Revlimid in Transfusion Dependent Patients: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Revlimid in Transfusion Dependent Patients (NCT01178814), a na phase 2 study. The primary outcome is Response Signature in Transfusion Dependent MDS Patients.

2. Clinical Question

The registry describes a prospective trial of Revlimid in patients with MDS who depend on transfusions. Its stated focus is not simply whether participants meet a response definition: the study aims to test whether expression of specific genes can help predict which patients respond to the study drug. The primary outcome is titled “Response Signature in Transfusion Dependent MDS Patients” and is recorded as a count of participants, with a timeframe of up to 12 weeks from baseline.

The registry’s outcome description also gives response criteria for target lesions assessed by MRI using RECIST v1.0. Complete response is defined as disappearance of all target lesions; partial response is defined as at least a 30% decrease in the sum of the longest diameter of target lesions. The record does not report how the response-signature result was categorized or how it was linked to these criteria. It also does not report the primary outcome result, so the registry record cannot establish how many participants met a response definition or how well the signature predicted response.

3. Trial Design

This is a prospective, single-group phase 2 trial sponsored by Columbia University. The registry lists an enrollment of 57 participants and describes the population as patients with transfusion-dependent, non-del(5q), low or intermediate-1 risk MDS. The study status is completed. Its listed start date is 2009-06-01 and its primary completion date is 2022-09-01.

Participants received Revlimid (lenalidomide), described as an oral capsule taken once a day. Masking is listed as none, and allocation is listed as not applicable. The registry identifies no control arm, so the study does not provide a randomized, like-for-like comparison of Revlimid with another treatment or with no treatment. Results from the single group therefore need to be read as observations within that group, not as a treatment contrast.

The study’s stated purpose was to test whether expression of specific genes could predict which patients would respond to Revlimid. The primary outcome is described as testing the response signature in the specified MDS population, confirming its predictive value, and establishing z-score boundaries that could be used to pre-select patients in future clinical studies. The registry does not report details of how the signature was measured, how its boundaries were set, or how prediction performance was evaluated.

The registry record does not report a sample-size justification, detailed eligibility criteria, dose modifications, treatment duration beyond the stated daily administration, or the statistical analysis plan. The single-group design and absence of masking are relevant context for interpreting any observed outcomes: without a concurrent comparison group, changes or responses cannot be separated from other influences using a randomized contrast. The record does not report whether any other design features addressed that limitation.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Overall Study

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

6. Primary Endpoint

The registry lists the primary outcome as Response Signature in Transfusion Dependent MDS Patients, measured over Up to 12 weeks from baseline.

The primary outcome is titled “Response Signature in Transfusion Dependent MDS Patients.” It is recorded as a count of participants, with a timeframe of up to 12 weeks from baseline. The study description says the trial aimed to test the Revlimid response signature in patients with transfusion-dependent, non-del(5q), low and intermediate-1 risk MDS, confirm the signature’s predictive value, and establish z-score boundaries for possible use in selecting patients for future clinical studies.

The registry also describes response criteria for target lesions assessed by MRI under RECIST v1.0. Complete response means disappearance of all target lesions. Partial response means at least a 30% decrease in the sum of the longest diameter of target lesions. These definitions describe response categories, but the record does not explain how they were incorporated into the response-signature endpoint or how the participant count was determined.

In general, a participant count can describe how many people meet a defined outcome, while a predictive signature asks whether a measured feature helps distinguish participants who do and do not meet that outcome. Those are related but different questions: a count alone does not quantify predictive accuracy. The registry record does not report the observed count, the signature’s predictive performance, or the result for the primary outcome. It therefore does not permit an assessment of the signature’s precision or its ability to classify response.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does a participant-count endpoint describe?

The primary outcome is recorded as a count of participants. A count is a way to report how many people met an outcome definition during the stated assessment period. Here, the timeframe is up to 12 weeks from baseline. To interpret such a count, readers need to know the outcome definition, the number assessed, and how participants without an assessment were handled. The record gives a response-signature description and lesion-response criteria, but does not report the observed count, the number assessed, or the handling of missing outcome data.

In general, a count is not the same as a rate or a comparative effect. A rate would require a denominator and a clearly defined observation period for the participants contributing data. A comparative effect would require a comparison between groups or another stated reference. The registry identifies a single group and does not report a comparator-based primary analysis. No rate, relative effect, or difference should be inferred from enrollment alone.

What is the distinction between response and prediction?

The study description frames its objective as testing whether expression of specific genes can help predict response to Revlimid. In general, prediction concerns how well information measured for a participant identifies or forecasts an outcome. An outcome count answers how many participants met a definition; it does not, by itself, indicate whether a signature predicted those outcomes. Evaluating prediction generally requires a defined signature, a defined response outcome, and a reported method for assessing how well predictions correspond to observed outcomes.

