This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record.
1. Trial at a Glance
ORATORIO (NCT05974839) is a completed clinical trial in patients with primary progressive multiple sclerosis evaluating the effect of ocrelizumab on cortical lesion number accumulation. The registry identifies cortical lesion number accumulation as the primary endpoint, measured using MRI between subjects on placebo and subjects who took ocrelizumab over a 5-year time frame.
| Feature | ORATORIO |
|---|---|
| Study acronym | ORATORIO |
| ClinicalTrials.gov identifier | NCT05974839 |
| Status | Completed |
| Condition | Primary Progressive Multiple Sclerosis |
| Enrollment | 732 |
| Intervention | Ocrelizumab |
| Primary endpoint | Cortical lesion number accumulation |
| Primary endpoint time frame | 5 years |
| Lead sponsor | State University of New York at Buffalo |
2. Clinical Question
The central statistical question is whether treatment with ocrelizumab is associated with a difference in cortical lesion number accumulation compared with placebo in patients with primary progressive multiple sclerosis over 5 years.
Population
Patients with primary progressive multiple sclerosis.
Intervention
Ocrelizumab.
Comparator
Placebo, as specified in the primary endpoint description.
Primary question
How does cortical lesion number accumulation measured using MRI differ between subjects on placebo and subjects who took ocrelizumab over 5 years?
3. Trial Design
The ClinicalTrials.gov record identifies ORATORIO as a completed study with an enrollment of 732 patients. The study began on 2023-10-01, and the primary completion date is listed as 2024-02-28. The lead sponsor is the State University of New York at Buffalo, classified in the record as an OTHER sponsor type.
4. Endpoints
| Endpoint | Registry definition | Time frame |
|---|---|---|
| Cortical lesion number accumulation | Cortical lesion accumulation as measured using MRI between subjects on placebo and subjects who took ocrelizumab. | 5 years |
The primary endpoint is a lesion-count outcome, rather than a conventional survival endpoint such as time to disability progression or death. The endpoint is also based on MRI assessment, making the measurement process an important part of its statistical interpretation.
What the endpoint is intended to capture
Cortical lesion number accumulation measures the accumulation of cortical lesions over the specified 5-year period. Conceptually, the analysis compares the amount of lesion accumulation observed in the ocrelizumab group with that observed in the placebo group.
The actual statistical estimand could be expressed in several ways depending on the prespecified analysis plan, including a difference in mean counts, a rate ratio, or another model-based contrast. The registry record does not specify the statistical estimand or model.
5. Planned Analysis
No formal statistical analyses were posted to ClinicalTrials.gov for the primary endpoint. The registry identifies what will be measured, but it does not provide an estimated treatment effect, confidence interval, or p-value for cortical lesion number accumulation.
Primary endpoint
Measured using MRI between subjects on placebo and subjects who took ocrelizumab
Time frame: 5 years
For an endpoint defined as the number of lesions accumulated over a fixed observation period, an appropriate statistical analysis would ordinarily depend on how the MRI measurements were collected and how the investigators defined the estimand. Count-data models such as Poisson or negative-binomial regression can be considered when the outcome is a count and the distributional assumptions are appropriate. If repeated MRI measurements contribute to the endpoint, longitudinal models may be appropriate because observations from the same patient are correlated.
The choice between these approaches is not interchangeable. A simple comparison of mean lesion counts, a count-regression rate ratio, and a longitudinal model answer somewhat different questions. The analysis should therefore be tied to the prespecified definition of cortical lesion accumulation and the timing and structure of the MRI assessments.
6. Statistical Methodology
Count outcomes require attention to the distribution
Cortical lesion number accumulation is a count. Counts can have distributions that differ substantially from the normal distribution, particularly when many observations have small values or when variability increases with the mean. A statistical model should therefore reflect the observed distribution rather than automatically applying a standard normal-theory comparison.
In a simple Poisson model, the exponentiated treatment coefficient can be interpreted as a rate ratio when the model represents event or lesion accumulation rates.
Overdispersion and negative-binomial models
A Poisson model assumes that the conditional variance is equal to the conditional mean. Count data can instead exhibit overdispersion, in which the variance is larger than the mean. A negative-binomial model introduces an additional dispersion parameter and can be useful when that assumption is more realistic.
A rate ratio below 1 would indicate a lower modeled lesion-accumulation rate in the ocrelizumab group relative to placebo; a rate ratio above 1 would indicate a higher modeled rate.
Longitudinal MRI measurements
If cortical lesion counts were assessed repeatedly during follow-up, the observations from an individual patient would not be statistically independent. Repeated measurements from the same person tend to be correlated, so an analysis that treats every MRI measurement as an independent observation could understate uncertainty.
