This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. View NCT00548405 on ClinicalTrials.gov.
1. Trial at a Glance
CARE-MS II was a randomized, parallel, single-masked phase 3 trial in relapsing-remitting multiple sclerosis. The registry reports 840 participants and three intervention arms: alemtuzumab 12 mg, alemtuzumab 24 mg, and interferon beta-1a. The posted primary efficacy analyses compare interferon beta-1a with alemtuzumab 12 mg.
| Feature | CARE-MS II |
|---|---|
| Trial name | CARE-MS II |
| Brief title | Comparison of Alemtuzumab and Rebif® Efficacy in Multiple Sclerosis, Study Two |
| Phase | Phase 3 |
| Condition | Multiple Sclerosis, Relapsing-Remitting |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Single |
| Primary purpose | Treatment |
| Enrollment | 840 |
| Interventions | Alemtuzumab 12 mg; Alemtuzumab 24 mg; Interferon beta-1a |
| Primary endpoints | Percentage of Participants With Sustained Accumulation of Disability (SAD); Annualized Relapse Rate |
| Primary endpoint types | Binary; Count / rate |
| Results posted | Yes |
| Statistical analyses posted | 6 |
| Lead sponsor | Genzyme, a Sanofi Company |
2. Clinical Question
The primary statistical question was whether alemtuzumab 12 mg differed from interferon beta-1a on two prespecified efficacy endpoints through 2 years: sustained accumulation of disability and annualized relapse rate.
Population
Participants with Multiple Sclerosis, Relapsing-Remitting enrolled in the phase 3 CARE-MS II trial.
Interventions
The trial included alemtuzumab 12 mg, alemtuzumab 24 mg, and interferon beta-1a.
Primary comparison
The posted primary efficacy analyses compare Interferon Beta-1a with Alemtuzumab 12 mg.
Primary question
Does alemtuzumab 12 mg improve the prespecified disability and relapse outcomes relative to interferon beta-1a over up to 2 years?
3. Trial Design
Alemtuzumab 12 mg
- Biological intervention
- Primary efficacy analyses were performed against interferon beta-1a
- Included in the FAS population used for the posted primary analyses
Alemtuzumab 24 mg
- Biological intervention
- Recruitment to this arm was closed early
- Efficacy analysis was not performed for this arm
Interferon beta-1a
- Biological intervention
- Comparator in the posted primary efficacy analyses
4. Analysis Population
The primary sustained-accumulation-of-disability analysis used the full analysis set (FAS), defined in the registry as all randomized participants who received at least 1 dose of study drug as per initial randomization. The annualized relapse rate analysis also used the FAS population.
| Analysis population | Role in the registry analyses |
|---|---|
| FAS | Primary efficacy population. Includes all randomized participants who received at least 1 dose of study drug as per initial randomization. |
| EDSS subset | For the Year 2 EDSS analysis, the analyzed population was the FAS subset with EDSS assessments at both baseline and end-of-study. |
| MSFC subset | For the Year 2 MSFC analysis, the analyzed population was a subset of the FAS with MSFC assessment at baseline, as described in the registry analysis. |
| T2 lesion-volume subset | For the Year 2 MRI-T2 analysis, the analyzed population was a subset of the FAS with T2 volume assessments at both baseline and end-of-study. |
5. Primary Endpoints
| Endpoint | Time frame | Registry definition / analysis |
|---|---|---|
| Percentage of Participants With Sustained Accumulation of Disability (SAD) | Up to 2 years | EDSS is an ordinal scale in half-point increments that qualifies disability in participants with MS. It assesses 7 functional systems as well as ambulation. EDSS total score ranges from 0, normal neurological examination, to 10, death due to MS. SAD was defined as an increase of at least 1.5 points for the registered endpoint. Primary analysis: Cox proportional-hazards regression. |
| Annualized Relapse Rate | Up to 2 years | Relapse was defined as new neurological symptoms or worsening of previous neurological symptoms with an objective change on neurological examination, attributable to multiple sclerosis, lasting at least 48 hours, present at normal body temperature, and preceded by at least 30 days of clinical stability. Annualized relapse rate was estimated through negative binomial regression in the registry outcome description; the posted statistical analysis reports proportional means regression with robust variance estimation and covariate adjustment for geographic region. |
6. Statistical Methodology
Time-to-event analysis for sustained disability
Sustained accumulation of disability is analyzed as a time-to-event endpoint because participants can reach the defined disability threshold at different times, while participants who do not reach it during observation do not necessarily provide an observed event time. The posted analysis used a Cox proportional-hazards regression model.
