A Dose-Escalation to Rash Study of Tarceva (Erlotinib) Plus Gemcitabine in Patients With Metastatic Pancreatic Cancer: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Dose-Escalation to Rash Study of Tarceva (Erlotinib) Plus Gemcitabine in Patients With Metastatic Pancreatic Cancer (NCT00652366), a randomized phase 2 study. The primary outcome is Percentage of Participants Who Died Assessed From Point of Randomization.

2. Clinical Question

3. Trial Design

This was a phase 2, randomized, open-label, parallel-group study of erlotinib plus gemcitabine in metastatic pancreatic cancer. The registry lists enrollment as 467. The study began in 2008-05, had primary completion in 2012-02, and is marked completed. The registry record does not report the original sample-size justification or the assumptions used to plan the trial.

All participants first received a 4-week run-in of erlotinib 100 mg/day with gemcitabine. After the run-in, participants who had no rash or only grade 1 rash were randomized to the standard-dose or escalating-dose erlotinib group. The escalating regimen began at 150 mg/day, with increases of 50 mg every 2 weeks up to 250 mg/day, stopping escalation at grade 2 rash or another dose-limiting toxicity. The standard regimen continued erlotinib at 100 mg/day. Both groups received gemcitabine on the schedule described in the registry.

The primary endpoint population is the full analysis set, defined as all randomized participants. The posted arm sizes for that endpoint are 75 in the standard-dose group and 70 in the escalating-dose group. These are the randomized outcome groups represented in the primary comparison; the enrollment count is not the denominator for that comparison. The registry also reports adverse-event groups that include participants who were not randomized into those two comparison groups, so those groups should not be read as equivalent randomized arms.

The study was open label, meaning treatment assignment was not masked. The registry record does not report additional details about concealment of allocation, site structure, or the original statistical design assumptions.

4. Participant Flow and Accounting

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

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Full analysis set (FAS): all randomized participants.

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.

Population: Safety analysis population (SAP): all participants who received at least 1 dose of the trial medication and had at least 1 safety follow-up, whether withdrawn prematurely or not.

6. Primary Endpoint

The registry lists the primary outcome as Percentage of Participants Who Died Assessed From Point of Randomization, measured over Randomization [Day 1 of Cycle 2 (4-week cycles)] and weekly thereafter for up to 46 months..

The primary endpoint is the percentage of participants who died, assessed from randomization. The registry defines overall survival as the time from randomization to death from any cause. Participants who remained alive at analysis were censored at the date they were last known to be alive. The stated assessment period began at randomization, which was Day 1 of Cycle 2, and continued weekly for up to 46 months.

A percentage of participants who died is an event proportion: it describes how many participants in each group had experienced death by the relevant analysis, relative to the group denominator. It is not itself a median survival time, and it does not show when deaths occurred. Time-to-event methods can use the timing of events and account for censoring; the registry’s endpoint definition includes this follow-up information, while the posted primary outcome is expressed as a percentage.

Overall survival is used as an outcome because it records death from any cause and does not depend on a judgment about whether a particular change in disease status counts as progression or response. Interpreting a death proportion still depends on follow-up duration and completeness. The posted primary comparison gives the observed proportions and an unadjusted risk difference with a confidence interval. The registry record does not report a primary-endpoint confidence interval for each arm separately or a reported sponsor analysis for this endpoint.

7. Results Posted in the Registry

• G+E Escalating Dose: Rash Grade < 2: 85.7% (60/70) versus 81.3% (61/75) on control; risk difference 4.4 percentage points (95% CI -7.7 to 16.4), risk ratio 1.05 (0.91–1.22); unadjusted two-proportion z-test p = 0.478.

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.

Percentage of Participants With Disease Progression or Death as Assessed From Point of Randomization

• 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:
  • G+E Escalating Dose: Rash Grade < 2: 88.6% (62/70) versus 90.7% (68/75) on control; risk difference -2.1 percentage points (95% CI -12.0 to 7.8), risk ratio 0.98 (0.87–1.09); unadjusted two-proportion z-test p = 0.679.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between G+E Escalating Dose: Rash Grade < 2 and the control. The interval gives the range of effects that remain compatible with the data.

PFS Assessed From Point of Randomization

• 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:
  • G+E Standard Dose: Rash Grade < 2 vs G+E Escalating Dose: Rash Grade < 2: hazard ratio 1.09 (95% CI 0.77–1.54), reported p = 0.630 (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 G+E Standard Dose: Rash Grade < 2 vs G+E Escalating Dose: Rash Grade < 2 and the control. The interval gives the range of effects that remain compatible with the data.

