This page separates reported trial results from statistical interpretation. Numerical results are restricted to the ClinicalTrials.gov record and the linked publication identifier.
1. Trial at a Glance
DASCERN was a randomized phase 2 trial evaluating dasatinib versus imatinib in patients with chronic phase chronic myeloid leukemia who did not have a favorable response to imatinib. The registry describes a parallel, unmasked, randomized treatment study with a superiority hypothesis.
| Feature | DASCERN |
|---|---|
| Trial name | DASCERN |
| Phase | 2 |
| Condition | Chronic Phase Chronic Myeloid Leukemia |
| Brief title | Study of Dasatinib vs Imatinib in Patients With Chronic Myeloid Leukemia (CML) Who Did Not Have Favorable Response to Imatinib |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 262 |
| Interventions | Imatinib; Dasatinib |
| Primary endpoints | 1 registered primary endpoint |
| Primary endpoint type | Binary |
| Hypothesis type | Superiority |
| Results posted | Yes |
| Outcome measures posted | 5 |
| Statistical analyses posted | 1 |
| Lead sponsor | Bristol-Myers Squibb |
| Sponsor type | Industry |
2. Clinical Question
The central statistical question was whether the proportion of patients achieving Major Molecular Response (MMR) after 12 months of CML treatment differed between patients randomized to imatinib at a dose of at least 400 mg and patients randomized to dasatinib at 100 mg.
Population
Patients with chronic phase chronic myeloid leukemia who did not have a favorable response to imatinib.
Intervention
Dasatinib (100 mg) in the primary statistical comparison.
Comparator
Imatinib (≥400 mg) in the primary statistical comparison.
Primary question
Does the percentage of patients achieving MMR after 12 months differ between the randomized treatment groups?
3. Trial Design
Imatinib
- Imatinib (≥400 mg)
- Comparator in the posted primary statistical analysis
Dasatinib
- Dasatinib (100 mg)
- Intervention in the posted primary statistical analysis
4. Trial Timing and Status
| Trial milestone | Date / status |
|---|---|
| Start | 2012-09-12 |
| Primary completion | 2017-11-08 |
| Status | Completed |
The registry identifies DASCERN as completed, with the study beginning on 2012-09-12 and reaching primary completion on 2017-11-08. These dates describe the registry timeline; they do not themselves provide information about the magnitude of the treatment effect.
5. Primary Endpoint
| Endpoint | Registry definition and time frame | Type |
|---|---|---|
| Percentage of Patients Achieving Major Molecular Response (MMR) After 12 Months of CML Treatment | At 12 months after Day 1 initiation of 1st line treatment with imatinib or imatinib at any dose, after less than optimal response to first-line imatinib. | Binary |
What counts as MMR?
The registry defines Major Molecular Response as a 3-log reduction in BCR-ABL transcripts from the standardized baseline, which represents 100% on the international scale. A 3-log reduction is fixed at 0.1% for MMR.
The registry definition also states that 95% confidence intervals are Clopper-Pearson (Exact) two-sided 95% confidence intervals and that the P-value is based on a Cochran-Mantel-Haenszel (CMH) test stratified by Sokal score, with the registry wording ending at “high, intermediate”.
6. Statistical Methodology
Cochran-Mantel-Haenszel test
The posted primary analysis used the Cochran-Mantel-Haenszel test. This is appropriate for a categorical or binary outcome when the comparison is intended to account for a stratification variable. Here, the registry identifies Sokal score as the stratification factor.
Rather than treating all observations as if they came from one unstructured 2 × 2 table, the CMH framework combines information across defined strata while accounting for the stratification variable.
Why stratification matters
Stratification is useful when a baseline characteristic is expected to be relevant to the treatment comparison. In DASCERN, the registry identifies Sokal score as the stratification variable for the primary CMH test. The resulting comparison therefore addresses treatment-group differences while accounting for the specified Sokal-score strata.
