Clinical Study Report — GLPI103-301

All efficacy, safety, and disposition figures are sourced from the generated analysis outputs (outputs/stats/, outputs/tfl/) and the ADaM datasets, and are verified by tools/reconcile_csr.py (23/23). Figures reflect the actual simulated enrollment (900 randomized to target). Cardiometabolic biomarker (NT-proBNP/hsCRP) and ECG endpoints are [MOCK — not in dataset].

Change History

1. Title Page

A Phase 3, Randomized, Double-Blind, Double-Dummy, Active-Controlled, Parallel-Group, Multicenter Trial Evaluating the Efficacy and Safety of GLPI-103 (Intravenous and Oral) versus Oral Semaglutide in Adults with Type 2 Diabetes Mellitus Inadequately Controlled on Metformin. Study GLPI103-301. Sponsor: Virtual Biopharma Inc. Phase of development: 3. Report status: Final.

2. Synopsis

3. Table of Contents

This report follows the 16-section structure of ICH E3 with appendices 16.1–16.4 (see §16). A navigable table of contents is provided in the rendered document.

4. Abbreviations

As defined in PROT-301 / SAP-301 (AE, SAE, TEAE, AESI, FAS, PPS, ANCOVA, MMRM, LS mean, CFB, CI, eGFR, FPG, HbA1c).

5. Ethics

The study was conducted in accordance with the ethical principles of the Declaration of Helsinki and ICH E6(R3) Good Clinical Practice. Independent ethics committee / institutional review board approvals and applicable regulatory authorizations were obtained before initiation at each site, and written informed consent (TMF-ICF) was obtained from every subject prior to any study-specific procedure. An independent Data Safety Monitoring Board provided unblinded safety oversight (TMF-DSMB).

6. Investigators and Study Administrative Structure

The study was conducted at approximately 17 centres across the Republic of Korea (site identifiers S001–S017, mapped to provinces/regions). A central laboratory provided all efficacy and safety analytes; randomization and drug supply were managed by a validated interactive web response system; data management and statistical analysis were performed under the Data Management Plan and SAP (TMF-DMP; SAP-301). The sponsor was Virtual Biopharma Inc.

7. Introduction

GLPI-103 is a first-in-class GLP-1 / apelin (APJ) receptor dual agonist developed for the treatment of type 2 diabetes mellitus. By engaging both the GLP-1 and APJ pathways, GLPI-103 is hypothesized to provide superior glycaemic and weight effects, with favourable cardiometabolic changes, relative to GLP-1 receptor agonism alone (PROT-301 §1). This pivotal Phase 3 study was designed to confirm superiority versus a guideline-recommended active comparator (oral semaglutide titrated to 14 mg once daily) on the change from baseline in HbA1c at Week 52, in adults inadequately controlled on metformin.

8. Study Objectives

The primary objective was to demonstrate superiority of GLPI-103 IV and GLPI-103 Oral versus oral semaglutide on HbA1c CFB at Week 52. Key secondary objectives addressed body-weight reduction, attainment of HbA1c <7.0%, and ≥10% body-weight reduction at Week 52; additional secondary objectives addressed FPG, blood pressure, and cardiometabolic biomarkers (PROT-301 §2).

9. Investigational Plan

9.1 Overall Design and Plan

This was a randomized, double-blind, double-dummy, active-controlled, parallel-group, multicenter study. After a 2-week screening period, eligible subjects were randomized in a 1:1:1 ratio to GLPI-103 IV (1→2→4 mg once weekly) plus oral placebo, GLPI-103 Oral (2→4→8 mg once daily) plus IV placebo, or oral semaglutide (3→7→14 mg once daily) plus IV placebo, and treated for 52 weeks under double-blind, double-dummy conditions, with a 4-week off-treatment safety follow-up. The double-dummy design maintained blinding across the two routes of administration.

9.2 Selection of Study Population

Adults aged 18–75 years with T2DM inadequately controlled on stable metformin (HbA1c 7.0–10.5%, BMI ≥25 kg/m², eGFR ≥45 mL/min/1.73 m²) were eligible; key exclusions included hepatic transaminases >3×ULN, pancreatitis history, personal/family history of MTC/MEN2, and recent GLP-1 receptor agonist use (PROT-301 §5). Eligibility was confirmed centrally.

