RE-sensitizing With Supraphysiologic Testosterone to Overcome REsistance (The RESTORE Study): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for RE-sensitizing With Supraphysiologic Testosterone to Overcome REsistance (The RESTORE Study) (NCT02090114), a non-randomized phase 2 study. The primary outcome is Prostate Specific Antigen (PSA) Response to Bipolar Androgen Therapy (BAT).

2. Clinical Question

The central question is descriptive: whether participants met the registry’s PSA response definition after BAT. The primary outcome is a count of participants, rather than a direct estimate of a treatment contrast against a concurrent control group. The registry record identifies four cohorts based on prior therapy or mutation status. These cohort distinctions matter because the participants entered with different treatment histories or molecular eligibility criteria.

The brief summary describes planned groups of men whose metastatic castration-resistant prostate cancer had progressed on enzalutamide, abiraterone acetate, or first-line castration-only therapy, plus a mutation-defined group. The study also listed secondary outcomes covering PSA progression, disease response, initiation of docetaxel chemotherapy, quality of life, and adverse events. The primary endpoint result itself is not reported in the registry record, so the clinical question cannot be answered from the posted outcome information alone.

3. Trial Design

RESTORE was a phase II, single-site, open-label study with a non-randomized, single-group design. The registry record lists the study as completed, with enrollment of 112. The recorded study start was 2014-08-25, and primary completion was 2021-11-11.

The study description identifies four cohorts. Cohort A included men with metastatic castration-resistant prostate cancer that had progressed on enzalutamide; Cohort B included men whose disease had progressed on abiraterone acetate; and Cohort C included men who had progressed on first-line castration-only therapy. Cohort D was defined by inactivating somatic or germline mutations in at least two of TP53, PTEN, or RB1. The brief summary lists planned cohort sizes of 30, 30, 30, and 20, respectively. The registry enrollment is an overall study figure and should not be mistaken for the number analyzed for every outcome.

The intervention list includes testosterone cypionate, testosterone enanthate, abiraterone acetate, and enzalutamide. The study summary describes parenteral testosterone followed by enzalutamide, abiraterone, or castration-only therapy. The available cohort descriptions are truncated, so the complete treatment sequence and rules for moving between therapies are not reported in the registry record.

Because allocation was non-randomized and there was no masking, the study does not provide a randomized, blinded comparison. Cohort outcomes may describe what occurred among participants with those entry characteristics, but differences between cohorts cannot be attributed to a treatment contrast without additional design and analysis information. The protocol’s detailed statistical plan and sample-size justification 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.

Before assignment: 16 patients failed screen.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Only 7/8 patients were evaluable from Cohort D.

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 Prostate Specific Antigen (PSA) Response to Bipolar Androgen Therapy (BAT), measured over up to 18 months.

The primary endpoint is the number of participants with at least a 50% reduction in PSA from the pre-BAT baseline, assessed over a timeframe of up to 18 months. It is a binary response measure at the participant level: each person either meets the stated reduction threshold or does not. The registry identifies the unit as participants and the outcome parameter as a count.

PSA is a biomarker used in prostate cancer research to track changes in disease-related activity. A threshold-based response definition turns a continuous measurement into a clinically interpretable category for summarizing how many participants reached a specified decrease. General statistical point: a binary endpoint is straightforward to count, but it does not show the size of each participant’s PSA change beyond whether the threshold was reached. It also does not, by itself, describe other aspects of disease status, quality of life, or safety.

The registry record does not report the primary response count or an estimate of its uncertainty. It notes that only 7/8 patients were evaluable from Cohort D. That qualification is important when interpreting the endpoint: the evaluable group for this cohort was not the full cohort count at risk reported in the safety information. The rules used to determine evaluability, and how unevaluable participants were handled in the primary analysis, are not reported in the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The available registry information supports a discussion of the endpoint’s form and the study’s design, but it does not document the full statistical analysis plan. The primary outcome is specified as a count of participants who achieved a PSA reduction threshold. No primary response count, inferential analysis, or uncertainty interval is reported in the registry record.

What does a binary endpoint represent?

A binary endpoint classifies each participant into one of two outcome categories. For this study’s primary endpoint, those categories are meeting the at-least-50% PSA reduction criterion or not meeting it. Counting participants in the response category gives a simple summary of how many met the endpoint definition. A participant-level classification does not preserve all information in the underlying PSA measurements: changes smaller or larger than the threshold are grouped into the same non-response or response category, respectively.

