Gemcitabine Hydrochloride or Pemetrexed Disodium and Carboplatin With or Without Celecoxib in Treating Patients With Advanced Non-Small Cell Lung Cancer: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Gemcitabine Hydrochloride or Pemetrexed Disodium and Carboplatin With or Without Celecoxib in Treating Patients With Advanced Non-Small Cell Lung Cancer (NCT01041781), a randomized phase 3 study. The primary outcome is Progression-free Survival.

2. Clinical Question

The clinical question is therefore focused on whether adding celecoxib changes progression-free survival in this treatment setting, not on comparing gemcitabine directly with pemetrexed. The registry describes both chemotherapy options in each arm, with carboplatin in both. The posted results do not provide separate primary comparisons for the two chemotherapy regimens or report whether effects differed by regimen.

3. Trial Design

The registry describes a randomized, double-blind, parallel-group phase III trial. Its official title identifies the study as an evaluation of selective COX-2 inhibition in COX-2-expressing advanced non-small cell lung cancer. The registry record reports enrollment of 313 and a status of terminated. The recorded start is 2010-02, and primary completion is dated 2017-01-17. The reason for termination and the timing or circumstances of that decision are not reported in the registry record.

Arm I combined celecoxib with standard chemotherapy, and Arm II combined placebo with standard chemotherapy. In both arms, chemotherapy included carboplatin and either gemcitabine hydrochloride or pemetrexed disodium. The descriptions specify gemcitabine on days 1 and 8 or pemetrexed on day 1, with carboplatin on day 1. Celecoxib or placebo was taken orally twice daily on days 1-21. The record does not report the criteria used to select between the chemotherapy options, nor does it describe treatment beyond these regimen details.

The posted primary outcome data include 154 participants in the celecoxib arm and 158 in the placebo arm. These endpoint analysis counts total 312, while the registry reports enrollment of 313; the relationship between enrollment and the primary endpoint analysis population is not explained in the registry record. The record does not report allocation concealment, the randomization sequence method, the planned sample-size justification, or the intended analysis population. It also does not report whether any interim analyses were planned or conducted.

Secondary outcomes include overall survival, response rate, toxicities, and analyses of urinary prostaglandin metabolite levels in relation to progression-free survival. The posted results include response rate and toxicity incidence alongside the primary endpoint. The record lists additional secondary outcomes, but does not provide posted results for all of them 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

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 Progression-free Survival, measured over Time between randomization and disease relapse or death from any cause, assessed up to 5 years.

Progression-free survival measures the time from randomization until the first of disease progression or death from any cause. The registry defines progression according to RECIST version 1.1. Its stated timeframe is up to 5 years. This endpoint combines disease worsening and death into a time-to-event measure; an event is counted when either specified event occurs first. Participants without a recorded event by the end of their available follow-up contribute information up to that point rather than being treated as though they had an event at that time.

The endpoint is useful for comparing how long participants remain alive without documented disease progression. It can retain information about the timing of events, rather than reducing follow-up to a simple count of whether an event occurred. The registry says the median and 95% confidence intervals are estimated using the Kaplan-Meier estimator. The median is the time by which the estimated event-free proportion has reached the midpoint of the distribution. It describes the centre of the observed time-to-event distribution, not the experience of every participant.

Progression-free survival is not the same as overall survival: the former includes progression as an event, whereas the secondary outcome list separately names overall survival. The posted primary result should therefore be interpreted as a comparison of progression-free survival, not as a direct estimate of survival alone. The registry record does not report further details about outcome assessment schedules, censoring rules, or independent review of progression.

7. Results Posted in the Registry

• Arm I (Arm A: Celecoxib + Standard Chemotherapy) vs Arm II (Arm B: Placebo + Standard Chemotherapy): hazard ratio 1.08 (95% CI 0.85–1.37), reported p = 0.535 (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.