9.3 Treatments and Randomization

Randomization was stratified by baseline HbA1c (<8.5 vs ≥8.5%), BMI (<30 vs ≥30 kg/m²), and Region, and implemented by permuted blocks within strata via the IWRS (TMF-RAND). Background metformin was continued unchanged. Conservative 8-week titration preceded maintenance dosing.

9.4 Efficacy and Safety Variables

The primary efficacy variable was CFB in HbA1c (%) at Week 52. Secondary variables included body weight, HbA1c <7.0% and ≥10% weight-loss responders, FPG, and blood pressure. Safety variables included treatment-emergent adverse events, adverse events of special interest, hypoglycaemia, clinical laboratory parameters, and vital signs (PROT-301 §9).

9.5 Statistical Methods (summary)

The primary endpoint was analysed by ANCOVA on the Full Analysis Set (CHG ~ TRT01A + BASE + STRAT_HBA1C + STRAT_BMI), with family-wise error controlled by a fixed-sequence testing hierarchy; longitudinal HbA1c was analysed by MMRM; responders by logistic/CMH methods; missing data primarily by MMRM under MAR with pattern-mixture, reference-based multiple-imputation, and tipping-point sensitivity analyses (SAP-301).

9.6 Changes in the Conduct or Planned Analyses

There were no substantive changes to the planned analyses (SAP-301 §15).

10. Study Patients

10.1 Disposition of Patients (Figure 1; Table 14.1.1)

Of 1,700 subjects screened, 776 were screen failures and 924 were eligible; enrollment closed upon reaching the target of 900 randomized (the remaining 24 eligible subjects were not randomized). The principal reasons for screen failure were HbA1c outside the 7.0–10.5% range (n=495), BMI <25 kg/m² (n=135), age outside 18–75 years (n=119), and eGFR <45 mL/min/1.73 m² (n=25). The observed screen-failure rate (45.6%) exceeded the planning assumption of approximately 30%, attributable to the realistic source demographics of the simulated population (noted per ICH E3).

Overall study discontinuation was 9.3% (84/900) and balanced across arms. Reasons for discontinuation were: lost to follow-up (n=28), withdrawal by subject (n=25), physician decision (n=18), adverse event (n=9), death (n=2), and lack of efficacy (n=2). 46 subjects (5.1%) initiated protocol-defined rescue medication for persistent hyperglycaemia (an intercurrent event handled per the primary estimand; §11.1).

10.2 Protocol Deviations

53 subjects (5.9%) had a major protocol deviation and were excluded from the Per-Protocol Set; consequently the PPS comprised 847 subjects (vs 900 in the FAS). The primary efficacy conclusions were consistent between the FAS and PPS analyses, supporting robustness to deviation-related exclusions (SAP-301 §6).

11. Efficacy Evaluation

11.1 Data Sets Analysed

The primary efficacy analysis was performed on the Full Analysis Set (N=900; FASFL="Y"), comprising all randomized subjects analysed according to randomized treatment.

11.2 Demographic and Baseline Characteristics (Table 14.1.3 / Table 1)

The treatment arms were well balanced with respect to demographic and baseline disease characteristics, region, education, and marital status, consistent with the stratified randomization (full table: outputs/tfl/table1_demographics.xlsx). Demographics were derived from real Korean population distributions (Nemotron-Personas-Korea; REF-002), yielding a population representative of older adults with established T2DM.

11.3 Primary Efficacy Result (Table 14.2.2 / Table 2)

The least-squares mean change from baseline in HbA1c at Week 52 (ANCOVA, FAS) was:

The pairwise comparisons versus the active comparator were:

Both GLPI-103 formulations demonstrated statistically significant and clinically meaningful superiority over oral semaglutide on the primary endpoint, with the intravenous formulation showing the greatest reduction. The pre-specified fixed-sequence testing hierarchy was satisfied at each step (SAP-301 §7.3), supporting confirmatory inference for the primary and key secondary endpoints. (Source: outputs/stats/ancova_results.csv.)

