossover (ulotaront 150 mg, placebo, moxifloxacin 400 mg). Sixty subjects with schizophrenia completed all periods. Ulotaront had no clinically relevant effect on heart rate, PR interval, or QRS duration. In by‐time‐point analysis (secondary analysis), the upper bound of the two‐sided 90% confidence interval for ΔΔQTcF (QT interval corrected for heart rate using Fridericia's formula) was below 10 ms at all time points for ulotaront. In concentration‐QTc analysis (primary analysis), a linear mixed‐effects model with ulotaront and its major metabolite SEP‐383103 was selected as the primary model based on prespecified criteria. Effect on ∆∆QTcF exceeding 10 ms can be excluded within observed ranges of ulotaront and SEP‐383103 plasma concentrations up to ~574 and ~272 ng/mL, respectively. The upper bound of 90% CI for ΔΔQTcF can be predicted to be below 10 ms at the highest anticipated clinical exposure, currently defined as steady‐state mean C max at ulotaront 100 mg/day in CYP2D6 poor metabolizers, ~416 and ~211 ng/mL for ulotaront and SEP‐383103, respectively. Assay sensitivity was demonstrated by the QTc effect caused by moxifloxacin. In conclusion, ulotaront is unlikely to cause clinically relevant QTc prolongation in patients with schizophrenia at the anticipated maximum therapeutic dose.


INTRODUCTION
Ulotaront (SEP-363856) is a trace amine-associated receptor 1 (TAAR1) agonist with 5-hydroxytryptamine type 1A (5-HT 1A ) receptor agonist activity in clinical development as a novel treatment for patients with schizophrenia. [1][2][3][4][5][6][7][8][9][10][11][12][13] Unlike the antipsychotic class, ulotaront does not mediate its effects via blockade of dopamine D2 or 5-HT 2A receptors. 4,5 In a randomized, double-blind, placebo-controlled, 4week phase II clinical trial enrolling 245 patients with an acute exacerbation of schizophrenia, ulotaront demonstrated a statistically significant improvement in change from baseline in the Positive and Negative Syndrome Scale (PANSS) total score compared to placebo. 10 In addition, in a 26-week, open-label extension study for subjects who completed the initial 4-week double-blind study, ulotaront was shown to be well-tolerated and effective in the long-term treatment of patients with schizophrenia. 3 A population pharmacokinetic (PK) analysis of ulotaront 6 was conducted using data from 7 phase I and 2 phase II studies. Following oral dosing, ulotaront is absorbed quickly, with median T max of 2.8 hours, and clears rapidly from plasma, with a median effective half-life of 7 hours. Ulotaront has a low accumulation ratio of 1.10 when administered once daily.
Multiple metabolites of ulotaront have been identified from preclinical and clinical studies. 2,13 SEP-383103 was the only major metabolite detected in the plasma of both preclinical and clinical studies. SEP-363854 was identified as an abundant metabolite from preclinical studies but as a minor metabolite in humans. The results of in vitro receptor binding studies indicated that major metabolite SEP-383103 is not pharmacologically active, but minor metabolite SEP-363854 does bind to TAAR1. Based on the effective half-life for each metabolite, there is no concern regarding metabolite accumulation upon repeated daily dosing of ulotaront.
This study aimed to evaluate the effects of orally administered ulotaront on electrocardiogram (ECG) parameters in adults with schizophrenia in accordance with the International Council for Harmonisation (ICH)-E14 guideline 14 and its Questions and Answers supplement (R3). 15 The effect of ulotaront on placeboadjusted change from baseline in QTc interval (ΔΔQTc) served as the primary endpoint using concentration-ΔΔQTc modeling of concentrations of ulotaront and its metabolites SEP-363854 and SEP-383103. 16,17 The intent was to determine whether ΔΔQTc >10 ms at the upper bound of the two-sided 90% confidence interval (CI) could be excluded at clinically relevant plasma levels of each analyte.
A single dose of 150 mg ulotaront was considered the highest practical dose for patients with schizophrenia, balancing thorough QT (TQT) study requirements with patient safety. Although TQT studies are typically performed in healthy subjects, 14 this study was conducted in the target patient population (i.e., patients with schizophrenia) because 150 mg ulotaront has not been tested in healthy volunteers. For certain drug classes, such as antipsychotics, tolerability has been reported as greater in subjects with schizophrenia than healthy subjects. 18 The dose of 150 mg ulotaront was 1.5 times the maximum dose of 100 mg that is being evaluated in phase III clinical trials. Median exposure of the 10th (TAAR1) agonist and does not block dopamine D2 or serotonin 5-HT 2A receptors. The effects of ulotaront on QT/QTc interval were unknown.

