Standard of Care Alone or in Combination With Ad-CEA Vaccine and Avelumab in People With Previously Untreated Metastatic Colorectal Cancer QUILT-2.004: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Standard of Care Alone or in Combination With Ad-CEA Vaccine and Avelumab in People With Previously Untreated Metastatic Colorectal Cancer QUILT-2.004 (NCT03050814), a randomized phase 2 study. The primary outcome is Progression Free Survival Between the Standard of Care (SOC) Arm A, and Standard of Care (SOC) Arm B + lead-in.

2. Clinical Question

3. Trial Design

This was a randomized phase II trial sponsored by the National Cancer Institute (NCI). The registry lists no masking, a sequential study model, and an overall enrollment of 30. The study is marked terminated. Its start date is 2017-04-05, and its primary completion date is 2021-08-25. The record does not state why the study was terminated.

The comparison involved standard care alone in Arm A and standard care plus Ad-CEA vaccine and avelumab in Arm B + Lead in. In Arm A, participants received FOLFOX with bevacizumab for up to 12 two-week cycles, followed by maintenance capecitabine and bevacizumab. In the registry’s arm description, Arm B included FOLFOX and bevacizumab with avelumab and Ad-CEA vaccine, followed by maintenance treatment that also included avelumab. The brief summary describes the experimental regimen and schedules in more detail, but the descriptions are not entirely aligned. The registry record does not explain the differences between those descriptions.

The primary outcome was progression-free survival, with a stated timeframe of up to 1.5 years. The outcome record defines progression using RECIST criteria and also counts death as an event. The posted outcome groups contain 10 participants for Arm A and 16 for Arm B + Lead in. The registry does not explain how these outcome counts relate to the overall enrollment of 30, nor does it report the analysis populations or reasons for any difference.

The study was randomized, but the randomization procedure, intended allocation ratio, and any stratification factors are not reported in the registry record. The original sample-size justification and the trial’s planned analysis details are also not reported in the registry record. In particular, the listed planning illustration is based on the observed effect and is not evidence of the study’s original design assumptions.

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:

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: 4 participants that were enrolled but not assigned to a treatment arm had baseline data collected and it is reported in this table. These participants were screen failures who were not treated.

6. Primary Endpoint

The registry lists the primary outcome as Progression Free Survival Between the Standard of Care (SOC) Arm A, and Standard of Care (SOC) Arm B + lead-in, measured over Up to 1.5 years.

The primary endpoint was progression-free survival, reported as a median in months. It measures the time from the start of treatment until disease progression or death, whichever occurs first. Progression was assessed using RECIST: the registry definition includes at least a 20% increase in the sum of target-lesion diameters, using the smallest sum recorded on study as the reference, or the appearance of one or more new lesions.

Time-to-event outcomes are useful when the event can occur at different times for different participants and when the timing of progression or death matters. The median gives a point on the time scale: it is the estimated time by which half of the relevant group has experienced the event, under the method used. It does not describe every participant’s experience or show how the outcome is distributed across the full follow-up period.

The registry reports median progression-free survival of 8.8 months for Arm A and 10.1 months for Arm B + Lead in. It also reports a hazard ratio for the comparison. These measures answer related but distinct questions: medians describe each group’s estimated event-time distribution at its midpoint, while a hazard ratio compares event rates over follow-up under a time-to-event analysis. Neither measure alone explains the full pattern of events, follow-up, or censoring. The registry record does not provide those participant-level details.

7. Results Posted in the Registry

• Standard of Care (SOC) - Arm A vs Standard of Care (SOC) - Arm B + Lead in: hazard ratio 1.06 (95% CI 0.38–2.97), reported p = 0.910 (as reported in the registry).

