This page separates reported trial information from statistical interpretation. The numerical trial information on this page is restricted to the ClinicalTrials.gov record data. The registry reports results for the registered primary endpoint, but the ClinicalTrials.gov record contains no formal statistical analyses with an estimate, confidence interval, or p-value.
1. Trial at a Glance
KarMMa was a completed phase 2, single-group study evaluating bb2121 in subjects with relapsed and refractory multiple myeloma. The ClinicalTrials.gov record reports 149 enrolled participants, one intervention arm, and a binary primary endpoint of overall response rate assessed from first dose to 24 months.
| Feature | KarMMa |
|---|---|
| Trial name | KarMMa |
| NCT ID | NCT03361748 |
| Brief title | Efficacy and Safety Study of bb2121 in Subjects With Relapsed and Refractory Multiple Myeloma |
| Therapeutic area | Hematology |
| Condition | Multiple Myeloma |
| Phase | Phase 2 |
| Status | Completed |
| Enrollment | 149 |
| Allocation | NA |
| Design model | Single-group |
| Masking | None |
| Primary purpose | Treatment |
| Intervention | bb2121 (biological) |
| Lead sponsor | Celgene |
| Sponsor type | Industry |
2. Clinical Question
The registered study evaluates the efficacy and safety of bb2121 in subjects with relapsed and refractory multiple myeloma. Because the registry-reported design is single-group with one intervention arm, the primary statistical question is not a randomized comparison between treatments. Instead, the principal efficacy quantity is the proportion of participants who achieve the prespecified response definition during the registered observation period.
Population
Subjects with relapsed and refractory multiple myeloma.
Intervention
bb2121, classified in the registry as a biological intervention.
Comparator
No comparator arm is listed. The design model is single-group.
Primary question
What proportion of participants achieve partial response or better according to the registered response criteria during the specified period?
3. Trial Design
The study started on December 13, 2017, and its listed primary completion date was December 20, 2023. The registry status in the ClinicalTrials.gov record is COMPLETED.
4. Endpoints
| Endpoint | Registry definition | Time frame | Type |
|---|---|---|---|
| Overall Response Rate | Number of participants who achieved partial response (PR) or better according to IMWG Uniform Response Criteria for Multiple Myeloma as assessed by an independent response committee (IRC). | From first dose to 24 Months | Binary |
The registry identifies Overall Response Rate as the sole registered primary endpoint. The endpoint is binary at the participant level: a participant either meets the response definition of partial response or better during the specified time frame or does not.
5. Primary Endpoint: Overall Response Rate
Registered primary endpoint
Binary endpoint · From first dose to 24 Months
Partial response (PR) or better according to IMWG Uniform Response Criteria for Multiple Myeloma, assessed by an independent response committee.
The ClinicalTrials.gov record indicates that results were posted for the primary endpoint. However, no formal statistical analyses were posted. In particular, no treatment-effect estimate, confidence interval, or p-value is provided in the ClinicalTrials.gov record.
6. Planned Analysis
Because the ClinicalTrials.gov record does not provide a formal statistical analysis record for the primary endpoint, the appropriate approach is to distinguish the reported measurement from the statistical method that would ordinarily be used to summarize it.
Primary analysis framework
Overall response rate is a binary proportion. For a single-group study, the fundamental estimate is
The numerator consists of participants meeting the registered response definition, while the denominator depends on the prespecified analysis population and handling of participants without an evaluable response. Those details are not provided in the ClinicalTrials.gov record.
A confidence interval for a single proportion would ordinarily be used to quantify statistical precision. Depending on the prespecified statistical analysis plan, an exact binomial interval or another binomial confidence interval could be used. The ClinicalTrials.gov record does not identify which confidence-interval construction was prespecified.
Why the single-group design matters
In a randomized two-arm trial, a response rate can be compared directly with the response rate in a control group. In a single-group study, the response proportion describes the observed performance of the treated cohort. A formal one-sample hypothesis test would require a prespecified reference response rate or other benchmark. No such benchmark is included in the ClinicalTrials.gov record.
What can be estimated
The proportion of enrolled or otherwise analysis-eligible participants meeting the registered response definition can be estimated, together with appropriate statistical uncertainty.
What is not reported
No formal treatment-effect estimate, confidence interval, p-value, or prespecified historical-control benchmark appears in the ClinicalTrials.gov record.
7. Why a Formal Comparison Is Not Reported Here
The absence of a comparator is not a minor formatting detail. It determines what can be learned from the statistical design. With one treatment arm, the observed response proportion answers a descriptive question about the study cohort. It does not by itself establish how the same participants would have responded under another treatment.
