A Study of AGS-16C3F vs. Axitinib in Metastatic Renal Cell Carcinoma: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study of AGS-16C3F vs. Axitinib in Metastatic Renal Cell Carcinoma (NCT02639182), a randomized phase 2 study. The primary outcome is Progression Free Survival (PFS) as Per Response Evaluation Criteria in Solid Tumors Version 1.1 (RECIST v1.1) Assessed by Investigator Radiologic Review.

2. Clinical Question

3. Trial Design

This was a randomized, open-label, parallel-group phase 2 study of AGS-16C3F versus axitinib in metastatic renal cell carcinoma. Enrollment was 133. The primary endpoint analysis lists 67 participants in the AGS-16C3F arm and 66 in the axitinib arm, and identifies the analysis population as the FAS Population. The registry record says the study is completed. The recorded start date is 2016-05-03, and primary completion was 2020-10-02.

Participants assigned to AGS-16C3F received 1.8 milligram per kilogram once every three weeks by single intravenous infusion. Participants assigned to axitinib received 2 to 10 milligram orally twice daily, following the product label and local institutional guidelines. Randomization supports a planned comparison between the groups, while the open-label design means treatment assignment was not masked.

The primary endpoint was progression-free survival by investigator radiologic review using RECIST v1.1. Its timeframe is described as randomization to the earliest of documented progression or death from any cause, up to 53 months. Secondary outcomes included progression-free survival assessed by blinded central radiology, investigator-assessed response and response duration, overall survival, disease control, adverse events, and pharmacokinetic measures. The registry record lists 22 outcome measures and six analyses; 21 secondary outcomes are posted. These listed outcomes show the range of assessments but do not establish which outcomes were prioritized in the protocol or how testing across them was planned. The statistical analysis plan, sample-size justification, and interim-analysis plan are not reported in the registry record.

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: Randomization was stratified by Eastern Cooperative Oncology Group (ECOG) performance status (0 or 1), the number of prior systemic RCC regimens (2 or \> 2) and Renal cell carcinoma (RCC), histology (clear or nonclear cell).

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: FAS Population

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: Full Analysis Set (FAS): All participants who were randomized to study drug.

6. Primary Endpoint

The registry lists the primary outcome as Progression Free Survival (PFS) as Per Response Evaluation Criteria in Solid Tumors Version 1.1 (RECIST v1.1) Assessed by Investigator Radiologic Review, measured over From date of randomization to the earliest of either documented disease progression or death from any cause (up to 53 months).

Progression-free survival is the time from randomization until the first of documented disease progression or death from any cause. It is used as a time-to-event endpoint because it combines two clinically relevant events into a measure of how long participants remain alive without recorded progression. The primary assessment was made through investigator radiologic review using RECIST v1.1.

In the registry definition, progressive disease includes a specified increase in the summed diameters of target lesions relative to the smallest sum on study, with an absolute increase requirement as well as a relative one. The appearance of one or more new lesions also counts as progression. The registry record describes participants as censored in several circumstances, including being alive without progression, having no post-baseline assessments, having consecutive missing assessments immediately before death or progression, or starting non-protocol anticancer treatment before progression or death. Censoring means that a participant's exact event time is not treated as observed under the analysis; the participant contributes follow-up up to the recorded censoring time.

Kaplan-Meier estimates were used to summarize progression-free survival. A median gives the time by which the estimated event-free survival curve reaches its midpoint. It is a summary of the observed time-to-event distribution, not a measure of how long every participant remained progression-free. The median alone does not describe the full curve, and interpretation also depends on follow-up, censoring, and uncertainty. The registry record reports 95% confidence intervals for each arm's median, but does not report a primary confidence interval for the between-arm comparison.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

Progression-free survival is a time-to-event outcome. Each participant's follow-up contributes information about the time until documented progression or death, if either event occurs during observation. If neither event is observed before follow-up ends or a censoring condition applies, the participant is censored rather than assigned an event time. Censored observations still contribute information up to their recorded follow-up time. The analysis therefore uses both event times and the timing of censoring, rather than treating all participants as if they had the same duration of observation.

What does a Kaplan-Meier median describe?

The Kaplan-Meier method estimates the proportion of participants who remain free of the endpoint over time while accounting for censoring. The median is the time at which that estimated proportion reaches the midpoint. It offers an interpretable summary of the event-time distribution, but it does not show the shape of the full curve or the timing of all events. In this trial, the primary medians were 2.9 months for AGS-16C3F and 5.7 months for axitinib. Their 95% confidence intervals quantify uncertainty in each arm's median estimate. They do not directly give the uncertainty in the difference between arms.

