This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record.
1. Trial at a Glance
ARAMIS was a randomized, parallel, quadruple-masked phase 3 trial evaluating darolutamide against placebo in men with high-risk non-metastatic castration-resistant prostate cancer. The registered primary endpoint was metastasis-free survival, a time-to-event endpoint analyzed using a log-rank framework and a stratified Cox regression model for the hazard ratio.
| Feature | ARAMIS |
|---|---|
| Trial name | ARAMIS |
| NCT identifier | NCT02200614 |
| Phase | Phase 3 |
| Condition | Prostate Cancer Non-Metastatic; Castration-Resistant |
| Population | Men with high-risk non-metastatic castration-resistant prostate cancer |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Quadruple |
| Primary purpose | Treatment |
| Enrollment | 1509 |
| Interventions | Darolutamide (Nubeqa, BAY1841788) versus placebo |
| Primary endpoint | Metastasis-Free Survival |
| Primary endpoint type | Time-to-event |
| Primary statistical method | Log-rank test with stratified Cox regression for the hazard ratio |
| Hypothesis type | Superiority |
| Lead sponsor | Bayer |
| Sponsor type | Industry |
2. Clinical Question
The primary statistical question was whether darolutamide, compared with placebo, was associated with a different time to metastasis or death from any cause in men with high-risk non-metastatic castration-resistant prostate cancer. The registered hypothesis type was superiority.
Population
Men with high-risk non-metastatic castration-resistant prostate cancer.
Intervention
Darolutamide (Nubeqa, BAY1841788).
Comparator
Placebo.
Primary question
Does darolutamide improve metastasis-free survival relative to placebo?
3. Trial Design
ARAMIS used a randomized, parallel-group design with quadruple masking. The registry reports an enrollment of 1509 and two intervention arms. The primary purpose was treatment.
Darolutamide
- Darolutamide (Nubeqa, BAY1841788)
- Randomized comparison against placebo
- Primary efficacy comparison based on metastasis-free survival
Placebo
- Placebo
- Randomized comparator arm
- Used as the reference group for the reported hazard ratios
4. Trial Timeline and Registry Status
Trial start
The registry lists 12 September 2014 as the study start date.
Primary completion
The registry lists 3 September 2018 as the primary completion date.
Registry status
the ClinicalTrials.gov record identifies ARAMIS as COMPLETED.
5. Endpoints
Primary Endpoint: Metastasis-Free Survival
| Endpoint | Registered definition and time frame | Analysis |
|---|---|---|
| Metastasis-Free Survival | Time from randomisation to evidence of metastasis or death from any cause, whichever occurs first. Cut-off date 15 Nov 2019. | Log-rank test; hazard ratio based on stratified Cox regression |
The registered primary time frame was from randomization to the time approximately 385 MFS events were observed (approximately 48 months). The endpoint therefore combines two clinically important event types: evidence of metastasis and death from any cause, with whichever occurs first determining the event time.
Secondary Time-to-Event Endpoints
| Endpoint | Registered time frame | Analysis type |
|---|---|---|
| Overall Survival - Primary Analysis | From randomization of the first subject to the time approximatively 140 death events were observed (approximately 48 months) | Log-rank test; stratified Cox regression |
| Time to Pain Progression - Primary Analysis | From randomization until last study treatment (assessed every 4 months) (approximately 48 months) | Log-rank test; stratified Cox regression |
| Time to Initiation of First Cytotoxic Chemotherapy for Prostate Cancer - Primary Analysis | From randomization until last study treatment (assessed every 4 months) (approximately 48 months) | Log-rank test; stratified Cox regression |
| Time to First Symptomatic Skeletal Event (SSE) - Primary Analysis | From randomization until last study treatment (assessed every 4 months) (approximately 48 months) | Log-rank test; stratified Cox regression |
| Overall Survival - Final Analysis | From randomization of the first subject to the time approximatively 254 death events were observed (approximately 56 months) | Log-rank test; stratified Cox regression |
| Time to Pain Progression - Final Analysis | From randomization until last study treatment (assessed every 4 months) (approximately 48 months) | Log-rank test; stratified Cox regression |
| Time to Initiation of First Cytotoxic Chemotherapy for Prostate Cancer - Final Analysis | From randomization until initiation of first cytotoxic chemotherapy treatment (approximately 59 months) | Log-rank test; stratified Cox regression |
| Time to First Symptomatic Skeletal Event (SSE) - Final Analysis | From randomization until occurrence of first SSE event (approximately 59 months) | Log-rank test; stratified Cox regression |
6. Statistical Methodology
Log-Rank Test
The registry identifies the log-rank test as the reported statistical method for the primary and secondary time-to-event analyses. The log-rank test compares the observed pattern of events between randomized groups over follow-up rather than comparing only a single fixed-time event rate.