A one-sample comparison can sometimes be appropriate when a study prospectively defines a clinically meaningful benchmark. For example, the statistical hypothesis could take the form
where p is the true response probability and p0 is a prespecified reference value. The registry-reported KarMMa data do not provide such a reference value.
Consequently, inserting a benchmark from another source would violate the data restriction for this page. The statistically defensible approach is to report the registered endpoint and explain how it would normally be analyzed without creating an unreported comparison.
8. Analysis Population and Denominator Considerations
the ClinicalTrials.gov record reports 149 enrolled participants. The ClinicalTrials.gov record also report 108 serious adverse-event cases among 137 participants at risk in the bb2121 arm. These are different quantities serving different analytical purposes.
| Quantity | Reported value | Statistical role |
|---|---|---|
| Enrollment | 149 | Overall study enrollment reported by the ClinicalTrials.gov record. |
| Serious adverse events | 108/137 | Safety count and denominator reported for the bb2121 arm. |
| Primary endpoint denominator | Not specified in the statistical analyses posted on ClinicalTrials.gov | The appropriate response-analysis denominator cannot be reconstructed from the ClinicalTrials.gov record alone. |
This distinction is important. It would be incorrect to assume that the denominator for the primary overall response rate must equal 149 solely because 149 participants were enrolled. The response analysis may use a prespecified evaluable or efficacy population, but the registry-reported statistical-analysis field does not define that population.
9. Statistical Methods Explained
What makes overall response rate a binary endpoint?
Each participant contributes a response classification rather than a continuously measured outcome. The registry defines the event as achieving partial response or better according to the specified response criteria and independent response committee assessment. The resulting participant-level variable is therefore naturally summarized as a proportion.
Why is an independent response committee important?
The registered definition specifies assessment by an independent response committee (IRC). Independent assessment can reduce the possibility that investigators' knowledge of treatment exposure influences response classification. It also gives the endpoint a standardized adjudication framework.
Why does the 24-month time frame matter?
The endpoint is defined from first dose to 24 months. A response occurring outside that registered time frame would not automatically belong to the same endpoint definition. Time windows are therefore part of the endpoint itself rather than merely descriptive follow-up information.
Why is the response rate not automatically a treatment effect?
A response rate describes the probability of response in the studied cohort. A treatment effect requires a comparison: against a randomized control group, a prespecified benchmark, or another explicitly defined counterfactual framework. KarMMa's registry-reported design is single-group, and no reference response rate is reported in the ClinicalTrials.gov record.
What would a confidence interval for ORR tell us?
A confidence interval around the response proportion would quantify statistical uncertainty attributable to sampling variation under the chosen binomial framework. It would not describe the range of responses that an individual participant might experience, and it would not establish comparative efficacy against another therapy.
A p-value depends on a defined null hypothesis and statistical procedure. In this record, no formal statistical analysis is reported and no reference response rate is identified. Reporting a p-value without those ingredients would imply a hypothesis test that is not documented in the ClinicalTrials.gov record.
10. Statistical Interpretation of a Single-Group Response Endpoint
If the observed ORR were reported as a proportion p, it would represent the fraction of the relevant response-analysis population who achieved partial response or better under the registered definition during the specified period.
The response proportion would not, by itself, quantify the causal effect of bb2121 relative to another treatment. It would also not imply that every participant had the same probability of response or that participants without a response necessarily experienced no clinical benefit.
The precision of an observed proportion depends on the number of participants contributing to the response analysis and the number of observed responses. A confidence interval therefore belongs alongside the point estimate when the formal analysis is available.
A response rate based on all enrolled participants and a response rate based on an evaluable population answer subtly different questions. The ClinicalTrials.gov record does not provide enough statistical-analysis detail to determine the primary endpoint denominator.
11. Safety Results
the ClinicalTrials.gov record reports serious adverse events for the bb2121 arm as 108/137. Because the study is single-group, this is a safety count within the treated study cohort rather than a between-arm safety comparison.
Serious adverse events
BB2121 arm · Affected / at risk
The denominator of 137 should be retained exactly as reported. It should not be replaced with the enrollment figure of 149, because the registry supplies the serious-adverse-event measure using 137 participants at risk.
12. Safety and Efficacy Are Different Statistical Questions
The KarMMa record illustrates why efficacy and safety endpoints should be analyzed separately. The primary efficacy endpoint is binary overall response rate, whereas the registry-reported safety measure is a count of participants affected by serious adverse events among those at risk.
Efficacy
The registered primary endpoint asks whether participants achieved partial response or better according to the specified response criteria.
Safety
The ClinicalTrials.gov record counts participants affected by serious adverse events among 137 participants at risk in the bb2121 arm.
Different denominators
The enrollment count of 149 and the serious-adverse-event denominator of 137 should not be treated as interchangeable.
