This page separates information reported in the ClinicalTrials.gov record from statistical interpretation and methodological explanation. No numerical efficacy estimate, confidence interval, or p-value is added beyond the ClinicalTrials.gov record. The ClinicalTrials.gov record indicates that results were posted, but no formal statistical-analysis results were provided in the statistical-analysis field.
1. Trial at a Glance
DREAMM-2 was a randomized, parallel, open-label phase 2 treatment study in multiple myeloma. The ClinicalTrials.gov record reports an enrollment of 221 participants and three study arms involving belantamab mafodotin frozen liquid and lyophilized powder formulations.
| Feature | DREAMM-2 |
|---|---|
| Trial name | DREAMM-2 |
| Phase | Phase 2 |
| Condition | Multiple Myeloma |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 221 |
| Study status | Completed |
| Start | 2018-06-18 |
| Primary completion | 2019-06-21 |
| Lead sponsor | GlaxoSmithKline |
| Sponsor type | Industry |
| ClinicalTrials.gov identifier | NCT03525678 |
2. Clinical Question
The registry description frames DREAMM-2 as a study of the efficacy and safety of two doses of GSK2857916 in participants with multiple myeloma who had failed prior treatment with an anti-CD38 antibody. The registered primary efficacy outcomes were overall response rate assessed by an Independent Review Committee.
Population
Participants with multiple myeloma who had failed prior treatment with an anti-CD38 antibody.
Intervention
Belantamab mafodotin, identified in the registry-reported intervention data as GSK2857916, administered as frozen liquid or lyophilized powder formulations.
Comparator
The ClinicalTrials.gov record describes three parallel study arms rather than a separate conventional placebo or untreated control arm.
Primary question
What proportion of participants achieved a confirmed partial response or better under the registered IMWG response definition, as assessed by the Independent Review Committee?
3. Trial Design
Randomization is an important statistical feature because it creates the framework for comparing outcomes between assigned treatment groups without relying on observational adjustment as the primary source of comparability. DREAMM-2 is also explicitly described as having no masking, so treatment assignment was not concealed through a blinded design.
Study timing
Study start
The ClinicalTrials.gov record identifies this as the study start date.
Primary completion
The ClinicalTrials.gov record identifies this as the primary completion date.
Current study status in the ClinicalTrials.gov record
the ClinicalTrials.gov record identifies DREAMM-2 as completed and indicates that results were posted.
4. Study Arms and Safety Population Information
The ClinicalTrials.gov record provides affected and at-risk counts for serious adverse events in the main study and in a PACT phase. These counts should be kept distinct from the primary efficacy populations because the ClinicalTrials.gov record does not provide a formal statistical comparison of serious adverse-event rates.
| Study cohort / treatment description | Serious adverse events affected | At risk |
|---|---|---|
| Main Study: GSK2857916 2.5 mg/kg, frozen | 43 | 95 |
| Main Study: GSK2857916 3.4 mg/kg, frozen | 53 | 99 |
| Main Study: GSK2857916 3.4 mg/kg, lyophilized | 15 | 24 |
| PACT Phase: GSK2857916 2.5 mg/kg, frozen | 1 | 1 |
| PACT Phase: GSK2857916 3.4 mg/kg, frozen | 2 | 2 |
Within the ClinicalTrials.gov record, the main-study denominators sum to 218 participants and the PACT-phase denominators sum to 3 participants, together corresponding to the reported enrollment of 221. This arithmetic describes the registry-reported counts; it does not establish that the safety denominator should be used as the denominator for every efficacy analysis.
5. Primary Endpoints
DREAMM-2 has two registered primary endpoints. Both are binary overall response rate measures assessed by an Independent Review Committee and both use a time frame of up to 48 weeks.
| Registered primary endpoint | Time frame | Endpoint type | Definition |
|---|---|---|---|
| Overall Response Rate (ORR) by Independent Review Committee (IRC) (Full Analysis Population) | Up to 48 weeks | Binary | ORR was determined according to the 2016 International Myeloma Working Group (IMWG) response criteria by IRC. ORR was calculated as the percentage of participants with a confirmed partial response (PR) or better, meaning PR, very good partial response (VGPR), complete response (CR), or stringent complete response (sCR). Confidence intervals were based on the exact method. |
| Overall Response Rate by Independent Review Committee (Efficacy Population) | Up to 48 weeks | Binary | ORR was determined according to the 2016 IMWG response criteria by IRC. ORR was calculated as the percentage of participants with a confirmed PR or better, meaning PR, VGPR, CR, or sCR. Confidence intervals were based on the exact method. The registry definition states that the Efficacy Population comprised the first 130 intent-to-treat participants whether or not randomized treatment (frozen solution) was administered. |
6. Results and Registry Analysis Status
the ClinicalTrials.gov record states that results were posted and that 62 outcome measures were posted. However, no formal statistical analyses were posted. Accordingly, there are no primary-endpoint treatment-effect estimates, confidence intervals, or p-values in the ClinicalTrials.gov record that can be reported here.