This is appropriate for endpoints such as metastasis-free survival, overall survival, pain progression, initiation of cytotoxic chemotherapy, and symptomatic skeletal events because the timing of the event matters and some participants may not experience the event during the observation period.
Hazard Ratio
The effect measure reported for the analyses is the hazard ratio (HR). A hazard ratio below 1 indicates a lower estimated instantaneous event rate in the darolutamide group relative to placebo under the fitted model.
An HR below 1 therefore represents a relative reduction in the estimated hazard. It does not directly represent an absolute probability, median survival difference, or the percentage of patients who benefit.
Stratified Cox Regression
For the primary endpoint, the registry states that the hazard ratio and 95% confidence interval were based on a Cox Regression Model stratified by prostate-specific antigen doubling time (PSADT) (≤ 6 months vs. > 6 months) and use of osteoclast-targeted therapy (No, Yes).
The same stratification factors are specified in the analysis notes for the posted secondary analyses. Stratification allows the time-to-event comparison to account for these prespecified categories without requiring them to enter the model as ordinary regression covariates.
Confidence Intervals
Every posted statistical analysis reports a two-sided 95% confidence interval for the hazard ratio. The interval describes statistical uncertainty around the estimated relative effect under the model and analysis framework. It should not be interpreted as a range containing the effects experienced by individual patients.
Superiority Testing
The primary hypothesis type is reported as superiority. The statistical question is therefore whether the treatment groups differ in the direction specified by the superiority framework, rather than whether darolutamide is merely no worse than placebo according to a non-inferiority margin.
7. Primary Result: Metastasis-Free Survival
ClinicalTrials.gov reports a formal statistical analysis for the primary endpoint. The comparison was between darolutamide (BAY1841788) and placebo using a log-rank test. The hazard ratio and confidence interval were obtained from a Cox regression model stratified by PSADT and use of osteoclast-targeted therapy.
Metastasis-Free Survival
95% CI: 0.341–0.500 · P < 0.000001
Two-sided 95% confidence interval · Superiority hypothesis
| Primary endpoint | Darolutamide vs placebo |
|---|---|
| Outcome | Metastasis-Free Survival |
| Analysis window | From randomization to the time approximately 385 MFS events were observed (approximately 48 months) |
| Method | Log Rank |
| Effect measure | Hazard Ratio (HR) |
| Estimate | 0.413 |
| 95% CI | 0.341–0.500 |
| P-value | <0.000001 |
| Hypothesis | Superiority |
The estimated HR of 0.413 means that, under the stratified Cox model, the estimated instantaneous rate of experiencing the MFS event was about 58.7% lower with darolutamide than with placebo. This follows directly from 1 − 0.413 = 0.587, but it should be described as a model-based relative hazard interpretation.
It does not mean that 58.7% of patients avoided metastasis, that 58.7% were cured, or that every patient experienced the same proportional reduction in risk. MFS is a time-to-event endpoint, and patients can contribute follow-up without experiencing an event.
The 95% CI of 0.341–0.500 quantifies uncertainty around the estimated hazard ratio under the specified statistical framework. Its width provides information about precision: the estimate is not a single certainty, but the observed data support a range of plausible model-based relative hazard values.
The p-value of <0.000001 addresses the evidence against the relevant null hypothesis within the reported testing framework. It does not measure the magnitude of the treatment effect, the clinical importance of the effect, or the probability that the treatment works.
Because the hazard ratio comes from a Cox regression model, its interpretation also depends on the model's time-to-event assumptions. The registry does not provide sufficient information in the ClinicalTrials.gov record to independently evaluate the proportional-hazards assumption.
