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
MIRROS was a randomized, double-blind phase 3 trial testing whether adding idasanutlin, an MDM2 antagonist whose mechanism depends on functional p53, to cytarabine improved overall survival in patients with relapsed or refractory AML. The primary analysis was restricted to patients whose tumors were TP53 wild-type. The trial was terminated because of futility.
| Feature | MIRROS |
|---|---|
| Official title | A Study of Idasanutlin With Cytarabine Versus Cytarabine Plus Placebo in Participants With Relapsed or Refractory Acute Myeloid Leukemia (AML) |
| Phase | Phase 3 |
| Condition | Leukemia, Myeloid, Acute |
| Design | Randomized, parallel-group, double-masked, placebo-controlled |
| Arms | Idasanutlin + cytarabine; placebo + cytarabine |
| Enrollment | 447 |
| Primary endpoint | Overall survival in the TP53 wild-type population |
| Primary analysis | Log-rank test; hazard ratio with two-sided 95% CI; TP53-WT ITT population |
| Dates | Start 2015-12-30; primary completion 2020-04-24 |
| Status | Terminated (futility); results posted |
| Lead sponsor | Hoffmann-La Roche (industry) |
| ClinicalTrials.gov | NCT02545283 |
2. Clinical Question
Idasanutlin is designed to block MDM2, a negative regulator of p53, and so is expected to act only in cells with a functional (wild-type) TP53 gene. MIRROS therefore asked a biomarker-focused question: among patients with relapsed or refractory AML and TP53 wild-type disease, does adding idasanutlin to cytarabine extend survival?
Population
Participants with relapsed or refractory acute myeloid leukemia; primary analysis in the TP53 wild-type population.
Intervention
Idasanutlin in combination with cytarabine.
Comparator
Placebo in combination with cytarabine.
Primary question
Does idasanutlin plus cytarabine improve overall survival compared with placebo plus cytarabine in TP53 wild-type patients?
3. Trial Design
Idasanutlin + cytarabine
- Idasanutlin
- Cytarabine
Placebo + cytarabine
- Placebo
- Cytarabine
The safety denominators (292 and 149) indicate unequal allocation, with roughly twice as many patients treated in the idasanutlin arm. The safety total of 441 is slightly below the enrollment of 447, reflecting randomized patients who did not receive study treatment.
4. Endpoints
| Endpoint (registry wording) | Description | Time frame |
|---|---|---|
| Overall Survival in TP53 WT Population | Overall survival in the TP53 wild-type population, compared between participants with R/R AML randomized to idasanutlin plus cytarabine and those randomized to cytarabine plus placebo. The study was terminated because of futility and did not reach the planned end of the study. | From randomization to death from any cause (up to approximately 4.5 years) |
Overall survival is the most objective oncology endpoint: death from any cause is unambiguous and is not affected by assessment schedules. The registry does not post statistical analyses for secondary endpoints.
5. Analysis Population
| Population | Definition / role |
|---|---|
| TP53-WT ITT population | Randomized participants with TP53 wild-type disease, analysed according to randomized assignment; used for the primary OS analysis |
| Safety population | Treated participants: 292 (idasanutlin + cytarabine) and 149 (placebo + cytarabine); used for serious adverse events |
Restricting the primary analysis to the TP53 wild-type population is a biomarker-enrichment strategy. Because TP53 status is measured before treatment, the TP53-WT subset is still a randomized comparison: randomization protects balance within any subgroup defined by a baseline characteristic. The registry does not report the number of patients in the TP53-WT ITT population in the posted analysis.
6. Primary Endpoint Result: Overall Survival in the TP53 WT Population
Hazard ratio for death
Two-sided 95% CI: 0.81–1.45 · P = 0.5752 (log-rank)
Groups compared: placebo + cytarabine vs idasanutlin + cytarabine · TP53-WT ITT population
What the estimate means. The estimated hazard ratio of 1.08 is close to 1, indicating no detectable difference in the rate of death between arms in the TP53 wild-type population. The registry lists the comparison as placebo plus cytarabine versus idasanutlin plus cytarabine without spelling out which arm forms the numerator. If the HR is expressed as idasanutlin relative to placebo, it corresponds to an 8% higher estimated hazard with idasanutlin; in either direction, the point estimate does not favour idasanutlin materially.
What the confidence interval says. The 95% CI of 0.81–1.45 includes 1. It is compatible with a modest reduction in hazard of up to 19% at one end and an increase of up to 45% at the other. The interval is not narrow, but it makes a large survival benefit unlikely, which is consistent with the futility decision.
What it does not mean. A non-significant HR is not proof that idasanutlin has no effect whatsoever. It shows that the trial did not find evidence of a survival benefit and that the plausible range of effects under the model includes no difference and some harm.
Why the p-value is not an effect size. P = 0.5752 indicates that the observed difference is well within what chance alone would produce if the arms had identical survival. It says nothing about how large or small any true effect is; the HR and its CI carry that information.
Cautions. The analysis was performed after the trial stopped early, so follow-up was shorter than planned and estimates carry more uncertainty than a completed trial would provide. The HR assumes proportional hazards; in AML, early treatment-related deaths and later disease-related deaths may produce hazards that change over time. The primary analysis applies to TP53 wild-type patients and not to the full enrolled population.
7. Safety: Serious Adverse Events
| Measure | Idasanutlin + cytarabine | Placebo + cytarabine |
|---|---|---|
| Patients with a serious adverse event / at risk | 173 / 292 (59.2%) | 72 / 149 (48.3%) |
Serious adverse events were common in both arms, as expected with cytarabine-based therapy in relapsed or refractory AML, and more frequent with idasanutlin. The registry reports these counts without a formal statistical comparison, and they are not adjusted for differences in treatment duration. Taken together with the neutral OS result, the safety data do not offset the absence of a survival benefit.
