This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical trial information on this page is restricted to the ClinicalTrials.gov record. The registry identifies the primary endpoint and states that results have been posted, but the ClinicalTrials.gov record contains no formal statistical analyses for that endpoint. Accordingly, this page distinguishes registered endpoint information from statistical interpretation of how that endpoint would normally be analysed.
1. Trial at a Glance
AURA2 was a phase 2, open-label, single-group study of AZD9291 in patients with non-small cell lung cancer after previous EGFR TKI therapy in EGFR and T790M mutation-positive tumours. The study enrolled 210 participants and registered objective response rate as its primary endpoint.
| Feature | AURA2 |
|---|---|
| Trial name | AURA2 |
| Phase | Phase 2 |
| Condition | Non Small Cell Lung Cancer |
| Brief title | Phase II AZD9291 Open Label Study in NSCLC After Previous EGFR TKI Therapy in EGFR and T790M Mutation Positive Tumours |
| Allocation | NA |
| Design model | SINGLE_GROUP |
| Masking | NONE |
| Primary purpose | TREATMENT |
| Intervention | AZD9291 (drug) |
| Enrollment | 210.0 |
| Arms | 1 |
| Primary endpoint type | Binary |
| Outcome measures posted | 4 |
| Results posted | Yes |
| Status | COMPLETED |
| Lead sponsor | AstraZeneca |
| Sponsor type | INDUSTRY |
| ClinicalTrials.gov | NCT02094261 |
2. Clinical Question
The AURA2 study addressed a treatment question in patients with non-small cell lung cancer after previous EGFR TKI therapy whose tumours were EGFR and T790M mutation positive. The registered intervention was AZD9291, and the primary endpoint was objective response rate.
Population
Patients with non-small cell lung cancer after previous EGFR TKI therapy in EGFR and T790M mutation-positive tumours.
Intervention
AZD9291 (drug), administered in the single study arm.
Comparator
The registry design is SINGLE_GROUP with 1 arm. There is therefore no randomized comparator arm in the ClinicalTrials.gov record.
Primary question
What proportion of patients experienced at least 1 visit response meeting the registered objective-response criteria of complete response or partial response?
3. Trial Design
AZD9291
- Intervention type: drug
- Registered intervention: AZD9291
- Serious adverse events: 42/210 in the AZD9291 80mg arm
Single-group study
- Arms: 1
- Randomized comparison: not part of the registry-reported design
- Between-arm treatment-effect estimates are therefore not available from the ClinicalTrials.gov record
Study timing
Study start
The ClinicalTrials.gov record lists May 20, 2014 as the study start date.
Primary completion
The ClinicalTrials.gov record lists May 1, 2015 as the primary completion date.
Current registry status in the ClinicalTrials.gov record
The study is identified as COMPLETED.
4. Primary Endpoint
The registered primary endpoint is Objective Response Rate (ORR), classified in the ClinicalTrials.gov record as a binary endpoint.
| Endpoint | Registry definition / time frame | Statistical type |
|---|---|---|
| Objective Response Rate (ORR) | Per Response Evaluation Criteria in Solid Tumours (RECIST v1.1) assessed by MRI or CT: Complete Response (CR): Disappearance of all target and non-target lesions and no new lesions; Partial Response (PR): ≥ 30% decrease in the sum of diameters of Target Lesions (compared to baseline) and no new lesions. ORR is the percentage of patients with at least 1 visit response of CR or PR (according to independent review) that was confirmed at least 4 weeks later, prior to progression or further anti-cancer therapy. | Binary |
| Time frame | RECIST tumour assessments every 6 weeks from first dose until objective disease progression, up to approximately 11 months (at the time of analysis) | Repeated tumour assessments |
What ORR measures
ORR is a binary patient-level outcome derived from tumour-response assessments. A patient is counted as an objective responder when the registered response framework identifies at least one visit response of complete response or partial response. The resulting ORR is expressed as a proportion or percentage of the analyzed patients.
The numerator is the number of patients meeting the registered CR or PR response criterion. The denominator must be defined by the prespecified analysis population. The ClinicalTrials.gov record does not provide the responder count, the formal denominator used for the posted outcome analysis, or a statistical estimate and confidence interval.
5. Planned Analysis
The ClinicalTrials.gov record states that results have been posted and identify ORR as the primary endpoint, but no formal statistical analyses were posted. Consequently, there is no numerical primary-endpoint result, confidence interval, or p-value available in the ClinicalTrials.gov record for this page to report.
Primary endpoint: Objective Response Rate
The registry data identify ORR as the primary binary endpoint but do not provide a statistical estimate, confidence interval, or p-value in the posted results.
