Study Of Adjuvant Lapatinib In High-Risk Head And Neck Cancer Subjects After Surgery: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Study Of Adjuvant Lapatinib In High-Risk Head And Neck Cancer Subjects After Surgery (NCT00424255), a randomized phase 3 study. The primary outcome is Disease Free Survival (DFS).

2. Clinical Question

3. Trial Design

This was a randomized, double-blind, placebo-controlled, multicentre, parallel-group Phase III trial. It enrolled 688 participants with high-risk head and neck cancer after surgery. The registry lists the study as completed and gives its start as 2006-12 and primary completion as 2013-03. The sponsor was GlaxoSmithKline.

Participants were assigned to lapatinib or placebo, with both groups receiving concurrent postoperative chemoradiotherapy. The lapatinib group was assigned lapatinib 1500 mg once daily during chemoradiotherapy for 6 to 7 weeks, followed by maintenance lapatinib once daily for one year. The placebo group received placebo during chemoradiotherapy and then maintenance placebo for one year. The arm descriptions specify chemoradiotherapy as a total dose of 66Gy over 6-7 weeks with cisplatin 100mg/m2 on days 1,2 and 43 of radiotherapy.

The primary outcome was disease-free survival (DFS), analyzed in the intent-to-treat population. This population included everyone randomized, whether or not study medication was received. The registry describes the primary time frame as randomization until the earliest date of recurrence or death, with an average of 101 study weeks. The trial also listed secondary outcomes covering survival, recurrence, second primary tumors, exposure, and adverse events.

The registry record does not report the original sample-size justification, the trial's planned significance level or power, the randomization ratio, or the detailed allocation procedure. The analysis-plan details beyond the reported methods 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.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Intent-to-Treat (ITT) Population: all participants who were randomized to study treatment, irrespective of whether they actually received study medication

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.

6. Primary Endpoint

The registry lists the primary outcome as Disease Free Survival (DFS), measured over From randomization until the earliest date of disease recurrence or death due to any cause (average of 101 study weeks).

Disease-free survival measures the time from randomization until the first qualifying event: disease recurrence or death due to any cause. Recurrence could include local, regional, or distant progression, or a second primary tumor. The recurrence assessment was based on radiological and clinical assessments by a blinded, independent reviewer.

Time-to-event outcomes retain information about when an event occurs, rather than treating every participant as having the same observation period. Participants who had not experienced recurrence or death could be censored at their last qualifying assessment. The record also describes censoring in specified circumstances: initiation of alternative anti-cancer therapy before recurrence or death, missing two or more consecutive disease assessments, and malignant disease at baseline. The exact rules differ by circumstance, as described in the endpoint definition.

DFS is used as a composite measure of recurrence or death. It therefore captures more than one kind of event, and a difference in DFS would not by itself identify which component drove the result. In this trial the reported summaries include median DFS and a hazard ratio. These describe different aspects of the time-to-event data: the median is a point on the survival-time distribution, while the hazard ratio compares event rates over follow-up under the analysis model. Neither summary alone describes the full pattern of event timing.

7. Results Posted in the Registry

• Placebo vs Lapatinib 1500 mg: hazard ratio 1.10 (95% CI 0.85–1.43) (as reported in the registry).

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.

Number of Participants With a Second Primary Tumor

• Endpoint type: Binary: the number of participants with the event, compared as proportions.
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Lapatinib 1500 mg: 2.6% (9/346) versus 1.5% (5/342) on control; risk difference 1.1 percentage points (95% CI -1.0 to 3.2), risk ratio 1.78 (0.60–5.25); unadjusted two-proportion z-test p = 0.290.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Lapatinib 1500 mg and the control. The interval gives the range of effects that remain compatible with the data.

The registry lists secondary outcomes spanning overall survival, disease-specific survival, recurrence timing, distant relapse, second primary tumors, treatment exposure, and adverse events. It records 20 secondary outcomes, but the detailed information available in the record is not equally complete for every outcome. In particular, no detailed numerical results for most of these secondary outcomes are reported in the registry record.

The posted second-primary-tumor counts and the associated unadjusted comparison provide one additional outcome to consider, but they do not establish a separate treatment effect. They should be interpreted alongside, not in place of, the primary DFS result. DFS includes recurrence or death and therefore overlaps conceptually with some secondary disease outcomes, while a second primary tumor is also included in the DFS definition as a recurrence event. The secondary outcome is not an independent confirmation of the primary finding.

