Glucophage Extended Release (GXR) China Bioequivalence Study (Nantong - Darmstadt): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Glucophage Extended Release (GXR) China Bioequivalence Study (Nantong - Darmstadt) (NCT03566810), a randomized phase 1 study. The primary outcome is Area Under the Plasma Concentration-Time Curve From Time Zero to the Last Sampling Time (AUC 0-t) of Metformin.

2. Clinical Question

The clinical question is a formulation and manufacturing-site comparison, not a comparison between treatment and no treatment. Both products are Glucophage Extended Release, and the study concerns whether the Nantong-manufactured product has comparable measured metformin exposure to the Darmstadt-manufactured product in healthy volunteers. Food condition is part of the question because the registry reports separate comparisons for fasting and fed administration.

The crossover structure means participants were assigned to a sequence involving both products, rather than to only one product throughout the study. Reversing the order across sequences allows the product comparison to be made within a crossover design. The available record does not give full details of sequence allocation or the statistical model used to account for the crossover structure. The reported comparisons therefore need to be read as the registry's analyses, with those modelling details not reported in the registry record.

3. Trial Design

This was a Phase 1, randomized, open-label, two-way crossover study. It enrolled 54 healthy participants and is recorded as completed. The study start date was 2018-10-11, and the primary completion date was 2018-11-28. The sponsor was Merck KGaA, Darmstadt, Germany.

The two products were single oral doses of 500 mg Glucophage Extended Release: the test product manufactured in Nantong, China, and the reference product manufactured in Darmstadt, Germany. The registry lists four sequences: test followed by reference under fasting conditions; reference followed by test under fasting conditions; test followed by reference under fed conditions; and reference followed by test under fed conditions. The listed sequence descriptions indicate a seven-day interval between treatment periods. The full details of the schedule and any other study procedures are not reported in the registry record.

Open-label means that masking was not used. Randomization concerns assignment to the treatment sequence; it does not mean that participants or study staff were masked to product identity. A crossover design can support a within-participant comparison because each participant may receive both products, but valid interpretation depends on the trial's conduct and analysis, including how period and sequence effects were handled. The registry record does not report the analysis method for the posted comparisons.

The primary pharmacokinetic measure was AUC 0-t for metformin, expressed in nanogram*hour per milliliter (ng*h/mL). Sampling was scheduled before dosing and at several post-dose times on Day 1 and Day 8, with further sampling on Day 2, Day 3, Day 9, and Day 10. AUC 0-t was defined as exposure from time zero to the last sampling time with a concentration at or above the lower limit of quantification. The record also lists secondary pharmacokinetic measures and adverse-event outcomes.

The registry reports 37 participants for each fasting product group and 16 for each fed product group in the posted primary outcome data. Those group counts should not be treated as separate parallel-arm enrollment totals: the trial is a crossover study, and the registry does not provide a full participant flow or explain the relationship between those analysis counts and the total enrollment.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Treatment Period 1 (Day 1-Day 3)

Reasons for not completing this period:

Treatment Period 2 (Day 8- Day 10)

Analysis population for the primary outcome, as defined in the registry: The Pharmacokinetic (PK) Analysis Set included all participants who completed the study with adequate study drug compliance, without any relevant protocol violations or events with respect to factors likely to affect comparability of PK results, and with sufficient evaluable data to determine primary outcome measures for both treatments.

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

Population: Safety population included all participants who received at least 1 dose of investigational medicinal product (IMP).

6. Primary Endpoint

The registry lists the primary outcome as Area Under the Plasma Concentration-Time Curve From Time Zero to the Last Sampling Time (AUC 0-t) of Metformin, measured over Pre-dose, 0.5, 1, 2, 3, 4, 5, 6, 7, 8, 10, 12, 14 post-dose on Day 1 and Day 8; 24, 30, 36 hours post-dose on Day 2 and 9; 48 hours post-dose on Day 3 and 10.

The primary endpoint, AUC 0-t, measures metformin exposure over time from dosing until the last sampling time at which the measured concentration remains at or above the lower limit of quantification. It is calculated from the concentration-time profile. In this study the registry states that the intended calculation used a mixed log-linear trapezoidal rule.

Exposure is relevant to a bioequivalence comparison because it summarizes the amount of drug present in the circulation over the measured interval. Comparing AUC for the test and reference products addresses whether their measured exposure profiles are similar under the study conditions. The endpoint is reported as a geometric mean, and the product comparison is reported as a ratio of geometric least-squares means. The ratio is expressed as a percentage, with the reference product serving as the comparison group in the reported analyses.

