A Study of LY3372689 to Assess the Safety, Tolerability, and Efficacy in Participants With Alzheimer's Disease: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study of LY3372689 to Assess the Safety, Tolerability, and Efficacy in Participants With Alzheimer's Disease (NCT05063539), a randomized phase 2 study. The primary outcome is Change From Baseline to End Time Point in Integrated Alzheimer's Disease Rating Scale (iADRS) (Intermediate (Low-medium) Tau Population).

2. Clinical Question

3. Trial Design

NCT05063539 is a completed, randomized, double-blind, parallel-group phase 2 trial sponsored by Eli Lilly and Company. The registry record lists Alzheimer disease as the condition and reports enrollment of 327 participants. The study began on 2021-09-16, and primary completion was recorded as 2024-07-09.

Participants were assigned to one of three groups: 0.75 mg LY3372689, 3 mg LY3372689, or placebo. All treatments were administered orally once daily during the double-blind period, which could continue for up to 124 weeks. The arm descriptions also state that participants entered a 4-week post-treatment follow-up period after their last blinded treatment. The record does not describe the allocation ratio or the method used to generate or conceal the allocation sequence.

The primary endpoint was change from baseline to Week 100 in iADRS among participants with intermediate baseline tau and baseline or post-baseline data for that outcome. Its analysis used a Bayesian disease progression model adjusted for age at baseline, AChEI/Memantine use at baseline, and pooled investigator. The posted arm counts for this endpoint were 86 for 0.75 mg, 85 for 3 mg, and 84 for placebo. These are endpoint analysis counts, not necessarily the full number randomized to each group.

The registry record lists 15 outcome measures, including the primary endpoint and secondary measures of cognition, function, and clinical status. It reports 52 analyses. Detailed information about the recruitment process, eligibility criteria beyond the disease and tau descriptions, adherence, and the handling of participants who discontinued treatment is not reported in the registry record.

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.

Before assignment: The study consists of a double-blind treatment period with a 124-week (wk) duration followed by a post treatment follow-up (PTFU) period for 4 weeks after last blinded treatment dose and Post treatment observational extension (PTOE) period for approximately 9 months on average since last blinded treatment dose.

Double-Blind Treatment Period

Reasons for not completing this period:

Post-Treatment Follow-Up Period

Reasons for not completing this period:

Post-Treatment Observational Extension

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: All randomized participants with baseline or post-baseline data for this outcome with baseline Intermediate Tau level.

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: All randomized participants.

6. Primary Endpoint

The registry lists the primary outcome as Change From Baseline to End Time Point in Integrated Alzheimer's Disease Rating Scale (iADRS) (Intermediate (Low-medium) Tau Population), measured over Baseline, Week 100.

The primary endpoint is change from baseline to Week 100 in the Integrated Alzheimer’s Disease Rating Scale, or iADRS, for participants with intermediate baseline tau. The scale combines the 13-item Alzheimer’s Disease Assessment Scale-Cognitive Subscale (ADAS-Cog13) with the Alzheimer’s Disease Cooperative Study-Instrumental Activities of Daily Living scale (ADCS-iADL). It is intended to capture cognitive and functional performance together, rather than treating those domains as entirely separate outcomes.

The registry describes the iADRS as ranging from 0 to 144. Lower scores indicate worse performance, while higher scores indicate better performance. A change score therefore represents movement from each participant’s baseline score by the specified assessment time. A between-group difference in change compares the model-estimated average change for a LY3372689 group with the corresponding placebo change. The direction of that contrast matters because positive and negative differences have different implications on a scale where higher scores indicate better performance.

The endpoint is restricted to the intermediate tau population, and its timeframe is baseline to Week 100. That distinction is important: the primary estimate does not represent every randomized participant or necessarily the overall trial population. The posted analysis includes participants with baseline or post-baseline data for this endpoint. Further details about how missing measurements were handled, how the intermediate tau category was defined, and whether the Week 100 measurement was imputed are not reported in the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does the primary estimate represent?