No composite conclusion
The ClinicalTrials.gov record does not support reducing efficacy and safety to a single statistical quantity or treatment-effect estimate.
13. Trial Timeline
Study start
The KarMMa trial began according to the ClinicalTrials.gov record.
Single-group treatment study
the ClinicalTrials.gov record identifies a phase 2 study with one arm, no masking, and primary purpose of treatment.
First dose to 24 months
The registered primary endpoint is overall response rate assessed during the period from first dose to 24 months.
Primary completion
the ClinicalTrials.gov record lists December 20, 2023 as the primary completion date.
Registry status
the ClinicalTrials.gov record in the ClinicalTrials.gov record lists the study status as COMPLETED.
14. What the Registry Results Do — and Do Not — Establish
| Question | What the ClinicalTrials.gov record supports | What they do not support |
|---|---|---|
| Was an efficacy endpoint registered? | Yes. Overall Response Rate was the single registered primary endpoint. | They do not supply an additional primary endpoint. |
| Was a result posted? | Yes. Results were posted for the primary endpoint. | The ClinicalTrials.gov record does not provide the numerical endpoint result. |
| Was a formal statistical analysis posted? | No statistical-analysis records are reported. | A treatment estimate, confidence interval, or p-value cannot be reconstructed. |
| Was there a comparator? | No. The design is single-group with one arm. | A randomized between-arm treatment effect cannot be reported. |
| Was safety reported? | Yes. Serious adverse events are reported as 108/137 for BB2121. | No comparator safety rate or formal safety comparison is reported. |
This distinction is central to responsible statistical reporting. A registry can contain a posted outcome measure while still lacking the formal analysis information required to reproduce a hypothesis test or confidence interval.
15. Design Features That Are Not Supported by the ClinicalTrials.gov record
Several statistical topics are important in clinical-trial methodology but cannot be attributed to this trial from the ClinicalTrials.gov record unless they are explicitly documented there.
| Topic | Status in the ClinicalTrials.gov record |
|---|---|
| Non-inferiority margin | Not reported. |
| Crossover | Not reported. |
| Factorial design | Not reported; the recorded design is single-group. |
| Multiplicity procedure | Not reported. |
| Interim analysis | Not reported. |
| Missing-data or imputation method | Not reported. |
| Stratification factors | Not reported. |
| Bayesian methods | Not reported. |
These omissions are analytically important. A complete statistical analysis should not silently substitute methods from another study, publication, protocol, or remembered version of a trial. For this page, the absence of a method in the ClinicalTrials.gov record is treated as an absence of evidence rather than an invitation to infer it.
16. Why This Trial Matters Statistically
KarMMa is a useful teaching example because it demonstrates a different statistical structure from a conventional randomized phase 3 trial. The central issue is not how to estimate a hazard ratio between two randomized groups. Instead, the statistical task begins with a single-group binary response endpoint and requires careful attention to the response definition, analysis denominator, time window, and uncertainty around a proportion.
| Concept | How it appears in KarMMa |
|---|---|
| Single-group design | The registry identifies a single-group design model with one arm. |
| Binary endpoint | The registered primary endpoint is Overall Response Rate. |
| Response criteria | Partial response or better according to IMWG Uniform Response Criteria for Multiple Myeloma. |
| Independent assessment | The primary endpoint specifies assessment by an independent response committee. |
| Time window | The registered primary endpoint is assessed from first dose to 24 months. |
| Proportion estimation | ORR is naturally expressed as the proportion meeting the response definition. |
| Confidence intervals | A formal single-group proportion analysis would ordinarily quantify uncertainty around the response estimate. |
| Reference benchmark | A formal one-sample hypothesis test would require a prespecified benchmark, which is not reported here. |
| Safety denominator | Serious adverse events are reported as 108/137 for BB2121. |
| Analysis transparency | The ClinicalTrials.gov record contains posted results but no statistical-analysis records. |
17. Statistical Methods Explained: A Deeper View
Why does the study design determine the estimand?
An estimand describes the quantity the analysis is intended to estimate. In a randomized trial, the treatment contrast can naturally be defined between randomized groups. In a single-group study, the primary descriptive estimand is instead the response probability within the studied treatment cohort. Without a comparator or benchmark, a causal treatment contrast is not identified by the design alone.
Why is a binary response endpoint different from survival analysis?
A binary endpoint reduces the observation to whether the prespecified event occurred within the defined assessment window. Survival analysis, by contrast, incorporates the timing of an event and censoring. The registered KarMMa primary endpoint is explicitly classified as binary, so a time-to-event hazard ratio should not be substituted for the registered endpoint.
