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
EUCROSS was a phase 2, single-group treatment study evaluating crizotinib in lung cancer, adenocarcinoma, and NSCLC. The registered primary endpoint was objective response rate (ORR), assessed by investigators according to RECIST v1.1.
| Feature | EUCROSS |
|---|---|
| Trial name | EUCROSS: European Trial on Crizotinib in ROS1 Translocated Lung Cancer |
| Phase | Phase 2 |
| Status | COMPLETED |
| Population | Lung Cancer; Adenocarcinoma; NSCLC |
| Design | Single-group treatment study |
| Masking | None |
| Allocation | NA |
| Enrollment | 34 |
| Arms | 1 |
| Intervention | Crizotinib |
| Primary endpoint | Objective Response Rate (ORR); investigator assessed RECIST v1.1 analysis |
| Lead sponsor | University of Cologne |
| Start | 2014-05 |
| Primary completion | 2018-07-24 |
| ClinicalTrials.gov | NCT02183870 |
2. Clinical Question
The central question was whether treatment with crizotinib could produce objective tumor responses in patients with advanced adenocarcinoma of the lung harbouring ROS1 fusion genes, with response assessed according to RECIST v1.1.
Population
The registry identifies the conditions as lung cancer, adenocarcinoma, and NSCLC. The primary endpoint definition specifies advanced adenocarcinoma of the lung harbouring ROS1 fusion genes.
Intervention
Crizotinib was the registered drug intervention.
Comparator
There was no separate comparator arm. The design model was SINGLE_GROUP.
Primary question
What objective response rate is observed when tumor response is evaluated by investigators using RECIST v1.1?
This distinction is fundamental statistically. A single-group response study addresses a different question from a randomized comparative trial. Without a concurrently randomized control group, the response proportion can describe the observed activity of treatment within the enrolled population, but it does not by itself provide a randomized estimate of treatment effect relative to another therapy.
3. Trial Design
Crizotinib
- Drug intervention: crizotinib
- Single-group study design
- No separate comparator arm
- Objective response assessed by investigators
- RECIST v1.1 evaluation framework
Single-Group Response Study
- 34.0 enrolled patients
- Binary primary endpoint
- Response evaluated during treatment
- CT/MRI scans used for efficacy evaluation
- No randomized treatment comparison
4. Endpoints
| Endpoint | Registry definition / assessment | Time frame |
|---|---|---|
| Objective Response Rate (ORR) | Objective Response Rate (ORR); evaluation criteria: investigator assessed RECIST v.1.1 analysis | From time of beginning of treatment until the documention of response according to RECIST v1.1 (expected average 12 months). |
The registry describes ORR as the primary outcome measure and specifies that CT/MRI scans will be performed to evaluate the efficacy of crizotinib treatment in advanced adenocarcinoma of the lung harbouring ROS1 fusion genes. The endpoint is classified as binary in the registry data.
What ORR measures
Objective response rate is fundamentally a proportion. Each evaluable patient is classified according to whether the prespecified response criterion is met under the stated assessment framework, and the response rate is the proportion classified as having an objective response.
For a binary endpoint, the statistical analysis therefore centers on estimating a proportion and quantifying the uncertainty around that proportion.
The registry does not provide a posted statistical estimate, confidence interval, p-value, or response count for the primary endpoint. Consequently, the ClinicalTrials.gov record describes how response was to be assessed but does not provide a numerical efficacy result.
5. Planned Analysis
The registered primary endpoint is a binary objective response rate assessed according to investigator-assessed RECIST v1.1 criteria. For an endpoint of this type, the primary descriptive analysis would ordinarily estimate the proportion of patients meeting the objective response criterion among the prespecified analysis population.
Primary estimand concept
The central quantity is the probability of objective response under the study's defined assessment process and population.
For a single-group binary endpoint, a confidence interval around the observed response proportion is an important component of interpretation. The point estimate describes the observed proportion, while the confidence interval communicates statistical uncertainty associated with estimating an underlying response probability from a finite sample.
The choice of confidence-interval method matters more in a study of this scale than the apparent simplicity of the endpoint might suggest. Binary proportions are bounded between 0 and 1, so methods that respect that boundary are generally preferable to an unrestricted normal approximation when sample sizes are limited or the response proportion is near either boundary.
Point estimate
The observed response proportion summarizes how many patients met the objective response criterion within the defined analysis population.
Confidence interval
A confidence interval quantifies uncertainty around the estimated response probability. It should be interpreted as statistical uncertainty, not as a range containing individual patient responses.
Binary classification
RECIST-based response converts longitudinal tumor-assessment information into a prespecified response classification for the endpoint.
Single-group interpretation
Because the design has one treatment group, the primary result would describe response under crizotinib rather than a randomized between-treatment effect.