What does the log-rank test assess?

The log-rank test compares observed event patterns between groups over the follow-up period, taking account of censoring. The registry record reports a primary log-rank p-value of 0.983. A p-value summarizes how compatible the observed test statistic is with the test's null hypothesis under its assumptions. It is not the probability that the null hypothesis is true, nor does it measure the probability that a result will replicate. It also does not establish equivalence. The registry record does not give a primary hazard ratio or a confidence interval for the comparison, so the relative magnitude and precision of a primary between-group effect are not available from the posted primary result.

How should confidence intervals be read?

A confidence interval represents uncertainty around an estimate under the statistical method used. The arm-specific 95% confidence intervals for the primary medians are 2.0 to 4.0 months for AGS-16C3F and 5.3 to 9.1 months for axitinib. These intervals concern each median separately. They are not an interval for a difference or ratio between the medians. General interpretation should distinguish estimates for individual groups from estimates of a direct comparison. The same distinction applies to secondary outcomes: an interval around an arm-specific median is not the interval for the hazard ratio.

What does censoring mean for these results?

The registry definition lists several censoring situations: participants alive without progression, those with no post-baseline assessments, those with consecutive missing assessments immediately before progression or death, and those who start non-protocol anticancer treatment before progression or death. These rules determine which follow-up time contributes to the time-to-event analysis. Censoring is not automatically a problem, but Kaplan-Meier and log-rank methods rely on assumptions about how censoring relates to event risk. If the reasons for ending observation are related to unobserved progression or death risk, estimates may be affected. The registry record does not describe sensitivity analyses for alternative censoring assumptions.

Why can over-dispersion matter for other endpoints?

General: count or rate outcomes can vary more than a model assumes, a condition often called over-dispersion. When present, it can make standard errors too small if the model does not allow for the extra variability, leading to confidence intervals that are too narrow and tests that are too optimistic. This is a general statistical concern, not a finding reported for the primary endpoint in this study. The primary endpoint is time-to-event, and the registry record describes Kaplan-Meier estimates and a log-rank comparison.

How should the planning illustrations be understood?

The sample-size illustrations are planning calculations based on an observed effect, not the trial's own design justification. They use a two-sided significance level of 0.05, equal allocation, and a normal approximation; the effect to detect is expressed as a fraction of the observed effect. Time-to-event illustrations state the total number of events, while other illustrations state participants per arm for 80% power or 90% power. These are hypothetical planning scenarios, not observed power and not a claim about the probability that the completed trial detected an effect. The registry record does not report that these assumptions were used to design the trial.

For time-to-event planning, an event count is central because the information for comparing groups depends on the number of observed events, not just the number enrolled. For a binary outcome, a participant count per arm is tied to the assumed event proportions and effect size. In either case, planning estimates depend on the assumed effect and analysis conditions. An observed effect used as an input after results are known is especially uncertain, so alternative fractions illustrate how required enrollment or event totals change under smaller target effects. The calculations should not be mistaken for a retrospective test of whether the study was adequately powered.

For the primary endpoint, the registry record identifies the FAS Population and a log-rank analysis, but does not report the protocol's sample-size justification, interim analyses, or full statistical analysis plan. Those omissions leave important design details unavailable, including whether any other prespecified analysis or adjustment was intended. The reported methods and estimates can still be described, but unreported procedures should not be inferred from the posted results.

10. Analyses Reported by the Sponsor

The sponsor-reported primary comparison is a log-rank test with p = 0.983. The primary arm summaries are the Kaplan-Meier medians and their 95% confidence intervals. No primary hazard ratio or confidence interval for the between-arm effect is reported in the registry record, so there is no primary comparison estimate to set alongside an unadjusted recalculation. The record does not flag a discrepancy between sponsor results and a recalculation.

For secondary progression-free survival by blinded central radiology, the sponsor reports a hazard ratio of 1.423, with a 95% confidence interval from 0.924 to 2.192 and p = 0.110. For investigator-assessed objective response, the sponsor reports an odds ratio of 0.4, with a 95% confidence interval from 0.1 to 1.1 and p = 0.062. For duration of response, the sponsor reports a hazard ratio of 0.376, with a 95% confidence interval from 0.031 to 4.482 and p = 0.439. These are sponsor analyses using the stated methods; they should not be replaced by simple calculations from arm-level summaries.

