A Study To Assess The Safety And Efficacy Of SU11248 In Patients With Gastrointestinal Stromal Tumor(GIST): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study To Assess The Safety And Efficacy Of SU11248 In Patients With Gastrointestinal Stromal Tumor(GIST) (NCT00075218), a randomized phase 3 study. The primary outcome is Time to Tumor Progression (TTP) as Assessed by Imaging Studies at End of Double-blind Treatment Phase.

2. Clinical Question

The central statistical question concerns the timing of documented progression, not simply whether progression occurred. The registry reports median time to progression and a hazard ratio from a log-rank analysis. The median describes the point by which the estimated event-time distribution reaches its midpoint; the hazard ratio compares the event rates over follow-up under the model used for the analysis. These are different summaries, so they should be read together rather than treated as interchangeable.

3. Trial Design

The registry describes a randomized, double-blind, placebo-controlled Phase III study with a crossover model. The sponsor was Pfizer, and the listed condition was gastrointestinal stromal tumor. Enrollment was 361. The record marks the study completed, with a start date of 2003-12 and primary completion in 2008-05.

Participants had malignant GIST that had failed imatinib therapy or for which they were intolerant to imatinib. The double-blind comparison was sunitinib versus placebo. The registry gives 207 participants for the sunitinib double-blind group and 105 for the placebo double-blind group for the primary outcome. Those outcome denominators should not be assumed to represent every randomized participant: the primary analysis description separately states that 82 sunitinib-group subjects and 67 placebo-group subjects who experienced progression during the blinded phase were included in the TTP analysis.

The crossover designation means that the record identifies a crossover model, but the sequence of treatment changes, timing, and rules for crossing over are not reported in the registry record. The arm descriptions are blank. Details such as the treatment schedule, allocation ratio, concealment procedures, and whether or how participants who crossed over were handled in each analysis are not reported in the registry record.

The primary endpoint was time to tumor progression assessed by imaging at the end of the double-blind treatment phase. Other posted outcomes include progression-free survival, overall survival, response measures, time to tumor response, and maintenance of performance status. The record also includes an open-label sunitinib group in its adverse-event results. That group is distinct from the two double-blind arms and should not be folded into the randomized comparison without further analysis details.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Before assignment: 361 subjects randomized to double-blind treatment in 2:1 ratio (sunitinib vs. Placebo). 255 subjects continued on or crossed over to Open-label treatment.

Double-Blind Treatment

Reasons for not completing this period:

Open-Label Treatment

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: From the Intent to Treat (ITT) population, 82 subjects on sunitinib treatment were observed to have disease progression during blinded phase and were included in TTP analysis. 67 subjects on placebo were observed to have disease progression during blinded phase.

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 Time to Tumor Progression (TTP) as Assessed by Imaging Studies at End of Double-blind Treatment Phase, measured over Day 28 of each 6-week cycle : duration of double-blind treatment phase.

Time to tumor progression (TTP) was defined as the time from randomization to the first documentation of objective tumor progression. Progression was based on imaging assessed by an independent, third-party laboratory using RECIST, the Response Evaluation Criteria in Solid Tumors. The assessment was scheduled for Day 28 of each 6-week cycle during the double-blind treatment phase.

A time-to-event endpoint uses both whether an event was recorded and when it was recorded. This can preserve information about the timing of progression across follow-up rather than reducing the outcome to a simple count of people with progression by a chosen cutoff. The median TTP describes the estimated time at which half of the relevant event-time distribution has been reached. The hazard ratio offers a comparative summary of event rates over the observed follow-up, under the assumptions of the analysis.

Imaging-based assessment and independent review are relevant because progression depends on applying criteria to scans. Independent review can reduce the influence of treatment knowledge on assessment, although the registry does not report the operational details of image review, disagreement resolution, or how incomplete scans were handled. TTP is not identical to progression-free survival: the registry posts these as separate outcomes, and their definitions and analysis populations may differ. The full protocol definitions and censoring rules are not reported in the registry record.