The primary endpoint is continuous: it is a score on a scale, and the registry identifies the reported measure as a mean. A mean describes the average value in a group, while a change-from-baseline measure focuses on how scores move relative to participants’ own baseline measurements. The primary comparison is a model-estimated difference in change between a dose group and placebo, not a direct comparison of final scores alone.

The posted primary estimates are posterior mean differences from a Bayesian disease progression model. In Bayesian analysis, a posterior distribution represents the uncertainty about an estimated quantity after combining the model and data with the specified prior information. A posterior mean is the average of that distribution. A credible interval gives a range containing a stated proportion of the posterior distribution under the model. The record reports 95% credible intervals, but the prior specification and further details of the Bayesian model are not reported in the registry record.

How should the interval be read?

An interval communicates uncertainty as well as the direction and scale of an estimate. If an interval includes zero, the reported model leaves room for a null difference and for differences in either direction. If it lies entirely on one side of zero, the model’s interval is consistent with a difference on that side. This does not by itself establish clinical importance, prove a mechanism, or show that the same pattern would occur in another population.

For the primary endpoint, the interval belongs to a Bayesian posterior analysis. The term “95% credible interval” has a Bayesian interpretation conditional on the model and its assumptions. It should not be silently relabeled as a confidence interval. In contrast, the secondary analyses report 95% confidence intervals for their model-based least-squares mean differences. Confidence intervals and credible intervals are both ways to express uncertainty, but they arise from different inferential frameworks and should be described using the terminology attached to each result.

What is adjusted in the primary model?

The registry says that the disease progression model adjusted for age at baseline, AChEI/Memantine use at baseline, and pooled investigator. Adjustment incorporates these listed characteristics into the model’s estimation. It can make the estimate conditional on those model terms, but it does not mean that every potential source of bias or imbalance has been removed. The record does not give further details of the model structure, prior distributions, or checks of model fit.

How do the secondary recalculations differ?

The secondary comparisons include posted least-squares means, standard errors, and sponsor-reported comparisons. The accompanying recalculated comparisons are simple, unadjusted calculations from those posted summary statistics. They are not replacements for the sponsor’s model-based analyses. For example, combining group-level estimates and their standard errors does not reproduce the participant-level model, its covariate adjustment, or any handling of incomplete observations. The reconstructed standard deviation is also approximate when inferred from standard errors and group sizes.

Unadjusted recalculations can be useful as a check on direction and rough scale, but they answer a narrower question. They do not account for the full analysis model or establish that the model assumptions are appropriate. Differences between a recalculation and a reported analysis may reflect adjustment, model specification, rounding, or analysis populations. For the primary outcome, the posted comparison is already model-based, and the record does not provide the group summaries needed to make an equivalent simple recalculation.

What do the planning illustrations mean?

The sample-size illustrations are prospective planning calculations based on the observed effect, using a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. Rows for fractions of the observed effect consider hypothetical smaller effects. They are not the trial’s original sample-size justification, which is not reported in the registry record, and they are not observed power. Observed power is not a useful substitute for interpreting an estimated effect and its uncertainty; the planning calculations instead show how sample-size requirements can change under stated assumptions about the effect to detect.

These illustrations give participants per arm for the stated power targets, except that time-to-event planning displays total events. The continuous-outcome illustrations are especially sensitive to the observed effect and variability used in the calculation. They should not be read as evidence that the trial was designed around those assumptions, or as proof that a study of a particular size would produce a particular result. The assumptions are explicit, but a full design rationale and the trial’s planned power calculation are not reported in the registry record.

Why does over-dispersion matter in general?

In general, over-dispersion occurs when observed variability is greater than a statistical model expects. When that happens, standard errors can be too small if the model does not account for the extra variation, making intervals appear narrower than the uncertainty warrants. The registry entries do not identify an over-dispersion issue for this trial, so this is a general point about interpreting model-based analyses, not a finding about LY3372689.

10. Analyses Reported by the Sponsor

The primary analysis is reported as a Bayesian disease progression model with adjustment for age at baseline, AChEI/Memantine use at baseline, and pooled investigator. Its results are posterior mean differences with 95% credible intervals. There is no separate simple recalculation of the primary endpoint from posted arm means because the record does not report the arm-level primary means or their dispersion.