Binary endpoint analysis
MMR is recorded as a binary outcome: a patient either achieves the prespecified MMR definition at the specified assessment point or does not. This differs fundamentally from a continuous endpoint, where an average difference might be the principal treatment contrast, and from a time-to-event endpoint, where follow-up and censoring are central components of the analysis.
Clopper-Pearson confidence intervals
The registry specifies exact two-sided 95% confidence intervals using the Clopper-Pearson method. This method is designed for binomial proportions and is commonly used when the goal is to obtain an exact interval for a proportion rather than relying on a normal approximation.
The primary endpoint is therefore a proportion-based outcome. The CMH test then provides the formal stratified comparison between randomized treatment groups.
Superiority hypothesis
The registry identifies the hypothesis type as superiority. Conceptually, a superiority analysis asks whether the treatment groups differ in the specified direction of interest rather than asking whether a new treatment is sufficiently close to a comparator to satisfy a non-inferiority margin.
7. Results: Primary Endpoint
The ClinicalTrials.gov record contains one posted formal statistical analysis for the primary endpoint. The analysis compares Imatinib (≥400 mg) with Dasatinib (100 mg) among All Randomized Participants using a Cochran-Mantel-Haenszel test.
Primary CMH test
Hypothesis type: Superiority · Analysis population: All Randomized Participants
Method: Cochran-Mantel-Haenszel test
| Primary analysis element | Reported value |
|---|---|
| Outcome measure | Percentage of Patients Achieving Major Molecular Response (MMR) After 12 Months of CML Treatment |
| Time frame | At 12 months after Day 1 initiation of 1st line treatment with imatinib or imatinib at any dose, after less than optimal response to first-line imatinib. |
| Analysis population | All Randomized Participants |
| Groups compared | Arm 1: Imatinib (≥400 mg) vs Arm 2: Dasatinib (100 mg) |
| Method | Cochran-Mantel-Haenszel |
| P-value | 0.005 |
| Hypothesis type | Superiority |
The reported P-value of 0.005 is the formal result of the posted CMH comparison of the randomized treatment groups for the 12-month MMR endpoint, with Sokal score used for stratification. Under the statistical framework represented by the registry analysis, this provides evidence against the null hypothesis of no treatment-group difference in the stratified binary outcome.
The P-value does not measure the size of the treatment effect. A P-value is a measure of compatibility between the observed data and the statistical null hypothesis under the specified analysis; it is not a percentage improvement, a probability that one treatment works, or a measure of clinical importance.
The ClinicalTrials.gov record does not provide the arm-specific MMR percentages, the corresponding confidence-interval values, or an absolute treatment difference. The registry does specify that confidence intervals for the proportions use the Clopper-Pearson exact two-sided 95% method, but the numerical interval estimates are not included in the registry-reported statistical-analysis entry. Consequently, the magnitude and precision of the observed MMR difference cannot be reconstructed from the reported P-value alone.
The analysis is based on All Randomized Participants, which preserves the treatment assignment defined by randomization. The interpretation also depends on the prespecified binary endpoint definition and the specified Sokal-score stratification.
8. Understanding the Primary Result
A P-value is not an effect size
The value 0.005 tells us about the statistical test, not how large the difference in MMR percentages was. Two studies can produce the same P-value with different effect sizes, and the same effect size can produce different P-values under different sample sizes and amounts of information.
What the CMH result does establish
The posted analysis establishes that a formal stratified categorical comparison was performed and that its reported P-value was 0.005. The test accounted for Sokal score according to the registry's description.
What the CMH result does not establish
The P-value alone does not provide the percentage achieving MMR in either arm, the absolute difference between arms, the relative difference, or a confidence interval for the treatment effect.
9. Analysis Population
| Population | Role in posted analysis |
|---|---|
| All Randomized Participants | Analysis population specified for the primary statistical comparison. |
Using all randomized participants for the primary analysis links the efficacy comparison to the original randomized allocation. This is an important design principle because randomization is intended to create comparable treatment groups at baseline, subject to chance variation.
The ClinicalTrials.gov record does not provide a separate per-protocol analysis, modified intention-to-treat definition, or additional efficacy analysis population. No such population is therefore introduced here.