In general, a recalculation from posted summaries may use a different method or lack adjustments used in the sponsor's analysis. When the registry record identifies a discrepancy, the sponsor's own analysis takes precedence. No such discrepancy is flagged in this record. The reported analyses remain the relevant registry results, while their interpretation should account for endpoint choice, uncertainty, and the registry record's unreported statistical-plan details.

11. Multiplicity, Missing Data and Analysis Populations

The registry record lists 22 outcome measures, 21 posted secondary outcomes, and six analyses. Multiple outcomes can be useful for describing a trial, but testing many outcomes raises a multiplicity issue. When many hypotheses are examined, the chance of at least one apparently notable result can increase if each test is interpreted without accounting for the others. A multiplicity procedure may control that issue by adjusting significance thresholds or by setting a hierarchy, but the registry record does not report such a procedure. It also does not report a testing hierarchy or identify which secondary tests, if any, were confirmatory.

The primary outcome's censoring rules are described. Participants could be censored if they were alive without progression, had no post-baseline assessments, had two or more consecutive missing assessments immediately before death or progression, or began non-protocol anticancer treatment before progression or death. Such rules are important because the event time is not observed for every participant. The Kaplan-Meier and log-rank methods use the available event and censoring information, and their interpretation depends on assumptions about censoring. The registry record does not describe sensitivity analyses for missing assessments, alternative censoring definitions, or the effect of missing data on the primary result.

The primary endpoint is reported for the FAS Population, but the registry record does not give further detail about how that population was defined or how each randomized participant was handled in every analysis. It does not report a full accounting of missing endpoint observations in the information available for this result. The registry record also does not report interim analyses, a sample-size justification, or a full statistical analysis plan. These absences limit assessment of whether analysis choices were prespecified and how multiplicity and missingness were handled across the whole study. They do not establish that a particular procedure was or was not used; those details are simply not reported in the registry record.

12. Safety Results

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

Deaths

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: dehydration (6.2%), abdominal pain (4.5%), diarrhoea (3.1%), pneumonia (3.1%), back pain (3.1%).

The posted serious adverse event figures are counts and percentages for specific serious-event terms. The registry record reports dehydration at 6.2%, abdominal pain at 4.5%, diarrhoea at 3.1%, pneumonia at 3.1%, back pain at 3.1%, and pain in extremity at 3.1% among the listed top serious terms. These figures describe reported events; they are not, by themselves, estimates of causal risk attributable to a treatment.

For the overall serious adverse event fields, 26 of 66 participants at risk in the AGS-16C3F group and 31 of 65 at risk in the axitinib group were recorded as affected. Other adverse events were recorded for 65 of 66 and 64 of 65 at risk, respectively. The registry record also lists 36 deaths in the AGS-16C3F group and 44 in the axitinib group. These are counts in the posted safety data, not adjusted comparisons. They do not establish why an event occurred or whether treatment caused it.

Safety interpretation depends on the event definitions, observation time, attribution process, and whether participants had comparable exposure and follow-up. The registry record does not provide enough detail in these fields to resolve those issues. The figures should therefore be read descriptively and in the context of the study's design and reporting, rather than as a ranking of treatment safety.

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

The primary result describes progression-free survival as assessed by investigators using RECIST v1.1. The estimated medians were 2.9 months with AGS-16C3F and 5.7 months with axitinib, and the reported log-rank p-value was 0.983. These statistics indicate that the registry's primary test did not identify a difference in the event-time patterns. They do not prove that the treatments have identical effects, nor do they establish that a clinically important difference is absent.

A nonsignificant test is not an equivalence test. Demonstrating equivalence or non-inferiority requires a suitable design, a prespecified margin, and an analysis structured around that question. The registry record does not report such a margin or identify this trial as an equivalence or non-inferiority study. It also does not provide a primary hazard ratio and confidence interval for the between-arm comparison, which would help express the range of relative effects compatible with the data.

The medians and their confidence intervals are estimates, not predictions for an individual participant. They summarize group-level follow-up under the analysis rules and do not show every participant's experience. The open-label design and investigator-based primary assessment are additional considerations. A blinded central radiology assessment was posted as a secondary outcome, but differing assessment methods do not make one result a substitute for another.

