Sorafenib Tosylate in Treating Patients With Progressive Metastatic Neuroendocrine Tumors: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Sorafenib Tosylate in Treating Patients With Progressive Metastatic Neuroendocrine Tumors (NCT00131911), a non-randomized phase 2 study. The primary outcome is Confirmed Response Rate.

2. Clinical Question

In plain terms, the trial was set up to describe whether patients in these disease groups experienced a confirmed tumor response while receiving sorafenib. It was a phase II study, not a randomized comparison of sorafenib with another treatment or with no treatment. The endpoint description specifies how response was defined and says the rate was estimated among evaluable patients. It also identifies patients not evaluated on consecutive cycles as excluded from this endpoint. The registry record does not report the outcome values needed to determine the observed response rate or its precision. Consequently, the clinical question can be described from the registry, but its numerical answer cannot be recovered from the posted record.

3. Trial Design

This was a phase II, non-randomized, unmasked, parallel-model trial sponsored by the National Cancer Institute. The registry lists enrollment as 93 and status as completed. The start is recorded as 2005-06 and primary completion as 2010-10. The study concerned progressive metastatic neuroendocrine tumors, including carcinoid and islet cell tumors. The registry lists several related tumor and syndrome conditions.

The two group descriptions distinguish patients by disease category: Group A comprised patients with carcinoid tumors, while Group B comprised patients with islet cell and other neuroendocrine tumors. Both groups are described as receiving 400 mg oral sorafenib twice daily on days 1-28. The record does not describe a separate placebo or active-comparator group. It explicitly identifies the allocation as non-randomized, so group membership should not be treated as if it arose from random assignment. The groups also differ by their described disease categories, which makes a direct comparison especially unsuitable as a controlled treatment contrast.

Masking is listed as none. The record does not report further details about how assessments were masked, how patients entered each group, or whether any other protocol procedures were used to reduce assessment bias. The primary endpoint was confirmed response rate; listed secondary outcomes include toxicity, overall survival, progression-free survival, and duration of response. The registry record lists five outcome measures and no analyses. It does not report a detailed statistical plan, a sample-size justification, or interim analyses. These omissions limit how fully the design and analysis choices can be reconstructed from the registry alone.

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: One carcinoid patient canceled prior to treatment and was excluded from all analyses.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Nine of the 50 carcinoid patients and 6 of the 42 Islet cell patients did not continue treatment past cycle 1. Therefore, these patients were not evaluated on consecutive cycles and were excluded from this endpoint.

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 Confirmed Response Rate, measured over Duration of Treatment (Up to 2 years).

The primary endpoint, confirmed response rate, is the percentage of evaluable participants who had a complete or partial response confirmed on subsequent scans at least 4 weeks apart. The registry defines complete response as disappearance of all target lesions and partial response as a 30% decrease in the sum of the longest diameters of target lesions. These definitions follow the RECIST framework named in the record. The endpoint timeframe is duration of treatment, up to 2 years.

Confirmation requires more than a response observed on a single scan: the response must also be seen on a later scan meeting the stated spacing criterion. This can make the endpoint less sensitive to a transient or uncertain scan finding than an unconfirmed response measure. It is a tumor-response endpoint, not a direct measure of survival, symptom relief, or quality of life. The registry lists those distinctions in part through its separate secondary outcomes.

The record says the rate was estimated as the number of successes divided by the total number of evaluable patients, and that a 95% confidence interval was estimated using the Duffy and Santner method. However, it does not report the number of confirmed responses, the resulting rate, or the confidence interval. The endpoint definition is therefore available, but the posted registry record does not give the numerical outcome.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The statistical method named for the primary endpoint is a confidence interval estimated by the Duffy and Santner method. The registry describes the response rate as the number of successes divided by the total number of evaluable patients. A success means a confirmed complete or partial response under the stated RECIST-based definition. The endpoint is expressed as a percentage of participants, and its timeframe is duration of treatment, up to 2 years. Although these specifications identify the intended summary and interval method, the registry record does not report the response count, response-rate estimate, or confidence interval.

What does a response-rate estimate describe?

General explanation: a response-rate estimate is a proportion, expressed as a percentage, of the defined analysis population meeting the response criterion. Its interpretation depends on the eligibility rules, the way responses are assessed, and which participants are included in the denominator. A confirmed response requires the response finding to persist on subsequent scans at least 4 weeks apart, according to the registry definition. The response-rate estimate would describe the evaluable participants under those rules; it would not automatically describe every enrolled patient or establish what would have happened under a different treatment.

