A Study of Lanabecestat (LY3314814) in Participants With Mild Alzheimer's Disease Dementia: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study of Lanabecestat (LY3314814) in Participants With Mild Alzheimer's Disease Dementia (NCT02783573), a randomized phase 3 study. The primary outcome is Change From Baseline in Alzheimer´s Disease Assessment Scale- Cognitive Subscale (ADAS-Cog13) Score.

2. Clinical Question

The primary comparison concerns the first 78 weeks, when the active-treatment arms are compared with placebo. ADAS-Cog13 scores range from 0 to 85, and an increase indicates cognitive worsening. The registry reports adjusted least-squares mean changes and mixed-model comparisons, alongside the number of participants contributing to each arm-level result.

This question is one part of a broader evaluation. The registry lists functional, cognitive, clinical progression, neuropsychiatric, and cerebrospinal-fluid biomarker outcomes. The study was randomized, double-blind, placebo-controlled, and delayed-start, with lanabecestat and placebo listed as interventions.

3. Trial Design

This was a phase 3, randomized, triple-masked, parallel-group study of lanabecestat in participants with mild Alzheimer's disease dementia. AstraZeneca is listed as the sponsor. The registry lists an enrollment of 1722 and a terminated study status. The recorded start date is 2016-07-01, and the primary completion date is 2018-09-28. The reason for termination is not reported in the registry record.

The listed treatment groups were placebo, lanabecestat 20 mg, and lanabecestat 50 mg for the primary week 78 comparison. The arm descriptions also include delayed-start groups: participants initially received placebo for 78 weeks and then began lanabecestat 20 mg or 50 mg. The descriptions say treatment continued until week 156. The primary outcome's stated timeframe is baseline to week 78.

Randomization assigns participants to treatment groups by chance; masking is intended to limit the influence of treatment knowledge on trial conduct and assessment. These are general features of trial design. The registry identifies the study as triple-masked but does not specify in the record which roles were masked. Details of sequence generation, allocation concealment, or the reason for early termination are not reported in the registry record.

The registry lists 17 outcome measures and 28 analyses. Its primary outcome is change in ADAS-Cog13 score. The posted sample sizes for that outcome are 460 for placebo, 471 for lanabecestat 20 mg, and 464 for lanabecestat 50 mg. These outcome-specific counts are distinct from 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.

Before assignment: In Period 1, per the protocol, placebo-controlled groups (Placebo for 78 weeks then Lanabecestat 20 mg; Placebo for 78 weeks then Lanabecestat 50 mg) were combined to form one placebo group. As study terminated early and very few participants entered into the period 2, the arms in period 2 are combined based on dose exposure for ease of comparison.

Placebo-Controlled Treatment Period

Reasons for not completing this period:

Delayed-Start Period

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: All randomized participants who received at least one dose of study drug and have baseline and at least one post-baseline data for ADAS-Cog13 measure.

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 who received at least one dose of study drug.

6. Primary Endpoint

The registry lists the primary outcome as Change From Baseline in Alzheimer´s Disease Assessment Scale- Cognitive Subscale (ADAS-Cog13) Score, measured over Baseline, Week 78.

The primary endpoint is change from baseline in the Alzheimer’s Disease Assessment Scale–Cognitive Subscale, 13-item version (ADAS-Cog13), at week 78. It is a psychometric measure covering word recall, following commands, constructional praxis, naming, ideational praxis, orientation, word recognition, memory, comprehension of spoken language, word-finding, and language ability. The description also includes delayed word recall and concentration or distractibility.

The scale ranges from 0 to 85, with higher scores indicating cognitive worsening. A change score compares a participant’s score with their own baseline score; the reported least-squares means are model-based group estimates, rather than simple averages. The registry reports these estimates with standard errors and uses a mixed-model repeated-measures analysis for the primary endpoint.

ADAS-Cog13 is used in this trial as a measure of cognition, one domain relevant to the study’s evaluation of lanabecestat in mild Alzheimer’s disease dementia. It is not a measure of every aspect of the condition. The registry therefore also lists outcomes covering daily function, clinical rating, neuropsychiatric symptoms, and a cerebrospinal-fluid biomarker. General interpretation: a score change is most informative when considered with its scale direction, the comparator result, and uncertainty around the estimated group contrast.

