Efficacy and Safety of Tislelizumab in Combination With Fruquintinib in Participants With Selected Solid Tumors: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Efficacy and Safety of Tislelizumab in Combination With Fruquintinib in Participants With Selected Solid Tumors (NCT04716634), a non-randomized phase 2 study. The primary outcome is Objective Response Rate (ORR) as Assessed by the Investigator Based on Response Evaluation Criteria in Solid Tumors (RECIST) Version (v)1.1.

2. Clinical Question

3. Trial Design

This was a multicenter, open-label Phase 2 study of tislelizumab combined with fruquintinib. The registry lists the allocation as non-randomized and the model as parallel. Enrollment was 84, and the study status is completed. The recorded study start was 2021-04-19, and primary completion was 2024-02-22.

The study included two parts. Part 1 was a safety run-in intended to determine dose-limiting toxicity and a recommended Phase 2 dose. Part 2 assessed preliminary efficacy and safety, including ORR. The registry describes separate groups for participants with gastric cancer, colorectal cancer, and PD-L1-positive NSCLC. Their reported numbers at risk in the adverse-event information are 31, 31, and 22, respectively. Those group sizes should not be treated as randomized arms in a comparative efficacy analysis.

Because treatment assignment was not randomized and there was no stated control group, the design does not establish a like-for-like comparison between the combination and another treatment. Open-label conduct means treatment was not masked, and the primary endpoint was assessed by investigators. The registry does not report further information about any steps used to limit assessment bias.

The record names secondary outcomes including progression-free survival, disease control rate, clinical benefit rate, duration of response, overall survival, and treatment-emergent adverse events. It lists 9 outcome measures but no analyses. Details of the statistical plan, including any additional design features or analysis procedures, are not reported in the registry record.

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.

Before assignment: The study consisted of 2 parts: Part 1 (Safety run-in) and Part 2 (Extension). Part 1 of the study with limited participants assessed dose-limiting toxicities (DLTs) and recommended Phase 2 dose (RP2D). These participants later moved to Part 2 of the study. Participant flow, baseline demographics and adverse event data were planned to be reported and analyzed for the combined Part 1 and Part 2 of study. All participants received same dose of study drugs (tislelizumab and fruquintinib).

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Analysis was performed on the DLT evaluable analysis set that included all participants in Part 1 who received at least 85% of the assigned total dose of fruquintinib and at least 67% of the assigned total dose of tislelizumab during the DLT assessment period.

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.

Population: Analysis was performed on safety analysis set that included all participants who received at least 1 dose of study drug(s).

6. Primary Endpoint

The registry lists the primary outcome as Objective Response Rate (ORR) as Assessed by the Investigator Based on Response Evaluation Criteria in Solid Tumors (RECIST) Version (v)1.1, measured over From date of randomization until the date of first documented progression or death from any cause, whichever came first (up to 2 years and 9 months).

The primary endpoint was investigator-assessed objective response rate, defined as the percentage of participants whose best overall response was complete response or partial response under RECIST v1.1. A complete response required disappearance of all target lesions, with pathological lymph nodes reduced to a short axis < 10 mm. A partial response required at least a 30% decrease in the sum of target-lesion diameters relative to the baseline sum.

ORR is a binary outcome at the participant level: each participant either meets the response definition or does not. It provides a direct summary of the proportion meeting a specified tumor-response criterion. It does not, by itself, describe how long a response lasts, how long participants remain free of progression, or overall survival; the registry lists those as separate outcomes.

The outcome timeframe is described as running from randomization until documented progression or death, up to 2 years and 9 months. That wording sits alongside a non-randomized allocation designation. The registry record does not explain this wording. It states that the ORR analysis population was the safety analysis set, consisting of participants who received at least 1 dose of study drug. No response count, percentage, or 95% confidence interval is reported in the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

How is ORR defined?

ORR is a proportion based on a prespecified classification of each participant’s best overall response. In this study, the registry defines the qualifying categories as complete response or partial response under RECIST v1.1. Complete response requires disappearance of target lesions and the specified reduction in pathological lymph-node short axis. Partial response requires at least a 30% reduction in the sum of target-lesion diameters relative to baseline. Participants whose best response does not meet either category do not meet the stated ORR definition.

