Study of Pegilodecakin (LY3500518) With FOLFOX Compared to FOLFOX Alone Second-line Tx in Participants With Metastatic Pancreatic Cancer: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Study of Pegilodecakin (LY3500518) With FOLFOX Compared to FOLFOX Alone Second-line Tx in Participants With Metastatic Pancreatic Cancer (NCT02923921), a randomized phase 3 study. The primary outcome is Overall Survival.

2. Clinical Question

The question is specific to this treatment setting and comparison: participants had metastatic pancreatic cancer following a first-line gemcitabine-containing regimen, and both groups received FOLFOX. The randomized design supports a comparison between assignment to the combination and assignment to FOLFOX alone, but the posted summary does not establish how every aspect of treatment exposure, follow-up, or analysis was handled beyond the details explicitly recorded.

The trial was open-label, so participants and study personnel were not masked according to the registry. That feature describes how the trial was conducted; it does not by itself determine whether the survival endpoint is biased. Death is an objective event, while open-label conduct can still matter for care decisions, assessment of other outcomes, and the way follow-up proceeds. Further details about those processes are not reported in the registry record.

3. Trial Design

This was a phase 3 randomized parallel-group trial sponsored by Eli Lilly and Company. The registry lists the study as completed, with enrollment of 567 participants. The recorded start date is 2017-03-01 and the primary completion date is 2019-09-09. These dates describe the registry record; they do not supply additional details about recruitment duration, follow-up completeness, or the timing of individual analyses.

Participants were assigned to pegilodecakin plus FOLFOX or FOLFOX alone. The arm sizes in the primary endpoint results were 283.0 and 284.0, respectively. The registry reports no masking. A parallel-group design means the two assigned groups were followed as separate groups rather than participants crossing between assigned treatments as part of the stated comparison. Details about allocation concealment, randomization strata, or any stratified analyses are not reported in the registry record.

The comparison was an add-on treatment question: FOLFOX was used in both arms, while pegilodecakin was included only in the combination arm. The FOLFOX description lists dl-leucovorin, oxaliplatin, and 5-FU, and says treatment began on Day 1 of 14-day cycles for up to 12 cycles or until disease progression. The pegilodecakin description refers to subcutaneous dosing on Days 1-5 and Days 8-12. The combination-arm record is truncated, so its full regimen details are not reported in the registry record.

The brief summary identifies overall survival as the efficacy measure for the main comparison. The primary endpoint is also described as time from randomization to death from any cause, with participants whose last known status was alive at the analysis cutoff censored at their last contact before that cutoff. The registry lists progression-free survival, objective response rate, disease control rate, duration of response, and the 12-month survival rate among the secondary outcomes. Estimates for duration of response and the 12-month survival rate are not reported in the registry record.

Randomization can balance measured and unmeasured baseline characteristics on average, but it does not guarantee identical groups in a particular trial. Baseline characteristics, reasons for exclusions after randomization, and detailed participant flow are not reported in the registry record. The posted primary population is all randomized participants, which is important context for interpreting the survival analysis.

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: In the Participant Flow, participants who completed were those who died due to any cause or were alive and on study at conclusion but off treatment.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: All randomized participants. The number of participants censored were Pegilodecakin + FOLFOX = 63 and FOLFOX = 73.

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: All randomized participants.

6. Primary Endpoint

The registry lists the primary outcome as Overall Survival, measured over Randomization to date of death from any cause (Up To 30 Months).

Overall survival is the time from randomization until death from any cause. The registry describes the outcome as a median number of months and reports a 95% confidence interval for each arm. Participants known to be alive at the data cutoff were censored at their last contact date before the cutoff. The stated observation timeframe was up to 30 months.

In general, overall survival is used because it captures a clear event—death from any cause—and does not require deciding whether a particular symptom or scan change represents progression. Measuring from randomization also provides a common starting point for comparing the assigned groups. The all-cause definition includes deaths regardless of attributed cause, as described in the registry.

Censoring is used when the event time is not observed by the analysis cutoff. A censored participant contributes information up to the last recorded contact, but the exact subsequent time of death is unknown. The registry reports 63 censored participants in the pegilodecakin plus FOLFOX group and 73 in the FOLFOX group. Those counts describe censoring for the primary outcome; they do not, by themselves, establish whether the censoring process was independent of prognosis.