The registry states that patients who did not continue beyond cycle 1 were not evaluated on consecutive cycles and were excluded from this endpoint. This is relevant because the analysis population is not simply the enrolled population. The record does not give the final evaluable denominator or describe additional exclusions. The number enrolled and the number eligible for a particular outcome can differ for reasons tied to outcome assessment and follow-up; the precise relationship must be reported to evaluate the analysis fully.

Why use a confidence interval?

General explanation: a confidence interval accompanies an estimate to show its statistical uncertainty under a specified method and assumptions. It provides a range of values compatible with the data and method, rather than a guarantee that the underlying value lies within that particular interval. It also helps readers judge how much precision the observed data support. The registry names a 95% confidence interval and the Duffy and Santner method, but without the interval or the underlying response count, readers cannot assess its width or the stability of the estimate.

General explanation: interval methods can differ in their performance, particularly when a proportion is based on a limited number of observations or is near the boundaries of its possible range. Naming the method improves transparency about how uncertainty was intended to be quantified. It does not replace reporting the estimate, denominator, and interval. The registry's method description is therefore informative about the planned presentation, but it is not a numerical result.

What can the group structure tell readers?

The trial is marked non-randomized, with no masking, and a parallel model. The groups are described by tumor type, and both received the same sorafenib regimen. General explanation: randomization can help balance measured and unmeasured characteristics across treatment groups, while non-randomized group assignment may leave group differences that complicate comparisons. Here, the group labels do not define a randomized comparison of different interventions. Group-specific response summaries, if available, would be descriptive and should not be interpreted as evidence that one treatment strategy is superior to another.

General explanation: masking may reduce the influence of knowledge about treatment assignment on some assessments, but its role depends on what is masked and how outcomes are measured. The record lists masking as none, and does not report further safeguards for response assessment. Since the primary outcome relies on scan-based criteria and confirmation, details about assessment procedures would help readers judge potential measurement bias. Those details are not reported in the registry record.

How should planning calculations be understood?

The registry describes sample-size assumptions for a planning illustration based on an observed effect: a two-sided significance level of 0.05, equal allocation between two arms, a normal approximation, and detection of a stated fraction of the observed effect. It distinguishes participants per arm for non-time-to-event outcomes from total events for time-to-event outcomes. These are planning assumptions, not a report of observed power. The registry record does not report the trial's own sample-size justification, and the numerical primary endpoint result needed to interpret an illustration based on the observed effect is not reported in the record.

General explanation: a sample-size calculation describes what a design might require under chosen assumptions; it does not tell readers the probability that a completed study detected an effect after the results are known. Observed power is not a substitute for an estimate and its confidence interval. Planning calculations are sensitive to assumptions about the target effect, allocation, significance level, and statistical approximation. Since the study was non-randomized and its two groups were disease-based, an equal-allocation two-arm illustration should not be mistaken for the actual group structure or a causal treatment comparison.

The registry lists no analyses. It does not specify a detailed statistical analysis plan, adjustments for multiple outcomes, interim analyses, handling of missing endpoint data beyond the stated exclusions, or subgroup analysis procedures. Those methods should not be inferred from the named confidence-interval approach. General explanation: transparent reporting of analysis populations, missing-data handling, prespecified comparisons, and uncertainty measures allows readers to distinguish the planned analysis from choices made after data collection. The record's outcome definition offers some methodological detail, but the absence of the numerical outcome and broader analysis information leaves substantial limits on independent statistical appraisal.

10. Exact Estimates for Each Arm

No two-arm comparison is shown for this trial, but each arm's result can still be summarized with exact interval estimates computed from the posted counts. The Clopper-Pearson interval is exact and conservative; the Wilson interval is a common closer-to-nominal alternative; the Bayesian credible interval uses a uniform Beta(1, 1) prior. For small groups the three can differ noticeably.

