Gemcitabine and Irinotecan in Treating Patients With Cancer of Unknown Primary: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Gemcitabine and Irinotecan in Treating Patients With Cancer of Unknown Primary (NCT00066781), a non-randomized phase 2 study. The primary outcome is Confirmed Response Rate (Partial or Complete Response for 2 Consecutive Evaluations at Least 4 Weeks Apart) as Measured by RECIST Criteria.

2. Clinical Question

The clinical question is descriptive rather than a randomized comparison of treatments: among eligible, treated patients with measurable cancer of unknown primary, what proportion had a confirmed response during the specified follow-up? The registry lists overall survival and time to disease progression as secondary outcomes, but outcome results for those measures are not reported in the registry record. The record therefore establishes the study’s question and intended response definition, but does not establish how many patients responded or how precisely a response rate was estimated.

3. Trial Design

This was a phase II, nonrandomized, parallel-model study with no masking. The registry lists enrollment of 31 and the status as completed. The study start is recorded as 2004-02, and primary completion as 2008-04. The sponsor was the Alliance for Clinical Trials in Oncology.

The record describes two treatment cohorts. In Cohort I, participants received intravenous gemcitabine over 30 minutes and intravenous irinotecan over 90 minutes on days 1, 8, 15, and 22. Courses repeated every 6 weeks unless disease progression or unacceptable toxicity occurred. Irinotecan dose could be escalated or de-escalated after course 1 depending on toxicity. Cohort I is marked closed to accrual 11/17/05. In Cohort II, the same infusion durations are listed, with treatment on days 1, 8, and 15 and courses repeating every 4 weeks unless disease progression or unacceptable toxicity occurred.

Because allocation was nonrandomized, the cohorts should not be interpreted as randomized treatment arms. Their schedules differ, and the registry record does not describe a prespecified comparative analysis between them. It also does not report how cohort membership was determined, whether the cohorts were intended as sequential regimens, or any adjustment strategy for comparisons. These details are not reported in the registry record.

The registry identifies measurable disease and previously untreated patients in the official study description. Its primary endpoint states that eligible patients who signed consent and began treatment would be evaluable for response. The number meeting that endpoint’s evaluability criteria is not reported in the registry record. Enrollment is not necessarily the same as the endpoint denominator.

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

Analysis population for the primary outcome, as defined in the registry: All evaluable patients

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 (Partial or Complete Response for 2 Consecutive Evaluations at Least 4 Weeks Apart) as Measured by RECIST Criteria, measured over Up to 2 years.

The primary endpoint was confirmed response rate, expressed as the percentage of patients with response. A response was defined as a complete response or partial response recorded as the objective status at two consecutive evaluations at least 4 weeks apart. The response assessment was to use RECIST criteria, and the stated timeframe was up to 2 years. The endpoint population was all evaluable patients; the record describes these as eligible patients who signed consent and began treatment.

Requiring confirmation at consecutive evaluations makes the endpoint more stringent than counting a response noted at only one assessment. In general, this approach helps distinguish a sustained response classification from a single observation, although it does not remove all uncertainty in measurement or follow-up. The registry record does not report assessment timing beyond the confirmation interval, the number of scheduled assessments, or how evaluations after treatment discontinuation were handled.

A response rate is a binary-outcome summary: each evaluable patient is classified as having or not having a confirmed response, and the rate describes the fraction classified as responders. A confidence interval conveys uncertainty around that estimated fraction. The registry record gives the intended confidence-interval type as 95%, but does not report the response estimate or interval limits. It is therefore not possible to describe the observed response frequency or the precision of its estimate from the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary measure is a binary response outcome. Each evaluable patient is classified as having a confirmed complete or partial response, or not meeting that confirmed-response definition. The response rate is the proportion of evaluable patients classified as responders. The registry record specifies the response definition and a 95% confidence interval, but does not report the estimate, interval limits, or the method used to calculate the interval.

What does a confidence interval convey?

In general, a confidence interval expresses the statistical uncertainty around an estimate. For a response proportion, a narrower interval indicates greater precision under the method’s assumptions, while a wider interval indicates less precision. The interval does not guarantee that the true value lies inside its bounds, and it does not resolve limitations in study design, outcome ascertainment, or the choice of analysis population. Since no bounds are reported in the registry record, the precision of the response estimate cannot be assessed from this record.

Why does confirmation matter?

The endpoint requires a complete or partial response at two consecutive evaluations at least 4 weeks apart. In general, requiring confirmation can reduce the influence of a response classification observed only once. It also makes the endpoint dependent on repeated assessment and continued availability of evaluable measurements. The registry record does not explain how missed assessments, conflicting evaluations, or discontinuation before confirmation were handled. Those rules are not reported in the registry record.

