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
LAPACT was a phase 2, single-group treatment study enrolling subjects with locally advanced pancreatic cancer. The registry describes nab-paclitaxel plus gemcitabine as the study treatment, with chemoradiation, capecitabine, and surgery also listed as interventions. The primary registered endpoint was Kaplan-Meier estimation of time to treatment failure.
| Feature | LAPACT |
|---|---|
| Phase | Phase 2 |
| Condition | Pancreatic Neoplasms |
| Study population | Subjects with locally advanced pancreatic cancer (LAPC) |
| Design | Single-group |
| Allocation | Not applicable |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 107 |
| Number of arms | 1 |
| Status | Completed |
| Start date | April 21, 2015 |
| Primary completion date | November 21, 2017 |
| Lead sponsor | Celgene |
| Sponsor type | Industry |
| ClinicalTrials.gov | NCT02301143 |
2. Clinical Question
The study addresses the treatment of subjects with locally advanced pancreatic cancer. Its design evaluates a single treatment strategy rather than comparing two randomized groups. The central statistical question is therefore different from that of a conventional randomized controlled trial: the analysis describes the time until treatment failure following initiation of the study treatment rather than estimating a between-group treatment effect.
Population
Subjects with locally advanced pancreatic cancer, within the broader registry condition of pancreatic neoplasms.
Study treatment
Nab-paclitaxel plus gemcitabine.
Additional interventions
Chemoradiation, capecitabine, and surgery are also listed in the registry as interventions.
Primary statistical question
How is time to treatment failure distributed after the first dose of study therapy, using Kaplan-Meier estimation?
3. Trial Design
The absence of a comparator is central to the interpretation of LAPACT. A Kaplan-Meier estimate can describe the observed distribution of treatment-failure times within the study group, but it does not by itself provide a randomized estimate of how the treatment changes treatment failure relative to another therapy.
4. Interventions
Nab-Paclitaxel Plus Gemcitabine
- Nab-paclitaxel
- Gemcitabine
Subsequent or Associated Interventions
- Chemoradiation
- Capecitabine
- Surgery
The registry lists all five interventions under the study: nab-paclitaxel, gemcitabine, chemoradiation, capecitabine, and surgery. The primary endpoint definition explicitly treats the start of a new non-protocol-defined anticancer therapy or surgery as a treatment-failure event.
5. Primary Endpoint
| Endpoint | Registry definition | Time frame | Statistical method |
|---|---|---|---|
| Kaplan-Meier Estimates for Time to Treatment Failure (TTF) | Time after the first dose of study therapy to discontinuation of study therapy due to disease progression, death by any cause, or the start of a new non-protocol-defined anticancer therapy/surgery. If a participant does not progress, die or start a new non-protocol-defined anticancer therapy, the participant was censored on the last tumor assessment date. | Day 1 of study treatment up to 28.75 months; maximum time for the last tumor assessment | Kaplan-Meier estimation; median and 90% CI estimated using the method of Brookmeyer and Crowley |
How TTF is defined
Time to treatment failure begins at the first dose of study therapy. A participant reaches the endpoint when study therapy is discontinued because of disease progression, death from any cause, or initiation of a new non-protocol-defined anticancer therapy or surgery.
This definition is broader than a progression-only endpoint. A participant can therefore experience TTF through several clinically distinct pathways. Death is an event, disease progression is an event, and treatment changes meeting the registry definition are also events.
Tumor assessment and progression
Tumor evaluations were performed using CT or MRI scans at the investigative sites, with response determined according to RECIST guidelines, version 1.1. Progressive disease was defined as a ≥ 20% increase in the sum of diameters of target lesions from nadir, with an absolute increase of ≥ 5 mm; progression of a non-target lesion or appearance of any new lesions was also considered progression.
Participants who had not experienced one of these events were censored at their last tumor assessment date.
6. Statistical Methodology
Kaplan-Meier estimation
The registered primary endpoint is explicitly a Kaplan-Meier estimate of TTF. Kaplan-Meier analysis is designed for time-to-event data in which some participants may not experience the event during the observation period. Those participants contribute information until the time at which they are censored.
Here, di represents the number of treatment-failure events at time ti, while ni represents the number of subjects at risk immediately before that time.