The registry record states that the study sought to confirm predictive value and establish z-score boundaries for future studies. It does not report the signature’s measurement procedure, the prediction method, the boundaries, or performance measures. The record also lists RECIST v1.0 criteria for target lesions assessed by MRI: complete response is disappearance of all target lesions, and partial response is at least a 30% decrease in the sum of the longest diameter of target lesions. It does not explain how those criteria were linked to the signature outcome. As a result, the prediction analysis cannot be reconstructed from the registry entry.

What can confidence intervals and significance tests add?

In general, a confidence interval describes uncertainty around an estimate under a specified statistical model and method. Its width helps readers judge precision: a narrower interval indicates less uncertainty under that method, while a wider interval indicates more. A significance test addresses how compatible the observed data are with a specified null hypothesis under its assumptions. Neither replaces a clear estimate, outcome definition, and analysis population.

The registry record does not report a primary-outcome estimate, confidence interval, or significance test result. It is therefore not possible to describe the magnitude or statistical uncertainty of the response-signature finding from this record. The absence of a reported result is not evidence that the result was null; it means the registry entry does not state the result.

How should the planning assumptions be read?

The record describes a sample-size illustration based on a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. It says the effect to detect is a stated fraction of the observed effect. For time-to-event calculations, the stated convention is a total number of events; for other calculations, the convention is participants per arm. These are planning assumptions, not observations from this single-group trial.

In general, a sample-size calculation illustrates how much information a proposed design might need under specified assumptions. A calculation that uses an effect observed in data is still a planning illustration; it does not show the probability that the completed study detected an effect. Such a probability, when calculated after observing results, is not a substitute for reporting the estimate and its uncertainty. The record does not report the trial’s own sample-size justification. The two-arm assumptions in the illustration should not be mistaken for the design of this study, which is listed as single-group with allocation not applicable.

Why do model assumptions matter?

In general, statistical methods rely on assumptions about how outcomes vary and how observations relate to one another. A normal approximation is a modeling choice for a calculation; it is not a statement that every clinical outcome follows a normal distribution. Over-dispersion means that observed variability is greater than a model expects. When present, it can make standard errors and intervals based on a simpler model too small, creating an overly precise impression. The registry record does not report a dispersion measure for the primary outcome or identify an outcome model for an analysis.

Because the primary endpoint is a participant count, interpretation would depend on the definition of the count, the data available, and the analysis approach. The record does not describe these details beyond the outcome title, unit, parameter type, and timeframe. No particular distributional model, adjustment, or uncertainty method should be assumed.

10. Multiplicity, Missing Data and Analysis Populations

The registry lists one outcome measure and zero analyses. It identifies the primary outcome, but does not report a complete statistical analysis plan, the number of hypotheses tested, or any adjustment for multiple testing. The stated aim includes testing a response signature and establishing z-score boundaries. The record does not explain whether these were treated as one analysis or several, or how the possibility of chance findings across multiple candidate measurements or boundaries was handled.

In general, multiplicity arises when several outcomes, subgroups, or statistical tests are considered. Repeated testing can increase the chance of an apparently notable result even when there is no corresponding underlying association. Adjustment methods and prespecified analysis choices can help address that issue, but the registry record does not report whether any such methods were used. It would be inappropriate to infer either that multiplicity was present in the completed analysis or that it was controlled.

The record also does not report missing primary-outcome data, the number of participants with complete assessments, or how any missing data were handled. Enrollment of 57 participants does not establish that all 57 contributed to the primary analysis. The analysis population is not reported in the registry record, nor are rules for including or excluding participants from an analysis. No assumptions about missingness, imputation, or a modified analysis population can be made from the available entry.

These omissions matter because outcome counts and predictive performance can depend on who was assessed and how incomplete observations were treated. Without the analysis population, missing-data handling, and multiplicity approach, the robustness and precision of any primary result cannot be evaluated. The registry record does not report interim analyses or other details of the statistical plan.

11. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: diarrhea (1.8%), dyspnea (1.8%), febrile neutropenia (1.8%).

The registry lists 57 participants at risk for each of its two adverse-event categories. It records 3 participants affected in the serious adverse-event category and 8 in the other adverse-event category. It also records 0 deaths. These are descriptive counts in the Revlimid group; the registry does not provide a control group for comparison.

The listed serious adverse-event terms are diarrhea, dyspnea, and febrile neutropenia, each reported at 1.8%. These entries describe the terms and percentages recorded in the registry. They should not be read as evidence that Revlimid caused those events, or as a comparison of risk with another treatment. The record does not report further details about event timing, duration, severity beyond the serious classification, or attribution.

In general, “at risk” indicates the group size used for reporting an event category, while the affected count indicates how many participants were recorded as having an event in that category. The serious and other adverse-event counts are separate registry entries and should not be added to infer a number of unique participants, since the record does not report whether participants appeared in both categories. The listed death count is descriptive and is not a comparative estimate.

12. What the Result Does, and Does Not, Mean

The registry record establishes the study’s stated aim: testing whether a gene-expression response signature could predict response to Revlimid among patients with transfusion-dependent, non-del(5q), low or intermediate-1 risk MDS. It also describes the primary outcome as a participant count assessed up to 12 weeks from baseline. However, the record does not report the primary outcome result, the observed number meeting the response definition, or a measure of predictive performance. The central research question therefore cannot be answered from the registry entry alone.

