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
ELARA was a completed phase 2, single-group study evaluating tisagenlecleucel in adults with refractory or relapsed follicular lymphoma. The registry identifies complete response rate, assessed by an independent review committee, as the primary efficacy endpoint at 1 year.
| Feature | ELARA |
|---|---|
| Phase | Phase 2 |
| Condition | Follicular lymphoma |
| Population | Adult patients with refractory or relapsed follicular lymphoma |
| Design | Single-group, non-masked study |
| Allocation | Not applicable |
| Primary purpose | Treatment |
| Enrollment | 98 |
| Number of arms | 1 |
| Intervention | Tisagenlecleucel |
| Study status | Completed |
| ClinicalTrials.gov | NCT03568461 |
| Lead sponsor | Novartis Pharmaceuticals |
| Sponsor type | Industry |
2. Clinical Question
The clinical question is whether tisagenlecleucel produces complete responses in adults with refractory or relapsed follicular lymphoma, with response determined by an independent review committee using the Lugano 2014 classification response criteria.
Population
Adult patients with refractory or relapsed follicular lymphoma.
Intervention
Tisagenlecleucel, identified in the registry as the study intervention and classified as a biological intervention.
Comparator
There was no comparator arm. The study used a single-group design.
Primary question
What proportion of participants achieve a best overall response of complete response within the registry-defined response assessment period?
3. Trial Design
Tisagenlecleucel
- Biological intervention.
- Evaluated in adults with refractory or relapsed follicular lymphoma.
- Efficacy assessment included CT and PET-based response evaluation.
- Best overall response was evaluated from tisagenlecleucel infusion until progressive disease or start of new anticancer therapy, whichever came first.
Single-group interpretation
- There was no randomized control group.
- There was no between-arm treatment effect estimate.
- There is therefore no treatment-versus-control hazard ratio or risk ratio to interpret.
- The primary efficacy question is centered on the observed complete response rate.
The single-group structure is central to the statistical interpretation of ELARA. In a randomized trial, the difference between treatment groups is the primary object of inference. Here, the principal efficacy quantity is a response proportion within one treated cohort. That changes both the appropriate statistical methods and the kinds of conclusions that can be drawn.
4. Trial Timeline
Study start
The registry records November 12, 2018 as the study start date.
Primary completion
The registry records November 24, 2020 as the primary completion date.
Study status
The registry identifies the study as completed.
5. Primary Endpoint
| Endpoint | Time frame | Registry definition and assessment |
|---|---|---|
| Complete Response Rate (CRR) Per Independent Review Committee (IRC) Assessment | 1 year | Complete response rate was defined as the percentage of participants with a best overall response (BOR) of complete response (CR) recorded from tisagenlecleucel infusion until progressive disease or start of new anticancer therapy, whichever came first. CRR was determined by an independent review committee (IRC) and was based on Lugano 2014 classification response criteria. |
How the response endpoint was constructed
The primary endpoint is not simply a single scan result. It is a participant-level classification based on the best overall response observed during the registry-defined assessment period. The response window begins at tisagenlecleucel infusion and ends at progressive disease or the start of new anticancer therapy, whichever occurs first.
The registry specifies that radiological response is first obtained from CT and PET studies according to the Lugano 2014 criteria. CT response incorporates anatomical measurements of index, non-index, and new lesions as well as spleen length.
PET response is evaluated using a 5-point scale, also described as the Deauville score. The possible PET outcomes are complete metabolic response, partial metabolic response, no metabolic response, or progressive metabolic disease.
| Response framework | Possible outcomes described by the registry |
|---|---|
| Overall response classification | Complete response (CR), partial response (PR), stable disease (SD), or progressive disease (PD) |
| PET response | Complete metabolic response (CMR), partial metabolic response (PMR), no metabolic response (NMR), or progressive metabolic disease (PMD) |
| CT assessment | Anatomical measurements of index/non-index/new lesions and spleen length |
| PET assessment | 5-point scale / Deauville score |
| Review | Independent review committee |
6. Statistical Methodology
Primary analysis of a response proportion
The primary endpoint is a binary participant-level outcome: each participant is classified according to whether the best overall response is complete response or not. The natural statistical summary is therefore the proportion of participants achieving CR.