The Kaplan-Meier curve estimates the probability of remaining free of the defined treatment-failure event over time. For TTF, this can be understood as the estimated probability that a participant has not yet experienced progression, death, or the specified treatment-change event by a given time.
Median TTF
The median TTF is the point at which the estimated Kaplan-Meier treatment-failure-free probability reaches 0.50. In a single-group study, this is a descriptive summary of the observed treatment-failure distribution rather than a comparison between treatment strategies.
Brookmeyer and Crowley confidence interval
The registry specifies that the median and its 90% confidence interval were estimated using the method of Brookmeyer and Crowley. This method uses the Kaplan-Meier survival function to construct an interval for the median time-to-event quantity.
A 90% confidence interval describes statistical uncertainty around the estimated median under the specified analysis framework. It does not describe the range of TTF values for individual participants, and it does not mean that 90% of participants have TTF values inside the interval.
Censoring
Censoring is explicitly incorporated into the endpoint definition. When a participant does not progress, die, or begin a new non-protocol-defined anticancer therapy, the participant is censored at the last tumor assessment date. Kaplan-Meier estimation allows those participants to contribute information up to that time without treating the censoring time itself as a treatment failure.
7. Planned Analysis
The registry specifies a Kaplan-Meier analysis of TTF from Day 1 of study treatment through 28.75 months, corresponding to the maximum time for the last tumor assessment. The primary summary is the median TTF with a 90% confidence interval estimated using the Brookmeyer and Crowley method.
| Analysis component | Registry specification |
|---|---|
| Starting point | First dose of study therapy |
| Event definition | Disease progression, death by any cause, or start of a new non-protocol-defined anticancer therapy/surgery |
| Censoring | Last tumor assessment date for participants without a defined treatment-failure event |
| Tumor assessment | CT or MRI; response determined according to RECIST version 1.1 |
| PD definition | ≥ 20% increase in the sum of diameters of target lesions from nadir with an absolute increase of ≥ 5 mm, or progression of a non-target lesion or appearance of any new lesions |
| Primary estimator | Kaplan-Meier |
| Summary measure | Median TTF |
| Confidence interval | 90% CI for the median using Brookmeyer and Crowley |
| Analysis window | Day 1 of study treatment up to 28.75 months |
8. Statistical Methods Explained
Why was Kaplan-Meier estimation used?
TTF is a time-to-event endpoint. Participants can have different lengths of follow-up, and some may reach the end of observation without experiencing the defined event. Kaplan-Meier estimation is designed for exactly this structure because it uses each participant's available event or censoring time rather than requiring every participant to experience treatment failure.
What does the Kaplan-Meier curve represent for TTF?
For this endpoint, the Kaplan-Meier function represents the estimated probability of remaining free of treatment failure through each point in time. Because TTF includes progression, death, and specified treatment changes, the curve is specifically a measure of freedom from that composite treatment-failure definition.
Why is death included as a TTF event?
The registry explicitly defines death by any cause as a treatment-failure event. Including death prevents a participant who dies before documented progression from being treated as if treatment remained successful simply because radiographic progression was not recorded first.
Why can starting another therapy or surgery count as treatment failure?
The endpoint is intended to capture discontinuation of the study treatment for several clinically meaningful reasons. The registry therefore includes initiation of a new non-protocol-defined anticancer therapy or surgery as an event rather than treating the participant as event-free indefinitely after changing treatment.
What does censoring at the last tumor assessment mean?
Suppose a participant has no qualifying TTF event by the last tumor assessment. That participant contributes information showing that treatment failure had not occurred through the assessment date. After that date, the participant's TTF status is not observed within the defined follow-up, so the observation is censored.
Why is the confidence interval for the median 90% rather than 95%?
The registry specifically states that the median and its 90% confidence interval were estimated using Brookmeyer and Crowley. A confidence interval quantifies uncertainty around the estimated median under the stated statistical procedure; its level is a property of the prespecified analysis rather than a measure of clinical importance.
Can TTF from this study be interpreted as a treatment effect?
Not by itself. LAPACT is a single-group study, so its Kaplan-Meier TTF estimate describes the experience of the enrolled study group. Without a randomized comparator, the estimate does not isolate a causal difference between nab-paclitaxel plus gemcitabine and another treatment strategy.
