This page separates reported trial information from statistical interpretation. The ClinicalTrials.gov record reports that results are posted, but they do not provide formal statistical analysis estimates, confidence intervals, or p-values for the registered primary endpoint.
View the ClinicalTrials.gov record for NCT02952534.
1. Trial at a Glance
TRITON2 was a completed phase 2, single-group study evaluating rucaparib in patients with metastatic castration-resistant prostate cancer and homologous recombination gene deficiency. The study enrolled 277 patients and used a binary primary efficacy endpoint based on confirmed objective response rate by gene in patients with measurable disease at baseline, assessed by central independent radiology review.
| Feature | TRITON2 |
|---|---|
| Phase | Phase 2 |
| Condition | Metastatic Castration Resistant Prostate Cancer |
| Population | Patients with metastatic castration-resistant prostate cancer and homologous recombination gene deficiency |
| Intervention | Rucaparib |
| Design | Single-group |
| Allocation | NA |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 277 |
| Primary endpoint type | Binary |
| Results posted | Yes |
| Outcome measures posted | 13 |
| Serious adverse events | 96/277, all patients |
| Lead sponsor | pharmaand GmbH |
| Sponsor type | Industry |
| ClinicalTrials.gov | NCT02952534 |
2. Clinical Question
The central statistical question in TRITON2 concerns the observed frequency of confirmed radiographic tumor response to rucaparib in the specified population. The registered primary endpoint is not a comparison between two randomized treatment groups; instead, it evaluates confirmed objective response rate by gene in patients with measurable disease at baseline using central independent radiology review.
Population
Patients with metastatic castration-resistant prostate cancer and homologous recombination gene deficiency.
Intervention
Rucaparib.
Comparator
The study is registered as a single-group design, so no concurrent randomized comparator arm is specified in the ClinicalTrials.gov record.
Primary question
What proportion of patients with measurable disease at baseline achieved a confirmed complete or partial radiographic response by gene under central independent radiology review?
3. Trial Design
Study timeline
Study start
TRITON2 began on February 15, 2017.
Primary completion
The registered primary completion date was July 18, 2021.
Completed
the ClinicalTrials.gov record lists the study status as completed.
4. Endpoints
| Endpoint | Registry definition | Endpoint type |
|---|---|---|
| Confirmed Objective Response Rate (ORR) by Gene in Patients With Measurable Disease at Baseline Per Central Independent Radiology Review (IRR) | The primary efficacy endpoint is confirmed radiographic ORR by central IRR. ORR is defined as the percentage of patients with a confirmed CR (complete response) or PR (partial response) by mRECIST (modified Response Evaluation Criteria in Solid Tumors) v1.1/PCWG3 (Prostate Cancer Working Group 3) criteria. The confirmed response is defined as a CR or PR on subsequent tumor assessment at least 28 days after first response documentation in the absence of confirmed progression in bone. CR is disappearance of all target and non-target lesions; any pathological lymph nodes (whether target or non-target) must have reduction in short axis to \<10 mm. PR is at least a 30% decrease in the sum of the diameters of target lesions, taking as reference the baseline sum diameters. | Binary |
What the primary endpoint actually measures
ORR is a binary-response endpoint summarized as a proportion. Each evaluable patient is classified according to whether the patient achieved a confirmed complete response or partial response under the specified response criteria. The resulting proportion answers a fundamentally different question from a time-to-event endpoint such as overall survival or progression-free survival.
The registry-reported TRITON2 data do not provide the numerator, denominator for the measurable-disease response population, gene-specific response counts, or a formal statistical estimate. They therefore do not support calculating an observed ORR.
5. Statistical Methodology
The ClinicalTrials.gov record establishes a binary primary endpoint, a single-group design, and central independent radiology review. They do not provide a statistical-analysis object containing an estimated effect, confidence interval, p-value, hypothesis-testing framework, or analysis population for the posted result.
Binary endpoint framework
For a single-group confirmed ORR endpoint, the fundamental statistical quantity is a proportion. Let Y denote the number of patients classified as confirmed responders and N the number of patients in the prespecified response-analysis population. The observed response rate is Y/N.
Here, p represents the underlying response probability in the population represented by the analysis. The binomial formulation is the natural starting point for inference about a single response proportion, although the registry-reported TRITON2 data do not specify which exact inferential procedure was used.
Confidence interval for a response proportion
A confidence interval would quantify statistical uncertainty around the observed response proportion. For a single-group binary endpoint, exact binomial intervals or other binomial proportion intervals can be used depending on the prespecified analysis plan.
The choice of interval method matters particularly when the number of responders is small or the response proportion is close to 0 or 1. The ClinicalTrials.gov record does not identify the confidence-interval procedure used for TRITON2, so no particular method is attributed to the registry.
