A Study of Lampalizumab Intravitreal Injections Administered Every Two Weeks or Every Four Weeks to Participants With Geographic Atrophy: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study of Lampalizumab Intravitreal Injections Administered Every Two Weeks or Every Four Weeks to Participants With Geographic Atrophy (NCT02288559), a randomized phase 2 study. The primary outcome is Change From Baseline in Geographic Atrophy (GA) Area, as Assessed by Fundus Autofluorescence (FAF) at Week 24.

2. Clinical Question

3. Trial Design

This was a phase II, multicenter, randomized, parallel-group, single-masked study with sham injection control. The registry lists enrollment of 96 and status as completed. The study began on 2015-03-30, and primary completion was recorded as 2017-06-02. The randomized treatment period lasted 24 weeks. The record describes a safety run-in assessment before enrollment in the randomized study, with participants in the run-in receiving lampalizumab every two weeks. The number included in the run-in is not reported in the registry record.

The randomized treatment groups received 10 mg lampalizumab intravitreally every two weeks, 10 mg every four weeks, or sham injections. Sham injections were intended to mimic the lampalizumab injections. The registry describes the study as single-masked but does not identify which study roles were masked or provide further details about masking procedures. Allocation was randomized; details of sequence generation and concealment are not reported in the registry record.

The main outcome was a continuous measure: change from baseline in geographic atrophy area assessed by fundus autofluorescence at Week 24. The registry reports mean changes and standard errors for the analysis groups. Its primary analysis population was modified intent-to-treat, defined as randomly assigned participants who had at least one post-baseline measurement of geographic atrophy area. The number analyzed was based on data available at the specified time point.

Other listed outcomes covered serum concentrations of lampalizumab for both dosing schedules, ocular and systemic adverse events, and anti-lampalizumab antibodies. The registry lists five secondary outcomes and six outcome measures overall. The sample-size justification, allocation ratio planned by the trial team, original power target, and detailed analysis protocol are 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: A total of 332 participants were screened and 92 participants were randomized out of which 3 participants from one site were removed from the randomized population due to serious good clinical practice (GCP) noncompliance. Out of 89, 4 participants were excluded from the randomized population as they did not have any post-baseline measurement. .

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Modified intent-to-treat (mITT) population included participants who were randomly assigned to study treatment and had at least one post-baseline GA area measurement. Number analyzed is the number of participants with data available for analysis at the given time-point.

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: Modified intent-to-treat (mITT) population included participants who were randomly assigned to study treatment and had at least one post-baseline geographic atrophy (GA) area measurement.

6. Primary Endpoint

The registry lists the primary outcome as Change From Baseline in Geographic Atrophy (GA) Area, as Assessed by Fundus Autofluorescence (FAF) at Week 24, measured over Baseline, Week 24.

The primary endpoint was change from baseline in geographic atrophy lesion area, measured by fundus autofluorescence at Week 24. Geographic atrophy involves loss of photoreceptors and surrounding retinal cells; the resulting lesions can cause vision loss. In this endpoint, the area of the lesion is the measured quantity, rather than visual acuity or another direct measure of vision. The registry notes that a positive change means the lesion area increased, indicating worsening or disease progression.

Fundus autofluorescence images were analyzed by a central reading center. Central image assessment provides a common measurement process across participants and sites, which is useful when an endpoint depends on image interpretation. The registry identifies baseline and Week 24 as the measurement time points for the reported change. It does not provide further details about image grading procedures, repeatability, or how readings were reconciled.

The outcome is continuous, and the registry reports group means with standard errors. A mean describes the average change in lesion area for the analyzed participants; it does not describe every participant's individual course. The reported analyses compare group means, with the sponsor's analyses using adjusted means. Change in lesion area is a structural measure of geographic atrophy. These results alone do not establish the degree of change in visual function or its importance to individual participants.

7. Results Posted in the Registry

• Sham Q4W: mean 1.12 versus 0.61 on control; difference 0.51 (95% CI -0.00 to 1.02), standardized effect d = 0.87; z-test p = 0.051.
• Lampalizumab Q2W: mean 1.05 versus 0.61 on control; difference 0.43 (95% CI 0.02 to 0.85), standardized effect d = 0.71; z-test p = 0.038.
• Lampalizumab Q4W: mean 0.91 versus 0.61 on control; difference 0.30 (95% CI -0.14 to 0.74), standardized effect d = 0.51; z-test p = 0.186.

