A Phase II Study to Evaluate the Efficacy and Safety of Oral Ceritinib in Patients With ALK-positive NSCLC Metastatic to the Brain and/or to Leptomeninges: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Phase II Study to Evaluate the Efficacy and Safety of Oral Ceritinib in Patients With ALK-positive NSCLC Metastatic to the Brain and/or to Leptomeninges (NCT02336451), a non-randomized phase 2 study. The primary outcome is Overall Response Rate (ORR) Per Investigator Assessment.

2. Clinical Question

3. Trial Design

This was a Phase II, multi-center, open-label study of oral ceritinib in ALK-positive NSCLC metastatic to the brain and/or leptomeninges. The registry lists non-randomized allocation and a parallel model, with five arms. Enrollment was 156. The recorded start date was 2015-04-01, the primary completion date was 2019-02-06, and the study status is completed.

The arm descriptions distinguish patients with brain metastases and no evidence of leptomeningeal carcinomatosis according to prior brain radiation and prior ALK-inhibitor exposure. A separate arm was for patients with leptomeningeal carcinomatosis, with or without active lesions at baseline imaging. The registry also describes restrictions on prior ALK-inhibitor treatment in some arms and notes that protocol amendments changed some eligibility details. These distinctions matter when interpreting arm-wise outcomes: the groups arose from different eligibility histories and were not randomly formed.

All participants received ceritinib; there is no registry-listed placebo or alternative-treatment comparator. Open-label means the study was not masked, while investigator assessment means the primary response classification was made by investigators. Some listed secondary intracranial response outcomes also used blinded independent review committee assessment, but that does not change the stated investigator-assessed primary endpoint.

The primary endpoint was overall response rate, defined as the percentage with a best overall confirmed complete or partial response in the whole body according to RECIST 1.1. The stated analysis population was the Full Analysis Set, consisting of patients who received at least one dose. The registry record does not report how many participants were included in that population, the arm-level primary endpoint results, or the method used to handle missing endpoint assessments.

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: Approximately 160 patients were planned to be enrolled.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: The Full Analysis Set (FAS) comprised of all patients who received at least one dose of ceritinib.

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: The Full Analysis Set (FAS) comprised of all patients who received at least one dose of ceritinib.

6. Primary Endpoint

The registry lists the primary outcome as Overall Response Rate (ORR) Per Investigator Assessment, measured over 43 months.

Overall response rate measures the proportion of participants whose best overall confirmed response met the criteria for complete response or partial response. In this study, the assessment covered the whole body and used RECIST 1.1, with the investigator assigning the response category. A complete response required disappearance of all non-nodal target and non-target lesions, normalization of target or non-target lymph node size, and no new lesion. A partial response required disappearance of all target lesions or at least a 30% decrease in the sum of target-lesion diameters from baseline, with non-target lesions not progressing or in complete response.

As a binary endpoint, each participant is classified as having or not having a qualifying confirmed response for the purpose of calculating the percentage. The endpoint is used in cancer studies to describe the frequency of measurable tumor shrinkage meeting prespecified response criteria. It does not by itself describe how long a response lasts, whether disease remains controlled, or how symptoms change. Those questions require other outcomes.

The registry lists a 43-month timeframe and specifies the Full Analysis Set as patients who received at least one dose of ceritinib. Although it names the endpoint and its 95% confidence interval, the response percentage and interval are not reported in the registry record. Consequently, the primary endpoint’s observed magnitude and statistical precision cannot be assessed from this record alone.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry identifies the primary outcome as a percentage of participants and specifies a 95% confidence interval. This is a binary-outcome summary: a participant either meets the response definition or does not. The response percentage describes the observed frequency in the analysis population; the confidence interval describes uncertainty around the estimate. A narrower interval generally indicates greater precision, while a wider interval indicates that a broader range of underlying response frequencies remains compatible with the data. The observed estimate and interval are not reported in the registry record.

What does investigator assessment mean?

