Sorafenib With or Without Gemcitabine in Treating Patients With Metastatic Pancreatic Cancer: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Sorafenib With or Without Gemcitabine in Treating Patients With Metastatic Pancreatic Cancer (NCT00114244), a randomized phase 2 study. The primary outcome is Objective Response (OR = CR or PR) as Determined by the RECIST Criteria.

2. Clinical Question

The registry record describes a randomized phase II study of sorafenib with or without gemcitabine in patients with metastatic pancreatic cancer. The primary question was whether the combination arm had a different frequency of objective response from the sorafenib-alone arm, using response defined by RECIST v1.0. The posted comparison is based on the number of participants with complete or partial response.

The posted result contains 0 responses among 15 participants in the sorafenib-alone arm and 1 among 37 participants in the combination arm. The estimated risk difference is 0.027, with a confidence interval from -0.025 to 0.079 and p = 0.52. These are sparse response data: the comparison estimates a difference, but the small number of responses leaves considerable uncertainty about its size and direction.

3. Trial Design

This was a randomized phase II study with no masking and a crossover model. The registry lists enrollment of 52 and a completed status. The sponsor was the National Cancer Institute. The recorded start was 2004-12 and primary completion was 2010-09.

Arm I received sorafenib tosylate, described as 400 mg orally twice daily on days 1-28. Participants experiencing disease progression could cross over to Arm II. Arm II received sorafenib as in Arm I plus gemcitabine hydrochloride, described as 1000 mg/m2 intravenously over 100 minutes on days 1, 8, and 15. The record therefore describes an initial comparison of sorafenib alone with the combination, while also allowing treatment to change after progression in Arm I.

The primary outcome was objective response, assessed every 6 weeks. The response definition was based on RECIST v1.0 and target lesions assessed by CT scan, MRI, or X-ray. Overall survival and progression-free survival are listed as secondary outcomes, but their results are not included in the posted outcome data.

The registry record does not report the randomization procedure, allocation concealment, eligibility criteria beyond the listed condition, planned analysis population, treatment exposure duration, or follow-up schedule for the secondary outcomes. It also does not report how crossover was handled in the primary analysis. Those details matter when interpreting a randomized comparison, particularly when participants in one arm may later receive the other arm’s treatment. The record identifies the design features but does not give enough detail to reconstruct the full protocol or its analysis plan.

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:

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 Objective Response (OR = CR or PR) as Determined by the RECIST Criteria, measured over Every 6 weeks..

The primary endpoint is objective response, defined as complete response or partial response under RECIST v1.0. A complete response means disappearance of all target lesions. A partial response is defined in the registry as a >=30% decrease in the sum of the longest diameters of target lesions. The record says that target lesions were assessed by CT scan, MRI, or X-ray, with assessment every 6 weeks.

Objective response is a binary outcome at the participant level for this comparison: each participant is counted according to whether a complete or partial response was recorded. This makes the endpoint directly interpretable as the frequency of participants meeting the response criteria in each arm. It does not describe the full course of disease, the duration of a response, or the time until progression or death. Those are distinct questions, and the registry lists progression-free survival and overall survival as secondary outcomes.

Response criteria provide a common framework for classifying changes in measurable target lesions. Their use can make outcome assessment more consistent, but the registry record does not report whether response assessment was blinded or independently reviewed. With only 1 posted response across the arms, the response comparison is especially sensitive to the limited number of observed events.

7. Results Posted in the Registry

• Arm II (Sorafenib Tosylate, Gemcitabine Hydrochloride): 2.7% (1/37) versus 0.0% (0/15) on control; risk difference 2.7 percentage points (95% CI -2.5 to 7.9); unadjusted two-proportion z-test p = 0.520.

8. Statistical Interpretation

9. Statistical Methods Explained

The central statistic is a risk difference: the response frequency in the combination arm minus the response frequency in the sorafenib-alone arm. A risk difference expresses an absolute contrast between two proportions, rather than a relative contrast. The registry reports an estimate of 0.027. The calculation method for that estimate is not reported in the registry record.

What does the confidence interval describe?