The endpoint is not a direct measure of clinical benefit or harm. It is a pharmacokinetic measure obtained in healthy participants after single doses. Its interpretation for a bioequivalence question depends on the statistical comparison and the prespecified acceptance limits. Those limits are not reported in the registry record, so the reported estimate and confidence interval can be described, but the formal decision rule cannot be independently checked from this record.

The registry specifies a pharmacokinetic analysis set, including completion, adequate study-drug compliance, absence of relevant protocol violations or events likely to affect comparability, and sufficient evaluable data. The remainder of the population definition is not reported in the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does a geometric mean describe?

Pharmacokinetic measurements are often analysed on a logarithmic scale because exposure values can vary multiplicatively and may have a skewed distribution. A geometric mean is a summary that is natural for data analysed on that scale. It is not the same summary as an arithmetic mean. In this record, the primary endpoint is described as a geometric mean, and the treatment comparisons are reported as ratios of geometric least-squares means.

A least-squares mean is a model-based estimate, adjusted according to the terms in the statistical model. In a crossover study, a model can account for the paired nature of the observations and may include design factors such as treatment, period, or sequence. Which factors were included, and how the data were modelled, are not reported in the registry record. The reported estimates should therefore be understood as the posted model-based comparison, rather than reconstructed from the group summaries alone.

How should the ratio be read?

The ratio is expressed as a percentage, with the test product compared with the reference product. A value of 100% would correspond to equal estimated geometric least-squares means. Values below or above 100% indicate that the test estimate is lower or higher, respectively, relative to the reference estimate. The fasting estimate is 98.69%, and the fed estimate is 99.39%. These estimates are close to 100%, but their interpretation must also consider the confidence intervals and the prespecified bioequivalence limits.

What does a confidence interval add?

A confidence interval represents uncertainty around an estimated effect under the assumptions of the statistical analysis. A narrower interval indicates greater precision; a wider interval indicates less precision. It is not a range guaranteed to contain the true value, and it does not capture every possible source of uncertainty, such as limitations in the study design or incomplete reporting.

The fasting 90% confidence interval is 92.20% to 105.64%; the fed 90% confidence interval is 93.15% to 106.05%. Each interval spans values both below and above 100%. Whether an interval supports a formal equivalence conclusion depends on whether it lies within the protocol's specified acceptance limits. Those limits are not reported in the registry record. No p-value is reported for these analyses, and a p-value is not required to describe the ratios and confidence intervals.

Why does the crossover design matter?

In a crossover study, treatment order is randomized and participants receive both products in different periods. Each participant can serve as their own comparison, which can reduce the influence of stable differences between people. The design also requires attention to period, sequence, and possible carryover effects. A washout interval may be part of a crossover schedule, but the available arm descriptions are truncated; the full procedural details are not reported in the registry record. The registry does not state how crossover features were handled in the statistical analysis.

The posted values do not permit a fresh reconstruction of a paired analysis. That would generally require participant-level paired observations or sufficient model output, not just separate group geometric means and coefficients of variation. The reported sponsor ratios and confidence intervals are therefore the appropriate figures to describe as the registry's primary comparison.

Why can variability affect precision?

Variability describes how dispersed measurements are around a summary. Greater variability can make an estimate less precise when other design features are unchanged, while a paired design can use within-participant information to improve precision when observations from the same person are related. The registry reports geometric coefficients of variation of 31.5 and 26.2 for the fasting product groups, and 17.1 and 18.7 for the fed product groups. These are descriptive summaries; they do not reveal the within-participant correlation needed to reconstruct precision for the crossover contrast.

Over-dispersion is a general statistical concern when observed variability exceeds what a model assumes. If a model fails to allow for that extra variation, standard errors may be too small and intervals may be too narrow. The registry does not report a dispersion assessment or enough model detail to judge whether this issue arose. It should not be inferred from the posted coefficients alone.

What can planning calculations tell a reader?

Planning illustrations based on an observed effect are hypothetical design calculations, not evidence about the power of the completed trial. They depend on stated assumptions, including the significance level, allocation, statistical approximation, and the fraction of an observed effect one aims to detect. Their purpose is to illustrate how assumptions and effect size affect a future study's required sample size. They do not show the probability that this completed study detected an effect, and they do not replace the observed confidence intervals. The registry does not report the trial's own sample-size justification, so the planning rationale for enrollment is not reported in the registry record.