The median survival estimates were 5.78 months for pegilodecakin plus FOLFOX and 6.28 months for FOLFOX, with overlapping reported confidence intervals. A median describes the point in the estimated survival distribution at which half the group has experienced the event; it does not describe the survival time of every participant. The comparison between groups is therefore not determined by comparing medians alone. The registry also reports a hazard ratio and log-rank test for the primary analysis, which use the event-time information across follow-up rather than only the two median estimates.

7. Results Posted in the Registry

• Pegilodecakin + FOLFOX vs FOLFOX: hazard ratio 1.04 (95% CI 0.86–1.26), reported p = 0.656 (as reported in the registry).

8. Secondary Efficacy Endpoints

The registry also posts results for secondary outcomes. For each outcome below, the endpoint type, the test applied to the posted summary statistics, the result and its interpretation are given in turn. Secondary endpoints are usually not adjusted for multiple testing, so they support rather than confirm the primary finding.

Progression Free Survival

• Endpoint type: Time to event: compared as a hazard ratio.
• Test: Hazard ratio as reported by the sponsor (Log Rank); it was not recalculated.
• Result:
  • Pegilodecakin + FOLFOX vs FOLFOX: hazard ratio 0.98 (95% CI 0.81–1.19), reported p = 0.814 (as reported in the registry).
• Interpretation:
  • The 95% confidence interval includes no difference (a ratio of 1), so these data do not show a statistically significant difference between Pegilodecakin + FOLFOX vs FOLFOX and the control. The interval gives the range of effects that remain compatible with the data.

Percentage of Participants Achieving Complete Response (CR) or Partial Response (PR) [Objective Response Rate (ORR)] That Assessed by Investigator

• Endpoint type: Binary: the percentage of participants with the event (counts rebuilt from the posted percentages and group sizes).
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • FOLFOX: 5.6% (16/284) versus 4.6% (13/283) on control; risk difference 1.0 percentage points (95% CI -2.6 to 4.7), risk ratio 1.23 (0.60–2.50); unadjusted two-proportion z-test p = 0.574.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between FOLFOX and the control. The interval gives the range of effects that remain compatible with the data.

Percentage of Participants With a Best Overall Response of Complete Response (CR), Partial Response (PR) or Stable Disease (SD): Disease Control Rate (DCR)

• Endpoint type: Binary: the percentage of participants with the event (counts rebuilt from the posted percentages and group sizes).
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • FOLFOX: 36.6% (104/284) versus 42.8% (121/283) on control; risk difference -6.1 percentage points (95% CI -14.2 to 1.9), risk ratio 0.86 (0.70–1.05); unadjusted two-proportion z-test p = 0.135.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between FOLFOX and the control. The interval gives the range of effects that remain compatible with the data.

The secondary endpoints add different perspectives to the overall-survival result. Progression-free survival concerns time to progressive disease or death, while objective response and disease control summarize tumor response classifications over the stated assessment periods. These outcomes can help describe disease course and response patterns, but they are not interchangeable with overall survival. The primary result remains the main randomized comparison identified in the registry.

Across the posted secondary estimates, the results do not form a consistent pattern of clear separation between the groups. The progression-free survival hazard ratio is 0.981, with a 95% confidence interval of 0.808 to 1.190 and p = 0.814. The sponsor’s reported response comparisons also have confidence intervals that include their null values. This broad pattern is compatible with the uncertainty in the primary comparison, but secondary outcomes do not independently settle the primary question.

The registry lists duration of response and the 12-month survival rate as secondary outcomes, but their estimates are not reported in the registry record. Their absence limits the completeness of the secondary-outcome picture. The registry also does not report detailed assessment timing, independent review procedures, or the handling of missing response assessments.

In general, multiple outcome comparisons create a multiplicity concern: the more hypotheses examined, the greater the chance that at least one result appears notable by chance if every test is interpreted separately. A prespecified testing sequence or multiplicity adjustment can address this issue, but such a plan is not reported in the registry record. For that reason, unadjusted secondary results should carry less weight than the designated primary endpoint and should be interpreted as supportive or exploratory evidence rather than as independent confirmation.

