To Evaluate Sipuleucel-T Manufactured With Different Concentrations of (PA2024) Antigen: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for To Evaluate Sipuleucel-T Manufactured With Different Concentrations of (PA2024) Antigen (NCT00715078), a randomized phase 2 study. The primary outcome is Cumulative CD54 Upregulation Ratio Between Each of the Cohorts..

2. Clinical Question

This is a question about a laboratory measure of the manufactured cell product, not a direct comparison of cancer outcomes. The primary outcome was defined as the sum of the final-product values across Infusion 1 through Infusion 3 for each infused subject. The endpoint description says subjects who received all 3 infusions were included in the analysis.

The registry reports comparisons of Cohort B and Cohort C with Cohort A. It does not report a direct comparison between Cohort B and Cohort C. The posted results therefore address whether each of those cohorts differed from Cohort A under the reported analyses; they do not establish which of all three concentrations is preferable for a clinical purpose. Clinical outcome comparisons are not reported in the registry record.

3. Trial Design

This was a randomized, multicenter, single-blind, parallel-group Phase 2 study. It was conducted in men with metastatic androgen independent prostate cancer. The registry record lists Dendreon as the sponsor, an enrollment of 122, a start in 2008-10, primary completion in 2012-04, and a completed status.

Participants were assigned to Cohort A, B, or C, with sipuleucel-T manufactured at different concentrations of PA2024 antigen. The concentrations were 10 μg/mL for Cohort A, 5 μg/mL for Cohort B, and 2 μg/mL for Cohort C. The cell suspension concentration was 1 x 10^7 PBMCs per mL in each cohort. The described regimen was 3 infusions at 2-week intervals.

The record identifies randomization and single blinding but does not report the randomization sequence, allocation concealment, or which people were blinded. The allocation ratio is not reported in the registry record. The arm descriptions give the infusion schedule and formulation concentrations; additional details about treatment delivery or adherence are not reported in the registry record.

The endpoint results list 40 participants in Cohort A, 36 in Cohort B, and 37 in Cohort C. The primary endpoint description specifies inclusion of subjects who received all 3 infusions, but the registry record does not provide a complete account of participant flow or explain the difference between the enrollment and the outcome group counts. The original sample-size rationale and the trial's planned statistical analysis document are not reported in the registry record.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Cumulative CD54 upregulation ratio will be the primary end point and calculated as the sum of Infusion 1 through Infusion 3 final product values for each infused subject.

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.

6. Primary Endpoint

The registry lists the primary outcome as Cumulative CD54 Upregulation Ratio Between Each of the Cohorts., measured over Baseline, Months 2, 4 and 6..

The primary endpoint was the cumulative CD54 upregulation ratio between the cohorts. CD54 upregulation is described as the fold increase in the final product (FP) from the BDS65 step. BDS65 refers to the sample collected after BDS77 and BDS65, but before ex vivo culture with PA2024 antigen; FP is the sample collected after ex vivo culture. The unit is given as the ratio of CD54 molecules on BDS65:FP cells.

For each infused subject, the cumulative measure was defined as the sum of the final-product values from Infusion 1 through Infusion 3. The timeframe lists baseline and Months 2, 4 and 6. The endpoint description says the analysis included subjects who received all 3 infusions. Thus, this outcome summarizes a product-related laboratory response across the infusion series, rather than measuring a patient-centered clinical outcome.

CD54 upregulation was used to characterize the manufactured cell product in relation to the different PA2024 concentrations. Comparing the endpoint across cohorts addresses whether the concentration groups differed on this laboratory measure. The registry record does not explain the biological or clinical threshold that would make a difference in this measure important, nor does it report a validated threshold linking the endpoint to patient benefit.

7. Results Posted in the Registry

• Cohort B: mean 30.94 versus 29.53 on control; difference 1.41 (95% CI -2.34 to 5.16), standardized effect d = 0.17; z-test p = 0.461.
• Cohort C: mean 26.67 versus 29.53 on control; difference -2.86 (95% CI -6.30 to 0.58), standardized effect d = -0.37; z-test p = 0.103.

8. Statistical Interpretation

9. Statistical Methods Explained

What does the sponsor's ANOVA compare?

The endpoint description says the primary analysis used an analysis of variance model for the log-transformed cumulative CD54 upregulation ratio, with antigen concentration cohort as the independent variable. In general, analysis of variance compares group means while accounting for variation among observations. Here the analysis was performed on a logarithmic scale, so the reported contrasts are expressed as ratios of geometric means rather than differences between arithmetic means.