R code and output

arm    <- c("Group A (Patients With Carcinoid Tumors)", "Group B (Islet Cell and Other Neuroendocrine Tumor")
n      <- c(41, 36)
events <- c(4, 5)
for (i in seq_along(n)) {
  ci <- binom.test(events[i], n[i])$conf.int          # Clopper-Pearson exact interval
  cr <- qbeta(c(0.025, 0.975), 1 + events[i], 1 + n[i] - events[i])   # Bayesian, uniform Beta(1,1) prior
  wi <- (function(x, n) { p <- x / n; z <- qnorm(0.975); d <- 1 + z^2 / n
    c((p + z^2 / (2 * n) - z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d, (p + z^2 / (2 * n) + z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d) })(events[i], n[i])
  cat(sprintf("ARM|%s|%d|%d|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f\n", arm[i], n[i], events[i], 100 * events[i] / n[i],
              100 * ci[1], 100 * ci[2], 100 * wi[1], 100 * wi[2], 100 * cr[1], 100 * cr[2]))
}

ARM  Group A (Patients With Carcinoid Tumors)  41  4  9.8  2.7  23.1  3.9  22.5  4.0  22.6
ARM  Group B (Islet Cell and Other Neuroendocrine Tumor  36  5  13.9  4.7  29.5  6.1  28.7  6.2  28.8

11. Multiplicity, Missing Data and Analysis Populations

The registry names confirmed response rate as the primary outcome and lists toxicity, overall survival, progression-free survival, and duration of response as secondary outcomes. It also lists five outcome measures, but the record does not describe how multiplicity was handled. In particular, it does not say whether statistical testing across outcomes or groups was adjusted, whether any outcomes had a hierarchy, or how the primary result would be distinguished from exploratory findings. No adjustment should be assumed. The record also does not report interim analyses or a detailed statistical analysis plan.

The endpoint population description identifies a specific exclusion: patients who did not continue treatment past cycle 1 were not evaluated on consecutive cycles and were excluded from the confirmed response endpoint. It gives 9 of 50 carcinoid patients and 6 of 42 islet cell patients as affected by this exclusion. The final evaluable denominator and the response outcomes among excluded patients are not reported in the registry record. The record does not say how other missing assessments, incomplete scans, or other missing endpoint information were handled.

General explanation: missing outcome information can affect an estimate if the participants with observed outcomes differ from those without them. Excluding participants may be reasonable for a defined evaluability rule, but readers need to know the analysis population and reasons for missingness to judge how the results might be affected. General explanation: multiplicity matters when several outcomes or comparisons are examined, because the chance of at least one apparently unusual result can increase as the number of tests grows. Without the analysis plan, prespecified hypotheses, and posted outcome values, it is not possible to assess these issues for this trial beyond noting that the registry record does not report those details.

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: abdominal pain (19.4%), fatigue (16.1%), nausea (14.0%), anorexia (10.8%), vomiting (8.6%).

The registry lists 93 patients at risk for both serious and other adverse events. It reports 52 affected for serious adverse events and 90 affected for other adverse events. These are participant-level affected figures as recorded; they should not be read as counts of all individual events. The record also lists serious adverse-event terms with percentages: abdominal pain, 19.4%; fatigue, 16.1%; nausea, 14.0%; anorexia, 10.8%; vomiting, 8.6%; and dehydration, 6.5%.

The registry record does not give the event-level detail needed to determine whether a participant experienced more than one event, the severity distribution beyond the serious classification, the timing of events, or the attribution of each event to treatment. The listed percentages should be read as the registry reports them; the record does not state the denominator for each term in the details. The death figure is not reported in the registry record.

General explanation: adverse-event summaries describe events observed during a study and do not by themselves establish that the intervention caused them. Interpretation also depends on definitions, follow-up, exposure, and how events were collected. These figures should not be compared with another population or study without comparable definitions and reporting. They are safety descriptions, not a treatment-effect estimate.

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

The study record describes a phase II investigation of sorafenib tosylate in progressive metastatic neuroendocrine tumors, with a primary endpoint based on confirmed tumor response. It gives a response definition, an evaluability rule, and the intended confidence-interval method. It does not provide the response count, percentage, or 95% confidence interval. The central numerical result and its precision are therefore not available in the registry record.

Even if response results were available, this design would support descriptive conclusions about outcomes in the analyzed participants rather than a randomized causal comparison. The groups were non-randomized and distinguished by disease category, and both group descriptions specify the same treatment. There is no described control group receiving another intervention or no study drug. Differences between groups would not isolate a treatment effect.

The confirmed response endpoint records whether specified scan changes were observed and subsequently confirmed. General explanation: a response endpoint is not interchangeable with overall survival, progression-free survival, duration of response, toxicity, symptoms, or quality of life. The registry lists several of these as secondary outcomes, but their results are not reported in the record. Nothing about an endpoint definition alone supplies the outcome values for those measures.

