This page separates reported registry information from statistical interpretation. The ClinicalTrials.gov record identifies a registered binary objective-response endpoint and confirm that results have been posted, but no formal statistical analyses are included in the posted results.
1. Trial at a Glance
CheckMate-142 is a phase 2, non-randomized, parallel-design oncology study evaluating nivolumab alone or nivolumab combination therapy in colorectal cancer. The registry lists 385.0 enrolled participants, 6 arms, and one registered primary endpoint: objective response rate by investigator assessment using RECIST v1.1.
| Feature | CheckMate-142 |
|---|---|
| Trial name | CheckMate-142 |
| Brief title | A Study of Nivolumab Alone or Nivolumab Combination Therapy in Colon Cancer That Has Come Back or Has Spread |
| Phase | Phase 2 |
| Status | COMPLETED |
| Allocation | NON_RANDOMIZED |
| Design model | PARALLEL |
| Masking | NONE |
| Primary purpose | TREATMENT |
| Enrollment | 385 |
| Arms | 6 |
| Lead sponsor | Bristol-Myers Squibb |
| Sponsor type | INDUSTRY |
| Start | 2014-03-12 |
| Primary completion | 2024-10-22 |
| ClinicalTrials.gov identifier | NCT02060188 |
2. Clinical Question
The registry describes a study of patients with colorectal cancer that includes microsatellite unstable colorectal cancer, microsatellite stable colorectal cancer, mismatch repair proficient colorectal cancer, and mismatch repair deficient colorectal cancer. The interventions listed are nivolumab, ipilimumab, cobimetinib, daratumumab, and BMS-986016.
Population
The registry conditions are Microsatellite Unstable Colorectal Cancer; Microsatellite Stable Colorectal Cancer; Mismatch Repair Proficient Colorectal Cancer; and Mismatch Repair Deficient Colorectal Cancer.
Interventions
The registry lists Ipilimumab, Nivolumab, Cobimetinib, Daratumumab, and BMS-986016 as drug interventions.
Comparator
The ClinicalTrials.gov record does not identify a randomized comparator. The allocation is explicitly recorded as NON_RANDOMIZED.
Primary question
The registered primary endpoint asks whether participants achieve a confirmed complete or partial response according to investigator-assessed RECIST v1.1 criteria.
3. Trial Design
Multi-cohort treatment study
- Six arms are listed in the registry.
- The overall design is recorded as parallel.
- Allocation is recorded as non-randomized.
- Masking is recorded as none.
Registered drug interventions
- Nivolumab
- Ipilimumab
- Cobimetinib
- Daratumumab
- BMS-986016
The registry extract does not provide enough arm-level information to reconstruct a complete six-arm treatment table. In particular, the available data do not establish the exact intervention composition, dose, schedule, or enrollment for every arm. Those details therefore are not inferred here.
4. Trial Population and Disease Definitions
The registry lists four colorectal-cancer condition categories. These definitions are important statistically because the biological classification of participants can affect how response rates are interpreted, particularly in a study that is not randomized.
| Registered condition | Role in the trial record |
|---|---|
| Microsatellite Unstable Colorectal Cancer | One of the conditions listed in the registry. |
| Microsatellite Stable Colorectal Cancer | One of the conditions listed in the registry. |
| Mismatch Repair Proficient Colorectal Cancer | One of the conditions listed in the registry. |
| Mismatch Repair Deficient Colorectal Cancer | One of the conditions listed in the registry. |
No baseline demographic table, baseline disease-characteristic table, or subgroup-specific response estimates are included in the ClinicalTrials.gov record. Those sections are therefore not reconstructed on this page.
5. Primary Endpoint
| Endpoint | Registered definition / time frame | Type |
|---|---|---|
| Objective Response Rate (ORR) by Response Evaluation Criteria in Solid Tumors (RECIST) v1.1 Per Investigator Assessment | Time Frame: From date of randomization to the date of objectively documented progression or the date of subsequent systemic cancer therapy, whichever occurs first (Up to approximately 127 months) | Binary |
The registry defines objective response rate as the number of randomized participants who achieve a best overall response (BOR) of confirmed complete response (CR) or confirmed partial response (PR), based on investigator assessments using RECIST v1.1, divided by the number of all randomized participants.