The findings also do not show that either treatment is appropriate for a particular person, or that the safety profile is favorable or unfavorable in general. The record reports event counts, but those counts alone do not establish treatment causation or comparative risk. Overall, the results are evidence about this trial's participants, treatments, endpoint definitions, and analyses. Generalization beyond that setting requires information that the registry result alone does not supply.

14. Why This Trial Matters Statistically

This trial illustrates why time-to-event results require more than a comparison of two median values. The primary medians summarize each arm, while the log-rank test compares event patterns across follow-up and accounts for censoring. These answer related but distinct questions. Without a reported primary hazard ratio and confidence interval for the comparison, the registry record offers limited information about the size and precision of a relative effect, even though it reports an inferential test and arm-specific estimates.

It also shows why endpoint assessment matters. The primary progression-free survival measure was based on investigator radiologic review in an open-label trial. A secondary outcome used blinded central radiology assessment, illustrating that an endpoint can be assessed through different processes. Statistical interpretation must keep those assessments distinct rather than treating all progression-free survival results as interchangeable.

The trial is also a useful example of the difference between observed results and planning calculations. Sample-size illustrations based on an observed effect can help explore hypothetical future planning assumptions, but they are not the trial's design rationale and do not describe observed power. For time-to-event outcomes, event totals are the planning quantity shown; other illustrations use participants per arm. Treating these as retrospective proof of adequacy would confuse planning with inference.

Finally, a broad set of secondary outcomes can enrich interpretation while raising multiplicity concerns. Without a reported adjustment strategy or hierarchy, isolated secondary p-values require caution. The trial's statistical meaning depends not only on the point estimates, but also on endpoint definitions, censoring, uncertainty, assessment methods, and reporting completeness.

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.

Important details for interpreting the primary comparison are not reported in the registry record. In particular, there is no primary hazard ratio or confidence interval for the between-arm effect, and the registry record does not provide a primary comparison interval that would quantify its precision directly. The medians and their separate confidence intervals should not be treated as that missing comparison estimate.

The trial was open label, and the primary endpoint depended on investigator radiologic review. This creates a potential for knowledge of assignment to affect assessment processes in general. A blinded central assessment was posted as a secondary outcome, but it is a distinct analysis and does not erase the limitations of the primary assessment. The registry record does not report how investigator assessments were independently monitored or adjudicated.

The record describes censoring rules but does not report sensitivity analyses for missing assessments or alternative censoring assumptions. It also does not provide a full account of missingness in the primary endpoint data. Since time-to-event estimates depend on follow-up and censoring, this limits assessment of how robust the results are to unobserved events or differences in follow-up.

Multiplicity handling for the many outcomes and analyses is not reported. The protocol's sample-size justification, interim-analysis plan, and full statistical analysis plan are also not reported in the registry record. Consequently, the reported results cannot establish from this record alone which analysis choices were prespecified or how secondary hypotheses were prioritized. These reporting gaps call for restraint, not an assumption that a particular method was used or omitted.

Lastly, the posted results concern the trial's defined population, treatment schedules, and follow-up. The registry record describes a completed phase 2 study, but does not by itself establish how results generalize to populations or settings beyond those studied. The secondary outcomes add context, yet differences in endpoint, assessment, and analysis mean they should not be combined into a single overall treatment claim.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: A Study of AGS-16C3F vs. Axitinib in Metastatic Renal Cell Carcinoma — NCT02639182.
• Registry record last updated 2024-11-18; status last verified 2024-10.
• Results first posted 2020-09-23.
• 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

NCT02639182 was a completed, randomized, open-label, parallel-group phase 2 study of AGS-16C3F versus axitinib in metastatic renal cell carcinoma. Enrollment was 133, and the primary analysis lists 67 participants assigned to AGS-16C3F and 66 assigned to axitinib. The primary endpoint was investigator-assessed progression-free survival under RECIST v1.1, analyzed using Kaplan-Meier estimates and a log-rank test.

Median progression-free survival was 2.9 months for AGS-16C3F and 5.7 months for axitinib; the reported log-rank p-value was 0.983. The registry record does not report a primary hazard ratio or confidence interval for the between-arm comparison. Secondary outcomes cover radiologic assessment, response, survival, safety, and pharmacokinetic measures. Multiplicity handling, the protocol's sample-size justification, interim analyses, and the full statistical analysis plan are not reported in the registry record. The findings should be interpreted as estimates and analyses from this specific trial, with attention to endpoint assessment, censoring, uncertainty, and missing reporting details.