WHAT QUESTION DID THIS STUDY ADDRESS?
This study was designed to evaluate potential effects of the highest practical dose of ulotaront on QT/QTc interval in subjects with schizophrenia.

WHAT DOES THIS STUDY ADD TO OUR KNOWLEDGE?
This study showed ulotaront is unlikely to cause clinically relevant QTc prolongation in subjects with schizophrenia at the anticipated maximum therapeutic dose.

HOW MIGHT THIS CHANGE CLINICAL PHARMACOLOGY OR TRANSLATIONAL SCIENCE?
This study provides evidence of cardiovascular safety of ulotaront, a novel TAAR1 agonist in development, which has been granted US Food and Drug Administration-Breakthrough Therapy designation for the treatment of patients with schizophrenia.
decile of a single dose of 150 mg ulotaront was estimated to be approximately two times higher than the geometric mean steady-state C max at the 100 mg dose level based on population PK simulation.

Study design
This study (NCT04369391) was conducted in accordance with the International Conference on Harmonisation Good Clinical Practices guidelines, the ethical principles of the Declaration of Helsinki, and applicable local law(s) and regulation(s). Study protocol and informed consent form (ICF) were approved by the Institutional Review Board (CGIRB/1019 39th Avenue Southeast Suite 120 Puyallup, Washington 98,374) before enrollment of any subjects into the study.
This study was a phase I, randomized, single-dose, active-and placebo-controlled, three-period crossover study of the effects of ulotaront 150 mg on ECG intervals in subjects with schizophrenia. Ulotaront and matching placebo were utilized in a double-blind fashion. Moxifloxacin was utilized as an active control in an open-label fashion. Subjects were enrolled from June 2020 to November 2020 at five sites in the United States. Prior to Period 1, subjects were randomly assigned on day 1 to one of the six sequences per William square design, in which ulotaront 150 mg, placebo, and moxifloxacin 400 mg were administered in different order from Period 1 through Period 3. Additional details are provided in the supplementary materials.

Subjects
Eligible subjects were males or females aged 18-65 years with a primary diagnosis of schizophrenia (as defined in the Diagnostic and Statistical Manual of Mental Disorders [DSM-5]), with a PANSS total score ≤ 80 and a Clinical Global Impressions Scale (CGI-S) score of ≤4; a body mass index of 18-35 kg/m 2 ; and were clinically stable for the past 3 months in the opinion of the Investigator at Screening.
Exclusion criteria included a DSM-5 diagnosis other than schizophrenia; any clinically significant unstable medical condition or any clinically significant chronic disease; any psychotropic medication (except those specified as rescue medications); and any other medications with a propensity for psychotropic effects within 3 days or five half-lives (whichever is longer) prior to day −1.

Cardiodynamic ECG assessments
The primary endpoint was placebo-adjusted changefrom-baseline QTc interval (ΔΔQTc). The default was QTcF (QTc interval corrected for heart rate using the Fridericia method). If a substantial HR effect was observed (i.e., the largest least squares mean ΔΔHR is greater than 10 beats per minute (bpm) in the bytime-point analysis), 16 other correction methods were planned, including an optimized HR-corrected QT interval (QTcI) and an individualized HR-corrected QT interval (QTcS). Secondary endpoints included changefrom-baseline HR, QTc (QTcF and QTc corrected with other methods if a substantial HR effect was observed), PR and QRS intervals (∆HR, ΔQTc, ΔPR, and ∆QRS); placebo-adjusted ΔHR, ΔPR, ΔQRS, and ΔQTc if not selected as the primary endpoint in case that a substantial HR effect was observed (ΔΔHR, ΔΔPR, ΔΔQRS, and ΔΔQTc); categorical outliers for QTc, HR, PR, and QRS; and frequency of treatment-emergent changes for T-wave morphology and U-wave presence. Additional details are provided in the supplemental materials.