7. Results Posted in the Registry

• Lanabecestat 20 mg: mean 8.93 versus 6.42 on control; difference 2.51 (95% CI -0.74 to 5.76), standardized effect d = 0.10; z-test p = 0.130.
• Lanabecestat 50 mg: mean 6.20 versus 6.42 on control; difference -0.22 (95% CI -3.76 to 3.32), standardized effect d = -0.01; z-test p = 0.903.

8. Secondary Efficacy Endpoints

The registry also posts results for secondary outcomes. For each outcome below, the endpoint type, the test applied to the posted summary statistics, the result and its interpretation are given in turn. Secondary endpoints are usually not adjusted for multiple testing, so they support rather than confirm the primary finding.

Change From Baseline in Alzheimer´s Disease Cooperative Study Activities of Daily Living Inventory Instrumental Items Score (ADCS-iADL)

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard errors.
• Result:
  • Lanabecestat 20 mg: mean -6.91 versus -3.95 on control; difference -2.96 (95% CI -6.51 to 0.59), standardized effect d = -0.12; z-test p = 0.102.
  • Lanabecestat 50 mg: mean -7.13 versus -3.95 on control; difference -3.18 (95% CI -6.88 to 0.52), standardized effect d = -0.13; z-test p = 0.093.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Lanabecestat 20 mg and the control. The interval gives the range of effects that remain compatible with the data.
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Lanabecestat 50 mg and the control. The interval gives the range of effects that remain compatible with the data.

Change From Baseline in Functional Activities Questionnaire (FAQ) Score

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard errors.
• Result:
  • Lanabecestat 20 mg: mean 5.16 versus 3.93 on control; difference 1.23 (95% CI -1.07 to 3.53), standardized effect d = 0.08; z-test p = 0.295.
  • Lanabecestat 50 mg: mean 3.41 versus 3.93 on control; difference -0.52 (95% CI -2.91 to 1.87), standardized effect d = -0.03; z-test p = 0.670.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Lanabecestat 20 mg and the control. The interval gives the range of effects that remain compatible with the data.
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Lanabecestat 50 mg and the control. The interval gives the range of effects that remain compatible with the data.

Change From Baseline on the Integrated Alzheimer's Disease Rating Scale (iADRS) Score

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard errors.
• Result:
  • Lanabecestat 20 mg: mean -15.22 versus -10.46 on control; difference -4.76 (95% CI -10.12 to 0.60), standardized effect d = -0.13; z-test p = 0.082.
  • Lanabecestat 50 mg: mean -12.43 versus -10.46 on control; difference -1.97 (95% CI -7.58 to 3.64), standardized effect d = -0.05; z-test p = 0.492.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Lanabecestat 20 mg and the control. The interval gives the range of effects that remain compatible with the data.
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Lanabecestat 50 mg and the control. The interval gives the range of effects that remain compatible with the data.

The secondary outcomes broaden the assessment beyond cognition to include daily activities, functional ability, integrated disease rating, clinical rating, progression, neuropsychiatric symptoms, a cognitive screening measure, and a cerebrospinal-fluid amyloid biomarker. The registry lists 16 secondary outcomes. Detailed posted results in the record include ADCS-iADL, FAQ, and iADRS changes at week 78.

Across these posted results, the point estimates do not form one consistent pattern across measures and doses. Some estimated changes are more negative for lanabecestat than placebo, while other contrasts are smaller or point in the other direction. The reported confidence intervals for these detailed secondary comparisons include zero. This broad pattern does not provide a clear, uniform separation from placebo across the reported measures. It should be considered alongside, not as a replacement for, the primary ADAS-Cog13 result.

General explanation: secondary outcomes can provide context about whether findings extend across related domains, but they are often numerous and may be exploratory. When many outcomes or comparisons are examined, the chance of at least one apparently notable result arising by chance increases unless multiplicity is addressed. The registry record does not report a multiplicity-adjustment procedure for these secondary analyses. Accordingly, unadjusted secondary p-values carry less confirmatory weight than a prespecified primary analysis with a stated multiplicity strategy.

The record lists additional secondary endpoints but does not include their detailed numerical results in the information available for this analysis. Their direction or statistical significance should not be inferred from the endpoint names alone. The secondary findings are best read as a broad set of measurements with uncertainty, rather than as independent confirmations of a treatment effect.