Percentage of Participants With a Best Overall Response (BOR) of Confirmed Complete Response (CR) or Partial Response (PR) According to RECIST

• 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:
  • G+E Escalating Dose: Rash Grade < 2: 8.6% (6/70) versus 14.7% (11/75) on control; risk difference -6.1 percentage points (95% CI -16.4 to 4.3), risk ratio 0.58 (0.23–1.50); unadjusted two-proportion z-test p = 0.254.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between G+E Escalating Dose: Rash Grade < 2 and the control. The interval gives the range of effects that remain compatible with the data.

The secondary outcomes extend the primary question in several directions: progression or death, progression-free survival, confirmed tumor response, disease-status categories, and outcomes measured from the start of the run-in as well as from randomization. Together, they give a broader description of disease events and response, but they do not all measure the same clinical process or use the same time origin. In particular, outcomes beginning at randomization apply to the post-run-in randomized population, while outcomes beginning at run-in start use a different point in participants’ course.

Across the posted comparisons, the estimates do not form a consistently precise pattern that clearly separates the randomized groups. The time-to-event comparison and the response comparison both have uncertainty that includes their respective null values. Progression-or-death proportions are also posted, but the registry record does not list reported analyses for that endpoint. This overall pattern is broadly consistent with the primary result’s uncertainty rather than supplying a clear, independent resolution of it.

Secondary outcomes can be useful for understanding whether the primary result aligns with other measures, and for describing outcomes not captured by the primary endpoint. However, testing many outcomes increases the chance that some may appear notable by chance. The registry record does not report a multiplicity adjustment or a hierarchy that would establish which secondary comparisons were confirmatory. Unadjusted secondary results therefore carry less weight than a prespecified primary analysis and should be interpreted cautiously, especially when their confidence intervals are broad.

These outcomes should not be combined informally into a single verdict. Differences in endpoint definition, follow-up origin, and statistical scale matter. A response proportion, a progression-free survival hazard ratio, and a death proportion answer related but distinct questions. Their role is to add context to the primary comparison, not to override it.

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

• Percentage of Participants With Disease Progression or Death as Assessed From Point of Randomization
• PFS Assessed From Point of Randomization
• Percentage of Participants With a Best Overall Response (BOR) of Confirmed Complete Response (CR) or Partial Response (PR) According to RECIST
• Percentage of Participants With a CR, PR, Stable Disease (SD), or PD According to RECIST
• Percentage of Participants With SD (Maintained for at Least 8 Weeks) or CR or PR (Maintained for at Least 4 Weeks) According to RECIST
• Percentage of Participants Who Died as Assessed From Start of 4-Week Run-In
• OS Assessed From Start of 4-Week Run-In
• Percentage of Participants With Disease Progression or Death as Assessed From the Start of 4-Week Run-In

9. Statistical Interpretation

10. Statistical Methods Explained

What does a risk difference describe?

A risk difference subtracts the event proportion in one group from the event proportion in the comparison group. In this study’s primary comparison, the posted estimate is 0.044 for escalating dose relative to standard dose. A positive estimate indicates a higher observed event proportion in the escalating-dose group under that direction of subtraction; it is not a statement about when events happened. The confidence interval expresses uncertainty in this estimated difference. Because its limits extend to both sides of no difference, the data are compatible with more than one direction of effect.

How is a confidence interval used?

A confidence interval is a range of values that reflects the statistical uncertainty around an estimate under the method used. It is not a guarantee that the true effect lies within those particular limits, nor does it measure the probability that a treatment is better. When an interval includes the null value, the data do not distinguish the estimate from no difference at the corresponding level of statistical compatibility. The breadth of the interval also matters: a wide interval leaves more plausible values unresolved than a narrow one.

What does the p-value add?

A p-value describes how unusual results at least as incompatible with a specified null hypothesis would be under that hypothesis and the assumptions of the test. It is not the probability that the null hypothesis is true, and it does not measure the size or clinical importance of an effect. The primary comparison’s unadjusted p-value is 0.478. Read alongside the interval, it does not provide strong evidence against no difference. The p-value cannot replace attention to the estimate, its uncertainty, study population, and endpoint definition.

Why are time-to-event methods different?

Progression-free survival is reported as a median in weeks, with a 95% confidence interval for each arm, and the between-group result is reported as a hazard ratio from a Log Rank analysis. A median time summarizes the point at which the estimated event distribution reaches its midpoint. A hazard ratio compares event rates over follow-up under the time-to-event model; it is not a ratio of medians and should not be interpreted as a direct difference in weeks. The reported hazard ratio is 1.09, with a 95% confidence interval from 0.77 to 1.54. The interval includes 1, the null value for a ratio, and the reported p-value is 0.630.