A geometric mean is a summary suited to positive measurements whose differences are naturally interpreted proportionally. After a logarithmic analysis is translated back to the original scale, a ratio of 1 represents equal geometric means; a ratio above or below 1 represents a proportional difference in one direction or the other. The reported Cohort B to Cohort A ratio was 1.046, and the Cohort C to Cohort A ratio was 0.907. These ratios describe the endpoint on the sponsor's analysis scale. They do not quantify a clinical benefit.

What is the role of the unadjusted recalculation?

The recalculated comparisons use the posted means and standard errors to express each contrast as a difference on the original outcome scale. Cohort B compared with Cohort A has a recalculated mean difference of 1.41, with p = 0.461. Cohort C compared with Cohort A has a recalculated mean difference of -2.86, with p = 0.103. These are simple, unadjusted comparisons, not the log-transformed ANOVA model reported by the sponsor.

In general, standard error describes the uncertainty in an estimated mean, while the pooled standard deviation used in an unadjusted comparison describes variability among individual observations across the compared groups. A mean difference and a ratio of geometric means answer related but distinct statistical questions. Their numerical values cannot be compared as if they were on the same scale. The recalculation is an additional descriptive analysis, not a correction to the sponsor's result.

How should the p-values be understood?

A p-value describes how compatible the observed data are with a specified null model, under the assumptions of the analysis. It is not the probability that the null hypothesis is true, nor does it measure the size or clinical importance of an effect. The sponsor's p-values were 0.502 for Cohort B versus Cohort A and 0.144 for Cohort C versus Cohort A. The recalculated p-values were 0.461 and 0.103, respectively. The registry record does not report the analysis's prespecified decision threshold, the test sidedness, or a multiplicity adjustment for these comparisons.

For a continuous outcome, a confidence interval gives a range of values compatible with the data and analysis assumptions. The sponsor's intervals are reported at 90% confidence and are on the ratio scale. The recalculated intervals are on the mean-difference scale; their confidence level is not stated in the registry record. An interval that includes the null value indicates that the data are compatible with no difference as well as with differences in either direction at that interval's level. It does not demonstrate equivalence. Equivalence or non-inferiority requires a prespecified margin and a design and analysis suited to that question; the margin is not reported in the registry record.

Why does the analysis population matter?

The endpoint description states that subjects who received all 3 infusions were included. That makes the endpoint analysis conditional on completing the infusion series, according to the stated description. The record does not report whether other analysis populations were used, how many participants were excluded from this endpoint analysis, or how missing outcome observations were handled. In general, excluding participants after randomization can affect comparability if the reasons for exclusion are related to outcomes or treatment group. Without participant flow and missing-data details, the extent of any such effect cannot be assessed.

The group summaries show counts of 40, 36, and 37 for Cohorts A, B, and C. The adverse-event information separately lists 40 at risk in each cohort. These counts concern different reported result sets and should not be assumed to represent an identical analysis population. The registry record does not explain the relationship between those counts.

What does the trial label mean for interpretation?

The reported analyses are pairwise comparisons of Cohort B with Cohort A and Cohort C with Cohort A. The record does not report a direct contrast between Cohorts B and C. When multiple comparisons are made, the chance of at least one false-positive finding can increase unless the analysis accounts for multiplicity. Whether an adjustment or a hierarchical testing plan was used is not reported in the registry record. The reported outcome type is labelled as non-inferiority or equivalence, but the margin and the detailed hypotheses are not reported. The label alone is insufficient to conclude that non-inferiority or equivalence was established.

General statistical interpretation also depends on model assumptions, including appropriate handling of the distribution of the outcome and independence of observations. The registry describes the ANOVA model and log transformation but does not report diagnostics, model checking, or sensitivity analyses. The posted results therefore allow a reading of the estimates and their uncertainty, but not a full assessment of every assumption behind the model.

10. Confidence Intervals and Precision

The sponsor reported 90% confidence intervals for the geometric-mean ratios. For Cohort B compared with Cohort A, the ratio was 1.046 and the interval extended from 0.936 to 1.168. For Cohort C compared with Cohort A, the ratio was 0.907 and the interval extended from 0.813 to 1.013. Both ranges include 1, the null value for a ratio. The range for the Cohort B comparison allows values below and above 1; the Cohort C range also reaches slightly above 1. These results leave uncertainty about the direction and size of the underlying differences.