Why a response endpoint is different from survival analysis
ORR is not a time-to-event endpoint in the same sense as overall survival or progression-free survival. The primary statistical object is a proportion rather than a hazard ratio or survival function. Although the registry specifies a time frame during which response is assessed, the endpoint itself is classified as binary. This makes the analysis conceptually closer to a binomial proportion than to a Kaplan-Meier analysis.
6. Statistical Methodology
Binary endpoint analysis
The primary endpoint is classified as binary, meaning that the response assessment ultimately produces an outcome such as response versus no response for the purpose of calculating ORR. The resulting data can be represented by a binomial model for the number of responders among the analysis population.
Here, X represents the number of patients classified as responders, n is the relevant analysis population, and p represents the underlying response probability.
The observed response rate is then the sample proportion X/n. The registry does not post the value of X or the resulting estimate, so the numerical response probability cannot be described from the ClinicalTrials.gov record.
RECIST v1.1 assessment
The registered endpoint specifies investigator-assessed RECIST v1.1 analysis. The registry also states that CT/MRI scans will be performed to evaluate efficacy. Thus, the statistical endpoint is built on a clinical and radiologic assessment process that precedes the binary response classification.
This distinction is useful when interpreting ORR. The statistical analysis of a proportion is relatively simple mathematically, but the validity of that proportion depends on the consistency of the underlying response assessments. A response rate is therefore not merely a count divided by a denominator; it is the result of applying predefined response criteria to longitudinal tumor assessments.
Confidence intervals for a proportion
A response rate without an uncertainty measure can give an incomplete picture of the evidence. The same observed proportion can imply different degrees of precision depending on the amount of information available. Confidence intervals provide a way to display that uncertainty alongside the point estimate.
For a binary endpoint, the response proportion and its confidence interval should normally be presented together.
Single-group analysis
The absence of a randomized comparator changes the interpretation of the statistical result. In a randomized two-arm study, a treatment effect can be defined through a contrast between groups. In a single-group study, the primary observed quantity is instead the response probability within the treated population.
A single-group response rate can still be scientifically informative, particularly when the treatment and population are defined around a specific biological or clinical characteristic. But statistical interpretation must remain aligned with the design: the response proportion is not itself a randomized causal contrast.
7. Statistical Methods Explained
What makes ORR a binary endpoint?
The registry classifies the primary endpoint as binary. Although tumor measurements are obtained through CT/MRI scans and evaluated according to RECIST v1.1, the primary endpoint ultimately records whether the prespecified objective-response criterion has been met. That produces a responder/non-responder classification suitable for proportion-based analysis.
Why is the response rate a proportion?
ORR summarizes the fraction of patients who meet the response criterion. If X patients respond among n patients in the relevant analysis population, the observed ORR is X/n. The statistical challenge is therefore to estimate the underlying response probability while accounting for sampling uncertainty.
Why does RECIST matter statistically?
RECIST provides the rules used to translate radiologic tumor assessments into standardized response classifications. Without a defined assessment framework, different investigators could classify similar tumor changes differently. The registry specifically identifies investigator-assessed RECIST v1.1 analysis as the evaluation criterion for the primary endpoint.
Why is a confidence interval important for ORR?
The response proportion is calculated from a finite number of patients and therefore varies from sample to sample. A confidence interval provides information about the precision of the estimate. A wide interval indicates that the underlying response probability is less precisely estimated than a narrow interval would suggest.
Why does the single-group design matter?
There is no randomized comparator against which the response rate can be contrasted. The statistical result therefore describes response within the crizotinib-treated study population rather than estimating a randomized treatment effect. Comparisons with external populations introduce additional assumptions that are not reported by the registry record.
Why is the endpoint's time frame important?
The registry defines the time frame from the beginning of treatment until documentation of response according to RECIST v1.1, with an expected average of 12 months. This establishes the period within which the response endpoint is evaluated. Time-frame definitions are important because changing the observation window can change which responses are captured.
8. Interpreting a Future ORR Result
If a numerical ORR were reported for this study, the first quantity to examine would be the observed response proportion. The second would be its confidence interval. These two pieces of information answer related but different questions: the point estimate describes the observed response frequency, while the interval describes uncertainty about the underlying response probability.
An ORR estimate represents the proportion of patients classified as having an objective response according to the registered investigator-assessed RECIST v1.1 framework. It does not mean that the same proportion of patients experienced identical tumor changes, nor does it establish how the response rate compares causally with another treatment.
A confidence interval would show the statistical precision of the observed response estimate. It would not describe the range of responses experienced by individual patients, and it would not automatically account for differences between this study population and an external population.
Because EUCROSS is recorded as a single-group study with one arm, the primary response result is inherently a within-study estimate. The absence of randomization means that an observed response rate should not be interpreted as a randomized estimate of treatment effect relative to an alternative therapy.
9. Important Limitations and Interpretation Issues
- No randomized comparator: the registry records a SINGLE_GROUP design with 1 arm. The primary endpoint therefore does not provide a randomized between-treatment comparison.