In general, a confidence interval represents the uncertainty associated with an estimated effect under a specified statistical procedure. A narrow interval indicates greater precision under that procedure, while a wide interval indicates less precision. Here, the posted interval runs from -0.025 to 0.079. It includes zero, so the data do not isolate a difference from no difference on the risk-difference scale. The confidence level and interval construction method are not reported in the registry record, so those features cannot be assessed from the posted information.

How should the p-value be read?

In general, a p-value summarizes how unusual data at least as incompatible with a null hypothesis would be under that hypothesis and the assumptions of the chosen test. It is not a measure of effect size, and it does not give the probability that the null hypothesis is true. The posted value is p = 0.52. The test, its assumptions, and whether the reported value is one-sided or two-sided are not reported in the registry record. The study’s planning illustration refers to a two-sided significance level of 0.05, but that planning assumption alone does not establish the analysis threshold used for this posted comparison.

Why do sparse outcomes matter?

Only 1 objective response was recorded across both arms. With so few observed responses, a small change in the event count could substantially alter the estimated arm contrast. This makes the estimate unstable and its uncertainty important to interpretation. The interval reflects uncertainty according to its calculation method; it does not account automatically for missing data, crossover, measurement differences, or other sources of bias.

For binary outcomes, the number of participants with an event and the number assessed in each arm are necessary to understand the observed proportions. The record lists 0 of 15 in Arm I and 1 of 37 in Arm II. The analysis population and handling of participants without an evaluable response are not reported in the registry record. It is therefore unclear whether the denominators represent all randomized participants, treated participants, or another population.

What does crossover change?

In general, crossover can make a comparison based on original assignment different from a comparison based on treatment actually received. If participants assigned to one arm later receive the other intervention, the analysis needs to specify how their outcomes are attributed. The registry says participants in Arm I experiencing disease progression cross over to Arm II, but it does not report the number who crossed over or the analytic approach. The posted arm counts should therefore be read alongside this design feature, without assuming a particular crossover adjustment.

The registry identifies the study as randomized, but the randomization procedure and allocation concealment are not reported. Randomization is generally intended to make baseline characteristics comparable across groups on average; it does not guarantee that every characteristic is balanced, particularly in a small study. Baseline characteristics are not included in the posted outcome record, so their distribution cannot be assessed.

What can planning calculations tell us?

In general, sample-size planning asks how many participants or events might be needed to detect a chosen effect under specified assumptions. Any planning illustration based on the observed effect is conditional on those assumptions; it is not observed power and does not describe the trial’s actual design justification. The registry record does not report the trial’s sample-size justification. The planning assumptions described for the illustration include a two-sided significance level of 0.05, equal allocation, and a normal approximation, with the effect framed as a fraction of the observed effect. Such illustrations should not be mistaken for evidence that the trial was adequately powered for its actual objective.

Over-dispersion is a general concern when data vary more than a chosen model expects, which can make model-based uncertainty too small if the extra variation is ignored. The posted primary outcome is a sparse binary count comparison, and the registry does not report a model-based analysis or an over-dispersion assessment. It would be inappropriate to infer that such a model was used.

The record contains no formal analysis entries for the posted outcome comparison. It reports a risk difference, interval, and p-value, but not the statistical test, adjustment variables, missing-data procedure, or prespecified analysis population. These omissions limit reproducibility and make it difficult to judge whether the reported estimate follows a prespecified plan.

10. Confidence Intervals and Precision

The estimated risk difference is 0.027, with the confidence interval extending from -0.025 to 0.079. The interval includes zero and includes estimates in both directions. Its span signals that the observed counts do not pin down the size or direction of the underlying contrast precisely. The p-value is 0.52, which is consistent with the lack of a sharply distinguished difference in this sparse comparison.

Precision is distinct from whether an estimate is clinically important. A confidence interval describes statistical uncertainty under the procedure used to calculate it; it does not identify which effects matter to patients or clinicians. The registry record does not state a threshold for a clinically important response difference, so the interval cannot be classified against such a threshold from the posted information.