Longitudinal generalized linear mixed models or generalized estimating equations are examples of approaches that can account for repeated observations. The appropriate method would depend on the actual measurement schedule, endpoint construction, distribution of the counts, and prespecified estimand.
Exposure time and accumulation rates
For an accumulation endpoint, follow-up duration can matter. If patients contribute different amounts of observable follow-up, an analysis based on an accumulation rate can incorporate time at risk or observation time. If every patient contributes a common fixed period to the primary endpoint, a direct comparison of the prespecified 5-year outcome may instead be appropriate.
MRI measurement is part of the endpoint
The endpoint depends on MRI measurement of cortical lesions. Consequently, statistical interpretation involves not only the comparison between treatment groups but also the reliability and consistency of lesion identification and counting. Differences in image acquisition, lesion classification, or assessment procedures can affect the measured endpoint even when the underlying biological process is unchanged.
7. Statistical Methods Explained
What kind of statistical variable is cortical lesion number accumulation?
It is fundamentally a count outcome: the endpoint records the number of cortical lesions accumulated over the specified period. Count outcomes are often modeled differently from continuous measurements because their distributions are constrained to nonnegative integer values.
Why might a Poisson model be considered?
Poisson regression is designed for count outcomes and models the expected count through a logarithmic link. If its assumptions are reasonable, the treatment coefficient can be translated into a multiplicative comparison between groups.
Why might a negative-binomial model be preferred?
Real-world count data frequently show more variability than a Poisson distribution permits. If the cortical lesion counts are overdispersed, a negative-binomial model can provide a more appropriate variance structure.
Why does repeated MRI assessment matter statistically?
If multiple MRI observations from the same patient contribute to the analysis, those observations are correlated. Treating them as independent can make standard errors too small. Longitudinal methods explicitly account for within-patient dependence.
What would a rate ratio mean?
A rate ratio compares modeled lesion-accumulation rates between the treatment groups. For example, a rate ratio of 0.70 would correspond to a modeled accumulation rate 30% lower in the treatment group than in the control group, assuming that rate ratio is the prespecified estimand and that the model is appropriate. This is an illustration of interpretation, not a reported result.
Why is the endpoint different from a hazard ratio?
A hazard ratio is ordinarily used for a time-to-event outcome, where the event is something such as progression or death and the analysis considers the timing of that event. Cortical lesion number accumulation is instead a count-based MRI endpoint. Unless the endpoint were explicitly redefined as time to development of a lesion, a hazard ratio would not be the natural primary measure of effect.
Why can the statistical model change the interpretation?
A mean difference, rate ratio, odds ratio, and longitudinal model coefficient do not represent the same estimand. The statistical analysis must therefore follow the precise endpoint definition. Because the registry record does not specify the model or estimand, a numerical treatment effect cannot be reconstructed from the endpoint description alone.
8. Interpreting the Absence of Posted Results
The registry identifies ORATORIO as completed, but no formal statistical analyses are posted for the primary endpoint in the ClinicalTrials.gov record. Completion of a study and availability of a statistical result are separate concepts: a study can have reached its listed completion milestone without the registry containing a numerical analysis of its primary outcome.
What is reported
The registry reports the condition, intervention, enrollment, study dates, status, and the definition and 5-year time frame of the primary endpoint.
What is not reported
The record does not provide a primary-endpoint treatment estimate, confidence interval, or p-value.
This distinction is important statistically. A completed trial does not, by itself, establish the direction or magnitude of an endpoint difference. Those conclusions require the actual prespecified analysis and its resulting estimates.
9. Limitations
- No posted primary analysis: the registry does not provide a numerical statistical analysis of cortical lesion number accumulation.
- Unspecified estimand: the record defines the endpoint but does not specify whether the principal comparison is a mean difference, rate ratio, model-based change, or another estimand.
- Unspecified statistical model: the registry does not identify the count-data or longitudinal model used for the endpoint.
- MRI measurement: cortical lesion accumulation depends on the detection and classification of lesions using MRI, so measurement procedures are relevant to interpretation.
- Potential repeated observations: if multiple MRI assessments contribute to the endpoint, within-patient correlation needs to be incorporated into the analysis.
- Count-data assumptions: Poisson-type models require distributional assumptions that may not hold if lesion counts are highly variable or overdispersed.
- Limited design information: the available registry information does not specify additional characteristics such as randomization ratio, masking, stratification, or analysis populations.