Concentration-QTc analysis (primary analysis)
The relationship between plasma concentrations of ulotaront and its metabolites SEP-363854 and SEP-383103 and change-from-baseline QTc (ΔQTcF or ΔQTc corrected with the method chosen as primary if a substantial HR effect was observed) was quantified using a linear mixed-effects modeling approach. A full model was considered with ΔQTc as the dependent variable, plasma concentrations of ulotaront, SEP-363854, and SEP-383103 as the explanatory variates (0 for placebo), centered baseline QTc (i.e., baseline QTc for individual subject minus the population mean baseline QTc for all subjects in the same period) as an additional covariate, treatment (active = 1 or placebo = 0), time (i.e., nominal time point) as fixed effects, and random intercept and slopes per subject. 16 From the model, the slopes (i.e., the regression parameters for SEP-363856, SEP-363854, and SEP-383103 concentrations) and the treatment effect-specific intercept (defined as the difference between active and placebo) were estimated together with the two-sided 90% CI. Additional details are provided in the supplemental materials.
Assay sensitivity would be demonstrated if the slope of the concentration-QTc (∆QTc) for moxifloxacin using a linear mixed-effects model was statistically significant at a 10% level for a two-sided test and the lower bound of the two-sided 90% CI of the predicted effect of ΔΔQTc was above 5 msec at the observed geometric mean C max .

By-time-point analysis (secondary analysis)
The analysis for QTc was based on a mixed model for repeated measures with ΔQTc as the dependent variable, period, sequence, time (i.e., categorical post-dose time point), treatment (ulotaront, moxifloxacin, and placebo), and time-by-treatment interaction as fixed effects, and baseline QTc as a covariate. An unstructured covariance matrix was specified for the repeated measures at postdose time points for subject within treatment period. From this analysis, the least-squares (LS) mean and two-sided 90% CI of ΔΔQTc was calculated for the contrast "ulotaront versus placebo" and "moxifloxacin versus placebo" at each post-dose time point, separately. Similar analysis models were also used on HR, PR, and QRS intervals.

Concentration-QTc analysis (Emax model)
The relationship between plasma concentrations of ulotaront and ΔQTcF was also investigated by an Emax model using an Emax function to replace the linear function of the plasma concentration of ulotaront in the abovementioned linear mixed-effect model. Additional details are provided in the supplemental materials.

Population pharmacokinetic analysis
As a reference point, the geometric mean of steady state C max at the anticipated maximum therapeutic dose of 100 mg was simulated by population PK modelling 6 of ulotaront using 1000 virtual schizophrenia subjects. The population PK model used for simulations included weight, patient status (healthy volunteer vs. patient with schizophrenia), Asian ethnicity, and gender as covariates on clearance; weight alone was included as a covariate on apparent central volume of distribution. This model included all available studies with the exclusion of the studies which had not been locked at the time of protocol development of this study.

Safety assessments
Safety and tolerability were assessed by clinical laboratory tests, vital signs, ECGs, physical examinations, Columbia-Suicide Severity Rating Scale, and monitoring of adverse events. Adverse events and serious AEs (SAEs) were monitored throughout the study at all visits.

Sample size calculation
A total of 68 subjects were randomized to obtain at least 50 evaluable subjects who complete the study. Using the concentration-QTc analysis method, 19,20 a sample size of 50 evaluable subjects provided more than 90% power to exclude that ulotaront causes more than a 10 msec QTc effect at clinically relevant plasma levels, as shown by the upper bound of the two-sided 90% CI of the model predicted QTc effect (∆∆QTc) at the observed geometric mean C max of ulotaront in the study. This power was estimated using a paired t-test. The calculation assumed a one-sided 5% significance level, a small underlying effect of ulotaront of 3 ms, and a standard deviation (SD) of the ΔQTc of 12 msec for both ulotaront and placebo.