Secondary outcomes are often examined in larger numbers and may involve different event definitions, follow-up requirements, or analysis populations. When multiple outcomes are tested, the chance of at least one apparently notable result can increase. Unless a plan for multiplicity is specified and accounted for, unadjusted secondary findings carry less evidential weight than a prespecified primary analysis. The registry record does not report a multiplicity adjustment or a detailed testing hierarchy.

It is therefore not appropriate to infer a coherent overall pattern of benefit or harm from isolated secondary entries. The available secondary results do not resolve the uncertainty in the primary hazard-ratio estimate. More complete reporting of the secondary estimates, their confidence intervals, analysis populations, and planned interpretation would be needed to assess how they relate to DFS.

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

• Overall Survival (OS)
• Disease Specific Survival (DSS)
• Time to Locoregional Recurrence (TTLR)
• Time to Distant Relapse (TTDR)
• Number of Participants With a Second Primary Tumor
• Extent of Exposure
• Number of Participants With Any Adverse Event (AE) or Serious Adverse Event (SAE)
• Number of Participants With the Indicated Chemistry Toxicities by Maximum Toxicity Grade (G3 and G4) at the Worst-case On-therapy Visit

9. Statistical Interpretation

10. Statistical Methods Explained

The primary outcome is a time-to-event measure. The registry reports a median for each arm where available, a confidence interval for the comparison's hazard ratio, and non-stratified log-rank tests. These methods address different features of event-time data and should be read together rather than treated as interchangeable results.

What does the median describe?

In general, the median event time is the time by which half of the estimated population has experienced the event, under the survival-analysis approach used. It is a summary of the event-time distribution, not an average follow-up time. It can be useful because it expresses time in familiar units, but it does not describe the full curve or the timing of events after or before the median. The registry record gives a median DFS of 53.6 months in the lapatinib group. The placebo median is not reported in the registry record, and neither arm's complete median confidence interval is available in the registry entries.

What does the hazard ratio describe?

In general, a hazard is the rate at which an event occurs among people who have not yet had the event, at a particular point in follow-up. A hazard ratio compares those rates between groups. The registry reports a hazard ratio of 1.10 for placebo versus lapatinib, with a 95% confidence interval of 0.85 to 1.43. The ratio is a relative measure; it is not an absolute difference in DFS months and does not state how many participants experienced recurrence or death.

A hazard ratio is often interpreted under a proportional-hazards framework, in which the relative hazards are treated as stable over time. Whether that assumption was assessed, and whether it was considered reasonable in this trial, is not reported in the registry record. The reported hazard ratio should therefore be understood as the registry's summary estimate, not as evidence that the relative difference was identical throughout follow-up.

Why use the log-rank test?

In general, the log-rank test compares event-time experience between groups while accounting for censoring. A non-stratified log-rank test makes the comparison without stratifying by specified factors. The registry lists two non-stratified log-rank analyses and gives a p-value for each. Their p-values do not constitute measures of effect size: they do not state how large a difference is, and they do not show the range of effects compatible with the data. The reason for the two tests and the distinction between them are not reported in the registry record.

A p-value is a measure of how unusual data at least as inconsistent with a specified null model would be if that model were true, under the test's assumptions. In general, it is not the probability that the null hypothesis is true, nor is it a measure of clinical importance. Here, the hazard ratio and confidence interval are more informative about the estimated direction and precision of the comparison than a p-value alone.

How do censoring and the analysis population matter?

Censoring means that a participant's event time is not observed past a particular point, even though the participant has not necessarily had the event. The endpoint definition specifies censoring at the last assessment in several circumstances. In general, survival methods use the observed time up to censoring, but their interpretation depends on the censoring process and the analysis rules. The registry describes the rules but does not report the number of participants censored for each reason or an evaluation of censoring assumptions.

The primary analysis population was intent-to-treat. This retains participants in their randomized groups irrespective of whether they took study medication. In general, an intent-to-treat analysis preserves the comparison created by random assignment and estimates the effect of assignment to the treatment strategy. The registry does not give the number who received treatment, discontinued treatment, or crossed over, so those aspects cannot be assessed from the record.

