A Study of Avastin (Bevacizumab) Added to a Chemotherapeutic Regimen in Patients With Metastatic Pancreatic Cancer: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study of Avastin (Bevacizumab) Added to a Chemotherapeutic Regimen in Patients With Metastatic Pancreatic Cancer (NCT01214720), a randomized phase 3 study. The primary outcome is Duration of Overall Survival - Percentage of Participants With an Event.

2. Clinical Question

3. Trial Design

This was a phase 3 randomized, double-blind, parallel-group trial comparing bevacizumab plus gemcitabine and erlotinib with placebo plus gemcitabine and erlotinib in metastatic pancreatic cancer. Hoffmann-La Roche is listed as the sponsor. The registry records a start date of 2005-07, primary completion in 2008-10, and a completed status.

The total enrollment was 607. For the posted primary outcome, the bevacizumab regimen group had 306 participants and the placebo regimen group had 301. The primary outcome population is identified as ITT. The study summary describes the intended treatment period as continuing until confirmed evidence of disease progression and gives a target sample size of 500+ individuals. The registry record does not report the original sample-size justification or the assumptions used to arrive at that target.

The primary outcome was the percentage of participants with an overall-survival event. The record defines overall survival as the time from randomization to death from any cause. Participants without an event were censored at the date they were last known to be alive. Participants randomized but not exposed to study drug, with no further follow-up, were censored at randomization. The listed primary-outcome assessment schedule includes the treatment-cycle visits, assessments after the last treatment, and follow-up at intervals extending up to 18 months from the last participant randomized.

The registry lists secondary outcomes including clinical benefit response, progression-free survival, response or stable disease at the first postbaseline tumor assessment, and bevacizumab concentration in the presence of gemcitabine and erlotinib. The posted results include five secondary outcome measures. Further information about randomization procedures, allocation concealment, stratification, adherence, and the detailed statistical analysis plan is 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.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: ITT 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: Intent-to-Treat (ITT) Population: All 607 participants randomized to treatment.

6. Primary Endpoint

The registry lists the primary outcome as Duration of Overall Survival - Percentage of Participants With an Event, measured over Randomization, Weeks 1-8 of Cycle 1, Weeks 1-3 of consecutive cycles, 28 days and 3 months after lst treatment, and every 3 months for up to 18 months from last participant randomized.

The primary endpoint is the percentage of participants with an overall-survival event. The registry defines the event as death from any cause and measures survival time from randomization to death. Participants without a recorded event are censored at their last known date alive. Those randomized but not exposed to study drug and without further follow-up are censored at randomization.

Overall survival is a time-to-event outcome, but the posted primary result is expressed as a percentage of participants with an event rather than as a median survival time or a hazard ratio. These summaries answer related but distinct questions: an event percentage describes how many participants experienced the defined event by the available follow-up, while a time-to-event analysis uses the timing of events and censoring. The primary result should therefore be read as the posted event proportions, not as a direct estimate of the length of survival.

General statistical explanation: censoring records that an event was not observed during a participant’s available follow-up; it does not establish what happened after that observation period. Censoring allows time-to-event methods to use partial follow-up, subject to assumptions about how censoring relates to the event. The registry gives the censoring rules, but does not report additional details about the assumptions or methods used for a primary time-to-event analysis.

7. Results Posted in the Registry

• Bevacizumab + Gemcitabine + Erlotinib: 72.2% (221/306) versus 77.4% (233/301) on control; risk difference -5.2 percentage points (95% CI -12.1 to 1.7), risk ratio 0.93 (0.85–1.02); unadjusted two-proportion z-test p = 0.141.