For a single-group binary endpoint such as ORR, the principal descriptive analysis would normally be the proportion of patients achieving CR or PR. Statistical uncertainty would commonly be expressed with a confidence interval for that proportion. If the protocol specified a reference response rate, a one-sample hypothesis test could additionally compare the observed ORR with that prespecified benchmark. The ClinicalTrials.gov record does not identify such a benchmark or provide the corresponding test.
6. Statistical Methodology
Binary endpoint analysis
ORR is classified as a binary endpoint in the ClinicalTrials.gov record. At the patient level, each participant can be classified according to whether the registered response criterion is met. The aggregate endpoint is therefore a proportion.
Here, x represents the number of patients classified as responders and n represents the prespecified analysis population. The ClinicalTrials.gov record provides neither x nor the formal outcome-analysis denominator.
Confidence intervals for ORR
A confidence interval for a single proportion quantifies uncertainty associated with estimating the underlying response probability from a finite sample. For a response endpoint, the interval is generally more informative than reporting the observed percentage alone because it shows how precisely the study estimates the response proportion.
For a binary endpoint, the appropriate interval method should be selected in accordance with the prespecified statistical analysis plan and the observed number of responders. The ClinicalTrials.gov record does not specify the confidence-interval method.
Why a single-group trial is statistically different
In a randomized comparative trial, the main efficacy quantity may be a treatment difference, ratio, or hazard ratio between groups. A single-group design has no randomized comparator. Its direct statistical output is instead the response proportion within the treated cohort and, where prespecified, its comparison with an external benchmark.
RECIST-based response assessment
The registered endpoint uses RECIST v1.1 and tumour assessments by MRI or CT. The registry-reported definition identifies complete response as disappearance of all target and non-target lesions with no new lesions, and partial response as at least a 30% decrease in the sum of diameters of target lesions compared with baseline with no new lesions.
Repeated assessments and patient-level classification
Assessments were scheduled every 6 weeks from first dose until objective disease progression, up to approximately 11 months (at the time of analysis) in the registry-reported time-frame text. Because ORR is defined as the percentage of patients with at least one visit response of CR or PR, the endpoint is not simply the percentage of individual tumour assessments that show response. The unit of analysis is the patient, using the registered patient-level response rule.
7. Statistical Methods Explained
What does a binary primary endpoint mean?
A binary endpoint places each patient into one of two outcome categories for the analysis. For AURA2, the registered primary endpoint is ORR, so the relevant classification is whether the patient has at least one qualifying CR or PR response under the registered RECIST framework.
Why is ORR expressed as a percentage?
Because ORR is a proportion, it can be expressed as a fraction of the analysis population or as a percentage. The percentage provides an intuitive description of how frequently the prespecified response outcome occurred in the study population.
Why does “at least 1 visit response” matter?
The registry definition does not require every tumour assessment to demonstrate response. It defines ORR as the percentage of patients with at least 1 visit response of CR or PR. This patient-level rule prevents repeated assessments of the same patient from being counted as independent response events.
Why would a confidence interval be useful?
An observed ORR is an estimate based on a finite number of patients. A confidence interval communicates statistical precision around that estimate. A narrow interval would indicate greater precision, while a wider interval would indicate greater uncertainty. The ClinicalTrials.gov record does not provide the interval itself.
Why can’t a single-group ORR be interpreted as a randomized treatment effect?
Without a randomized comparator, the observed response proportion does not isolate the effect of AZD9291 from all other differences between the study population and a hypothetical untreated or alternative-treatment population. A comparison against an external response benchmark, if prespecified, answers a different question from a randomized treatment comparison.
Why does the analysis population matter?
The ORR denominator can materially affect the reported response rate. For example, an analysis based on all enrolled patients differs from one restricted to patients who meet prespecified response-evaluation criteria. The ClinicalTrials.gov record reports enrollment of 210.0 but do not specify the formal denominator used for the posted ORR outcome.
Why should repeated imaging not be treated as repeated independent observations?
The registry describes tumour assessments every 6 weeks. These repeated measurements come from the same patients and are therefore correlated. The registered ORR definition addresses this by reducing the longitudinal response information to a patient-level responder/non-responder classification.
8. Safety
the ClinicalTrials.gov record reports serious adverse events by arm as 42/210 for the AZD9291 80mg arm.
| Safety measure | AZD9291 80mg |
|---|---|
| Serious adverse events | 42/210 |
Why efficacy and safety use different denominators in many trials
Efficacy endpoints and safety outcomes can use different analysis populations because they answer different questions. Efficacy generally follows the prespecified endpoint-analysis population, while safety is often linked to actual treatment exposure. For AURA2, the ClinicalTrials.gov record gives 42 affected patients out of 210 at risk for the AZD9291 80mg arm, but the ClinicalTrials.gov record does not provide additional population definitions.