8. Secondary Endpoint Results
ClinicalTrials.gov reports eight secondary statistical analyses in the ClinicalTrials.gov record. All use the same general survival-analysis framework: log-rank testing, with the hazard ratio and 95% confidence interval based on a Cox regression model stratified by PSADT and use of osteoclast-targeted therapy.
Overall Survival — Primary Analysis
Overall Survival
95% CI: 0.501–0.994 · P = 0.045210
The HR of 0.706 corresponds to an estimated instantaneous rate of death approximately 29.4% lower with darolutamide than with placebo under the reported Cox model. This is a relative hazard interpretation, not a statement that 29.4% of patients avoided death or that individual patients experienced a 29.4% reduction in risk.
The 95% CI of 0.501–0.994 expresses uncertainty around the estimate. The interval is comparatively broad relative to the point estimate, so the point estimate should not be treated as exact. The p-value of 0.045210 describes statistical evidence under the specified analysis; it does not quantify the size or clinical importance of the effect.
The registry identifies the analysis as a superiority comparison and specifies a stratified Cox model. As with the other hazard ratios, the interpretation depends on the time-to-event model and censoring framework.
Time to Pain Progression — Primary Analysis
Time to Pain Progression
95% CI: 0.533–0.785 · P = 0.000008
The HR of 0.647 indicates an estimated instantaneous event rate approximately 35.3% lower in the darolutamide group than in the placebo group under the reported model. It is not a direct estimate of the percentage of patients who experienced pain progression or a statement about the magnitude of pain improvement in an individual patient.
The 95% CI of 0.533–0.785 shows the uncertainty around the hazard ratio. The p-value of 0.000008 addresses statistical evidence under the reported comparison and should not be used as a substitute for the effect estimate or confidence interval.
Time to Initiation of First Cytotoxic Chemotherapy for Prostate Cancer — Primary Analysis
Time to First Cytotoxic Chemotherapy
95% CI: 0.314–0.595 · P < 0.000001
The HR of 0.433 corresponds to an estimated instantaneous event rate approximately 56.7% lower with darolutamide than with placebo under the stratified Cox model. The endpoint is time to initiation of first cytotoxic chemotherapy for prostate cancer; the hazard ratio therefore concerns the timing of that event, not an absolute probability of receiving chemotherapy.
The 95% CI of 0.314–0.595 describes uncertainty around the relative hazard estimate. The p-value of <0.000001 provides evidence against the relevant null hypothesis within the reported testing framework but does not itself describe effect size.
Time to First Symptomatic Skeletal Event — Primary Analysis
Time to First SSE
95% CI: 0.218–0.842 · P = 0.011262
The HR of 0.428 corresponds to an estimated instantaneous event rate approximately 57.2% lower in the darolutamide group than in the placebo group under the reported Cox model. This does not mean that 57.2% of patients avoided an SSE or that the absolute probability of an SSE was reduced by 57.2%.
The 95% CI of 0.218–0.842 is relatively wide, indicating meaningful uncertainty around the point estimate. The p-value of 0.011262 measures statistical evidence under the reported test; it does not measure the size or clinical importance of the effect.
Overall Survival — Final Analysis
Overall Survival
95% CI: 0.533–0.881 · P = 0.003048
The final-analysis HR of 0.685 corresponds to an estimated instantaneous rate of death approximately 31.5% lower with darolutamide than with placebo under the reported stratified Cox model.
The 95% CI of 0.533–0.881 quantifies uncertainty around this estimate. The p-value of 0.003048 describes statistical evidence under the reported analysis and is not a measure of effect size. This final analysis also uses a different reported event target and time frame from the earlier Overall Survival primary analysis, so the two HRs should not be silently treated as though they came from the same analysis cutoff.
Time to Pain Progression — Final Analysis
Time to Pain Progression
95% CI: 0.533–0.785 · P = 0.000008
The final-analysis HR of 0.647 corresponds to an estimated instantaneous event rate approximately 35.3% lower in the darolutamide group under the reported model. The estimate is identical to the registry-reported primary-analysis HR for this endpoint, but the registry identifies the entries as separate primary and final analyses.