- Limited sample size: enrollment was 34.0 patients. Binary endpoints estimated from a finite population have sampling uncertainty, and the amount of information available for estimating a response probability is directly related to the analysis population.
- Investigator assessment: the primary endpoint is explicitly based on investigator-assessed RECIST v1.1 analysis. Interpretation therefore depends on consistent application of the specified response criteria.
- Response is not survival: ORR measures objective tumor response rather than time to death or another time-to-event outcome. It should not be interpreted as a survival measure.
- Observation window: the registered time frame extends from beginning of treatment until documentation of response according to RECIST v1.1, with an expected average of 12 months. Response interpretation therefore depends on the defined assessment period.
- Generalizability: the registry identifies a specific oncology population involving lung cancer, adenocarcinoma, and NSCLC, with the endpoint definition referring to advanced adenocarcinoma of the lung harbouring ROS1 fusion genes. Results from a narrowly defined population may not directly represent other patient populations.
- No posted numerical analysis: the ClinicalTrials.gov record does not report a statistical estimate, confidence interval, p-value, or response count for the primary endpoint.
10. Why This Trial Matters Statistically
EUCROSS provides a useful teaching example because the statistical structure is much simpler than a large randomized survival trial, yet the interpretation of the primary endpoint still requires several important distinctions. The study combines a single-group design, a binary efficacy endpoint, radiologic assessment, and a defined response-assessment window.
| Concept | How it appears in EUCROSS |
|---|---|
| Single-group design | The registry records a SINGLE_GROUP design model and 1 arm. |
| Binary endpoint | The primary endpoint is classified as Binary and is objective response rate. |
| Proportion estimation | ORR is naturally summarized as the proportion of patients meeting the response criterion. |
| RECIST assessment | The registered endpoint specifies investigator-assessed RECIST v1.1 analysis. |
| Radiologic assessment | CT/MRI scans are specified for evaluation of efficacy. |
| Confidence intervals | An estimated response proportion is most informative when accompanied by an uncertainty interval. |
| Analysis population | The denominator used for ORR is a critical component of interpreting a binary endpoint. |
| Comparator interpretation | Because there is no separate comparator arm, the response rate is not a randomized treatment-effect estimate. |
The most important statistical lesson is that a simple endpoint does not eliminate the need for careful interpretation. A response proportion is easy to calculate, but understanding what it means requires attention to the study population, response definition, assessment process, observation window, uncertainty, and absence of a randomized comparator.
11. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
The registered primary endpoint is a binary objective response rate assessed using investigator-assessed RECIST v1.1 criteria. The appropriate statistical object is therefore a response proportion, accompanied by an appropriate measure of uncertainty.
Clinical interpretation
The clinical meaning of an eventual response estimate depends on the definition of objective response, the population in which it was measured, the assessment time frame, and the absence of a randomized comparator. The registry does not post a numerical response outcome.
12. What the Registry Reports — and What It Does Not
| Item | Registry status |
|---|---|
| Study phase | Phase 2 |
| Study status | COMPLETED |
| Enrollment | 34 |
| Intervention | Crizotinib |
| Primary endpoint | Objective Response Rate (ORR) |
| Assessment criterion | Investigator assessed RECIST v1.1 analysis |
| Endpoint type | Binary |
| Response assessment | CT/MRI scans |
| Primary endpoint results | Not posted in the ClinicalTrials.gov record |
| Posted confidence interval | Not reported in the ClinicalTrials.gov record |
| Posted p-value | Not reported in the ClinicalTrials.gov record |
| Posted response count | Not reported in the ClinicalTrials.gov record |
The distinction between study completion and posted statistical results is important. A registry can identify a study as completed while still not displaying a numerical analysis of the registered endpoint. For statistical interpretation, the presence of a defined endpoint and completed study status should therefore not be confused with the availability of an analyzed efficacy result.
13. Related Tutorials
Learn more about the methods used in this trial:
The registry does not identify additional topics for related tutorial or calculator cards.
14. Related Calculators
No related calculator topics are identified for this trial in the registry-based record.
15. Sources
- ClinicalTrials.gov: EUCROSS, NCT02183870.
Continue through Clinical Biostats
Explore additional clinical trial results, biostatistics tutorials, and statistical analysis tools.
16. Record Summary
EUCROSS is a completed phase 2, single-group study evaluating crizotinib in an oncology population involving lung cancer, adenocarcinoma, and NSCLC. The registered primary endpoint is objective response rate, assessed by investigators using RECIST v1.1, with CT/MRI scans specified for efficacy evaluation and a time frame extending from the beginning of treatment until documentation of response, with an expected average of 12 months.
Statistically, the study is best understood as a single-group binary-endpoint study. The principal quantity is an objective response proportion, and its interpretation depends on the analysis population, RECIST assessment process, observation window, and statistical uncertainty around the proportion. Because the registry records no separate comparator arm, the endpoint does not constitute a randomized treatment-effect comparison.