Demographics
A total of 68 subjects were randomized, 66 subjects received the first dose of study drug, and 60 subjects (88.2%) completed the study. Among the 66 subjects in the safety analysis population, mean age was 48.3 years (range 24-64 years). Fifty-two (78.8%) subjects were male and 14 (21.2%) subjects were female. Mean body weight was 84.21 kg (range 57.8-122.8 kg). Mean PANSS total score was 57.5 (range 35-80) and mean CGI-S score was 2.9 (range 1-4) at baseline. Mean time since initial onset of schizophrenia was 24.72 years (Q1, Q3: 16.1, 33.1). Eight (11.8%) subjects discontinued the study. The reasons for discontinuation were withdrawal by the subject and protocol deviation for 3 (4.4%) subjects each and AE for 2 (2.9%) subjects. The 66 subjects who received the first study drug dose were included in the Safety, QTc, and PK/QTc Analysis Populations. Two subjects discontinued after receiving only placebo and therefore were not included in the PK Analysis Population.

Effect on heart rate
Ulotaront at the studied dose did not have a clinically relevant effect on HR. LS mean change-from-baseline HR (ΔHR) and LS mean placebo-adjusted ΔHR (ΔΔHR) across post-dose time points are shown in Figure 1. The largest LS mean ΔΔHR was below 5 bpm, demonstrating that ulotaront had no substantial HR effect. Based on the lack of meaningful HR effect, QTc using the Fridericia correction (QTcF) was chosen as primary endpoint. Additional HR correction methods, QTcI and QTcS, were therefore not applied.

By-time-point analysis
After dosing with ulotaront, the upper bound of the twosided 90% CI for ΔΔQTcF values was below 10 msec at all time points for ulotaront with the largest increase in mean ΔΔQTcF (6.7 msec; 90% CI: 3.98-9.50) at 2.5 h post-dose ( Figure 2 and Table 1). After dosing with moxifloxacin, the upper bound of the two-sided 90% CI for ΔΔQTcF values was above 10 msec, with the largest increase in mean ΔΔQTcF (9.1 msec; 90% CI: 6.35-11.88) at 2.5 h post-dose ( Figure 2 and Table 1).