The 95% CI of 0.533–0.785 describes uncertainty around the estimate, while the p-value of 0.000008 addresses statistical evidence rather than effect magnitude.
Time to Initiation of First Cytotoxic Chemotherapy — Final Analysis
Time to First Cytotoxic Chemotherapy
95% CI: 0.444–0.755 · P = 0.000044
The final-analysis HR of 0.579 corresponds to an estimated instantaneous event rate approximately 42.1% lower with darolutamide than with placebo under the reported Cox model.
The 95% CI of 0.444–0.755 describes uncertainty around the relative hazard. The p-value of 0.000044 provides statistical evidence within the reported analysis but does not indicate the magnitude or clinical importance of the treatment effect.
Time to First Symptomatic Skeletal Event — Final Analysis
Time to First SSE
95% CI: 0.287–0.815 · P = 0.005294
The final-analysis HR of 0.484 corresponds to an estimated instantaneous event rate approximately 51.6% lower with darolutamide than with placebo under the reported Cox model.
The 95% CI of 0.287–0.815 indicates uncertainty around the estimate, with a relatively broad interval. The p-value of 0.005294 addresses statistical evidence under the reported analysis rather than effect size.
9. Results Summary
The posted statistical analyses form a coherent time-to-event analysis program. The primary endpoint, MFS, had an HR of 0.413 with a 95% CI of 0.341–0.500 and a p-value of <0.000001. The secondary analyses also reported hazard ratios below 1 for overall survival, pain progression, initiation of first cytotoxic chemotherapy, and first symptomatic skeletal event.
| Endpoint | Analysis | HR | 95% CI | P-value |
|---|---|---|---|---|
| Metastasis-Free Survival | Primary | 0.413 | 0.341–0.500 | <0.000001 |
| Overall Survival | Primary Analysis | 0.706 | 0.501–0.994 | 0.045210 |
| Time to Pain Progression | Primary Analysis | 0.647 | 0.533–0.785 | 0.000008 |
| Time to Initiation of First Cytotoxic Chemotherapy | Primary Analysis | 0.433 | 0.314–0.595 | <0.000001 |
| Time to First SSE | Primary Analysis | 0.428 | 0.218–0.842 | 0.011262 |
| Overall Survival | Final Analysis | 0.685 | 0.533–0.881 | 0.003048 |
| Time to Pain Progression | Final Analysis | 0.647 | 0.533–0.785 | 0.000008 |
| Time to Initiation of First Cytotoxic Chemotherapy | Final Analysis | 0.579 | 0.444–0.755 | 0.000044 |
| Time to First SSE | Final Analysis | 0.484 | 0.287–0.815 | 0.005294 |
10. Stratification and Why It Matters
The registry-reported analysis notes repeatedly identify two stratification factors for the Cox regression model: PSADT (≤ 6 months vs. > 6 months) and use of osteoclast-targeted therapy (No, Yes).
| Stratification factor | Reported categories | Statistical role |
|---|---|---|
| Prostate-specific antigen doubling time (PSADT) | ≤ 6 months vs. > 6 months | Stratification factor in the Cox regression model |
| Use of osteoclast-targeted therapy | No, Yes | Stratification factor in the Cox regression model |
Stratification is important because the reported hazard ratios are not simply unadjusted ratios of two crude event rates. The Cox analysis estimates the treatment effect while preserving the specified stratified structure. Consequently, the HR should be interpreted as the treatment comparison produced by that model rather than as a generic ratio of two raw probabilities.
Why PSADT matters statistically
The analysis distinguishes PSADT categories of ≤ 6 months and > 6 months. The ClinicalTrials.gov record identifies this factor as part of the stratified Cox model.
Why osteoclast-targeted therapy matters
Use of osteoclast-targeted therapy is also incorporated through model stratification, with categories of No and Yes.
11. Statistical Methods Explained
Why was a log-rank test used?
The endpoints are time-to-event outcomes. A log-rank test compares the survival experience of the randomized groups over follow-up and makes use of the timing of events. This is more appropriate for MFS or overall survival than a simple comparison of the proportion of patients with an event at one arbitrary time point.