7. Results Posted in the Registry

• Sunitinib Double-Blind Treatment vs Placebo Double-Blind Treatment: hazard ratio 0.33 (95% CI 0.23–0.47), reported p < 0.001 (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.

Progression Free Survival (PFS)

• Endpoint type: Time to event: compared as a hazard ratio.
• Test: Hazard ratio as reported by the sponsor (Log Rank); it was not recalculated.
• Result:
  • Sunitinib Double-Blind Treatment vs Placebo Double-Blind Treatment: hazard ratio 0.35 (95% CI 0.25–0.47), reported p < 0.001 (as reported in the registry).
• Interpretation:
  • The 95% confidence interval excludes no difference (a ratio of 1), so the hazard of the event is statistically lower for Sunitinib Double-Blind Treatment vs Placebo Double-Blind Treatment than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Overall Survival

• Endpoint type: Time to event: compared as a hazard ratio.
• Test: Hazard ratio as reported by the sponsor (Log Rank); it was not recalculated.
• Result:
  • Sunitinib Double-Blind Treatment vs Placebo Double-Blind Treatment: hazard ratio 0.88 (95% CI 0.68–1.13), reported p = 0.306 (as reported in the registry).
• Interpretation:
  • The 95% confidence interval includes no difference (a ratio of 1), so these data do not show a statistically significant difference between Sunitinib Double-Blind Treatment vs Placebo Double-Blind Treatment and the control. The interval gives the range of effects that remain compatible with the data.

Overall Survival Based on the Rank Preserving Structural Failure Time Method

• Endpoint type: Time to event: compared as a hazard ratio.
• Test: Hazard ratio as reported by the sponsor (Rank Preserving Structural Failure Time); it was not recalculated.
• Result:
  • Sunitinib Double-Blind Treatment vs Placebo Double-Blind Treatment: hazard ratio 0.51 (95% CI 0.26–1.13), reported p = 0.306 (as reported in the registry).
• Interpretation:
  • The 95% confidence interval includes no difference (a ratio of 1), so these data do not show a statistically significant difference between Sunitinib Double-Blind Treatment vs Placebo Double-Blind Treatment and the control. The interval gives the range of effects that remain compatible with the data.

The posted secondary outcomes span progression-free survival, overall survival, response, time to response, and performance-status maintenance. The overall pattern is not uniform across these outcome types. The progression-free survival comparison is directionally similar to the primary TTP comparison, while the overall-survival analysis does not show the same clearly separated statistical estimate. The record also includes an overall-survival analysis using a method intended to address crossover. These related outcomes provide context, but they do not replace the primary endpoint or remove uncertainty about the analysis choices.

Secondary endpoints can differ from TTP in their event definitions, follow-up, susceptibility to crossover, and analysis populations. A result from one endpoint cannot automatically be transferred to another. Response outcomes, for example, concern tumor response rather than the timing of first progression. Overall survival concerns death and may be influenced by treatment received after the blinded comparison. The registry lists additional outcome measures, but only a subset of secondary results is summarized in the record.

Multiplicity matters because examining many outcomes or analyses creates more opportunities for a result to appear unusual by chance. Unless testing is controlled through a prespecified hierarchy or adjustment, secondary p-values are generally less persuasive as stand-alone confirmatory evidence. The record lists 14 outcome measures and 10 analyses, but does not report a multiplicity strategy or endpoint hierarchy. Accordingly, secondary findings are best treated as supportive or exploratory unless the complete protocol and analysis plan establish otherwise. Their role is to add context and identify questions for further evaluation, not to independently establish a definitive conclusion.

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

• Progression Free Survival (PFS)
• Overall Survival Status of Subjects
• Overall Survival
• Overall Survival Based on the Rank Preserving Structural Failure Time Method
• Best Overall Tumor Response During Double-blind Treatment Phase
• Confirmed Objective Response (CR or PR) in Subjects
• Time to Tumor Response (TTR)
• Duration of Performance Status Maintenance

9. Statistical Interpretation

10. Statistical Methods Explained

How does a time-to-event analysis work?