10. Why This Trial Matters Statistically
ORATORIO provides a useful teaching example because its primary outcome is different from the time-to-event endpoints that dominate many clinical-trial analyses. Instead of asking only whether and when a clinical event occurs, the primary endpoint asks about the accumulation of a measurable biological lesion burden over a defined period.
| Concept | How it appears in ORATORIO |
|---|---|
| Count outcome | Cortical lesion number accumulation is expressed as a lesion count. |
| Between-group comparison | The registry describes MRI-measured cortical lesion accumulation between placebo and ocrelizumab groups. |
| Longitudinal measurement | MRI assessment over a 5-year endpoint time frame creates the possibility of repeated measurements contributing information about accumulation. |
| Count regression | Poisson or negative-binomial approaches are examples of models that can be considered for count outcomes when appropriate. |
| Overdispersion | Excess variability relative to the Poisson assumption can motivate a negative-binomial approach. |
| Within-patient correlation | Repeated MRI observations from the same patient require methods that account for dependence if they are analyzed longitudinally. |
| MRI measurement | The endpoint depends on consistent identification and counting of cortical lesions. |
| Estimand selection | The interpretation depends on whether the prespecified analysis targets a count difference, accumulation rate, or another treatment contrast. |
| Registry results reporting | The study is listed as completed, but no formal statistical analyses are posted for the primary endpoint. |
11. Statistical Interpretation of the Primary Endpoint
The registry defines the primary endpoint as cortical lesion number accumulation, measured using MRI between subjects on placebo and subjects who took ocrelizumab over 5 years. The endpoint therefore focuses on the accumulation of cortical lesions rather than directly measuring a time-to-event outcome.
A formal treatment effect would quantify the difference in cortical lesion accumulation between the ocrelizumab and placebo groups according to the prespecified statistical estimand. The effect estimate should be accompanied by a measure of uncertainty, such as a confidence interval, so that the magnitude and precision of the comparison can be evaluated together.
Even if a p-value were available, it would address evidence against a specified null hypothesis rather than describe the magnitude of the lesion-accumulation difference. A complete statistical interpretation would consider the estimated effect, its confidence interval, the analysis population, endpoint definition, and assumptions of the statistical model.
12. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
The primary endpoint is a 5-year MRI-based comparison of cortical lesion number accumulation between placebo and ocrelizumab groups. The registry does not report the resulting numerical treatment effect or formal statistical test.
Clinical interpretation
The endpoint is intended to characterize cortical lesion accumulation in patients with primary progressive multiple sclerosis. Its clinical meaning depends on how the MRI lesion measure relates to disease biology and clinical outcomes, which cannot be established from the endpoint definition alone.
13. What Would a Complete Statistical Result Contain?
For a count-based primary endpoint such as cortical lesion number accumulation, a complete statistical report would ordinarily identify the analysis population, exact endpoint construction, treatment-effect estimand, statistical model, uncertainty measure, and hypothesis-testing procedure.
| Component | Purpose |
|---|---|
| Endpoint definition | Specifies exactly how cortical lesion accumulation is calculated. |
| Analysis population | Defines which enrolled or treated patients contribute to the primary analysis. |
| Effect measure | Quantifies the difference or ratio between treatment groups. |
| Confidence interval | Describes statistical uncertainty around the estimated effect. |
| Statistical model | Defines how the count data and its variability are represented. |
| Covariates | Identifies any prespecified adjustment variables. |
| Handling of repeated MRI measurements | Accounts for within-patient correlation when applicable. |
| Missing-data strategy | Defines how incomplete MRI follow-up affects the primary analysis. |
| Multiplicity strategy | Explains how type I error is controlled when multiple formal hypotheses are tested. |
These components matter because the numerical value of a treatment effect is meaningful only in relation to the estimand and analysis that generated it. Two analyses of the same underlying MRI observations can produce different estimates if they answer different statistical questions.
14. Trial Timeline
Study start
The ClinicalTrials.gov record lists October 1, 2023 as the study start date.
Primary completion
The registry lists February 28, 2024 as the primary completion date.
Current study status
The ClinicalTrials.gov record lists the ORATORIO study status as completed.
15. Sources
- ClinicalTrials.gov: ORATORIO (NCT05974839).
Continue through the Clinical Biostats knowledge graph
Explore the broader Clinical Biostats collection of clinical-trial analyses, statistical tutorials, and analytical tools.
16. Record Summary
ORATORIO is a completed clinical trial enrolling 732 patients with primary progressive multiple sclerosis and evaluating ocrelizumab. Its registered primary endpoint is cortical lesion number accumulation, measured using MRI between subjects on placebo and subjects who took ocrelizumab over a 5-year time frame. Statistically, the endpoint presents a count-data problem in which the appropriate analysis depends on the precise definition of accumulation, MRI assessment schedule, distribution of lesion counts, and prespecified treatment-effect estimand.
The registry provides the core endpoint definition and study information but does not post a formal statistical analysis of the primary endpoint. Consequently, the central statistical lesson is methodological: cortical lesion accumulation should be analyzed according to its count structure and, when applicable, its longitudinal measurement structure, with the treatment effect interpreted together with its uncertainty and the assumptions underlying the selected model.