The interval’s confidence level and construction method are not reported in the registry record. It is also not stated whether the estimate accounts for crossover, missing outcome assessments, or any covariates. The endpoints of the interval should therefore be interpreted as the reported uncertainty interval, without attributing a particular calculation method or coverage level to it.

The limited number of responses is a key reason to avoid overinterpreting the point estimate. The interval is not proof that the true difference is exactly zero, and the positive point estimate is not proof of a reliable advantage. Rather, these posted data leave substantial uncertainty. More complete interpretation would require the prespecified analysis plan, the analysis population, response-assessment completeness, and the handling of crossover.

11. Multiplicity, Missing Data and Analysis Populations

The registry lists overall survival and progression-free survival as secondary outcomes in addition to objective response. It does not post their outcome estimates or analysis results in this record. The relationship between the primary and secondary outcomes, including any hierarchy or confirmatory role, is not reported.

When a study tests multiple outcomes or performs multiple comparisons, the chance of at least one apparently unusual result can increase. In general, multiplicity procedures can control that issue by adjusting thresholds or using a prespecified testing sequence. The registry record does not report whether multiplicity was addressed, whether there were interim analyses, or whether the primary response comparison was part of a broader testing strategy. No adjustment should be assumed.

Missing-data handling is also not reported. The posted denominators are 15 and 37, but the record does not state whether all randomized participants had an evaluable response, whether anyone was excluded from the primary analysis, or what rule was used for participants with missing assessments. It does not report response-assessment completion by arm. These details are important for a binary outcome because the denominator determines the observed response frequency.

The analysis population is not reported in the registry record. It is therefore not possible to determine whether the reported groups reflect all randomized participants, participants who received assigned treatment, or another defined population. The record also does not describe protocol deviations or how crossover participants were analyzed. These are not minor technical details: they influence what question the posted estimate answers.

The registry does not report interim monitoring, stopping rules, or a detailed statistical analysis plan. Without those details, readers cannot establish whether the reported comparison was prespecified in its current form or whether other analyses were conducted but not posted. The absence of these details is a limitation of the registry record, not evidence that a particular procedure was or was not used.

12. 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 0% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: thrombosis (26.7%), disease progression (13.3%), vascular access complication (8.1%), sinus tachycardia (6.7%), abdominal pain (6.7%).

The registry reports serious adverse events among 8 of 15 participants in Arm I and 15 of 37 participants in Arm II. It also reports other adverse events among 15 of 15 participants in Arm I and 37 of 37 participants in Arm II. These counts describe participants recorded with events in each category; they do not establish that an event was caused by study treatment.

The listed serious adverse-event terms and percentages are thrombosis, 26.7%; disease progression, 13.3%; vascular access complication, 8.1%; sinus tachycardia, 6.7%; abdominal pain, 6.7%; and constipation, 6.7%. The registry record does not specify the denominator or observation window for these term percentages in the posted information. Disease progression is included among the listed serious terms, so the list should not be read as a list of treatment toxicities alone.

The record does not report death counts. It also does not provide enough information to determine event timing, duration, recurrence, severity grading beyond the serious-event classification, or attribution to study treatment. Differences in counts between arms should not be interpreted as comparative safety effects without those details and an appropriate analysis.

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

The posted primary result answers a limited question: how many participants in each recorded arm had an objective response under the stated RECIST criteria, and what risk-difference estimate was reported. The counts are 0 of 15 in the sorafenib-alone arm and 1 of 37 in the combination arm. The estimate is 0.027, with an interval from -0.025 to 0.079 and p = 0.52.

Those figures do not establish that the treatments have the same response frequency. The interval includes zero, but it also allows nonzero differences in either direction. Nor does the positive point estimate establish that adding gemcitabine produces a reliable increase in response. The limited event count leaves the estimate imprecise.

The result also does not answer whether any response lasted, whether progression-free survival or overall survival differed, or how quality of life or other outcomes compared. Although the registry lists overall survival and progression-free survival as secondary outcomes, outcome estimates for them are not reported in the posted results. The response endpoint is not interchangeable with these time-to-event outcomes.