11. Confidence Intervals and Precision

The 95% confidence interval for the reported hazard ratio runs from 0.85 to 1.43 around the point estimate of 1.10. In general, a confidence interval describes a range of estimates compatible with the data and the method used, while reflecting sampling uncertainty. It is not a guarantee that the true value lies inside this particular interval, nor does it quantify every source of uncertainty in the study.

Because the interval includes 1, the posted estimate is compatible with no relative difference in hazards. Its endpoints also allow estimates on either side of 1, so the data do not pin down the direction with high precision. The range is not narrow enough to establish a single, precise effect size. The registry's log-rank analyses provide p-values, but these do not replace the interval's information about the range of effect estimates.

The arm-specific median results are less complete. The lapatinib median is 53.6 months, and its reported lower confidence limit is 45.8 months; an upper confidence limit is not reported in the registry record. For placebo, the record gives a lower confidence limit of 54.6 months but does not report the median or upper confidence limit. Without complete median estimates, the arm medians cannot be compared fully from the registry entries.

Confidence intervals capture statistical uncertainty under the analysis method. They do not address possible limitations such as missing outcome assessments, departures from model assumptions, or incomplete reporting. The registry record does not include enough information to assess all of those issues. The appropriate reading is therefore that the hazard ratio is an imprecise estimate, while the incomplete median information further limits interpretation of absolute event timing.

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 sample-size display is a planning illustration based on the observed effect, not the trial's original design assumptions. The trial's own sample-size justification and original design assumptions are not reported in the registry record. For the time-to-event comparison, the figures are total numbers of events, not participants per arm. For the second-primary-tumor comparison, the figures are participants per arm.

The illustrations assume a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. They consider effects equal to 100%, 75%, and 50% of the observed effect. The time-to-event planning uses the Schoenfeld method. In the 80% power column, the event or participant count is the planning amount associated with an 80% chance of detecting the specified effect under those assumptions, if that effect and model apply. The 90% power column uses a higher target chance of detecting that effect and therefore has larger planning counts.

The pattern across the rows reflects a general design principle: smaller effects are harder to distinguish from chance variation, so detecting them with the same target power requires more information. For a time-to-event endpoint, information is represented by events; for the binary endpoint, it is represented by participants in each arm. These counts are hypothetical planning requirements for the stated effect fractions and assumptions, not observed trial outcomes.

This is not observed power. The figures do not show the probability that the completed trial detected its true effect, and they are not a retrospective assessment of the trial's success. They are planning illustrations generated from the observed effect and the stated assumptions.

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

d is the total number of events needed (Schoenfeld's formula) with 1:1 allocation and proportional hazards.

Worked calculation for this trial

1. Hazard ratio HR = 1.100, so ln(HR) = 0.0953.
2. d = 4 × (1.96 + 0.842)² / [ln(1.100)]² = 3,456.1, rounded up to 3,457 events in total 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
hr <- 1.1                              # observed hazard ratio
events_needed <- function(power, frac = 1) ceiling(round(4 * (z_a + qnorm(power))^2 / log(hr^frac)^2, 6))   # Schoenfeld, 1:1 allocation
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: events = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(events_needed(0.80, f), big.mark = ","), format(events_needed(0.90, f), big.mark = ",")))

# Simulation: given d total events under proportional hazards, events in the treatment arm are Binomial(d, hr / (1 + hr)).
# The score (log-rank type) statistic is (D1 - d / 2) / sqrt(d / 4).
set.seed(424255)
d    <- events_needed(0.80)
reps <- 20000
D1 <- rbinom(reps, d, hr / (1 + hr))
z  <- (D1 - d / 2) / sqrt(d / 4)
cat(sprintf("Simulated power with %s events: %.3f (%d replicates)\n", format(d, big.mark = ","), mean(abs(z) > z_a), reps))

Output from running the code

100% of observed effect: events = 3,457 (80% power), 4,627 (90% power)
 75% of observed effect: events = 6,145 (80% power), 8,226 (90% power)
 50% of observed effect: events = 13,825 (80% power), 18,508 (90% power)
Simulated power with 3,457 events: 0.797 (20000 replicates)

The simulation uses the fact that, with d events in total and proportional hazards, the number of events in the treatment arm is binomial. It is an approximation that conditions on the total number of events, and it should land close to the target power.