10. Analyses Reported by the Sponsor

The registry's primary comparisons are reported as ratios of geometric least-squares means with 90% confidence intervals. The fasting comparison is 98.69% (90% confidence interval 92.20% to 105.64%); the fed comparison is 99.39% (90% confidence interval 93.15% to 106.05%). These are the sponsor-reported analyses and are the figures to use when describing the study's formal posted comparisons.

The record also lists separate geometric mean AUC values and geometric coefficients of variation for the product groups. Those summaries are descriptive and do not reproduce the sponsor's crossover model. A simple recalculation from arm-level summaries, where available, is unadjusted and cannot account for within-participant pairing or model terms unless the required information is reported. The registry does not report the method used for the comparisons, and no recalculated ratio or interval is available in the record.

No discrepancy between a sponsor analysis and a recalculation is flagged in the registry record. The sponsor's reported analysis therefore remains the primary basis for interpreting the posted comparison. The registry does not report enough detail to independently verify the model specification or the acceptance limits used for the bioequivalence decision.

11. Multiplicity, Missing Data and Analysis Populations

The registry lists 14 outcome measures and reports 8 analyses. It also lists 12 posted secondary outcomes. These counts describe the recorded outcomes and analyses; they do not establish how many tests were planned, whether all planned tests were reported, or how any multiplicity was handled. A testing hierarchy, family-wise error adjustment, or other multiplicity strategy is not reported in the registry record.

The primary endpoint is specified as AUC 0-t, with separate reported comparisons under fasting and fed conditions. The record does not state whether these two comparisons were treated as co-primary, whether one was primary and the other supportive, or how their joint interpretation was defined. It also does not report a prespecified analysis plan or the study's formal equivalence margins.

The primary analysis population is described as participants who completed the study with adequate study-drug compliance, without relevant protocol violations or events likely to affect comparability of pharmacokinetic results, and with sufficient evaluable data. Full eligibility rules for that analysis set are not reported in the registry record. The record does not provide a complete participant flow, numbers excluded from each analysis, reasons for exclusion, or a comparison of analysis populations.

Missing-data handling is not reported in the registry record. The record does not state how missing concentrations, incomplete profiles, or values below the lower limit of quantification were handled beyond defining the last sampling time for AUC 0-t as the last time concentration was at or above that limit. That endpoint definition alone does not specify all rules for missing or unevaluable observations.

The methods used for the analysis, any interim analyses, the full statistical analysis plan, and the sample-size justification are not reported in the registry record. This limits assessment of prespecification, multiplicity control, and robustness to exclusions or missing data. The absence of these details does not show that a procedure was or was not used; it means the registry record does not document it.

12. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

Most frequent non-serious adverse event terms

The registry reports no serious treatment-emergent adverse events in any of the four listed product-condition groups. The number at risk and the number with a serious event are 37 and 0 for test fasting, 37 and 0 for reference fasting, 16 and 0 for test fed, and 16 and 0 for reference fed. No deaths are reported in these groups.

For other treatment-emergent adverse events, the registry reports 10 participants for test fasting, 11 for reference fasting, 3 for test fed, and 4 for reference fed. These are counts of participants recorded with events, not a measure of event severity or a complete description of the events. The registry does not list the event terms in the information, and attribution to study product is not reported.

These counts should be read as descriptive safety observations from this study, not as evidence that an event cannot occur. The study involved healthy participants, single doses, and the posted analysis groups differ in size across food conditions. Because the design is crossover, the group counts may include participants exposed to both products; they should not be added as if they were necessarily independent participant groups. The registry does not provide further detail about exposure time, event timing, withdrawals due to events, or the method used to compare safety outcomes.

13. What the Result Does, and Does Not, Mean

The posted results describe pharmacokinetic comparisons of single doses of two Glucophage Extended Release products in healthy volunteers. The sponsor-reported test-to-reference ratios are 98.69% under fasting conditions and 99.39% under fed conditions. Their 90% confidence intervals are 92.20% to 105.64% and 93.15% to 106.05%, respectively. The estimates are close to 100%, while the intervals show uncertainty around those estimates.

Those numbers do not, by themselves, establish whether the study met its formal bioequivalence criterion. The decision depends on prespecified acceptance limits, which are not reported in the registry record. The registry also does not report the statistical model or the full analysis plan needed to evaluate all modelling choices. A careful reading distinguishes the reported estimates from a formal conclusion that cannot be checked using the registry information alone.