8. Statistical Interpretation

9. Statistical Methods Explained

Several statistical ideas help interpret the outcome and comparisons. The primary measure is continuous: lesion area change in mm^2. For a continuous outcome, a group mean gives the average measured change among participants included in that analysis. It does not imply that each participant had the same change, and it can be influenced by the distribution of individual measurements. The registry reports standard errors alongside the means. A standard error describes uncertainty in the estimated group mean; it is not the spread of individual participant measurements.

What does a mean difference represent?

A mean difference compares the mean outcome in one group with the mean in another. The posted recalculations compare each lampalizumab group with the corresponding sham group using the available group summaries. A positive difference in these recalculations means the lampalizumab-group mean change is larger than the sham-group mean change; a negative difference means it is smaller. Since positive lesion-area change denotes an increase, the sign is interpreted in relation to that endpoint definition. A difference is an estimate, not a certainty about the underlying group values.

The unadjusted calculations use the posted means and standard errors to estimate a between-group difference and its uncertainty. They do not account for the repeated measurements or other model features that may be used in the sponsor's analysis. The Q2W recalculated estimate is 0.435, while the Q4W recalculated estimate is 0.297. The corresponding unadjusted p-values are 0.038 and 0.186. These recalculations are descriptive checks based on summary statistics, not replacements for the registered trial analysis. A recalculation cannot reproduce covariate adjustment or a longitudinal model when the necessary participant-level data and model details are unavailable.

What is a repeated-measures model?

The sponsor reports use of a Mixed-Effect Model Repeated Measures analysis for the Q2W comparison and MMRM for the Q4W comparison. A repeated-measures model is generally used when outcomes are collected from the same participants at multiple time points. In general, such a model can represent the correlation among repeated measurements and estimate group differences while using the longitudinal data structure. The registry does not report the model's covariates, covariance structure, handling of missing observations, or other implementation details. The specific adjustment embodied in the reported estimates therefore cannot be fully assessed from the registry record.

The sponsor's adjusted mean difference for Q2W is 0.435, with an 80% confidence interval from 0.162 to 0.707 and p = 0.043. For Q4W, the adjusted mean difference is -0.210, with an 80% confidence interval from -0.491 to 0.071 and p = 0.336. A model-based estimate can differ from a raw difference because it is adjusted and reflects the model's use of repeated observations. The Q4W adjusted estimate has the opposite sign from the simple unadjusted comparison of the displayed Week 24 group means. The discrepancy indicator is true; accordingly, the sponsor's own analysis takes precedence over the unadjusted recalculation.

How should p-values be understood?

A p-value describes how compatible the observed data are with a specified null hypothesis under the analysis assumptions. It is not the probability that the null hypothesis is true, and it does not measure clinical importance. The sponsor-reported p-values are 0.043 for Q2W and 0.336 for Q4W. The unadjusted recalculations report p = 0.038 and p = 0.186, respectively. Because the analyses differ, their p-values should not be treated as interchangeable. The registry does not report whether, or how, the analyses were adjusted for multiple comparisons.

General statistical interpretation also depends on assumptions, including the appropriateness of the model and the way missing data are handled. Those details are not reported in the registry record. The endpoint analysis population required at least one post-baseline measurement, and the number analyzed reflects availability at the time point. This definition clarifies who could contribute to the analysis but does not reveal how many randomized participants lacked measurements or why data were missing.

The outcome means are reported with standard errors, and confidence intervals are reported for comparisons. The standard errors describe precision of group mean estimates, whereas a confidence interval conveys uncertainty about a between-group estimate. Neither by itself describes variation in individual responses. No dispersion measure for individual participant changes is reported in the endpoint summary.

10. Confidence Intervals and Precision

The sponsor's Q2W adjusted mean difference is 0.435, with an 80% confidence interval from 0.162 to 0.707. The interval is entirely above zero, so the reported model-based estimate is compatible with a positive difference across the interval's range. Its width also shows that the estimate is not exact: plausible values under the analysis extend from a smaller positive difference to a larger one. The sponsor reports p = 0.043 for this comparison.

