This page separates information posted in the ClinicalTrials.gov record from statistical interpretation. The registry identifies a primary time-to-event endpoint and reports that results are posted, but the ClinicalTrials.gov record contains no formal statistical analyses for that endpoint.
1. Trial at a Glance
IPASS was a randomized, parallel-group, open-label phase 3 treatment trial comparing gefitinib with carboplatin and paclitaxel in non-small cell lung cancer. The registry reports an enrollment of 1329 participants and one registered primary endpoint: median progression-free survival in months.
| Feature | IPASS |
|---|---|
| Trial name | IPASS |
| NCT identifier | NCT00322452 |
| Brief title | First Line IRESSA™ Versus Carboplatin/Paclitaxel in Asia |
| Phase | Phase 3 |
| Status | COMPLETED |
| Condition | Non-Small Cell Lung Cancer |
| Allocation | RANDOMIZED |
| Design model | PARALLEL |
| Masking | NONE |
| Primary purpose | TREATMENT |
| Enrollment | 1329 |
| Lead sponsor | AstraZeneca |
| Sponsor type | INDUSTRY |
2. Clinical Question
The registered clinical question concerns first-line treatment of non-small cell lung cancer: how does gefitinib compare with carboplatin/paclitaxel with respect to the registered primary time-to-event outcome, median progression-free survival?
Population
Patients enrolled in the IPASS phase 3 trial with the registered condition of non-small cell lung cancer.
Intervention
Gefitinib.
Comparator
Carboplatin and paclitaxel.
Primary question
How does the randomized comparison relate to median progression-free survival in months?
3. Trial Design
Gefitinib
- Intervention: gefitinib.
- Drug intervention.
Carboplatin/Paclitaxel
- Interventions: carboplatin and paclitaxel.
- Both are registered as drug interventions.
The registry describes the trial as randomized, parallel, unmasked, and intended for treatment. These design characteristics are important because randomization establishes the basis for comparing assigned treatment groups, while the absence of masking can affect how outcomes are assessed or reported when outcomes are susceptible to knowledge of treatment assignment.
4. Endpoints
| Endpoint | Registry definition | Time frame |
|---|---|---|
| Median Progression Free Survival (PFS) in Months | PFS was defined as the interval from the date of randomization to the date of objective disease progression (as per RECIST) or the date of death (from any cause) in the absence of objective disease progression. The median PFS in months is presented here. | Tumour assessments as per RECIST were performed at baseline and then every 42 days ± 7 days from randomization until data cut off (14th April 2008). |
Why PFS is a time-to-event endpoint
Progression-free survival is not simply a binary response variable. Each participant contributes a time from randomization until an event occurs or until follow-up ends without an observed event. That structure creates censoring, unequal follow-up, and a need for methods designed for time-to-event data.
The registered definition also makes the event rule explicit: objective disease progression according to RECIST or death from any cause in the absence of objective disease progression. Consequently, death can count as a PFS event even when objective progression has not previously been recorded.
5. Statistical Methodology
The ClinicalTrials.gov record identifies the primary endpoint as time-to-event, but no formal statistical analyses were posted. Therefore, the statistical methods below describe the standard analytical framework appropriate for this endpoint rather than claiming that these methods were formally reported for IPASS in the ClinicalTrials.gov record.
Kaplan-Meier estimation
For a time-to-event endpoint such as PFS, Kaplan-Meier estimation is typically used to describe the distribution of event times while retaining information from participants who are censored before an event.
Here, di represents the number of events at event time ti, while ni is the number of participants at risk immediately before that time.
Median progression-free survival
The registered endpoint specifically requests median PFS in months. In a Kaplan-Meier analysis, the median is the time at which the estimated survival function reaches 0.50, provided that this point is estimable within the observed follow-up.
The median is therefore a descriptive summary of the time-to-event distribution. It is different from a hazard ratio, which compares event hazards between treatment groups over the modeled follow-up.