This kind of endpoint is useful because it converts tumor assessments into a clinically interpretable fraction of participants meeting a response criterion. It is not a time-to-event measure. It does not indicate response duration, time to progression, or survival. The registry lists those as separate outcomes. The analysis details for those outcomes are not reported in the registry record.

What does a confidence interval add?

The registry specifies a 95% confidence interval for ORR. In general, a confidence interval describes the statistical uncertainty around an estimate, rather than guaranteeing that an individual participant’s outcome will fall within a particular range. A narrow interval indicates greater precision under the analysis assumptions; a wide interval indicates less precision. The interval depends on the observed responses and the analyzed number of participants. Since neither an ORR estimate nor its interval is posted, the precision of the primary result cannot be discussed for this study.

General statistical interpretation also depends on how the analysis population and missing assessments are handled. The stated primary population is the safety analysis set: participants who received at least 1 dose of study drug. The record does not say how participants without evaluable tumor assessments, or with missing response information, were treated in the ORR calculation. It also does not report the number analyzed for ORR.

What can the study design establish?

The trial is described as non-randomized, open-label, and parallel. In general, randomization helps balance measured and unmeasured characteristics between treatment groups, while masking can reduce the influence of treatment knowledge on assessment. Neither a randomized comparator nor masking is reported for this study. Its tumor-specific groups are not a control group for one another: gastric cancer, colorectal cancer, and PD-L1-positive NSCLC differ by the population being studied. Differences between their outcomes, if reported, would not by themselves show that treatment effects differ by tumor type.

The endpoint was investigator-assessed. In general, tumor response classification depends on applying the stated criteria to assessments over time. Open-label conduct can make assessment processes more vulnerable to expectation or knowledge of treatment. The registry record does not report whether a blinded independent review was used, so no such procedure can be assumed.

What does the sample-size information mean?

The record describes a planning illustration based on a fraction of the observed effect, with a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. It specifies that non-time-to-event illustrations are expressed as participants per arm. These assumptions describe an illustration, not the trial’s own design justification. The numerical planning results and the protocol’s sample-size justification are not reported in the registry record. Such planning values, when available, are design illustrations and not observed power. The registry does not report a randomized two-arm comparison for this study, so the equal-allocation assumption should not be mistaken for the actual study allocation.

Why can over-dispersion matter?

General explanation: for some outcomes, observed variability is greater than a simple statistical model expects. This is called over-dispersion. If a method assumes less variability than the data contain, uncertainty may be understated. ORR is a binary participant-level outcome, and its uncertainty is usually tied to the response count and analysis population. The registry does not report the analysis method beyond the stated confidence-interval measure, so it is not possible to assess whether any particular variance adjustment was used or needed.

The registry lists 9 outcome measures and no analyses. Detailed statistical methods, including the calculation method for the confidence interval, are not reported in the registry record. No further inferential procedure should be attributed to the trial without that information.

10. Exact Estimates for Each Arm

No two-arm comparison is shown for this trial, but each arm's result can still be summarized with exact interval estimates computed from the posted counts. The Clopper-Pearson interval is exact and conservative; the Wilson interval is a common closer-to-nominal alternative; the Bayesian credible interval uses a uniform Beta(1, 1) prior. For small groups the three can differ noticeably.

R code and output

arm    <- c("Gastric Cancer (GC): Tislelizumab and Fruquintinib", "Colorectal Cancer (CRC): Tislelizumab and Fruquint", "PD-L1 + NSCLC: Tislelizumab and Fruquintinib")
n      <- c(31, 31, 22)
events <- c(4, 3, 9)
for (i in seq_along(n)) {
  ci <- binom.test(events[i], n[i])$conf.int          # Clopper-Pearson exact interval
  cr <- qbeta(c(0.025, 0.975), 1 + events[i], 1 + n[i] - events[i])   # Bayesian, uniform Beta(1,1) prior
  wi <- (function(x, n) { p <- x / n; z <- qnorm(0.975); d <- 1 + z^2 / n
    c((p + z^2 / (2 * n) - z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d, (p + z^2 / (2 * n) + z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d) })(events[i], n[i])
  cat(sprintf("ARM|%s|%d|%d|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f\n", arm[i], n[i], events[i], 100 * events[i] / n[i],
              100 * ci[1], 100 * ci[2], 100 * wi[1], 100 * wi[2], 100 * cr[1], 100 * cr[2]))
}