For the secondary outcomes with detailed summary statistics, the recalculated contrasts are unadjusted approximations, whereas the registry’s own comparisons are reported as least-squares mean differences using NCS2. Those approaches need not produce identical estimates, intervals, or p-values. The figures are broadly similar in direction and scale in some comparisons, but the appropriate interpretation of the posted outcome analysis is the registry’s own reported analysis, which uses its stated method and analysis population. No discrepancy flag is recorded for these results.

The recalculated p-values and planning illustrations should therefore be treated as supplementary statistical checks, not as the trial’s prespecified or sponsor-reported inferential results. In particular, a difference between an unadjusted calculation and an adjusted model result is not, by itself, evidence of an error. The registry record does not provide enough detail to reconstruct all model specifications or explain every difference between the approaches.

11. Multiplicity, Missing Data and Analysis Populations

The record lists 15 outcome measures and 52 analyses, creating multiple opportunities to compare groups, populations, and endpoints. The primary endpoint is identified, but a family-wise error strategy, hierarchical testing sequence, false-discovery procedure, or other multiplicity adjustment is not reported in the registry record. Without such information, secondary p-values should be interpreted cautiously: the more analyses considered, the greater the chance that some nominally notable findings may occur by chance. The record does not allow a determination of which secondary findings, if any, were intended to be confirmatory.

Missing-data handling is also not reported in the registry record. The primary population is described as randomized participants with baseline or post-baseline data for the outcome and intermediate baseline tau. That definition identifies who contributed relevant data, but it does not explain what happened to participants with no post-baseline measurement, how intermittent missing visits were treated, or whether a particular imputation or likelihood-based approach was used. The secondary outcome summaries likewise do not establish the assumptions made about missing observations.

Analysis populations differ across outcomes: the primary result uses the intermediate tau population, while some secondary measures include an overall population as well as an intermediate tau population. The record does not describe a full intention-to-treat analysis set, a per-protocol set, or rules for inclusion in each analysis beyond the primary population definition. It also does not report a sensitivity analysis assessing how missing data or analysis-population choices might affect conclusions.

These omissions matter because both multiplicity and missingness can change how much confidence to place in an isolated estimate. They do not demonstrate that the analyses were flawed; they mean the registry record does not provide enough information to assess those aspects of the statistical plan in detail.

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 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: prostate cancer (3.4%), atrial fibrillation (2.7%), pneumonia (1.9%), pulmonary embolism (1.9%), fall (1.8%).

During the double-blind period, the record reports serious adverse events for 13 of 108 participants in the 0.75 mg group, 29 of 110 in the 3 mg group, and 16 of 108 in the placebo group. These are counts of participants affected among those at risk in each group, as reported in the registry. They should not be interpreted as evidence that a particular event was caused by treatment. The record does not provide event-level timing, exposure-adjusted rates, or a causal assessment for these counts.

Deaths during the double-blind period were reported for 0 participants in the 0.75 mg group, 1 in the 3 mg group, and 1 in the placebo group. The post-treatment follow-up and observational extension groups have separate serious adverse event counts and death counts. Those periods should be kept distinct from the double-blind comparison because they represent a different follow-up period and may include different numbers of participants at risk.

The registry also lists serious adverse event terms including prostate cancer, atrial fibrillation, pneumonia, pulmonary embolism, fall, and angina unstable, with the percentages stated in the record. Such term-level percentages describe reported events; they do not establish that the study drug caused them. Comparisons of safety require attention to exposure, follow-up, event definitions, and participant-level context, details that are not all reported in the registry record.

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

The primary result describes a model-estimated difference in iADRS change between each LY3372689 group and placebo among participants with intermediate baseline tau who had baseline or post-baseline data. The estimates point in different directions for the two doses: the 0.75 mg point estimate is positive, while the 3 mg estimate is negative. Their credible intervals also differ in whether they include zero. This is the statistical pattern reported for the primary endpoint, not a conclusion about every participant or every outcome.