Log-rank comparison
For a randomized comparison of two time-to-event distributions, a log-rank test is a standard method for testing whether the survival experience differs between groups. A stratified version can be used when the design includes prespecified stratification factors. The ClinicalTrials.gov record does not identify stratification factors or report a formal log-rank analysis for IPASS.
Cox proportional-hazards model
A Cox proportional-hazards model is commonly used to estimate a hazard ratio comparing two treatment groups. Unlike the median, the hazard ratio is a relative measure of the instantaneous event rate under the fitted model.
A hazard ratio would summarize relative event hazards; it would not directly state the difference in median PFS or the absolute probability of remaining progression-free at a particular time.
Analysis population
Randomization makes an intention-to-treat comparison the natural framework for estimating the effect of assigned treatment. However, the ClinicalTrials.gov record does not specify the analysis populations used for IPASS or identify an ITT population in the statistical-analysis records.
6. Planned Analysis
The ClinicalTrials.gov record reports that results have been posted and identifies one registered primary endpoint, but the posted statistical analyses contains no formal comparison estimates. No hazard ratio, confidence interval, p-value, median PFS value, or treatment-group PFS estimate is provided in the ClinicalTrials.gov record.
Primary endpoint
Registered endpoint: Median Progression Free Survival (PFS) in Months.
The registry defines PFS from randomization to objective disease progression according to RECIST or death from any cause in the absence of objective disease progression.
For this endpoint, a typical analysis would begin by estimating Kaplan-Meier PFS curves separately by randomized treatment group. The median PFS would then be obtained from each estimated curve when estimable. A between-group comparison would commonly use a log-rank test, and a Cox model could provide a hazard ratio with a confidence interval.
What a complete PFS analysis would contain
| Component | Purpose |
|---|---|
| Kaplan-Meier estimate | Describe the progression-free survival distribution over time while accommodating censoring. |
| Median PFS | Summarize the time at which the estimated PFS curve reaches 50%, when estimable. |
| Log-rank test | Provide a hypothesis test comparing the time-to-event distributions between randomized groups. |
| Cox model | Estimate a relative hazard measure such as a hazard ratio and its confidence interval. |
| Event and censoring rules | Determine how progression, death, and incomplete follow-up contribute to the analysis. |
7. Statistical Methods Explained
Why is PFS analyzed as a time-to-event endpoint?
PFS records not only whether progression or death occurs but also when it occurs. Participants who have not experienced progression or death by the end of their available follow-up may be censored rather than treated as if they had experienced no event indefinitely. Time-to-event methods are designed for this structure.
Why does the analysis start at randomization?
The registry definition specifies that PFS is measured from the date of randomization. Starting the clock at randomization aligns the endpoint with the randomized treatment assignment and establishes a common origin for comparing participants in the two treatment groups.
Why does death count as a PFS event?
The registered definition treats death from any cause as a PFS event when objective disease progression has not occurred. This prevents a participant who dies without a recorded progression assessment from being treated as indefinitely progression-free.
Why are tumour assessments performed repeatedly?
The registry specifies tumour assessments at baseline and then every 42 days ± 7 days from randomization. Repeated assessments provide opportunities to identify objective disease progression according to the registered RECIST framework.
Why would Kaplan-Meier estimation be appropriate?
Kaplan-Meier estimation is designed for event-time data with censoring. It uses the information available from each participant up to the time of an event or censoring rather than requiring every participant to be followed until progression or death.
What would a hazard ratio add beyond median PFS?
The median describes one point on each treatment group's estimated time-to-event distribution. A hazard ratio instead summarizes the relative event hazard under a Cox model over the analyzed follow-up. The two measures answer related but different questions and should not be treated as interchangeable.
Why is the absence of a posted p-value important?
A p-value cannot be reconstructed from the enrollment count alone. It requires an appropriate comparison statistic and the corresponding event-time information. Because the ClinicalTrials.gov record contains no formal comparison, reporting a p-value would require information not provided in the trial data.