For the Q4W adjusted mean difference of -0.210, the 80% confidence interval runs from -0.491 to 0.071. This interval includes zero and spans both negative and positive values. It therefore leaves uncertainty about the direction of the adjusted difference under the reported analysis. The accompanying p-value is 0.336. The interval does not establish that the groups are identical; rather, it reflects uncertainty that includes no difference as well as differences in either direction.

The recalculated unadjusted comparisons use 95% confidence intervals: 0.023 to 0.847 for Q2W and -0.143 to 0.737 for Q4W. These intervals should not be compared with the sponsor's intervals as though they were the same analysis: the confidence levels differ, and the sponsor's estimates are adjusted. The Q2W recalculated interval is entirely above zero, while the Q4W recalculated interval includes zero. The recalculated results are simple unadjusted checks, and the reported discrepancy means the sponsor's analysis takes precedence.

In general, a confidence interval describes the range of estimates reasonably compatible with the data and the method used, under its assumptions. It is not a range containing a specified fraction of individual participants' outcomes. Precision depends on factors such as the amount of information and variability; the smaller analyzed groups, particularly the sham groups, can make comparisons less precise. The registry does not report confidence intervals for the individual group means.

11. 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 sample-size figures are planning illustrations based on the observed effect, not the original trial design calculation. They are participants per arm for 80% power and participants per arm for 90% power. They are not total-event counts: the event-count convention applies only to time-to-event planning, whereas these comparisons concern a continuous outcome. The trial's own original sample-size assumptions and justification are not reported in the registry record.

The illustrations use a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. They consider detecting the full observed effect, 75% of the observed effect, or 50% of the observed effect. For example, the full-effect Q2W planning row gives 32 participants per arm for 80% power and 42 participants per arm for 90% power. These figures are hypothetical planning requirements under the stated assumptions, not the sizes actually analyzed in the trial.

The 80% power column means that, if the assumed effect and other planning assumptions were correct, a study with that number of participants per arm would have an 80% chance of rejecting the null hypothesis using the specified test. The 90% column expresses the same planning idea with a higher target chance of detecting that assumed effect. The figures are not evidence that the completed study had a particular probability of finding a result, and this is not observed power.

Smaller effects require more participants to detect under the same assumptions. Thus the planning rows for a fraction of the observed effect call for larger numbers than the row using the full observed effect. These calculations rely on the observed estimates, which are uncertain, and the normal-approximation planning assumptions. They do not account for unknown details of the original protocol, attrition, or any design features not stated in the registry record.

12. 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 = 0.507; pooled standard deviation σ = 0.580.
2. n = 2 × (1.96 + 0.842)² × 0.580² / 0.507² = 20.6, rounded up to 21 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 <- 0.507                      # observed difference in means
sigma <- 0.580452409763281                     # 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(2288559)
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 = 21 (80% power), 28 (90% power)
 75% of observed effect: n per arm = 37 (80% power), 49 (90% power)
 50% of observed effect: n per arm = 83 (80% power), 111 (90% power)
power.t.test: 21.6 
Simulated power at n = 21 per arm: 0.788 (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.

13. Analyses Reported by the Sponsor

The registry reports two sponsor analyses using repeated-measures methods. For Lampalizumab Q2W compared with Sham Q2W, the difference in adjusted means is 0.435, with an 80% confidence interval from 0.162 to 0.707 and p = 0.043. For Lampalizumab Q4W compared with Sham Q4W, the difference in adjusted means is -0.210, with an 80% confidence interval from -0.491 to 0.071 and p = 0.336.

The unadjusted recalculations from the posted group summaries give 0.435 for Q2W and 0.297 for Q4W. The Q2W point estimates match, but the uncertainty measures and p-values differ because the sponsor used a repeated-measures adjusted analysis, while the recalculation is unadjusted. For Q4W, the point estimates also differ in direction. A raw comparison of the displayed group means cannot reproduce the sponsor's model-based estimate without participant-level data and full model details.