The complete response rate describes the proportion of the analyzed cohort achieving the prespecified response category. Unlike a hazard ratio, it is an absolute proportion rather than a relative time-to-event effect.
For a single-group phase 2 study, a confidence interval for the response proportion is important because the observed proportion alone does not convey sampling uncertainty. An exact binomial interval or another prespecified binomial confidence interval can be used for this type of endpoint, depending on the statistical analysis plan.
Why the independent review committee matters
The primary endpoint was determined by an independent review committee rather than solely by investigators at the treating sites. Independent review can reduce the influence of knowledge of treatment administration and local assessment practices when classifying radiological response.
That is particularly relevant when response classification depends on multiple imaging modalities and a structured disease-response framework. The statistical endpoint remains a proportion, but the validity of that proportion depends on applying the response criteria consistently.
Why Lugano 2014 matters statistically
The response definition determines who enters the numerator of the CRR. A change in response criteria can change the classification of individual participants and therefore change the estimated response rate. The use of a prespecified disease-response framework makes the endpoint operational rather than subjective in the statistical sense: the analysis depends on the classification rules defined for the study.
Time frame and competing treatment events
The registry-defined observation period ends at progressive disease or the start of new anticancer therapy, whichever comes first. That rule is important because a new anticancer therapy terminates the specified window for determining the best overall response. The endpoint therefore incorporates both the response classification and the protocol-defined observation rule.
7. Statistical Methods Explained
Why is CRR analyzed as a proportion?
The endpoint asks how many participants achieve a particular categorical outcome: complete response. Each participant contributes a response classification, so the primary summary is the fraction of the analysis population whose best overall response is CR.
What does a complete response rate actually measure?
CRR measures the proportion of participants classified as having a best overall response of complete response during the specified assessment period. It does not measure how quickly responses occurred, how long individual responses lasted, or overall survival unless those outcomes are analyzed separately.
Why is an independent review committee used?
An independent review committee provides a separate assessment of the imaging-based endpoint. In a single-group study, there is no randomized control arm to balance differences in assessment behavior between groups. Independent review therefore provides an additional layer of standardized endpoint determination.
Why are both CT and PET relevant?
The registry describes complementary components of response assessment. CT evaluates anatomical disease characteristics, including lesion measurements and spleen length, while PET provides metabolic response information using a 5-point scale or Deauville score. The overall response framework incorporates these assessments under the Lugano 2014 criteria.
Why does the denominator matter?
A response percentage is meaningful only when the analysis population is clearly defined. Enrollment is 98, but the serious-adverse-event safety denominator reported in the registry is 97. Those are different quantities and should not be substituted for one another. The appropriate CRR denominator is the population specified for the primary efficacy analysis.
Why is a confidence interval necessary for a response rate?
An observed response proportion is estimated from a finite number of participants. The confidence interval expresses the statistical uncertainty surrounding that estimate. A wider interval indicates less precision, while a narrower interval indicates greater precision under the chosen statistical framework.
Why is there no hazard ratio for the primary endpoint?
A hazard ratio compares instantaneous event rates between groups over time. ELARA is a single-group study with no randomized comparator, and the primary endpoint is a response proportion rather than a time-to-event comparison. A hazard ratio therefore is not the natural effect measure for the primary endpoint.
8. Planned Analysis
The registry defines CRR as the percentage of participants whose best overall response is complete response, determined by an independent review committee according to Lugano 2014 criteria. For this type of endpoint, the principal analysis would normally estimate the complete response proportion in the prespecified efficacy analysis population and accompany that estimate with an appropriate confidence interval.
The statistical question is therefore fundamentally different from the treatment-effect question in a randomized comparative trial. The relevant estimate is the observed probability of complete response within the treated study population, rather than a ratio comparing tisagenlecleucel with another treatment.
What is measured
The proportion of participants with a best overall response of complete response during the registry-defined period.
How it is assessed
Independent review committee assessment using CT and PET studies under the Lugano 2014 response framework.
Typical statistical summary
A binomial response proportion with a confidence interval appropriate for a single-group efficacy endpoint.
What the design does not provide
A randomized between-treatment comparison, because the study contains a single treatment arm.
