A Study of Nab-Paclitaxel and Gemcitabine With or Without Olaratumab (LY3012207) in Participants With Metastatic Pancreatic Cancer: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study of Nab-Paclitaxel and Gemcitabine With or Without Olaratumab (LY3012207) in Participants With Metastatic Pancreatic Cancer (NCT03086369), a randomized phase 1, phase 2 study. The primary outcome is Phase 1b: Number of Participants With Dose Limiting Toxicities (DLTs).

2. Clinical Question

The central design question in phase 2 was whether adding olaratumab to the chemotherapy backbone changed clinical outcomes compared with adding placebo. The phase 1b question was different: whether dose-limiting toxicities occurred during the first cycle at the studied olaratumab doses. These questions should not be conflated. A count of early toxicities in the phase 1b groups does not establish an efficacy effect in phase 2, and phase 2 outcomes do not replace the dose-finding safety endpoint.

3. Trial Design

The study is titled as a phase 1b open-label and phase 2 randomized, double-blinded study of nab-paclitaxel and gemcitabine with or without olaratumab in first-line metastatic pancreatic cancer. The registry records randomized allocation, double masking and a parallel model. Those design descriptors apply to the randomized comparison; the phase 1b part is expressly described as open-label and its dose groups should not be read as blinded randomized arms.

In phase 1b, participants received olaratumab 15 mg/kg or 20 mg/kg alongside nab-paclitaxel and gemcitabine. A cohort expansion at 20 mg/kg was added following a protocol amendment to confirm safety before phase 2 opened. The primary endpoint population comprised phase 1b participants who received at least one dose of olaratumab.

In phase 2, the two groups received either olaratumab or placebo alongside nab-paclitaxel and gemcitabine. The registry describes olaratumab as a 20 mg/kg loading dose in cycle 1 followed by 15 mg/kg in subsequent cycles. The placebo group received the same chemotherapy backbone. The study is recorded as completed, with a start date of 2017-06-22 and primary completion date of 2021-01-05.

Total enrollment was 184. The registry record does not report the allocation ratio, the sample-size justification for the original design, or detailed procedures for randomization and masking. These should not be inferred from the fact that phase 2 was randomized and double-blinded.

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: Completers included participants who died from any cause.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: All participants in phase 1b who received at least one dose of Olaratumab.

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 participants in phase 1b/2.

6. Primary Endpoint

The registry lists the primary outcome as Phase 1b: Number of Participants With Dose Limiting Toxicities (DLTs), measured over Cycle 1 (Up to 28 days).

The primary endpoint was the number of phase 1b participants with a dose-limiting toxicity during cycle 1, up to 28 days. The registry defines a dose-limiting toxicity as an adverse event considered likely related to study medication or the combination and meeting specified severity criteria. These include febrile neutropenia, severe or prolonged blood-count abnormalities, non-blood toxicity of grade 3 or higher with stated exceptions, and other significant toxicities judged to limit dosing.

This kind of endpoint is used in early-phase research to identify whether notable toxicities arise during an initial treatment period at a studied dose. It is a count, not a rate comparison designed to estimate an efficacy effect. The reported counts were zero in each of the three phase 1b groups: 0 of 3 participants at 15 mg/kg, 0 of 7 at 20 mg/kg, and 0 of 12 in the 20 mg/kg cohort expansion.

Zero observed events means no participant in those groups met the registry’s DLT definition during the specified window. It does not mean that no adverse events occurred: the registry separately records adverse events, including serious events, in these groups. Nor does a zero count establish that the risk of DLT is zero. With small groups and a limited observation window, the count alone gives limited precision about how often such events might occur in a broader population or over longer treatment.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry uses different statistical summaries because the outcomes ask different questions. The primary phase 1b measure is a count of participants with a defined event. Progression-free survival and time to pain worsening are time-to-event outcomes, for which the registry reports medians and hazard ratios. Objective response is a binary outcome summarized as a percentage, with an unadjusted comparison also calculated from posted summary statistics.

What does a median time describe?

For a time-to-event outcome, the median is the time by which half of the participants in the analyzed group have had the event, according to the time-to-event analysis. It is not the average follow-up time, and it does not state when every participant experienced the event. The progression-free survival endpoint is defined from baseline to disease progression or death, up to 26 months. The time-to-first-worsening outcome concerns worsening of the Brief Pain Inventory Short Form Modified worst pain score, with follow-up up to 21 months.