The registry record identifies a discrepancy. In keeping with that distinction, the sponsor's own analysis takes precedence; the recalculations should be treated as supplementary, unadjusted checks. The record does not provide sufficient model details to independently reproduce the sponsor estimates or determine the source of the Q4W difference.

14. Multiplicity, Missing Data and Analysis Populations

The primary analysis population is described as modified intent-to-treat. It includes participants randomly assigned to study treatment who had at least one post-baseline geographic atrophy area measurement. The number analyzed is defined as the number with data available for analysis at the given time point. The registry reports endpoint group counts of 10 for Sham Q2W, 10 for Sham Q4W, 43 for Lampalizumab Q2W, and 22 for Lampalizumab Q4W. These analysis counts do not account for all enrollment, and the record does not give a complete flow of participants through screening, run-in, randomization, follow-up, and analysis.

The missing-data method for the primary outcome is not reported in the registry record. In particular, the record does not explain how missing Week 24 outcomes were handled, whether participants with incomplete follow-up contributed earlier measurements to the repeated-measures analysis, or whether sensitivity analyses assessed assumptions about missingness. The model name alone does not answer these questions.

The registry reports comparisons for two lampalizumab schedules against corresponding sham groups, as well as secondary outcomes concerning serum concentrations, adverse events, and antibodies. It does not report a multiplicity adjustment, a testing hierarchy, or how the family of outcomes and comparisons was controlled. Without that information, the p-values should be read as reported for their individual analyses, not as evidence that a stated overall type I error rate was maintained across all outcomes and comparisons.

The record does not report the original statistical analysis plan, whether analyses were prespecified in their reported form, or whether interim analyses were conducted. It also does not report detailed sensitivity analyses or subgroup methods. These omissions limit assessment of how analysis decisions, incomplete observations, and multiple comparisons may affect interpretation. They do not change the arithmetic of the reported estimates, but they limit the available context for evaluating their robustness.

15. 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: non-hodgkins lymphoma recurrent (9.1%), presyncope (9.1%), postural orthostatic tachycardia syndrome (4.5%), influenza (4.5%), fall (4.5%).

The registry reports serious adverse event counts among participants at risk in each safety group. For Lampalizumab Q2W, 7 of 46 participants were reported with serious adverse events; for Lampalizumab Q4W, 3 of 22. For Sham Q2W, 0 of 10 were reported, and for Sham Q4W, 1 of 11. These are descriptive group counts over the study's reported safety assessment, not adjusted comparisons. The safety denominators differ from some endpoint analysis counts, including the Sham Q4W denominator, so they should not be substituted for the primary-outcome analysis populations.

The registry lists serious-event terms and percentages including Non-hodgkins lymphoma recurrent at 9.1%, Presyncope at 9.1%, Postural orthostatic tachycardia syndrome at 4.5%, Influenza at 4.5%, Fall at 4.5%, and Embolic stroke at 4.5%. The denominator and event-count details for these term percentages are not reported in the registry record. The registry also reports 2 deaths in Lampalizumab Q2W and none in the other listed groups. The information reported does not establish whether deaths or adverse events were caused by study treatment. Counts alone do not support a causal comparison, particularly without further detail on event definitions, exposure, timing, and assessment.

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

The primary outcome records change in geographic atrophy lesion area, with positive values indicating an increase in area. The group means were positive for each schedule and control group. The adjusted Q2W comparison was estimated at 0.435, with its 80% confidence interval above zero, while the adjusted Q4W comparison was -0.210, with an 80% confidence interval that includes zero. These are estimates for the structural endpoint at Week 24, not direct measures of how participants saw or functioned.

The reported model-based comparisons and the simple unadjusted recalculations do not have the same analytic basis. In particular, the Q4W point estimate differs in direction between the sponsor's adjusted analysis and the unadjusted calculation. The registry flags a discrepancy, so the sponsor's analysis takes precedence. The recalculations can help readers see how raw group summaries relate to simple comparisons, but they cannot replace the sponsor's repeated-measures analysis.

Neither an interval that excludes zero nor a p-value considered small establishes clinical importance. The record does not provide a threshold for a clinically meaningful change in lesion area, nor does it report a direct visual-function outcome as the primary endpoint. Conversely, an interval containing zero does not prove there is no difference; it indicates uncertainty that includes zero under the reported method.