The unadjusted recalculated mean differences were 1.41 for Cohort B versus Cohort A and -2.86 for Cohort C versus Cohort A. Their reported intervals were -2.341 to 5.161 and -6.298 to 0.578, respectively. Both include zero, the null value for a difference. The confidence level for these recalculated intervals is not reported in the registry record, so they should not be described as having the same confidence level as the sponsor's intervals.

In general, a confidence interval communicates both an estimate and its statistical precision. A wider interval permits a broader range of effects; a narrower one is more precise, conditional on the analysis assumptions. Here, the ranges do not isolate a single direction of difference for either pairwise comparison. They also do not establish that the cohorts are equivalent: that conclusion would require an appropriate prespecified equivalence margin and interpretation against it, which is not reported in the registry record.

The group standard errors are 1.285 for Cohort A, 1.418 for Cohort B, and 1.194 for Cohort C. Standard errors indicate uncertainty in the group mean estimates, not the spread of individual participant measurements. The underlying individual-level distributions and additional precision details are not reported in the registry record.

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 planning figures are participants per arm, not total events. They illustrate the arm size that would be needed under the stated planning assumptions to detect a chosen fraction of the observed effect in a two-arm comparison. The assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The original trial's own sample-size justification and design assumptions are not reported in the registry record.

The 80% power column describes a planning target where, if the assumed effect and other planning assumptions held, the calculation would give an 80% chance of detecting that effect under the specified test. The 90% power column uses a higher probability target under the same type of assumptions. Power is conditional on the assumed effect size, analysis, and other planning choices; it is not a guarantee that a future study will detect a difference.

For the Cohort B versus Cohort A comparison, the planning row using 100% of the observed effect gives 546 participants per arm for 80% power and 730 participants per arm for 90% power. Planning for 75% of the observed effect gives 970 and 1,298 participants per arm, respectively; planning for 50% gives 2,181 and 2,920. For Cohort C versus Cohort A, the corresponding figures are 115 and 154 for 100% of the observed effect, 204 and 273 for 75%, and 459 and 614 for 50%.

Smaller effects are harder to distinguish from random variation, so the planning figures increase as the assumed fraction of the observed effect gets smaller. These are planning illustrations based on observed effects, not observed power. They do not show the power of the completed trial or establish that the trial's original design used these assumptions.

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

n is the number of participants in each arm, Δ is the difference in means to detect and σ is the common standard deviation.

Worked calculation for this trial

1. Observed difference Δ = 1.410; pooled standard deviation σ = 8.309.
2. n = 2 × (1.96 + 0.842)² × 8.309² / 1.410² = 545.2, rounded up to 546 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
delta <- 1.41                      # observed difference in means
sigma <- 8.30940792114577                     # pooled standard deviation
n_per_arm <- function(power, frac = 1) {
  ceiling(round(2 * (z_a + qnorm(power))^2 * sigma^2 / (delta * frac)^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 (t-based, so very slightly larger)
cat("power.t.test:", round(power.t.test(delta = delta, sd = sigma, power = 0.80)$n, 1), "\n")

# Simulation: exact two-sample t-test using simulated means and variances
set.seed(715078)
n    <- n_per_arm(0.80)
reps <- 20000
m_diff <- rnorm(reps, delta, sigma * sqrt(2 / n))             # difference in sample means
s2     <- sigma^2 * rchisq(reps, 2 * (n - 1)) / (2 * (n - 1))   # pooled sample variance
tstat  <- m_diff / sqrt(s2 * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(tstat) > qt(0.975, 2 * (n - 1))), reps))

Output from running the code

100% of observed effect: n per arm = 546 (80% power), 730 (90% power)
 75% of observed effect: n per arm = 970 (80% power), 1,298 (90% power)
 50% of observed effect: n per arm = 2,181 (80% power), 2,920 (90% power)
power.t.test: 546.1 
Simulated power at n = 546 per arm: 0.805 (20000 replicates)

The simulation draws many sets of group means and variances at the planned sample size and counts how often the two-sample t-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.

13. Analyses Reported by the Sponsor

The sponsor's reported primary comparisons used ANOVA on the log-transformed cumulative CD54 endpoint and expressed the results as ratios of geometric means. The reported estimates were 1.046 for Cohort B versus Cohort A and 0.907 for Cohort C versus Cohort A, with 90% confidence intervals and p-values of 0.502 and 0.144, respectively.