Central independent radiology review
The registered primary endpoint specifies central independent radiology review. This is important statistically because the response classification is not simply an investigator-reported clinical impression. The endpoint is anchored to an independent radiologic assessment framework using mRECIST v1.1/PCWG3 criteria as specified in the registry definition.
Gene-specific analysis
The primary endpoint is explicitly defined as confirmed ORR by gene. That wording means that the response endpoint is structured to describe response in relation to the relevant homologous recombination gene category rather than treating all patients as a single undifferentiated response group.
The ClinicalTrials.gov record does not provide the individual gene categories, gene-specific sample sizes, gene-specific responder counts, or gene-specific confidence intervals. Those quantities therefore cannot be reconstructed here.
What the registry data do not establish
the ClinicalTrials.gov record does not provide a formal statistical comparison between rucaparib and a control treatment. It also does not provide a non-inferiority margin, crossover specification, factorial structure, multiplicity procedure, interim-analysis rule, missing-data/imputation method, stratified analysis, or Bayesian model.
6. Statistical Methods Explained
Why is ORR a binary endpoint?
For the primary endpoint, each patient ultimately contributes a response classification based on whether the patient achieved a confirmed CR or PR under the registered criteria. The analysis then summarizes those binary classifications as a proportion. This is different from measuring a continuous tumor-size change or a time-to-progression variable.
Why does the study design matter for interpreting ORR?
TRITON2 is registered as a single-group study. Therefore, an observed response proportion describes response within the treated study population. Without a concurrent randomized comparator in the registry-reported design, the response rate itself does not isolate a randomized between-treatment effect.
What does "confirmed" add to ORR?
Why use central independent radiology review?
Radiographic response can involve judgment about lesion measurements and response categories. A central independent review creates a prespecified review framework separate from local investigator assessment. For TRITON2, the registered primary endpoint explicitly assigns the efficacy assessment to central independent radiology review.
Why does the confidence interval matter?
An observed response proportion is calculated from a finite sample and therefore has sampling uncertainty. A confidence interval communicates that uncertainty around the estimated response proportion. It does not describe the range of responses an individual patient might experience, and it is not a measure of clinical importance.
Why can't the enrollment of 277 be used as the ORR denominator?
The registered primary endpoint specifically concerns patients with measurable disease at baseline. The overall enrollment is 277, but the ClinicalTrials.gov record does not state how many of those patients belonged to the measurable-disease response population. Therefore, 277 cannot automatically be substituted for the response-analysis denominator.
7. Planned Analysis
Primary endpoint: confirmed ORR by gene
Confirmed CR or PR by mRECIST v1.1/PCWG3 criteria, assessed by central independent radiology review.
Results are posted for the trial, but the registry-reported TRITON2 data contain no formal statistical analyses. Consequently, there is no reported numerical primary endpoint estimate in the registry record that can be presented as the trial's observed ORR.
How this endpoint would normally be analyzed
- Define the prespecified response-analysis population, consistent with the requirement for measurable disease at baseline.
- Classify each patient according to the central independent radiology review as having a confirmed CR, confirmed PR, or not meeting the confirmed-response definition.
- Calculate the observed ORR as the number of confirmed CR or PR patients divided by the prespecified response-analysis population.
- Calculate a confidence interval for the single response proportion using the method specified in the statistical analysis plan.
- Present response results by the relevant homologous recombination gene categories because the registered endpoint is explicitly defined "by Gene."
- Keep the response-analysis population distinct from the overall enrolled population unless the registry or statistical analysis plan explicitly defines them as identical.
The key statistical feature of TRITON2 is that its primary efficacy question is proportion-based rather than comparative. An ORR estimate would describe the fraction of the relevant measurable-disease population achieving a confirmed CR or PR. It would not, by itself, establish that rucaparib produced more responses than another treatment because the registry-reported design is single-group.
A confidence interval would show the precision of the estimated response proportion. A narrower interval would indicate greater statistical precision, while a wider interval would indicate more uncertainty. A p-value, if one were prespecified for a comparison against a reference response rate, would address compatibility with that specific null hypothesis; it would not measure the magnitude or clinical importance of the observed ORR.
8. Secondary Endpoints and Posted Outcomes
the ClinicalTrials.gov record states that 13 outcome measures are posted and that results are posted for the study. However, the ClinicalTrials.gov record does not enumerate the 13 outcome measures or provide their numerical results.
| Registry information | What can be established from the ClinicalTrials.gov record |
|---|---|
| Outcome measures posted | 13 |
| Results posted | Yes |
| Formal statistical analyses reported | None |
| Primary endpoint results reported | No numerical estimate, confidence interval, or p-value |
| Secondary endpoint results reported | No secondary endpoint definitions or numerical results are included in the ClinicalTrials.gov record |
Because the ClinicalTrials.gov record does not enumerate the secondary outcomes, this page does not reconstruct or attribute secondary efficacy results from the linked publications. Doing so would violate the numerical-source rule for this analysis.