9. What Is Known — and What Is Not Reported in the Supplied Data
| Topic | Supplied information | Statistical implication |
|---|---|---|
| Primary endpoint | ORR | Binary patient-level response endpoint |
| Formal statistical analysis | None in the posted results | No reported estimate, CI, or p-value can be reproduced from the registry-reported analysis field |
| Results posted | Yes | Registry profile indicates posted results, but the registry-reported statistical-analysis field is empty |
| Enrollment | 210.0 | Provides the study enrollment count, not necessarily the formal ORR analysis denominator |
| Comparator | None; single-group | No randomized between-arm treatment effect is available |
| Serious adverse events | 42/210 for AZD9291 80mg | Descriptive safety count |
| Subgroups | Not reported | No subgroup effect estimates should be inferred |
| Median survival | Not reported | No survival analysis should be added |
| Hazard ratios | Not reported | No Cox or time-to-event treatment effect should be reported |
| Multiplicity | Not reported | No multiplicity adjustment should be inferred |
| Interim analysis | Not reported | No interim-monitoring procedure should be inferred |
| Missing-data method | Not reported | No imputation procedure should be inferred |
| Bayesian methods | Not reported | No Bayesian analysis should be inferred |
10. Design Topics Not Supported by the Supplied Data
Non-inferiority
No non-inferiority margin is reported. The study is identified as a single-group phase 2 study, so a non-inferiority framework should not be assumed.
Crossover
No crossover procedure is reported. A crossover analysis therefore should not be inferred.
Factorial design
The design model is SINGLE_GROUP with 1 arm. No factorial structure is reported.
Multiplicity
No multiplicity procedure is reported. No adjustment for multiple endpoints or multiple testing should be inferred.
Interim analysis
No interim-analysis plan or alpha-spending procedure is reported.
Missing data
No imputation or missing-data strategy is reported in the ClinicalTrials.gov record.
11. Limitations
- No randomized comparator: the SINGLE_GROUP design does not provide a randomized control group against which the response rate can be directly compared.
- Formal statistical analyses not reported: no formal statistical analyses were posted, so no formal ORR estimate, confidence interval, or p-value can be reported from that field.
- Incomplete endpoint text: the registry definition and time frame end mid-sentence. This page does not infer the missing wording.
- Analysis denominator: enrollment is reported as 210.0, but the ClinicalTrials.gov record does not identify the prespecified denominator for the posted ORR outcome.
- No comparative effect measure: a hazard ratio, risk ratio, odds ratio, or treatment difference cannot be derived without information that is not reported.
- No subgroup analyses reported: subgroup-specific response estimates and interaction analyses should not be inferred.
- No survival results reported: the provided primary endpoint is binary ORR, and the ClinicalTrials.gov record contains no survival estimates or time-to-event analyses.
- Safety detail is limited: the ClinicalTrials.gov record identifies serious adverse events as 42/210 for the AZD9291 80mg arm but do not provide further safety breakdowns.
- External validity: interpretation of a single-group phase 2 response rate depends on the characteristics of the enrolled population and the clinical context against which the response proportion is compared. Those broader comparative characteristics are not reported here.
12. Why This Trial Matters Statistically
AURA2 is useful as a statistical teaching case because it illustrates a different structure from a conventional randomized phase 3 comparison. Its primary endpoint is binary, its design is single-group, and its response assessment is longitudinal but ultimately summarized at the patient level.
| Concept | How it appears in AURA2 |
|---|---|
| Single-group design | One study arm with 210.0 enrolled participants |
| Binary endpoint | Objective Response Rate is the registered primary endpoint type |
| Patient-level response | ORR is defined as the percentage of patients with at least 1 visit response of CR or PR |
| RECIST v1.1 | Response classification uses registered RECIST v1.1 criteria assessed by MRI or CT |
| Repeated assessment | Tumour assessments are scheduled every 6 weeks from first dose until objective disease progression |
| Proportion estimation | The primary efficacy quantity is a response proportion rather than a between-arm effect measure |
| Confidence intervals | A confidence interval would quantify uncertainty around the estimated response proportion |
| Analysis population | The formal ORR denominator is not reported, illustrating why enrollment and analysis population should not automatically be treated as identical |
| Safety analysis | Serious adverse events are reported descriptively as 42/210 for AZD9291 80mg |
| Registry interpretation | Results are identified as posted, while no formal statistical analyses were posted |
13. A Statistical Walkthrough of ORR
Step 1: Define the patient-level outcome
The registered endpoint starts with tumour-response assessments under RECIST v1.1. A patient is considered a responder if at least one visit response meets the registered CR or PR criteria.
Step 2: Count responders
The number of patients satisfying the response rule becomes the numerator of the ORR. The ClinicalTrials.gov record does not provide that responder count.