11.4 Longitudinal Analysis (Figure 2; Table 14.2.3)

The mixed model for repeated measures (AR(1) covariance, the pre-specified fallback applied on non-convergence of the unstructured model) showed a progressive reduction in HbA1c that emerged by Week 4 and was maintained through Week 52, with consistent separation favouring GLPI-103. The LS-mean change from baseline by visit was:

The Week-52 LS means were consistent with the ANCOVA primary result, and the between-arm separation was apparent from the first post-baseline visit and widened through the maintenance period. Because the MMRM uses all observed longitudinal data under a missing-at-random assumption, it appropriately accommodates the 9.3% discontinuation and intermittent missing visits (outputs/stats/mmrm_results.csv).

11.5 Key Secondary and Other Endpoints (Tables 14.2.4–14.2.5)

GLPI-103 produced substantially greater body-weight reduction and higher responder rates than semaglutide on both glycaemic-target and weight-loss endpoints, supporting the dual-mechanism hypothesis. Blood-pressure changes were small and similar across arms (SBP CFB ranging from approximately 0 to −0.8 mmHg). Cardiometabolic biomarker endpoints (NT-proBNP, hsCRP) were pre-specified but not captured in the simulated dataset [MOCK].

11.6 Subgroup Analyses (Figure 3)

The HbA1c treatment effect favouring GLPI-103 was consistent across the pre-specified subgroups defined by baseline HbA1c, BMI, age, and sex, with overlapping confidence intervals and no qualitative interaction, supporting the generalizability of the primary result (outputs/tfl/figure3_forest_plot.pdf).

11.7 Efficacy Conclusions

GLPI-103 (IV and Oral) was superior to oral semaglutide on glycaemic control and weight reduction at Week 52, with high responder rates and consistent subgroup effects, confirming the efficacy hypothesis underlying the development program.

12. Safety Evaluation

12.1 Extent of Exposure

All randomized subjects who received at least one dose constituted the Safety Set (N=900; 300 / 301 / 299). Subjects were exposed for up to 52 weeks following an 8-week titration.

12.2 Overall Adverse Events (Table 14.3.1 / Table 3)

Any treatment-emergent adverse event was reported by 227 (75.7%) subjects in the GLPI-103 IV arm, 204 (67.8%) in the GLPI-103 Oral arm, and 216 (72.2%) in the semaglutide arm. Adverse events were predominantly mild (649 events) or moderate (337 events), with 87 severe events; 752 of 1,073 events were assessed as treatment-related. The distribution by system organ class (subject incidence) was:

Gastrointestinal disorders were the most frequently affected class, with the highest incidence in the higher-exposure intravenous arm (driven by nausea), consistent with the expected GLP-1 tolerability gradient; decreased appetite and vomiting were more frequent with semaglutide.

12.3 Common Adverse Events by Preferred Term

Nausea was significantly more frequent with GLPI-103 IV (32.3%) than with semaglutide (21.4%) — a difference of +10.9 percentage points (p=0.003; outputs/stats/ae_prop_test.csv). This reflects the expected gastrointestinal tolerability trade-off of greater GLP-1 receptor engagement in the higher-exposure intravenous arm and was generally manageable with the protocol-defined titration. Injection-site reactions were, as anticipated, most frequent in the intravenous arm (5.0%). Hypoglycaemia was infrequent and numerically higher with semaglutide (6.4%) than with GLPI-103 (~3%).

12.4 Adverse Events of Special Interest

• Gastrointestinal events: as described; predominantly mild–moderate and titration-manageable.
• Hypoglycaemia: infrequent; numerically higher with semaglutide (6.4%) than GLPI-103 (~3%); no severe (Level 3) events of concern.
• Acute pancreatitis and thyroid C-cell neoplasia (MTC): none recorded; monitored as class risks.
• Cardiovascular / heart-rate (APJ-related): no clinically meaningful signal in vital-sign data.

12.5 Deaths, Serious Adverse Events, and Discontinuations

10 serious adverse events were reported, including 2 deaths (none assessed as treatment-related). Overall study discontinuation was 9.3% (84/900), balanced across arms, with adverse-event-related discontinuations infrequent (n=9). The serious events and deaths showed no pattern by treatment and are consistent with the background morbidity expected in an older T2DM population (§10.1).

12.6 Clinical Laboratory and Vital Signs

Hepatic (ALT/AST) and renal (eGFR) parameters were stable, with mean changes near zero across arms and no pattern of treatment-related hepatotoxicity or nephrotoxicity. Vital signs, including heart rate, showed no clinically meaningful change. ECG data were not captured in the simulated dataset [MOCK].