The trial’s single-group design also limits what can be inferred comparatively. In general, observations within one group can describe what was recorded among enrolled participants, but they cannot by themselves estimate how outcomes would differ under a concurrent alternative. No randomized comparison is reported. The record does not report an adjusted comparison or a control-group result.

The safety information is also descriptive: 3 serious adverse-event participants, 8 participants in the other adverse-event category, and 0 deaths are recorded among 57 at risk in each category. These entries do not establish whether an event was caused by the study drug, and the absence of a comparison group prevents a comparative safety interpretation from this record.

The study’s stated intention to establish z-score boundaries for future studies should not be confused with a reported, validated selection rule. The registry does not report the boundaries, their performance, or validation in another group. Likewise, the sample-size illustration is a planning exercise under two-arm assumptions, not an observed efficacy result or evidence of the trial’s statistical power. The completed status indicates the registry lists the study as completed; it does not fill in the unreported primary findings.

Thus, the record provides useful information about the population, intervention, design, intended endpoint, and selected safety counts. It does not provide enough primary-outcome or analysis information to determine whether the response signature predicted response, how precise any estimate was, or how findings would compare with another treatment or course of care.

13. Why This Trial Matters Statistically

This trial is statistically notable for the distinction between a biomarker prediction question and a simple response count. The stated objective is to test whether gene expression can help predict which participants respond to Revlimid. That question concerns the relationship between a signature and an outcome, not merely the number of participants with an outcome. The registry record states the objective but does not report the result needed to assess that relationship.

The single-group design is important context. It can describe outcomes among participants who received Revlimid, but a predictive finding and a comparative treatment effect are different claims. In general, prediction can be evaluated within a study population, while determining how outcomes compare with an alternative requires an appropriate comparison. The registry lists no comparator and no allocation to a control group.

The stated aim to establish z-score boundaries for future studies also highlights the difference between developing a selection approach and demonstrating that it works in new data. In general, a boundary selected in one dataset may perform differently in another population. Independent validation can assess how well a rule carries over, but the registry record does not report a validation result or the proposed boundaries.

The record also illustrates why transparent reporting of analysis populations, missing data, uncertainty, and multiplicity matters. Without the primary result, its uncertainty, and the analysis details, readers cannot judge precision or robustness. The registry’s safety counts add descriptive context but do not answer the predictive question or provide a comparison of risk. The statistical significance of the study’s aim lies in the kind of evidence it sought to develop; the registry entry does not report whether that evidence was obtained.

14. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

Several limitations specific to this trial are apparent from the registry record. The most consequential is the absence of a reported primary outcome result. Although the outcome title, unit, and timeframe are listed, the record gives no observed count and no estimate of the response signature’s predictive value. The z-score boundaries the study aimed to establish are also not reported.

The response outcome is not fully operationalized in the entry. It describes RECIST v1.0 target-lesion criteria assessed by MRI, but does not explain how those criteria were applied to the response-signature endpoint in this MDS population or how a participant was classified for the primary count. The registry does not report details of the gene-expression measurement or the predictive analysis.

The design is single-group, with masking listed as none and allocation not applicable. Consequently, the registry does not provide a randomized comparison with another group. The record also does not report the analysis population, missing primary-outcome data, handling of missing observations, or a detailed statistical analysis plan. It does not report whether any prespecified approach addressed multiple testing or the selection of z-score boundaries.

The enrollment is 57, but enrollment should not be treated as the number analyzed for the primary endpoint: the number with evaluable outcomes is not reported. The safety entries are also limited to category counts, listed serious-event terms and percentages, and deaths; event timing and attribution are not reported. Finally, the sample-size illustration uses two-arm assumptions that do not describe the listed single-group trial design. The trial’s own sample-size justification is not reported in the registry record.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Revlimid in Transfusion Dependent Patients — NCT01178814.
• Registry record last updated 2024-06-24; status last verified 2024-06.
• Results first posted 2021-07-20.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

18. Record Summary

This completed, prospective phase 2 study was a single-group trial sponsored by Columbia University, with 57 participants enrolled. It evaluated Revlimid (lenalidomide), described as an oral capsule taken once a day, in patients with transfusion-dependent, non-del(5q), low or intermediate-1 risk MDS. Allocation was not applicable and masking was none.

The stated objective was to test whether a gene-expression response signature could predict response and help establish z-score boundaries for future patient selection. The primary outcome is a participant count assessed up to 12 weeks from baseline. The registry describes RECIST v1.0 response criteria for target lesions assessed by MRI but does not report the primary outcome result or the signature’s predictive performance.

The safety record lists 3 participants affected in the serious adverse-event category, 8 in the other adverse-event category, and 0 deaths, with 57 at risk in each category. The registry does not report the trial’s sample-size justification, detailed analysis plan, analysis population, or missing-data handling.