Concentration-QTc analysis
The time courses of mean plasma concentrations of ulotaront, SEP-363854, and SEP-383103 following treatment with 150 mg ulotaront are presented in Figure S1. The highest mean plasma concentrations of ulotaront, SEP-363854, and SEP-383103 were observed at 5, 6, and 6 hours post-dose for the 150 mg ulotaront treatment group, respectively. Geometric mean C max of ulotaront, SEP-363854, and SEP-383103 were 458 ng/mL, 11.3 ng/ mL, and 213 ng/mL, respectively. Concentration of ulotaront ranged up to 835 ng/mL ( Figure S2) which covers clinically relevant exposure with approximately 2.6-fold above the estimated geometric mean C max at the anticipated maximum therapeutic dose of 100 mg (318 ng/mL). Geometric mean t 1/2 of ulotaront, SEP-363854, and SEP-383103 were 6.28 h, 7.4 h, and 10.0 h, respectively.
The time courses of mean plasma concentrations versus LS mean ∆∆QTc are shown in Figure S2. The patterns of mean plasma concentration versus LS mean ∆∆QTc were similar for all three analytes, with the largest mean ΔΔQTcF occurring before the highest mean plasma concentration, demonstrating that there is no delayed effect (i.e., there is an absence of hysteresis).
The scatter plot of concentrations of ulotaront, SEP-363854, and SEP-383103 versus ΔQTcF with a linear regression line and a locally weighted scatter plot smoothing (LOESS regression line) is given in Figure S2. The regression slopes were shallow for all three analytes. The linear regression line and LOESS regression line were close across the observed concentrations for ulotaront and SEP-383103, falling within or slightly above the 90% CI of the LOESS line, except for SEP-363854 at high concentration levels due to sparse observations. This indicates that a linear model for the concentration-QTc relationship may be appropriate for ulotaront and SEP-383103, but may not be the case for SEP-363854.
The relationships between plasma concentrations and ΔQTcF were investigated using seven linear mixed-effects models (Table S1). Treatment effect-specific intercepts for all seven models were not statistically significant (absolute t-value <1.95). Model C with ulotaront and SEP-383103 as analytes was chosen as the primary model, since it had the smallest Akaike Information Criterion (AIC) value among these seven models. An assessment of the adequacy of the linear mixed-effects model was given by the goodness-offit plots for Models E (ulotaront alone) and G (SEP-383103 alone) in Figure 3. These plots indicated that predicted F I G U R E 1 Change-from-baseline heart rate (ΔHR, left panel) and placebo-adjusted ΔHR (ΔΔHR, right panel) across time points. bpm, beats per minute; CI, confidence interval; Δ, change-from-baseline; ΔΔ, placebo-adjusted change-from-baseline; HR, heart rate; LS, least squares. LS mean and 90% CI based on a linear mixed-effects model: ΔHR = Period + Sequence + Time + Treatment + Treatment × Time + Baseline HR.  ΔΔQTcF values are close to the observed mean ΔΔQTcF across most of the plasma concentrations. Therefore, it can be concluded that the observed data are captured in an acceptable way for each of the analytes, ulotaront and SEP-383103, separately. The estimated slopes of ulotaront and SEP-383103 were shallow: 0.011 and 0.0090 ms per ng/mL, respectively, with a small, not statistically significant treatment-effect intercept of −1.03 ms (Table S1).
Using the primary model with ulotaront and SEP-383103 as analytes, the effect on ∆∆QTcF can be predicted to 5.17 ms (90% CI: 2.98-7.36) and 4.44 ms (90% CI: 1.59-7.30) at the geometric mean C max of ulotaront (458 ng/mL) and of SEP-383103 (213 ng/mL), respectively, also accounting for the effect of SEP-383103 and ulotaront at the same time points (Table 2). Prediction results from other linear mixed-effects models were similar to those from the primary model. The effect on ∆∆QTcF exceeding 10 ms can be excluded at plasma concentrations of ulotaront, SEP-363854, and SEP-383103 up to ~574, 15.3, and ~ 272 ng/mL, respectively.
A similar linear mixed-effects model with a treatment effect-specific intercept was used in the assay sensitivity analysis for moxifloxacin. The slope of the relationship was positive and statistically significant: 0.0044 ms per ng/mL (90% CI: 0.00299-0.00585; p < 0.0001) with a treatment effect-specific intercept of 2.58 ms. The lower bound of the two-sided 90% CI of the predicted effect on ΔΔQTcF (10.32 ms [90% CI: 8.50-12.14]) at the geometric mean peak moxifloxacin concentration (1750 ng/mL) was above 5 ms, thereby demonstrating assay sensitivity.

Emax model
The goodness-of-fit plots in Figure 3 suggest that the observed mean ΔΔQTcF in the highest three deciles (8th, 9th, and 10th deciles) for ulotaront and SEP-383103 seemed to plateau ( Figure 3). An assessment of the adequacy of the Emax model was given by the goodness-of-fit plot.
The Emax model indicated that the observed mean ΔΔQTcF at 10th decile fell within the 90% CI of the model-predicted ΔΔQTcF. The AIC value from the Emax is 9657.8, larger than the AIC values from the six linear mixed-effects Models B-G (Table S1). Moreover, the treatment effect-specific intercept from the Emax model is moderate (−3.17 ms) and significant (the absolute t-value for the treatment effect-specific intercept estimate >1.95), while this intercept from the linear mixed-effects models is small and not significant. Based on this, the Emax should not be seen as the primary model. The predicted effect on ΔΔQTcF at the geometric mean C max of ulotaront (458 ng/mL) was 4.75 ms (90% CI: 2.75-6.76). The effect on ΔΔQTcF exceeding 10 ms can be excluded within the full observed range of ulotaront plasma concentration, ~835 ng/mL (Figure 3 and Figure S2) based on the Emax model.

