This page separates reported registry information from statistical interpretation. The trial record reports that results have been posted, but the ClinicalTrials.gov record does not contain formal statistical-analysis estimates for the registered primary endpoint.
1. Trial at a Glance
TRANSFORM was a randomized, parallel-group, open-label phase 3 study comparing standard of care with JCAR017 in adult subjects with high-risk, transplant-eligible relapsed or refractory aggressive B-cell non-Hodgkin lymphomas.
| Feature | TRANSFORM |
|---|---|
| Trial name | TRANSFORM |
| NCT identifier | NCT03575351 |
| Phase | Phase 3 |
| Condition | Lymphoma, Non-Hodgkin |
| Population | Adult subjects with high-risk, transplant-eligible relapsed or refractory aggressive B-cell non-Hodgkin lymphomas |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 184 |
| Arms | 2 |
| Interventions | Standard of Care (drug); JCAR017 (genetic) |
| Trial status | Completed |
| Lead sponsor | Celgene |
| Sponsor type | Industry |
2. Clinical Question
The central question was whether JCAR017 should be compared with standard of care for efficacy and safety in adults with high-risk, transplant-eligible relapsed or refractory aggressive B-cell non-Hodgkin lymphomas, using event-free survival as the registered primary endpoint.
Population
Adult subjects with high-risk, transplant-eligible relapsed or refractory aggressive B-cell non-Hodgkin lymphomas.
Intervention
JCAR017, classified in the ClinicalTrials.gov record as a genetic intervention.
Comparator
Standard of Care, classified in the ClinicalTrials.gov record as a drug intervention.
Primary question
How does event-free survival, as defined by the independent review committee endpoint, compare between the randomized treatment strategies?
3. Trial Design
Standard of Care
- Standard of Care intervention.
- Registry classification: drug.
- Serious adverse events reported for the pre-crossover period: 21/30.
JCAR017
- JCAR017 intervention.
- Registry classification: genetic.
- Serious adverse events reported for the JCAR017 arm only: 43/92.
The design is explicitly randomized, parallel-group, and unmasked. These features matter statistically because randomization creates the principal basis for comparing the treatment strategies, while the absence of masking can affect operational and assessment processes. The ClinicalTrials.gov record does not provide additional stratification variables, allocation ratios, or details of the randomization procedure.
4. Primary Endpoint
| Endpoint | Registered definition | Endpoint type |
|---|---|---|
| Event-free Survival (EFS) Per Independent Review Committee (IRC) | Time from randomization to death, progressive disease (PD), failure to achieve complete response (CR) or partial response (PR) by 9 weeks, or start of new antineoplastic therapy, whichever occurs first. CR: Target nodes masses must regress to ≤ 1.5cm in LDi, no extralymphatic sites, no new lesions. PR: ≥ 50% decrease in sum of diameters of up to 6 target nodes and extranodal sites, no new lesions. | Time-to-event |
The registry-reported registered time-frame field begins: “From randomization to death from any cause, PD, failure to achieve CR or PR by 9 weeks post randomization, or start of new antineoplastic therapy due to efficacy concerns, whichever occurs first (Up to 36 months)”. The more complete endpoint definition in the ClinicalTrials.gov record specifies that the event is whichever of the defined events occurs first.
Why this endpoint is statistically unusual
EFS is not simply time to death and is not identical to conventional progression-free survival. It combines several clinically defined events into a composite time-to-event endpoint. Here, the event definition includes death, progressive disease, failure to achieve CR or PR by 9 weeks, or initiation of new antineoplastic therapy.
That construction has an important statistical consequence: an EFS event can occur for different underlying reasons. The treatment comparison therefore describes time until the first qualifying event, not time until one particular biological or clinical event.
The endpoint is defined from randomization, and the first qualifying event determines the observed event time.
5. Statistical Methodology
The ClinicalTrials.gov record classifies the primary endpoint as a time-to-event outcome but do not provide a formal statistical-analysis record for it. Therefore, no treatment-effect estimate, confidence interval, p-value, hazard ratio, median EFS, or event count should be inferred from the ClinicalTrials.gov record.
What would typically be used for EFS?