Response Rate

• Endpoint type: Binary: the percentage of participants with the event (counts rebuilt from the posted percentages and group sizes).
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Arm II (Arm B: Placebo + Standard Chemotherapy): 34.8% (55/158) versus 40.3% (62/154) on control; risk difference -5.4 percentage points (95% CI -16.2 to 5.3), risk ratio 0.86 (0.65–1.15); unadjusted two-proportion z-test p = 0.320.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Arm II (Arm B: Placebo + Standard Chemotherapy) and the control. The interval gives the range of effects that remain compatible with the data.

Incidence of Toxicities as Assessed by NCI CTCAE v. 4.0

• Endpoint type: Binary: the percentage of participants with the event (counts rebuilt from the posted percentages and group sizes).
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Arm II (Arm B: Placebo + Standard Chemotherapy): 55.1% (87/158) versus 61.0% (94/154) on control; risk difference -6.0 percentage points (95% CI -16.9 to 5.0), risk ratio 0.90 (0.75–1.09); unadjusted two-proportion z-test p = 0.285.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Arm II (Arm B: Placebo + Standard Chemotherapy) and the control. The interval gives the range of effects that remain compatible with the data.

The secondary outcomes broaden the focus beyond progression-free survival to overall survival, response, toxicities, and the prognostic value of urinary prostaglandin metabolite levels. The posted results include response rate and toxicity incidence. Other listed secondary outcomes do not have posted results in the registry record, so no conclusion about their observed patterns can be drawn from the registry record.

The response-rate estimates are 40.0% for celecoxib and 35.0% for placebo, with respective 95% confidence intervals of 32.0% to 48.0% and 27.0% to 43.0%. The toxicity-incidence estimates are 61.04% and 55.06%, respectively; confidence intervals are not reported for that outcome. These endpoints do not measure the same event or time scale as progression-free survival, so their estimates should be interpreted on their own terms rather than as confirmation or contradiction of the primary result.

General explanation: when many outcomes or subgroup questions are tested, the chance of at least one apparently notable result arising by chance can increase. Unless a multiplicity strategy is specified, secondary findings—especially unadjusted comparisons—usually carry less inferential weight than a prespecified primary endpoint. The registry record does not report a multiplicity adjustment or a hierarchy for the secondary analyses. The p-values associated with unadjusted secondary comparisons should therefore be regarded as exploratory, not as independent confirmatory evidence.

The record lists overall survival and several urinary prostaglandin metabolite analyses, including comparisons by baseline quartile thresholds, but the relevant posted estimates are not reported in the registry record. Their presence in the outcome list does not establish that an effect was found. The primary progression-free survival result remains the principal posted comparative finding, with secondary results adding context but not changing the uncertainty in the primary estimate.

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

• Overall Survival
• Response Rate
• Incidence of Toxicities as Assessed by NCI CTCAE v. 4.0
• Prognostic Value of Urinary Prostaglandin Metabolites (PGE-M) Levels for Worse PFS for Patients Who Had Baseline Urinary PGE-M Above/Below the First Quartile (Q1)
• Prognostic Value of Urinary Prostaglandin Metabolites (PGE-M) Levels for Worse PFS for Patients Who Had Baseline Urinary PGE-M Above/Below the Median Quartile (Q2)
• Prognostic Value of Urinary Prostaglandin Metabolites (PGE-M) Levels for Worse PFS for Patients Who Had Baseline Urinary PGE-M Above/Below the Third Quartile (Q3)

9. Statistical Interpretation

10. Statistical Methods Explained

The primary endpoint is a time-to-event outcome. General explanation: this type of analysis uses both whether an event occurred and the time at which it occurred. Participants who have not experienced the event by their last observed follow-up are commonly treated as censored, meaning their observed time contributes information up to that point. The exact censoring rules and analysis population used in this trial are not reported in the registry record.

What does the median represent?

The registry states that progression-free survival medians were estimated with the Kaplan-Meier estimator. General explanation: the Kaplan-Meier method estimates the proportion remaining event-free over time while accounting for participants whose follow-up ends before an event is observed. The median is the time at which the estimated event-free proportion reaches the midpoint. It is a summary of the estimated distribution, not an estimate that every participant experiences progression or death at that time. The posted medians are 5.16 months for celecoxib and 5.26 months for placebo.