10. Sokal-Score Stratification
The primary P-value is based on a Cochran-Mantel-Haenszel test stratified by Sokal score. The registry definition posted on ClinicalTrials.gov for the endpoint identifies the score categories in the wording “high, intermediate” but does not provide a complete category description beyond that registry-reported phrase.
Why stratify?
A stratified categorical analysis can account for a prespecified classification variable rather than ignoring that variable when combining the treatment-group comparisons.
Why CMH?
The CMH procedure is designed to combine information from multiple strata for a categorical treatment comparison.
What is being preserved?
The analysis retains the randomized treatment comparison while incorporating the specified Sokal-score stratification.
What is not provided?
The ClinicalTrials.gov record does not provide stratum-specific MMR percentages or stratum-specific effect estimates.
11. Safety Results
The ClinicalTrials.gov record provides serious adverse events by treatment arm. These figures are reported as affected participants / participants at risk.
| Group | Serious adverse events | Affected / at risk |
|---|---|---|
| Arm 1: Imatinib | Serious adverse events | 11/86 |
| Arm 2: Dasatinib | Serious adverse events | 48/171 |
| Dasatinib after crossover from imatinib | Serious adverse events | 8/46 |
The crossover group is reported separately in the ClinicalTrials.gov record as Dasatinib After Crossover From Imatinib: 8/46. It should not be treated as a third randomized treatment arm when interpreting the randomized efficacy comparison.
12. Crossover and Treatment Interpretation
The safety data explicitly identify a group described as Dasatinib After Crossover From Imatinib. This indicates that crossover occurred and that some participants who began in the imatinib group subsequently received dasatinib.
Crossover is statistically important because randomized assignment and treatment actually received become different concepts after participants switch treatment. For an intention-to-treat efficacy analysis, participants remain associated with their randomized group. For treatment-exposure summaries, however, subsequent treatment can affect the observed safety and treatment-exposure experience.
The ClinicalTrials.gov record does not provide the timing of crossover, the reasons for crossover, or a crossover-adjusted efficacy analysis. No quantitative adjustment for crossover is therefore attempted here.
13. Statistical Methods Explained
Why was a Cochran-Mantel-Haenszel test used?
MMR is a binary outcome, so a categorical-data method is appropriate. The CMH procedure is particularly useful when the comparison is stratified. In DASCERN, the registry specifies Sokal score as the stratification variable, allowing the primary treatment comparison to account for that factor.
What does a binary endpoint mean here?
The endpoint classifies patients according to whether they achieved the prespecified MMR definition at 12 months. The statistical question is therefore about the distribution of responders and nonresponders between the treatment groups rather than about a mean molecular-response measurement.
Why is Sokal score included in the analysis?
The registry explicitly states that the CMH P-value is stratified by Sokal score. Stratification allows the treatment comparison to be performed within the specified strata and then combined into an overall adjusted categorical test.
What does P = 0.005 mean?
It is the reported P-value from the primary CMH test. Under the specified statistical framework, it represents the evidence against the null hypothesis of no treatment-group difference in the stratified MMR outcome. It does not quantify the size of the difference.
Why are confidence intervals important?
A confidence interval adds information about precision that a P-value alone cannot provide. The registry specifies exact two-sided 95% Clopper-Pearson confidence intervals for the MMR proportions. The numerical confidence intervals themselves are not included in the ClinicalTrials.gov record.
Why should the P-value not be converted into a treatment benefit?
A P-value and an effect estimate answer different questions. The P-value addresses the statistical test, whereas an effect estimate describes the magnitude of the observed treatment difference. Without the arm-specific MMR percentages or another reported effect measure, the magnitude cannot be derived from 0.005 alone.
14. Randomization and Causal Interpretation
Randomization is one of the central statistical features of DASCERN. Participants were randomly allocated, creating the basis for comparing outcomes between treatment groups without relying solely on adjustment for measured baseline characteristics.