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is a time-to-event measure. Each participant’s event time begins at treatment start and ends at progression or death, whichever occurs first. When the event has not been recorded by a participant’s last observation, that participant’s follow-up is generally treated as censored in time-to-event methods. General statistical explanation: censoring allows the analysis to use the period during which a participant was observed without assuming that an event occurred after observation stopped. The registry record does not provide the censoring rules or participant-level follow-up information for this trial.

The outcome record lists Kaplan-Meier as the method for the reported comparison. In general, Kaplan-Meier methods estimate the event-time distribution while accounting for participants who have different observed follow-up times. The estimated curve can be used to derive a median event time, when the curve reaches the midpoint. A median is a group-level summary and does not mean that every participant had an event at that time. The registry does not report the plotted curve’s risk sets, event marks, or uncertainty at each time point in this discussion of the outcome.

The sponsor’s comparison is expressed as a hazard ratio of 1.061 for Arm A versus Arm B + Lead in. In general, a hazard ratio compares the instantaneous event rates between groups over the observed follow-up, subject to the assumptions and model used. A value near 1 indicates that the estimated rates are similar at the point estimate, but a point estimate is not a complete account of uncertainty. The registry identifies Kaplan-Meier as the analysis method and reports a hazard ratio, but does not describe the model used to estimate that ratio, its covariates, or any checks of model assumptions. Those details are not reported in the registry record.

The 95% confidence interval for the hazard ratio is 0.380 to 2.966. In general, a confidence interval describes a range of estimates compatible with the data and the method under repeated sampling assumptions. It communicates precision: a broad interval indicates that a range of materially different values remains compatible with the estimate. It is not a range that contains a fixed proportion of individual participants’ outcomes. The interpretation also depends on the analysis assumptions, which are not fully specified in the registry record.

The reported p-value is 0.91. In general, a p-value measures how unusual results at least as incompatible with a specified null hypothesis would be under that hypothesis and the analysis assumptions. It is not the probability that the null hypothesis is true, nor does it measure clinical importance. Here, the registry record does not state the precise null hypothesis, test procedure, or whether the reported p-value is adjusted. The value should therefore be read as the sponsor-reported result, not as a standalone measure of evidence or effect size.

General explanation: a two-sided test considers departures in either direction from its null value. The planning illustration lists a two-sided significance level of 0.05 and a normal approximation. Those are assumptions for the planning illustration, not confirmed details of the trial’s original design. The registry record does not report the original sample-size justification, the original target effect, or the trial’s planned power.

Time-to-event data can be affected by unequal follow-up and censoring. Participants who have not experienced progression or death by the end of observation may still contribute information up to that point. If censoring is related to prognosis in ways not addressed by the analysis, estimates may be affected. This is a general consideration; the registry does not report enough detail to assess whether censoring was independent of outcome in this trial.

Multiplicity refers generally to the increased chance of false-positive findings when many outcomes, comparisons, or analyses are tested. The record lists one primary outcome and identifies a secondary outcome concerning hospitalization because of adverse events attributed to disease progression. It also reports three outcome measures and one analysis, but does not describe a multiplicity strategy or the full planned testing hierarchy. Whether any adjustment was intended is not reported in the registry record.

Analysis populations describe which participants are included in a particular analysis. For example, a protocol may define a population based on randomized assignment or treatment received, but the registry record does not state which approach was used. The primary outcome groups list 10 and 16 participants, while overall enrollment is 30; the record does not explain the inclusion or exclusion of participants from this outcome analysis.

Missing outcome data and withdrawals can affect estimates, especially when the reasons for missingness relate to participants’ outcomes. General methods can include explicit assumptions or sensitivity analyses, but the registry record does not describe missing-data handling, withdrawal handling, or sensitivity analyses for progression-free survival. Interim analyses, stopping rules, stratification, and covariate adjustment are also not reported in the registry record. These absences do not show that such methods were or were not used; they mean the registry does not document them.