For a randomized parallel-group trial with a time-to-event primary endpoint such as EFS, a conventional analysis would typically estimate the event-time distributions using the Kaplan-Meier method and compare randomized treatment groups with a log-rank test. A Cox proportional-hazards model is commonly used to estimate a hazard ratio and its confidence interval when appropriate.
Here, di represents events at an event time and ni represents subjects at risk immediately before that time. This describes the general Kaplan-Meier framework; it is not a reconstruction of TRANSFORM results.
Why randomization matters
Randomization is the central design feature supporting a causal comparison between JCAR017 and standard of care. In a well-conducted randomized trial, treatment assignment is determined independently of participants' baseline characteristics in expectation. That allows differences in outcomes between randomized groups to be interpreted within the causal framework of the trial rather than as a simple observational association.
Why EFS requires censoring rules
Time-to-event analyses frequently include participants who have not experienced an event by the time their outcome information ends. Such observations are generally right-censored. The validity of the resulting survival estimates depends on prespecified definitions for events, censoring, and follow-up. The ClinicalTrials.gov record defines the EFS events but do not provide the complete censoring rules or missing-data procedures.
Independent Review Committee assessment
The registered endpoint specifies assessment by an Independent Review Committee. Independent assessment can reduce the potential for treatment assignment to influence interpretation of disease status, although the ClinicalTrials.gov record does not provide the committee's detailed operating procedures or adjudication rules.
6. Planned Analysis
ClinicalTrials.gov indicates that results have been posted for TRANSFORM, but the ClinicalTrials.gov record does not contain statistical-analysis entries for the registered primary endpoint. Consequently, this page presents the planned statistical framework rather than an observed efficacy result.
How the registered primary endpoint would normally be analyzed
- Time origin: randomization.
- Event definition: the first occurrence of death, progressive disease, failure to achieve CR or PR by 9 weeks, or start of new antineoplastic therapy.
- Survival estimation: Kaplan-Meier estimation is a standard approach for describing the distribution of a time-to-event endpoint.
- Between-group comparison: a log-rank test is commonly used for comparing survival distributions between randomized groups.
- Effect estimation: a Cox proportional-hazards model is commonly used to estimate a hazard ratio with a confidence interval when its assumptions are appropriate.
- Analysis population: randomized treatment groups would ordinarily be retained according to the prespecified analysis population rather than being redefined according to subsequent treatment received.
7. Statistical Methods Explained
Why was randomization important?
Randomization assigns participants to treatment strategies according to a trial-controlled mechanism rather than allowing treatment selection to depend on prognosis or clinician preference. Its purpose is not to make the two groups numerically identical; rather, it provides the foundation for an unbiased treatment comparison in expectation and helps balance both measured and unmeasured prognostic factors through the randomization process.
Why is EFS a time-to-event endpoint?
EFS records not only whether a qualifying event occurs but also when the first event occurs. A participant who remains event-free for a longer period contributes information about the timing of the endpoint. Participants without an event by the end of available follow-up can contribute censored information rather than being treated as if they had experienced the event.
Why would Kaplan-Meier estimation be useful?
The Kaplan-Meier estimator provides an estimate of the probability of remaining event-free over time while accommodating right-censored observations. For TRANSFORM, that framework is naturally aligned with an EFS endpoint measured from randomization.
What would a hazard ratio mean for EFS?
If a Cox model were used and produced a hazard ratio below 1 for JCAR017 versus standard of care, it would indicate a lower estimated instantaneous rate of experiencing the defined EFS event in the JCAR017 group under the fitted model. It would not mean that the same percentage of every individual patient's risk was reduced, nor would it directly represent an absolute difference in the probability of remaining event-free.
Why does the EFS definition matter?
The definition determines exactly what counts as failure. Because TRANSFORM's EFS definition includes death, progressive disease, failure to achieve CR or PR by 9 weeks, and initiation of new antineoplastic therapy, the resulting treatment effect would summarize the time to the first of these events. Changing the event definition could change the estimand and therefore the interpretation of the analysis.
Why does independent review matter?
The endpoint is explicitly defined as EFS per Independent Review Committee. Independent assessment can provide a standardized approach to determining disease-related events. The statistical interpretation should nevertheless remain tied to the exact adjudication and censoring rules specified by the trial protocol and analysis plan, which are not included in the ClinicalTrials.gov record.