For the primary endpoint, investigators evaluated response using RECIST 1.1. Such criteria provide rules for classifying changes in tumor lesions, rather than relying on an informal impression of improvement. The registry describes complete and partial response requirements and specifies that the response had to be confirmed. In an open-label study, the investigators were not masked to treatment assignment. The registry does not report additional safeguards or review procedures for the primary endpoint, so the influence of assessment processes cannot be evaluated from this record.

How should the five arms be compared?

The parallel design means the study included separate groups, but non-randomized allocation means participants were not randomly assigned among those groups. The arm descriptions indicate different histories of brain radiation and ALK-inhibitor exposure, along with a leptomeningeal disease group. These features may also be associated with differences in prognosis or disease presentation. As a general statistical principle, comparisons between non-randomized groups can reflect both treatment-related factors and differences in who entered each group. The registry itself cautions that these arms are not a like-for-like control comparison. No adjusted comparative analysis is reported in the registry record.

What do the secondary outcome assessments add?

The registry lists overall intracranial response and intracranial disease control outcomes, including assessments by investigators and by a blinded independent review committee. Independent blinded review can provide an assessment less directly influenced by knowledge of treatment, as a general methodological consideration. It is a distinct assessment source from investigator review; it does not transform the non-randomized treatment allocation into a randomized comparison. The registry also lists time to intracranial tumor response, which concerns the timing of response rather than only whether response occurred. Numerical results and analysis details for these outcomes are not reported in the registry record.

What can a confidence interval tell the reader?

A confidence interval expresses statistical uncertainty in an estimate under the assumptions of the method used to construct it. It is not a guarantee that the true value lies within the interval, nor does it measure all sources of uncertainty. For this study, the registry names a 95% confidence interval for the primary percentage but does not report its bounds or the method used to calculate it. The precision of the response estimate therefore cannot be described quantitatively.

What are the planning illustrations?

Planning illustrations are hypothetical calculations based on an observed effect, not observed outcomes or evidence about the original study’s design. The described assumptions specify a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation, with the effect to detect expressed as a fraction of the observed effect. For time-to-event calculations, the stated quantity is a total number of events; for other calculations, it is participants per arm. These assumptions do not establish that the trial was planned using those calculations. The registry record does not report a sample-size justification, and any planning illustration should not be interpreted as observed power.

Over-dispersion is a general issue for count outcomes when variability exceeds what a simple probability model expects. It can make standard errors too small if the extra variability is ignored. The primary endpoint is a response percentage, not a count-rate endpoint, and the registry does not report an over-dispersion model for it. More broadly, the statistical analysis plan, model choices, covariate adjustment, interim analyses, and multiplicity procedures are not reported in the registry record. No assumptions about those methods should be made.

10. Exact Estimates for Each Arm

No two-arm comparison is shown for this trial, but each arm's result can still be summarized with exact interval estimates computed from the posted counts. The Clopper-Pearson interval is exact and conservative; the Wilson interval is a common closer-to-nominal alternative; the Bayesian credible interval uses a uniform Beta(1, 1) prior. For small groups the three can differ noticeably.

R code and output

arm    <- c("Arm 1 (PrALKi=Y, PrBRad=Y)", "Arm 2 (PrALKi=Y, PrBRad=N)", "Arm 3 (PrALKi=N, PrBRad=Y)", "Arm 4 (PrALKi=N, PrBRad=N)", "Arm 5 (LepDis)")
n      <- c(42, 40, 12, 44, 18)
events <- c(15, 12, 6, 26, 3)
for (i in seq_along(n)) {
  ci <- binom.test(events[i], n[i])$conf.int          # Clopper-Pearson exact interval
  cr <- qbeta(c(0.025, 0.975), 1 + events[i], 1 + n[i] - events[i])   # Bayesian, uniform Beta(1,1) prior
  wi <- (function(x, n) { p <- x / n; z <- qnorm(0.975); d <- 1 + z^2 / n
    c((p + z^2 / (2 * n) - z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d, (p + z^2 / (2 * n) + z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d) })(events[i], n[i])
  cat(sprintf("ARM|%s|%d|%d|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f\n", arm[i], n[i], events[i], 100 * events[i] / n[i],
              100 * ci[1], 100 * ci[2], 100 * wi[1], 100 * wi[2], 100 * cr[1], 100 * cr[2]))
}