Time-to-event methods analyze the elapsed time until a defined event, such as the first documented tumor progression. Participants may have different lengths of follow-up, and some may not have a documented event before observation ends. Such observations are generally treated as censored: they contribute information up to the last time their event status is known, without being counted as if an event occurred then. This is a general explanation; the registry does not report the specific censoring rules used for the TTP analysis.

The reported median is a summary of the estimated event-time distribution. It is not an average, and it does not mean that each participant experienced progression at that time. The hazard ratio summarizes the relative event rates between groups over follow-up. A hazard ratio of 1 corresponds to equal event rates under the comparison; values on either side indicate the direction of the estimated comparison. Interpreting a single hazard ratio across follow-up is most straightforward when the relative rates are reasonably stable over time. Whether that condition was assessed is not reported in the registry record.

What does the log-rank test assess?

The log-rank method compares event-time experience between groups while accounting for the timing of events and censoring. The registry names this method for the primary comparison and reports a hazard ratio with its confidence interval and p-value. The test result is evidence about whether the observed event-time patterns are compatible with its null hypothesis; it does not by itself describe clinical importance, explain why the groups differ, or resolve concerns about study conduct or analysis choices. The registry does not report diagnostic checks or additional details about how the hazard ratio was estimated.

What does the imaging assessment contribute?

The primary definition relies on the first documented objective progression, based on imaging reviewed by an independent, third-party laboratory using RECIST. This makes the event a defined assessment rather than a general clinical impression. However, the record does not provide the imaging schedule beyond Day 28 of each 6-week cycle, the version or operational details of RECIST, adjudication procedures, or rules for missing and unevaluable scans. Any such details could affect the date on which progression is assigned.

What is the role of the confidence interval?

A confidence interval gives a range of values compatible with the data under the statistical procedure and its assumptions. For the primary hazard ratio, the reported 95% interval is 0.233 to 0.466. The interval is on the relative event-rate scale; it is not an interval for median weeks or a range containing most individual participants’ outcomes. Precision depends on the amount and distribution of event information, not only on the total number enrolled. The registry does not post an event-time dataset or enough information to reproduce the interval independently.

How should crossover be considered?

The trial is identified as having a crossover model. Crossover can complicate interpretation of outcomes measured after participants switch from their assigned treatment, particularly overall survival, because later outcomes may reflect exposure to more than the initially assigned treatment. The registry posts a rank-preserving structural failure time method for one overall-survival analysis. In general, such methods aim to estimate treatment effects while accounting for switching, but they depend on assumptions about how treatment affects event times. The assumptions, crossover timing, and implementation details are not reported in the registry record. The method should therefore be understood as an additional analysis, not as a guarantee that crossover-related uncertainty has been removed.

What can and cannot be inferred from the posted analysis?

The reported primary analysis supports a comparison of TTP between the named double-blind groups under the analysis used. It does not establish that every endpoint was tested under a shared confirmatory procedure. The record lists 14 outcome measures and 10 analyses, but does not report the full prespecified statistical analysis plan, hierarchy of endpoints, covariate adjustments, or multiplicity strategy. Nor does it report whether interim analyses or stopping rules were used. Those omissions limit the ability to judge how the primary result fits into the trial’s complete decision process.

11. Confidence Intervals and Precision

The primary hazard ratio is 0.329, and its 95% confidence interval is 0.233 to 0.466. The interval is relatively distant from 1, the reference value for equal event rates in a hazard-ratio comparison. The reported p-value is < 0.001. These results indicate that the estimated comparison is not merely a single point value: the interval gives a range of estimates compatible with the data and the method’s assumptions.