Time-to-event methods can use information from participants whose event has not been observed by the end of their available follow-up, provided their observation is handled as censored. Censoring means that the exact event time is not known within the observed period; it is not itself an event. The registry record does not report enough detail to describe the censoring rules, assessment schedule, or missing-data handling for these outcomes.

How should a hazard ratio be read?

A hazard ratio compares the event rates over time between groups, as represented by the fitted time-to-event analysis. A value of 1 indicates equal estimated hazards; values on either side indicate different estimated hazards, with direction depending on the event and group ordering. It is not a ratio of median times, nor does it directly state the chance that an individual will experience the event. The reported progression-free survival hazard ratio is 1.192, while the time-to-first-worsening hazard ratio is 0.390. Their confidence intervals convey substantial uncertainty for the former and a range below 1 for the latter. Interpretation also depends on the outcome’s definition, the analysis population and model assumptions.

The registry reports log-rank analyses for these time-to-event comparisons. In general, a log-rank test compares the observed and expected event patterns between groups across follow-up. The hazard ratio provides an estimate of relative event rates, while the log-rank p-value assesses compatibility with a null comparison under the test’s assumptions. A p-value does not measure the size or importance of an effect. A p-value above a conventional threshold does not prove no difference, and one below a threshold does not by itself establish that a finding is definitive.

What does the response comparison estimate?

Objective response is the percentage of participants achieving complete or partial response. The unadjusted recalculation compares the proportion with response in each group. A risk difference is the first group’s response proportion minus the comparator’s proportion under the specified ordering. Its confidence interval describes uncertainty in that difference under the calculation’s assumptions. The posted unadjusted phase 2 comparison has a risk difference of -0.033 and a 95% interval from -0.176 to 0.111. Because this interval includes zero, it is compatible with no difference as well as with effects in either direction.

The ORR recalculation uses the posted group summary statistics and is unadjusted. It is not an adjusted model estimate and should not be mistaken for the sponsor’s own analysis. The registry record lists no sponsor-reported ORR analysis. The phase 1b percentages are based on groups of different sizes and were not randomized comparisons; a numerical contrast among those groups would not isolate a dose effect.

Why does over-dispersion matter?

In general, over-dispersion occurs when observed variation exceeds what a statistical model expects. It can make standard errors too small and confidence intervals too narrow if ignored. The primary endpoint is a participant count, and the registry record gives no dispersion estimate or model-based comparison for it. No claim about over-dispersion can be made from the posted count alone; the concept matters when interpreting count models, not as an additional result from this study.

What are the planning illustrations?

The sample-size calculations are planning illustrations based on observed effects, not the study’s original design calculations and not observed power. The stated assumptions are a two-sided significance level of 0.05, equal allocation, and a normal approximation. For time-to-event outcomes, the illustrations express the total number of events needed; for other outcomes they express participants per arm. The displayed scenarios use 100%, 75% and 50% of the observed effect. These are hypothetical target effects. They do not show that the study had a particular probability of detecting an effect, and they are not a reason to treat a non-significant result as evidence of absence.

For progression-free survival, the illustrations range from 1,018 total events for 80% power and 1,363 for 90% power at 100% of the observed effect, to 4,072 and 5,451 at 50%. For the ORR comparison, the corresponding participants-per-arm illustrations range from 3,215 and 4,304 at 100% to 13,032 and 17,445 at 50%. The registry’s stated assumptions and the large illustrative targets emphasize that planning calculations depend strongly on the effect one aims to detect. They should not be read as actual trial enrollment goals or as a retroactive justification of the design.