Step 3: Identify the prespecified denominator
The analysis denominator should come from the prespecified endpoint-analysis population. the ClinicalTrials.gov record reports 210.0 enrolled participants, but it does not state that 210.0 is the formal denominator used for the posted ORR outcome.
Step 4: Calculate the response proportion
The result can then be expressed as a proportion or percentage.
Step 5: Quantify uncertainty
A confidence interval should accompany the point estimate when inferential precision is required. The exact interval method should be taken from the statistical analysis plan or registry-posted statistical information. No such method is reported here.
Step 6: Ask whether a formal benchmark exists
For a single-arm oncology study, a protocol may define a response-rate benchmark against which the observed ORR is tested. Such a benchmark would need to be prespecified to support a formal hypothesis test. The registry-reported AURA2 data do not provide one, so no such comparison is reported here.
14. Interpretation of the Serious Adverse Event Count
The ClinicalTrials.gov record is 42/210 serious adverse events for the AZD9291 80mg arm. This identifies the number affected and the reported at-risk denominator.
The count does not by itself establish a causal treatment effect, a comparative safety advantage or disadvantage, or a rate relative to another therapy. There is no randomized comparator arm in the ClinicalTrials.gov record.
The denominator provides the population against which the affected count is interpreted. Because the ClinicalTrials.gov record gives 42/210, the descriptive safety quantity is anchored to those two reported counts rather than to an inferred exposure population.
15. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
AURA2 is a single-group phase 2 study with a binary primary endpoint. The appropriate core statistical description is an estimate of the proportion of patients meeting the registered response criterion, accompanied by an appropriate measure of uncertainty.
Clinical interpretation
The registered clinical question concerns objective tumour response after previous EGFR TKI therapy in EGFR and T790M mutation-positive tumours. The ClinicalTrials.gov record does not provide enough formal statistical results to quantify the magnitude or precision of that response endpoint.
16. Statistical Questions a Reader Should Ask
| Question | Why it matters |
|---|---|
| Who was included in the ORR denominator? | The denominator determines the reported response proportion. |
| How many patients had CR or PR? | This determines the ORR numerator. |
| What confidence-interval method was prespecified? | Different methods can produce different interval estimates, especially for binary outcomes with limited event counts. |
| Was there a prespecified response-rate benchmark? | A benchmark would determine whether a formal one-sample hypothesis test was appropriate. |
| How were patients with incomplete assessments handled? | Missing or unevaluable response assessments can affect both numerator and denominator. |
| Was confirmation of response required? | The registry-reported endpoint definition specifies at least 1 visit response but does not provide the missing remainder of the registry wording. |
| Were response assessments independent? | Repeated assessments come from the same patient and should not be treated as independent observations. |
| Was the analysis prespecified? | Prespecification distinguishes a planned confirmatory or descriptive analysis from an analysis chosen after seeing results. |
17. Related Statistical Concepts
Learn more about the methods used in this trial:
18. Related Statistical Calculators
19. Important Statistical Takeaways
1. The endpoint is binary
ORR classifies patients according to whether they meet the registered CR or PR response rule.
2. The design is single-group
There is one study arm and no randomized comparator in the ClinicalTrials.gov record.
3. Enrollment is not automatically the analysis denominator
The trial enrolled 210.0 participants, but the ClinicalTrials.gov record does not identify the formal ORR analysis population.
4. A response percentage needs context
The response estimate should be considered alongside its denominator, uncertainty interval, and any prespecified benchmark.
5. Repeated imaging is not repeated independent evidence
Patients undergo repeated tumour assessments, but ORR is defined at the patient level.
6.
No formal statistical analyses were posted, so this page does not fabricate an ORR estimate, confidence interval, or p-value.
20. Sources
- ClinicalTrials.gov: AURA2 — NCT02094261.
- PubMed: PMID 40311309.
- PubMed: PMID 31345012.
- PubMed: PMID 30512189.
- PubMed: PMID 29293889.
- PubMed: PMID 27751847.
Continue through Clinical Biostats
Explore the statistical methods behind clinical-trial endpoints, study designs, and quantitative analysis workflows.
21. Record Summary
AURA2 is a completed phase 2, open-label, single-group study of AZD9291 in non-small cell lung cancer after previous EGFR TKI therapy in EGFR and T790M mutation-positive tumours. The study enrolled 210.0 participants and registered objective response rate as its primary binary endpoint, using RECIST v1.1 tumour assessments by MRI or CT every 6 weeks from first dose until objective disease progression, up to approximately 11 months (at the time of analysis) in the registry text. The ClinicalTrials.gov record identifies results as posted, but contain no formal statistical analyses. The appropriate statistical framework is therefore centered on patient-level response proportions and their uncertainty rather than randomized treatment-effect measures. The ClinicalTrials.gov record reports serious adverse events in 42/210 patients in the AZD9291 80mg arm.