The median estimates also have uncertainty. For sunitinib, the reported median TTP is 27.3 weeks, with a 95% confidence interval of 16.0 to 32.1 weeks. For placebo, it is 6.4 weeks, with a 95% confidence interval of 4.4 to 10.0 weeks. These intervals apply to the group-specific median estimates, whereas the hazard-ratio interval applies to a relative comparison. Overlap or non-overlap between separate median intervals is not a replacement for the reported comparison test.

Precision is influenced by the number and timing of events, the amount of follow-up, and the pattern of censoring. The registry reports 82 progression events in the sunitinib group and 67 in the placebo group for the primary analysis description, but does not provide the event-time data or detailed risk sets. The outcome denominators are 207 and 105. Those figures describe different aspects of the record and should not be treated as interchangeable.

A confidence interval does not account for every possible source of uncertainty. It reflects the statistical procedure under its assumptions, not uncertainty from unreported analysis choices, measurement limitations, missing information, or applicability to other populations. The registry record does not report enough detail to assess those sources fully. Precision in the posted estimate therefore should not be mistaken for certainty about all aspects of the result.

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 planning illustrations use a stated fraction of the observed effect and report the total number of events for time-to-event outcomes. They are not participant counts per arm. For an illustration using the observed effect, the primary TTP calculation lists 26 total events for 80% power and 35 total events for 90% power. For planning at 75% of the observed effect, it lists 46 and 61 total events; at 50% of the observed effect, it lists 102 and 137 total events. The secondary outcomes have their own event-based illustrations.

The stated planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is the stated fraction of the observed effect. For outcomes other than time-to-event outcomes, the illustrations are expressed as participants per arm; for time-to-event outcomes, they are expressed as total events. The 80% power column describes the event or participant count calculated to give an 80% chance of detecting the specified effect under those assumptions if that effect were true. The 90% power column gives the corresponding planning count for a 90% chance.

Smaller effects require more information to detect reliably, which is why the illustrations for a smaller fraction of the observed effect show larger event counts. The calculations are planning illustrations based on the observed effect, not the trial’s original design assumptions. The original sample-size justification and design assumptions are not reported in the registry record. These figures are not observed power: they do not measure the probability that the completed trial detected its true effect.

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 = 0.329, so ln(HR) = -1.1117.
2. d = 4 × (1.96 + 0.842)² / [ln(0.329)]² = 25.4, rounded up to 26 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 <- 0.329                              # 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(75218)
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 = 26 (80% power), 35 (90% power)
 75% of observed effect: events = 46 (80% power), 61 (90% power)
 50% of observed effect: events = 102 (80% power), 137 (90% power)
Simulated power with 26 events: 0.826 (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. Group-Sequential Design Analysis

Many time-to-event trials, especially in oncology, stop or report early when interim data are overwhelming, so a single fixed-sample number understates the real design problem. The calculation below is an illustrative group-sequential design built from this trial's posted control median and hazard ratio. The registry record does not state the planned number of looks or the spending function, so the design is a planning illustration, not a reconstruction of the protocol. Assumed: one-sided α = 0.025, 90% power, three equally spaced looks on information, Lan-DeMets O'Brien-Fleming spending, 24 months of accrual, 60 months in total and a 1% monthly dropout hazard.

Under these assumptions the design needs about 36 participants in total, with the boundaries below.

Fixed design versus group-sequential alternatives

Group-sequential designs cost a small increase in maximum sample size and events compared with the fixed design, in exchange for the chance to stop early for overwhelming efficacy. O'Brien-Fleming spending keeps early boundaries very conservative and the final boundary close to the fixed-design value, whereas Pocock spending spends more alpha early.

Sensitivity to a smaller true effect

Planning for exactly the observed effect is optimistic, because observed effects in published trials tend to be somewhat larger than the true effect. The table repeats the main design (three looks, O'Brien-Fleming spending, 90% power) for smaller true hazard ratios.

For context, this trial enrolled 361 participants.

Cross-check, upper boundaries: gsDesign 3.71, 2.511, 1.993 | rpact 3.71, 2.511, 1.993. Agreement between the two packages confirms the boundary calculation.