The result does not by itself establish why the groups differed, whether any estimated difference is clinically important, or whether a similar pattern would occur in another study population. The registry record does not state a threshold for clinical importance. Nor does the primary comparison answer whether participants experienced a particular symptom, how individual trajectories varied, or what happened after the reported assessment timeframe.

Secondary outcomes and adverse-event counts broaden the description of the study, but they answer different questions and use different populations and timeframes. A secondary result should not be substituted for the designated primary endpoint, and a safety count is not a causal attribution. The presence of multiple outcome analyses also means isolated secondary findings need careful interpretation, particularly because a multiplicity strategy is not reported in the registry record.

Finally, these are registry results and descriptions. The record includes the trial status, outcome definitions, estimates, and selected safety counts, but it does not report all protocol and analysis details needed to evaluate every assumption. The findings can be described precisely within their stated population, measure, and timeframe, while important uncertainties about generalizability, missing data, model details, and clinical importance remain.

14. Why This Trial Matters Statistically

This trial illustrates why the scale, population, and analysis method must be read together. The primary endpoint combines cognitive and functional components, but the estimate is specific to participants with intermediate baseline tau and to change through Week 100. A result in that group is not automatically a result for the overall enrolled population. The secondary outcomes broaden the measurement scope, yet their populations and timeframes do not all match the primary analysis.

It also demonstrates that a point estimate alone is not enough. The 95% credible intervals convey uncertainty around the primary posterior means, and the two dose comparisons have different interval patterns and directions. Statistical interpretation should retain that uncertainty rather than reduce the result to a single label. The adjusted Bayesian analysis is distinct from the simple unadjusted calculations that can be made from some secondary summaries.

Finally, the registry contains many analyses, while the multiplicity strategy and important missing-data details are not reported in the registry record. This makes transparent reporting of endpoints, populations, model assumptions, and analysis plans especially important. The trial’s posted results support a bounded statistical description, but do not justify treating every outcome comparison as equally confirmatory or extending a finding beyond the population and timeframe studied.

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.

Several trial-specific details needed for a fuller statistical assessment are not reported in the registry record. These include the allocation ratio and sequence-generation or concealment procedures; detailed eligibility criteria; the definition and measurement procedure for the intermediate tau category; and the full contents of the protocol and statistical analysis plan. The record identifies the primary model’s adjustment variables but does not provide its full specification, prior distributions, model diagnostics, or sensitivity analyses.

The primary endpoint results are not accompanied by arm-level means or dispersion estimates, limiting independent reconstruction of the primary comparison. The posted primary estimates are model-based, and no primary p-values are reported. For some secondary outcomes, recalculations rely on group summaries and approximate variability, so they cannot reproduce the sponsor’s participant-level analyses. The reported secondary analyses also cover different populations and assessment timeframes.

Further details about attrition, treatment adherence, exposure duration, handling of missing measurements, and the reasons for discontinuation are not reported in the registry record. The record does not state a multiplicity strategy despite listing many analyses, nor does it provide a threshold for clinical importance on the primary scale. These gaps limit assessment of robustness and clinical interpretation. They do not establish that the underlying analyses were inappropriate; they constrain what can be concluded from the registry record alone.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: A Study of LY3372689 to Assess the Safety, Tolerability, and Efficacy in Participants With Alzheimer's Disease — NCT05063539.
• Registry record last updated 2026-06-12; status last verified 2026-05.
• Results first posted 2025-07-28.
• 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

NCT05063539 is a completed randomized, double-blind, parallel-group phase 2 trial of oral LY3372689 versus placebo in participants with early symptomatic Alzheimer’s disease. Enrollment was 327. The study began on 2021-09-16, with primary completion on 2024-07-09. Participants received 0.75 mg, 3 mg, or placebo once daily during a double-blind period lasting up to 124 weeks, followed by a stated 4-week post-treatment follow-up.

The primary endpoint was change from baseline to Week 100 in iADRS among participants with intermediate baseline tau and baseline or post-baseline outcome data. The reported Bayesian model estimates differ in direction for the two dose comparisons, and the uncertainty intervals are not alike. Secondary outcomes and safety counts provide additional, but not definitive, context. Important details about multiplicity, missing data, and the full statistical plan are not reported in the registry record.