9. Understanding the Primary Endpoint Statistically
Suppose a single-group study produces an observed CRR of p. The estimate is simply the number of participants with complete response divided by the number of participants in the relevant analysis population. The statistical challenge is not calculating that fraction; it is quantifying uncertainty and defining exactly which participants belong in the numerator and denominator.
Here, X represents the number of participants achieving complete response among n participants in the relevant analysis population, while p represents the underlying probability of complete response under the statistical model.
The distinction between the observed proportion and the underlying response probability is important. The observed CRR is a sample estimate. The confidence interval describes uncertainty about the underlying probability represented by that estimate under the assumptions of the chosen interval method.
Because ELARA has no concurrent control arm, an observed CRR cannot by itself establish how much better tisagenlecleucel is than another therapy. Any comparative conclusion would require a valid external benchmark or a randomized comparator, neither of which is part of the single-group design described in the registry.
10. Safety
The registry reports serious adverse events by arm as 52 affected participants among 97 at risk in the CTL019 arm.
| Safety measure | Tisagenlecleucel / CTL019 |
|---|---|
| Serious adverse events | 52/97 |
Because this is a single-group study, the safety result is a descriptive frequency within the treated cohort rather than a comparative safety estimate. There is no control-arm serious-adverse-event rate against which the 52/97 figure can be contrasted.
The reported 52/97 figure means that 52 participants were affected among 97 participants at risk for the reported serious-adverse-event measure. It does not provide a relative risk versus another treatment because there is no comparator arm.
The denominator also illustrates why safety and efficacy populations must be kept distinct. The study enrolled 98 participants, whereas the registry's serious-adverse-event measure uses 97 as the number at risk. A safety denominator should not automatically be substituted for the efficacy denominator when interpreting CRR.
11. Endpoint Assessment: CT, PET, and Best Overall Response
The primary endpoint depends on a sequence of clinical and statistical classifications. First, radiological response is obtained from CT and PET studies. CT contributes anatomical information, while PET contributes metabolic response information. These observations are then evaluated within the Lugano 2014 response framework.
| Stage | Statistical relevance |
|---|---|
| CT assessment | Provides anatomical measurements of index, non-index, and new lesions and spleen length. |
| PET assessment | Provides metabolic response classification using a 5-point scale or Deauville score. |
| Lugano 2014 framework | Defines the disease-response categories used for overall response assessment. |
| Best overall response | Determines the participant's best response during the specified observation period. |
| CRR | Summarizes the proportion whose best overall response is complete response. |
This hierarchy matters because the primary analysis is performed on the final participant-level response classification, not on every individual lesion measurement separately. The imaging measurements are inputs into the endpoint determination.
12. What the Primary Endpoint Does — and Does Not — Mean
CRR is an absolute proportion. If a participant is classified as having a best overall response of complete response, that participant contributes to the numerator; participants without that classification do not.
The endpoint therefore answers a specific question: what proportion of the analyzed cohort achieved complete response under the prespecified response framework during the defined observation period?
A complete response rate is not the same as overall survival, progression-free survival, duration of response, or the probability of remaining disease-free at a future time point. Those are different estimands requiring different statistical definitions.
The response estimate is based on a finite sample. Its confidence interval communicates statistical precision around the estimated response probability. It does not describe the range of responses experienced by individual participants.
If a hypothesis test is used for a response-rate endpoint, its p-value describes evidence against the specified null hypothesis under the statistical model. It does not describe the magnitude of the response rate itself. The response proportion and its confidence interval are therefore essential for interpreting the size and precision of the observed effect.
13. Single-Group Design and Causal Interpretation
Randomization is one of the principal mechanisms used to separate treatment effects from differences in prognosis and other characteristics between treatment groups. ELARA does not use random allocation because the registry identifies the allocation as not applicable and the design model as single-group.
That design can be appropriate for a phase 2 investigation in which the principal objective is to characterize activity within a defined patient population. Statistically, however, the interpretation differs from a randomized comparative trial.
| Question | Single-group interpretation |
|---|---|
| What is directly estimated? | The response experience within the treated cohort. |
| Is there a concurrent control group? | No. |
| Can randomized treatment differences be estimated? | No, because there is only one treatment arm. |
| Can CRR be estimated? | Yes, as a proportion within the appropriate analysis population. |
| Can the result establish superiority over another treatment? | Not from the single-group comparison alone. |
This distinction is not a limitation of the arithmetic. It is a feature of the design. A very precisely estimated response proportion can still lack the causal interpretation of a randomized treatment comparison because the study does not contain a contemporaneous randomized comparator.