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 (PFS) - Percentage of Participants With an Event

• 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:
  • Bevacizumab + Gemcitabine + Erlotinib: 84.0% (257/306) versus 92.4% (278/301) on control; risk difference -8.4 percentage points (95% CI -13.5 to -3.3), risk ratio 0.91 (0.86–0.96); unadjusted two-proportion z-test p = 0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the proportion with the event is statistically lower for Bevacizumab + Gemcitabine + Erlotinib than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Progression-Free Survival (PFS) - Time to Event

• 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:
  • Bevacizumab + Gemcitabine + Erlotinib vs Placebo + Gemcitabine + Erlotinib: hazard ratio 0.73 (95% CI 0.61–0.86), reported p < 0.001 (as reported in the registry).
• Interpretation:
  • The 95% confidence interval excludes no difference (a ratio of 1), so the hazard of the event is statistically lower for Bevacizumab + Gemcitabine + Erlotinib vs Placebo + Gemcitabine + Erlotinib than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Percentage of Participants With Complete Response (CR), Partial Response (PR), or Stable Disease (SD) at First Postbaseline Tumor Assessment

• 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:
  • Bevacizumab + Gemcitabine + Erlotinib: 62.1% (190/306) versus 58.5% (176/301) on control; risk difference 3.6 percentage points (95% CI -4.2 to 11.4), risk ratio 1.06 (0.93–1.21); unadjusted two-proportion z-test p = 0.362.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Bevacizumab + Gemcitabine + Erlotinib and the control. The interval gives the range of effects that remain compatible with the data.

The secondary outcomes provide information about disease progression, tumor response or stable disease, and other measures beyond the primary overall-survival event percentage. Their overall pattern is mixed rather than uniformly aligned with the primary comparison. The PFS event percentage and the reported PFS time-to-event comparison favor the bevacizumab regimen in direction, while the response-or-stable-disease comparison is not clearly separated between groups by its reported interval. These outcomes measure different endpoints and should not be treated as interchangeable confirmations of one another.

The PFS time-to-event result is the secondary outcome with a reported log-rank analysis and a hazard ratio whose 95% confidence interval lies below 1. The response analysis has a reported chi-squared comparison, but its confidence interval includes no difference. The primary result itself is an unadjusted recalculation from posted counts and has an interval that includes zero. This contrast illustrates why interpretation should consider the endpoint hierarchy and the type of analysis rather than selecting whichever outcome has the strongest statistical signal.

Unadjusted secondary results carry less weight for confirmatory interpretation when many outcomes are examined. General statistical explanation: testing multiple outcomes increases the chance that at least one result appears statistically unusual by chance, even if all underlying null hypotheses are true. A prespecified multiplicity procedure can control that risk, but such a procedure is not reported in the registry record. Without that information, secondary p-values should be regarded as nominal evidence rather than as definitive, multiplicity-controlled conclusions.

The registry lists additional secondary measures, including clinical benefit response and bevacizumab concentration, but the posted outcome summaries do not include results for every measure in the list. The number and scope of outcome measures make it particularly important not to interpret the reported secondary findings in isolation. Overall, the secondary data add context on PFS and response-related outcomes, but do not resolve the uncertainty in the posted primary overall-survival event comparison.

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

• Clinical Benefit Response (CBR)
• Progression-Free Survival (PFS) - Percentage of Participants With an Event
• Progression-Free Survival (PFS) - Time to Event
• Percentage of Participants With Complete Response (CR), Partial Response (PR), or Stable Disease (SD) at First Postbaseline Tumor Assessment
• Bevacizumab Concentration in the Presence of Gemcitabine and Erlotinib

9. Statistical Interpretation

10. Statistical Methods Explained

The posted results include binary event proportions, a time-to-event comparison, interval estimates, and a sample-size planning illustration. Each method summarizes a different feature of the trial data. The primary outcome is a binary event percentage; the PFS time-to-event result is reported using a hazard ratio and a log-rank test; and the response outcome is compared as a difference in response rates. Methods and assumptions not described in the registry record cannot be inferred from these summaries.

What does an event percentage describe?

An event percentage is the proportion of participants in a group who experienced the event defined for that outcome. For overall survival, the registry defines the event as death from any cause. For PFS, the posted outcome is described as the percentage of participants with an event, but the event definition is not included in the available outcome description. These percentages are straightforward to read as group summaries, but they do not convey when events occurred or how long participants were followed.