The registry-reported definition continues beyond the available extract with the definition of complete response and related response criteria, but the ClinicalTrials.gov record terminate partway through that text. The endpoint is therefore presented using the registry wording available in the ClinicalTrials.gov record rather than completing the definition from an external source.
Conceptually, ORR is a proportion. Unlike a hazard ratio, it does not describe an instantaneous event rate over time. It classifies each participant according to whether the prespecified response criterion was achieved.
6. Results Availability and Planned Analysis
What the registry reports
The available results information establishes that outcome measures have been posted and that the registered primary endpoint is a binary objective response rate based on investigator assessment using RECIST v1.1.
What a formal analysis would normally involve
For a binary response endpoint, the statistical analysis would ordinarily begin with the number and proportion of participants achieving the prespecified response definition. Depending on the design and prespecified estimand, an analysis could also provide a confidence interval for the response proportion and, where a formal comparator or benchmark is defined, a hypothesis test against that prespecified reference.
For a non-randomized multi-arm study, the statistical interpretation is different from that of a randomized comparative trial. Response proportions can describe activity within cohorts, but differences between cohorts may reflect differences in patient characteristics, disease classification, eligibility, treatment selection, follow-up, or other factors rather than treatment effects alone.
7. How to Analyze Objective Response Rate
Objective response rate is naturally expressed as a binomial proportion. If x participants meet the confirmed CR or PR definition among n participants in the analysis set, the estimated response rate is:
The numerator is the number of participants meeting the prespecified response definition, while the denominator is determined by the analysis population specified for the endpoint.
Confidence intervals
A confidence interval is important because a response proportion calculated from a finite number of participants is an estimate rather than a known population parameter. For a binomial endpoint, an exact or score-based interval may be preferable to a simple normal approximation, particularly when the sample size is modest or the observed response proportion is close to 0 or 1.
Why the analysis population matters
The registered endpoint definition uses all randomized participants as its denominator. Yet the overall trial allocation is recorded as non-randomized. That creates an important interpretive issue: the precise analysis population used for the posted result should be verified against the underlying registry results and statistical analysis plan rather than inferred from the trial's general enrollment number.
8. Statistical Methodology
Binary outcome analysis
The primary endpoint is explicitly classified as Binary. A binary endpoint records whether each participant satisfies a prespecified condition. Here, the condition is achieving a confirmed complete response or confirmed partial response according to investigator assessment using RECIST v1.1.
In a simple binomial formulation, X represents the number of responders among n analyzed participants and p represents the underlying response probability.
Confidence intervals for a proportion
The uncertainty around an ORR estimate should be communicated with a confidence interval rather than relying only on the point estimate. The interval quantifies sampling uncertainty under the chosen statistical method; it does not describe the range of response probabilities for individual participants.
Comparative testing
A comparative hypothesis test requires a clearly specified comparison. In a randomized two-arm trial, that comparison could be a difference in response probabilities, a response-rate ratio, or another prespecified effect measure. In a non-randomized multi-arm study, the meaning of a between-cohort comparison depends critically on how treatment assignment and the populations being compared were defined.
Investigator assessment
The registered endpoint specifies investigator assessment. This matters because response classification is an assessment process rather than a purely administrative outcome. Any statistical interpretation of ORR must therefore retain the endpoint's prespecified assessment framework rather than treating response as an unspecified binary event.
RECIST v1.1
The endpoint explicitly references Response Evaluation Criteria in Solid Tumors (RECIST) v1.1. Thus, the binary response variable is constructed from a standardized radiographic response framework rather than from an unspecified clinical judgment.
9. Statistical Methods Explained
Why is ORR a binary endpoint?