14. Analysis Population and Denominator Discipline
Clinical trial statistics often depend as much on population definitions as on the statistical test itself. ELARA illustrates this principle because the registry reports 98 participants enrolled but 97 participants at risk for the serious-adverse-event measure.
Enrollment
The study enrollment is 98 participants.
Safety denominator
The registry reports 97 participants at risk for the serious-adverse-event measure.
Efficacy denominator
The appropriate denominator for CRR is the prespecified efficacy analysis population for the endpoint.
Why this matters
Different endpoints can use different analysis populations. Their denominators should therefore be reported with the corresponding endpoint rather than assumed to be identical.
This is particularly important for response endpoints. A response percentage without a clearly identified denominator can be misleading because the numerical percentage depends directly on who is included in the analysis population.
15. Missing Data and Response Assessment
The registry definition of CRR specifies the period over which best overall response is recorded, ending at progressive disease or the start of new anticancer therapy, whichever comes first. That rule establishes an important boundary for endpoint assessment.
For a response endpoint, missing or incomplete assessments can affect whether a participant's best overall response can be established. The statistical handling of such participants should therefore follow the prespecified analysis plan and should be interpreted together with the endpoint definition and analysis population.
The registry information presented here does not specify a separate imputation method for missing primary-endpoint assessments. Consequently, the most important statistical principle is to distinguish the observed response classification from assumptions that might otherwise be made about an unobserved assessment.
16. Multiplicity and Hypothesis Testing
The registry identifies one primary endpoint: complete response rate per independent review committee assessment, measured at 1 year. Unlike a multi-arm randomized trial with several co-primary endpoints, the central statistical problem here is not the comparison of multiple randomized treatment effects.
Multiplicity can nevertheless arise in clinical development when multiple endpoints, time points, response definitions, or subgroup analyses are examined. The statistical interpretation of any additional hypothesis tests depends on whether those tests were prespecified and how the overall type I error was controlled.
| Concept | ELARA context |
|---|---|
| Primary efficacy endpoint | Complete Response Rate per Independent Review Committee Assessment |
| Primary time frame | 1 year |
| Multiple randomized treatment comparisons | Not applicable; the design has one treatment arm. |
| Primary effect measure | Response proportion rather than a treatment-effect ratio. |
17. Why No Hazard Ratio Is Needed for CRR
Hazard ratios are particularly useful when the endpoint is the time until an event such as progression or death and the analysis accounts for censoring. CRR has a different structure. It asks whether a participant achieved a specified response during a defined observation window.
The statistical method should follow the estimand. A response proportion summarizes a binary outcome; a hazard ratio summarizes a relative instantaneous event rate between groups.
Using a hazard ratio for the primary endpoint would therefore answer a different question from the one specified by the registry. The appropriate analysis focuses on the proportion achieving complete response and the uncertainty around that proportion.
18. Limitations
- No randomized comparator: the single-group design does not provide a concurrent randomized control group for estimating a treatment-versus-control effect.
- Response endpoint: CRR captures complete response but does not by itself describe the duration of response, progression-free survival, or overall survival.
- Endpoint classification: CRR depends on imaging assessment and application of the Lugano 2014 response framework, including CT and PET information.
- Observation rule: best overall response is defined from tisagenlecleucel infusion until progressive disease or the start of new anticancer therapy, whichever occurs first.
- Denominator discipline: enrollment and safety denominators are not necessarily identical, as illustrated by 98 enrolled participants versus 97 at risk for the reported serious-adverse-event measure.
- External comparisons: comparison with historical or other-study response rates can be affected by differences in patient selection, response assessment, follow-up, and endpoint definitions.
- Statistical precision: the precision of a response estimate depends on the number of participants contributing to the relevant efficacy analysis and should be assessed with its confidence interval.