12.7 Safety Conclusions

GLPI-103 demonstrated a manageable, class-consistent safety profile dominated by gastrointestinal adverse events, with a tolerability trade-off (higher nausea) in the higher-exposure intravenous formulation addressed by titration; no new or serious safety signals attributable to treatment were identified.

13. Discussion and Overall Conclusions

In this pivotal Phase 3 study, GLPI-103, in both intravenous and oral formulations, achieved statistically significant and clinically meaningful superiority over oral semaglutide on the primary endpoint of HbA1c change from baseline at Week 52 (IV −0.68%, Oral −0.34%; both p<0.001). The benefit was evident early (by Week 4 in the longitudinal analysis), sustained through Week 52, and consistent across pre-specified subgroups. Importantly, the glycaemic benefit was accompanied by substantially greater weight reduction (−14.5 and −10.3 kg for IV and Oral versus −7.6 kg for semaglutide) and markedly higher responder rates for both glycaemic-target and clinically meaningful weight-loss endpoints, supporting the therapeutic rationale of combined GLP-1 and APJ agonism.

The safety profile was consistent with the GLP-1 receptor agonist class and was dominated by gastrointestinal adverse events. Nausea was more frequent with the intravenous formulation (32.3% vs 21.4% for semaglutide; p=0.003), reflecting the efficacy–tolerability trade-off of greater receptor engagement at higher exposure and managed by the protocol-defined titration; vomiting and decreased appetite were more frequent with semaglutide, and hypoglycaemia was infrequent (numerically higher with semaglutide). Ten serious adverse events and two deaths were reported, with no pattern by treatment and none assessed as treatment-related, consistent with the background morbidity of an older T2DM population.

The trial incorporated realistic clinical-trial complexity that strengthens the analytical interpretation: enrollment to a 900-subject target (1,700 screened), a 9.3% discontinuation rate with documented reasons, intermittent (non-monotone) missing visits, between-site heterogeneity, protocol-defined rescue medication (handled as an intercurrent event under the estimand framework), major protocol deviations defining a Per-Protocol Set (847 of 900) distinct from the Full Analysis Set, and consequent missing data handled by the mixed-model and sensitivity analyses (SAP-301). Residual limitations of the simulated dataset are stated transparently: cardiometabolic biomarker and ECG endpoints were not captured, and the effect sizes are at the favourable end of the plausible range. Every quantitative result reported is reconciled to the source datasets.

In conclusion, GLPI-103 (IV and Oral) demonstrated a favourable benefit–risk profile in adults with T2DM inadequately controlled on metformin, supporting its marketing application (M2.5; M2.5.6).

14. Tables, Figures, and Listings Referred to in the Report

Referenced tables and figures are provided in outputs/tfl/ (Table 1 demographics, Table 2 primary efficacy, Table 3 TEAE summary; Figure 1 CONSORT, Figure 2 longitudinal HbA1c, Figure 3 subgroup forest) and statistical outputs in outputs/stats/; shells are specified in TLF-301. In-text table numbers follow the ICH E3 14.x convention.

15. Reference List

ICH E3; PROT-301; SAP-301; ADaM-SPEC-301; REF-002; DEF-301 (define.xml / SDRG / ADRG); ISE-301; ISS-301.

16. Appendices

• 16.1 Study information: protocol and amendments (PROT-301), statistical analysis plan (SAP-301), sample case report form (CRF-301) and annotated CRF (ACRF-301), randomization specification (TMF-RAND), ethics and investigator documentation (TMF-CTA).
• 16.2 Patient data listings: subject disposition, protocol deviations, adverse events, serious adverse events and deaths, and laboratory abnormalities (derived from the ADaM datasets).
• 16.3 Narratives of deaths, other serious adverse events, and adverse-event discontinuations: provided in the companion appendix CSR-301-N (the 2 deaths, 10 serious adverse events, and 9 adverse-event discontinuations, narrated per ICH E3 §12.3.2).
• 16.3 Case report forms: blank (CRF-301) and annotated (ACRF-301).
• 16.4 Individual patient data listings: ADaM datasets with define.xml and reviewer's guides (DEF-301).
• Reconciliation: all quantitative efficacy, safety, and disposition figures in §10–12 are verified against outputs/ by tools/reconcile_csr.py (23/23 key values).