What does an HR of 0.413 mean?
An HR of 0.413 means that the fitted model estimates the instantaneous MFS event rate in the darolutamide group to be 0.413 times that in the placebo group. Expressed as a relative reduction in estimated hazard, this is approximately 58.7%. It does not mean that 58.7% of patients were protected from metastasis or death.
Why is the confidence interval important?
The point estimate is only one estimate of the treatment effect. The 95% CI of 0.341–0.500 for MFS communicates the uncertainty around the HR under the model and sampling framework. A confidence interval also helps distinguish a precise estimate from an estimate that is compatible with a much wider range of effects.
Why doesn't the p-value measure effect size?
A p-value addresses evidence against a null hypothesis under a specified statistical model and testing procedure. It does not tell us whether an HR is 0.9, 0.5, or 0.1. The effect size is described by the hazard ratio and its confidence interval; the p-value provides complementary information about statistical evidence.
Why was stratified Cox regression used?
The registry specifically states that the hazard ratio and 95% CI were based on a Cox regression model stratified by PSADT and use of osteoclast-targeted therapy. This allows the treatment comparison to be estimated within the specified strata while accounting for the stratified analysis structure.
Why should the primary and final analyses not be combined?
The registry identifies separate primary and final analyses for several secondary endpoints and supplies different event targets or time frames for some of them. For example, Overall Survival has a primary analysis based on approximately 140 death events and a final analysis based on approximately 254 death events. These are therefore distinct analyses and should retain their separate labels.
12. Understanding the Primary MFS Endpoint
MFS is especially instructive statistically because it combines progression to metastasis with death from any cause. The registered definition states that MFS is the time from randomisation to evidence of metastasis or death from any cause, whichever occurs first.
A participant who experiences either component first has an MFS event at that time. A participant without either event by the relevant follow-up point contributes time-to-event information subject to the trial's censoring framework.
This construction makes MFS different from overall survival. Overall survival uses death from any cause as the event, whereas MFS reaches its event definition when metastasis or death occurs first. The two endpoints can therefore provide related but non-identical information about disease progression and survival.
13. Primary vs Secondary Analyses
the ClinicalTrials.gov record identifies one registered primary endpoint and eight posted secondary statistical analyses. The distinction between primary and secondary analyses matters because statistical interpretation depends not only on the numerical estimate but also on the role assigned to the endpoint in the trial design.
| Role | Endpoint | Number of posted analyses |
|---|---|---|
| Primary | Metastasis-Free Survival | 1 |
| Secondary | Overall Survival | 2 |
| Secondary | Time to Pain Progression | 2 |
| Secondary | Time to Initiation of First Cytotoxic Chemotherapy for Prostate Cancer | 2 |
| Secondary | Time to First Symptomatic Skeletal Event | 2 |
The ClinicalTrials.gov record does not provide a multiplicity-adjustment scheme, alpha-spending rule, interim-analysis plan, or endpoint hierarchy beyond identifying the primary and secondary roles. Those design details therefore should not be inferred from the reported p-values.
14. Safety: Serious Adverse Events
The ClinicalTrials.gov record reports serious adverse events by arm using affected and at-risk counts. Because the data provide these counts in several exposure categories, they are reproduced exactly rather than combined into an independently calculated overall rate.
| Group | Serious adverse events affected / at risk |
|---|---|
| Darolutamide (DB) | 251 / 954 |
| Darolutamide (DB+OL) | 367 / 954 |
| Placebo (DB) | 121 / 554 |
| Placebo+Darolutamide (CO) | 43 / 170 |
The safety data also illustrate why efficacy and safety analyses answer different questions. The efficacy analysis is based on randomized treatment comparison and time-to-event methods, whereas the ClinicalTrials.gov record is presented as affected and at-risk counts in specified treatment/exposure categories.
15. What the Hazard Ratios Do — and Do Not — Mean
The primary MFS HR of 0.413 is a relative measure of the event rate estimated by the stratified Cox model. It does not directly provide the probability that an individual patient will develop metastasis or die.
The ClinicalTrials.gov record does not provide median MFS, Kaplan-Meier survival probabilities at specified time points, or absolute event rates by randomized arm. Those quantities should not be inferred from the hazard ratio.