This is an important distinction between results being posted and a formal comparative statistical analysis being available in the ClinicalTrials.gov record. The primary endpoints themselves are clearly defined, including their response criteria, analysis populations, time frame, and exact-confidence-interval method. The ClinicalTrials.gov record does not provide the resulting ORR percentages.
7. Planned Analysis
Because the primary endpoints are binary response outcomes, the natural statistical object is a proportion: the percentage of participants who achieve a confirmed PR or better by the specified response criteria within the stated time frame.
The registry definition explicitly states that confidence intervals for ORR were based on the exact method. The ClinicalTrials.gov record does not state the numerical ORR estimate or its resulting confidence interval.
Full Analysis Population
For the Full Analysis Population endpoint, the registry specifies that ORR is the percentage of participants with a confirmed PR, VGPR, CR, or sCR. The appropriate descriptive analysis is therefore a binomial proportion with an exact confidence interval, using the registry-defined Full Analysis Population as the denominator.
Efficacy Population
The second primary endpoint uses the registry-defined Efficacy Population. The registry-reported definition specifies that this population comprised the first 130 intent-to-treat participants whether or not randomized treatment with the frozen solution was administered. The same binary response definition and exact confidence-interval approach apply.
Why the population definition matters
Changing the denominator can change the estimated response rate even if the underlying response events are unchanged. For that reason, the Full Analysis Population result and the Efficacy Population result should not be treated as interchangeable summaries. A statistically careful report would identify the population before interpreting the response proportion.
8. Statistical Methodology
Binary endpoint analysis
Both registered primary endpoints are classified in the ClinicalTrials.gov record as binary. Each participant is conceptually classified according to whether the prespecified response criterion has been achieved. The resulting response proportion summarizes the frequency of confirmed PR or better.
Exact confidence intervals
The registry explicitly states that confidence intervals for ORR were based on the exact method. For a binomial response proportion, an exact interval avoids relying on a large-sample normal approximation when calculating uncertainty around the observed proportion.
Here, x represents the number of participants meeting the confirmed-response definition and n represents the number of participants in the specified analysis population. The ClinicalTrials.gov record does not provide the value of x for either primary endpoint.
Independent Review Committee assessment
The primary response endpoints were determined by an Independent Review Committee. From a statistical perspective, an independent review framework is intended to provide a standardized assessment of response rather than relying solely on treatment-site assessment. The ClinicalTrials.gov record identifies the IRC as the assessment source but do not provide further details about review adjudication, discordant assessments, or missing-assessment handling.
2016 IMWG response criteria
The response definition is based on the 2016 International Myeloma Working Group criteria. For the registered endpoint, the relevant threshold is confirmed PR or better, including PR, VGPR, CR, and sCR. The binary endpoint therefore collapses several ordered response categories into one clinically defined responder/non-responder outcome.
Randomization
Randomization is the principal design feature supporting treatment-group comparison. When allocation is randomized, observed differences in response can be interpreted within the randomized comparison framework rather than as simple associations between treatment selection and outcome. DREAMM-2 is explicitly identified as randomized and parallel, although the ClinicalTrials.gov record does not provide the randomization ratio or a formal stratification scheme.
9. Statistical Methods Explained
What does ORR measure?
ORR measures the proportion of participants who achieve a confirmed PR or better under the specified IMWG response criteria. It is a binary endpoint even though the underlying response categories include PR, VGPR, CR, and sCR.
Why use an exact confidence interval?
A response rate is a binomial proportion. An exact interval is a way of quantifying uncertainty around that proportion without depending on the same large-sample approximation used by a simple normal-theory interval. The registry specifically states that the confidence intervals were based on the exact method.
Why does the analysis population matter?
The two primary endpoints use different population definitions. The Full Analysis Population endpoint and the Efficacy Population endpoint therefore answer closely related questions using different denominators. A response percentage without its population definition is incomplete statistical reporting.
Why does randomization matter for interpretation?
Randomization establishes treatment assignment before outcome information is observed and, in expectation, balances measured and unmeasured baseline characteristics across randomized groups. That is why a randomized comparison provides a stronger framework for causal inference than a simple observational comparison. Randomization does not, however, eliminate statistical uncertainty or guarantee identical groups in every sample.
What does an ORR confidence interval tell us?
The confidence interval quantifies uncertainty around the estimated response proportion under the specified statistical framework. It does not describe the range of responses an individual patient can experience, and it does not itself establish whether two study arms differ.
Why is a posted result not the same as a posted statistical comparison?