Secondary endpoints can still be informative when considered with their definitions, estimates, and intervals. They should not be selected selectively based only on which appears most favorable. The available record does not provide enough detail to determine whether a formal hierarchy or adjustment governed secondary testing.

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

• Progression Free Survival
• Percentage of Participants Achieving Complete Response (CR) or Partial Response (PR) [Objective Response Rate (ORR)] That Assessed by Investigator
• Percentage of Participants With a Best Overall Response of Complete Response (CR), Partial Response (PR) or Stable Disease (SD): Disease Control Rate (DCR)
• Duration of Response (DOR)
• Percentage of Participants Alive at 1 Year (12-Month Survival Rate)

9. Statistical Interpretation

10. Statistical Methods Explained

What does the hazard ratio describe?

The primary comparison reports a hazard ratio of 1.045 for pegilodecakin plus FOLFOX versus FOLFOX, using a log-rank analysis. In general, a hazard ratio compares the instantaneous event rates between groups over follow-up, conditional on participants having remained event-free until each time. It is not a direct comparison of median survival, and it does not mean that one group lived a particular number of months longer or shorter.

A hazard ratio of 1 is the reference value for no relative difference in event rates under the hazard-ratio model. The reported 95% confidence interval, 0.863 to 1.265, includes that reference. This range communicates uncertainty in the estimated relative event rate. The registry record does not describe the model details beyond the reported log-rank method and hazard ratio, so assumptions about proportional hazards or other model diagnostics are not reported in the registry record.

How does the log-rank test use follow-up?

In general, the log-rank test compares the observed event experience between groups across event times, accounting for participants who are still at risk at each point and for censoring. It is a time-to-event method rather than a comparison of only the final event counts or median values. The registry identifies the log-rank method for the primary result and reports p = 0.656. Detailed test implementation, including any stratification or handling of ties, is not reported in the registry record.

The test result and the hazard-ratio estimate answer related but not identical questions. The hazard ratio expresses an estimated relative event rate, while the log-rank p-value summarizes the evidence against a null comparison under the test. Neither quantity alone conveys the full range of plausible effects. The confidence interval is useful for that reason, while the endpoint definition and censoring rules explain what data contribute to the comparison.

What does the median and its interval mean?

The median is a summary of the estimated time-to-event distribution. The registry reports 5.78 months with a 95% confidence interval of 5.45 to 6.64 for pegilodecakin plus FOLFOX, and 6.28 months with a 95% confidence interval of 5.62 to 7.39 for FOLFOX. In general, these intervals express uncertainty in each arm’s median estimate; they are not ranges containing most participants’ individual survival times.

The arm-specific intervals are not a substitute for the interval around the treatment comparison. A formal comparison uses the reported hazard ratio and its interval. Comparing whether separate arm-specific intervals overlap is not, by itself, a formal test of the difference between groups.

How are response outcomes analyzed?

Objective response rate and disease control rate are binary outcomes: each participant is classified according to whether the specified response criterion was met within the stated assessment period. The registry describes objective response as complete or partial response assessed by the investigator. Disease control includes complete response, partial response, or stable disease. The posted arm percentages are descriptive summaries, while the sponsor’s reported comparisons use the Cochran-Mantel-Haenszel method and odds ratios.

For objective response, the reported sponsor odds ratio is 0.8, with a 95% confidence interval of 0.4 to 1.7 and p = 0.704. For disease control, the sponsor reports an odds ratio of 1.3, with a 95% confidence interval of 0.9 to 1.8 and p = 0.146. In general, an odds ratio compares the odds of an outcome between groups; it is not the same as a difference in percentages. The registry does not report the strata or adjustment details behind the Cochran-Mantel-Haenszel analyses.

The unadjusted recalculations for binary outcomes use the posted participant counts and response totals. They are simple comparisons and do not replace the sponsor’s reported analyses. For objective response, the recalculated risk difference is 0.010, with a 95% confidence interval of -0.026 to 0.047. For disease control, it is -0.061, with a 95% confidence interval of -0.142 to 0.019. These are risk differences on the proportion scale; they should not be interpreted as odds ratios.