R code and output

suppressWarnings(suppressMessages({ library(gsDesign); library(rpact) }))
med_control <- 1.47692307692308          # control-arm median, months (converted from the posted unit)
hr          <- 0.329          # hazard ratio used for planning
lambdaC     <- log(2) / med_control
eta         <- 0.01               # dropout hazard per month (assumption)
acc_months  <- 24                 # accrual duration (assumption)
study_months<- 60                 # total study duration (assumption)

design <- function(k, sfu, beta, h = hr) gsSurv(k = k, test.type = 1, alpha = 0.025, beta = beta, sfu = sfu,
  lambdaC = lambdaC, hr = h, hr0 = 1, eta = eta, gamma = 1, R = acc_months, T = study_months, minfup = study_months - acc_months)

# Main design: 3 looks, Lan-DeMets O'Brien-Fleming spending, one-sided alpha 0.025, 90% power
x <- design(3, sfLDOF, 0.10)
cat(sprintf("N_TOTAL|%s\n", format(ceiling(sum(x$gamma * x$R)), big.mark = ",")))
for (i in 1:3) cat(sprintf("GS|%d|%.1f|%.0f|%.3f|%.5f|%.5f\n", i, x$T[i], x$n.I[i], x$upper$bound[i], pnorm(-x$upper$bound[i]), cumsum(x$upper$spend)[i]))
cat("\n"); print(gsBoundSummary(x))

# Comparison of designs
cmp <- list(list("Fixed design, one analysis", 1, sfLDOF), list("Two looks, O'Brien-Fleming", 2, sfLDOF),
            list("Three looks, O'Brien-Fleming", 3, sfLDOF), list("Three looks, Pocock", 3, sfLDPocock))
cat("\n")
for (b in c(0.20, 0.10)) for (d in cmp) {
  y <- design(d[[2]], d[[3]], b)
  cat(sprintf("DES|%s|%d|%.0f|%s|%.1f\n", d[[1]], round(100 * (1 - b)), tail(y$n.I, 1), format(ceiling(sum(y$gamma * y$R)), big.mark = ","), tail(y$T, 1)))
}

# Sensitivity: what if the true effect is smaller than the observed one? (3 looks, O'Brien-Fleming, 90% power)
cat("\n")
for (fr in c(1, 0.75, 0.5)) {
  y <- design(3, sfLDOF, 0.10, exp(log(hr) * fr))
  cat(sprintf("SENS|%.2f|%.3f|%.0f|%s\n", fr, exp(log(hr) * fr), tail(y$n.I, 1), format(ceiling(sum(y$gamma * y$R)), big.mark = ",")))
}

# Independent cross-check of the boundaries with rpact
r <- getDesignGroupSequential(kMax = 3, alpha = 0.025, beta = 0.10, sided = 1, typeOfDesign = "asOF", informationRates = x$timing)
cat(sprintf("\nCross-check, upper boundaries: gsDesign %s | rpact %s\n", paste(round(x$upper$bound, 3), collapse = ", "), paste(round(r$criticalValues, 3), collapse = ", ")))

Method: LachinFoulkes 
   Analysis               Value Efficacy
  IA 1: 33%                   Z   3.7103
      N: 18         p (1-sided)   0.0001
 Events: 12        ~HR at bound   0.1110
  Month: 12    P(Cross) if HR=1   0.0001
            P(Cross) if HR=0.33   0.0338
  IA 2: 67%                   Z   2.5114
      N: 30         p (1-sided)   0.0060
 Events: 23        ~HR at bound   0.3493
  Month: 20    P(Cross) if HR=1   0.0060
            P(Cross) if HR=0.33   0.5603
      Final                   Z   1.9930
      N: 36         p (1-sided)   0.0231
 Events: 35        ~HR at bound   0.5058
  Month: 60    P(Cross) if HR=1   0.0250
            P(Cross) if HR=0.33   0.9000



Cross-check, upper boundaries: gsDesign 3.71, 2.511, 1.993 | rpact 3.71, 2.511, 1.993

15. Analyses Reported by the Sponsor

The registry reports the sponsor’s primary comparison as a log-rank hazard ratio of 0.329, with a 95% confidence interval of 0.233 to 0.466 and p < 0.001. The displayed recalculation is not a competing sponsor analysis: it is a simple, unadjusted recalculation from posted summary statistics. The record indicates no discrepancy between that recalculation and the sponsor’s reported analysis.