8. Safety Results
The ClinicalTrials.gov record reports serious adverse events by treatment group and crossover status. These figures are affected by the denominators and treatment-period definitions shown in the registry data and should not be converted into a single overall randomized comparison.
| Safety group | Serious adverse events | Interpretation of denominator |
|---|---|---|
| SOC Arm (Pre-crossover Only) | 21/30 | Affected / at risk for the specified pre-crossover group. |
| Liso-cel Arm Only | 43/92 | Affected / at risk for the specified JCAR017-related group. |
| SOC-JCAR017 Arm Pre- and Post-Crossover | 19/61 | Affected / at risk for the specified SOC-JCAR017 group before and after crossover. |
These denominators are not directly interchangeable. In particular, the safety categories distinguish a pre-crossover SOC group, a JCAR017-only group, and a SOC-JCAR017 group incorporating pre- and post-crossover information. A formal randomized safety comparison would require a clearly specified analysis population, exposure definition, follow-up period, and treatment attribution rule.
9. Crossover and Treatment Attribution
The ClinicalTrials.gov record explicitly include a pre- and post-crossover category for the SOC-JCAR017 group. That makes crossover an important design consideration for interpreting both efficacy and safety.
For efficacy, crossover can complicate the relationship between randomized assignment and subsequent treatment exposure. The primary randomized comparison remains anchored to the original treatment assignment, whereas an analysis based only on treatment actually received can answer a different question and can introduce post-randomization selection.
For safety, treatment attribution becomes especially important because an adverse event occurring after crossover may be associated with exposure to a different intervention than the one initially assigned. The ClinicalTrials.gov record demonstrate that the registry distinguishes these periods, but do not provide enough information to reconstruct a full exposure-adjusted safety analysis.
10. Missing Data, Censoring, and Analysis Population
Because the registered primary endpoint is time-to-event, missingness and censoring are conceptually different from simply having a missing numerical measurement at a scheduled visit.
Censoring
A participant who has not experienced an EFS event by the end of observed follow-up may contribute information up to the censoring time rather than being counted as an event.
Missing assessments
Missing disease assessments can affect the determination of progressive disease or response-related components of EFS. The ClinicalTrials.gov record does not specify the trial's complete imputation or censoring rules.
Randomized analysis
Randomization provides the principal basis for the treatment comparison. Reassigning participants according to later treatment exposure can answer a different causal question.
Registry limitation
The ClinicalTrials.gov record does not specify a detailed missing-data strategy, imputation method, censoring algorithm, or formal analysis population definition.
11. Non-Inferiority, Multiplicity, and Bayesian Methods
The ClinicalTrials.gov record does not report a non-inferiority margin, a factorial design, Bayesian methods, a multiplicity strategy, or an interim-analysis procedure. These features therefore cannot be characterized as components of the TRANSFORM statistical design on the basis of the ClinicalTrials.gov record.
| Design topic | What the ClinicalTrials.gov record supports |
|---|---|
| Non-inferiority margin | Not reported in the ClinicalTrials.gov record. |
| Factorial design | Not reported; the design model is identified as parallel. |
| Multiplicity | No formal multiplicity procedure is reported in the ClinicalTrials.gov record. |
| Interim analysis | No interim-analysis procedure is reported in the ClinicalTrials.gov record. |
| Missing-data / imputation method | No formal imputation strategy is reported in the ClinicalTrials.gov record. |
| Stratification | No stratification factors are reported in the ClinicalTrials.gov record. |
| Bayesian methods | No Bayesian analysis is reported in the ClinicalTrials.gov record. |
This distinction is important. Absence of a feature from the ClinicalTrials.gov recordset is not evidence that a protocol or statistical analysis plan could not contain it; it means only that the feature cannot be responsibly described here as a documented component of the registry-reported TRANSFORM record.
12. Interpreting the EFS Endpoint
EFS is a composite time-to-event endpoint. Its interpretation concerns the time from randomization until the first qualifying event, not merely mortality and not merely radiographic progression.
EFS does not mean that every event has the same clinical meaning. Death, progressive disease, failure to achieve CR or PR by 9 weeks, and initiation of new antineoplastic therapy represent different types of events that are combined into one endpoint according to the registered definition.