Secondary outcomes registered for this trial, for reference (only those above were recalculated):

• Change From Baseline in Alzheimer´s Disease Cooperative Study Activities of Daily Living Inventory Instrumental Items Score (ADCS-iADL)
• Change From Baseline in Functional Activities Questionnaire (FAQ) Score
• Change From Baseline on the Integrated Alzheimer's Disease Rating Scale (iADRS) Score
• Change From Baseline in the Clinical Dementia Rating - Sum of Boxes (CDR-SB) Score
• Time to Progression as Measured by Loss of Clinical Dementia Rating (CDR) Global Score Stage
• Change From Baseline in Neuropsychiatric Inventory (NPI) Score
• Change From Baseline on the Mini-Mental State Examination (MMSE)
• Percent Change From Baseline in Concentration of Cerebrospinal Fluid (CSF) Biomarker Amyloid Beta (Aβ)1-42

9. Statistical Interpretation

10. Statistical Methods Explained

The primary result is based on a mixed-model repeated-measures (MMRM) analysis. General explanation: this class of model uses measurements across visits to estimate group patterns over time and can use available repeated observations without requiring every participant to have a measurement at every visit. Its conclusions depend on the model specification and assumptions. The registry gives the primary analysis factors and covariates but does not report all implementation details.

What does the model adjust for?

The registry says the MMRM included treatment, visit, and treatment-by-visit interaction as factors. It also included acetylcholinesterase inhibitor use at baseline and pooled site. The listed covariates were baseline ADAS-Cog13 total score, age at baseline, and the interaction between baseline ADAS-Cog13 total score and visit. In general, adjustment can account for measured variables specified in the model and can improve precision when appropriate. It does not eliminate uncertainty or account automatically for unmeasured factors.

The treatment-by-visit term allows the modeled treatment contrast to vary across visits rather than forcing one common contrast at every visit. The baseline-score-by-visit term allows the relationship between baseline score and outcome to vary by visit. These are general explanations of the listed terms; further details of their coding and estimation are not reported in the registry record.

What is a least-squares mean?

A least-squares mean is a model-based estimate of a group mean, adjusted for the other variables included in the model. It is not necessarily identical to the arithmetic average of observed participant values. The registry reports primary arm estimates of 6.42 for placebo, 8.93 for lanabecestat 20 mg, and 6.2 for lanabecestat 50 mg, with standard errors. These are least-squares means for the stated outcome and analysis.

The registry’s reported contrasts are labeled “LS Mean Difference (Final Values)” and use mixed models analysis. For the 20 mg comparison, the reported contrast is 2.51; for the 50 mg comparison, it is -0.21. Their confidence intervals and p-values express uncertainty under the sponsor-reported model. General explanation: a standard error describes the estimated sampling variability of an estimate, while a confidence interval presents a range reflecting that uncertainty in the original measurement units.

How does the recalculation differ?

The recalculated comparisons are simple, unadjusted comparisons derived from the posted arm-level summary statistics. They are not a reconstruction of the sponsor’s MMRM: the recalculation does not reproduce the model’s visit structure, covariate adjustment, or repeated-measures analysis. The pooled standard deviations used in the recalculation are approximate, as indicated in the posted calculation inputs. These recalculations are useful as a secondary check on the broad scale and direction of contrasts, not as a replacement for the registry-reported model.

General explanation: over-dispersion occurs when outcome variability is greater than a model assumes. If ignored, uncertainty can be understated and inferential conclusions can become too confident. The registry record does not report an over-dispersion assessment for this continuous outcome, so no conclusion about such an issue can be drawn for this trial.

What does the sample-size illustration assume?

The planning illustration uses a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. It considers an effect equal to a stated fraction of the observed effect. For these continuous outcomes, the figures are participants per arm. General explanation: power is the probability, under a specified effect and model, of detecting that effect at the chosen significance level. It is a design property under assumptions, not a measure of whether the observed trial result is true.

The registry does not report the trial’s original sample-size justification, its planned effect size, or its own design assumptions. The posted illustration should therefore be read as a retrospective planning aid based on observed results, not as documentation of the trial’s original statistical plan. Interim analyses, stopping boundaries, and a full prespecified statistical analysis plan are not reported in the registry record.

11. Confidence Intervals and Precision

For lanabecestat 20 mg compared with placebo, the registry reports an estimated ADAS-Cog13 contrast of 2.51, with a 95% confidence interval from -0.752 to 5.776. For lanabecestat 50 mg compared with placebo, the estimate is -0.21, with a 95% confidence interval from -3.725 to 3.296. Both intervals include zero, and each extends across values in both directions.