The MFS p-value of <0.000001 indicates strong statistical evidence under the reported analysis framework. It does not mean there is a probability of less than 0.000001 that the null hypothesis is true, nor does it measure clinical magnitude.
The MFS 95% CI of 0.341–0.500 provides a more informative description of uncertainty than the point estimate alone. Confidence intervals should be considered together with the HR and the endpoint definition.
16. Censoring and Time-to-Event Interpretation
Time-to-event analysis differs from an ordinary binary outcome because participants may have different amounts of observed follow-up. The registered endpoints specify time from randomization to a defined event, and the reported methods are survival-analysis methods.
In a Kaplan-Meier framework, participants who have not experienced the event at their last usable follow-up can be censored rather than treated as though the event never occurred. The ClinicalTrials.gov record does not provide the individual event and censoring records needed to reproduce the survival curves, so the page does not attempt to reconstruct them.
Two groups can have similar numbers of events but different event timing, or similar event timing among observed events but different censoring patterns. The log-rank and Cox methods are designed for this time-to-event structure.
17. What the Registry Data Do Not Establish
The ClinicalTrials.gov record is sufficiently detailed to analyze the registered primary endpoint and eight secondary statistical analyses, but several common trial-analysis elements are not provided. They should therefore not be reconstructed from memory or inferred from the reported hazard ratios.
| Topic | What can be stated from the ClinicalTrials.gov record |
|---|---|
| Median MFS or OS | Not provided in the ClinicalTrials.gov record. |
| Kaplan-Meier coordinates | Not provided. |
| Baseline characteristics | Not provided. |
| Subgroup hazard ratios | Not provided. |
| Forest plot estimates | Not provided. |
| Multiplicity adjustment | Not provided. |
| Interim-analysis plan | Not provided. |
| Missing-data or imputation strategy | Not provided. |
| Non-inferiority margin | Not applicable to the reported superiority hypothesis in the ClinicalTrials.gov record. |
| Bayesian methods | No Bayesian method is reported; the analyses posted on ClinicalTrials.gov use log-rank and Cox methods. |
This distinction is important for reproducible statistical interpretation. A trial-results page should distinguish what the registry actually reports from methodological features that might commonly appear in similar trials but are not documented in the ClinicalTrials.gov record.
18. Limitations
- Registry-level reporting: the ClinicalTrials.gov record provides summary estimates and model descriptions but not individual-level event and censoring data.
- No absolute survival estimates: median survival and time-specific Kaplan-Meier estimates are not included in the ClinicalTrials.gov record, so the hazard ratios cannot be translated into absolute survival differences here.
- Model dependence: the hazard ratios are based on Cox regression and therefore represent model-based relative effects rather than simple raw risk ratios.
- Proportional-hazards assumption: the ClinicalTrials.gov record does not report an assessment of the proportional-hazards assumption. A single HR should therefore not be interpreted as a complete description of the treatment-effect trajectory over time.
- Stratified analysis: the reported estimates account for PSADT and osteoclast-targeted therapy through stratification, so they should not be interpreted as unadjusted crude comparisons.
- Multiple endpoints: the trial has one registered primary endpoint and multiple secondary analyses. The ClinicalTrials.gov record does not specify a multiplicity-adjustment scheme for the secondary p-values.
- Separate analysis cutoffs: the primary and final secondary analyses use different event targets or time frames in the registry and should not be combined into one undifferentiated result.
- Safety denominators: serious adverse-event counts are reported under several exposure labels and should not be merged into a single rate without additional information.
- Limited subgroup information: subgroup estimates and interaction tests are not included in the ClinicalTrials.gov record, so treatment-effect heterogeneity cannot be evaluated from this record alone.