The registry reports a Cox proportional-hazards regression model with robust variance estimation using treatment group and geographic region as covariates.
Rate analysis for relapses
Annualized relapse rate is a count/rate endpoint rather than a single binary event. The registry outcome description states that the annualized relapse rate was estimated through negative binomial regression. The posted statistical analysis reports proportional means regression with robust variance estimation and covariate adjustment for geographic region, with the treatment effect expressed as a rate ratio.
The registry-reported analysis reports robust variance estimation and adjustment for geographic region. The resulting rate ratio compares relapse rates rather than time until a first relapse.
Covariate adjustment
Geographic region appears explicitly in both primary analysis models. For SAD, the registry states that treatment group and geographic region were included in the Cox model. For annualized relapse rate, geographic region was included as a covariate in the proportional means regression model.
Repeated-measures analyses
Two secondary endpoints evaluated change from baseline at Year 2: EDSS score and MSFC score. The analysis text reports the Wei-Lachin method for non-parametric analysis of repeated measures. The registry reports the Wei-Lachin method for these analyses.
Ranked ANCOVA for MRI-T2 lesion volume
The MRI-T2 endpoint used a ranked ANCOVA. The model was adjusted for geographic region and baseline T2 lesion volume. This illustrates an important distinction between an endpoint's clinical scale and the statistical method used to compare treatment groups: ranking can be used when the raw outcome does not meet the assumptions preferred for a conventional linear-model analysis.
7. Primary Results: Sustained Accumulation of Disability
The first primary endpoint was the percentage of participants with sustained accumulation of disability through up to 2 years. The posted analysis compared Interferon Beta-1a with Alemtuzumab 12 mg in the FAS population using Cox proportional-hazards regression.
Hazard ratio for sustained accumulation of disability
95% CI: 0.38–0.87 · P = 0.0084
Analysis: Cox proportional-hazards regression with robust variance estimation and geographic-region covariate adjustment.
A hazard ratio of 0.58 means that, under the fitted Cox model, the estimated instantaneous hazard of reaching the defined sustained-disability event was 42% lower for Alemtuzumab 12 mg relative to Interferon Beta-1a, because \(1-0.58=0.42\).
The HR does not mean that 42% of participants avoided disability, nor does it mean that every participant experienced a 42% reduction in individual risk. It is a relative, model-based time-to-event measure.
The 95% CI of 0.38–0.87 describes uncertainty around the estimated hazard ratio under the analysis framework. It does not describe the range of individual treatment effects across patients.
The P-value of 0.0084 addresses evidence against the relevant null hypothesis under the prespecified superiority framework; it does not measure the magnitude or clinical importance of the treatment effect. The HR and its confidence interval provide that effect-size information.
Because this is a Cox-model result, interpretation also depends on the proportional-hazards framework. A single HR is most straightforward when the relative hazards are reasonably represented by a common ratio over time.
8. Primary Results: Annualized Relapse Rate
The second primary endpoint was annualized relapse rate through up to 2 years. The posted analysis compared Interferon Beta-1a with Alemtuzumab 12 mg using proportional means regression with robust variance estimation and adjustment for geographic region.
Rate ratio for annualized relapse rate
95% CI: 0.39–0.65 · P < 0.0001
Effect measure: Rate ratio
A rate ratio of 0.51 means that the estimated relapse rate under Alemtuzumab 12 mg was 51% of the corresponding rate under Interferon Beta-1a in the posted model. Equivalently, this corresponds to an estimated 49% lower relapse rate because \(1-0.51=0.49\).
The rate ratio does not mean that 49% of participants had no relapses, and it is not the same measure as the hazard ratio for sustained disability. The two endpoints summarize different aspects of disease activity.
The 95% CI of 0.39–0.65 quantifies uncertainty around the rate-ratio estimate. Because the interval is relatively narrow compared with the full range of possible rate ratios, it provides a more precise statistical description of the estimated relative relapse rate than the point estimate alone.