The recalculated comparisons use the posted arm means and standard errors to produce unadjusted mean differences on the original scale. They are 1.41 for Cohort B versus Cohort A and -2.86 for Cohort C versus Cohort A, with p-values of 0.461 and 0.103. Because the sponsor's model and the recalculation use different scales and methods, their estimates, intervals, and p-values are not expected to be identical.

The registry record does not flag a discrepancy between the posted summary statistics and the sponsor analyses. The sponsor's own analysis is the principal reported analysis for interpreting the trial result; the simple recalculation is a supplementary view, not a substitute. Neither analysis alone establishes clinical importance, and the registry does not report a clinical threshold or non-inferiority margin for the endpoint.

14. Multiplicity, Missing Data and Analysis Populations

The registry record reports two pairwise primary-outcome comparisons: Cohort B with Cohort A and Cohort C with Cohort A. It does not report a direct comparison between Cohort B and Cohort C. The record does not state whether these comparisons were prespecified as a family, whether multiplicity was controlled, or whether a hierarchical testing procedure was used. When more than one hypothesis is tested, general statistical practice recognizes that the chance of at least one false-positive result can increase if no adjustment or testing strategy is used. The impact cannot be determined from the posted information.

The endpoint description says that subjects who received all 3 infusions were included. It does not report the numbers screened, randomized, treated, completing each infusion, or included in each analysis beyond the outcome group counts. The adverse-event risk-set counts are different from some of the endpoint group counts, but the reason for this is not reported in the registry record. The record also does not describe how missing endpoint values were handled, whether any values were imputed, or whether sensitivity analyses assessed missing-data assumptions.

The registry does not report an intention-to-treat analysis, a per-protocol population, or another analysis population definition beyond the stated inclusion of subjects who received all 3 infusions. In general, analyzing only participants who complete treatment can alter the balance created by randomization if inclusion is related to outcomes or treatment group. The available record does not provide enough participant-flow or missingness information to assess whether that concern affected these results.

The statistical plan, including the prespecified primary hypothesis, analysis threshold, handling of deviations, and planned sensitivity analyses, is not reported in the registry record. The non-inferiority or equivalence label does not supply those details. In particular, an equivalence or non-inferiority conclusion requires a margin specified in advance and an analysis interpreted against that margin. The margin is not reported in the registry record, so the reported ratios and intervals cannot by themselves establish equivalence or non-inferiority.

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 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: anaemia (7.5%), asthenia (7.5%), fatigue (5.0%), cerebrovascular accident (5.0%), hydronephrosis (5.0%).

The adverse-event summary lists serious adverse events for 10 of 40 participants at risk in Cohort A, 9 of 40 in Cohort B, and 11 of 40 in Cohort C. It also lists other adverse events for 38 of 40, 40 of 40, and 40 of 40, respectively. Death counts are listed as 29, 36, and 36 in the three cohorts. These are separate registry summaries; the record does not explain their relationship to the primary endpoint analysis counts or provide timing and attribution details.

The registry lists serious adverse-event terms with percentages, including ANAEMIA at 7.5%, ASTHENIA at 7.5%, FATIGUE at 5.0%, CEREBROVASCULAR ACCIDENT at 5.0%, HYDRONEPHROSIS at 5.0%, and DISSEMINATED INTRAVASCULAR COAGULATION at 2.5%. The record does not assign these listed percentages to individual cohorts in the summary. They should therefore be read as the posted serious-event term summary, not as cohort-specific rates.

These counts describe reported safety events and deaths in the registry record. They do not establish that an event was caused by treatment, and the record does not provide enough detail to compare event timing, severity beyond the serious-event classification, or exposure duration across cohorts. No medical conclusion should be drawn from these summaries alone.

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

The results describe a laboratory product measure in three randomized cohorts that differed in the PA2024 concentration used to manufacture sipuleucel-T. The sponsor's ANOVA results and the unadjusted recalculations both leave uncertainty around the cohort differences: the reported sponsor intervals include 1, and the recalculated intervals include zero. These statistics concern cumulative CD54 upregulation, not a direct measure of cancer control or patient experience.

The registry reports no direct comparison between Cohorts B and C. It also does not report clinical outcomes such as tumor response, survival, or symptoms. Therefore, the posted primary outcome cannot establish whether a particular concentration changes those outcomes. Nor can a result on this laboratory measure alone be taken as evidence that a treatment is clinically beneficial, ineffective, equivalent, or non-inferior.