Censoring occurs when a participant’s event time is not observed during the available follow-up. For overall survival, the registry says participants still alive at analysis were censored at their last-known-alive date. Time-to-event estimates rely on appropriate handling of censoring; if follow-up differs in ways related to outcome, interpretation can be affected. The registry record does not report further details of censoring patterns or sensitivity analyses.

What does the open-label design imply?

In an open-label trial, participants and study personnel know the assigned treatment. Knowledge of assignment can influence care decisions, reporting, or outcome assessment, especially for outcomes requiring judgment. Death is less dependent on subjective assessment than many clinical outcomes, but open-label status remains relevant to the overall study context. The registry record does not report whether outcome assessors were separately blinded.

Why can over-dispersion matter?

General: for count outcomes, over-dispersion means the observed variability is greater than a model’s assumed variability. If a model does not allow for that extra variation, standard errors can be too small and confidence intervals too narrow. The primary endpoint is reported as a participant percentage, not a count-rate outcome, and the registry does not report an over-dispersion adjustment. This general issue should not be mistaken for a reported feature of this trial’s primary analysis.

How should the secondary response analysis be read?

The response endpoint uses confirmed complete or partial response according to RECIST. The registry includes a reported difference in response rates with a 95% confidence interval and a chi-squared p-value of 0.254, as well as a reported odds ratio. These are distinct summaries: a difference in rates is on an additive scale, whereas an odds ratio compares odds. Neither should be treated as a direct measure of survival. The registry record does not report an adjustment for testing multiple secondary outcomes.

11. Confidence Intervals and Precision

The primary risk-difference estimate is 0.044, with a 95% confidence interval from -0.077 to 0.164. This range includes zero, so the result is compatible with no difference as well as with effects in either direction. Its width signals limited precision: the estimate alone should not be read as a definitive measure of how far apart the groups are. The posted p-value of 0.478 is consistent with the interval’s failure to exclude the null value.

For progression-free survival, the standard-dose median is 19.4 weeks, with a 95% confidence interval from 16.0 to 26.9 weeks; the escalating-dose median is 15.3 weeks, with a 95% confidence interval from 11.3 to 19.1 weeks. The reported hazard ratio is 1.09, with a 95% confidence interval from 0.77 to 1.54. The interval includes the null ratio, and the reported Log Rank p-value is 0.630. These estimates do not provide a precise separation between groups.

The confirmed-response percentages also have broad arm-specific intervals: 14.7% with a 95% confidence interval from 7.6% to 24.7% for standard dose, and 8.6% with a 95% confidence interval from 3.2% to 17.7% for escalating dose. The reported difference in response rates has a 95% confidence interval from -17.2 to 5.0, with a reported p-value of 0.254. The registry reports these values at their posted precision; greater numerical detail would imply a level of precision not recorded.

Confidence intervals describe statistical uncertainty, not every source of uncertainty. They do not resolve potential limitations from eligibility after run-in, open-label conduct, missing outcome information, or multiplicity. The registry record does not report enough information to quantify those additional sources.

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 are based on the observed effect, not on the trial’s original design. For binary outcomes, the figures are participants per arm. For time-to-event outcomes, the figures are total events, not participants per arm. The progression-free survival planning illustration is therefore an event target, whereas the primary death-proportion illustration is expressed as participants per arm.

The stated 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 set to the stated fraction of the observed effect. The 80% power column describes a planning scenario with an 80% chance of detecting an effect of that specified size if the assumptions hold and the effect is real. The 90% power column uses a higher target chance of detection under the same kind of assumptions. These are design-planning quantities, not probabilities that the posted result is correct.

For the primary binary outcome, the planning entries are 1,125 participants per arm for 80% power and 1,505 for 90% power when planning for 100% of the observed effect. Planning for 75% of the observed effect gives 2,053 and 2,748 participants per arm; planning for 50% gives 4,738 and 6,342 participants per arm. The other endpoint illustrations have their own values, which should be read according to whether they count participants per arm or total events.

Smaller effects are harder to distinguish from random variation, so planning to detect a smaller fraction of the observed effect requires more participants or events. These illustrations are not observed power. They do not indicate the probability that the trial detected an effect after seeing its results, and they do not establish the assumptions used in the original trial design. The registry record does not report that original design justification.

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

n is the number of participants in each arm, p₀ and p₁ are the control and treatment proportions, and p̄ is their average.