General statistical explanation: a risk difference subtracts one group’s event probability from the other’s, according to the order chosen for the comparison. A value below zero means the event probability is lower in the first group when the difference is calculated as first group minus comparator; a value above zero means it is higher. For an outcome where an event is adverse, the direction may be easy to describe, but the estimate alone does not establish why the groups differed or whether the difference is important in clinical practice. The primary risk difference is an unadjusted recalculation from the posted summary counts.

How should a confidence interval be understood?

General statistical explanation: a confidence interval is a range of estimates generated by a procedure intended to have a stated coverage rate over repeated samples, if its assumptions hold. It communicates uncertainty around an estimate and is not a range containing a fixed proportion of individual participant outcomes. When a difference interval includes zero, the data do not rule out no difference at the interval’s stated confidence level. The primary recalculated interval is listed without a confidence level in the registry summary, so a confidence percentage should not be assigned to it based on that summary alone.

The PFS time-to-event result reports a hazard ratio of 0.73, a 95% confidence interval from 0.61 to 0.86, and a log-rank p-value of < 0.001. General statistical explanation: a hazard ratio compares event rates over time between groups under the model or method used; it is not the ratio of the groups’ median times, nor does it directly state an absolute difference in event percentages. The reported confidence interval is wholly below 1, indicating that the range of hazard-ratio values compatible with the reported interval lies below the no-relative-difference value. The registry names the log-rank method, but further details of the time-to-event analysis are not reported.

Why do binary outcomes need careful interpretation?

Binary outcomes reduce each participant’s follow-up to whether the defined event was observed. They can be useful for a clear group comparison, but the result depends on the event definition, observation window, and completeness of follow-up. The primary outcome description includes censoring rules, yet the primary summary is an event percentage. That percentage should not be read as a survival curve or as a substitute for a time-to-event analysis. The posted PFS event percentage also represents a separate outcome and should not be combined with the overall-survival event percentage.

For the response-or-stable-disease endpoint, the registry reports a 95% confidence interval for the difference in response rates and a chi-squared analysis. The reported difference is 3.62, with a 95% confidence interval from -4.3 to 11.6 and p = 0.362. The unadjusted recalculation gives a risk difference of 0.036, an interval from -0.042 to 0.114, and p = 0.362. These are expressed on different scales in the registry and the recalculation: the reported difference is in percentage points, while the recalculated risk difference is on the probability scale. The registry-reported analysis should be distinguished from the unadjusted recalculation.

What does over-dispersion mean?

General statistical explanation: over-dispersion occurs when observed variation is greater than a statistical model expects. It can make standard errors too small and confidence intervals too narrow if the model assumes less variability than the data contain. The primary outcome is a binary percentage, and no dispersion estimate or over-dispersion assessment is reported for it. No claim about over-dispersion or a model adjustment is justified from the posted primary summary.

Finally, the analysis population matters. The registry identifies the primary population as ITT, a framework that generally analyzes participants according to randomized assignment. The record does not describe all population-specific exclusions or analysis details. The posted counts and outcome summaries should therefore be interpreted using the registry’s stated population label, without inferring additional analysis rules.

11. Confidence Intervals and Precision

The primary event percentages are 72.2% for the bevacizumab regimen and 77.4% for the placebo regimen. Their unadjusted recalculated risk difference is -0.052, with interval limits of -0.121 and 0.017. Because that interval crosses zero, the posted primary comparison is imprecise enough to include no difference as well as differences in either direction. The registry summary does not state a confidence level for this recalculated interval, so it is best described simply as the reported interval estimate.

The PFS time-to-event result has a hazard ratio of 0.73 and a 95% confidence interval from 0.61 to 0.86. This interval is narrower in its conclusion about direction than the primary interval: its reported limits are both below 1, the no-relative-difference value for a hazard ratio. The response-rate comparison, by contrast, has a 95% confidence interval for the reported difference from -4.3 to 11.6 and p = 0.362; that interval spans zero.

General statistical explanation: precision concerns how narrowly an estimate is bounded by uncertainty, not whether an estimate is important or free from bias. Wider intervals usually leave a broader range of values compatible with the data; narrower intervals leave a smaller range, conditional on the method and assumptions. Sample size, event frequency, variability, and analysis choices can affect interval width. The registry record does not report enough detail to attribute the observed precision to any one factor.