The causal interpretation of a randomized comparison depends on maintaining the distinction between assignment and treatment received. The primary statistical analysis is explicitly based on all randomized participants, while the safety record separately identifies participants who received dasatinib after crossover from imatinib.
The strength of the randomized design comes from the assignment mechanism. Subsequent treatment changes do not erase the original randomized allocation used for the primary efficacy analysis.
15. Endpoint Timing and Interpretation
The primary endpoint is evaluated at a specific time point: 12 months after Day 1 initiation of 1st line treatment, using the registry wording posted on ClinicalTrials.gov for the endpoint.
This makes the endpoint different from a time-to-event outcome. The primary analysis asks whether the binary MMR status at the specified 12-month assessment differs between randomized groups. The analysis therefore does not require a Kaplan-Meier curve or a Cox proportional-hazards model for the primary endpoint.
Time point
12 months after Day 1 initiation of 1st line treatment as specified in the registry.
Outcome type
Binary MMR achievement.
Statistical test
Cochran-Mantel-Haenszel test.
Stratification
Sokal score.
16. What the Posted Result Can and Cannot Tell Us
| Question | Supported by the ClinicalTrials.gov record? |
|---|---|
| Was a formal primary statistical test posted? | Yes. |
| What was the primary statistical method? | Cochran-Mantel-Haenszel. |
| What was the reported P-value? | 0.005. |
| Was the test a superiority test? | Yes. |
| What analysis population was used? | All Randomized Participants. |
| Was Sokal score used for stratification? | Yes. |
| What were the arm-specific MMR percentages? | Not included in the ClinicalTrials.gov record. |
| What was the absolute MMR difference? | Not included in the ClinicalTrials.gov record. |
| What were the numerical 95% confidence intervals? | The method is specified as Clopper-Pearson exact two-sided 95% CI, but numerical intervals are not included in the ClinicalTrials.gov record. |
This distinction is important for responsible interpretation. The reported P-value is a real result and can be interpreted as such, but it cannot legitimately be transformed into an unreported response percentage or effect size.
17. Important Limitations and Interpretation Issues
- Incomplete numerical effect reporting in the registry-reported analysis: the primary statistical-analysis entry provides the P-value but not the arm-specific MMR percentages or numerical confidence intervals.
- P-value versus effect size: P = 0.005 establishes the result of the formal hypothesis test but does not quantify the magnitude of the treatment difference.
- Confidence-interval limitation: the registry specifies Clopper-Pearson exact two-sided 95% confidence intervals, but the numerical intervals are not reported in the posted statistical-analysis data provided here.
- Stratification: the CMH analysis accounts for Sokal score as specified by the registry, but the ClinicalTrials.gov record does not provide stratum-specific results.
- Crossover: the safety record identifies dasatinib after crossover from imatinib. The ClinicalTrials.gov record does not provide a crossover-adjusted efficacy analysis.
- Safety scope: serious adverse-event counts are provided, but the ClinicalTrials.gov record does not contain a complete safety table or formal comparative safety analysis.
- Analysis population: the posted primary analysis uses all randomized participants, while the registry-reported serious-adverse-event denominators are reported separately by safety group.
- Endpoint timing: the primary endpoint is defined at a 12-month assessment point, so it should not be interpreted as a time-to-event endpoint.
- Registry wording: the registry-reported endpoint definition ends with the partial phrase “high, intermediate” when describing the Sokal-score stratification categories. No additional category wording is reported and none is inferred here.
18. Why This Trial Matters Statistically
DASCERN is a useful teaching example because it demonstrates a different statistical pathway from trials whose primary endpoints are survival times. Its primary endpoint is binary, the analysis is stratified, and the formal comparison uses the Cochran-Mantel-Haenszel test rather than a time-to-event model.
