This page separates reported registry information from statistical interpretation. The ClinicalTrials.gov record contains a registered primary time-to-event endpoint and posted outcome measures, but no formal statistical-analysis results.
1. Trial at a Glance
INFORM was a randomized, parallel, double-blind phase 4 trial evaluating gefitinib maintenance therapy against placebo in patients with locally advanced or metastatic (Stage IIIB/IV) non-small cell lung cancer. The registry records 296 enrolled participants and one primary endpoint: progression-free survival.
| Feature | INFORM |
|---|---|
| Trial name | INFORM |
| NCT identifier | NCT00770588 |
| Therapeutic area | Oncology |
| Condition | Non-small Cell Lung Cancer (NSCLC) |
| Brief title | Assess the Efficacy, Safety and Tolerability of Gefitinib (Iressa® 250mg) as Maintenance Therapy in Locally Advanced or Metastatic (Stage IIIB/IV) Non Small Cell Lung Cancer (NSCLC) |
| Phase | Phase 4 |
| Status | COMPLETED |
| Enrollment | 296 |
| Allocation | RANDOMIZED |
| Design model | PARALLEL |
| Masking | DOUBLE |
| Primary purpose | TREATMENT |
| Interventions | Gefitinib (drug); Placebo (drug) |
| Lead sponsor | AstraZeneca |
| Sponsor type | INDUSTRY |
2. Clinical Question
The registered clinical question concerns whether gefitinib used as maintenance therapy produces a difference in progression-free survival compared with placebo in patients with locally advanced or metastatic (Stage IIIB/IV) non-small cell lung cancer.
Population
Patients with locally advanced or metastatic (Stage IIIB/IV) non-small cell lung cancer.
Intervention
Gefitinib, described in the brief title as Iressa® 250mg, used as maintenance therapy.
Comparator
Placebo.
Primary question
How does gefitinib maintenance therapy compare with placebo with respect to progression-free survival?
3. Trial Design
Gefitinib
- Gefitinib (drug)
- Maintenance therapy
- Brief title specifies Iressa® 250mg
Placebo
- Placebo (drug)
- Comparator arm
- Used for the randomized comparison
4. Randomization and Blinding
Randomization and double masking are central statistical features of INFORM. Randomization creates the basis for comparing outcomes between treatment assignments while reducing systematic differences attributable to measured and unmeasured baseline characteristics in expectation. The registry identifies the allocation as randomized and the design as double-blind.
For a time-to-event endpoint such as PFS, these design features matter because the statistical comparison is intended to reflect differences associated with the assigned treatment rather than differences created by selective treatment allocation or knowledge of treatment assignment.
Why randomize?
Randomization provides the fundamental basis for a causal comparison of the treatment assignments, assuming the trial is conducted and analysed appropriately.
Why blind the trial?
Double blinding can reduce the influence of treatment knowledge on participant behavior, investigator decisions, and other aspects of trial conduct.
5. Primary Endpoint
| Endpoint | Registered definition | Endpoint type |
|---|---|---|
| Progression Free Survival (PFS) | From date of randomization until the date of first documented progression or date of death from any cause, whichever came first.The primary analysis of PFS will be performed when at least 265 events have occurred, which is expected to occur approximately. | Time-to-event |
How progression was defined
The registry states that PFS will be calculated from tumor measurements collected at each tumor assessment per the RECIST criteria and/or the date of patient death. Progression is defined, using RECIST, as a measurable increase in the smallest dimension of any target or non-target lesion, or the appearance of new lesions, since baseline.
Thus, the endpoint combines two clinically important event pathways: documented tumor progression and death from any cause. A participant who experiences either event can reach the PFS endpoint. Participants who have not experienced an event at the relevant follow-up point contribute time until censoring under the applicable analysis rules.
The qualifying event described by the registry is first documented progression or death from any cause. The actual statistical analysis requires the event and censoring records for individual participants; those data are not included in the registry summary.
6. Results Status
ClinicalTrials.gov reports that results have been posted and that six outcome measures are posted. However, the ClinicalTrials.gov record contains no formal statistical analyses for the primary endpoint. Consequently, the ClinicalTrials.gov record does not provide a formal comparative estimate for PFS such as a hazard ratio, confidence interval, p-value, or median PFS.
7. Planned Analysis
Because the ClinicalTrials.gov record does not contain a formal statistical analysis for PFS, the appropriate focus is the analysis that would ordinarily be used for a randomized time-to-event endpoint of this type.
Kaplan-Meier estimation
The Kaplan-Meier method is commonly used to estimate the PFS distribution over time. It accommodates right-censored observations, allowing participants who have not yet experienced progression or death at their last evaluable follow-up to contribute information up to that point.