Why does independent response assessment matter for statistical interpretation?
The response definition specifies an independent response committee. This means the endpoint is tied to an adjudication process rather than simply an investigator's informal assessment. The statistical analysis still depends on how participants are classified and which population forms the denominator.
Why can the denominator change the apparent response rate?
Suppose the number of responders remained fixed while the analysis denominator changed. The resulting proportion would change even though the responder count did not. This is why the analysis population must be specified before interpreting ORR. The ClinicalTrials.gov record does not provide the formal denominator for the primary response analysis.
Why is a historical-control comparison different from randomization?
A historical benchmark comes from participants observed under different circumstances, potentially involving different eligibility criteria, supportive care, disease characteristics, assessment procedures, or calendar periods. A prespecified historical-control hypothesis can be statistically valid in an appropriate design, but it does not create the same protection against confounding as random allocation. No historical-control benchmark is in the ClinicalTrials.gov record.
What does a confidence interval contribute beyond a point estimate?
The point estimate gives the observed proportion. A confidence interval describes the uncertainty associated with estimating the underlying response probability from a finite sample under the selected statistical framework. A narrow interval indicates greater statistical precision than a wide interval, all else being equal; it does not indicate that every patient has a similar individual outcome.
18. Limitations
- No randomized comparator: the registry-reported design is single-group, so the primary endpoint does not directly estimate a randomized treatment contrast.
- No formal statistical-analysis records reported: the ClinicalTrials.gov record states that results are posted, but no formal statistical analyses were posted.
- Numerical ORR result not reported here: the primary endpoint is identified as having results posted, but its numerical result is not included in the ClinicalTrials.gov record.
- Analysis denominator not reported: the enrollment count of 149 should not automatically be treated as the primary response-analysis denominator.
- No benchmark reported: a formal one-sample hypothesis test would require a prespecified reference value, which is not provided.
- Safety denominator differs from enrollment: serious adverse events are reported as 108/137, so the 137 at-risk denominator should not be replaced by 149.
- Unreported methodology: the ClinicalTrials.gov record does not specify the confidence-interval construction, missing-data rules, multiplicity strategy, interim monitoring, or other detailed statistical procedures.
- External publication details are limited: the ClinicalTrials.gov record identifies linked PubMed records by PMID but does not provide publication-specific numerical findings for incorporation into this page.
19. What a Complete Statistical Result Would Ideally Contain
For a binary primary endpoint in a single-group phase 2 study, a statistically complete presentation would ordinarily allow the reader to identify the response count, the analysis denominator, the resulting response proportion, and a confidence interval using the prespecified statistical method. If a formal benchmark hypothesis were part of the protocol, the benchmark and corresponding test procedure would also be stated.
Observed quantity
Number of participants achieving PR or better divided by the prespecified response-analysis population.
Precision
A confidence interval around the response proportion using the prespecified binomial method.
Hypothesis test
Only appropriate if a reference response probability was prospectively defined.
Interpretation
The observed response probability should be distinguished from a causal treatment effect or randomized comparison.
Because these formal numerical components are not included in the ClinicalTrials.gov record, this page deliberately stops short of reconstructing them from outside sources.
20. Sources
- ClinicalTrials.gov: NCT03361748 — KarMMa.
- Linked PubMed record: PMID 40198886.
- Linked PubMed record: PMID 37087950.
- Linked PubMed record: PMID 36794354.
- Linked PubMed record: PMID 36705644.
- Linked PubMed record: PMID 35485211.
The quantitative content of this page is restricted to the registry-reported KarMMa trial data. The linked PubMed records are provided as source-navigation references because they were identified in the ClinicalTrials.gov record; publication-specific numerical results have not been added where they were not present in the ClinicalTrials.gov record.
21. Related Tutorials
Learn more about the statistical concepts used in this trial:
22. Related Calculators
Continue through Clinical Biostats
Explore statistical tutorials and calculators for clinical-trial endpoints, binary outcomes, confidence intervals, and study design.
23. Record Summary
KarMMa is a completed phase 2, single-group study of bb2121 in subjects with relapsed and refractory multiple myeloma. The ClinicalTrials.gov record reports 149 enrolled participants, one intervention arm, no masking, and a binary primary endpoint of Overall Response Rate measured from first dose to 24 months. The endpoint is defined as the number of participants achieving partial response or better according to IMWG Uniform Response Criteria for Multiple Myeloma as assessed by an independent response committee.
The principal statistical lesson is that a single-group binary endpoint should be interpreted as a response proportion rather than as a randomized treatment effect. A complete formal analysis would require the response count, the prespecified analysis denominator, an appropriate confidence interval, and—if applicable—a prospectively defined reference response rate for hypothesis testing.