10. Confidence Intervals and Precision

The reported hazard ratio is 1.061, with a 95% confidence interval from 0.380 to 2.966. This interval is broad, extending to values that could represent lower or higher event rates depending on the comparison direction. Its breadth means that the point estimate should not be read as a precise estimate of the relative time-to-event contrast. The interval includes 1, the value corresponding to equal hazards under the usual hazard-ratio interpretation.

Confidence intervals express uncertainty in an estimate. In general, a 95% confidence interval is constructed by a method that, under its assumptions and repeated use, would cover the target quantity in 95% of such intervals. It does not mean that there is a 95% probability that the true value lies within this particular interval. The exact interpretation depends on the statistical method and assumptions. The registry record does not give enough detail about the hazard-ratio model to evaluate all of those assumptions.

The arm-specific median estimates also have broad uncertainty: 8.8 months for Arm A, with a 95% confidence interval of 3.3 to 17.0 months; and 10.1 months for Arm B + Lead in, with a 95% confidence interval of 3.6 to 16.1 months. The intervals overlap substantially. Comparing overlap is not a formal test of the difference between medians, and it does not replace the reported hazard-ratio analysis.

The primary-outcome groups list 10 and 16 participants. With group counts of this size, individual event times and censoring can have substantial influence on time-to-event estimates. The registry does not report event counts or the distribution of follow-up, so the contribution of those features to precision cannot be assessed from the record. The confidence intervals are the clearest posted indication of uncertainty, and they show that the estimated effects are not tightly determined.

11. 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 display is a planning illustration based on the observed effect, not the trial’s original design justification. Because the outcome is time-to-event, its figures are total events, not participants per arm. The listed assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation; the event calculations use the Schoenfeld approach. The registry record does not report the trial’s own original sample-size assumptions.

The rows consider planning for 100%, 75%, and 50% of the observed effect. The columns describe the total number of events needed under the illustration for 80% or 90% power. In plain terms, those power targets describe the chance of detecting the specified effect under the planning assumptions if that effect is present. The 90% target requires more events than the 80% target for a given effect level.

As the effect to be detected becomes smaller, distinguishing it from the null becomes more difficult, so the planning illustration requires more events. This is why the totals increase as the assumed fraction of the observed effect decreases. These are event targets, not participant counts; the number of participants needed to generate a given number of events depends on follow-up and event occurrence, which are not specified by this illustration.

The displayed figures are planning quantities, not observed power. They do not measure the probability that the completed trial detected an effect, and they do not establish whether the trial was adequately sized under its original plan. No original sample-size rationale is reported in the registry record.

12. 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 = 1.061, so ln(HR) = 0.0592.
2. d = 4 × (1.96 + 0.842)² / [ln(1.061)]² = 8,954.7, rounded up to 8,955 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 <- 1.061                              # 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(3050814)
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 = 8,955 (80% power), 11,988 (90% power)
 75% of observed effect: events = 15,920 (80% power), 21,312 (90% power)
 50% of observed effect: events = 35,819 (80% power), 47,952 (90% power)
Simulated power with 8,955 events: 0.799 (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.

13. Analyses Reported by the Sponsor

The registry reports the sponsor’s hazard-ratio analysis as 1.061, with a 95% confidence interval of 0.380 to 2.966 and p = 0.91. The primary outcome lists Kaplan-Meier as the method. The same estimate, interval, and p-value appear in the reported analysis, so there is no flagged discrepancy between the posted comparison and the sponsor-reported result.

The statistical reading therefore centers on the sponsor-reported hazard ratio rather than a replacement estimate. The record’s interpretation is limited by the broad confidence interval and by the absence of details about the model, analysis population, and censoring. The trial has small posted outcome-group counts, and the registry does not explain how those counts relate to total enrollment.

The planning illustration is a separate exercise based on the observed effect. It does not revise the sponsor’s analysis, reconstruct the trial’s original sample-size plan, or offer an alternative treatment-effect estimate. The original design assumptions and sample-size justification are not reported in the registry record.