What are the planning calculations?

The planning illustrations use the observed effect as a starting point and show hypothetical sample sizes for detecting stated fractions of that effect. In general, power is the chance that a planned test detects an effect of a specified size under the assumptions used to calculate the design. These calculations do not tell whether the completed trial had adequate power under its original plan. The original sample-size justification and statistical plan are not reported in the registry record.

For time-to-event outcomes, the planning values are total event counts; for binary outcomes, they are participants per arm. They use a two-sided significance level of 0.05, equal allocation, and a normal approximation. The planning calculations use the stated fraction of the observed effect as the effect to detect. Since a smaller target effect is harder to distinguish from random variation, the planning illustration requires more participants or events as the target effect gets smaller. These assumptions describe the posted illustration, not necessarily the trial’s original design.

For the primary endpoint, the calculations use the reported hazard ratio as the observed effect. The illustration gives total events for 80% and 90% power under the stated effect fractions. For binary outcomes, the corresponding entries are participants per arm. Such quantities are useful for understanding the scale of a hypothetical future design, but they depend on assumptions about the effect and statistical approach. They do not establish what sample size would be appropriate for another study without its own design choices and clinical objectives.

11. Confidence Intervals and Precision

The primary hazard-ratio estimate is 1.045, and its 95% confidence interval is 0.863 to 1.265. In general, a confidence interval describes a range of estimates compatible with the data and method under repeated use of the same procedure. It does not assign a probability to the true effect being inside this particular interval. The interval includes 1, so the posted estimate is compatible with both a lower and a higher relative event rate for the combination group under the reported comparison.

The breadth of the interval matters alongside the point estimate. The estimate is close to 1, but the interval extends on both sides of 1; the data therefore do not pin the relative effect to a narrow value. The interval is not proof of no difference, and it does not establish equivalence. The registry record does not report a prespecified equivalence or noninferiority margin against which to interpret the bounds.

The median estimates also have uncertainty. For pegilodecakin plus FOLFOX, the median was 5.78 months with a 95% confidence interval of 5.45 to 6.64. For FOLFOX, the median was 6.28 months with a 95% confidence interval of 5.62 to 7.39. These intervals describe uncertainty for each arm’s median, not the range of survival times among participants. The comparison interval for the hazard ratio is more directly relevant to the treatment contrast.

In the secondary outcomes, sponsor-reported odds-ratio intervals are broad: the objective-response interval is 0.4 to 1.7, and the disease-control interval is 0.9 to 1.8. The recalculated risk-difference intervals also cross zero for both outcomes. In general, intervals that include the null value indicate that the data are compatible with no difference under the relevant measure, while also allowing effects in either direction. These results are not precise evidence that the groups have identical response outcomes.

Confidence intervals reflect sampling uncertainty under the analysis method; they do not account automatically for every potential bias, missing-data problem, or design limitation. Interpretation also depends on outcome definition, analysis population, censoring, and multiplicity. Details on several of these topics are not reported in the registry record.

12. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The planning values distinguish between time-to-event outcomes and binary outcomes. For overall survival and progression-free survival, each value is the total number of events needed across the comparison, not the number of participants per arm. For objective response and disease control, each value is participants per arm. The figures are planning illustrations based on the observed effect, not the trial’s original design calculation.

The stated assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, a normal approximation, and an effect to detect equal to the specified fraction of the observed effect. For the time-to-event calculations, the Schoenfeld approach is used for total events. The original trial’s sample-size assumptions and justification are not reported in the registry record.

The 80% power column means the hypothetical design would have an 80% chance of detecting the specified effect if the assumptions and effect were correct. The 90% power column describes a design with a higher chance of detecting that same specified effect, and generally requires more information. These are planning targets, not probabilities that the completed trial detected an effect and not observed power.

The effect fractions are alternative planning scenarios. Detecting a smaller effect is harder, so the illustration requires more participants or events when the target is a smaller fraction of the observed effect. For example, the overall-survival illustration gives 16,205 total events for 80% power when planning for 100% of the observed effect, and 64,818 total events when planning for 50% of the observed effect. Those values are hypothetical event requirements under the stated calculation assumptions; they are not the observed event count in the trial.