Additional ECG parameters
Categorical outliers for QTc, HR, PR, and QRS are shown in Table 3. A single subject experienced an outlier event of ΔQTcF >480 ms at one time point following moxifloxacin treatment. There were no outlier events of ΔQTcF >500 ms and ΔQTcF >60 ms. Ulotaront at the studied dose did not have a clinically relevant effect on cardiac conduction (i.e., the PR and QRS intervals) ( Figure S3). There were no PR or QRS outliers. There were 14 instances following ulotaront treatment of treatment emergent T-wave morphology changes, 16 instances following placebo and 9 instances following moxifloxacin F I G U R E 2 Change-from-baseline QTcF (ΔQTcF, left panel) and placebo-adjusted ΔQTcF (ΔΔQTcF, right panel) across time points. CI, confidence interval; Δ, change-from-baseline; ΔΔ, placebo-adjusted change-from-baseline; HR, heart rate; LS, least squares; QTcF, QT interval corrected for heart rate using Fridericia's formula. LS mean and 90% CI based on a linear mixed-effects model:  No deaths were reported in the study. Two subjects discontinued the study due to an AE: one subject after receiving ulotaront (anxiety, possibly related to study drug) and one subject after receiving moxifloxacin (schizophrenia [exacerbation of], not related to study drug). One subject experienced an SAE after receiving ulotaront The red-filled circles with vertical bars denote the mean placebocorrected ΔQTcF (ΔΔQTcF) with 90% CI displayed at the median plasma concentration within each decile. The solid black line with gray shaded area denotes the model-predicted mean placebo-adjusted ΔQTcF with 90% CI. The horizontal red line with notches shows the range of concentrations divided into deciles. The distance between each decile represents the point at which 10% of the data are present; the first notch to second notch denotes the first 10% of the data, the second notch to third notch denotes the 10%-20% of the data, and so on. The black circle with vertical bars denotes the mean placebo-corrected ΔQTcF with 90% CI for placebo at a concentration of 0.  Note: Based on a linear mixed-effects model with ΔQTcF as the dependent variable, time-matched ulotaront and SEP-383103 plasma concentration as explanatory variates, centered baseline as an additional covariate, treatment (active = 1 or placebo = 0) and nominal time point as fixed effects, and a random intercept and slope per subject. T max is on individual level, not population level. Plasma concentrations in placebo treatment period are set to zero. In nonplacebo treatment period, plasma concentrations below the quantifiable limit at predose are set to zero and after dosing are set to half of the lower limit of quantitation.

Location of prediction
(hypotension), which was considered a medically important event, and was assessed by the Investigator as severe and related to study drug. The hypotension resolved the same day without treatment. There were no clinically meaningful changes in laboratory values and no substantial differences among the treatment groups or study visits. Mean changes in vital signs parameters were generally small and not considered to be clinically significant. No AEs of QT prolongation were reported; the frequency of ECG observations meeting protocol-defined markedly abnormal criteria was similar between ulotaront and placebo treatment and was observed for more subjects following moxifloxacin treatment. There were no positive findings on the C-SSRS at baseline and postdose. Overall, a single dose of ulotaront 150 mg was generally well-tolerated in adults with schizophrenia in this study.