Interpretation is constrained by the absence of a reported clinical threshold for an important change in CD54 upregulation, a non-inferiority or equivalence margin, and detailed missing-data and analysis-population information. The record states that the primary analysis included subjects who received all 3 infusions, but it does not fully describe participant flow. The listed outcome counts also differ from the adverse-event risk-set counts, and the record does not explain that relationship.

The appropriate statistical reading is consequently narrow: the posted estimates summarize differences in the stated laboratory endpoint, with considerable uncertainty about their size and direction. Their meaning for clinical decisions or patient outcomes is not reported in the registry record. A lack of strong statistical evidence for a difference is not the same as proof that the groups are identical, and an interval that includes the null value does not establish equivalence.

17. Why This Trial Matters Statistically

This study is statistically informative because it compares a continuous laboratory endpoint across randomized groups while varying a manufacturing concentration. It illustrates how the scale of analysis matters: the sponsor used a log-transformed ANOVA and reported ratios of geometric means, while a simple recalculation from the posted summaries describes arithmetic mean differences. These are related but not interchangeable ways of expressing group contrasts.

The case also shows why uncertainty is central to interpreting a point estimate. The sponsor's estimates are accompanied by 90% confidence intervals, and the recalculated estimates by intervals whose confidence level is not specified in the registry record. Each set of intervals includes its corresponding null value. Point estimates alone would conceal the range of effects compatible with the analyses.

It also highlights the distinction between a laboratory endpoint and a clinical outcome. A product measure can help characterize the manufacturing process, but the posted CD54 results do not directly answer questions about tumor response, survival, symptoms, or other patient outcomes. A statistically described difference is not automatically clinically meaningful; interpretation needs a relevant threshold or clinical linkage, neither of which is reported in the registry record.

Finally, the endpoint description's completion criterion, the different counts in the endpoint and safety summaries, and the absence of reported multiplicity and missing-data methods show why population definitions and analysis plans matter. The reported statistics can be read, but their broader interpretation depends on information that the registry record does not include.

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.

Several trial-specific limitations affect how fully the results can be interpreted. The posted primary endpoint is a cumulative laboratory measure, and the registry record does not report its relationship to clinical outcomes or a threshold for a clinically meaningful difference. Thus, the statistical comparisons do not establish the importance of the observed endpoint values for patients.

The sponsor reports two comparisons with Cohort A, but no direct Cohort B versus Cohort C comparison. The record does not describe whether multiplicity was addressed, and it does not report a non-inferiority margin despite labelling the analysis type as non-inferiority or equivalence. It is therefore not possible to assess a formal equivalence or non-inferiority claim from the posted information.

The primary endpoint description says subjects who received all 3 infusions were included. Full participant flow, the handling of missing measurements, and the use of other analysis populations are not reported in the registry record. The endpoint summaries list group counts of 40, 36, and 37, while adverse-event summaries list 40 at risk in each cohort; the record does not explain the difference in these reported counts.

Further statistical details are unavailable, including the original sample-size justification, the prespecified significance threshold, the randomization procedure, blinding details, model diagnostics, and sensitivity analyses. The recalculated comparisons are unadjusted and use a different scale from the sponsor's log-transformed ANOVA. They offer a supplementary description but do not resolve the missing design and analysis details. These limitations make the results best interpreted as evidence about the posted laboratory endpoint and its uncertainty, rather than broader clinical conclusions.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: To Evaluate Sipuleucel-T Manufactured With Different Concentrations of (PA2024) Antigen — NCT00715078.
• Registry record last updated 2017-05-23; status last verified 2017-04.
• Results first posted 2014-05-12.
• 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

NCT00715078 was a completed randomized, multicenter, single-blind, parallel-group Phase 2 study sponsored by Dendreon. The registry describes men with metastatic androgen independent prostate cancer and lists enrollment of 122. The study compared sipuleucel-T manufactured with PA2024 concentrations of 10 μg/mL, 5 μg/mL, and 2 μg/mL.

The primary endpoint was cumulative CD54 upregulation across the final products from 3 infusions. The sponsor reported log-transformed ANOVA comparisons of Cohort B and Cohort C with Cohort A. The reported geometric-mean ratios were 1.046 and 0.907, with 90% confidence intervals that included 1. The unadjusted recalculations also had intervals that included zero. Clinical outcome results, a meaningful endpoint threshold, and several design and analysis details are not reported in the registry record.