For interpretation, the sponsor’s reported analysis is the primary reference. A recalculation from aggregate summaries cannot reproduce participant-level modeling, verify the analysis population, inspect censoring, or test the assumptions used in the original analysis. Agreement in the reported comparison does not establish that every analytic detail is independently verified.

The registry does not report the complete statistical analysis plan, so it is not possible to establish whether the reported method and endpoint were prespecified in the precise form used, or whether other analysis choices were made. Nor does the record provide enough information to recreate the log-rank analysis from individual event and censoring times. The posted hazard ratio, confidence interval, and p-value should therefore be read as the sponsor-reported result, with the aggregate recalculation serving only as a limited cross-check.

16. Multiplicity, Missing Data and Analysis Populations

The registry record does not report whether the trial used a prespecified hierarchy of outcomes, adjusted significance thresholds, or another method to control multiplicity. It lists 14 outcome measures and 10 analyses, including several related time-to-event and response outcomes. When many outcomes or analyses are considered, the chance of at least one apparently notable result can increase. A prespecified testing strategy can help clarify which comparisons are confirmatory and how later tests should be interpreted. Without that information, the primary and secondary findings cannot be placed confidently within a complete family of statistical tests.

The record describes the primary analysis population as the intent-to-treat population, then notes that 82 subjects on sunitinib and 67 on placebo who experienced progression during the blinded phase were included in the TTP analysis. It also gives outcome group sizes of 207 and 105. The registry does not explain the relationship among these figures in enough detail to reconstruct all analysis populations, nor does it report a participant flow diagram or full reasons for exclusions from particular analyses.

Missing-data procedures are not reported in the registry record. For a time-to-event endpoint, the handling of incomplete follow-up, missed imaging, withdrawal, and censoring can matter. The record does not state how censoring was defined, whether sensitivity analyses were performed, or how missing tumor assessments were treated. It also does not report whether an interim analysis occurred or whether stopping rules were used.

These gaps do not negate the posted estimates, but they limit independent assessment of their robustness and confirmatory status. The registry record is not a substitute for the complete protocol, statistical analysis plan, or participant-level data. Without those materials, the full multiplicity, missing-data, and analysis-population decisions remain uncertain.

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

The posted primary result describes a difference in the time-to-tumor-progression distributions for the double-blind sunitinib and placebo groups. The reported hazard ratio is 0.329, with a 95% confidence interval of 0.233 to 0.466 and p < 0.001. The group-specific median TTP estimates are 27.3 weeks and 6.4 weeks, respectively. Taken together, these summaries describe the observed comparison under the registry’s reported analysis.

The result does not mean that every participant experienced the group median, nor does the hazard ratio give an individual participant’s chance of progression. It does not quantify a difference in overall survival. The posted overall-survival analysis has a different endpoint and reports a hazard ratio of 0.876, with a 95% confidence interval of 0.679 to 1.129 and p = 0.306. Crossover is relevant to interpreting later outcomes, and the record includes a separate rank-preserving structural failure time analysis, but does not report enough details to assess its assumptions fully.

The finding is specific to the population described in the registry: people with malignant GIST whose disease had failed imatinib therapy or who were intolerant to imatinib. It cannot, on the posted information alone, be generalized to other diseases, patient groups, treatment settings, or comparisons. The record’s crossover designation and the presence of an open-label treatment group also mean that treatment exposure and follow-up beyond the blinded phase require careful interpretation.