The calculations should be read as scale comparisons rather than recommendations. A future design would need its own target effect, type I error approach, allocation plan, endpoint assumptions, and feasibility assessment. The registry does not report whether the original trial used these planning assumptions.

13. How the Sample Size Is Derived

Power is the probability that a study of a given size rejects the null hypothesis when a specific true effect exists. It depends on four things: the effect to detect, the variability of the outcome, the significance level (here a two-sided α of 0.05, so z_1−α/2 = 1.960) and the desired power (z_1−β = 0.842 for 80% and 1.282 for 90%).

Formula

d is the total number of events needed (Schoenfeld's formula) with 1:1 allocation and proportional hazards.

Worked calculation for this trial

1. Hazard ratio HR = 1.045, so ln(HR) = 0.0440.
2. d = 4 × (1.96 + 0.842)² / [ln(1.045)]² = 16,204.3, rounded up to 16,205 events in total for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
hr <- 1.045                              # observed hazard ratio
events_needed <- function(power, frac = 1) ceiling(round(4 * (z_a + qnorm(power))^2 / log(hr^frac)^2, 6))   # Schoenfeld, 1:1 allocation
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: events = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(events_needed(0.80, f), big.mark = ","), format(events_needed(0.90, f), big.mark = ",")))

# Simulation: given d total events under proportional hazards, events in the treatment arm are Binomial(d, hr / (1 + hr)).
# The score (log-rank type) statistic is (D1 - d / 2) / sqrt(d / 4).
set.seed(2923921)
d    <- events_needed(0.80)
reps <- 20000
D1 <- rbinom(reps, d, hr / (1 + hr))
z  <- (D1 - d / 2) / sqrt(d / 4)
cat(sprintf("Simulated power with %s events: %.3f (%d replicates)\n", format(d, big.mark = ","), mean(abs(z) > z_a), reps))

Output from running the code

100% of observed effect: events = 16,205 (80% power), 21,693 (90% power)
 75% of observed effect: events = 28,808 (80% power), 38,566 (90% power)
 50% of observed effect: events = 64,818 (80% power), 86,772 (90% power)
Simulated power with 16,205 events: 0.797 (20000 replicates)

The simulation uses the fact that, with d events in total and proportional hazards, the number of events in the treatment arm is binomial. It is an approximation that conditions on the total number of events, and it should land close to the target power.

14. Analyses Reported by the Sponsor

The sponsor’s primary analysis reports a log-rank hazard ratio of 1.045, with a 95% confidence interval of 0.863 to 1.265 and p = 0.656. The primary endpoint summary uses this reported time-to-event comparison; no separate unadjusted recalculation is used in place of it. The registry record does not flag a discrepancy for the primary result.

For the binary secondary outcomes, the sponsor reports Cochran-Mantel-Haenszel odds ratios, while the simple recalculations use unadjusted risk differences from the posted counts. These measures and methods are different. For objective response, the sponsor reports an odds ratio of 0.8, while the recalculated risk difference is 0.010. For disease control, the sponsor reports an odds ratio of 1.3, while the recalculated risk difference is -0.061. Those estimates should not be compared as if they were the same statistic.

The sponsor’s analyses take precedence when describing the trial’s reported inferential results. The recalculations are supplementary unadjusted comparisons and do not reproduce the sponsor’s stratified method or any unreported analysis details. The registry does not describe the strata used for the Cochran-Mantel-Haenszel analyses, so the basis for any difference between adjusted and unadjusted estimates cannot be examined further from the record.

For both binary outcomes, the sponsor-reported confidence intervals are broad, and the reported p-values do not provide strong evidence against the corresponding null comparison. The appropriate reading remains tied to the reported method and its uncertainty, without treating a single p-value as a measure of clinical importance or as proof of no effect.