General statistical explanation: a response proportion is often calculated by dividing the number meeting the criterion by the number included in the analysis. That proportion is interpretable only when the analysis denominator and the rules for missing or unevaluable outcomes are clear. The primary endpoint result and analysis denominator are not reported in the registry record. The note that 7/8 patients were evaluable from Cohort D signals that evaluability was relevant, but the record does not describe the criteria or how non-evaluable cases were treated.

What does non-randomized allocation imply?

In a randomized comparison, chance assignment is used to make groups comparable on average. This study is listed as non-randomized and single-group, so it does not create a randomized control arm. The four cohorts were defined by prior treatment experience or mutation status, not by random assignment to comparable treatment alternatives. Consequently, differences in outcomes across these groups would be descriptive, and interpretation would need to account for the reasons participants entered each cohort.

The study was also open label, with no masking. This means masking is not part of the recorded design. The registry does not report whether endpoint assessment involved additional safeguards or independent review. Without details on assessment procedures, the potential influence of knowledge of treatment on outcome assessment cannot be evaluated from the record.

What can a confidence interval add?

General statistical explanation: a confidence interval expresses the uncertainty around an estimated quantity under a specified statistical method. For a binary response estimate, an interval can help show whether the available data support a relatively precise or imprecise estimate. It does not correct for selection bias, missing outcome information, or the absence of a suitable comparator. No confidence interval for the primary endpoint is reported in the registry record.

A p-value, when used, summarizes how compatible data are with a specified statistical null model. It is not a measure of the size or clinical importance of an effect, and it does not establish causation. The registry record lists 0 analyses, and no p-value or inferential analysis for the primary endpoint is reported. A statistical testing plan, including any prespecified null hypothesis, is not reported in the registry record.

How should planning calculations be read?

General statistical explanation: sample-size calculations are planning exercises based on assumptions about the effect to detect, variability, allocation, and error rates. They are not observed results and do not tell readers whether the study demonstrated an effect. The stated planning illustration uses a two-sided significance level of 0.05, equal allocation between two arms, a normal approximation, and an effect defined as a stated fraction of the observed effect. Those assumptions describe an illustration, not the trial’s own design. The record does not report the trial’s sample-size justification.

When a planning illustration uses the observed effect as its starting point, the result depends on that estimate and can be unstable, especially when the estimate is imprecise. General statistical explanation: powering a future study to detect a fraction of an observed estimate is a planning choice, not evidence that the original study had that power or that the estimate is reliable. No claim of observed power follows from a planning calculation.

What does over-dispersion mean in general?

General statistical explanation: over-dispersion occurs when observed variability is greater than a statistical model assumes. It can make standard errors too small if ignored, which can make confidence intervals appear narrower and tests more decisive than warranted. The primary outcome in this study is a participant count, but the registry record does not report a model, a dispersion estimate, or an analysis that would allow over-dispersion to be assessed. It should not be assumed that a particular count model or correction was used.

The registry lists 12 outcome measures, including PSA progression, disease response, chemotherapy initiation, quality-of-life measures, and adverse events. The record does not report how those outcomes were analyzed, whether continuous measures were summarized by means or other statistics, or how repeated observations over time were handled. Those choices can affect both the estimate and its uncertainty, so the outcome definitions alone are not enough to reconstruct the analysis.

10. Exact Estimates for Each Arm

No two-arm comparison is shown for this trial, but each arm's result can still be summarized with exact interval estimates computed from the posted counts. The Clopper-Pearson interval is exact and conservative; the Wilson interval is a common closer-to-nominal alternative; the Bayesian credible interval uses a uniform Beta(1, 1) prior. For small groups the three can differ noticeably.