ARM  Arm 1 (PrALKi=Y, PrBRad=Y)  42  15  35.7  21.6  52.0  23.0  50.8  23.0  50.9
ARM  Arm 2 (PrALKi=Y, PrBRad=N)  40  12  30.0  16.6  46.5  18.1  45.4  18.1  45.5
ARM  Arm 3 (PrALKi=N, PrBRad=Y)  12  6  50.0  21.1  78.9  25.4  74.6  25.1  74.9
ARM  Arm 4 (PrALKi=N, PrBRad=N)  44  26  59.1  43.2  73.7  44.4  72.3  44.3  72.3
ARM  Arm 5 (LepDis)  18  3  16.7  3.6  41.4  5.8  39.2  6.1  39.6

11. Multiplicity, Missing Data and Analysis Populations

The registry lists a primary endpoint and numerous secondary outcomes, including disease control, intracranial response, intracranial disease control at specified assessments and overall, and time to intracranial response. Multiple endpoints and assessment sources can raise a multiplicity issue: the more statistical questions examined, the greater the chance of a nominally unusual finding occurring by chance if each is tested separately. A prespecified strategy can address that issue, but the registry record does not report a multiplicity adjustment or testing hierarchy. It also reports no analyses in the analysis-count field. That is not evidence that no analyses were conducted; it means the record does not document them there.

The primary analysis population is defined as the Full Analysis Set, comprising patients who received at least one dose of ceritinib. The registry does not report the number in that population, arm-specific analysis denominators for the primary endpoint, or the handling of participants without an evaluable response assessment. Missing-data rules, censoring rules, and sensitivity analyses are not reported in the registry record.

Missing assessments can affect a percentage endpoint if participants without a recorded response are handled differently across analyses. As a general principle, readers need the analysis population, denominators, and missing-data approach to understand how the reported estimate was constructed. Those details are not available for the primary endpoint in this record. The stated population definition should therefore not be taken to mean that every enrolled participant was included in the primary analysis.

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: pneumonia (11.1%), staphylococcal infection (8.3%), blood creatinine increased (8.3%), hyperglycaemia (8.3%), hyponatraemia (8.3%).

The registry reports serious adverse-event counts by arm, along with the number at risk, counts for other adverse events, and deaths. For all participants, 75 were reported as affected by serious adverse events among 156 at risk; 154 were reported as affected by other adverse events among 156 at risk; and 31 deaths were recorded. These are descriptive registry counts, not estimates of treatment effect. The record does not report whether events or deaths were considered related to ceritinib, so they should not be interpreted as treatment-caused.

The registry lists pneumonia at 11.1% as a serious adverse-event term and lists Staphylococcal infection, blood creatinine increased, hyperglycaemia, hyponatraemia, and acute kidney injury at 8.3% each. These percentages describe the terms as reported in the registry; the record does not provide further detail about event timing, severity grading, or causal attribution.

Arm-level counts should be read in the context of the non-randomized design and different eligibility histories. Counts alone do not show how long participants were observed or whether exposure time differed. Those details are not reported in the registry record. The safety summaries are therefore useful as descriptions of events recorded, but they do not establish comparative risk between arms.

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

The record documents a completed Phase II study of oral ceritinib in patients with ALK-positive NSCLC involving the brain and/or leptomeninges. It describes a whole-body, investigator-assessed response endpoint and gives a definition based on confirmed complete or partial response under RECIST 1.1. It also records several secondary intracranial outcomes and adverse-event summaries.

What the record does not establish is the observed primary response percentage. The estimate and its 95% confidence interval are not reported in the registry record, so neither the magnitude of the response nor its statistical precision can be assessed from the information summarized. The record also does not provide numerical secondary efficacy results in this information. A study’s endpoint definitions and completed status cannot substitute for the outcome estimates.

The design does not support treating the five arms as randomized treatment comparisons. Participants entered arms defined by clinical histories and disease presentation, and all arms received ceritinib. Any arm differences could reflect the characteristics that defined eligibility, among other factors. The registry does not report adjusted analyses that would help address such differences.