| Concept | How it appears in DASCERN |
|---|---|
| Randomization | Participants were randomized to treatment groups. |
| Parallel design | The registry identifies a parallel design model. |
| Binary endpoint | MMR achievement at 12 months is the registered primary endpoint. |
| Superiority hypothesis | The posted analysis is classified as a superiority test. |
| Cochran-Mantel-Haenszel test | The primary P-value is based on the CMH method. |
| Stratification | The primary CMH test is stratified by Sokal score. |
| Exact confidence intervals | The registry specifies Clopper-Pearson exact two-sided 95% confidence intervals. |
| Analysis population | The posted primary analysis uses all randomized participants. |
| Crossover | The safety data identify dasatinib after crossover from imatinib. |
| Safety analysis | Serious adverse events are reported by randomized arm and for the crossover group. |
19. Statistical Interpretation of the Primary P-value
The reported P = 0.005 is the result of the prespecified-type CMH categorical comparison reported for the primary MMR endpoint. The test is stratified by Sokal score and uses all randomized participants.
The P-value is not the probability that the null hypothesis is true. It is also not the probability that dasatinib is effective, the percentage of patients benefiting, or the percentage difference in MMR between the arms.
A complete clinical interpretation normally places the hypothesis-test result alongside the observed MMR proportions and their confidence intervals. Because the ClinicalTrials.gov record does not provide those numerical values, the size and precision of the treatment-group difference cannot be stated from the P-value alone.
The P-value is not an isolated number. It is produced by a particular statistical procedure. Here, the procedure is the Cochran-Mantel-Haenszel test, and the registry specifies Sokal-score stratification. Changing the analysis framework can change the resulting test statistic and P-value.
20. Statistical Design Topics Not Reported in the ClinicalTrials.gov record
The registry-reported DASCERN data support discussion of randomization, parallel allocation, superiority testing, CMH stratification, and crossover. They do not provide a non-inferiority margin, factorial structure, Bayesian method, interim-analysis procedure, multiplicity strategy, or missing-data/imputation method.
| Design topic | What can be stated from the ClinicalTrials.gov record |
|---|---|
| Non-inferiority margin | Not provided; the posted hypothesis type is superiority. |
| Factorial design | Not identified; the registry design model is parallel. |
| Multiplicity | No multiplicity procedure is provided in the ClinicalTrials.gov record. |
| Interim analysis | No interim-analysis method is provided in the ClinicalTrials.gov record. |
| Missing-data/imputation method | No method is provided in the ClinicalTrials.gov record. |
| Bayesian methods | No Bayesian method is identified. |
| Stratification | Sokal score is identified for the primary CMH test. |
| Crossover | Dasatinib after crossover from imatinib is identified in the safety data. |
21. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
The primary posted analysis reports a CMH P-value of 0.005 for the 12-month MMR endpoint, using all randomized participants and Sokal-score stratification, under a superiority hypothesis.
Clinical interpretation
The ClinicalTrials.gov record establishes that MMR at 12 months was the primary clinical endpoint and that a formal treatment-group comparison was posted. It does not supply the arm-specific MMR percentages needed to quantify the magnitude of the observed difference.
22. Related Tutorials
Learn more about the methods used in this trial:
23. Related Statistical Calculators
24. Sources
- ClinicalTrials.gov: DASCERN, NCT01593254.
- Linked publication: PubMed record for PMID 32265500.
Continue through the Clinical Biostats statistical pathway
Use the related tutorials and calculators to explore the categorical-data, hypothesis-testing, and randomized-trial concepts illustrated by DASCERN.
25. Record Summary
DASCERN provides a focused example of a randomized clinical-trial analysis in which the primary endpoint is binary rather than time-to-event. The registry identifies MMR at 12 months as the primary endpoint, defines MMR as a 3-log reduction in BCR-ABL transcripts from the standardized baseline, and reports a Cochran-Mantel-Haenszel test stratified by Sokal score. The formal primary analysis uses all randomized participants and reports P = 0.005 under a superiority hypothesis.
The statistical interpretation should remain proportional to the information actually reported. The ClinicalTrials.gov record provides the test method and P-value, and specifies the Clopper-Pearson approach for 95% confidence intervals, but it does not provide the arm-specific MMR percentages or numerical confidence intervals. Those missing effect estimates mean that the magnitude and precision of the treatment difference cannot be reconstructed from the P-value alone.