Precision should also be kept separate from the p-value. A p-value summarizes compatibility with a null hypothesis under a test; it is not an estimate of effect size or a measure of the probability that the null is true. The effect estimate and its interval give more direct information about the range and direction of plausible group differences. For the primary outcome, the interval’s inclusion of zero is central to interpreting the unadjusted comparison.

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 sample-size illustrations are planning calculations based on stated fractions of the observed effect, not the trial’s original design calculation. For the binary outcomes, the figures are participants per arm for 80% power and participants per arm for 90% power. For the time-to-event outcome, the figures are total events rather than participants. These units matter: an event target is not interchangeable with an enrollment target because the number of events depends on follow-up and event occurrence.

The listed planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. Each row considers detecting the stated fraction of the observed effect. The 80% power column means the planning calculation is designed to have an 80% chance of detecting an effect of that specified size under the assumptions, if that effect is real. The 90% power column describes a higher planned chance of detecting that specified effect, and therefore generally requires more information. These are design targets, not guarantees about any future study.

The rows for 100%, 75%, and 50% of the observed effect illustrate how planning requirements change when the effect to detect is smaller. A smaller target effect requires more participants per arm for binary outcomes, or more total events for time-to-event outcomes, under the listed assumptions. The illustrations are therefore sensitive to the chosen effect size and should not be mistaken for observed trial results.

These figures are not observed power. They do not measure the probability that the completed trial’s result is true, nor do they explain the trial’s original sample-size choice. The original design assumptions and justification are not reported in the registry record. The sample-size curve should be read as a visual expression of the same planning relationship between target effect and required information, rather than as evidence of what the completed trial achieved.

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

n is the number of participants in each arm, p₀ and p₁ are the control and treatment proportions, and p̄ is their average.

Worked calculation for this trial

1. Control proportion p₀ = 0.774; treatment proportion p₁ = 0.722; average p̄ = 0.748.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.6139 = 1.2032.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.6128 = 0.5157.
4. n = (1.2032 + 0.5157)² / (-0.0519)² = 1,098.4, rounded up to 1,099 participants per arm 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
p0 <- 0.774086378737541                          # control proportion
p1 <- 0.722222222222222                          # treatment proportion
n_per_arm <- function(power, frac = 1) {
  q1 <- p0 + (p1 - p0) * frac               # smaller effect = fraction of the observed difference
  pbar <- (p0 + q1) / 2
  z_b <- qnorm(power)
  ceiling(round((z_a * sqrt(2 * pbar * (1 - pbar)) + z_b * sqrt(p0 * (1 - p0) + q1 * (1 - q1)))^2 / (q1 - p0)^2, 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

# Check against R's built-in function (unrounded n per arm)
cat("power.prop.test:", round(power.prop.test(p1 = p0, p2 = p1, power = 0.80)$n, 1), "\n")

# Simulation: how often does the z-test reject at the 80%-power sample size?
set.seed(1214720)
n    <- n_per_arm(0.80)
reps <- 20000
x0 <- rbinom(reps, n, p0); x1 <- rbinom(reps, n, p1)
ph <- (x0 + x1) / (2 * n)
z  <- ((x1 - x0) / n) / sqrt(ph * (1 - ph) * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(z) > z_a, na.rm = TRUE), reps))

Output from running the code

100% of observed effect: n per arm = 1,099 (80% power), 1,470 (90% power)
 75% of observed effect: n per arm = 1,920 (80% power), 2,570 (90% power)
 50% of observed effect: n per arm = 4,244 (80% power), 5,680 (90% power)
power.prop.test: 1098.4 
Simulated power at n = 1,099 per arm: 0.796 (20000 replicates)

The simulation draws many pairs of event counts at the planned sample size and counts how often the two-proportion z-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

14. Multiplicity, Missing Data and Analysis Populations

The registry lists multiple secondary outcomes, and the posted results include several distinct comparisons. No multiplicity adjustment, gatekeeping procedure, hierarchical testing plan, or other control of the family-wise false-positive risk is reported in the registry record. Therefore, the p-values for secondary outcomes should be read as unadjusted unless a reported analysis explicitly states otherwise. The record’s lack of a reported multiplicity strategy does not establish that no strategy existed; it means that the strategy cannot be evaluated from this record.