General explanation: a confidence interval describes the uncertainty around an estimate under the analysis model. A narrower interval indicates greater precision on the measurement scale; a wider interval indicates less precision. An interval that includes zero leaves open the possibility of no group contrast, but it can also include nonzero contrasts. It should not be interpreted as proof of equivalence or as a range containing every plausible value with certainty.

The precision should be understood in the context of the outcome-specific sample sizes: 460 participants contributed to the placebo result, 471 to the 20 mg result, and 464 to the 50 mg result. These counts do not establish that every randomized participant had complete follow-up. The primary analysis population required baseline and at least one post-baseline ADAS-Cog13 observation, and the registry does not report a complete account of missing measurements or the extent to which missingness differed by group.

The displayed confidence intervals are from the sponsor-reported mixed-model analyses, not the unadjusted recalculation. Their width reflects sampling uncertainty as represented by that analysis; it does not include every possible source of uncertainty, such as uncertainty about model assumptions or the relevance of the measure to patients. No clinical-importance threshold is reported in the registry record, so the intervals cannot be classified against such a threshold.

12. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The planning figures for these outcomes are participants per arm, not totals across arms or numbers of events. The calculation assumes a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. It evaluates effects set at the stated fraction of the observed effect. These are planning assumptions for the illustration; the trial’s original design assumptions and sample-size justification are not reported in the registry record.

The 80% power column gives the number of participants per arm that the planning calculation estimates would be needed to have an 80% chance of detecting the specified effect under those assumptions. The 90% power column gives the corresponding estimate for a 90% chance. These are hypothetical design calculations, not descriptions of what happened in the trial. This is not observed power.

Rows that use a smaller fraction of the observed effect require more participants per arm in the illustration. General explanation: smaller effects are harder to distinguish from random variation when other design features stay fixed, so a larger sample is generally needed to achieve the same planned power. A large planning figure based on a very small observed effect should not be read as a recommendation for a new study without considering the scientific question and other design inputs.

The calculation is based on an observed estimate and a simplified normal approximation. It does not reproduce the trial’s MMRM or establish the original target enrollment. The registry gives no basis for treating these illustrations as the sponsor’s prospective design plan.

13. How the Sample Size Is Derived

Power is the probability that a study of a given size rejects the null hypothesis when a specific true effect exists. It depends on four things: the effect to detect, the variability of the outcome, the significance level (here a two-sided α of 0.05, so z_1−α/2 = 1.960) and the desired power (z_1−β = 0.842 for 80% and 1.282 for 90%).

Formula

n is the number of participants in each arm, Δ is the difference in means to detect and σ is the common standard deviation.

Worked calculation for this trial

1. Observed difference Δ = 2.510; pooled standard deviation σ = 25.248.
2. n = 2 × (1.96 + 0.842)² × 25.248² / 2.510² = 1,588.3, rounded up to 1,589 participants per arm for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
delta <- 2.51                      # observed difference in means
sigma <- 25.2476151342599                     # pooled standard deviation
n_per_arm <- function(power, frac = 1) {
  ceiling(round(2 * (z_a + qnorm(power))^2 * sigma^2 / (delta * frac)^2, 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

# Check against R's built-in function (t-based, so very slightly larger)
cat("power.t.test:", round(power.t.test(delta = delta, sd = sigma, power = 0.80)$n, 1), "\n")

# Simulation: exact two-sample t-test using simulated means and variances
set.seed(2783573)
n    <- n_per_arm(0.80)
reps <- 20000
m_diff <- rnorm(reps, delta, sigma * sqrt(2 / n))             # difference in sample means
s2     <- sigma^2 * rchisq(reps, 2 * (n - 1)) / (2 * (n - 1))   # pooled sample variance
tstat  <- m_diff / sqrt(s2 * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(tstat) > qt(0.975, 2 * (n - 1))), reps))

Output from running the code

100% of observed effect: n per arm = 1,589 (80% power), 2,127 (90% power)
 75% of observed effect: n per arm = 2,824 (80% power), 3,781 (90% power)
 50% of observed effect: n per arm = 6,354 (80% power), 8,506 (90% power)
power.t.test: 1589.3 
Simulated power at n = 1,589 per arm: 0.801 (20000 replicates)

The simulation draws many sets of group means and variances at the planned sample size and counts how often the two-sample t-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

14. Analyses Reported by the Sponsor

The registry-reported analyses use mixed models analysis and provide least-squares mean differences, 95% confidence intervals, and p-values. The recalculations are simple unadjusted comparisons based on posted arm-level summary statistics. They are methodologically different, so modest differences between their estimates or uncertainty measures are expected and do not by themselves indicate an error.