Each participant is ultimately classified according to whether the prespecified response criterion is met. Confirmed CR or confirmed PR qualifies as response under the registered definition; participants who do not satisfy that definition do not contribute to the responder numerator. This makes the primary endpoint naturally suitable for binomial-proportion methods.
What does ORR measure?
ORR measures the proportion of participants achieving the specified objective tumor response. It is a measure of response frequency, not a direct measure of overall survival, progression-free survival, duration of response, or quality of life.
Why is the confidence interval important?
Two studies can have the same observed response proportion but different statistical precision if their analysis populations differ in size. A confidence interval makes that uncertainty visible. A narrow interval indicates greater precision under the specified statistical framework; a wider interval indicates greater uncertainty.
Why does a p-value not measure the size of an ORR?
A p-value addresses compatibility with a specified null hypothesis under the chosen statistical model. It does not tell the reader how large the response rate is, how clinically important a response is, or what proportion of individual patients will respond. Those questions require the response estimate and its uncertainty, together with clinical context.
Why is non-randomization important?
Randomization helps balance measured and unmeasured prognostic factors between treatment groups in expectation. Because the registry records CheckMate-142 as non-randomized, a difference in response between treatment cohorts cannot automatically be interpreted as a causal treatment effect. Patient selection and cohort composition can contribute to observed differences.
Why should the registry's denominator be checked carefully?
The registered endpoint definition states that the response rate is calculated using randomized participants as the denominator, while the overall allocation is recorded as non-randomized. That apparent inconsistency makes verification of the actual analysis population especially important before interpreting a posted ORR.
10. Safety Results
| Safety information reported in the registry extract | Affected / at risk |
|---|---|
| Serious adverse events by arm | Cohort 1: Nivolumab Monotherapy: 46/74; Cohort 2: Nivolumab 3 mg/kg + Ipilimumab: 72/119; Cohort 3: Nivolumab 3 mg/kg Q2W with Ipi: 22/45; MSS Nivolumab 3 mg/kg + Ipilimumab 1 mg/: 8/10; Non-MSI-H: Nivolumab 1 mg/kg + Ipilimuma: 3/3; Non-MSI-H: Nivolumab 1 mg/kg + Ipilimuma: 9/10; Cohort 4: Nivoluma |
The serious-adverse-event data also illustrate why denominators matter. A count such as “46/74” communicates both the number of affected participants and the number at risk. A count without its denominator would not provide enough information to compare event frequency across cohorts.
11. Interpretation of a Non-Randomized Multi-Arm Study
Within-cohort description
An ORR can describe the observed frequency of response within a defined cohort, provided the analysis population and response rules are clearly specified.
Between-cohort comparison
Observed differences between non-randomized cohorts can reflect treatment as well as differences in the populations receiving each intervention.
Confounding
Without random allocation, measured and unmeasured prognostic factors may differ between treatment groups and complicate causal interpretation.
Endpoint scope
ORR captures tumor response according to the registered RECIST v1.1 framework. It does not by itself establish durability of response or a time-to-event benefit.
This distinction is central to the statistical reading of CheckMate-142. A response rate is a useful descriptive endpoint, but the inferential question changes depending on whether the analysis is estimating activity within a cohort, comparing cohorts, or testing against a prespecified benchmark.
12. Registered Design Features and What They Imply Statistically
| Design feature | Registry information | Statistical implication |
|---|---|---|
| Allocation | NON_RANDOMIZED | Treatment-group differences cannot automatically be attributed to treatment because assignment is not randomized. |
| Design model | PARALLEL | The study is organized into parallel treatment cohorts rather than a crossover design. |
| Masking | NONE | The registry records an unmasked study; the primary endpoint is investigator assessed. |
| Primary purpose | TREATMENT | The registry classifies the principal purpose as treatment. |
| Arms | 6 | The study contains multiple treatment cohorts, making arm-specific denominators and analysis populations important. |
| Primary endpoint | Binary ORR | The principal registered outcome is naturally analyzed as a response proportion. |
13. Multiplicity and Multiple Arms
The trial has 6 arms, but the ClinicalTrials.gov record does not provide a multiplicity strategy. Consequently, this page does not assume a particular familywise-error procedure, hierarchical testing sequence, adjustment method, or number of confirmatory pairwise comparisons.