Worked calculation for this trial

1. Control proportion p₀ = 0.813; treatment proportion p₁ = 0.857; average p̄ = 0.835.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.5246 = 1.0282.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.5237 = 0.4408.
4. n = (1.0282 + 0.4408)² / (0.0438)² = 1,124.4, rounded up to 1,125 participants per arm 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
p0 <- 0.813333333333333                          # control proportion
p1 <- 0.857142857142857                          # treatment proportion
n_per_arm <- function(power, frac = 1) {
  q1 <- p0 + (p1 - p0) * frac               # smaller effect = fraction of the observed difference
  pbar <- (p0 + q1) / 2
  z_b <- qnorm(power)
  ceiling(round((z_a * sqrt(2 * pbar * (1 - pbar)) + z_b * sqrt(p0 * (1 - p0) + q1 * (1 - q1)))^2 / (q1 - p0)^2, 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

# Check against R's built-in function (unrounded n per arm)
cat("power.prop.test:", round(power.prop.test(p1 = p0, p2 = p1, power = 0.80)$n, 1), "\n")

# Simulation: how often does the z-test reject at the 80%-power sample size?
set.seed(652366)
n    <- n_per_arm(0.80)
reps <- 20000
x0 <- rbinom(reps, n, p0); x1 <- rbinom(reps, n, p1)
ph <- (x0 + x1) / (2 * n)
z  <- ((x1 - x0) / n) / sqrt(ph * (1 - ph) * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(z) > z_a, na.rm = TRUE), reps))

Output from running the code

100% of observed effect: n per arm = 1,125 (80% power), 1,505 (90% power)
 75% of observed effect: n per arm = 2,053 (80% power), 2,748 (90% power)
 50% of observed effect: n per arm = 4,738 (80% power), 6,342 (90% power)
power.prop.test: 1124.4 
Simulated power at n = 1,125 per arm: 0.803 (20000 replicates)

The simulation draws many pairs of event counts at the planned sample size and counts how often the two-proportion z-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

14. Multiplicity, Missing Data and Analysis Populations

The registry record lists a primary endpoint and multiple secondary outcomes, but a multiplicity strategy is not reported in the registry record. It is therefore not possible to determine from the record whether the original analysis controlled the overall chance of false-positive findings across secondary endpoints, prespecified a testing hierarchy, or treated those analyses as exploratory. Unadjusted p-values for secondary comparisons do not account for the number of outcomes examined.

The primary population is defined as the full analysis set, all randomized participants. The primary endpoint description says participants still alive at analysis were censored at the last date they were known to be alive. That provides the stated handling of surviving participants for the time-to-death definition, but other details about missing observations, withdrawals, or sensitivity analyses are not reported in the registry record. The handling of missing response assessments and disease-status data is also not reported in the registry record.

The randomization occurred only after the 4-week run-in and only among participants with no rash or grade 1 rash. This creates a defined population for the randomized comparison, but the trial’s reported primary result should not be generalized automatically to participants who were not eligible for randomization. The registry includes additional adverse-event categories outside the two randomized outcome groups; their existence does not change the primary analysis population.

The registry record does not report interim analyses, stopping rules, or a detailed statistical analysis plan. It also does not report the original sample-size justification. Without those details, the posted results can be interpreted using the outcomes and comparisons recorded, but it is not possible to assess whether all analysis choices were prespecified or how missing information was addressed beyond the stated censoring definition.

15. Safety Results

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

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: abdominal pain (5.6%), anaemia (5.6%), pyrexia (5.6%), cerebrovascular accident (5.6%), diarrhoea (5.2%).

The serious adverse-event figures are reported across five registry groups, including the two rash-eligible dose groups and groups defined by rash status or early dropout. These categories are not all randomized treatment arms and should not be treated as a direct randomized safety comparison. The serious adverse-event counts are 39 of 105 at risk in the group labeled rash at least grade 2, 24 of 77 in the standard-dose rash-below-grade-2 group, 20 of 71 in the escalating-dose rash-below-grade-2 group, 64 of 178 in the no-rash non-eligible group, and 22 of 36 in the early-dropout group.

The registry lists abdominal pain, anaemia, pyrexia, and cerebrovascular accident at 5.6% each among the top serious terms, followed by diarrhoea at 5.2% and pneumonia at 3.4%. These figures describe reported terms in the safety record; they do not establish that treatment caused an event. The registry record does not report deaths in the adverse-event group summary. Differences in group eligibility, follow-up, and denominators limit direct comparisons across these categories.