The P-value < 0.0001 measures evidence against the relevant null hypothesis under the statistical testing framework. It is not a measure of how large the rate reduction is; the rate ratio itself describes the estimated magnitude.
The analysis uses robust variance estimation and geographic-region adjustment. Interpretation therefore belongs to the fitted regression framework rather than to a simple unadjusted comparison of raw relapse counts.
| Primary endpoint | Comparison | Method | Effect | 95% CI | P-value |
|---|---|---|---|---|---|
| Sustained Accumulation of Disability | Interferon Beta-1a vs Alemtuzumab 12 mg | Cox proportional-hazards regression | HR 0.58 | 0.38–0.87 | 0.0084 |
| Annualized Relapse Rate | Interferon Beta-1a vs Alemtuzumab 12 mg | Proportional means regression | Rate ratio 0.51 | 0.39–0.65 | <0.0001 |
9. Secondary Endpoint Results
The ClinicalTrials.gov record contains four secondary statistical analyses in addition to the two primary analyses. These results are reported below without adding effect estimates or confidence intervals that are not contained in the ClinicalTrials.gov record.
| Secondary endpoint | Time frame | Method | Reported result |
|---|---|---|---|
| Percentage of Participants Who Were Relapse Free at Year 2 | Year 2 | Cox proportional-hazards regression | HR 0.53; 95% CI 0.41–0.69; P < 0.0001 |
| Change From Baseline in Expanded Disability Status Scale (EDSS) Score at Year 2 | Baseline, Year 2 | Wei-Lachin method for non-parametric analysis of repeated measures | P < 0.0001 |
| Change From Baseline in Multiple Sclerosis Functional Composite (MSFC) Score at Year 2 | Baseline, Year 2 | Wei-Lachin method for non-parametric analysis of repeated measures | P = 0.0022 |
| Percent Change From Baseline in Magnetic Resonance Imaging Time Constant 2 (MRI-T2) Hyperintense Lesion Volume at Year 2 | Baseline, Year 2 | Ranked ANCOVA | P = 0.1371 |
Relapse-free status at Year 2
Hazard ratio for the relapse-free endpoint
95% CI: 0.41–0.69 · P < 0.0001
This secondary endpoint again uses a time-to-event framework, with the posted analysis using Cox proportional-hazards regression, robust variance estimation, and geographic-region adjustment. The hazard ratio should be interpreted within that model rather than as a simple difference in the percentage of participants who were relapse free.
EDSS change at Year 2
The analysis of change from baseline in EDSS score at Year 2 used the Wei-Lachin method for non-parametric analysis of repeated measures and reported P < 0.0001. The analyzed population was a subset of the FAS with EDSS assessment at both baseline and end-of-study.
MSFC change at Year 2
The analysis of change from baseline in MSFC score at Year 2 used the Wei-Lachin method and reported P = 0.0022. The endpoint is expressed as a Z-score in the registry data.
MRI-T2 lesion-volume change
Percent change from baseline in MRI-T2 hyperintense lesion volume at Year 2 was analyzed using ranked ANCOVA, with adjustment for geographic region and baseline T2 lesion volume. The posted analysis reported P = 0.1371.
10. Serious Adverse Events
The ClinicalTrials.gov record reports serious adverse events by arm as affected participants over participants at risk. These counts should be interpreted as safety summaries rather than as efficacy comparisons.
| Arm | Affected / at risk |
|---|---|
| Interferon Beta-1a | 44/202 |
| Alemtuzumab 12 mg | 85/435 |
| Alemtuzumab 24 mg | 30/161 |
| Alemtuzumab (Pooled) | 115/596 |
The pooled alemtuzumab count combines the two alemtuzumab arms as reported in the ClinicalTrials.gov record. The underlying denominators differ across the arms, so raw affected-participant counts alone are not appropriate for comparing the frequency of serious adverse events.
11. Statistical Methods Explained
Why was a Cox proportional-hazards model used for sustained accumulation of disability?
SAD is a time-to-event endpoint: participants can experience the defined disability event at different points during follow-up, while others may remain event-free through their available observation. A Cox model uses this timing information and accommodates right-censoring. The CARE-MS II analysis also incorporated geographic region as a covariate and used robust variance estimation.
What does an HR of 0.58 mean?