8. Understanding Censoring in PFS
Censoring is central to the interpretation of a progression-free survival endpoint. A participant may remain without a recorded progression or death during the period in which they are observed. The statistical analysis must distinguish this incomplete observation from an actual event.
Observed event
Objective disease progression according to RECIST, or death from any cause without objective disease progression, meets the registered PFS event definition.
Censored observation
A participant without an observed PFS event contributes information up to the relevant censoring point under the prespecified analysis rules.
This distinction explains why simply dividing the number of participants by the number of events is not a valid way to calculate median PFS. The timing of events and censoring matters.
9. Interpretation of a Hypothetical Hazard Ratio
If a formal Cox analysis reported a hazard ratio below 1 for gefitinib versus carboplatin/paclitaxel, that would indicate a lower estimated instantaneous hazard of progression or death for the gefitinib group under the fitted model. The registry-reported IPASS data do not provide such a hazard ratio, so no numerical treatment effect is reported here.
A hazard ratio would not mean that the same percentage of individual patients avoided progression, nor would it directly represent the difference between the two median PFS values. It is a relative time-to-event measure rather than an absolute risk difference.
If a hazard ratio were reported, its confidence interval would quantify statistical uncertainty around the estimated relative hazard under the model and sampling framework. It would not describe the range of treatment effects experienced by individual patients.
A p-value would address evidence against a specified null hypothesis under the statistical model. It would not tell the reader how large the treatment difference is. Effect size and uncertainty should be evaluated using measures such as median differences, hazard ratios, confidence intervals, and absolute time-specific estimates when available.
10. Safety
The ClinicalTrials.gov record reports serious adverse events by treatment arm. These figures are presented as affected participants divided by the corresponding number at risk.
| Safety measure | Gefitinib | Carboplatin/Paclitaxel |
|---|---|---|
| Serious adverse events, affected / at risk | 110 / 607 | 92 / 589 |
The denominators in the registry data are different from the overall enrollment of 1329. Therefore, the serious-adverse-event figures should be interpreted using their stated at-risk denominators rather than assuming that they represent all enrolled participants.
11. Trial Timeline
Trial start
The registered IPASS trial began in March 2006.
Primary completion
The registry records primary completion in April 2008.
Registry status
The ClinicalTrials.gov record identifies the trial status as completed and reports results as posted.
12. What the Registered Endpoint Tells Us
The primary endpoint is specifically Median Progression Free Survival (PFS) in Months. That wording matters statistically. It identifies the target summary measure as a median rather than, for example, a response proportion or a fixed-time survival probability.
| Question | What the registered endpoint addresses |
|---|---|
| What is measured? | Progression-free survival. |
| When does measurement begin? | The date of randomization. |
| What ends the PFS interval? | Objective disease progression according to RECIST or death from any cause in the absence of objective disease progression. |
| What summary is requested? | Median PFS in months. |
| How are tumours assessed? | At baseline and then every 42 days ± 7 days from randomization according to the registry time frame. |
The endpoint therefore combines clinical assessment rules with a time-to-event statistical framework. Both components matter: changing the event definition or the timing of assessments could change the resulting PFS distribution even if the underlying disease process were unchanged.
13. Important Limitations and Interpretation Issues
- No formal statistical-analysis records reported: the trial data identify results as posted but contain an empty statistical-analysis array. Numerical efficacy estimates, confidence intervals, and p-values therefore cannot be reproduced from the ClinicalTrials.gov record.
- Primary endpoint only: the ClinicalTrials.gov record identifies one registered primary endpoint. They do not provide statistical analyses for additional endpoints among the 16 outcome measures posted.
- Incomplete time-frame wording: the registry-reported primary-endpoint time frame ends with “until dat,” so the final words of the registry time frame are not available in the ClinicalTrials.gov record.
- Censoring: PFS requires appropriate handling of participants without an observed progression or death during available follow-up.