R code and output

arm    <- c("Cohort A:Post-enzalutamide", "Cohort B: Post-abiraterone", "Cohort C: Castration Only", "Cohort D: Mutation")
n      <- c(30, 29, 29, 7)
events <- c(9, 5, 4, 2)
for (i in seq_along(n)) {
  ci <- binom.test(events[i], n[i])$conf.int          # Clopper-Pearson exact interval
  cr <- qbeta(c(0.025, 0.975), 1 + events[i], 1 + n[i] - events[i])   # Bayesian, uniform Beta(1,1) prior
  wi <- (function(x, n) { p <- x / n; z <- qnorm(0.975); d <- 1 + z^2 / n
    c((p + z^2 / (2 * n) - z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d, (p + z^2 / (2 * n) + z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d) })(events[i], n[i])
  cat(sprintf("ARM|%s|%d|%d|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f\n", arm[i], n[i], events[i], 100 * events[i] / n[i],
              100 * ci[1], 100 * ci[2], 100 * wi[1], 100 * wi[2], 100 * cr[1], 100 * cr[2]))
}

ARM  Cohort A:Post-enzalutamide  30  9  30.0  14.7  49.4  16.7  47.9  16.7  48.0
ARM  Cohort B: Post-abiraterone  29  5  17.2  5.8  35.8  7.6  34.5  7.7  34.7
ARM  Cohort C: Castration Only  29  4  13.8  3.9  31.7  5.5  30.6  5.6  30.7
ARM  Cohort D: Mutation  7  2  28.6  3.7  71.0  8.2  64.1  8.5  65.1

11. Multiplicity, Missing Data and Analysis Populations

The registry lists 12 outcome measures, including the primary PSA response endpoint and secondary measures of PSA progression, disease response, chemotherapy initiation, quality of life, and safety. When a study assesses multiple outcomes, the chance of at least one apparently notable result can increase if each outcome is tested separately without a plan for multiplicity. General statistical explanation: multiplicity procedures can control this risk, while a prespecified hierarchy can define which findings are primary and which are exploratory. The registry record does not report a multiplicity strategy, a hierarchy among outcomes, or a statistical testing plan.

The registry lists 0 analyses. It therefore does not provide analysis results or explain which participants were included in each outcome analysis. The primary endpoint description notes that only 7/8 patients were evaluable from Cohort D, but the criteria for evaluability and the handling of participants without an evaluable result are not reported in the registry record.

Missing-data methods are also not reported in the registry record. This matters because follow-up extends up to 18 months, and the listed secondary outcomes include measures that may be collected over time. Without information about how incomplete follow-up or unavailable measurements were handled, the potential effect of missing observations on estimates cannot be assessed.

The analysis populations are not reported in the registry record. In particular, the record does not specify whether the primary endpoint used all enrolled participants, an evaluable group, or another defined population. The listed overall enrollment and the cohort-specific safety counts should not be treated as a substitute for an endpoint-specific analysis denominator.

12. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: urinary tract infection (12.5%), urine retention (3.4%), disseminated intravascular coagulation (3.4%), neurological symptoms (3.4%), pneumonia (3.4%).

The registry’s serious adverse event information is presented as counts of participants at risk and participants affected within each cohort. Cohort A lists 30 participants at risk and 3 affected; Cohort B lists 29 at risk and 4 affected; Cohort C lists 29 at risk and 5 affected; and Cohort D lists 8 at risk and 1 affected. Death counts are listed as 1, 0, 1, and 0, respectively. These are descriptive counts, not a comparison of event rates from randomized groups.

The registry also lists participants affected by other adverse events: 30 of 30 in Cohort A, 29 of 29 in Cohort B, 29 of 29 in Cohort C, and 8 of 8 in Cohort D. These participant counts do not indicate how many separate events occurred, how severe each event was, or whether an event was considered related to treatment.

Among the listed serious-event terms, urinary tract infection is reported as 12.5%; urine retention, disseminated intravascular coagulation, neurological symptoms, pneumonia, and myocardial infarction are each listed as 3.4%. The record does not report the denominator, event timing, exposure duration, or whether these terms refer to participants or events for those percentages. Serious adverse events and deaths should therefore be read as safety descriptions in this study, not as evidence of treatment causation or a comparison against another group.

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

The registry record defines a focused question: how many participants had at least a 50% reduction in PSA from the pre-BAT baseline during follow-up of up to 18 months. It also describes a phase II, open-label, non-randomized, single-group study in men with metastatic castration-resistant prostate cancer, organized into cohorts by prior treatment history or mutation status.

What the posted information does not establish is the number of participants who achieved the primary endpoint. The response count and its analysis denominator are not reported in the registry record. Without them, neither the frequency of meeting the PSA criterion nor the precision of that estimate can be assessed. The registry record also does not provide numerical results for the listed secondary outcomes.