Within the fitted Cox model, an HR of 0.58 corresponds to an estimated instantaneous event hazard that is 58% of the comparator hazard. It can therefore be described as an estimated 42% lower hazard. It does not mean that 42% of participants avoided the event or that every patient's individual risk was reduced by exactly 42%.
Why is the relapse endpoint expressed as a rate ratio?
Annualized relapse rate concerns the frequency of qualifying relapse events over participant-time rather than simply whether a participant ever relapsed. A rate ratio compares estimated rates between treatment groups. The posted CARE-MS II analysis used proportional means regression with robust variance estimation and geographic-region adjustment.
What does a rate ratio of 0.51 mean?
A rate ratio of 0.51 indicates that the estimated relapse rate under Alemtuzumab 12 mg was 51% of the rate under Interferon Beta-1a in the fitted analysis. The complementary statement is an estimated 49% lower rate. This is different from saying that 49% of patients were protected from relapse.
Why adjust for geographic region?
Geographic region can be incorporated as a covariate to account for systematic regional differences within the statistical model. In CARE-MS II, the registry analysis explicitly identifies geographic region as a covariate in both primary models. Adjustment can improve the precision and specification of the treatment comparison when the covariate contributes relevant variation.
Why use Wei-Lachin for EDSS and MSFC?
The registry analysis reports the Wei-Lachin method for non-parametric analysis of repeated measures. These endpoints involve measurements at baseline and Year 2, so the analysis must account for repeated observations rather than treating every measurement as independent.
Why was ranked ANCOVA used for MRI-T2 lesion volume?
The registry reports ranked ANCOVA with adjustment for geographic region and baseline T2 lesion volume. ANCOVA incorporates baseline information while comparing follow-up outcomes, and ranking provides a non-parametric approach when the raw outcome scale is not well suited to standard model assumptions. The posted analysis reported a P-value of 0.1371 but no effect estimate or confidence interval in the ClinicalTrials.gov record.
12. Confidence Intervals and P-values
Confidence intervals
For the two primary endpoints, the registry supplies 95% two-sided confidence intervals. The SAD HR is 0.58 with a 95% CI of 0.38–0.87; the relapse rate ratio is 0.51 with a 95% CI of 0.39–0.65.
P-values
The primary P-values are 0.0084 for SAD and <0.0001 for annualized relapse rate. These quantify evidence under the relevant statistical testing framework rather than effect magnitude.
These three quantities are related but answer different questions. A complete interpretation should not substitute a P-value for the treatment-effect estimate or its uncertainty interval.
13. Primary Analysis Logic
| Question | Statistical representation |
|---|---|
| Did the time to sustained disability differ? | Cox proportional-hazards model; hazard ratio 0.58; 95% CI 0.38–0.87; P = 0.0084. |
| Did relapse frequency differ? | Proportional means regression; rate ratio 0.51; 95% CI 0.39–0.65; P < 0.0001. |
| Was geographic region incorporated? | Yes. It was included as a covariate in both primary analyses. |
| Was robust variance estimation used? | Yes. It was reported for both primary analyses. |
| Was the Alemtuzumab 24 mg arm included in the primary efficacy comparison? | No. The ClinicalTrials.gov record states that efficacy analysis was not performed for this arm because recruitment was closed early. |
14. What the Primary Results Do — and Do Not — Establish
The HR of 0.58 provides a relative time-to-event estimate for the defined SAD endpoint. Its 95% CI of 0.38–0.87 describes uncertainty around that estimate. The result does not provide an absolute probability of disability at a particular time because the registry analysis does not provide such a survival probability.
The rate ratio of 0.51 describes the relative relapse rate estimated by the posted regression analysis. Its 95% CI of 0.39–0.65 describes uncertainty around the rate-ratio estimate. It does not establish that every participant experienced the same proportional change in relapse frequency.
The primary P-values of 0.0084 and <0.0001 provide evidence against their respective null hypotheses under the superiority testing framework. They should not be read as probabilities that the null hypothesis is true, probabilities that treatment is effective, or numerical measures of clinical importance.
