A Study of Fruquintinib in Combination With Tislelizumab in Advanced Solid Tumors: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study of Fruquintinib in Combination With Tislelizumab in Advanced Solid Tumors (NCT04577963), a non-randomized phase 1, phase 2 study. The primary outcome is Part 2: Objective Response Rate (ORR).

2. Clinical Question

The study included a safety lead-in followed by dose expansion. The expansion cohorts were Cohort A for triple negative breast cancer treated with immuno-oncology therapy in the metastatic setting, Cohort B for immuno-oncology-naïve triple negative breast cancer in the metastatic setting, Cohort C for endometrial cancer, and Cohort D for microsatellite-stable metastatic colorectal cancer. The record describes Cohort C as immuno-oncology-naïve in the study summary.

3. Trial Design

This was an open-label, non-randomized Phase 1b/2 study with a sequential model. It evaluated fruquintinib in combination with tislelizumab in patients with locally advanced or metastatic solid tumors. The registry lists 52 enrolled participants. The study started on 2021-08-09, had a primary completion date of 2024-06-18, and is listed as terminated.

Part 1 was a safety lead-in. Approximately 6-12 patients with locally advanced or metastatic solid tumors were to receive the combination and be assessed for dose-limiting toxicities during the 28-day observation period. The stated purpose of this part was to determine the recommended Phase 2 dose.

Part 2 was a dose-expansion phase. Patients were enrolled into tumor-specific cohorts receiving the recommended Phase 2 dose. These included two triple negative breast cancer cohorts distinguished by immuno-oncology treatment history, an endometrial cancer cohort, and a microsatellite-stable metastatic colorectal cancer cohort. The registry describes the study as multicenter.

Because allocation was non-randomized and masking was absent, the design does not support a randomized comparison of the combination against a control treatment. The cohorts also represent different diseases and treatment histories. Their results should therefore be understood as cohort-specific observations rather than as a direct comparison between equivalent groups.

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 a safety lead-in phase (Part 1) and a dose expansion phase (Part 2). A total of 52 patients were enrolled in this study (6 patients received study treatment in Part 1 and 46 patients received study treatment in Part 2). The study was terminated early based upon the strategic evaluation of clinical development of fruquintinib in the United States. This change was not based on any concern for patient safety or efficacy relative to fruquintinib and/or tislelizumab treatment.

Part 1

Part 2

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: The DLT-evaluable analysis set included all patients enrolled in Part 1 of the study who received at least 1 dose of fruquintinib or tislelizumab and were considered DLT-evaluable based on the safety review committee review, according to the pre-specified criteria from the protocol.

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: The safety analysis set (SAS) included all enrolled patients who received at least 1 dose of fruquintinib or tislelizumab. 1 patient enrolled in Part 1 was also included in Part 2: Cohort A: TNBC (IO-Treated in the Metastatic Setting) and 5 patients enrolled in Part 1 were also included in Part 2: Cohort B: TNBC (IO-Naïve in the Metastatic Setting).

6. Primary Endpoint

The registry lists the primary outcome as Part 2: Objective Response Rate (ORR), measured over Tumor assessments performed every 8 weeks (+/-1 week) until PD, up to a maximum of approximately 34 months.

The primary endpoint was Part 2 objective response rate (ORR), expressed as the percentage of patients with a confirmed best overall response of complete response or partial response. The investigator assessed response using RECIST v1.1. Best overall response was the best response recorded from the start of study treatment until documented progression under RECIST v1.1 or the start of new anticancer therapy, whichever came first.

Under the endpoint definition, complete response required disappearance of all target lesions; pathological lymph nodes, whether target or non-target, had to decrease to less than 10 millimeters in short axis. Partial response required at least a 30% decrease in the sum of target-lesion diameters, using the baseline sum as the reference. Tumor assessments were scheduled every 8 weeks (+/-1 week) until progression, for up to approximately 34 months.

ORR is a binary-response summary: each patient is classified according to whether a confirmed complete or partial response occurred. It is used to describe observed tumor response under the specified assessment rules. It does not by itself describe how long a response lasts, whether disease remains controlled without meeting response criteria, or survival. The registry record does not report a numerical ORR estimate or confidence interval.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does a response percentage describe?

ORR is a binary endpoint: a patient either meets the endpoint definition for a confirmed complete or partial response, or does not. The percentage summarizes the share of the analysis population classified as responders under the stated response rules. It does not distinguish complete from partial responses, and it does not measure the duration of response. The registry record defines response using investigator assessment under RECIST v1.1 and specifies how best overall response is determined.