Here, di represents events at an event time and ni represents the number at risk immediately before that time. The resulting curve estimates the probability of remaining event-free through time t.
Comparing randomized groups
For a randomized two-arm PFS comparison, a log-rank test is commonly used to compare the survival distributions between treatment groups. A Cox proportional-hazards model is also commonly used to estimate a hazard ratio and its confidence interval.
Those methods are described here as standard statistical approaches for this endpoint, not as reported analyses of INFORM. The ClinicalTrials.gov record does not state that a particular log-rank test, Cox model, stratified model, or specific variance estimator was used.
What a hazard ratio would answer
A hazard ratio would summarize the relative instantaneous event rate between randomized groups under the fitted model.
What it would not answer
A hazard ratio would not by itself provide the absolute probability of remaining progression-free at a particular time or describe the experience of every individual participant.
What Kaplan-Meier adds
A Kaplan-Meier curve would provide an estimated PFS distribution over follow-up and make censoring and the timing of events visible.
What the p-value would add
A hypothesis-test p-value would address evidence against a specified null hypothesis; it would not measure the magnitude or clinical importance of an effect.
8. How to Interpret a PFS Analysis
PFS is a time-to-event endpoint, so its interpretation differs from a simple proportion of patients who responded. The relevant question is not merely whether progression occurred, but when progression or death occurred relative to randomization.
This distinction is important because two treatment groups can have similar proportions of events but different timing of those events. Conversely, an apparently different event-free proportion at one time point does not by itself summarize the entire follow-up distribution.
A hazard ratio below 1 would indicate a lower estimated instantaneous rate of progression or death in the gefitinib group relative to placebo under the fitted model. It would not mean that the same percentage of every patient experienced a reduction in risk, nor would it directly equal the relative difference in median PFS or the absolute difference in PFS probability.
The confidence interval would describe statistical uncertainty around the estimated treatment effect under the specified model and sampling framework. A narrow interval would indicate greater statistical precision than a wide interval, all else equal. It would not describe the range of outcomes expected for individual patients.
The p-value would quantify how compatible the observed data are with the specified null hypothesis under the assumptions of the test. It would not measure effect size, clinical importance, probability that the treatment works, or probability that the null hypothesis is true.
9. Statistical Methods Explained
Why is PFS a time-to-event endpoint?
PFS records both whether a qualifying event occurs and the time from randomization to that event. The timing component means that ordinary methods designed for a single binary outcome do not fully capture the information. Time-to-event methods are designed to use the event timing while accommodating censoring.
Why is randomization important for PFS?
Randomization establishes the treatment groups before their subsequent outcomes are observed. In a properly conducted randomized comparison, systematic differences in prognosis should not be intentionally allocated to one treatment arm. This makes the difference in PFS between randomized groups interpretable as evidence about the treatment assignment, subject to the usual assumptions and conduct of the trial.
Why use Kaplan-Meier estimation?
Not every participant necessarily experiences progression or death during the observation period. Kaplan-Meier estimation allows participants who remain event-free at their last evaluable follow-up to contribute information without treating them as if they had experienced the event at that time.
What would a Cox hazard ratio mean?
A Cox model would estimate the relative hazard of progression or death between treatment groups. A hazard ratio of 1 would represent equal estimated hazards under the model, whereas a value below 1 would indicate a lower estimated hazard in the numerator treatment group. The interpretation depends on the model and, for a conventional Cox interpretation, the proportional-hazards assumption.
Why is censoring important?
Censoring occurs when a participant's event-free follow-up ends before a qualifying event is observed. Time-to-event analyses rely on assumptions about the relationship between censoring and the event process. The registry-reported INFORM data do not describe the censoring rules in sufficient detail to evaluate them directly.
Why should a missing formal analysis not be replaced with a reconstructed result?
A valid treatment-effect estimate requires the underlying analysis data, the event definitions, censoring rules, model specification, and inferential procedure. Simply knowing that results were posted and that PFS was measured is not enough to recover a defensible hazard ratio, confidence interval, or p-value. A statistical result should be reported only when the source actually provides it or when a transparent independent analysis can be performed from adequate underlying data.
10. Safety Results
The ClinicalTrials.gov record reports serious adverse events by arm. The reported counts are 10 affected participants among 147 at risk in the gefitinib group and 5 among 148 at risk in the placebo group.
| Safety measure | Gefitinib | Placebo |
|---|---|---|
| Serious adverse events | 10/147 | 5/148 |
These figures should be interpreted as reported counts rather than as a formal comparative safety analysis. The ClinicalTrials.gov record does not include a statistical analysis, confidence interval, p-value, exposure-adjusted incidence rate, or detailed definitions beyond the designation of serious adverse events.