What does the hazard ratio compare?

The sponsor’s reported primary comparison is a hazard ratio, analysed with a log-rank method. General explanation: a hazard is the instantaneous event rate among participants who have not yet experienced the event, conditional on the time already spent event-free. A hazard ratio compares those rates between arms over follow-up. It is not a ratio of median times and should not be read as a direct percentage change in the median. The posted estimate is 1.076 for celecoxib versus placebo. Its interpretation depends on the direction of the comparison as specified in the registry and on the assumptions underlying the time-to-event analysis.

General explanation: the log-rank test compares the observed distribution of events over time between groups with the distribution expected under a no-difference hypothesis. It uses event timing, rather than only comparing the two medians. The sponsor’s reported p-value is 0.535. The registry record does not report further model details, such as whether covariates were included, or the handling of departures from model assumptions.

How should confidence intervals be read?

A 95% confidence interval expresses statistical uncertainty around an estimate. General explanation: under repeated use of the same method and assumptions, intervals constructed in this way would cover the underlying parameter at the stated rate. It is not a guarantee that a particular interval contains the true value, and it is not a range containing 95% of participant outcomes. The primary hazard-ratio interval is 0.853 to 1.367. It includes 1, the value corresponding to no relative difference in hazard. The interval also allows for values in either direction, so the point estimate alone is not a precise account of the possible effect.

The response-rate and toxicity-incidence comparisons are binary outcomes: each participant is classified according to whether the defined outcome occurred. General explanation: comparing proportions discards event timing, unlike a time-to-event analysis. The posted response-rate estimates are 40.0% and 35.0%, and the posted toxicity-incidence estimates are 61.04% and 55.06%. The registry record does not report sponsor analyses for these secondary outcomes in the registry record. Any unadjusted recalculations associated with secondary comparisons should be distinguished from sponsor-reported analyses and do not account for other comparisons or covariates.

General explanation: over-dispersion occurs when variability is greater than a model expects. Accounting for it can matter when a chosen model otherwise understates uncertainty. The registry record does not report an over-dispersion model or indicate that one was used, and the posted primary analysis is described as a log-rank comparison. No claim about over-dispersion is needed to interpret that reported result.

The registry record does not report the original statistical analysis plan, the planned sample-size justification, interim analyses, multiplicity procedures, missing-data methods, or details of the analysis population. These items affect how completely the analysis can be evaluated, but their absence does not change the figures that are posted. The primary inference should remain anchored to the sponsor-reported hazard ratio, its confidence interval, and its log-rank p-value.

11. Confidence Intervals and Precision

The primary hazard ratio is 1.076, with a 95% confidence interval from 0.853 to 1.367. The interval contains 1, so it is compatible with no relative difference in hazard as well as with effects in either direction. Its span indicates that the estimate is not highly precise: the data do not pin down a narrow range of plausible relative hazards. The p-value, 0.535, is consistent with the uncertainty conveyed by the interval, but the two statistics answer different questions. The interval describes estimation uncertainty; the p-value assesses the data against a specified null hypothesis under the test assumptions.

The arm-specific median estimates also have uncertainty. The celecoxib median is 5.16 months, with an interval from 4.4 to 5.78 months; the placebo median is 5.26 months, with an interval from 4.4 to 6.08 months. Those intervals overlap. Overlap alone is not a formal test of the between-arm difference; the sponsor-reported hazard ratio and log-rank analysis provide the posted primary comparison.

The response-rate intervals are 32.0% to 48.0% for celecoxib and 27.0% to 43.0% for placebo. For toxicity incidence, confidence intervals are not reported in the registry record. Confidence intervals are also not reported for the posted serious adverse event counts. Consequently, the precision of those safety comparisons cannot be assessed from the counts alone.