This is an important distinction. Multiple treatment cohorts can generate many possible comparisons, but the existence of several arms does not by itself establish that every possible comparison was intended to be confirmatory. The appropriate interpretation depends on the prespecified statistical analysis plan.
If several hypotheses are tested within one confirmatory family, the probability of at least one false-positive conclusion can exceed the nominal level unless the testing strategy accounts for multiplicity.
14. Interim Analysis, Bayesian Methods, and Missing Data
The ClinicalTrials.gov record does not report an interim-analysis strategy, alpha-spending method, Bayesian model, missing-data imputation method, or formal multiplicity procedure.
| Methodological topic | What can be established from the ClinicalTrials.gov record |
|---|---|
| Interim analysis | No interim-analysis method is reported. |
| Alpha spending | No alpha-spending method is reported. |
| Bayesian methods | No Bayesian statistical method is reported. |
| Missing-data / imputation | No missing-data or imputation method is reported. |
| Multiplicity | No formal multiplicity strategy is reported. |
| Stratification | No statistical stratification factors are reported. |
These omissions do not mean that such methods were necessarily absent from the complete protocol or statistical analysis plan. They mean only that the ClinicalTrials.gov record does not document them, so they should not be presented as established features of the analysis.
15. Analysis Populations
The registered primary endpoint definition specifies “randomized participants” as its denominator, but the overall trial allocation is recorded as NON_RANDOMIZED. The ClinicalTrials.gov record does not provide a complete definition of the actual efficacy analysis population.
| Population concept | Status in the ClinicalTrials.gov record |
|---|---|
| Randomized population | Referenced in the registered endpoint definition, although overall allocation is recorded as non-randomized. |
| Efficacy analysis population | Not separately defined in the ClinicalTrials.gov record. |
| Safety population | Arm-level serious-adverse-event affected/at-risk information is reported, but a complete population definition is not. |
| Per-protocol population | Not reported. |
For a binary response endpoint, changing the denominator can materially change the estimated ORR. Therefore, a response percentage should always be interpreted alongside the population from which the numerator and denominator were derived.
16. Clinical Biostats Interpretation of the Primary Endpoint
An ORR estimate would represent the proportion of participants in the specified analysis population who achieved a confirmed complete response or confirmed partial response according to investigator assessment using RECIST v1.1.
ORR would not mean that the same proportion of participants were cured, remained progression-free for a specified period, or experienced a particular survival benefit. Response is one clinical endpoint with a specific definition.
A confidence interval around ORR would describe statistical uncertainty around the estimated response proportion under the selected binomial framework. It would not describe the probability that an individual participant will respond.
If a formal hypothesis test were specified, its p-value would quantify evidence against a particular null hypothesis under the chosen model. It would not measure the magnitude of ORR or establish clinical importance by itself.
Because allocation is recorded as non-randomized, a difference in observed response between cohorts would require caution. Differences in participant characteristics or treatment selection could contribute to the observed result.
17. Limitations
- Non-randomized allocation: treatment assignment is recorded as NON_RANDOMIZED, limiting the causal interpretation of differences between treatment cohorts.
- Six-arm structure: multiple cohorts make it important to identify the exact analysis population and prespecified comparison strategy for each reported estimate.
- Registry-definition inconsistency: the registry-reported endpoint definition refers to randomized participants while the ClinicalTrials.gov record records allocation as non-randomized.
- No formal statistical analyses reported: no formal statistical analyses were posted, so no formal treatment-effect estimate, confidence interval, or p-value can be attributed to the registry extract used here.