Missing-data information is also limited. For overall survival, the record specifies how participants without an event are censored and how randomized participants without exposure or further follow-up are censored. These are stated rules for the primary outcome, but the registry does not report the extent of missing follow-up, the reasons for missingness, sensitivity analyses, or a broader missing-data plan. The PFS and response descriptions include assessment schedules, but do not give a complete account of missing assessments or how they were handled.

The primary population is identified as ITT. This is an important label because it indicates that the primary analysis is associated with randomized assignment, but the record does not give further details of the population definition, exclusions, or implementation. It does not report a per-protocol analysis or a detailed statistical analysis plan. The number of analyses listed in the registry is not enough to determine whether all planned analyses were reported or whether the displayed results were prespecified.

General statistical explanation: multiplicity and missingness affect different parts of interpretation. Multiplicity concerns the increased opportunity for chance findings when multiple hypotheses are tested. Missingness concerns whether incomplete observations could distort estimates, depending on why data are absent and how the analysis handles them. Neither issue can be resolved from p-values alone. Their assessment requires details about prespecification, data availability, analysis decisions, and sensitivity checks, which are not reported in the registry record.

15. 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 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 (5.4%), pulmonary embolism (4.2%), pneumonia (3.0%), sepsis (2.7%), vomiting (2.7%).

The registry reports serious adverse events for 297 participants at risk in the bevacizumab regimen group, with 139 affected, and for 286 participants at risk in the placebo regimen group, with 119 affected. It also reports other adverse events for those same at-risk group sizes: 290 affected in the bevacizumab regimen group and 276 in the placebo regimen group. These are counts, not percentages, and no additional percentage calculations are made.

The listed serious adverse event terms and percentages include pyrexia at 5.4%, pulmonary embolism at 4.2%, pneumonia at 3.0%, sepsis at 2.7%, vomiting at 2.7%, and deep vein thrombosis at 1.7%. The registry summary does not identify these listed percentages by treatment group in the information available, so they should not be assigned to one group or used to compare groups.

These figures describe recorded adverse events in the registry and are not a complete safety analysis. The death counts are not reported in the registry record. The record does not provide the full event definitions, severity breakdown, exposure-adjusted rates, or details needed to interpret differences in event burden. The at-risk denominators also differ from the primary outcome group sizes, so the safety counts should be read in their own context.

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

The primary result describes a lower observed overall-survival event percentage in the bevacizumab regimen group than in the placebo regimen group. The unadjusted recalculated risk difference is negative, but its interval includes zero and the unadjusted p-value is 0.141. Statistically, this leaves uncertainty about the direction and size of the group difference. The result is not evidence that the two regimens are identical, and it does not establish a clear difference in the primary endpoint.

The primary endpoint is posted as an event percentage, not a time-to-event estimate. Although the registry defines overall survival using time from randomization to death, the available primary result does not report a median, hazard ratio, or survival curve. It therefore cannot establish how event timing differed between groups. Censoring is described, but the record does not report enough detail to assess follow-up patterns or their influence on the primary summary.

The secondary PFS findings point in a more favorable direction for the bevacizumab regimen, including a reported hazard ratio with an interval below 1. However, PFS is not the primary endpoint, and the registry does not report a multiplicity strategy. The response comparison is also not clearly separated by its reported interval. These results should be considered as separate secondary findings rather than combined into a single conclusion about overall survival.

General statistical explanation: statistical significance, precision, and clinical importance are different concepts. A p-value assesses compatibility with a null hypothesis under a test; an interval describes uncertainty around an estimate; and clinical importance requires context not established by statistical results alone. The registry record does not provide a threshold for a clinically important difference or enough information to determine the practical meaning of the estimates.