For lanabecestat 20 mg versus placebo, the recalculated estimate is 2.51, while the registry-reported mixed-model estimate is also 2.51. For lanabecestat 50 mg versus placebo, the recalculated estimate is -0.22 and the registry-reported estimate is -0.21. The reported mixed-model p-values are 0.129 and 0.903, respectively. The registry record does not flag a discrepancy between its analyses and the recalculations.

The sponsor-reported mixed-model results should be used as the principal inferential results because they reflect the analysis described for the primary outcome. The recalculated comparisons are an approximate, unadjusted cross-check and should not be substituted for the sponsor’s model-based analysis. Neither set of results, by itself, establishes clinical importance.

15. Multiplicity, Missing Data and Analysis Populations

The registry lists 17 outcome measures and 28 analyses, including a primary outcome and multiple secondary outcomes. Testing several outcomes, doses, or time points creates a multiplicity issue: even if each test is considered separately, the chance of at least one false-positive finding can rise across a collection of tests. General statistical practice may address this through prespecified testing hierarchies, adjusted significance thresholds, or other procedures, but the registry record does not report a multiplicity plan for the posted results.

The record identifies the primary endpoint and describes its MMRM factors and covariates. It does not report the full statistical analysis plan, whether all analyses were prespecified, how the family of comparisons was defined, or whether any secondary analyses were adjusted for multiplicity. Therefore, the reported p-values should be read as presented and not assumed to have a multiplicity correction.

The primary analysis population consisted of randomized participants who received at least one dose and had baseline and at least one post-baseline ADAS-Cog13 measurement. This is the population definition reported for the primary endpoint. It is not a statement that every randomized participant was included in that analysis. The registry does not report enough information to characterize all exclusions from the primary analysis population.

MMRM can use available repeated observations under its modeling framework, but that general feature does not establish how missing data arose or whether the assumptions were appropriate in this trial. The extent, timing, and reasons for missing outcome data, and sensitivity analyses addressing missingness, are not reported in the registry record. Nor are details of interim analyses, stopping rules, or a prespecified plan for handling treatment discontinuation reported. These omissions limit assessment of the robustness of the inferential results, without changing the estimates and intervals the registry reports.

16. 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: psychotic disorder (8.3%), fall (0.9%), syncope (0.5%), epididymitis (0.5%), orchitis (0.5%).

The registry’s serious adverse event summaries are divided by treatment and period. In Period 1, the numbers at risk were 558 for placebo, 588 for lanabecestat 20 mg, and 568 for lanabecestat 50 mg. Serious adverse events affected 50, 50, and 46 participants in those respective groups. The listed deaths were 5, 2, and 3. These are counts, not rates, and the record’s group sizes differ.

For Period 2, the delayed-start lanabecestat 20 mg group had 14 participants at risk, with no serious adverse events and no listed deaths. The delayed-start 50 mg group had 12 participants at risk, with 1 serious adverse event and no listed deaths. The small numbers at risk in these groups should be kept in view when interpreting their counts.

The registry lists psychotic disorder at 8.3% among the top serious adverse event terms, followed by fall at 0.9%, and syncope, epididymitis, orchitis, and prostate cancer at 0.5% each. The record does not provide enough context in these figures alone to establish causation or compare risk while accounting for exposure duration. These summaries describe reported events; they do not determine whether a treatment is safe or unsafe.

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

The posted primary results describe week 78 ADAS-Cog13 changes and the modeled contrasts with placebo. The estimate for lanabecestat 20 mg is positive, while the estimate for 50 mg is close to zero and slightly negative. Both reported confidence intervals include zero. The data therefore do not provide a precise separation from placebo in either dose comparison under the reported analysis.

This is a statistical description of the registry results, not a conclusion about whether lanabecestat does or does not work. The p-values and confidence intervals do not establish that the groups are equivalent, nor do they show that every value within an interval is equally likely. The record does not report a threshold for a clinically important ADAS-Cog13 difference, so statistical estimates alone cannot establish clinical importance.