11. Safety Interpretation
Safety and efficacy answer different questions. The PFS endpoint concerns the time to progression or death, whereas serious adverse events concern an important category of adverse safety outcomes. A statistical comparison of PFS cannot be used as a substitute for a safety assessment, and a safety count cannot by itself establish an overall treatment benefit or harm.
Observed count
The registry data report 10 serious-adverse-event cases among 147 at risk with gefitinib and 5 among 148 at risk with placebo.
What is not reported
No formal statistical comparison of serious adverse events is included in the ClinicalTrials.gov record.
Why denominators matter
Safety frequencies depend on the population at risk and the definition of the safety analysis set. The registry-reported denominators should therefore be retained exactly.
Why exposure matters
Simple participant counts do not account for differing durations of treatment or observation. Exposure-adjusted analyses would require information not reported here.
12. What the Registry Does and Does Not Establish
The INFORM record establishes several important design and endpoint facts: it was randomized, parallel, and double-blind; it enrolled 296.0 participants; it had two drug interventions, gefitinib and placebo; and it registered PFS as its sole primary endpoint.
The registry-reported statistical-analysis field does not establish the magnitude of the PFS treatment effect. In particular, the ClinicalTrials.gov record does not contain a hazard ratio, confidence interval, p-value, median PFS, Kaplan-Meier estimates, event counts for the primary endpoint, subgroup analyses, or formal statistical model output.
| Question | Supported by the ClinicalTrials.gov record? |
|---|---|
| Was the study randomized? | Yes — allocation is recorded as RANDOMIZED. |
| Was the design parallel? | Yes — design model is PARALLEL. |
| Was the study double-blind? | Yes — masking is DOUBLE. |
| Was PFS a primary endpoint? | Yes — one registered primary endpoint is PFS. |
| Were results posted? | Yes — results are recorded as posted. |
| Is a formal PFS treatment-effect estimate reported? | No — no formal statistical analyses were posted. |
| Is a PFS hazard ratio reported? | No. |
| Is a PFS confidence interval reported? | No. |
| Is a PFS p-value reported? | No. |
| Are serious adverse-event counts reported? | Yes — 10/147 for gefitinib and 5/148 for placebo. |
13. Limitations
- No formal efficacy analysis in the ClinicalTrials.gov record: no formal statistical analyses were posted, so a comparative PFS estimate cannot be reported from the ClinicalTrials.gov record.
- No event-level data: without individual event and censoring information, an independent Kaplan-Meier analysis cannot be reconstructed.
- No reported hazard ratio: the registry summary in the ClinicalTrials.gov record does not provide a hazard ratio for PFS.
- No reported confidence interval or p-value: statistical uncertainty and formal hypothesis-test results cannot be reconstructed without additional analysis information.
- Limited design detail: the ClinicalTrials.gov record identifies randomization, parallel design, and double blinding but does not provide stratification factors, randomization ratios, or detailed allocation procedures.
- Limited safety detail: only serious-adverse-event counts by arm are reported. The data do not provide a complete adverse-event profile or exposure-adjusted safety analysis.
- Safety denominators differ from total enrollment: the reported denominators of 147 and 148 do not sum to 296.0, and the ClinicalTrials.gov record does not explain the difference.
- No subgroup analysis: subgroup estimates and interaction tests are not included in the ClinicalTrials.gov record.
- No multiplicity information: the ClinicalTrials.gov record does not specify an alpha-control strategy or hierarchy beyond identifying one primary endpoint.
- No interim-analysis information: the ClinicalTrials.gov record does not describe interim looks, stopping boundaries, or alpha spending.
- No missing-data or imputation rules: the ClinicalTrials.gov record does not specify how missing tumor assessments or other incomplete observations were handled.
14. Why This Trial Matters Statistically
INFORM is useful as a statistical teaching case because its registered design and endpoint illustrate the basic structure of a randomized time-to-event comparison. At the same time, the record demonstrates an important principle in evidence reporting: a trial can have posted results without every formal statistical analysis being available in the extracted registry data.