A registry can contain outcome-measure results without supplying every component of a formal inferential analysis. A complete comparative analysis might include treatment-group estimates, effect measures, confidence intervals, and a prespecified hypothesis test. The registry-reported DREAMM-2 data contain the registered endpoint definitions but no entries in the statistical-analysis field, so those inferential quantities cannot be reported from the ClinicalTrials.gov recordset.
10. Analysis Populations and Estimands
The distinction between the two primary endpoints provides a useful example of why statistical reporting begins with an explicit estimand and analysis population. The Full Analysis Population endpoint asks for ORR in one defined population, while the Efficacy Population endpoint uses a narrower registry-defined population.
| Component | Full Analysis Population endpoint | Efficacy Population endpoint |
|---|---|---|
| Outcome | Confirmed PR or better | Confirmed PR or better |
| Assessment | Independent Review Committee | Independent Review Committee |
| Response criteria | 2016 IMWG | 2016 IMWG |
| Time frame | Up to 48 weeks | Up to 48 weeks |
| Endpoint type | Binary | Binary |
| Confidence interval | Exact method | Exact method |
| Population definition reported | Full Analysis Population | First 130 intent-to-treat participants whether or not randomized treatment (frozen solution) was administered |
These definitions also illustrate why a statistical result cannot safely be transferred from one population to another. Even when two endpoints use the same clinical response criterion, their statistical interpretation depends on who is included in the denominator and how that population was constructed.
11. Safety
The ClinicalTrials.gov record reports serious adverse events by study cohort. The counts are presented as affected participants over participants at risk.
| Cohort | Affected / at risk | Descriptive percentage |
|---|---|---|
| Main Study: GSK2857916 2.5 mg/kg, frozen | 43 / 95 | 45.3%* |
| Main Study: GSK2857916 3.4 mg/kg, frozen | 53 / 99 | 53.5%* |
| Main Study: GSK2857916 3.4 mg/kg, lyophilized | 15 / 24 | 62.5%* |
| PACT Phase: GSK2857916 2.5 mg/kg, frozen | 1 / 1 | 100%* |
| PACT Phase: GSK2857916 3.4 mg/kg, frozen | 2 / 2 | 100%* |
*The percentages in this table are simple arithmetic descriptions of the registry-reported affected/at-risk counts. They are not additional registry-reported estimates and are not accompanied by confidence intervals or hypothesis tests in the ClinicalTrials.gov record.
Safety and efficacy also answer different questions. ORR asks whether a participant met a defined response criterion, while serious adverse-event reporting describes an important safety outcome. Neither endpoint should be used as a substitute for the other.
12. Design Features the Supplied Data Do and Do Not Specify
| Design topic | What the ClinicalTrials.gov record reports | Statistical implication |
|---|---|---|
| Randomization | Randomized | Supports comparison according to assigned treatment. |
| Parallel design | Parallel | Participants are evaluated within concurrent study arms rather than a crossover structure. |
| Masking | None | The ClinicalTrials.gov record identifies the study as open-label; response assessment is nevertheless attributed to an IRC. |
| Multiplicity | Not specified in the ClinicalTrials.gov record | No multiplicity adjustment should be inferred. |
| Interim analysis | Not specified in the ClinicalTrials.gov record | No interim alpha-spending or stopping rule should be inferred. |
| Missing-data method | Not specified in the ClinicalTrials.gov record | No imputation rule should be inferred. |
| Stratification | Not specified in the ClinicalTrials.gov record | No stratified analysis should be assumed. |
| Bayesian methods | Not specified in the ClinicalTrials.gov record | No Bayesian prior or posterior analysis should be inferred. |
| Non-inferiority margin | Not specified; the primary endpoints are ORR measures | No non-inferiority framework should be applied to the primary endpoints. |
| Crossover | Not specified in the ClinicalTrials.gov record | No crossover adjustment should be inferred. |
| Factorial design | Not specified; design model is parallel | No factorial interaction analysis should be inferred. |
This distinction is important for an independent statistical analysis. A method may be common for a particular clinical-trial design, but common practice is not evidence that the method was actually prespecified in this trial. Where the ClinicalTrials.gov record is silent, this page does not attribute a specific method to DREAMM-2.
13. Interpreting the Serious-Adverse-Event Counts
The registry-reported main-study counts range from 43/95 to 53/99 and 15/24 across the listed treatment cohorts. These are descriptive frequencies of participants affected by serious adverse events among those at risk in the registry-reported cohorts.
The counts do not by themselves establish a statistically significant difference between doses or formulations. A formal comparison would require a prespecified comparison, an analysis population, an estimand, and an appropriate uncertainty measure.
The PACT-phase denominators are 1 and 2 in the ClinicalTrials.gov record. Such small denominators make the corresponding observed proportions highly unstable and unsuitable for direct comparison with much larger cohorts without accounting for statistical uncertainty and the purpose of the PACT phase.