What can be inferred from the design?

Nonrandomized allocation means treatment assignment was not described as random. In general, randomization helps balance known and unknown prognostic characteristics between groups; without it, differences in participant characteristics or treatment context may influence observed outcome differences. This study also has cohort schedules that differ in treatment days and course intervals. Thus, the cohorts do not constitute a randomized comparison of otherwise identical regimens. No comparative analysis method is reported in the registry record.

The registry record identifies a parallel model and no masking. In general, masking can reduce the influence of knowledge of treatment assignment on assessment or reporting. The record does not provide details on how RECIST assessments were conducted or whether assessors were independent. Such procedures are not reported in the registry record.

What do the planning assumptions mean?

The record describes sample-size planning illustrations based on the observed effect, a two-sided significance level of 0.05, equal allocation, and a normal approximation. It states that the effect to detect is a specified fraction of the observed effect; no numeric fraction or illustration output is reported in the registry record. The record distinguishes participants per arm for non-time-to-event outcomes from total events for time-to-event outcomes. These are planning illustrations, not observed power and not a substitute for the study’s own design rationale. The trial’s own sample-size justification is not reported in the registry record.

In general, over-dispersion matters when data vary more than a statistical model expects. For a simple binary response proportion, the main uncertainties include the number of evaluable participants and the number with confirmed response; the registry record supplies neither response count nor response estimate. No model accounting for over-dispersion is described, and no model-based analysis is reported in the registry record.

The registry lists three outcome measures and zero analyses. That record-level information does not specify statistical tests, confidence-interval calculation methods, covariate adjustments, or handling of multiple outcomes. These methods are not reported in the registry record. Statistical interpretation should therefore remain descriptive and limited to the design, endpoint definition, and adverse-event counts that are actually recorded.

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("Cohort I", "Cohort II")
n      <- c(11, 15)
events <- c(1, 2)
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  Cohort I  11  1  9.1  0.2  41.3  1.6  37.7  2.1  38.5
ARM  Cohort II  15  2  13.3  1.7  40.5  3.7  37.9  4.0  38.3

11. Multiplicity, Missing Data and Analysis Populations

The registry lists confirmed response rate as the primary endpoint and lists overall survival and time to disease progression as secondary outcomes. It also records three outcome measures and zero analyses. The statistical plan for controlling multiplicity across outcomes or across any repeated analyses is not reported in the registry record. Without that plan or reported outcome estimates, it is not possible to assess whether statistical testing was conducted or how any resulting uncertainty would have been interpreted.

The primary endpoint definition identifies the intended response population as eligible participants who signed consent and began treatment. The number evaluable under that definition is not reported in the registry record. The record does not describe how missing response assessments, incomplete follow-up, withdrawal, or treatment discontinuation were handled. It also does not report whether all enrolled participants were included in any analysis population or whether any exclusions occurred.

For the secondary outcomes, the record does not report event counts, follow-up summaries, censoring rules, or analysis populations. Missing-data and censoring methods are not reported in the registry record. In general, such details matter because participants with incomplete outcome information may contribute differently to an analysis depending on the endpoint and the prespecified rules.

No imputation method, sensitivity analysis, or missing-data assumption is described. These details are not reported in the registry record. The absence of those details does not establish that data were missing; it means the registry record does not explain whether missingness occurred or how it would have been addressed.

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 non-serious adverse event terms

The serious adverse-event counts are zero affected among 14 participants at risk in Cohort I and zero affected among 17 participants at risk in Cohort II. The registry also lists other adverse events affecting 14 of 14 at risk in Cohort I and 16 of 17 at risk in Cohort II. These figures describe the recorded categories and denominators; the registry record does not provide further detail about event severity, duration, recurrence, or attribution.

Zero affected participants in the serious-event category means no participant in the listed at-risk denominator was recorded as affected in that category. It does not establish that the underlying risk is zero. In general, observed counts are uncertain estimates of event occurrence, especially when the number at risk is limited. The registry record does not report confidence intervals for these safety figures.

The cohorts had different schedules and were not randomized comparison groups. The counts should therefore be read descriptively, not as evidence that one schedule is safer than the other. Death counts are not reported in the registry record. The list of top serious-event terms is empty, so specific serious adverse-event terms are not reported in the registry record.

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

The record establishes that a phase II study evaluated a gemcitabine-and-irinotecan regimen in previously untreated patients with measurable cancer of unknown primary, with confirmed response as the primary endpoint. It also defines the response criteria and describes two cohort schedules. These are statements about study purpose and design, not evidence of an observed treatment effect.