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("Phase1b: Olaratumab 15 mg/kg + Nab-paclitaxel + Ge", "Phase1b: Olaratumab 20 mg/kg + Nab-paclitaxel + Ge", "Phase1b (Cohort Expansion): Olaratumab 20 mg/kg + ")
n      <- c(3, 7, 12)
events <- c(0, 0, 0)
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  Phase1b: Olaratumab 15 mg/kg + Nab-paclitaxel + Ge  3  0  0.0  0.0  70.8  0.0  56.1  0.6  60.2
ARM  Phase1b: Olaratumab 20 mg/kg + Nab-paclitaxel + Ge  7  0  0.0  0.0  41.0  0.0  35.4  0.3  36.9
ARM  Phase1b (Cohort Expansion): Olaratumab 20 mg/kg +   12  0  0.0  0.0  26.5  0.0  24.2  0.2  24.7

11. Multiplicity, Missing Data and Analysis Populations

Multiplicity concerns the increased chance of obtaining at least one apparently unusual result when many outcomes or comparisons are tested. This trial has a primary phase 1b safety endpoint and a range of secondary endpoints across the phase 1b and phase 2 parts. The registry record does not report a multiplicity strategy, such as a hierarchy of tests, an adjustment procedure, or a plan to control the overall type I error rate. It also does not report whether the time-to-first-worsening analysis was designated as confirmatory or exploratory.

This matters when interpreting a nominal p-value for a secondary endpoint. The p-value of 0.017 for time to first worsening is the reported result for that analysis, but without a documented multiplicity approach it cannot be assumed to represent a multiplicity-adjusted finding. The phase 2 progression-free survival p-value and the unadjusted objective response calculation address separate endpoints and do not resolve that uncertainty. A lack of reported adjustment is not proof that no adjustment was performed; it means the registry record does not establish what was done.

Missing-data methods are also not reported in the registry record. For time-to-event outcomes, the rules governing censoring, withdrawal, and incomplete follow-up affect interpretation. For response outcomes, the handling of participants whose response status is unknown can affect the denominator and estimate. The posted summaries do not provide enough information to assess these procedures or their possible impact.

Analysis populations are not fully described for every reported phase 2 outcome. The primary endpoint population is explicitly phase 1b participants who received at least one dose of olaratumab, but the registry record does not give equivalent detailed population definitions for all secondary analyses. It does not report whether those analyses used an intention-to-treat population, a treated population, or another defined set. These omissions limit assessment of the analysis choices and prevent a full evaluation of how missing observations and protocol deviations were handled.

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: pancreatic cyst (33.3%), pyrexia (33.3%), device related bacteraemia (33.3%), device related infection (33.3%), postoperative wound infection (33.3%).

The registry records serious adverse events separately from the dose-limiting-toxicity endpoint. In phase 1b, serious adverse events were reported for 3 of 3 participants at 15 mg/kg, 3 of 7 at 20 mg/kg, and 8 of 12 in the 20 mg/kg cohort expansion. In phase 2, the counts were 50 of 81 in the olaratumab group and 41 of 78 in the placebo group. These figures describe participants recorded as having serious adverse events; they are not the number of distinct events.

The registry also records other, non-serious adverse events for 3 of 3, 7 of 7 and 12 of 12 participants in the phase 1b groups, and 79 of 81 and 78 of 78 participants in the phase 2 groups. Death counts are listed as 2, 7 and 12 in phase 1b and 60 and 63 in phase 2. These counts alone do not establish cause, timing, or treatment attribution. The registry record does not report enough detail in these summaries to determine whether a particular event or death was caused by study treatment.

Serious-term entries include pancreatic cyst, pyrexia, device-related bacteraemia, device-related infection, postoperative wound infection and septic shock, each listed at 33.3%. The denominator and attribution for those percentages are not reported alongside the terms. Safety counts should therefore be read descriptively, taking account of group size and the study’s adverse-event definitions, rather than as adjusted comparisons of risk.

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

The findings describe what was recorded in this study; they do not establish that one regimen is superior or inferior overall. The primary endpoint was a phase 1b count of dose-limiting toxicities in cycle 1. No such events were recorded in the three phase 1b groups, but the small groups and limited time window mean that the result is not a precise estimate of uncommon or later toxicity.

The randomized phase 2 results provide comparisons for selected efficacy and symptom outcomes. For progression-free survival, the reported confidence interval includes the null value for the hazard ratio, and the p-value does not indicate a statistically clear difference. For response, the unadjusted comparison is also compatible with no difference. These results do not prove that the groups are equivalent: a confidence interval that includes a null value can still include potentially meaningful effects in either direction.

The time-to-first-worsening result has an estimate and confidence interval on one side of the null, with a nominal p-value below 0.05. It remains a secondary outcome, and the multiplicity strategy is not reported in the registry record. It should not be elevated into a definitive conclusion without information about prespecification, the full analysis plan, and other relevant results.