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

The posted primary result describes death proportions after randomization among participants who completed the run-in and met the rash eligibility condition. It shows observed percentages of 81.3% and 85.7%, with a risk-difference confidence interval that includes zero. The result does not establish a clear difference between the groups, but it also does not demonstrate that the regimens are equivalent. The interval remains compatible with effects in either direction.

The findings concern erlotinib dose strategies given with gemcitabine in this particular randomized population. They do not answer whether either regimen is preferable for every person with metastatic pancreatic cancer, including people who did not meet the post-run-in criteria. Nor do the posted percentages alone describe the timing of deaths. The separate progression-free survival analysis provides time-to-event information, but its reported estimate also has substantial uncertainty.

Secondary response and progression outcomes can help characterize the study, yet they have distinct definitions and are subject to the limitations of multiple unadjusted comparisons. The record does not report a multiplicity plan or comprehensive missing-data procedures. Those omissions limit how confidently secondary signals can be treated as confirmatory evidence.

Safety figures describe reported adverse events across groups that do not all correspond to randomized arms. They are useful for understanding what was recorded, but they cannot be used as a simple causal comparison between the two assigned dose strategies. Overall, the registry results are descriptive and comparative evidence for the enrolled study population, with uncertainty that should remain visible in any interpretation.

17. Why This Trial Matters Statistically

This trial raises a statistical design issue that is specific to its treatment strategy: randomization followed a run-in, and eligibility for randomization depended on rash status. Consequently, the randomized comparison estimates outcomes for a selected group that remained eligible after initial treatment, rather than for everyone who entered the study. Recognizing that distinction is essential when defining the population to which the comparison applies.

The primary outcome is a death proportion, while the registry also reports time-to-event and tumor-response outcomes. These measures answer related but nonidentical questions. Reading them together requires attention to their time origins, endpoint definitions, and statistical scales rather than treating them as interchangeable confirmations.

The posted primary estimate has a confidence interval that includes both no difference and effects in either direction. This illustrates why statistical interpretation should consider both the estimated effect and its precision, rather than relying on a single p-value or on the direction of the point estimate. The secondary outcomes likewise show why numerous results without a reported multiplicity strategy must be interpreted with restraint.

Finally, safety summaries span randomized and nonrandomized categories. A clear description of denominators and group formation is necessary before making a comparative claim. The registry record allows readers to distinguish the randomized efficacy population from broader safety categories, while leaving important original design and analysis details unreported.

18. 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.
• Participant counts were reconstructed from posted percentages and arm sizes, so small rounding differences are possible.

The primary comparison is based on the randomized participants who were eligible after a 4-week run-in. Participants who developed more than grade 1 rash or otherwise did not enter randomization are outside that primary comparison. This selection limits how directly the results apply to all enrolled participants or to populations not defined by the same eligibility process.

The study was open label, and the registry record does not report whether outcome assessors were blinded. Knowledge of treatment assignment can influence care and assessment, particularly for outcomes involving clinical judgment. The primary outcome of death is less subjective than response assessment, but open-label conduct remains relevant context.

The primary result is posted as an event percentage, while the endpoint is defined using time from randomization to death with censoring for participants last known alive. The registry record does not provide further details about the distribution of follow-up, censoring, or sensitivity analyses. It also does not report the handling of missing assessments for secondary outcomes.

Several secondary outcomes are listed, but a multiplicity strategy, interim analyses, and the original statistical analysis plan are not reported in the registry record. The posted unadjusted comparisons therefore have limited confirmatory weight. The adverse-event groups also include categories that are not all randomized treatment groups, restricting direct safety comparisons. The registry record does not report the original sample-size justification or additional details about allocation procedures.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: A Dose-Escalation to Rash Study of Tarceva (Erlotinib) Plus Gemcitabine in Patients With Metastatic Pancreatic Cancer — NCT00652366.
• Registry record last updated 2015-02-11; status last verified 2015-01.
• Results first posted 2015-02-11.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

22. Record Summary

The registry describes a completed phase 2, randomized, open-label, parallel-group study of erlotinib plus gemcitabine in metastatic pancreatic cancer. After a 4-week run-in, participants meeting the rash eligibility condition were randomized to standard-dose erlotinib or a dose-escalation strategy. The primary outcome was the percentage of randomized participants who died from the point of randomization.

The posted primary comparison reports death in 81.3% of the standard-dose group and 85.7% of the escalating-dose group. The risk difference is 0.044, with a 95% confidence interval from -0.077 to 0.164 and an unadjusted p-value of 0.478. The interval does not establish a clear difference and leaves uncertainty in both directions. Secondary outcomes and safety results add context, but their interpretation is limited by endpoint differences, unadjusted comparisons, and the distinct populations represented in the registry.