19. Why This Trial Matters Statistically
ELARA is a useful teaching example because it demonstrates how statistical analysis changes when a phase 2 clinical trial uses a single treatment arm and a response-based primary endpoint rather than a randomized time-to-event comparison.
| Concept | How it appears in ELARA |
|---|---|
| Phase 2 design | The study evaluates tisagenlecleucel in adults with refractory or relapsed follicular lymphoma. |
| Single-group design | One treatment arm with no randomized comparator. |
| Response proportion | CRR is the primary endpoint. |
| Independent review | CRR is determined by an independent review committee. |
| Objective response classification | Response is evaluated under Lugano 2014 criteria. |
| Multimodal assessment | CT and PET studies contribute to response assessment. |
| Best overall response | The primary endpoint uses the participant's best overall response during a defined period. |
| Confidence intervals | Response proportions should be accompanied by an appropriate measure of statistical uncertainty. |
| Analysis populations | Enrollment and endpoint-specific denominators must be distinguished. |
| Safety description | Serious adverse events are summarized descriptively within the single treatment arm. |
20. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
The primary endpoint is a single-group complete response proportion determined by independent review using the Lugano 2014 response framework. Its appropriate interpretation depends on the analysis population, the response classification, and the uncertainty around the estimated proportion.
Clinical interpretation
The study evaluates tisagenlecleucel in adults with refractory or relapsed follicular lymphoma and describes complete response as its primary efficacy outcome. Safety is a separate component of the evidence and includes the reported serious-adverse-event experience.
The two perspectives should not be conflated. Statistical analysis describes the magnitude and precision of the observed endpoint under the study design. Clinical interpretation considers what that endpoint represents for the population studied, while recognizing that a single-group study does not provide a randomized comparison.
21. Key Statistical Takeaways
1. The estimand is a proportion
The central quantity is the percentage of participants whose best overall response is complete response.
2. There is no treatment-effect ratio
With one treatment arm, there is no randomized comparator and therefore no treatment-versus-control hazard ratio or risk ratio.
3. Endpoint definition matters
CRR depends on the Lugano 2014 response framework and the specified CT and PET assessment process.
4. Review independence matters
The primary endpoint is determined by an independent review committee, providing a separate assessment of response.
5. Confidence intervals add essential information
The observed response proportion should be interpreted together with its confidence interval to understand statistical precision.
6. Safety uses its own denominator
The registry reports 52 serious-adverse-event cases among 97 participants at risk, distinct from the 98 participants enrolled.
22. Statistical Interpretation of the Study Design
A useful way to understand ELARA is to separate three questions that are sometimes blended together in clinical-trial reporting.
| Question | What ELARA addresses | Statistical object |
|---|---|---|
| Did participants achieve complete response? | Yes, through the primary CRR endpoint. | Response proportion |
| How precisely is the response rate estimated? | Addressed through an appropriate confidence interval for the response proportion. | Confidence interval |
| Is tisagenlecleucel better than a concurrent alternative treatment? | Not directly addressed by the single-group design. | No randomized between-arm effect estimate |
This separation is fundamental. A study can provide useful information about response activity without providing the same type of comparative evidence generated by a randomized controlled trial. The strength and meaning of an estimate depend on the question the design was built to answer.
23. Sources
- ClinicalTrials.gov: ELARA, NCT03568461.
- PubMed: Publication indexed under PMID 38194692.
- PubMed: Publication indexed under PMID 35973192.
- PubMed: Publication indexed under PMID 34921238.
Continue through the Clinical Biostats knowledge graph
Explore the broader Clinical Biostats collection of clinical trial results, statistical tutorials, and analysis resources.
24. Record Summary
ELARA is a completed phase 2, single-group study of tisagenlecleucel in adults with refractory or relapsed follicular lymphoma, with 98 participants enrolled and 1 treatment arm. Its primary endpoint is complete response rate per independent review committee assessment at 1 year, with complete response defined through best overall response from tisagenlecleucel infusion until progressive disease or the start of new anticancer therapy, whichever came first.
Statistically, the central feature of ELARA is that its primary efficacy endpoint is a response proportion rather than a randomized treatment comparison. The response assessment incorporates CT and PET studies under the Lugano 2014 classification response criteria, with independent review committee determination. The registry separately reports serious adverse events affecting 52 of 97 participants at risk in the CTL019 arm.