If a hazard ratio and 95% confidence interval were available, the interval would quantify statistical uncertainty around the estimated relative event rate under the chosen model. It would not describe the range of outcomes that individual patients should be expected to experience.
A p-value addresses evidence against a specified null hypothesis under the statistical model. It does not tell the reader how large or clinically important a treatment effect is. Effect estimates and confidence intervals are needed to understand magnitude and precision.
13. Important Limitations and Interpretation Issues
- No registry-reported formal efficacy analysis: no formal statistical analyses were posted, so this page does not report a numerical EFS treatment comparison.
- Results posted does not equal numerical ClinicalTrials.gov record: the ClinicalTrials.gov record says results have been posted, but the ClinicalTrials.gov record does not include the corresponding formal statistical estimates.
- Composite endpoint: EFS combines several event types, so an overall EFS effect would not identify which component drove a treatment difference without component-specific analyses.
- Crossover: the safety data explicitly include a SOC-JCAR017 pre- and post-crossover category, making treatment attribution an important consideration.
- Safety denominators differ: the registry-reported serious-adverse-event figures use different treatment-period definitions and denominators and therefore cannot be interpreted as a simple three-group comparison.
- Censoring rules are not reported: complete interpretation of a time-to-event analysis requires the prespecified censoring and follow-up rules.
- Model assumptions: if a Cox proportional-hazards model were used, the proportional-hazards assumption would need consideration. A single hazard ratio is not automatically an adequate summary if hazards vary substantially over time.
- Independent review: the endpoint is defined per Independent Review Committee, but the ClinicalTrials.gov record does not provide the complete adjudication procedures.
- Generalizability: the study population was restricted to adult subjects with high-risk, transplant-eligible relapsed or refractory aggressive B-cell non-Hodgkin lymphomas.
14. Why This Trial Matters Statistically
TRANSFORM provides a useful teaching example because its primary endpoint illustrates how modern randomized trials can combine response criteria, progression, mortality, and subsequent treatment into a single time-to-event estimand.
| Concept | How it appears in TRANSFORM |
|---|---|
| Randomization | The study is identified as randomized. |
| Parallel-group design | The design model is parallel with 2 arms. |
| Time-to-event analysis | The registered primary endpoint is EFS. |
| Composite endpoint | EFS is defined using several possible first events. |
| Independent review | The primary endpoint is EFS per Independent Review Committee. |
| Response criteria | Failure to achieve CR or PR by 9 weeks contributes to the EFS definition. |
| Crossover | The ClinicalTrials.gov record distinguishes a SOC-JCAR017 group before and after crossover. |
| Safety analysis | Serious adverse events are reported using treatment- and crossover-specific denominators. |
| Analysis transparency | No formal statistical analyses were posted. |
15. Related Tutorials
Learn more about the methods used in this trial:
16. Related Statistical Calculators
17. Sources
- ClinicalTrials.gov: TRANSFORM, NCT03575351.
- PubMed: Publication indexed under PMID 42044466.
- PubMed: Publication indexed under PMID 40623279.
- PubMed: Publication indexed under PMID 38547003.
- PubMed: Publication indexed under PMID 37544410.
- PubMed: Publication indexed under PMID 36542826.
Continue through the Clinical Biostats statistical learning pathway
Explore tutorials and statistical calculators that provide additional background on randomized clinical-trial methods and treatment allocation.
18. Record Summary
TRANSFORM is a completed randomized phase 3, parallel-group study enrolling 184 adult subjects with high-risk, transplant-eligible relapsed or refractory aggressive B-cell non-Hodgkin lymphomas. The trial compares standard of care with JCAR017 and defines its primary endpoint as event-free survival per an Independent Review Committee. EFS is measured from randomization to the first occurrence of death, progressive disease, failure to achieve complete response or partial response by 9 weeks, or initiation of new antineoplastic therapy.
The statistical structure is therefore centered on a randomized time-to-event comparison with a composite endpoint. Kaplan-Meier estimation, log-rank comparison, and Cox modeling are standard approaches for an endpoint of this type, but the ClinicalTrials.gov record does not provide the formal TRANSFORM statistical-analysis results needed to report a hazard ratio, confidence interval, p-value, median EFS, or event count. The safety information registry-reported likewise requires careful attention to its treatment-period and crossover-specific denominators.