14. Why This Trial Matters Statistically
DREAMM-2 is a useful statistical teaching case because it combines randomized treatment allocation with binary response endpoints, distinct analysis populations, independent review, exact confidence intervals, multiple formulations and doses, and descriptive safety data.
| Statistical concept | How it appears in DREAMM-2 |
|---|---|
| Randomization | The trial is explicitly randomized. |
| Parallel design | The registry identifies a parallel three-arm design. |
| Binary endpoint | Both registered primary endpoints are ORR measures. |
| Analysis population | The two primary endpoints use different population definitions. |
| Exact confidence interval | The registry explicitly states that ORR confidence intervals use the exact method. |
| Independent review | Both primary endpoints are assessed by an Independent Review Committee. |
| Response criteria | ORR uses the 2016 IMWG response criteria and requires confirmed PR or better. |
| Safety denominator | Serious adverse events are reported as affected participants over participants at risk. |
| Open-label design | The registry-reported masking field is "None." |
| Separation of evidence from inference | The record reports results but supplies no formal statistical-analysis entries. |
The most instructive feature is perhaps the last one. A registry record can contain detailed endpoint definitions while leaving the formal inferential comparison unavailable in a particular data extract. A rigorous analysis therefore distinguishes what can be calculated or interpreted from what is actually reported.
15. Limitations
- No formal statistical-analysis results reported: The statistical-analysis array in the ClinicalTrials.gov record is empty, so no treatment-effect estimate, confidence interval, or p-value is available for either primary endpoint.
- No primary ORR counts reported: The registry-reported endpoint definitions describe what constitutes a response but do not provide the number of responders in either primary analysis population.
- Different analysis populations: The Full Analysis Population and Efficacy Population endpoints should not be treated as interchangeable.
- Limited design detail: The ClinicalTrials.gov record does not specify randomization ratio, stratification factors, missing-data handling, multiplicity procedures, interim-analysis rules, or Bayesian methods.
- Safety analysis limitations: Serious-adverse-event counts are reported descriptively without comparative effect measures or uncertainty intervals.
- Small PACT denominators: The PACT-phase safety cohorts contain very small numbers at risk, making their observed proportions particularly unstable.
- Open-label design: The trial is identified as unmasked. The primary response endpoints were nevertheless assessed by an Independent Review Committee, according to the registry-reported definitions.
- Registry scope: The ClinicalTrials.gov record contains 62 posted outcome measures, but only the two registered primary endpoints and the registry-reported serious-adverse-event counts are described in detail here.
16. What Can and Cannot Be Concluded From the Supplied Record
What is documented
DREAMM-2 was a completed, randomized, parallel phase 2 study with 221 enrolled participants and three arms. Its two registered primary endpoints were binary ORR measures assessed by an IRC over up to 48 weeks.
What is methodologically specified
The ORR definition requires confirmed PR or better under the 2016 IMWG criteria, and the registry states that exact confidence intervals were used.
What is reported for safety
Serious adverse-event affected/at-risk counts are available for five listed study cohorts spanning the main study and PACT phase.
What is not reported
The statistical-analysis field contains no formal analyses. Numerical primary ORR estimates, treatment contrasts, confidence intervals, and p-values are therefore not reported on this page.
17. Sources
- ClinicalTrials.gov: NCT03525678 — DREAMM-2.
- Linked publication record: PubMed record — PMID 38798125.
- Linked publication record: PubMed record — PMID 37465991.
- Linked publication record: PubMed record — PMID 34465265.
- Linked publication record: PubMed record — PMID 34314018.
- Linked publication record: PubMed record — PMID 34039952.
The publication links above are provided because they are listed in the ClinicalTrials.gov record. This page does not import numerical results from those publications; the numerical trial facts presented here are restricted to the ClinicalTrials.gov record.
18. Related Tutorials
Learn more about the methods used in this trial:
19. Related Statistical Calculators
Continue through the Clinical Biostats statistical pathway
Explore the underlying trial-design and statistical methods through tutorials, then use the appropriate calculator when a numerical analysis is required.
20. Record Summary
DREAMM-2 provides a useful example of how a randomized phase 2 trial can be analyzed through binary response endpoints. The registry defines two primary ORR endpoints, both assessed by an Independent Review Committee according to the 2016 IMWG response criteria over up to 48 weeks, with exact confidence intervals. The two endpoints differ in their analysis populations, making population definition an important part of interpretation. The ClinicalTrials.gov record also identifies a three-arm parallel design, no masking, and 221 enrolled participants, and it provides descriptive serious-adverse-event counts for main-study and PACT-phase cohorts.
The most important statistical limitation of the ClinicalTrials.gov record is that the statistical-analysis field is empty. Results are identified as posted, but the ClinicalTrials.gov record does not contain numerical ORR results or formal comparative statistical analyses.