- Assessment timing: tumour assessments were scheduled at baseline and every 42 days ± 7 days, so the observed timing of progression is connected to the assessment schedule.
- Unmasked design: the registry identifies masking as NONE. This is a design characteristic that should be considered when interpreting outcomes, particularly outcomes involving assessment or clinical decision-making.
- Hazard-ratio assumptions: if a Cox proportional-hazards model were used, interpretation of a single hazard ratio would depend on the proportional-hazards framework. The ClinicalTrials.gov record does not report whether this assumption was formally assessed.
- Analysis population: the ClinicalTrials.gov record does not specify the formal efficacy or safety analysis populations. The serious-adverse-event denominators are reported separately as 607 and 589.
- Randomization does not eliminate sampling uncertainty: randomized allocation provides the basis for the comparison, but treatment-effect estimates still have statistical uncertainty that should be quantified with confidence intervals.
14. Why This Trial Matters Statistically
IPASS is a useful teaching example because the registry presents a straightforward randomized treatment comparison while its primary outcome has the more complex structure of a time-to-event endpoint. Understanding the trial therefore requires moving beyond a simple comparison of proportions.
| Concept | How it appears in IPASS |
|---|---|
| Randomization | The trial uses randomized allocation. |
| Parallel design | The registry identifies a parallel design model with two arms. |
| Time-to-event analysis | The primary endpoint is classified as time-to-event. |
| Progression-free survival | PFS runs from randomization to objective progression or death under the registered definition. |
| RECIST assessment | Tumour assessments are specified at baseline and then every 42 days ± 7 days. |
| Kaplan-Meier estimation | A standard framework for estimating PFS distributions in the presence of censoring. |
| Median survival time | The registered primary endpoint specifically requests median PFS in months. |
| Hazard ratio | A Cox model would commonly provide a relative time-to-event effect estimate, although no IPASS HR is reported here. |
| Confidence interval | Would quantify uncertainty around an estimated treatment effect if a formal analysis were available. |
| Safety denominators | Serious adverse events are reported as 110/607 and 92/589 by arm. |
15. Related Tutorials
Learn more about the method used in this trial:
16. Related Statistical Calculators
17. Sources
- ClinicalTrials.gov: NCT00322452 — IPASS.
- PubMed record: PMID 31920008.
- PubMed record: PMID 28212993.
- PubMed record: PMID 24361280.
- PubMed record: PMID 23540718.
- PubMed record: PMID 22897752.
The five PubMed records above are the publications identified in the ClinicalTrials.gov record. No publication-specific numerical results are incorporated into this page because the ClinicalTrials.gov record contains no statistical-analysis estimates for the primary endpoint.
Continue through the Clinical Biostats trial-analysis pathway
Use the broader Clinical Biostats resources to explore randomized-trial design and statistical methods for time-to-event outcomes.
18. Record Summary
IPASS is a completed randomized phase 3 trial in non-small cell lung cancer comparing gefitinib with carboplatin/paclitaxel. The ClinicalTrials.gov record reports 1329 participants, two parallel treatment arms, no masking, and one registered primary endpoint: median progression-free survival in months. PFS is defined from randomization to objective disease progression according to RECIST or death from any cause in the absence of objective disease progression, with tumour assessments scheduled at baseline and then every 42 days ± 7 days from randomization under the registry-reported time frame.
The statistical structure is therefore fundamentally a time-to-event problem. Kaplan-Meier estimation would ordinarily describe the PFS distribution, the median would summarize the time at which the estimated PFS curve reaches 50%, and a log-rank test or Cox model could ordinarily support a randomized treatment comparison.
The safety information that is available is similarly specific: serious adverse events are reported as 110/607 for gefitinib and 92/589 for carboplatin/paclitaxel. Those figures are presented as reported in the registry, without converting them into derived percentages or assigning a formal statistical comparison that is not present in the registry data.