The findings should be understood within the study's limited assessment timeframe, the analyzed populations, and the incomplete reporting of model and missing-data details. They do not establish effects beyond Week 24, and they do not determine effects in populations not described by this trial. The registry record describes a study of exposure-response and safety; the reported outcome information alone is not a basis for treatment advice or for concluding that an intervention works or does not work.

17. Why This Trial Matters Statistically

This trial offers a useful statistical example because it compares two dosing schedules against corresponding sham schedules while using a continuous imaging outcome. The endpoint's direction is explicit: positive change means lesion area increased. That definition is essential for interpreting whether a positive or negative mean difference reflects a larger or smaller average change in the measured structure.

It also illustrates why the analysis method matters. The sponsor's results are based on repeated-measures models, whereas the recalculations use group summary statistics and are unadjusted. For Q2W, the sponsor and recalculation give the same point estimate but different confidence intervals and p-values. For Q4W, their point estimates differ in direction. Reading the estimates without distinguishing their methods could obscure this important limitation. The discrepancy flag supports giving the sponsor's analysis precedence while retaining the recalculation as a secondary check.

The study also makes the distinction between an outcome-specific analysis population and overall enrollment visible. The modified intent-to-treat definition requires at least one post-baseline measurement, and reported group counts vary across endpoint and safety information. Such differences matter when interpreting precision and when considering how missing measurements may affect results.

Finally, the registry supplies planning illustrations based on observed effects, but those calculations are not the trial's original design justification and are not observed power. This is a reminder that planning estimates and completed-trial results answer different questions. The registry's missing details about multiplicity, missing-data handling, and model specifications leave important uncertainty about robustness, even though the main estimates and safety counts are available.

18. 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 record does not report the full protocol or statistical analysis plan. The original sample-size justification, intended power, and design assumptions are not reported, so the planning illustrations based on observed effects should not be mistaken for the study's original rationale. The record also does not state whether interim analyses occurred.

Although sponsor analyses are described as repeated-measures models, details of their implementation are absent. The covariates, covariance structure, model diagnostics, and missing-data assumptions are not reported. This restricts independent evaluation and reproduction of the adjusted comparisons. The observed discrepancy between the sponsor's Q4W estimate and the simple unadjusted estimate further emphasizes that the group summaries do not capture all features of the sponsor analysis.

The primary analysis population requires at least one post-baseline outcome measurement. The record does not give a complete account of participants excluded from analysis, reasons for unavailable measurements, or sensitivity analyses. The endpoint counts also differ from safety denominators for Sham Q4W. These distinctions make it important not to assume that every randomized participant contributed to every outcome.

The reported primary outcome is a structural measurement of lesion area at Week 24. The registry does not give a clinical-importance threshold for the measure or report a primary visual-function outcome. The available information therefore cannot establish the practical importance of the estimated differences. The safety information is also limited: serious-event counts and selected terms are available, but detailed event narratives, timing, causal assessments, and full term denominators are not reported in the registry record.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: A Study of Lampalizumab Intravitreal Injections Administered Every Two Weeks or Every Four Weeks to Participants With Geographic Atrophy — NCT02288559.
• Registry record last updated 2019-09-25; status last verified 2019-09.
• Results first posted 2019-02-19.
• 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.

22. Record Summary

NCT02288559 is a completed phase II, multicenter, randomized, parallel-group, single-masked, sham injection-controlled study of intravitreal lampalizumab in participants with geographic atrophy. Enrollment was 96. The randomized treatment period lasted 24 weeks and compared 10 mg lampalizumab every two weeks or every four weeks with corresponding sham schedules. A safety run-in preceded the randomized study.

The primary outcome was change from baseline in geographic atrophy area assessed by fundus autofluorescence at Week 24. The endpoint population was modified intent-to-treat, requiring random assignment and at least one post-baseline measurement. Sponsor-reported repeated-measures analyses and unadjusted recalculations are both available, with a discrepancy identified for interpretation. The registry also lists serum concentration, adverse-event, and antibody outcomes. Details of the original sample-size rationale, multiplicity approach, missing-data handling, and full model specifications are not reported in the registry record.