8. Statistical Methodology
Kaplan-Meier estimation
Overall survival is summarised with the Kaplan-Meier estimator, which handles right-censoring. In a trial stopped for futility, many surviving patients are censored at the date of the final data cut-off, so the tail of the survival curve is estimated from fewer patients.
di is the number of deaths at time ti and ni the number still at risk just before ti.
Log-rank test
The primary comparison used the log-rank test, which compares the observed number of deaths in each arm with the number expected if survival were identical, accumulated across all event times. It is most powerful when the hazard ratio is roughly constant over time and loses power if curves cross.
Hazard ratio
The hazard ratio of 1.08 with a two-sided 95% CI summarises the relative rate of death between arms, typically estimated with a Cox proportional-hazards model. The registry lists the statistical method as log-rank and reports the HR as the effect measure; it does not describe stratification factors.
An HR is a relative measure averaged over follow-up. It does not describe the proportion of patients who benefit or the difference in median survival.
Intention-to-treat within a biomarker population
The TP53-WT ITT population analyses all randomized TP53 wild-type patients as randomized. This preserves the protection of randomization within the biomarker-defined group, while the safety population contains only treated patients.
Futility monitoring
Futility rules allow a trial to stop early when accumulating data suggest the primary objective will not be met. Unlike efficacy stopping, futility stopping does not inflate the type I error, because it can only prevent a positive conclusion. It does mean that final estimates are based on less information than planned.
9. Statistical Methods Explained
What does a hazard ratio of 1.08 mean?
It indicates that the estimated instantaneous death rates in the two arms were very similar. If read as idasanutlin relative to placebo, the estimated hazard of death was 8% higher with idasanutlin, but the confidence interval (0.81–1.45) is wide enough that this small difference in either direction is fully compatible with chance.
Why does the direction of the comparison matter?
An HR of 1.08 for A versus B is equivalent to about 0.93 for B versus A. The registry lists the groups as placebo plus cytarabine versus idasanutlin plus cytarabine. When direction is not explicit, the safest summary is the one that does not depend on it: the CI includes 1, so no survival difference was demonstrated.
Why was the primary analysis limited to TP53 wild-type patients?
Idasanutlin's mechanism requires functional p53, so the treatment effect was expected to be concentrated in TP53 wild-type disease. Focusing the primary test on that group increases power if the hypothesis is right. Because TP53 status is a baseline characteristic, the comparison within this group remains randomized.
Why is a large p-value like 0.5752 not evidence that the drugs are equivalent?
A p-value measures how compatible the data are with no difference; it cannot confirm that no difference exists. Equivalence would require a prespecified margin and a confidence interval lying entirely within it. Here the CI extends from 0.81 to 1.45, too wide to establish equivalence.
What does stopping for futility imply statistically?
A futility stop means an interim look showed a low probability of reaching a positive result. The final HR and CI reflect the data at that point. Because futility stopping cannot create a false-positive claim, the result is interpretable, but the reduced follow-up leaves more uncertainty than a completed trial would have.
10. Limitations and Interpretation Issues
- Early termination: the trial stopped for futility before its planned end, so survival follow-up was shorter and estimates less precise than designed.
- Direction of the HR: the registry's group ordering does not make the numerator arm explicit.
- Biomarker population: the primary result applies to TP53 wild-type patients; the number of patients in that population is not stated in the posted analysis.
- Proportional hazards: the HR summarises the whole follow-up; early and late hazards may differ in AML.
- Limited posted analyses: the registry posts no statistical analyses of secondary endpoints, so response and other outcomes cannot be evaluated here.
- Safety data: serious adverse events are unadjusted counts without a formal comparison or exposure adjustment.
11. Why This Trial Matters Statistically
MIRROS illustrates how to read a neutral phase 3 result: a hazard ratio near 1, a confidence interval that includes both modest benefit and meaningful harm, a large p-value that does not prove equivalence, and early termination for futility within a biomarker-selected population.
| Concept | How it appears in MIRROS |
|---|---|
| Time-to-event endpoints | Overall survival from randomization to death from any cause |
| Kaplan-Meier estimation | Survival distributions with censoring at the final data cut-off |
| Log-rank test | Primary comparison, P = 0.5752 |
| Hazard ratio | HR 1.08 in the TP53 wild-type population |
| Confidence intervals | 0.81–1.45, including 1 and ruling out a large benefit |
| P-values | A non-significant result is not evidence of equivalence |
| Intention-to-treat analysis | TP53-WT ITT population analysed as randomized |
12. Related Tutorials
Learn more about the methods used in this trial:
13. Related Calculators
14. Sources
- ClinicalTrials.gov: NCT02545283 – A Study of Idasanutlin With Cytarabine Versus Cytarabine Plus Placebo in Participants With Relapsed or Refractory Acute Myeloid Leukemia (AML).
- PubMed: PMID 32167393.
Explore more trial analyses
Each trial page connects its endpoints and methods to statistical tutorials and calculators.
15. Record Summary
MIRROS enrolled 447 patients with relapsed or refractory AML and compared idasanutlin plus cytarabine with placebo plus cytarabine. In the TP53 wild-type population, the overall survival hazard ratio was 1.08 (95% CI 0.81–1.45; P = 0.5752), and the trial was terminated for futility. Serious adverse events occurred in 59.2% versus 48.3% of treated patients. The most informative reading focuses on the confidence interval, which excludes a large survival benefit, on the limits of a futility-terminated trial, and on the distinction between absence of evidence and evidence of equivalence.