15. Multiplicity, Missing Data and Analysis Populations

The registry identifies overall survival as the primary endpoint and lists several secondary outcomes. Testing more than one outcome can increase the chance of a false-positive result when each comparison is interpreted at its own threshold. In general, a protocol can manage this by specifying a testing hierarchy, allocating the overall error rate, or using another multiplicity strategy. The statistical plan for multiplicity is not reported in the registry record, so it is not possible to determine how the secondary p-values were intended to be interpreted collectively.

The primary analysis population is described as all randomized participants. That is an important feature of the analysis description, but the record does not state whether every participant was included in every secondary analysis or explain exclusions and missing outcome assessments. The handling of missing data for response outcomes is not reported in the registry record. For time-to-event outcomes, the record does specify censoring for participants whose last known status was alive at the cutoff, using last contact before the cutoff.

The registry reports 63 and 73 censored participants for the primary endpoint, but does not provide reasons for censoring, follow-up summaries by group, or sensitivity analyses for alternative censoring assumptions. In general, time-to-event methods rely on assumptions about how censoring relates to the chance of experiencing the event. If those assumptions do not hold, estimates may be affected. The posted counts alone cannot assess whether that occurred.

Other statistical details are also not reported in the registry record, including the original sample-size justification, interim analyses, a detailed analysis plan, allocation concealment, and the handling of protocol deviations. Their absence from the registry record does not show that these procedures were not used; it means they cannot be evaluated from the reported information.

These gaps matter because a p-value and confidence interval are interpreted within a specified analysis plan and data-handling approach. The posted results support a reading of the reported comparisons, but do not allow a complete audit of multiplicity control, missing-data assumptions, or deviations from the planned analysis.

16. 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: pyrexia (4.7%), abdominal pain (4.0%), sepsis (3.6%), vomiting (2.5%), pneumonia (2.5%).

The registry reports serious adverse event summaries for groups at risk of 278 participants in the pegilodecakin plus FOLFOX group and 251 in the FOLFOX group. Serious adverse events affected 123 and 95 participants, respectively. The record also lists other adverse events affecting 271 and 240 participants, and deaths of 217 and 188. These are reported counts; no rates should be inferred from them without additional context about event definitions, observation time, and the relationship between the safety populations and randomized populations.

The listed serious adverse event terms include pyrexia at 4.7%, abdominal pain at 4.0%, sepsis at 3.6%, vomiting at 2.5%, pneumonia at 2.5%, and cholangitis at 2.2%. These percentages describe the listed terms in the registry safety results. The record does not provide enough information in this summary to determine whether participants could contribute to more than one term, how attribution was assessed, or the timing and duration of events.

Safety counts require cautious interpretation. The at-risk groups differ from the randomized arm sizes, and the registry summary does not explain why. The recorded deaths should not be interpreted as treatment-related deaths from these counts alone. The adverse-event information describes events reported in the trial; it does not establish causation or support a treatment recommendation.

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

The results describe a randomized comparison of pegilodecakin plus FOLFOX with FOLFOX alone in participants with metastatic pancreatic cancer after progression during or following a first-line gemcitabine-containing regimen. The primary endpoint was overall survival. The reported hazard ratio is 1.045, and its 95% confidence interval is 0.863 to 1.265. The log-rank p-value is 0.656. These results do not show a clear separation in overall survival in this comparison, while the confidence interval leaves uncertainty about the size and direction of a possible difference.

That is not the same as proving that the groups are identical or that no meaningful difference exists. A failure to establish a difference is not an equivalence finding. The registry record does not report an equivalence or noninferiority margin, and the interval includes values on either side of the hazard-ratio reference value. Precision and the design question both matter when interpreting a result that does not show clear separation.

The reported medians are 5.78 months and 6.28 months, but median comparisons alone do not summarize the full time-to-event evidence. The hazard ratio and its interval are the reported comparative statistics. The trial’s stated follow-up was up to 30 months, and censoring was recorded for participants last known to be alive before the analysis cutoff. The registry does not provide detailed follow-up completeness or sensitivity analyses.

The secondary outcomes offer related but distinct information. Progression-free survival and response measures describe disease progression and response classifications, not the same outcome as death from any cause. Secondary comparisons should be interpreted cautiously because the multiplicity plan is not reported in the registry record. Duration of response and 12-month survival are listed, but their estimates are not reported in the registry record.