9. Results Reporting
The ClinicalTrials.gov record indicates that results have been posted for LAPACT. However, no formal statistical analyses are posted for the registered primary endpoint in the record. The available registry information therefore establishes the analysis framework for TTF but does not provide a posted formal statistical estimate, confidence interval, or p-value for that endpoint.
What is specified
The endpoint definition, observation window, censoring rule, Kaplan-Meier estimator, median summary, and 90% Brookmeyer-Crowley confidence interval are specified.
What is not formally analyzed in the record
No posted formal statistical analysis is available for the primary TTF endpoint, including a reported p-value or treatment-effect estimate.
This distinction is particularly important for a single-group study. The absence of a comparator means that the natural inferential target is the estimated TTF distribution itself rather than a hazard ratio comparing two randomized treatment groups.
10. Safety Results
The registry reports serious adverse events by study arm. Because LAPACT has one study arm, the safety information is reported for the nab-paclitaxel plus gemcitabine group rather than as a randomized treatment comparison.
| Study arm | Serious adverse events affected | At risk |
|---|---|---|
| Nab-Paclitaxel Plus Gemcitabine (Inducti | 38 | 106 |
| Nab-Paclitaxel Plus Gemcitabine (Overall | 39 | 106 |
Serious adverse events
These are the affected/at-risk counts reported by the registry for the two listed nab-paclitaxel plus gemcitabine safety entries.
The denominators in the safety reporting are 106, whereas overall study enrollment was 107. The registry therefore distinguishes the population contributing to these serious-adverse-event summaries from the enrolled population.
A serious-adverse-event count in a single-group study describes the observed frequency within the treated study population. It does not establish how that frequency compares with another therapy, because there is no randomized comparator in the study design.
11. Understanding the Endpoint More Deeply
TTF is broader than progression-free survival
TTF and progression-free survival can look superficially similar because both are time-to-event outcomes and can incorporate progression and death. Their definitions are not necessarily interchangeable. Here, the registry explicitly includes discontinuation of study therapy due to the start of a new non-protocol-defined anticancer therapy or surgery as a TTF event.
That distinction changes the estimand. The question is not simply whether the tumor progressed or the participant died. It is whether the participant remained free from the broader set of events that the registry defines as treatment failure.
The starting point matters
The clock begins at the first dose of study therapy. Consequently, TTF is measured from treatment initiation rather than from diagnosis, enrollment alone, or a later tumor assessment. Defining time zero precisely is essential because changing the origin of a survival analysis changes the quantity being estimated.
The observation window matters
The registered time frame extends from Day 1 of study treatment up to 28.75 months, described as the maximum time for the last tumor assessment. The analysis therefore has a defined observation horizon rather than an unrestricted claim about treatment failure at any future point.
12. Why This Trial Matters Statistically
LAPACT is a useful teaching example because its statistical structure differs substantially from a randomized phase 3 comparison. The central analysis is not a treatment-effect estimate between arms but a carefully defined time-to-event description within a single treatment group.
| Concept | How it appears in LAPACT |
|---|---|
| Single-group design | One study arm with 107 enrolled subjects. |
| Time-to-event analysis | TTF is measured from the first dose of study therapy. |
| Kaplan-Meier estimation | Used to estimate the TTF distribution. |
| Composite event definition | Progression, death, or specified new treatment/surgery can define treatment failure. |
| Censoring | Participants without a qualifying event are censored at the last tumor assessment date. |
| RECIST assessment | Tumor evaluations use CT or MRI and RECIST version 1.1 criteria. |
| Median time-to-event | The registered summary measure is median TTF. |
| Brookmeyer-Crowley interval | The registry specifies this method for the 90% confidence interval around the median. |
| Single-arm interpretation | TTF describes the observed study population rather than a randomized treatment contrast. |
| Safety denominators | Serious-adverse-event summaries use 106 as the at-risk denominator. |
13. Limitations
- No randomized comparator: The single-group design prevents a randomized comparison of TTF between nab-paclitaxel plus gemcitabine and another treatment.
- Descriptive interpretation: A Kaplan-Meier estimate from one treatment group describes the observed time-to-treatment-failure distribution but does not by itself establish comparative treatment efficacy.