The findings concern measured metformin exposure after single doses in healthy participants under the study's fed and fasting conditions. They are not direct evidence about clinical outcomes, long-term use, other patient populations, or repeated dosing. The study does not compare the product with no treatment or with a different active ingredient.

The safety information records zero serious treatment-emergent adverse events and zero deaths in each listed group, alongside counts of participants with other treatment-emergent adverse events. These observations describe what was recorded in the study; they do not rule out uncommon events or establish comparative safety beyond the recorded conditions and sample.

Overall, the registry supports reporting the exposure estimates and their confidence intervals, but the missing acceptance limits and analysis details constrain how far the result can be interpreted. It is appropriate to distinguish what the posted data quantify from conclusions that require information not reported in the registry record.

14. Why This Trial Matters Statistically

This trial illustrates why bioequivalence interpretation relies on an estimate and its uncertainty rather than on a visual comparison of two group averages. The reported ratios are near 100%, but the confidence intervals describe the range of values compatible with the sponsor's analysis under its assumptions. A formal equivalence assessment also requires prespecified margins. Without those margins in the registry record, numerical closeness to 100% is informative but not a complete decision rule.

The crossover design is another important statistical feature. Comparing products within participants can account for stable participant-level differences, but the analysis must reflect the paired structure and relevant period or sequence effects. The registry provides the sponsor's ratios and intervals but not the method used to derive them. Separate group summaries cannot reproduce a paired, model-based comparison without the necessary within-person information.

The distinction between descriptive variability and uncertainty in the treatment contrast also matters. Geometric coefficients of variation describe dispersion within the listed groups, while the confidence intervals around the ratios quantify uncertainty in the comparisons. Those quantities answer different questions. They should not be treated as interchangeable.

Finally, the study separates fasting and fed comparisons and reports different analysis counts for those conditions. Complete participant flow and missing-data details would help readers understand why the counts differ and how the analysis sets were formed. These reporting limits do not invalidate the posted estimates, but they narrow what can be independently assessed. The trial is a useful example of how a clear statistical reading can preserve the reported results while keeping the limits of the registry record explicit.

15. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The registry does not report the formal bioequivalence acceptance limits. This is a central limitation for interpreting whether the reported confidence intervals satisfy the study's decision rule. The estimates and intervals can be stated, but an independent formal conclusion cannot be reached from the available record alone.

The statistical method for the reported ratios is not specified. The record does not provide the model terms, handling of the crossover structure, or the estimation procedure. This limits independent verification of the sponsor's comparisons and prevents reconstruction from the posted geometric means and coefficients of variation.

The record reports 37 participants for each fasting product group and 16 for each fed product group, but does not provide full participant flow or explain the difference in analysis counts. It also does not give detailed reasons for exclusions, missing concentration data, or all rules for handling profiles that could not be evaluated. The stated primary analysis-set criteria are not a complete account of eligibility and analysis decisions.

Multiplicity procedures, interim analyses, the sample-size justification, and the full statistical analysis plan are not reported in the registry record. The posted outcome and analysis counts do not establish how many comparisons were planned or how results were selected for reporting. The interpretation of the two food-condition analyses therefore has to remain tied to the information actually recorded.

The study was open-label and involved healthy participants receiving single doses. The registry reports no serious adverse events and no deaths in the listed groups, but the safety information does not give event terms, timing, exposure details, or comparative testing. These features limit conclusions about broader safety, clinical outcomes, repeated use, and populations beyond the study participants.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Glucophage Extended Release (GXR) China Bioequivalence Study (Nantong - Darmstadt) — NCT03566810.
• Registry record last updated 2020-01-09; status last verified 2019-12.
• Results first posted 2020-01-09.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

19. Record Summary

NCT03566810 is a completed Phase 1 randomized, open-label, two-way crossover study sponsored by Merck KGaA, Darmstadt, Germany. It enrolled 54 healthy participants and compared single 500 mg doses of Glucophage Extended Release manufactured in Nantong, China, with the reference product manufactured in Darmstadt, Germany, under fasting and fed conditions.

The primary endpoint was metformin AUC 0-t. The registry reports a test-to-reference ratio of 98.69% (90% confidence interval 92.20% to 105.64%) for fasting administration and 99.39% (90% confidence interval 93.15% to 106.05%) for fed administration. The acceptance limits and analysis method are not reported in the registry record.

No serious treatment-emergent adverse events or deaths are reported in the listed groups. Other treatment-emergent adverse events are recorded for 10, 11, 3, and 4 participants across the test fasting, reference fasting, test fed, and reference fed groups, respectively.