The safety figures show the numbers recorded as affected for serious and other adverse events and list several serious adverse-event terms and percentages. Those data do not establish causation. The registry record does not include enough detail to reconstruct timing, attribution, participant-level overlap, or the death count. As a result, the available safety information is descriptive and should not be extended beyond what it reports.

General explanation: a study result is most informative when an estimate is paired with its uncertainty, its analysis population, and a design capable of addressing the intended comparison. Here, the record supports explanation of the endpoint and study structure, but not a numerical conclusion about the primary endpoint or a comparative claim about treatment effects. Completion status indicates that the study is listed as completed; it does not fill in unreported outcome results.

14. Why This Trial Matters Statistically

This trial is statistically informative as an example of why a clearly defined endpoint and a reported result are both necessary. The registry specifies confirmed response using RECIST-based criteria, requires confirmation on subsequent scans at least 4 weeks apart, identifies an endpoint timeframe, and names a confidence-interval method. Those features help explain what the intended estimate would represent. Yet the response count, percentage, and interval are not reported in the registry record, so the principal outcome cannot be numerically interpreted from the record alone.

The trial also illustrates the importance of distinguishing treatment arms from disease-defined groups. It is labeled non-randomized, and the group descriptions refer to different tumor categories receiving the same intervention. Describing them as parallel groups does not make them a randomized comparator design. General explanation: causal comparisons depend on how groups are formed and on whether differences other than treatment can plausibly explain observed outcome differences. The design classification is therefore essential context for interpreting any group-specific summaries.

The endpoint population description makes the analysis denominator another important issue. The record identifies patients who were excluded because they did not continue beyond cycle 1 and were not evaluated on consecutive cycles, but does not report the final evaluable denominator or the response results. General explanation: readers need both the numerator and denominator to interpret a proportion, and need a clear account of exclusions to understand who the estimate describes.

Finally, this record demonstrates that listing secondary outcomes is not the same as reporting their results. It names toxicity, overall survival, progression-free survival, and duration of response, but the information available does not give results for them or explain multiplicity procedures. Statistical transparency requires outcome values and uncertainty as well as definitions. The record offers meaningful design and methods information, while leaving the principal numerical findings and several analysis details unreported.

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 main limitation for statistical interpretation is that the registry record does not report the primary response count, rate, or confidence interval. Without these values, neither the observed magnitude nor its precision can be evaluated. Although the endpoint definition and Duffy and Santner interval method are specified, they cannot substitute for the actual estimate and denominator.

The outcome population is also incompletely described. The record states that 9 of 50 carcinoid patients and 6 of 42 islet cell patients did not continue treatment past cycle 1 and were excluded from the endpoint. It does not provide the final evaluable counts or describe any further exclusions or missing scans. It also does not explain how the endpoint group counts relate to the total enrollment of 93.

The non-randomized design and disease-based group definitions limit comparisons. Both groups are described as receiving the same sorafenib regimen, and no randomized control group is reported. Therefore, any apparent differences by group would not establish a treatment effect. Masking is listed as none; further details about response assessment procedures are not reported in the registry record.

The record does not report a detailed statistical analysis plan, a trial-specific sample-size justification, interim analyses, or multiplicity procedures. It lists secondary outcomes but does not report their results in the available record. For safety, participant counts and selected serious adverse-event terms are listed, but the death count, event-level detail, attribution, and the denominator for each term percentage are not reported. These limitations mean the registry record supports a description of the study design and intended measures, but only limited assessment of its findings.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Sorafenib Tosylate in Treating Patients With Progressive Metastatic Neuroendocrine Tumors — NCT00131911.
• Registry record last updated 2014-11-17; status last verified 2014-06.
• Results first posted 2014-11-17.
• 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

NCT00131911 is a completed phase II, non-randomized, unmasked parallel-model trial sponsored by the National Cancer Institute. The registry lists enrollment of 93 and describes sorafenib tosylate for progressive metastatic neuroendocrine tumors. Group A comprised patients with carcinoid tumors and Group B comprised patients with islet cell and other neuroendocrine tumors; both group descriptions specify 400 mg oral sorafenib twice daily on days 1-28.

The primary outcome was confirmed response rate during treatment, up to 2 years, defined using RECIST and assessed among evaluable patients. The record identifies patients excluded from this endpoint because they did not continue beyond cycle 1. It names the Duffy and Santner method for a 95% confidence interval but does not report the response estimate or interval. Secondary outcomes include toxicity, overall survival, progression-free survival, and duration of response. The registry record does not report their results or a detailed analysis plan.