The design does not include a randomized control group. Even if cohort-specific outcome estimates were available, differences between cohorts would not, on their own, show that one treatment sequence caused a better or worse result. Cohorts were selected according to prior progression or mutation status, and those characteristics could be associated with different outcomes. A causal comparison would require a design and analysis capable of addressing those differences.

The available safety information gives counts of participants affected by serious and other adverse events and counts of deaths. These figures can describe events recorded in the study, but they do not establish that the intervention caused them. The registry record does not report event attribution, timing, treatment exposure duration, or a concurrent control group for comparison.

General statistical explanation: a biomarker response threshold captures one defined aspect of outcome. It does not, by itself, establish an effect on every other measure of disease or quality of life. The study listed outcomes beyond PSA response, but the numerical findings for those outcomes are not reported in the registry record. The appropriate reading is therefore limited to the design and the specific descriptive information available, without inferring efficacy or treatment causation.

14. Why This Trial Matters Statistically

Statistically, RESTORE illustrates how a clearly defined endpoint can still be difficult to interpret when the numerical result and analysis details are unavailable. The primary outcome has a specific threshold and timeframe: at least a 50% PSA reduction from pre-BAT baseline, with follow-up up to 18 months. Yet a threshold and timeframe do not supply the response count, denominator, or uncertainty around the estimate. Those are necessary to understand the observed frequency and its precision.

The cohort structure also highlights the importance of distinguishing descriptive group summaries from comparative evidence. Participants were grouped by prior treatment progression or mutation status, rather than randomized to comparable alternatives. The design can describe outcomes within these groups, but it does not by itself separate treatment effects from differences in who entered each cohort.

The registry’s safety counts provide another example of why denominators and context matter. Counts of affected participants can help describe the reported safety experience, while event attribution, timing, exposure, and comparison information are also needed to interpret those observations. The listed adverse-event counts do not answer whether events were caused by a study intervention.

Finally, the record lists a broad set of outcome measures, including PSA progression, disease response, chemotherapy initiation, quality of life, and safety. Multiple outcomes raise questions about prespecification, missing data, and multiplicity. These methodological details are not reported in the registry record. Recognizing what is and is not documented is part of sound statistical interpretation: it prevents study design, endpoint definitions, and safety counts from being mistaken for a complete analysis of treatment effects.

15. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

Several limitations specific to this study follow directly from its design and the available registry information. The study was non-randomized, open label, and single group. It therefore lacks a randomized concurrent control for estimating a treatment contrast, and cohort differences cannot be interpreted as effects of treatment sequence alone.

The primary PSA response count, endpoint-specific denominator, and uncertainty estimate are not reported in the registry record. The note that 7/8 patients were evaluable from Cohort D identifies an evaluability issue but does not define the criteria or explain how participants without an evaluable result were handled. The analysis populations and missing-data methods are not reported in the registry record.

The brief summary describes four cohorts with planned sizes, while the overall registry enrollment is 112. The record does not explain cohort-level enrollment and disposition across all outcomes. The available cohort descriptions are incomplete, and the full protocol details, statistical analysis plan, and sample-size justification are not reported in the registry record.

Although the record lists 12 outcome measures, it does not provide numerical results for the primary or secondary endpoints, or an analysis plan addressing multiple outcomes. Safety information is limited to participant counts, selected serious-event terms and percentages, and deaths. Event attribution, timing, duration of exposure, and the denominators used for the listed percentages are not reported in the registry record. These gaps limit evaluation of outcome precision, robustness, and safety context.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: RE-sensitizing With Supraphysiologic Testosterone to Overcome REsistance (The RESTORE Study) — NCT02090114.
• Registry record last updated 2022-06-27; status last verified 2022-06.
• Results first posted 2022-06-27.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

19. Record Summary

The RESTORE Study (NCT02090114) is listed as a completed phase II study sponsored by Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins. It enrolled 112 participants and was described as a single-site, open-label, non-randomized, single-group study in men with prostate cancer.

The study focused on bipolar androgen therapy followed by other listed therapies in cohorts defined by prior treatment progression or mutation status. Its primary endpoint was the count of participants with at least a 50% PSA reduction from the pre-BAT baseline over up to 18 months. The registry lists 12 outcome measures and 0 analyses.

The primary endpoint result, statistical analysis details, missing-data methods, and multiplicity strategy are not reported in the registry record. Available safety information includes cohort-level counts of participants affected by serious and other adverse events and counts of deaths.