19. Why This Trial Matters Statistically
ARAMIS is a useful teaching example because its registered primary endpoint, statistical analysis, and secondary outcomes all center on time-to-event methodology. The trial therefore illustrates how a modern randomized study can use one statistical framework across several clinically different event definitions.
| Statistical concept | How it appears in ARAMIS |
|---|---|
| Randomization | The study uses randomized allocation in a parallel design. |
| Quadruple masking | The registry identifies the trial as quadruple masked. |
| Time-to-event endpoints | MFS is the registered primary endpoint, with OS, pain progression, chemotherapy initiation, and SSE among the secondary endpoints. |
| Log-rank test | The reported statistical method for the primary and secondary time-to-event comparisons. |
| Hazard ratio | The reported effect measure for all nine posted statistical analyses. |
| Cox regression | Used to obtain the reported HRs and 95% confidence intervals. |
| Stratification | Models were stratified by PSADT and use of osteoclast-targeted therapy. |
| Confidence intervals | Each posted analysis includes a two-sided 95% CI. |
| Superiority | The primary hypothesis type is reported as superiority. |
| Multiple analyses | One primary endpoint and eight secondary statistical analyses are posted. |
20. Statistical Interpretation vs Clinical Interpretation
Statistical interpretation
The reported primary MFS analysis produced an HR of 0.413 with a two-sided 95% CI of 0.341–0.500 and P < 0.000001 using a log-rank comparison and stratified Cox regression.
Clinical interpretation
The statistical result concerns time to metastasis or death as defined by the registered MFS endpoint. The ClinicalTrials.gov record does not provide median MFS, absolute survival probabilities, or individual patient outcomes, so those quantities cannot be inferred here.
The distinction matters. A hazard ratio is a compact relative summary of a time-to-event comparison, whereas clinical interpretation often requires absolute event probabilities, event timing, duration of follow-up, treatment exposure, and safety information. Only some of those elements are present in the ClinicalTrials.gov record.
21. A Worked Reading of the Primary Result
Consider the primary estimate step by step:
MFS is time from randomisation to evidence of metastasis or death from any cause, whichever occurs first.
Darolutamide (BAY1841788) vs placebo.
Hazard ratio, estimated using a stratified Cox regression model.
An HR below 1 indicates a lower estimated instantaneous event rate in the darolutamide group under the model.
The confidence interval communicates uncertainty around the estimated hazard ratio.
The p-value indicates statistical evidence under the reported superiority testing framework; it does not measure the size of the treatment effect.
22. Secondary Endpoint Pattern
The secondary results provide a useful illustration of why effect estimates should be read individually rather than summarized only by their p-values. The reported hazard ratios range from 0.428 to 0.706 across the registry-reported primary secondary analyses and from 0.484 to 0.685 across the registry-reported final analyses.
These values describe different endpoints and, in some cases, different analysis time frames. An HR of 0.706 for an overall-survival analysis cannot be interpreted as though it were the same quantity as an HR of 0.433 for time to initiation of first cytotoxic chemotherapy. The endpoint definition, event process, analysis cutoff, and clinical meaning are different.
23. Related Tutorials
Learn more about the methods used in this trial:
24. Related Calculators
25. Sources
- ClinicalTrials.gov: ARAMIS (NCT02200614), the official trial registry record used for the trial design, endpoint definitions, statistical analyses, and safety data presented on this page.
- PubMed: PubMed record 41450200.
- PubMed: PubMed record 40132221.
- PubMed: PubMed record 38394384.
- PubMed: PubMed record 37660438.
- PubMed: PubMed record 37222151.
Continue through the Clinical Biostats statistical pathway
This trial connects randomized clinical-trial design with survival analysis, hazard ratios, confidence intervals, log-rank testing, and stratified Cox regression.
26. Record Summary
ARAMIS is a clear example of a randomized phase 3 time-to-event trial in which the primary endpoint and multiple secondary endpoints were analyzed using a consistent survival-analysis framework. The primary MFS analysis reported an HR of 0.413 with a two-sided 95% CI of 0.341–0.500 and P < 0.000001. The registry describes a log-rank comparison and a stratified Cox regression model using PSADT and use of osteoclast-targeted therapy as stratification factors.
The secondary analyses similarly report hazard ratios below 1 for overall survival, pain progression, initiation of first cytotoxic chemotherapy, and first symptomatic skeletal event. The most statistically informative way to read these findings is to keep the endpoint definition, analysis time frame, hazard ratio, confidence interval, and p-value together rather than reducing each result to a single significance label.