The crossover design further limits a simple interpretation of the arms as sustained, distinct treatment strategies. Participants in Arm I could move to Arm II after disease progression, but the record does not describe crossover frequency or analysis handling. A comparison by initial assignment, a comparison by treatment received, and an analysis accounting for crossover can address different questions; the record does not identify which approach underlies the posted result.

These data are therefore best understood as a sparse, registry-posted response comparison with substantial uncertainty and incomplete analysis detail. They do not justify a treatment recommendation or a conclusion about efficacy or safety beyond the specific counts and estimates reported.

14. Why This Trial Matters Statistically

This trial is statistically informative as an example of how a randomized comparison can still yield a highly uncertain estimate when the primary outcome is rare. The point estimate alone can appear to tell a simple story, but the interval and the observed counts show why precision must accompany interpretation. Here, one response across both arms is the basis for a risk-difference estimate with a confidence interval that includes zero and extends in both directions.

The design also illustrates the importance of distinguishing randomized assignment from treatment received. The registry describes crossover from sorafenib alone to the combination after disease progression. This feature may be relevant to how outcomes are attributed, yet the record does not say how crossover was handled in the primary comparison. Without that information, readers should avoid assuming that arm labels correspond to uninterrupted exposure.

Another statistical lesson is that a posted p-value is not a substitute for effect size, uncertainty, or analysis transparency. The risk difference and interval convey information that the p-value alone cannot. At the same time, missing details about the test, analysis population, missing outcomes, and multiplicity restrict independent assessment of the result.

The registry’s secondary-outcome listings also show why endpoints should be read separately. Objective response, progression-free survival, and overall survival address different aspects of outcome. The record lists the latter two but does not post their results in this outcome record, so the primary response comparison cannot stand in for them.

For biostatistical interpretation, the key lesson is to connect the estimate to its event counts, design features, uncertainty, and reporting limits. The evidence in the registry is a description of observed responses and an imprecise contrast, not a complete account of every outcome or analysis in the trial.

15. 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 primary outcome result is based on very few responses, with 0 recorded in Arm I and 1 in Arm II. This makes the estimated contrast sensitive to small changes in event classification or assessment completeness. The registry record does not report whether any responses were confirmed, whether assessment was independently reviewed, or how missing assessments were classified.

The study permitted crossover after disease progression in Arm I, but crossover counts and analytic handling are not reported. As a result, the posted comparison cannot be fully interpreted as either a comparison by initial assignment or by treatment actually received. The registry also does not report the randomization method, allocation concealment, baseline characteristics, or analysis population.

Although overall survival and progression-free survival are listed as secondary outcomes, their results are not reported in the posted outcome data. The registry record does not provide a detailed statistical analysis plan, test specification, confidence-interval method or confidence level, missing-data procedure, multiplicity strategy, interim-analysis information, or sample-size justification. These gaps limit the ability to verify the analysis and understand whether it matched the planned approach.

Safety interpretation has similar constraints. Serious adverse-event counts and selected terms are reported, but attribution, timing, observation windows, and the denominators for the term percentages are not specified in the posted information. Death counts are not reported in the registry record. Consequently, the adverse-event listings describe recorded events without providing a complete comparative safety analysis.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Sorafenib With or Without Gemcitabine in Treating Patients With Metastatic Pancreatic Cancer — NCT00114244.
• Registry record last updated 2015-01-14; status last verified 2013-10.
• Results first posted 2015-01-14.
• 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.

19. Record Summary

NCT00114244 is a completed randomized phase II study sponsored by the National Cancer Institute. It enrolled 52 participants with metastatic pancreatic cancer and compared sorafenib alone with sorafenib plus gemcitabine, with crossover from Arm I to Arm II permitted after disease progression.

The primary outcome was objective response by RECIST v1.0, assessed every 6 weeks. The registry reports 0 responses among 15 participants in Arm I and 1 among 37 in Arm II. The risk difference is 0.027, with an interval from -0.025 to 0.079 and p = 0.52. The estimate is imprecise, and important details about the analysis population, missing data, crossover handling, and statistical methods are not reported in the registry record.