14. Analyses Reported by the Sponsor

The primary comparison is presented using the sponsor's reported hazard ratio and confidence interval: 1.10, with a 95% confidence interval from 0.85 to 1.43, for placebo versus lapatinib 1500 mg. The sponsor's reported result is the primary basis for interpreting the time-to-event comparison. The registry does not mark a discrepancy between the reported result and the comparison data.

The record also lists two p-values from non-stratified log-rank tests. It does not explain why two tests are reported or specify how they relate to the hazard-ratio analysis. These values should therefore be read as separate posted test results, not as replacements for the hazard ratio or its confidence interval.

For the secondary binary outcome of second primary tumors, the comparison includes simple recalculations from the posted group summaries. Those are unadjusted calculations and are not a substitute for a sponsor analysis. No sponsor-reported analysis for that secondary outcome is listed. The distinction matters because recalculations based on summary counts may not incorporate the trial's full analysis procedures or any adjustments that were part of a separate analysis. The registry record does not provide such adjusted results for this outcome.

15. Multiplicity, Missing Data and Analysis Populations

Multiplicity arises when several outcomes, comparisons, or analyses are examined. In general, each additional test creates another opportunity for a result to appear notable by chance. A prespecified testing hierarchy or an adjustment for multiple testing can help control that issue, but the appropriate approach depends on the trial's analysis plan.

The registry lists a primary DFS outcome, 20 secondary outcomes, three analyses, and two p-values from non-stratified log-rank tests. It does not explain the relationship between the two log-rank tests, identify a testing hierarchy, or report a multiplicity adjustment. The original statistical analysis plan and the protocol's multiplicity strategy are not reported in the registry record. Consequently, the two log-rank p-values and any secondary findings should not be treated as if they were a single, fully specified confirmatory testing sequence.

The primary analysis population was intent-to-treat, including all randomized participants irrespective of whether they received study medication. The registry's censoring rules describe how certain circumstances were handled for DFS, including alternative anti-cancer therapy before recurrence or death, missed disease assessments, and malignant disease at baseline. The numbers affected by these rules, the amount of missing outcome information, and sensitivity analyses for missingness are not reported in the registry record.

In general, censoring methods rely on assumptions about the relationship between the reason observation ends and the subsequent event experience. The endpoint description alone cannot establish whether those assumptions held. The registry also does not report details of missing-data handling for the secondary outcomes or a complete account of analysis deviations. These gaps limit assessment of the robustness of the posted findings, but do not change the definition of the reported intent-to-treat population.

16. Safety Results

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

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: lymphopenia (5.2%), mucosal inflammation (4.9%), vomiting (3.4%), hyponatraemia (3.4%), dysphagia (3.2%).

The adverse-event summaries give the number of participants at risk and the number with at least one serious adverse event or any adverse event. In the placebo group, 133 of 336 participants at risk had a serious adverse event, and 321 of 336 had any adverse event. In the lapatinib 1500 mg group, 169 of 349 participants at risk had a serious adverse event, and 337 of 349 had any adverse event. These are participant counts, not counts of separate event episodes.

The registry also lists the most frequent serious adverse-event terms with percentages: lymphopenia, 5.2%; mucosal inflammation, 4.9%; vomiting, 3.4%; hyponatraemia, 3.4%; dysphagia, 3.2%; and neutropenia, 3.0%. The listed terms are not a complete description of all adverse events, and the registry record does not give arm-specific values for each listed term in the registry record.

These summaries describe observed adverse events in the posted record; they do not establish that a treatment caused an event or provide a full comparative safety analysis. The registry does not report deaths in the adverse-event summary, and the methods for attributing events are not reported in the registry record.

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

The central statistical finding is a reported hazard ratio of 1.10 for placebo versus lapatinib 1500 mg, with a 95% confidence interval from 0.85 to 1.43. This is an estimate of the relative event hazard under the analysis, not an absolute difference in months and not a direct count of recurrences or deaths. The interval includes 1 and spans values in either direction, leaving uncertainty about the size and direction of the comparison.

The reported median DFS for the lapatinib group is 53.6 months, but its upper confidence limit is not reported. The placebo median is not reported, although its lower confidence limit is 54.6 months. The medians therefore cannot be fully compared from the registry record. The two listed log-rank p-values are additional test results, but the record does not explain why both were conducted or how they map to separate analysis decisions.