The response result itself is not reported in the registry record. There is no response count, percentage, or confidence interval in the record, so the magnitude and precision of the primary endpoint cannot be described. The listed secondary outcomes, overall survival and time to disease progression, also have no reported results in the registry record. Consequently, the record does not establish the observed outcome pattern for either primary or secondary endpoints.

The nonrandomized design does not support treating the cohorts as randomized alternatives. Their schedules differ, and there is no reported comparative analysis. Even if cohort-specific outcomes were available, differences could reflect factors other than the schedule. The record does not report an adjustment plan that would address such differences.

The adverse-event counts show no recorded serious adverse events among the stated at-risk denominators and frequent events in the separate other-adverse-event category. Those counts are descriptive and do not quantify comparative risk or provide a complete account of safety. Death counts and detailed serious-event terms are not reported in the registry record.

In general, a trial endpoint can answer a focused statistical question only when the result, analysis population, and uncertainty are available for interpretation. Here, the registry record documents the question but not the corresponding outcome estimates. It therefore cannot support a conclusion about the size, direction, or precision of response or survival outcomes.

14. Why This Trial Matters Statistically

This study illustrates the distinction between a clearly defined endpoint and a reported endpoint result. The registry specifies confirmed response, the RECIST basis, the requirement for consecutive evaluations at least 4 weeks apart, and a timeframe of up to 2 years. Yet the response estimate and confidence interval are not reported in the registry record. Statistical interpretation requires both an operational definition and outcome data with an identifiable denominator.

The record also highlights why design labels matter. A nonrandomized, parallel study with two different schedules is not a randomized head-to-head comparison. Treating cohort outcomes as directly comparable could overstate what the design can establish. The registry explicitly notes that the arms are not a like-for-like control comparison.

Safety counts demonstrate another useful statistical distinction: zero observed serious events is not the same as proving zero risk. The at-risk denominators are reported, but no uncertainty interval accompanies the counts, and death totals are not reported in the registry record. Careful description separates what was observed from what the data can establish.

Finally, the record lists secondary outcomes and sample-size planning assumptions while reporting no analyses. This underscores that endpoint selection, planning assumptions, analysis methods, and actual results are separate parts of a statistical account. When methods or results are absent, they cannot be reconstructed from the study title, treatment schedule, or enrollment figure.

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 interpreting outcomes is that the registry record does not report the primary response result, its confidence interval, or results for the listed secondary outcomes. It therefore gives no direct estimate of response frequency, survival, or time to disease progression. The number of participants evaluable for response is also not reported in the registry record, so the endpoint denominator cannot be checked against the enrollment of 31.

The nonrandomized design limits causal and comparative interpretation. Cohort I and Cohort II received schedules differing in treatment days and course intervals, and the registry identifies the arms as not being a like-for-like control comparison. The record does not describe cohort assignment or a strategy for adjusting comparisons. Cohort-level differences should not be interpreted as effects attributable solely to schedule.

The registry record does not describe response-assessment procedures beyond the stated RECIST definition and confirmation requirement. It does not report missing-assessment rules, analysis populations beyond the primary endpoint description, or methods for handling incomplete outcome information. The analysis methods, multiplicity plan, and trial-specific sample-size justification are not reported in the registry record.

The safety summary supplies at-risk counts and affected counts for two adverse-event categories, but does not report confidence intervals, detailed event characteristics, or deaths. These omissions limit assessment of the precision and completeness of the safety information. The record’s completion status does not itself supply the missing outcome estimates or analytical details.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Gemcitabine and Irinotecan in Treating Patients With Cancer of Unknown Primary — NCT00066781.
• Registry record last updated 2017-04-04; status last verified 2017-02.
• Results first posted 2017-04-04.
• 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

The registry describes a completed phase II, nonrandomized, unmasked, parallel-model study of gemcitabine and irinotecan in previously untreated patients with measurable carcinoma of unknown primary. Enrollment was 31. Two cohorts received different intravenous treatment schedules, with Cohort I closed to accrual 11/17/05.

The primary endpoint was confirmed complete or partial response by RECIST, requiring two consecutive evaluations at least 4 weeks apart, with assessment planned up to 2 years. Overall survival and time to disease progression were listed as secondary outcomes. Primary and secondary outcome results, analysis methods, and the number evaluable for response are not reported in the registry record.

The safety summary records zero serious adverse events among 14 at risk in Cohort I and among 17 at risk in Cohort II. Other adverse events were recorded for 14 of 14 and 16 of 17, respectively. Death counts are not reported in the registry record.