ARM  Gastric Cancer (GC): Tislelizumab and Fruquintinib  31  4  12.9  3.6  29.8  5.1  28.9  5.3  29.0
ARM  Colorectal Cancer (CRC): Tislelizumab and Fruquint  31  3  9.7  2.0  25.8  3.3  24.9  3.5  25.0
ARM  PD-L1 + NSCLC: Tislelizumab and Fruquintinib  22  9  40.9  20.7  63.6  23.3  61.3  23.2  61.5

11. Multiplicity, Missing Data and Analysis Populations

The record lists multiple outcomes: the primary ORR endpoint and secondary measures including progression-free survival, disease control rate, clinical benefit rate, duration of response, overall survival, and several treatment-emergent adverse-event categories. When multiple outcomes are tested, the chance of at least one apparently unusual result can increase. General approaches to multiplicity include defining a hierarchy, adjusting thresholds, or treating analyses as descriptive. The registry record does not report a multiplicity strategy or explain how confirmatory claims across outcomes were controlled.

The registry lists 9 outcome measures and 0 analyses. It does not provide an ORR result, a confidence interval, or outcome-specific analysis details in the information available. The statistical analysis plan, any prespecified subgroup analyses, and any interim analyses are not reported in the registry record. No particular testing sequence or adjustment can therefore be inferred.

The primary analysis population is described as the safety analysis set, including participants who received at least 1 dose of study drug. The record does not report how many participants were included in the ORR analysis, whether every participant in that set had an evaluable assessment, or how missing and unevaluable assessments were handled. It also does not describe the extent or pattern of missing outcome data. These details matter because different handling rules can change the denominator and the resulting response proportion.

The endpoint timeframe refers to randomization even though allocation is listed as non-randomized. The registry record does not clarify that wording. The analysis population and timing conventions should therefore be read as stated, without assuming additional methods.

12. 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 3% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: death (9.7%), general physical health deterioration (9.7%), hypopituitarism (9.1%), hepatic function abnormal (6.5%), covid-19 (6.5%).

The adverse-event information reports counts at risk and affected for serious events and other events in each tumor group. For gastric cancer, the serious-event figures are 13 of 31, and the other-event figures are 28 of 31. For colorectal cancer, the corresponding figures are 14 of 31 and 29 of 31. For PD-L1-positive NSCLC, they are 13 of 22 and 19 of 22. These are descriptive counts; they do not establish that events were caused by study treatment.

The record also lists death counts of 24, 24, and 8 for the three groups, respectively. The definitions and attribution for these death counts are not reported in the registry record. They should not be interpreted as treatment-related deaths or as deaths caused by adverse events without that information.

Among the listed serious adverse-event terms are Death (9.7%), General physical health deterioration (9.7%), Hypopituitarism (9.1%), Hepatic function abnormal (6.5%), COVID-19 (6.5%), and Cardiac failure acute (4.5%). The record does not provide context for the denominators or whether terms may overlap within participants. A participant can experience more than one event, so term percentages should not be added to obtain a participant-level risk. The registry lists treatment-emergent adverse events and related categories as outcomes, but does not report further causal assessment details in this record.

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

The record establishes that a completed Phase 2 study evaluated tislelizumab plus fruquintinib in selected advanced or metastatic, unresectable solid tumors. It also specifies an investigator-assessed ORR endpoint, its RECIST v1.1 definition, and the intended analysis population. Those details clarify what the study set out to measure.

They do not establish the observed response rate. No primary response count, percentage, or confidence interval is reported in the registry record. The design description, enrollment figure, and endpoint definition cannot substitute for the missing outcome estimate. The registry’s completion status also does not indicate whether the endpoint met any particular threshold.

Even if a response estimate were available, this non-randomized study without a stated control group would describe outcomes among treated participants rather than provide a randomized estimate of comparative effect. The open-label design and investigator assessment are additional features relevant to interpretation. The registry record does not report whether independent blinded assessment or other bias-reduction procedures were used.