The endpoint combines recurrence and death, so it does not isolate the effect on any one event type. The registry's censoring rules also mean that observed follow-up and assessment availability contribute to the analysis. The number of censored participants and the distribution of event types are not reported in the registry record.

These results do not establish that lapatinib improves or worsens outcomes. They also do not establish equivalence between the strategies: a confidence interval that includes no difference is not proof that the true effect is exactly zero. The statistical reading is that the posted estimate is uncertain, with incomplete reporting of the arm-specific medians and some analysis details. Clinical decisions require considerations beyond these registry statistics and are not addressed by this interpretation.

18. Why This Trial Matters Statistically

This trial is statistically informative because it combines a randomized comparison with a time-to-event primary outcome in a postoperative, high-risk cancer population. Randomization and double blinding are important design features for reducing systematic differences between assigned groups and limiting bias in treatment knowledge. The registry identifies the intent-to-treat population and specifies how disease recurrence and censoring were defined, giving readers a basis for understanding the primary analysis.

At the same time, the record illustrates why effect estimates and their precision matter more than a simple focus on whether a test crosses a threshold. The hazard ratio is close to 1, but its confidence interval spans values below and above 1. The median information is incomplete, and the two log-rank results are not explained in relation to one another. Those details make clear that a single headline statistic cannot convey all of the uncertainty or reporting limitations.

The trial also illustrates a distinction between a study's original design and retrospective planning illustrations. The registry does not report the original sample-size justification, while the posted planning calculations use fractions of the observed effect. Those calculations describe hypothetical event or participant requirements under stated assumptions; they do not establish the trial's original power or its probability of detecting an effect.

Finally, the breadth of secondary outcomes makes transparent reporting of multiplicity and analysis plans important. Without a stated hierarchy or adjustment strategy, secondary findings should be treated cautiously. These are general statistical lessons illustrated by the information recorded for this trial, rather than conclusions about the clinical value of either strategy.

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 limitations are specific to the registry record. The placebo median DFS and complete confidence intervals for both arm-specific medians are not reported. The hazard-ratio interval is available, but the registry does not provide the number of DFS events, a detailed survival curve description, or the event counts by type. Those omissions make it difficult to understand how the composite outcome was distributed.

The registry lists two non-stratified log-rank p-values but does not explain their relationship or the reason for conducting both tests. It does not report the original statistical analysis plan, the trial's sample-size justification, whether the proportional-hazards assumption was assessed, or whether sensitivity analyses were performed. The reported hazard ratio therefore cannot be evaluated against those analysis details using the registry record alone.

For the secondary outcomes, most detailed numerical results are not reported in the registry record. The available second-primary-tumor comparison is an unadjusted recalculation from posted summaries, and a sponsor-reported analysis is not listed. The registry does not report a multiplicity strategy, so isolated secondary results cannot be given the same weight as a clearly defined primary analysis.

The adverse-event information includes aggregate participant counts and a list of serious-event terms, but does not provide a full account of event timing, duration, attribution, or treatment discontinuation. The record also does not report demographic and baseline clinical details in the registry record, participant-level follow-up, numbers censored for each reason, or the amount of missing data. These gaps limit independent assessment of generalizability, robustness, and the full safety profile.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Study Of Adjuvant Lapatinib In High-Risk Head And Neck Cancer Subjects After Surgery — NCT00424255.
• Registry record last updated 2014-07-18; status last verified 2014-06.
• Results first posted 2014-02-11.
• 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

NCT00424255 is a completed, randomized, double-blind, placebo-controlled Phase III trial sponsored by GlaxoSmithKline. It enrolled 688 participants with high-risk resected squamous cell carcinoma of the head and neck. Both groups received postoperative chemoradiotherapy; the assigned study treatment was lapatinib or placebo, followed by maintenance treatment.

The primary outcome was DFS in the intent-to-treat population. The registry reports a hazard ratio of 1.10 for placebo versus lapatinib 1500 mg, with a 95% confidence interval from 0.85 to 1.43. The arm-specific median DFS reporting is incomplete, and the reason for two listed log-rank tests is not reported. Multiple secondary outcomes are listed, but detailed results are not available for most of them in the registry record.

The posted estimates should be read with their uncertainty and reporting limitations. The registry record does not establish a definitive direction or magnitude for the primary comparison.