- Composite endpoint: TTF combines progression, death, and specified treatment changes. These events may have different clinical meanings even though they are treated as one endpoint.
- Treatment changes: Because initiation of a new non-protocol-defined anticancer therapy or surgery can constitute treatment failure, the endpoint incorporates treatment-management decisions as well as disease-related events.
- Censoring assumptions: Kaplan-Meier estimation relies on assumptions about the relationship between censoring and the underlying event process. The registry specifies the censoring rule but does not provide a posted formal analysis assessing those assumptions.
- Limited formal statistical reporting: The registry identifies the primary estimator and confidence-interval method but does not post a formal statistical analysis for the primary endpoint.
- Safety population: The reported serious-adverse-event denominators are 106, compared with 107 enrolled subjects. Safety frequencies should therefore be interpreted using their reported at-risk denominator rather than the overall enrollment number.
- Generalizability: The applicability of the observed TTF and safety experience depends on the study's enrolled population and the specific eligibility criteria represented in the registry.
14. Statistical Interpretation vs Clinical Interpretation
Statistical interpretation
The primary endpoint is a right-censored time-to-event quantity analyzed using Kaplan-Meier estimation. The registry specifies median TTF and a 90% confidence interval calculated using Brookmeyer and Crowley.
Clinical interpretation
TTF summarizes how long subjects remained free from the registry-defined combination of progression, death, and specified treatment changes. Because the study is single-group, that observation is not a randomized comparison with another treatment strategy.
15. What Kaplan-Meier TTF Does — and Does Not — Mean
A Kaplan-Meier TTF estimate describes the estimated probability of remaining free of the defined treatment-failure event through a particular time. The event definition includes progression, death, and initiation of a new non-protocol-defined anticancer therapy or surgery.
A Kaplan-Meier TTF estimate is not a probability that an individual patient will necessarily experience treatment failure at that exact time. It is also not a randomized estimate of benefit relative to another treatment.
The median gives a single summary of the time-to-treatment-failure distribution. It is especially useful when the mean is difficult to interpret because of censoring or because the event-time distribution is skewed.
The 90% confidence interval expresses uncertainty around the estimated median. A wider interval indicates less precision in the median estimate, while a narrower interval indicates greater statistical precision under the stated analysis framework.
16. Censoring and Follow-Up
Participants who do not experience disease progression, death, or a qualifying new anticancer therapy or surgery are censored at the last tumor assessment date. This means their exact subsequent TTF is not observed within the defined analysis window, but their observed follow-up remains informative through the censoring date.
| Participant status at last observation | TTF treatment |
|---|---|
| Progression occurs | Counted as a TTF event. |
| Death occurs | Counted as a TTF event. |
| New non-protocol-defined anticancer therapy or surgery begins | Counted as a TTF event. |
| No qualifying event by last tumor assessment | Censored at the last tumor assessment date. |
This structure is why simple averages of observed treatment durations would not reproduce the registered analysis. Kaplan-Meier estimation explicitly incorporates the timing of events and censoring.
17. RECIST and the Definition of Progression
Progression within the TTF definition is based on tumor evaluation by CT or MRI and response determination according to RECIST version 1.1. The registry specifies a target-lesion progression threshold of a ≥ 20% increase in the sum of diameters from nadir together with an absolute increase of ≥ 5 mm.
Progression of a non-target lesion or appearance of any new lesions is also considered progression.
Because progression is one of the pathways to TTF, the radiologic definition is directly connected to the primary statistical endpoint. The statistical analysis therefore depends not only on the survival-analysis method but also on the clinical rules used to classify progression.
18. Trial Timeline
Study start
The LAPACT study began.
Primary completion
The registry records November 21, 2017 as the primary completion date.
28.75 months
The registered TTF time frame extends from Day 1 of study treatment up to 28.75 months, identified as the maximum time for the last tumor assessment.
19. How to Read the Safety Data Alongside TTF
Safety and TTF answer different questions. TTF describes time until a predefined treatment-failure event, while serious-adverse-event reporting describes an important category of adverse safety outcomes.
TTF
A time-to-event endpoint incorporating progression, death, and specified treatment changes.