- Investigator assessment: the primary endpoint relies on investigator assessment under RECIST v1.1, so the endpoint reflects that prespecified assessment process.
- Unreported methodological details: the ClinicalTrials.gov record does not establish the interim-analysis, missing-data, Bayesian, multiplicity, or stratification methods.
- Limited endpoint scope: ORR is a binary response measure and does not by itself characterize duration of response or time-to-event outcomes.
- Generalizability: the four registered colorectal-cancer condition categories define the disease populations represented in the registry record; the ClinicalTrials.gov record does not provide a complete baseline-characteristic table for assessing external applicability.
18. Why This Trial Matters Statistically
CheckMate-142 is a useful statistical teaching case because it demonstrates how the interpretation of a clinical-trial result depends not only on the endpoint but also on allocation, cohort structure, analysis populations, and the prespecified inferential framework.
| Concept | How it appears in CheckMate-142 |
|---|---|
| Binary endpoint | Objective response rate is the registered primary endpoint type. |
| RECIST-based response | ORR is based on investigator assessment using RECIST v1.1. |
| Non-randomized design | Allocation is explicitly recorded as NON_RANDOMIZED. |
| Multi-arm study | The registry lists 6 arms. |
| Denominator selection | The endpoint definition specifies randomized participants, creating an important population-definition issue because the study is recorded as non-randomized. |
| Confidence intervals | A proportion estimate should be interpreted together with its statistical uncertainty when a formal analysis is available. |
| Multiplicity | Multiple arms can create multiple potential comparisons, but the ClinicalTrials.gov record does not specify an error-control strategy. |
| Safety denominators | Serious adverse events are reported as affected/at-risk counts for several cohorts. |
| Results interpretation | Results are posted, but the registry-reported statistical-analysis array contains no formal analyses. |
19. Trial Timeline
Study start
The registry lists 2014-03-12 as the study start date.
Clinical development phase
The trial is recorded as phase 2 with treatment as its primary purpose.
Enrollment
The registry lists enrollment of 385.0 participants across 6 arms.
Primary completion
The registry lists 2024-10-22 as the primary completion date.
Current registry status in the ClinicalTrials.gov record
The trial is recorded as COMPLETED, with results posted in the ClinicalTrials.gov record.
20. Four Questions a Statistician Should Ask Before Reading the ORR
1. What is the denominator?
Confirm exactly which participants are included in the analysis population and reconcile that definition with the registered wording referring to randomized participants.
2. What is the response definition?
Confirm that the response classification follows the registered investigator-assessed RECIST v1.1 framework and the required confirmation rules.
3. Is there a comparator?
Determine whether the reported analysis is descriptive within a cohort or makes a prespecified comparison. the ClinicalTrials.gov record records non-randomized allocation.
4. What uncertainty is reported?
Look for a confidence interval and the method used to construct it. A response percentage without uncertainty provides an incomplete statistical picture.
21. Statistical Concepts in This Trial
Learn more about the methods used in this trial:
22. Related Statistical Calculators
Explore calculators that support the statistical concepts behind this trial:
23. Sources
- ClinicalTrials.gov: CheckMate-142, NCT02060188.
- PubMed: Publication indexed under PMID 38821718.
- PubMed: Publication indexed under PMID 34637336.
- PubMed: Publication indexed under PMID 29355075.
- PubMed: Publication indexed under PMID 28734759.
Continue through Clinical Biostats
Connect this trial's response endpoint, clinical-trial design, and safety concepts with broader statistical tutorials and practical calculators.
24. Record Summary
CheckMate-142 is a phase 2, non-randomized, parallel-design oncology study with 385.0 enrolled participants, 6 arms, and a registered binary primary endpoint measuring objective response rate by investigator assessment using RECIST v1.1. The registry reports results as posted, but the ClinicalTrials.gov record contains no formal analysis records. The statistically appropriate interpretation is therefore to distinguish the registered endpoint and design from any efficacy conclusion that would require a verified result, analysis population, confidence interval, and prespecified comparison framework.