These results do not support medical advice. They describe the posted outcomes for the randomized groups under the registry’s definitions and analyses. Conclusions beyond those data would require fuller information about the protocol, analysis plan, missing data, and the clinical context.

17. Why This Trial Matters Statistically

This trial illustrates why the outcome definition and the reported effect measure must be read together. The primary outcome is overall survival, defined using time from randomization to death, yet the posted primary summary is an event percentage. A percentage of participants with an event and a time-to-event estimate answer different questions. Readers should not assume that one summary gives the timing information of the other.

It also shows why an effect estimate and its uncertainty are more informative than a point estimate alone. The primary event percentages differ in observed direction, but the recalculated risk-difference interval includes zero. The secondary PFS time-to-event result has a different pattern and a reported interval below the hazard-ratio null value. Looking at these measures side by side makes the endpoint hierarchy, differing outcome definitions, and multiplicity concerns important to interpretation.

The sample-size illustrations provide another useful statistical distinction: planning for a future study is not the same as evaluating the completed study. The illustrations use the observed effects and stated assumptions, whereas the trial’s original design assumptions are not reported in the registry record. Their requirements change as the target effect changes, and the time-to-event calculations use total events rather than participants per arm.

Finally, the registry’s reported details establish some important features—randomization, double masking, the ITT label for the primary analysis, and the censoring definition—but leave other methods and data-handling details unreported. This is a practical example of how to separate what a registry result directly documents from what cannot be concluded without a protocol or fuller analysis report.

18. 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.
• Participant counts were reconstructed from posted percentages and arm sizes, so small rounding differences are possible.

The registry record does not report the original sample-size justification, detailed statistical analysis plan, randomization procedures, allocation concealment, stratification factors, or interim analyses. It also does not state whether the primary endpoint was analyzed using a time-to-event method in addition to the posted event percentage. These omissions limit assessment of how closely the reported analyses correspond to the planned analyses.

The primary comparison is an unadjusted recalculation from posted summary counts rather than a reported sponsor analysis. The available primary summary does not state a confidence level for its interval and does not provide a primary p-value from a reported analysis. The event percentage also does not convey event timing, and the registry does not report a primary overall-survival hazard ratio or median.

The registry lists multiple secondary outcomes, but does not report a multiplicity strategy. Results for all listed outcomes are not present in the posted summaries. The outcome descriptions also vary in detail; for example, the PFS event percentage is given without a full event definition in the available information. Missing follow-up and assessment data, reasons for missingness, and sensitivity analyses are not reported, beyond the stated overall-survival censoring rules.

Safety information is limited to counts of participants with serious and other adverse events, plus selected serious-event terms and percentages. The listed term percentages are not attributed to arms in the available summary, and deaths are not reported. These limitations prevent a complete comparison of adverse-event profiles from these figures alone. Finally, the registry summaries cannot by themselves establish the clinical importance of observed differences or address aspects of trial conduct that are not recorded.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: A Study of Avastin (Bevacizumab) Added to a Chemotherapeutic Regimen in Patients With Metastatic Pancreatic Cancer — NCT01214720.
• Registry record last updated 2014-08-13; status last verified 2014-07.
• Results first posted 2014-08-13.
• 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.

22. Record Summary

NCT01214720 is a completed phase 3 randomized, double-blind, parallel-group study sponsored by Hoffmann-La Roche. It enrolled 607 participants with metastatic pancreatic cancer and compared bevacizumab plus gemcitabine and erlotinib with placebo plus gemcitabine and erlotinib.

The primary outcome was the percentage of participants with an overall-survival event, defined as death from any cause. The posted event percentages were 72.2% and 77.4%, respectively. The unadjusted recalculated risk difference interval includes zero, leaving uncertainty about the primary comparison. A secondary PFS time-to-event analysis reported a hazard ratio of 0.73, with a 95% confidence interval from 0.61 to 0.86 and p < 0.001.

The record identifies the primary population as ITT and gives censoring rules, but does not report the original sample-size justification, multiplicity strategy, or several details needed to assess missing data and analysis plans. The secondary and safety summaries add context but do not remove those limitations.