Statistical evidence is not the same as a clinical recommendation. The posted analyses do not establish a complete balance of benefits and harms, and no treatment advice follows from these summaries. The registry reports adverse-event counts and selected serious-event terms, but does not provide all exposure-adjusted rates, event definitions, or full safety-analysis details. The result should be understood as an estimate with uncertainty and limitations, rather than as a universal or definitive statement.

18. Why This Trial Matters Statistically

This trial is statistically instructive because it combines a randomized, double-blind, placebo-controlled comparison with a time-to-event primary endpoint and a crossover model. TTP is not simply a count of participants with progression: it uses event timing and follow-up, making the median and hazard ratio complementary summaries. The registry’s imaging-based definition and independent third-party assessment also illustrate how endpoint measurement procedures form part of the statistical interpretation.

The result highlights why effect estimates, confidence intervals, and p-values answer different questions. The hazard ratio communicates a relative event-rate comparison, its confidence interval conveys estimation uncertainty, and the log-rank p-value summarizes evidence against a null comparison under that test. None alone conveys every aspect of the clinical meaning or the quality of the analysis.

Crossover gives the secondary overall-survival analyses additional complexity. The registry posts a rank-preserving structural failure time analysis, which makes explicit that treatment switching may affect interpretation after the blinded phase. Yet method labels alone cannot establish whether assumptions were appropriate; those details are not reported in the registry record.

Finally, the record demonstrates the importance of separating the primary endpoint from numerous secondary outcomes and understanding the role of multiplicity. The primary TTP result is clear in the reported analysis, but incomplete information about the testing strategy, missing-data procedures, and analysis populations limits a full assessment of robustness. Statistical interpretation therefore requires both attention to the estimates and careful recognition of what the record does not describe.

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 leaves important details of the trial design and analysis unspecified. Although the study is identified as a crossover model, it does not describe the crossover sequence, timing, or rules. The arm descriptions are blank, and the treatment schedule and allocation ratio are not reported in the registry record. These omissions restrict understanding of treatment exposure and how participants moved through the study.

The primary outcome information includes group denominators of 207 and 105 and states that 82 and 67 subjects with progression during the blinded phase were included in the TTP analysis. The record does not fully explain the analysis population flow or provide the underlying event and censoring data. It is therefore not possible to independently reconstruct the time-to-event analysis or assess how participants without documented progression contributed.

The full protocol, statistical analysis plan, sample-size justification, interim-analysis procedures, stopping rules, multiplicity strategy, and missing-data methods are not reported in the registry record. The record also does not give detailed checks of model assumptions or sensitivity analyses. These gaps limit assessment of prespecification, robustness, and the confirmatory status of secondary findings.

Safety information includes counts of affected participants and selected serious-event terms, but the record does not provide the at-risk denominators for those group summaries or enough detail to compare exposure-adjusted rates. The existence of an open-label sunitinib group further means that its safety observations should not be interpreted as part of the double-blind randomized comparison without additional information. These record-level limitations qualify interpretation; they do not themselves determine the direction or validity of the reported primary estimate.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: A Study To Assess The Safety And Efficacy Of SU11248 In Patients With Gastrointestinal Stromal Tumor(GIST) — NCT00075218.
• Registry record last updated 2009-09-28; status last verified 2009-08.
• Results first posted 2009-09-28.
• 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 record describes a completed, randomized, double-blind, placebo-controlled Phase III study with a crossover model in people with malignant gastrointestinal stromal tumor resistant or intolerant to imatinib. Enrollment was 361, and Pfizer is listed as the sponsor. The primary outcome was time to tumor progression by imaging during the double-blind treatment phase.

The reported primary comparison was a log-rank hazard ratio of 0.329, with a 95% confidence interval of 0.233 to 0.466 and p < 0.001. Median TTP was 27.3 weeks for double-blind sunitinib and 6.4 weeks for placebo. Secondary results include progression-free survival and overall survival, with crossover-related analysis also posted. The registry does not report the full statistical plan, multiplicity approach, or detailed missing-data procedures, limiting assessment beyond the posted estimates.