These findings apply to the trial’s defined population and assigned regimens. They do not answer questions about other treatment settings, different patient populations, or treatment combinations. Nor do the posted results by themselves establish how an individual should be treated. They are statistical summaries of the randomized trial comparison, with limitations arising from uncertainty and details that are not reported in the registry record.

18. Why This Trial Matters Statistically

This trial illustrates why the endpoint and the treatment comparison must be kept clear. Both randomized groups received FOLFOX, while pegilodecakin was added in one group. The primary question was whether the combination changed overall survival relative to FOLFOX alone in the stated second-line metastatic pancreatic cancer setting. The hazard ratio, interval, and log-rank test address that comparison; arm-specific medians alone do not.

The results also show the value of interpreting an estimate together with its precision. The primary hazard ratio is close to 1, but its confidence interval extends on both sides of 1. A point estimate without its interval could create an impression of certainty that the data do not support. Conversely, an interval containing 1 is not proof of equivalence or proof that all potentially important effects are excluded.

The trial provides examples of distinct statistical measures across endpoint types. Overall and progression-free survival are time-to-event outcomes, for which censoring and follow-up matter. Response outcomes are binary classifications, and their sponsor-reported odds ratios are not interchangeable with unadjusted risk differences. Reading each measure on its own scale helps avoid misleading comparisons.

The record also highlights the importance of analysis-plan information. Multiplicity handling, the original sample-size justification, interim analyses, and several missing-data details are not reported in the registry record. Without those details, the primary posted result can be described, but the full statistical decision process cannot be reconstructed. That distinction is central to responsible interpretation of registry results.

19. 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 registry record does not report baseline characteristics, so the balance of prognostic factors between randomized groups cannot be assessed from these results. It also does not report detailed participant flow, treatment exposure, adherence, reasons for treatment discontinuation, or subsequent anticancer therapies. Those details could help explain how participants contributed to outcomes over follow-up.

The record identifies the study as open-label and does not describe allocation concealment. Overall survival is an objective endpoint, but other outcomes—particularly investigator-assessed response and progression—may depend more on assessment procedures. Independent review, imaging schedules, and adjudication details are not reported in the registry record.

The primary outcome includes censoring for participants last known to be alive before the cutoff, and the registry reports censoring counts. Reasons for censoring, follow-up duration by group, and sensitivity analyses are not reported. These omissions limit assessment of how censoring may affect the estimates. The registry also does not report the full statistical analysis plan, multiplicity procedures, or interim analysis details.

Some listed secondary outcomes do not have estimates reported in the registry record, including duration of response and the 12-month survival rate. The available safety summary gives counts and selected serious adverse event terms, but does not include enough detail to assess event timing, attribution, exposure-adjusted rates, or overlap between event categories.

Finally, the combination-arm dosing description is incomplete in the registry record. That limits the ability to reproduce the regimen from the available description. These limitations concern what can be evaluated from the registry record; they do not establish that unreported procedures or information were absent from the full trial documentation.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Study of Pegilodecakin (LY3500518) With FOLFOX Compared to FOLFOX Alone Second-line Tx in Participants With Metastatic Pancreatic Cancer — NCT02923921.
• Registry record last updated 2020-10-19; status last verified 2020-04-15.
• Results first posted 2020-10-19.
• 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.

23. Record Summary

NCT02923921 is a completed phase 3 randomized, open-label, parallel-group study sponsored by Eli Lilly and Company. It enrolled 567 participants with metastatic pancreatic cancer and compared pegilodecakin plus FOLFOX with FOLFOX alone after progression during or following a first-line gemcitabine-containing regimen.

The primary endpoint was overall survival, measured from randomization to death from any cause. The reported hazard ratio was 1.045, with a 95% confidence interval of 0.863 to 1.265 and a log-rank p-value of 0.656. Median overall survival was 5.78 months and 6.28 months in the respective groups.

The posted results do not show a clear separation in overall survival, and the confidence interval indicates uncertainty around the comparison. The registry does not report several details needed for a complete assessment of the analysis plan, multiplicity, missing data, and trial conduct.