General explanation: a confidence interval reflects sampling uncertainty under the method used, but it does not capture every source of uncertainty, such as unreported analysis choices or limitations in outcome measurement. Here, the primary interval is informative but broad enough that a modest point estimate should not be treated as a precise estimate of the effect. The registry record does not report additional sensitivity analyses that might show how robust the estimate is to alternative analytic choices.

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 calculations are planning illustrations based on the observed effect, not the trial’s original design. For the primary time-to-event comparison, the figures are total events, not participants per arm. The listed total events for 80% power are 5,852 when planning for 100% of the observed effect, 10,403 for 75% of the observed effect, and 23,405 for 50% of the observed effect. For 90% power, the corresponding figures are 7,834, 13,926, and 31,333 total events.

The assumptions for these illustrations are a two-sided significance level of 0.05, equal allocation between the two arms, a normal approximation, and an effect to detect equal to the stated fraction of the observed effect. The time-to-event calculations use the Schoenfeld approach. For the non-time-to-event illustrations, the reported quantities are participants per arm rather than total events. These assumptions describe the planning illustrations only; the trial’s own original sample-size assumptions and justification are not reported in the registry record.

In plain language, the 80% power column gives the required event or participant count for an 80% chance of detecting the specified effect under the stated assumptions, if that effect is present. The 90% power column represents a higher chance of detection under those same assumptions, and therefore requires a larger count. Smaller effects are harder to distinguish from random variation, so planning for a smaller fraction of the observed effect requires more events or participants. These counts are not observed power; they are forward-looking illustrations and should not be read as a measure of what the completed trial achieved.

The figures are much larger when the planning effect is reduced from 100% to 75% or 50% of the observed effect. That pattern is a general feature of power planning: subtle effects require more information than larger effects. The registry’s enrollment count and the posted endpoint analysis counts are not interchangeable with the event totals in these calculations. The original target, event assumptions, and whether the study was designed around the posted effect are not reported in the registry record.

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 = 1.076, so ln(HR) = 0.0733.
2. d = 4 × (1.96 + 0.842)² / [ln(1.076)]² = 5,851.2, rounded up to 5,852 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 <- 1.076                              # 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(1041781)
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 = 5,852 (80% power), 7,834 (90% power)
 75% of observed effect: events = 10,403 (80% power), 13,926 (90% power)
 50% of observed effect: events = 23,405 (80% power), 31,333 (90% power)
Simulated power with 5,852 events: 0.804 (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 sponsor’s own primary analysis is the log-rank comparison reporting a hazard ratio of 1.076, a 95% confidence interval from 0.853 to 1.367, and p = 0.535. That result is the appropriate primary statistical result to cite. The posted arm-specific Kaplan-Meier medians are 5.16 months for celecoxib and 5.26 months for placebo, but those medians are not substitutes for the sponsor’s time-to-event comparison.

No discrepancy is flagged between the sponsor result and the posted primary comparison. Any simple unadjusted recalculation from summary statistics is not a replacement for the sponsor’s analysis. Such recalculations do not reproduce all details of the original analysis and may not use the same participant-level information or methods. The sponsor’s own analysis takes precedence if a discrepancy is present; no discrepancy is indicated for this result.

The registry record identifies the primary analysis method as log-rank and gives the hazard ratio and its interval, but does not report a fuller statistical analysis plan. It is therefore not possible to establish from the registry record whether the sponsor used additional adjustment, particular censoring conventions, or other prespecified analytic choices. The stated primary result should be reported with its method, estimate, interval, and p-value, without implying that an unadjusted reanalysis validates or supersedes it.

15. Multiplicity, Missing Data and Analysis Populations

Multiplicity arises when several statistical questions are examined. General explanation: testing more outcomes, subgroups, or time points can increase the chance that at least one result appears statistically notable even if none reflects a real effect. Strategies such as prioritizing a primary endpoint, setting a hierarchy, or adjusting significance thresholds can help control that issue. The registry record lists several secondary outcomes, including overall survival, response rate, toxicity, and urinary prostaglandin metabolite analyses. It does not report a multiplicity procedure, a hierarchy among secondary outcomes, or whether particular comparisons were considered exploratory.