The adverse-event and death counts describe what was recorded, but do not establish causation. The registry does not report attribution, exposure duration, or enough detail to make comparative safety estimates from the counts alone. Likewise, the record does not describe missing-data handling, the statistical analysis plan, or the approach to multiple endpoints.

In statistical terms, this record identifies what was intended to be measured and provides some descriptive safety information, but it leaves the central efficacy estimate and several analysis details unreported. It therefore cannot, on its own, support a conclusion about the size or precision of the primary response estimate or about causal treatment effects.

14. Why This Trial Matters Statistically

This study raises statistical questions that are especially relevant to research in a population defined by disease location and prior treatment history. The five-arm structure separated participants according to brain radiation, prior ALK-inhibitor exposure, and leptomeningeal disease. Those distinctions help describe the populations studied, but they also mean arm outcomes are not straightforward substitutes for a randomized comparison.

The endpoint definition illustrates how a clinical response becomes a statistical outcome. RECIST 1.1 supplies criteria for classifying complete or partial response, and the overall response rate turns those classifications into a percentage. Reporting the estimate with its confidence interval would allow readers to assess both the observed response frequency and the uncertainty around it. The registry names that interval but does not report its value, leaving an important part of the statistical interpretation unavailable.

The study also lists investigator and blinded independent review committee assessments for some intracranial outcomes. Different assessment sources can be informative, but they should be understood as distinct ways of classifying outcome, not as independent treatment groups. The record’s range of primary and secondary outcomes also makes it important to know how multiplicity was handled; that method is not reported.

Finally, the safety record shows why counts need context. Arm-specific numbers at risk and affected participants are useful descriptions, but interpretation also depends on attribution, observation time, event definitions, and missingness. The registry record does not report several of these details. For biostatisticians and clinical researchers, the case emphasizes the difference between a clearly specified endpoint and a fully interpretable result: the former is documented, while key estimates and analytic details are not.

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 most important limitation for interpreting efficacy is that the registry record does not report the primary overall response rate or its 95% confidence interval. Although the endpoint, assessment criteria, population definition, and timeframe are stated, the result and its precision cannot be evaluated from this record.

The non-randomized, five-arm design limits direct comparisons among arms. The groups were defined by differences in prior ALK-inhibitor exposure, prior brain radiation, and leptomeningeal disease status, so they were not like-for-like controls. The registry record does not report adjusted comparative analyses. It also does not report a separate control treatment.

The study was open-label, and the primary outcome was investigator-assessed. The record does not report whether the primary assessment had independent review, how disagreements were resolved, or other procedures that might help readers evaluate assessment consistency. Although blinded independent review committee assessment is listed for some secondary intracranial outcomes, numerical results for those outcomes are not reported in this record.

Important analysis details are also absent: the statistical analysis plan, sample-size justification, missing-data handling, multiplicity procedures, and interim analyses are not reported in the registry record. The Full Analysis Set is defined, but the primary endpoint denominator and the number included in that set are not stated. Safety counts are available, but event attribution and observation duration are not reported. These gaps constrain interpretation of both efficacy and safety summaries.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: A Phase II Study to Evaluate the Efficacy and Safety of Oral Ceritinib in Patients With ALK-positive NSCLC Metastatic to the Brain and/or to Leptomeninges — NCT02336451.
• Registry record last updated 2020-04-21; status last verified 2020-04.
• Results first posted 2020-02-21.
• 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

NCT02336451 was a completed, non-randomized, open-label Phase II study of oral ceritinib in ALK-positive NSCLC metastatic to the brain and/or leptomeninges. The registry lists a parallel five-arm design and enrollment of 156. The primary endpoint was investigator-assessed whole-body overall response rate under RECIST 1.1 in the Full Analysis Set, with a stated timeframe of 43 months.

The registry record defines the primary endpoint and lists secondary intracranial outcomes, but does not report the primary response estimate or its 95% confidence interval. It reports serious and other adverse-event counts and deaths by arm, without reported causal attribution. The record also does not report key analysis details such as missing-data handling or multiplicity procedures. Because allocation was non-randomized and arm eligibility differed, the arms are not like-for-like controls.