15. Multiplicity, Superiority, and the Two Primary Endpoints
the ClinicalTrials.gov record identifies the primary hypothesis type as superiority and identifies two registered primary endpoints. The ClinicalTrials.gov record does not provide an alpha-allocation scheme, multiplicity procedure, or formal hierarchy between these two primary endpoints.
| Feature | Supported interpretation |
|---|---|
| Primary endpoint count | 2 |
| Hypothesis type | Superiority |
| Primary endpoint 1 | Sustained Accumulation of Disability through up to 2 years |
| Primary endpoint 2 | Annualized Relapse Rate through up to 2 years |
| Multiplicity procedure | Not specified in the ClinicalTrials.gov record |
| Alpha allocation | Not specified in the ClinicalTrials.gov record |
Because the ClinicalTrials.gov record does not specify how multiplicity across the two primary endpoints was handled, the reported P-values should be presented exactly as registered rather than applying an invented adjustment. The fact that both primary analyses have small reported P-values is descriptive of the posted analyses; it does not supply information about an unreported multiplicity strategy.
16. Missing Data, Censoring, and Analysis-Set Considerations
Time-to-event analysis naturally incorporates participants who are censored before experiencing the event, provided the censoring assumptions are appropriate. The ClinicalTrials.gov record identifies the FAS and several endpoint-specific subsets but do not provide a separate missing-data or imputation strategy.
FAS definition
The FAS includes randomized participants who received at least 1 dose of study drug as per initial randomization.
EDSS subset
The Year 2 EDSS analysis used FAS participants with EDSS assessments at both baseline and end-of-study.
MSFC subset
The Year 2 MSFC analysis used a subset of the FAS with the relevant MSFC assessments.
MRI-T2 subset
The Year 2 MRI-T2 analysis used FAS participants with T2 volume assessments at both baseline and end-of-study.
17. Interpretation of the Alemtuzumab 24 mg Arm
The presence of three intervention arms does not mean that the posted efficacy results support a three-way statistical comparison. The ClinicalTrials.gov record explicitly state that efficacy analysis was not performed for Alemtuzumab 24 mg because recruitment to that arm was closed early.
| Arm | Included in posted primary efficacy comparison? | Reason / role |
|---|---|---|
| Alemtuzumab 12 mg | Yes | Compared with Interferon Beta-1a for the primary efficacy analyses. |
| Alemtuzumab 24 mg | No | Recruitment was closed early; efficacy analysis was not performed. |
| Interferon Beta-1a | Yes | Comparator in the posted primary efficacy analyses. |
This distinction is statistically important. An intervention can be part of the randomized study design without contributing to a particular reported efficacy estimate. The correct interpretation of the primary HR and rate ratio is therefore the comparison specified in the analysis record: Interferon Beta-1a versus Alemtuzumab 12 mg.
18. Design Features That Shape Interpretation
Randomization
Randomized allocation provides the structural basis for comparing treatment groups while reducing systematic allocation differences.
Parallel design
Participants were assigned to parallel intervention groups rather than repeatedly crossing between treatments as part of the study design.
Single masking
The registry classifies the study as single masked. The ClinicalTrials.gov record does not specify the masked party.
Early arm closure
Early closure of the Alemtuzumab 24 mg recruitment arm directly affects which treatment comparisons were formally analyzed.
19. Why This Trial Matters Statistically
CARE-MS II is a useful teaching case because it combines several common clinical-trial analysis problems within one randomized phase 3 study: a time-to-event disability endpoint, a recurrent-event rate endpoint, covariate adjustment, robust variance estimation, repeated-measures analyses, ranked ANCOVA, endpoint-specific analysis populations, and an intervention arm for which efficacy analysis was not performed.
| Concept | How it appears in CARE-MS II |
|---|---|
| Randomization | Randomized parallel-group phase 3 design. |
| Time-to-event endpoint | Sustained accumulation of disability through up to 2 years. |
| Hazard ratio | Primary SAD effect measure: HR 0.58 with 95% CI 0.38–0.87. |
| Cox model | Primary SAD analysis and secondary relapse-free analysis. |
| Rate ratio | Primary annualized relapse-rate effect measure: 0.51 with 95% CI 0.39–0.65. |
| Robust variance | Used in both primary models. |
| Covariate adjustment | Geographic region was included in the primary analyses. |
| Repeated measures | Wei-Lachin analysis for Year 2 EDSS and MSFC changes. |
| ANCOVA | Ranked ANCOVA for MRI-T2 lesion-volume change. |
| Analysis populations | FAS for primary efficacy, with endpoint-specific subsets for several secondary outcomes. |
| Design qualification | Alemtuzumab 24 mg efficacy analysis was not performed after early recruitment closure. |
20. Limitations
- Alemtuzumab 24 mg efficacy: the ClinicalTrials.gov record explicitly state that efficacy analysis was not performed for this arm because recruitment was closed early.