9. Safety
the ClinicalTrials.gov record provides one safety statistic: 96 of 277 all patients were affected by serious adverse events.
| Safety measure | Reported population | Affected / at risk |
|---|---|---|
| Serious adverse events | All Patients | 96/277 |
Serious adverse events
All patients in the registry-reported trial-profile safety summary.
Because TRITON2 has a single study arm in the registry-reported design, there is no randomized arm-to-arm serious-adverse-event comparison to report. The 96/277 figure is a descriptive safety count, not a treatment-versus-control effect estimate.
The value 96/277 identifies the number of patients affected by serious adverse events relative to the stated at-risk population. It does not establish whether those events were caused by rucaparib, and it does not quantify the severity, duration, timing, or individual clinical consequences of those events.
10. Statistical Interpretation of a Single-Group Response Rate
A single-group ORR is a descriptive measure of response frequency within the study population. It should not be relabeled as a relative risk reduction, hazard ratio, odds ratio, or randomized treatment effect.
If a confidence interval were reported around the ORR, it would quantify sampling uncertainty around the estimated population response probability under the chosen statistical framework. It would not indicate that individual patient responses are distributed uniformly across that interval.
A p-value only has meaning relative to a specified null hypothesis. For a single-arm ORR study, a formal test could compare the response probability with a prespecified reference value, but the registry-reported TRITON2 data do not state such a reference value or provide a p-value.
The registered endpoint is specifically limited to patients with measurable disease at baseline. That restriction means the overall enrollment of 277 should not automatically be interpreted as the denominator for the primary ORR calculation.
11. Central Review and Response Assessment
The registered primary efficacy endpoint specifies central independent radiology review (IRR). The endpoint definition further specifies response according to mRECIST v1.1/PCWG3 criteria.
Central review
The primary efficacy assessment is based on central independent radiology review rather than an unspecified local assessment.
Response criteria
The registry specifies mRECIST v1.1/PCWG3 criteria for classification of CR and PR.
Confirmation
The registry-reported definition requires a subsequent tumor assessment at least 28 days after first response documentation in the absence of confirmed progression in bone. CR is disappearance of all target and non-target lesions; any pathological lymph nodes (whether target or non-target) must have reduction in short axis to \<10 mm. PR is at least a 30% decrease in the sum of the diameters of target lesions, taking as reference the baseline sum diameters. after the qualifying response.
Assessment schedule
Why confirmation changes the statistical endpoint
A confirmed-response endpoint is not simply the percentage of patients who ever have a scan interpreted as partial or complete response. The confirmation requirement introduces a longitudinal component into what is ultimately recorded as a binary endpoint: the response classification depends on subsequent assessment.
This distinction matters for statistical analysis because patients need sufficient follow-up and assessment opportunities to be classified under the confirmation rule. A patient's lack of a confirmed response is therefore not necessarily equivalent to an immediate biological absence of tumor shrinkage.
12. Analysis Population and Denominator Issues
The most important denominator issue visible in the ClinicalTrials.gov record is the distinction between overall enrollment and the population specified by the primary endpoint.
| Population concept | What the ClinicalTrials.gov record establishes |
|---|---|
| Overall enrollment | 277 patients |
| Primary response population | Patients with measurable disease at baseline |
| Gene-specific population | The endpoint is defined by gene, but the ClinicalTrials.gov record does not give gene-specific sample sizes |
| Safety population | The serious-AE summary uses all patients, 96/277 |
This distinction prevents a common statistical mistake: treating the trial's total enrollment as if it were automatically the denominator for every outcome. Different endpoints can legitimately use different analysis populations when the endpoint definition requires it.
13. What Is Not Supported by the ClinicalTrials.gov record
the ClinicalTrials.gov record is sufficient to characterize the broad study design and primary endpoint, but several commonly reported statistical features are absent from the ClinicalTrials.gov record.
| Topic | TRITON2 information reported |
|---|---|
| Non-inferiority margin | Not reported |
| Crossover | Not reported |
| Factorial design | Not reported; the registered design model is single-group |
| Multiplicity adjustment | Not reported |
| Interim analysis | Not reported |
| Missing-data / imputation method | Not reported |
| Bayesian methods | Not reported |
| Stratified statistical analysis | Not reported |
| ORR estimate | Not reported in the ClinicalTrials.gov record |
| ORR confidence interval | Not reported in the ClinicalTrials.gov record |
| ORR p-value | Not reported in the ClinicalTrials.gov record |
These omissions are analytically important. A high-quality statistical interpretation should distinguish information that is genuinely absent from information that could be guessed from the type of endpoint.
