Statistical Methods definition

Statistical Methods. The primary endpoint for this clinical trial is the percent of patients with radiographic disease progression according to RECIST 1.1 at 6 months. Without maintenance therapy, 92% of patients are expected to progress by 6 months. Therefore, we assume 1/12 (0.083) pembrolizumab + placebo arm show no progression versus 7/12 pembrolizumab + paricalcitol. This gives 90% statistical power with one side alpha = (0.05). TABLE OF CONTENTS

Statistical Methods. Sample size: Phase 2: With 30 patients, the Phase 2 portion of the study will have over 80% power to detect a change from baseline in eGFR relative to zero. The power calculation, which was based on a 2-sided t-test, assumes the following: • Two-sided Type I error rate of 0.05 • 5% of the patients will not complete at least 12 weeks of study treatment • A change from baseline in eGFR of approximately 4.3 mL/min/1.73 m2 • Standard deviation of change from baseline in eGFR of 8 mL/min/1.73 m2 Phase 3: Primary endpoint With 150 patients enrolled (75 in each group), the study will have approximately 80% power to detect a difference between the two treatment groups in change from baseline in eGFR of 3.1 mL/min/1.73 m2 at Week 48. The power calculation, which was based on mixed-model repeated measures (MMRM) analysis, assumes the following: • 9 repeated measurements (Weeks 1, 2, 4, 6, 8, 12, 24, 36, and 48) having compound symmetry covariance structure • The correlation between observations on the same subject is 0.7 • Two-sided Type I error rate of 0.05 • Standard deviation of change from baseline in eGFR of 8 mL/min/1.73 m2 • Analyses at Week 48 are based on the intent-to-treat (ITT) population • Missing data for the Week 48 analysis will be imputed using Jump to Reference (J2R) multiple imputation (Ratitch, 2013) based on available data collected from patients discontinuing from study treatment but continuing in the study. Key secondary endpoint With 150 patients enrolled and at least 140 patients having available Week 52 data after completing 48 weeks of treatment, the study will have a minimum detectable difference between the two treatment groups in change from baseline in eGFR of approximately 2.2 mL/min/1.73 m2 at Week 52. The power calculation was based on the same analysis method and assumptions as the primary endpoint, with the following exceptions: • With the addition of Week 52, the model has 10 repeated measurements (Weeks 1, 2, 4, 6, 8, 12, 24, 36, 48, and 52) • Analyses at Week 52 are based on available Week 52 data from the subset of patients who complete the 48-week treatment course. Since the Phase 2 and Phase 3 cohorts are independent sets of patients, the Phase 2 analysis will not impact the type I error rate for the Phase 3 analysis. The analysis of efficacy will use an unstructured covariance structure, which is expected to have approximately the same power as the analysis with compound symmetry used for study planning. The method for main...

Statistical Methods. Analysis Populations: Randomized: This population will be comprised of all subjects who were initially randomized. This population will be used for summaries of subject disposition and baseline subject characteristics. Intent-to-treat (ITT)/Safety: This population will be comprised of all subjects who were initially randomized and received at least one dose of study drug. This will be the primary population for all summaries of subject disposition and baseline characteristics, efficacy analyses, and safety analyses for purposes of regulatory submissions. Comparisons in the primary endpoint of change from the baseline BMI between treatment groups will be assessed using a mixed effects model with repeated measures (MMRM) with factors of treatment, visit, treatment by visit interaction, baseline BMI value, age stratification, and gender stratification. Appropriate contrast will be applied for treatment comparisons at Week 56. The 3 comparisons of interest are 1) top-dose vs. placebo; 2) mid-dose vs. placebo; and 3) top-dose vs. mid-dose. Sensitivity analyses will be conducted to examine the impact of missing data on the robustness of statistical conclusions. In previous studies in adults, PHEN/TPM 7.5 mg/46 mg dose resulted in a placebo-subtracted BMI reduction of 2.4 units with a standard deviation of 2.9. Assuming a similar effect size, with enrollment of 200 subjects (100 randomized to the top-dose, 50 randomized to the mid-dose, and 50 randomized to the placebo), the present study will have greater than 90% power to detect a significant difference in BMI reduction between the top-dose and the placebo, and approximately 80% power to detect a significant difference between the mid-dose group and the placebo. The above power calculation assumes very conservative differences between the active doses and placebo and a worst case 30% dropout rate.

Statistical Methods. Analysis Populations The statistical analysis of the HEROES PBC studies will follow a nested randomized trial emulation approach (▇▇▇▇▇▇ 2008, ▇▇▇▇▇▇ 2013) using a treatment decision design to identify index events (▇▇▇▇▇▇▇▇▇ 2015). Due to the lack of randomization, non-randomized observational studies will instead be conducted, with the goal of mimicking the hypothetical randomized trial as closely as possible. All patients will be evaluated over time from the start of the study period to identify evidence of inadequate treatment response to UDCA. Each date of inadequate response identified in the data will be considered a treatment decision point, and that date will be considered an index date, with a corresponding record created in the analytic data for that patient index. Each patient can contribute multiple index dates and thus patient indices, corresponding to each date they have a treatment decision point, and therefore each patient may contribute person-time to multiple patient index records. It is likely that OCA-treated and non–OCA-treated patient indices will differ on confounding variables, ie, variables that influence both treatment and outcome. To address differences in the covariate distribution between treatment groups, the Sponsor will first conduct descriptive analyses on all baseline variables (demographics, clinical characteristics, and treatment history) for the OCA-treated and non–OCA-treated groups at index. Standardized mortality/morbidity ratio (SMR) weights then will be used to create a pseudo-population of non–OCA-treated indices with the same covariate distribution as the OCA-treated patients at the time of OCA initiation (▇▇▇▇ 2003), and unweighted and weighted standardized mean differences will be computed. Primary Analysis: The objective of the primary analysis is to estimate the as-treated effect of OCA treatment versus non–OCA-based treatment on the composite endpoint of all-cause death, liver transplant, or decompensation requiring hospitalization. The primary analysis outcome will be assessed with the hazard ratio comparing the hazard of the first event of the composite endpoint among the OCA-treated patients and standardized morbidity ratio-weighted non–OCA-treated PBC patient indices. The hazard ratio will be estimated using a ▇▇▇ proportional hazards model. The nonparametric bootstrap will be used to estimate the standard error of the estimate, which will then be used to generate 95% CI for the hazard ratio and to perf...