- Incomplete effect reporting for secondary endpoints: the registry analyses provide P-values for EDSS, MSFC, and MRI-T2 endpoints without corresponding effect estimates and confidence intervals.
- Analysis-population subsets: several Year 2 continuous endpoints use subsets of the FAS requiring relevant assessments at baseline and/or follow-up.
- Multiplicity information: the ClinicalTrials.gov record identifies two primary endpoints and superiority testing but do not specify an alpha-allocation or multiplicity procedure.
- Missing-data strategy: the ClinicalTrials.gov record does not specify an imputation method, so no particular imputation approach should be attributed to the trial.
- Cox-model assumptions: hazard-ratio interpretation relies on the proportional-hazards framework. The ClinicalTrials.gov record does not provide a diagnostic assessment of that assumption.
- Rate-model interpretation: the annualized relapse result is a model-based rate ratio, not an individual-level probability or a first-event hazard ratio.
- Registry-level evidence: this page is limited to the ClinicalTrials.gov record and does not supplement it with additional numerical results from publications.
21. A Statistical Reading of the Trial
The most informative way to read the CARE-MS II results is to keep the endpoint types separate. The SAD analysis asks about the relative timing of a defined disability event and produces a hazard ratio. The annualized relapse analysis asks about relapse frequency and produces a rate ratio. Although both estimates are below 1, they are not interchangeable statistics.
The two primary estimates describe different outcome processes. The first is a time-to-event relative hazard; the second is a relative event rate. Their confidence intervals and P-values should be interpreted within their respective models.
The secondary results broaden the analysis. Relapse-free status at Year 2 was also analyzed with a Cox model and produced an HR of 0.53 with a 95% CI of 0.41–0.69 and P < 0.0001. EDSS and MSFC change were evaluated with a repeated-measures method, while MRI-T2 lesion-volume change was evaluated using ranked ANCOVA.
Finally, the early closure of the Alemtuzumab 24 mg recruitment arm is not a minor administrative detail. It changes the scope of the efficacy evidence: the registry-reported primary analyses support a specific comparison between Interferon Beta-1a and Alemtuzumab 12 mg, rather than a three-arm efficacy conclusion.
22. Statistical Concepts in This Trial
Learn more about the methods used in this trial:
23. Related Statistical Calculators
24. Sources
- ClinicalTrials.gov: CARE-MS II, NCT00548405.
- Linked publication: PubMed record — PMID 39935588.
- Linked publication: PubMed record — PMID 37745914.
- Linked publication: PubMed record — PMID 37272540.
- Linked publication: PubMed record — PMID 36619856.
- Linked publication: PubMed record — PMID 34882037.
Continue through the Clinical Biostats statistical pathway
Explore the statistical methods behind randomized trials, survival analysis, repeated measures, regression, and treatment-effect estimation.
25. Record Summary
CARE-MS II is a phase 3 randomized parallel-group trial in relapsing-remitting multiple sclerosis with 840 enrolled participants and three intervention arms. The ClinicalTrials.gov analyses report two primary efficacy comparisons between Interferon Beta-1a and Alemtuzumab 12 mg: a Cox-model hazard ratio of 0.58 for sustained accumulation of disability, with a 95% CI of 0.38–0.87 and P = 0.0084; and a proportional-means-regression rate ratio of 0.51 for annualized relapse rate, with a 95% CI of 0.39–0.65 and P < 0.0001.
The secondary analyses extend the statistical picture to relapse-free status, EDSS change, MSFC change, and MRI-T2 lesion-volume change. The methods include Cox proportional-hazards regression, Wei-Lachin repeated-measures analysis, and ranked ANCOVA. The serious-adverse-event summaries are reported separately by treatment arm. Most importantly for interpretation, the ClinicalTrials.gov record states that efficacy analysis was not performed for the Alemtuzumab 24 mg arm after its recruitment was closed early.