14. Limitations
- Single-group design: the registry-reported design does not include a concurrent randomized comparator, so the primary response proportion is not itself a randomized comparative treatment effect.
- Response-population restriction: the primary endpoint concerns patients with measurable disease at baseline, while the overall enrollment is 277. The ClinicalTrials.gov record does not give the size of the measurable-disease population.
- Gene-specific interpretation: the endpoint is defined by gene, but the ClinicalTrials.gov record does not provide gene-specific response counts or denominators.
- No formal statistical-analysis records reported: although results are listed as posted, the registry-reported statistical-analysis array is empty. Numerical efficacy estimates, confidence intervals, and p-values therefore cannot be independently reproduced from in the registry record.
- Central review does not eliminate all uncertainty: independent radiologic review standardizes assessment but does not eliminate measurement variability or the statistical uncertainty associated with a finite response population.
- Confirmation requirement: confirmed response depends on subsequent assessment, so response classification incorporates follow-up beyond the initial qualifying scan.
- Safety interpretation: the registry-reported serious-adverse-event count is descriptive and does not provide a comparator or attribution analysis.
15. Why This Trial Matters Statistically
TRITON2 is a useful teaching case because it illustrates a different statistical structure from a conventional randomized phase 3 survival trial. Its central efficacy question is a single-group binary response problem, with response adjudicated by central independent radiology review and defined using specific radiologic criteria.
| Concept | How it appears in TRITON2 |
|---|---|
| Single-group design | One registered study arm and no concurrent randomized comparator in the ClinicalTrials.gov record. |
| Binary endpoint | The primary endpoint is confirmed objective response rate. |
| Response proportion | ORR is defined as the percentage of patients with confirmed CR or PR. |
| Measurable-disease population | The primary endpoint applies to patients with measurable disease at baseline. |
| Central review | Primary efficacy assessment is by central independent radiology review. |
| Response criteria | mRECIST v1.1/PCWG3 criteria are specified. |
| Confirmation | A subsequent tumor assessment at least 28 days after first response documentation in the absence of confirmed progression in bone. CR is disappearance of all target and non-target lesions; any pathological lymph nodes (whether target or non-target) must have reduction in short axis to \<10 mm. PR is at least a 30% decrease in the sum of the diameters of target lesions, taking as reference the baseline sum diameters. after the qualifying response is part of the registry-reported definition. |
| Gene-specific analysis | The registered endpoint is confirmed ORR by gene. |
| Statistical uncertainty | A response proportion would ordinarily be accompanied by an appropriate confidence interval. |
| Denominator discipline | The overall enrollment of 277 is not automatically the primary-response denominator. |
| Safety analysis | Serious adverse events are reported as 96/277 among all patients. |
16. Related Statistical Tutorials
Learn more about the methods used in this trial:
17. Related Statistical Calculators
18. Sources
- ClinicalTrials.gov: TRITON2 (NCT02952534).
- PubMed: Publication associated with TRITON2 (PMID 37277275).
- PubMed: Publication associated with TRITON2 (PMID 36898948).
- PubMed: Publication associated with TRITON2 (PMID 35397664).
- PubMed: Publication associated with TRITON2 (PMID 32203306).
Continue through the Clinical Biostats knowledge graph
Explore the statistical methods behind binary clinical-trial endpoints, response rates, confidence intervals, and single-arm study designs.
19. Record Summary
TRITON2 is a completed phase 2 single-group study of rucaparib in patients with metastatic castration-resistant prostate cancer and homologous recombination gene deficiency. The trial enrolled 277 patients, and its registered primary endpoint is confirmed objective response rate by gene in patients with measurable disease at baseline, assessed by central independent radiology review using mRECIST v1.1/PCWG3 criteria. The endpoint is binary and requires confirmation through a subsequent tumor assessment under the registry definition.
The statistical structure is therefore fundamentally different from a randomized time-to-event trial. The primary quantity is a response proportion, and appropriate inference would ordinarily involve a confidence interval and, if specified by the analysis plan, a hypothesis test against a prespecified reference value. The ClinicalTrials.gov record does not identify those inferential procedures or provide a formal statistical-analysis record.
the ClinicalTrials.gov record does provide a safety count of 96 serious-adverse-event cases among 277 all patients. That statistic is descriptive and cannot be converted into a randomized treatment effect because the registry-reported design contains a single study arm.