Statistical Methods. The number of pediatric subjects for the dose escalation part of the study will be determined based in part on the number of dose escalations required to determine the protocol-defined MTD and/or RP2D.

Statistical Methods. Efficacy: In order to compare the primary efficacy parameter, duration of diarrhea between the treatment groups a ▇▇▇▇▇▇ ▇▇▇▇▇ analysis will be performed. The log-rank test will be used to test whether the difference of the duration of diarrhea between two treatment groups is statistically significant, i.e. p-value <0.

Statistical Methods. Analysis Populations • Safety Population (SAFETY): All patients who receive at least 1 dose of investigational product, analyzed as treated. • Intention-to-Treat Population (ITT): All randomized patients who receive at least 1 dose of study drug, analyzed as randomized. Efficacy Analyses The primary efficacy analysis will use a Mixed Model Repeated Measures (MMRM) approach to estimate mean change from (Visit 2) in the Total MADRS score evaluated at the Week 6 visit in the TNX-601 ER and placebo arms. The model will include all participants in the ITT population, and the dependent variable will be the observed change from Baseline (Visit 2) in total MADRS score at each postdose visit. Covariates in the model will include the fixed categorical effects of treatment, site, visit, and treatment by visit interaction, as well as the continuous fixed covariates of baseline score and baseline score by visit interaction. Missing data will be imputed via multiple imputation (MI), and details regarding the MI approach will be described in the Statistical Analysis Plan (SAP). The primary analysis will be followed by several sensitivity analyses. The key secondary efficacy outcomes will be measured and compared between treatment groups and will be adjusted for multiplicity using a fixed sequence procedure. Further details on sensitivity and secondary analyses can be found in the SAP, as well as the order of the key secondary endpoints. Safety Analyses Informed consent X Optional Pharmacogenomics consent Xg Clinical Trial Subject Registry Consent Xk Placebo response traininga X X X X X Inclusion/exclusion criteria X X Psychiatric history (MINI 7.0.2) X MSI-BPD X Prior and concomitant medications X X X X X X Demographics and medical history X Xl Vital signs, weight, heightb, and BMI X X X X X X Physical examination, including a brief neurological exam X X 12-lead ECG X X X Pregnancy testc X X X X X X Urine drug screend X Clinical laboratory assessments X X X PK X X Randomization X Dispense study drug Xf X X Study drug return, compliance, and accountability X X X Blood sample collection Xg * * * * MADRS X X X X X CGI-S X X X X CGI-I X X X HARS X X X X SDS X X X X PGIC X X X PGIS X X X X AEs X X X X X MADDERS® Xh Xh Xh Xi C-SSRS X X X X X X CSFQ-14 X X SOWS X Xj = ▇▇▇▇▇▇ Screening Instrument for Borderline Personality Disorder; PGIC = Patient Global Impression of Change; PGIS = Patient Global Impression of Severity; PK = pharmacokinetic; SDS = ▇▇▇▇▇▇▇ Disability Scale; MADD...

Statistical Methods. Efficacy: In order to compare the primary efficacy parameter, duration of diarrhea between the treatment groups a ▇▇▇▇▇▇ ▇▇▇▇▇ analysis will be performed. The primary efficacy parameter, duration of diarrhea, will be analyzed using descriptive statistics. In addition, a ▇▇▇▇▇▇-▇▇▇▇▇ plot will be generated for which summary statistics will be presented as well. Secondary efficacy parameters will be analyzed similarly. All parameters will be summarized using descriptive statistics. Safety: The safety sample will be used for the analysis of the safety and tolerability data. Treatment emergent AEs are summarized. Severity and drug-event relationship of treatment emergent AEs are summarized separately. Vitals signs, including changes from baseline will be Name of Sponsor: ▇▇▇▇▇▇ Taiwan Name of Finished Product: Hidrasec Infants Granules for Oral Suspension 10 mg Hidrasec Children Granules for Oral Suspension 30 mg Name of Active Ingredient(s): Racecadotril summarized. A frequency table will be presented for markedly abnormal values. Sample size In total, 40 subjects will be allocated to receive Racecadotril treatment, 20 subjects in each age group (< 24 months of age and ≥ 24 months of age). Results of a recently finalized study in Russian children showed that all children under Racecadotril treatment recovered within 3 days of drug treatment. A sample size of 40 subjects is sufficient to estimate a two-sided 95% confidence interval with the precision of 3.1 percentage points for a recovery of at least 99% of the subjects after 3 days of Racecadotril treatment. In order to account for drop-outs, maximum 48 subjects will be recruited. Recruitment will be stopped once 40 subjects have completed the study. 1.1 Independent Ethics Committee or Institutional Review Board 14