The primary endpoint is identified as progression-free survival, and the sponsor analysis is reported for that endpoint. The registry does not report the original statistical analysis plan, so the prespecified status of all analytic decisions cannot be checked from the record. The number of analyses listed in the record is four, but the record does not explain how that count maps to endpoints, subgroups, or analytic procedures. It would be inappropriate to infer a multiplicity correction from that count.

Missing-data and analysis-population details are also limited. The registry reports enrollment of 313 and primary outcome counts of 154 in the celecoxib arm and 158 in the placebo arm. It does not explain the difference between the enrollment total and the primary endpoint analysis counts. The criteria for inclusion in the endpoint analysis, handling of withdrawals, censoring rules, and treatment of missing outcome data are not reported in the registry record. It is also not reported whether the primary analysis followed an intention-to-treat principle or another population definition.

The registry status is terminated, but the reason for termination is not reported. The record does not state whether termination was related to recruitment, safety, efficacy, funding, or another matter, and none should be inferred. Interim monitoring plans and any interim analyses are also not reported. These omissions limit evaluation of the trial’s operational and analytic context; they do not justify altering the interpretation of the posted estimate or assigning a reason to the termination.

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 0% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: anemia (15.1%), fatigue (13.7%), platelet count decreased (11.0%), neutrophil count decreased (6.1%), hyponatremia (5.5%).

The registry reports serious adverse events affecting 31 of 146 participants at risk in the celecoxib arm and 21 of 148 at risk in the placebo arm. It also reports deaths of 15 and 9, respectively, in the adverse-event information. These are counts in the registry record; they do not by themselves establish that the events or deaths were caused by celecoxib, placebo, or chemotherapy. The record does not report a causal attribution for these counts or provide confidence intervals for their comparison.

The most frequently listed serious adverse event terms include anemia at 15.1%, fatigue at 13.7%, platelet count decreased at 11.0%, neutrophil count decreased at 6.1%, hyponatremia at 5.5%, and vomiting at 4.8%. The registry excerpt does not identify the arm-specific denominators for these term percentages, so they should not be used to infer an arm-to-arm difference.

The separate toxicity-incidence outcome is reported as 61.04% in the celecoxib arm and 55.06% in the placebo arm. That outcome is distinct from the serious adverse event counts. The registry record does not provide enough detail to equate all toxicities with serious events or to assess event severity, timing, recurrence, or attribution. These figures describe reported safety data; they do not determine an individual’s risk or establish a causal comparison.

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

The posted result addresses whether progression-free survival differed between celecoxib and placebo when each was combined with carboplatin and either gemcitabine hydrochloride or pemetrexed disodium in advanced non-small cell lung cancer. The primary sponsor analysis reports a hazard ratio of 1.076 with a 95% confidence interval from 0.853 to 1.367 and p = 0.535. The arm-specific median estimates are 5.16 months and 5.26 months. Together, these figures describe a comparison whose point estimate is close to 1, but whose interval remains compatible with relative hazards on either side of 1.

These data do not establish that the two approaches are equivalent. A non-significant p-value does not demonstrate equivalence or rule out a meaningful difference; equivalence requires an appropriate design and a prespecified margin, neither of which is reported in the registry record. The interval also does not prove that a difference exists. It expresses uncertainty around the estimated comparison, and the posted result does not identify a clear direction.

The result is specific to the studied treatment context and endpoint. It does not compare gemcitabine with pemetrexed, because the record describes both chemotherapy options within each randomized arm and does not report separate estimates by chemotherapy regimen. It does not establish a result for overall survival, because that is a separate secondary outcome and its posted result is not reported in the registry record. Nor does the primary endpoint answer every safety question.

The registry lists a COX-2-expressing population in the official title, but the posted primary results do not report a treatment-effect analysis by COX-2 expression level. No conclusion about biomarker subgroups can be drawn from the primary estimate alone. The terminated status also does not explain why the trial ended; the reason is not reported in the registry record. The figures support a careful account of the posted statistical comparison, not a treatment recommendation or a claim about unreported outcomes.