14. Multiplicity, Missing Data and Analysis Populations

The registry identifies progression-free survival as the primary outcome and lists the number of participants hospitalized because of adverse events attributed to disease progression as a secondary outcome. It also records three outcome measures and one analysis. These listings do not establish the full testing plan, the order in which outcomes were assessed, or whether other comparisons were considered.

Multiplicity is a general concern when several outcomes, comparisons, time points, or analyses are tested. With multiple opportunities to find a small p-value, the chance of at least one false-positive result can increase. Some study plans address this by assigning a hierarchy or adjusting significance thresholds. Whether this trial used such an approach is not reported in the registry record. The record also does not state whether the reported p-value was adjusted for multiplicity.

The registry reports overall enrollment of 30 and primary-outcome groups of 10 and 16. It does not describe which participants were included in each analysis, how randomized participants were handled, or the reasons for any difference between enrollment and outcome counts. The analysis populations are not reported in the registry record.

Missing-data methods, handling of withdrawals, and sensitivity analyses are not reported in the registry record. For a time-to-event outcome, the treatment of censoring and incomplete follow-up can affect interpretation; the record does not describe the censoring rules or whether the reasons for incomplete follow-up were examined. Interim analyses, stopping rules, stratification factors, and adjustments for baseline characteristics are also not reported in the registry record. These gaps limit assessment of the analysis plan, but do not by themselves show that a particular procedure was omitted.

The registry names hospitalization due to adverse events attributed to disease progression as a secondary outcome, but results for that outcome are not reported in the registry record. Consequently, the posted primary-outcome comparison does not provide a complete account of the study’s listed outcomes.

15. 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 0% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: febrile neutropenia (20.0%), neutrophil count decreased (20.0%), fatigue (12.5%), infusion related reaction (12.5%), pain (12.5%).

The serious adverse-event summary lists 10 participants at risk in Arm A, of whom 9 were affected, and 16 at risk in Arm B + Lead in, of whom 10 were affected. The registry also lists 10 of 10 participants in Arm A and 16 of 16 in Arm B + Lead in as affected by other adverse events. These are participant counts as reported; the registry record does not give the event-level timing or explain the relationship between these categories.

One death is reported in each arm. The record does not state in this summary whether these deaths were attributed to treatment, disease progression, or another cause. Serious adverse events and other adverse events should not be interpreted as the same category, and the counts do not indicate how many events each participant experienced.

Among the listed serious adverse-event terms, febrile neutropenia and decreased neutrophil count are each reported at 20.0%; fatigue, infusion related reaction, and pain are each reported at 12.5%; and abdominal pain is reported at 10.0%. The registry does not provide the denominator or further event details alongside these term percentages in the information summarized. These figures describe reported safety events in this trial and do not establish comparative safety beyond the posted data.

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

The posted results describe progression-free survival in a randomized comparison of standard care alone with standard care plus Ad-CEA vaccine and avelumab. Median progression-free survival was 8.8 months in Arm A and 10.1 months in Arm B + Lead in. The reported hazard ratio was 1.061, with a 95% confidence interval from 0.380 to 2.966 and p = 0.91. Together, these results show an imprecise estimate of the relative time-to-event contrast.

The results do not establish that the groups had the same outcome. A non-small p-value does not prove equality, and the confidence interval includes a range of possible hazard ratios. The results also do not establish a difference in either direction. The estimate is compatible with materially different relative event rates, and the registry record does not provide further analysis details to narrow that interpretation.

These data should not be interpreted as evidence that either regimen works or does not work. The analysis is a comparison within this trial, and the registry record does not report the full protocol, analysis populations, event counts, or follow-up distribution. Overall enrollment was 30, while the primary-outcome groups list 10 and 16 participants; the reason for that distinction is not documented. The study is marked terminated, but the reason is not reported.

The safety summary records serious adverse events, other adverse events, and one death in each group. Those counts are descriptive and do not establish causation or provide a complete comparative safety assessment. The listed secondary hospitalization outcome has no result reported in the registry record.