| Concept | How it appears in INFORM |
|---|---|
| Randomization | The registry identifies the allocation as RANDOMIZED. |
| Parallel design | The design model is PARALLEL with two intervention arms. |
| Double blinding | The masking classification is DOUBLE. |
| Time-to-event analysis | The sole registered primary endpoint, PFS, is classified as time-to-event. |
| Kaplan-Meier estimation | A standard method for estimating PFS over time, although no formal Kaplan-Meier output is reported here. |
| Hazard ratio | A standard way to summarize a randomized time-to-event comparison, although no INFORM HR is reported here. |
| Censoring | Relevant because PFS is measured over time and not every participant necessarily experiences an observed event during follow-up. |
| RECIST assessment | The registry defines progression using tumor measurements and RECIST criteria. |
| Safety analysis | Serious adverse events are reported as 10/147 versus 5/148 by arm. |
| Evidence transparency | The empty the posted statistical analyses illustrates why posted outcome data should not be treated as equivalent to a complete formal statistical report. |
15. Statistical Interpretation: The Most Important Lessons
PFS is fundamentally a timing endpoint. The analysis should preserve information about when progression or death occurs rather than reducing the outcome immediately to a simple event/no-event comparison.
Because INFORM is recorded as randomized, the treatment assignment provides the structural basis for comparing PFS between the gefitinib and placebo groups. The validity of that interpretation still depends on appropriate trial conduct and analysis.
The absence of a registry-reported hazard ratio or confidence interval is not evidence for either treatment. It means that the requested comparative estimate cannot be responsibly recovered from the ClinicalTrials.gov record.
Even if a p-value were available, it would not communicate the magnitude of the treatment effect by itself. A complete PFS report would ideally pair a treatment-effect estimate with an uncertainty interval and an appropriate description of the survival distributions.
The serious-adverse-event counts are reported against arm-specific denominators of 147 and 148. Those denominators should not be silently replaced with the overall enrollment of 296.0.
16. What a Complete PFS Analysis Would Normally Report
For an educational statistical report of a randomized PFS endpoint, several complementary pieces of information are useful. The registry-reported INFORM data do not contain these results, but they illustrate what a complete analysis would ordinarily seek to communicate.
| Component | Purpose |
|---|---|
| Kaplan-Meier curves | Describe the estimated PFS distribution over time in each randomized group. |
| Number of PFS events | Describe how much event information contributes to the comparison. |
| Median PFS | Provide a time-point summary when the estimated survival curve reaches 0.50. |
| Hazard ratio | Summarize relative event hazard under a Cox-type model. |
| Confidence interval | Quantify uncertainty around the treatment-effect estimate. |
| Hypothesis-test result | Address the prespecified null hypothesis using the selected inferential procedure. |
| Time-specific PFS estimates | Provide absolute event-free probabilities at clinically relevant time points when specified. |
| Censoring information | Help readers understand how incomplete follow-up contributes to the survival analysis. |
None of these additional numerical PFS results should be inferred from the enrollment total, the number of arms, or the existence of posted registry outcomes. Doing so would turn a methodological description into an unsupported efficacy claim.
17. A Practical Framework for Reading INFORM
Start with the estimand
The registered endpoint asks about time from randomization to first documented progression or death from any cause. That defines what the primary analysis is intended to measure.
Check the design
Randomized, parallel, and double-blind design features establish the framework for a comparative treatment analysis.
Check the analysis
Look for the event definition, censoring rules, statistical model, treatment-effect estimate, confidence interval, and hypothesis-test procedure.
Separate evidence from inference
When the formal analysis is absent, standard statistical methods can explain how the endpoint would normally be analysed, but they cannot create a treatment effect that the ClinicalTrials.gov record does not report.
18. Related Tutorials
Learn more about the methods used in this trial:
19. Related Calculators
20. Sources
- ClinicalTrials.gov: INFORM, NCT00770588.
- PubMed: PubMed record for PMID 25546556.
- PubMed: PubMed record for PMID 22512843.
Continue through the Clinical Biostats statistical library
Use the related tutorials and calculators to explore randomized trial design, time-to-event endpoints, Kaplan-Meier estimation, and hazard-ratio methodology.
21. Record Summary
INFORM is recorded as a completed, randomized, parallel, double-blind phase 4 trial with 296.0 enrolled participants and two drug interventions: gefitinib and placebo. Its registered primary endpoint is progression-free survival, defined from randomization to first documented progression or death from any cause, with progression assessed using tumor measurements and RECIST criteria.
The most important statistical feature of the ClinicalTrials.gov record is the distinction between posted results and reported formal statistical analysis. The ClinicalTrials.gov record indicates that results were posted and six outcome measures are available, but no formal statistical analyses were posted.
The safety information that is reported consists of serious adverse events by arm: 10/147 for gefitinib and 5/148 for placebo. Because the denominators differ from the overall enrollment and the ClinicalTrials.gov record does not explain the safety analysis population, those figures are reproduced exactly rather than recalculated.