Serious adverse events
A safety outcome reported using affected and at-risk counts for the study group.
Different denominators
Enrollment was 107, while the reported serious-adverse-event entries use 106 as the at-risk denominator.
No randomized safety contrast
Because the study has one arm, the reported serious-adverse-event counts do not provide a randomized treatment comparison.
20. What a Formal TTF Analysis Would Usually Report
For a Kaplan-Meier time-to-event endpoint such as TTF, a complete formal statistical presentation would ordinarily describe the number of events and censored observations, provide the Kaplan-Meier estimate over time, report the median when estimable, and provide the confidence interval around the median. The registry specifies the median and 90% confidence interval as the principal summary quantities for LAPACT.
A Kaplan-Meier analysis is more informative than a simple count of treatment failures because the timing of each event matters. Two studies could have the same number of events but very different treatment-failure curves if the events occurred at different points in follow-up.
21. Why the Absence of a Comparator Changes the Statistics
In a randomized trial, a hazard ratio can summarize the relative difference in event hazards between treatment groups. In a single-group study, there is no second randomized risk set against which to calculate such a treatment contrast.
The statistical emphasis consequently shifts from comparative inference to time-to-event estimation. The Kaplan-Meier curve describes the observed event-free experience, while the median and its confidence interval summarize the distribution without requiring an external control group.
| Question | What the LAPACT design can address | What requires a comparator |
|---|---|---|
| When does treatment failure occur? | Kaplan-Meier TTF estimation. | Not required. |
| What is the median TTF? | Kaplan-Meier median with the specified 90% CI. | Not required. |
| Is TTF different from another treatment? | Not established by the single-group design. | Requires an appropriate comparator. |
| What is the hazard ratio versus another treatment? | Not a registered single-group comparison. | Requires event-time data from a comparison group and an appropriate model. |
22. Important Interpretation Issues
- Endpoint definition drives interpretation: TTF includes events beyond radiographic progression, so it should not automatically be described as equivalent to a progression-only endpoint.
- Death is an event: Death by any cause is explicitly included in the registered TTF definition.
- Treatment changes are events: Starting a new non-protocol-defined anticancer therapy or surgery can terminate treatment-failure-free follow-up.
- Single-arm evidence is descriptive: Without randomization, the observed TTF distribution cannot isolate the causal effect of the study treatment from other influences.
- Confidence intervals describe precision: The 90% interval concerns uncertainty around the estimated median, not variability among individual participants.
- Censoring requires assumptions: Kaplan-Meier inference depends on assumptions about the censoring mechanism.
- RECIST classification matters: Because progression is one route to TTF, the rules for identifying progressive disease directly affect the endpoint.
- Safety and efficacy are distinct: Serious-adverse-event counts should be interpreted independently from the time-to-treatment-failure analysis.
23. Sources
- ClinicalTrials.gov: NCT02301143 — LAPACT.
- PubMed: PubMed record associated with LAPACT.
Continue through the Clinical Biostats knowledge graph
Explore additional Clinical Biostats resources on clinical-trial design, survival analysis, endpoints, and statistical methodology.
24. Record Summary
LAPACT is a phase 2, single-group study of nab-paclitaxel plus gemcitabine in subjects with locally advanced pancreatic cancer. Its registered primary endpoint, Kaplan-Meier Estimates for Time to Treatment Failure, illustrates the statistical structure of a single-arm survival analysis: time begins at the first dose of study therapy, treatment failure includes disease progression, death, or specified initiation of new anticancer therapy or surgery, and participants without such an event are censored at their last tumor assessment date.
The registry specifies Kaplan-Meier estimation, with the median TTF and its 90% confidence interval estimated using the Brookmeyer and Crowley method. Tumor progression is determined using CT or MRI assessment and RECIST version 1.1 criteria, including the specified target-lesion increase from nadir and the other registry-defined progression criteria.
The key statistical distinction is that LAPACT does not contain a randomized comparator. Its primary analysis therefore describes the time-to-treatment-failure experience of the enrolled study group rather than estimating a treatment effect between randomized groups. The serious-adverse-event summaries likewise describe the single study group, with 38/106 and 39/106 affected/at-risk counts reported for the two listed nab-paclitaxel plus gemcitabine entries.