In short, the registry contains a primary time-to-event estimate and a safety summary, but substantial uncertainty and reporting gaps remain. The confidence intervals, participant counts, and undocumented analysis details are central to understanding what can and cannot be inferred.

17. Why This Trial Matters Statistically

This trial illustrates why a reported effect estimate needs to be read alongside its uncertainty. A hazard ratio of 1.061 is a single point estimate, but the 95% confidence interval from 0.380 to 2.966 spans a wide range. The point estimate alone could make the comparison seem settled; the interval shows that the data leave considerable uncertainty about the relative event rates.

It also illustrates that median survival summaries and hazard ratios are not interchangeable. The medians describe an estimated midpoint in each group’s time-to-event distribution, whereas a hazard ratio compares event rates over follow-up. Reading the measures together gives more context than treating either as a complete account. Neither replaces information about event counts, follow-up, and censoring, which is not reported in the registry record.

The trial’s overall enrollment is 30, while its primary-outcome groups list 10 and 16 participants. This makes clear why outcome-specific counts and analysis-population definitions matter. Without an explanation of how those counts were assembled, readers cannot determine from the registry record how all enrolled participants contributed to the primary analysis.

The sample-size illustration also demonstrates the distinction between planning and observation. It uses the observed effect to show event targets under stated assumptions; it does not reveal the original design rationale or the trial’s observed power. The original design assumptions are not reported in the registry record. Statistical reporting is most informative when design assumptions, analysis populations, event counts, and uncertainty are all available alongside the headline estimate.

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.

The primary-outcome groups contain 10 participants in Arm A and 16 in Arm B + Lead in, while overall enrollment is 30. The registry record does not explain this difference, define the analysis populations, or state how many participants contributed events. This limits assessment of the completeness of the primary analysis.

Progression-free survival is reported with median estimates and a hazard ratio, but event counts, follow-up distributions, censoring details, and assessment schedules are not reported in the registry record. The exact model used to estimate the hazard ratio and any covariate adjustments are also not described. These omissions make it harder to evaluate the analysis assumptions and the influence of incomplete follow-up.

The study is marked terminated, but the reason for termination and its relation to planned enrollment or follow-up are not reported. The record describes the study as randomized, unmasked, and sequential, but does not explain the randomization procedure, intended allocation ratio, or the meaning of the sequential model. The original sample-size justification, interim-analysis plans, and stopping rules are not reported in the registry record.

The brief summary and arm descriptions do not align perfectly on treatment schedules and maintenance details. The registry record does not explain those differences. The reported secondary outcome concerning hospitalization because of adverse events attributed to disease progression has no result listed in the record. Safety counts are posted, but the summarized information does not specify the attribution of deaths or provide event-level detail.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Standard of Care Alone or in Combination With Ad-CEA Vaccine and Avelumab in People With Previously Untreated Metastatic Colorectal Cancer QUILT-2.004 — NCT03050814.
• Registry record last updated 2021-12-20; status last verified 2021-12.
• Results first posted 2021-12-09.
• 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

NCT03050814 is a randomized, unmasked phase II trial sponsored by the National Cancer Institute (NCI), comparing standard care alone with standard care plus Ad-CEA vaccine and avelumab for previously untreated metastatic or unresectable colorectal cancer. The registry lists enrollment of 30, a sequential model, and terminated status.

The primary outcome was progression-free survival, defined from treatment start to progression or death. Median estimates were 8.8 months for Arm A and 10.1 months for Arm B + Lead in. The sponsor-reported hazard ratio was 1.061, with a 95% confidence interval of 0.380 to 2.966 and p = 0.91.

The estimate is imprecise, and the registry record does not explain the difference between total enrollment and the primary-outcome group counts. Original design assumptions, analysis-population details, and the reason for termination are not reported in the registry record.