18. Why This Trial Matters Statistically

This trial illustrates why time-to-event results should not be reduced to a comparison of medians alone. The medians, 5.16 months for celecoxib and 5.26 months for placebo, are intuitive summaries, but the sponsor’s log-rank analysis uses the event-time information across follow-up and reports a hazard ratio of 1.076. The confidence interval and p-value are essential to understanding how uncertain the comparison is.

It also shows the distinction between a point estimate and precision. A hazard ratio close to 1 may appear straightforward, but its 95% confidence interval from 0.853 to 1.367 spans values on either side of 1. The interpretation should therefore describe both the estimate and its uncertainty rather than treating the point estimate as definitive. The result does not establish equivalence, and a non-significant test is not proof of no effect.

The registry’s multiple secondary outcomes demonstrate why results need an outcome hierarchy and clear multiplicity reporting. Response, toxicity, overall survival, and biomarker analyses address different questions. Without reported multiplicity procedures and complete secondary results, isolated secondary patterns should not outweigh the primary analysis.

Finally, the record distinguishes the trial’s posted results from planning illustrations based on the observed effect. Those calculations show how required information grows as the target effect becomes smaller, but they do not reveal the trial’s original design assumptions. The original justification is not reported in the registry record. That distinction matters when readers assess whether enrollment and observed event information were adequate for the question the trial intended to answer.

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.

The registry record does not report the reason the trial was terminated. Its status therefore cannot be used to infer that the decision reflected efficacy, safety, recruitment, or any other cause. The record also does not report the original target sample size or its justification, so the adequacy of enrollment relative to the intended design cannot be assessed from the posted enrollment count alone.

The primary endpoint analysis counts are 154 and 158, while reported enrollment is 313. The registry record does not explain the difference or define the analysis population. Details about missing outcomes, withdrawals, censoring, treatment adherence, and analysis by assigned group are not reported. The registry gives the primary method and estimate but not a full statistical analysis plan, covariate strategy, or sensitivity analyses.

The treatment arms each allowed gemcitabine hydrochloride or pemetrexed disodium with carboplatin. The posted primary result is an overall arm comparison, and the registry record does not report separate outcomes by chemotherapy regimen or tests of interaction. It is therefore not possible to determine from these results whether the comparison varied across the two chemotherapy options. The record also does not give detailed results by COX-2 expression level, despite the official title’s focus on COX-2-expressing disease.

Some secondary outcomes are listed without posted results in the registry record. The toxicity-incidence comparison has no confidence interval, and the serious adverse event counts have no reported comparative uncertainty or causal attribution. Multiplicity methods are not reported. These gaps limit the strength and scope of conclusions that can be drawn, but they do not change the numerical primary result or justify conclusions about unreported analyses.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Gemcitabine Hydrochloride or Pemetrexed Disodium and Carboplatin With or Without Celecoxib in Treating Patients With Advanced Non-Small Cell Lung Cancer — NCT01041781.
• Registry record last updated 2018-07-23; status last verified 2018-06.
• Results first posted 2018-07-23.
• 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

The registry describes a randomized, double-blind, parallel-group phase III trial of celecoxib versus placebo alongside carboplatin and either gemcitabine hydrochloride or pemetrexed disodium for advanced non-small cell lung cancer. Enrollment was 313, and the registry status is terminated. The reason for termination is not reported in the registry record.

The primary endpoint was progression-free survival, defined as time from randomization to progression or death from any cause. The posted medians were 5.16 months with celecoxib and 5.26 months with placebo. The sponsor-reported log-rank comparison gave a hazard ratio of 1.076, a 95% confidence interval from 0.853 to 1.367, and p = 0.535.

The interval includes 1 and allows for relative hazards in either direction. The posted result does not establish equivalence, nor does it establish a clear difference. Secondary outcomes include response and toxicity, with additional outcomes listed but not reported in the registry record. The original sample-size justification, multiplicity procedures, analysis population details, and reason for termination are not reported.