The outcome is one measure of cognition, even though it covers multiple cognitive tasks. It does not capture every aspect of Alzheimer's disease dementia or the full experience of participants. The registry’s other outcome domains can add context, but multiple secondary analyses and unreported multiplicity handling limit their confirmatory weight.

The study is listed as terminated. The reason for termination is not reported in the registry record and should not be inferred from the trial status or the posted estimates. The registry’s stated primary analysis population and available outcome counts also mean the results should not be interpreted as a complete account of all randomized participants’ follow-up.

General interpretation: a trial result is most useful when estimates are considered alongside uncertainty, design, outcome definitions, missing-data information, and the prespecified analysis plan. Several of those details, including the original sample-size justification and full multiplicity strategy, are not reported in the registry record. The posted results support describing the estimated contrasts and their precision, but not making conclusions beyond those data.

18. Why This Trial Matters Statistically

This trial is statistically informative as an example of why an estimate should be read together with its uncertainty and analysis method. The two dose comparisons have different point estimates, but both sponsor-reported confidence intervals include zero. Reporting the estimates in the original ADAS-Cog13 units makes the possible range of group contrasts more interpretable than relying on p-values alone.

The record also distinguishes a model-based primary analysis from an approximate unadjusted recalculation. The MMRM incorporates repeated outcome measurements and specified baseline covariates, while the simple recalculation uses summary statistics. Understanding that distinction prevents a rough cross-check from being mistaken for a reproduction of the trial’s analysis.

The delayed-start structure is relevant to the timing of interpretation. The primary comparison is at week 78, when placebo is the comparator; the listed delayed-start groups then begin lanabecestat. Results at that primary time point answer a different question from any later comparison after the delayed-start participants have begun active treatment.

Finally, the registry lists many outcome measures and analyses, but does not report multiplicity procedures, a full statistical analysis plan, or the original sample-size justification. These gaps illustrate why registry results are best interpreted with attention to what is reported and what remains undocumented. The posted planning illustrations can aid understanding of sample-size sensitivity, but they do not substitute for prospective design assumptions.

19. 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 limitations affect interpretation. The study is listed as terminated, but the reason for termination is not reported in the registry record. Without that information, the status cannot be used to explain the observed results or the availability of outcome data.

The primary outcome analysis includes randomized participants who received at least one dose and had baseline and at least one post-baseline ADAS-Cog13 measure. The record does not provide a complete account of how many randomized participants were excluded from this population, the timing and reasons for missing observations, or sensitivity analyses for missing data. This restricts assessment of how missingness may affect the estimates.

The registry describes the MMRM factors and covariates but does not report a full statistical analysis plan, multiplicity handling, interim analyses, or stopping rules. Its original sample-size justification and design assumptions are also not reported. The retrospective planning illustration is based on observed effects and should not be treated as the trial’s original plan.

The detailed secondary results cover only some of the listed secondary outcomes in the record. Multiple outcomes are listed, and no multiplicity adjustment is reported. Their results therefore warrant cautious interpretation and cannot be assumed to provide independent confirmation of the primary outcome.

Finally, the reported primary contrasts concern one cognitive scale at week 78. The registry does not report a clinical-importance threshold for the scale, and the outcome does not encompass all aspects of disease or function. These limitations narrow the conclusions that can be drawn from the posted statistical results.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: A Study of Lanabecestat (LY3314814) in Participants With Mild Alzheimer's Disease Dementia — NCT02783573.
• Registry record last updated 2019-12-03; status last verified 2019-10.
• Results first posted 2019-08-06.
• 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.

23. Record Summary

NCT02783573 is a phase 3 randomized, triple-masked, parallel-group, placebo-controlled and delayed-start study of lanabecestat in participants with mild Alzheimer's disease dementia. The registry lists 1722 enrolled participants and a terminated status. Its recorded start date is 2016-07-01, and primary completion is dated 2018-09-28.

The primary outcome is change from baseline to week 78 in ADAS-Cog13 score. Registry-reported mixed-model comparisons with placebo were 2.51 for lanabecestat 20 mg and -0.21 for lanabecestat 50 mg; both 95% confidence intervals include zero. The registry lists multiple secondary outcomes and adverse event summaries. The reason for termination, original sample-size rationale, multiplicity plan, and complete missing-data details are not reported in the registry record.