DISCUSSION
PK characteristics for single-dose administration of ulotaront 150 mg were consistent with results from previous studies. 6 In this study, geometric mean C max of ulotaront 150 mg was 458 ng/mL (Table 2), which was ~1.44fold above the estimated geometric mean of steady-state C max at the anticipated maximum therapeutic dose of 100 mg (318 ng/mL). PK characteristics for single oral dose administration of 400 mg moxifloxacin were consistent with the literature for sufficient positive control for TQT study. 19,21 The highest anticipated clinical exposure of ulotaront is currently defined as steady-state mean C max for 100 mg once daily in CYP2D6 poor metabolizers based on results of a phase I clinical drug-drug interaction study where co-administration of a CYP2D6 strong inhibitor increased ulotaront C max and AUC 0-inf by 31% and 72%, respectively. Coadministration of the strong CYP2D6 inhibitor served as a proxy for poor CYP2D6 metabolism in subjects who otherwise are normal CYP2D6 metabolizers. In the primary analysis, the upper bound of the two-sided 90% CI for ΔΔQTcF was below 10 ms at the highest anticipated clinical exposure (~416 ng/mL and ~ 211 ng/mL for ulotaront and SEP-383103, respectively), indicating ΔΔQTcF is unlikely to exceed 10 ms in patients with schizophrenia at the anticipated maximum therapeutic dose of 100 mg ulotaront once daily. Additional results from ongoing phase III and renal/hepatic impairment studies will further inform estimation of the highest clinically relevant plasma exposure.
Based on model-independent checks of the appropriateness of the linear assumption about the concentration-QTc relationship (Figure S2), the linear regression lines slightly separated from the LOESS regression lines at higher concentration levels (i.e., from ~446 ng/mL of ulotaront; from ~12.2 ng/mL of SEP-363854; and from ~213 ng/mL of SEP-383103). In addition, based on goodness-of-fit plots (Figure 3), the observed mean ΔΔQTcF seemed to plateau at the higher concentrations of ulotaront, suggesting that the linear assumption may overestimate the effect on ΔΔQTcF at the high end of ulotaront concentrations. An Emax model was therefore also fitted according to a predefined statistical analysis plan (Data S1) 22 and provided an alternative evaluation that suggests QT prolongation may remain under 10 msec up to ~835 ng/mL of ulotaront, although data are limited in the concentration range ( Figure 3).
Ulotaront was generally well tolerated and no AEs of QTc prolongation were reported in this study. The safety profile of ulotaront observed in this study is consistent with what has been observed in other studies across the ulotaront clinical development program. In a randomized, double-blind, placebo-controlled, 4-week phase II clinical trial enrolling 245 patients with an acute exacerbation of schizophrenia, no post-baseline clinically significant ECG abnormalities, including prolongation in the QTcF interval, were observed. 10 Neither ulotaront or SEP-363854 or SEP-383103 affected hERG channel current at physiologically relevant concentrations. The half maximal inhibitory concentration (IC 50 ) values for the inhibitory effect of ulotaront and SEP-363854 on hERG channel current were 958.5 μM and 296.8 μM, respectively (unpublished data). The IC 50 values are T A B L E 4 Adverse events occurring in ≥5% subjects by system organ class and preferred term (safety analysis population). Percentages were calculated based on the number of safety analysis population subjects in each treatment group.

SOC
AEs were sorted alphabetically by SOC and descending frequency of PT within each SOC in the ulotaront treatment group.
Abbreviations: AE, adverse event; MedDRA, Medical Dictionary for Regulatory Activities; m, number of events; N, number of subjects in the treatment group; n, number of subjects in the subgroup; PT, preferred term; SOC, system organ class.
>380fold and > 4000-fold higher than the geometric mean C max achieved at the 150 mg dose in this study for ulotaront and SEP-363854, respectively. SEP-380103 inhibited hERG current by 6.3% at 30 μM, the highest concentration tested, and no IC 50 value could be calculated (unpublished data). The highest concentration assessed (30 μM) is >25fold higher than the geometric mean C max of SEP-383103 achieved at the 150 mg dose in this study.
As demonstrated in the current study, ulotaront is unlikely to cause clinically relevant QTc prolongation in patients with schizophrenia at the anticipated maximum therapeutic dose of 100 mg once daily. Lack of a clinically relevant effect of antipsychotic drugs on the QTc interval is important because patients treated with antipsychotic drugs have been reported to have an increased risk of sudden cardiac death. [23][24][25][26] Several antipsychotic drugs have been shown to prolong the QTc interval, 27,28 and drug-induced QTc prolongation is associated with higher risk of cardiac arrhythmias, such as torsades de pointes, which can lead to sudden death. 29,30 In this study, concentration-QTc analysis was used as the primary analysis, which enabled the assessment of QTc changes over a wide range of concentrations. This type of analysis is useful in predicting potential QTc effects at different plasma concentrations not directly studied, for example, in specific populations or conditions (such as drug interactions) even though not directly evaluated. 17 Also, concentration-QTc analysis required a smaller sample size than by-time-point analysis, 31 which enabled a more efficient study.
In conclusion, single-dose administration of ulotaront 150 mg in subjects with schizophrenia had no clinically relevant effect on heart rate, PR interval, and QRS duration. Ulotaront is unlikely to have a clinically relevant effect on the QTc interval in patients with schizophrenia at the anticipated maximum therapeutic dose of 100 mg once daily.

ACKNOWLEDGMENTS
Sunovion discovered ulotaront in collaboration with PsychoGenics based in part on a mechanism-independent approach using the in vivo phenotypic SmartCube® platform and associated artificial intelligence algorithms. We thank all the subjects, investigators, and site staffs for participating in this trial.

FUNDING INFORMATION
This study was fully funded by Sunovion Pharmaceuticals Inc.