More generally, a treatment effect estimate is not a guarantee about an individual participant. A confidence interval conveys statistical uncertainty under the analysis assumptions; it does not account for every possible bias, measurement issue, or limitation in the record. The registry reports adverse-event counts, but those counts alone do not establish causality or permit a complete benefit-risk assessment.

Nothing in these results is medical advice. They should be understood as registry-reported outcomes from a specific study design and population, not as a recommendation for treatment or a conclusion about every person with metastatic pancreatic cancer. The registry record does not report enough methodological detail to resolve all questions about missing data, analysis populations, or multiplicity.

14. Why This Trial Matters Statistically

This trial is statistically instructive because its primary question and its randomized phase 2 questions are not the same. The primary endpoint is a count of early dose-limiting toxicities in open-label phase 1b groups, while the phase 2 part compares treatment groups using time-to-event and response outcomes. Separating those objectives prevents a safety count from being mistaken for evidence of efficacy.

The study also illustrates why point estimates need uncertainty intervals. The phase 2 progression-free survival estimate is not a stand-alone verdict: its confidence interval spans the null and values on either side. The phase 2 response comparison likewise needs to be read with its interval, rather than relying on the percentages alone. In contrast, the pain-worsening analysis has a nominally notable result, which raises the separate issue of how multiple secondary tests should be interpreted.

Planning calculations based on observed effects are another useful statistical distinction. Their event and participant targets illustrate how demanding a future study could be under the stated assumptions, especially if the target effect is smaller than the observed estimate. They are not observed power, and they do not recreate the original protocol’s design rationale, which is not reported in the registry record.

Finally, the study highlights the difference between a completed trial and a complete statistical account. The registry gives key outcomes but does not report every analysis detail needed to evaluate missing data, analysis populations, and multiplicity. Careful interpretation depends on recognizing both what the posted estimates show and what remains undocumented.

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.

Several limitations specific to this record shape interpretation. First, the primary endpoint comes from phase 1b groups with small participant counts, and it covers cycle 1 only. A zero DLT count in that setting does not characterize the full range of treatment-emergent toxicity over longer exposure. The separate adverse-event counts show that DLTs and adverse events are not interchangeable measures.

Second, the phase 1b dose groups were open-label and were not the randomized phase 2 comparison. Their response percentages should not be interpreted as dose comparisons or combined with phase 2 estimates as if drawn from one randomized population. The 20 mg/kg cohort expansion was added following a protocol amendment, which further distinguishes it from the original phase 1b dose groups.

Third, the phase 2 efficacy results include secondary endpoints, and the registry record does not report a multiplicity plan or complete analysis-population definitions. The ORR comparison is an unadjusted recalculation, with no sponsor-reported analysis listed. The time-to-event summaries do not include all details needed to assess censoring and missing observations. These limitations constrain how strongly the numerical estimates can be interpreted.

Fourth, the registry record does not provide the original sample-size justification, detailed randomization procedures, or a full statistical analysis plan. The planning illustrations are based on observed effects and stated assumptions, so they cannot fill that gap. Finally, safety summary counts do not provide causal attribution or detailed timing. The available information supports descriptive reading of recorded outcomes, not a complete reconstruction of the trial’s conduct and analysis.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: A Study of Nab-Paclitaxel and Gemcitabine With or Without Olaratumab (LY3012207) in Participants With Metastatic Pancreatic Cancer — NCT03086369.
• Registry record last updated 2022-06-28; status last verified 2022-06.
• Results first posted 2022-01-11.
• 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

NCT03086369 is a completed phase 1b/phase 2 study of nab-paclitaxel and gemcitabine with or without olaratumab in first-line metastatic pancreatic cancer. The phase 1b part was open-label and assessed dose-limiting toxicities during cycle 1; none were recorded in the three reported dose groups. The phase 2 part was randomized, double-blinded and parallel, comparing olaratumab plus chemotherapy with placebo plus chemotherapy.

For phase 2, the registry reports progression-free survival, objective response and time to first worsening of worst pain. The first two comparisons do not establish a statistically clear difference; the pain-worsening result is a secondary finding whose multiplicity context is not reported. The registry also lists other secondary outcomes and adverse-event summaries. Missing-data procedures, full analysis populations, multiplicity methods and the original sample-size justification are not reported in the registry record.