The three tumor groups should be understood as separate populations, not as treatment-versus-control arms. The record does not report response results by tumor group, so it does not support claims about consistency or variation across them. It also does not provide the missing-data rules, the number analyzed for ORR, or details of the statistical analysis plan.

Safety information is posted as group-level counts and a list of serious-event terms. These data describe recorded events but do not, on their own, establish causation. The registry’s death counts are not accompanied by attribution details in this record. Overall, the posted information supports a description of the trial’s design and planned outcome definitions, but not a quantitative judgment about the primary efficacy result or a causal comparison.

14. Why This Trial Matters Statistically

This trial is statistically informative as an example of how study design and outcome reporting constrain interpretation. It is a Phase 2, open-label, non-randomized study, with separate tumor-specific groups and no like-for-like control comparison. These design features determine what an observed outcome could mean: without randomization and a control group, a response proportion would be descriptive of the treated participants rather than a direct estimate of comparative treatment effect.

The primary endpoint is clearly defined as investigator-assessed ORR under RECIST v1.1. A prespecified response definition makes the endpoint more interpretable by specifying which response categories count. The registry also identifies a 95% confidence interval as the intended measure of uncertainty. Yet the record does not post the estimate or interval. This illustrates why endpoint definitions and completed-study status are not enough for readers to assess the magnitude or precision of a result.

The study’s distinct disease groups also matter. Gastric cancer, colorectal cancer, and PD-L1-positive NSCLC are presented as separate populations, not randomized comparison arms. The registry’s group-specific adverse-event counts do not answer whether efficacy differs across those populations. Such conclusions would require the relevant outcome estimates and a suitable analysis, neither of which is reported in the record.

Finally, the record lists numerous efficacy and safety outcomes but no analyses. Without reported methods for multiplicity, missing outcomes, analysis populations, or any interim analyses, readers cannot reconstruct the inferential framework. The statistical lesson is to separate what a registry establishes about design and measurement from what requires posted numerical results and analysis details.

15. 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 trial-specific limitations affect interpretation of the registry record. First, it is non-randomized and has no stated control group. The tumor-specific groups are different populations, not comparable treatment arms. Consequently, the record does not support a controlled comparison of the combination with another treatment or a causal treatment-effect estimate.

Second, the study was open-label and the primary endpoint was investigator-assessed. The record does not report whether response assessments underwent independent review or whether other procedures were used to reduce assessment bias. The primary endpoint timeframe also refers to randomization despite the non-randomized allocation designation; the record does not explain this wording.

Most importantly, the registry record does not report the ORR result, confidence interval, or number included in the ORR analysis. It also does not report handling of missing or unevaluable assessments, the statistical analysis plan, multiplicity procedures, subgroup analyses, or interim analyses. These omissions prevent assessment of the result’s precision, robustness, and analysis choices.

The safety information includes group-level counts and serious-event terms, but the record does not specify the attribution and denominator context for every listed item. Death counts are not characterized by cause or treatment relationship. The observed safety counts therefore should not be read as proof of causation. These limitations apply to interpretation of the posted registry information and do not establish whether further results exist outside the registry record.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Efficacy and Safety of Tislelizumab in Combination With Fruquintinib in Participants With Selected Solid Tumors — NCT04716634.
• Registry record last updated 2025-03-25; status last verified 2025-03.
• Results first posted 2025-03-25.
• 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.

19. Record Summary

NCT04716634 is a completed, multicenter, open-label Phase 2 study of tislelizumab plus fruquintinib in selected advanced or metastatic, unresectable gastric cancer, colorectal cancer, and PD-L1-positive NSCLC. The study was non-randomized and included a safety run-in and a part assessing preliminary efficacy and safety. Enrollment was 84.

The primary outcome was investigator-assessed ORR under RECIST v1.1, analyzed in participants who received at least 1 dose of study drug. The record specifies a 95% confidence interval but does not report an ORR estimate or interval. It lists secondary efficacy and safety outcomes, 9 outcome measures, and 0 analyses.

Adverse-event counts are reported for each tumor group, along with selected serious-event terms. The record does not provide enough analysis detail to establish comparative effects, response precision, or the causal relationship of the listed safety events.