The registry lists the endpoint unit as percentage of patients and its dispersion as a 95% confidence interval. A confidence interval expresses uncertainty around an estimated quantity: narrower intervals indicate greater precision, while wider intervals indicate less precision. It is not a guarantee that a particular interval contains the underlying value. No numerical ORR estimate or confidence interval is reported in the registry record, so the observed response and its precision cannot be described from this record.

Why do cohort definitions matter?

The expansion cohorts were not interchangeable groups. They covered triple negative breast cancer, endometrial cancer, and microsatellite-stable metastatic colorectal cancer, with triple negative breast cancer further separated by immuno-oncology treatment history. Disease type and treatment history can shape the clinical context in which an outcome is observed. A difference between these cohorts would not, on its own, show that treatment had a different effect across groups.

The design was non-randomized and open-label, with sequential parts. Random allocation is one way to make comparison groups more alike in expectation; this study did not use it. Open-label assessment means treatment masking was absent. These design features do not make descriptive results unusable, but they limit causal comparisons and may affect how outcomes are assessed. There was no like-for-like randomized control arm recorded.

What does the analysis population mean?

The primary endpoint population is described as the safety analysis set: enrolled patients who received at least 1 dose of fruquintinib or tislelizumab. The registry also states that, as prespecified in the statistical analysis plan, all Part 1 patients were included in the relevant Part 2 cohorts based on tumor type. This affects how the endpoint population is defined and how Part 1 participants may contribute to Part 2 cohort summaries. The record does not report the participant-level outcome data needed to reconstruct the ORR analysis.

How should safety counts be read?

The safety entries give counts of participants at risk and affected for serious adverse events and for other adverse events, along with deaths, by Part 2 cohort. A count of affected participants is a participant-level summary, not necessarily a count of every event experienced. One participant may experience more than one event, and the categories do not identify severity, timing, or treatment relationship beyond the recorded classification. The listed serious-event terms and percentages provide additional entries, but the registry record does not give enough detail to treat these as a full account of event frequency or exposure-adjusted risk.

General statistical principle: event percentages depend on their denominators and on the event definitions and follow-up period. Comparing raw percentages across groups can be misleading when groups differ in size, follow-up, or clinical composition. The posted figures should therefore be read by cohort and in the context of the registry’s definitions, not as evidence from a controlled safety comparison.

What can planning calculations tell readers?

General statistical principle: planning calculations can illustrate the number of participants per arm that might be needed to detect a specified effect under stated assumptions. They are design illustrations, not observed trial outcomes and not observed power. The record’s planning assumptions describe a two-sided significance level of 0.05, equal allocation between two arms, a normal approximation, and an effect defined as a fraction of the observed effect. For time-to-event outcomes, such planning calculations use a total number of events; other calculations refer to participants per arm.

Those illustrations do not turn this non-randomized study into a two-arm controlled comparison. Nor do they supply a missing ORR estimate. The registry record does not report the trial’s own sample-size justification. It also does not report a detailed account of how uncertainty, missing outcome assessments, or multiple endpoint testing was handled. Calculations based on an observed effect should not be interpreted as evidence that the study had a particular probability of detecting an effect.

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("Part 2: Cohort A: TNBC (IO-Treated in the Metastat", "Part 2: Cohort B: TNBC (IO-Naïve in the Metastatic", "Part 2: Cohort C: EC (IO-Naïve)", "Part 2: Cohort D: MSS mCRC (IO-Naïve)")
n      <- c(1, 11, 1, 39)
events <- c(0, 3, 0, 2)
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  Part 2: Cohort A: TNBC (IO-Treated in the Metastat  1  0  0.0  0.0  97.5  0.0  79.3  1.3  84.2
ARM  Part 2: Cohort B: TNBC (IO-Naïve in the Metastatic  11  3  27.3  6.0  61.0  9.7  56.6  9.9  57.2
ARM  Part 2: Cohort C: EC (IO-Naïve)  1  0  0.0  0.0  97.5  0.0  79.3  1.3  84.2
ARM  Part 2: Cohort D: MSS mCRC (IO-Naïve)  39  2  5.1  0.6  17.3  1.4  16.9  1.6  16.9

11. Multiplicity, Missing Data and Analysis Populations

The registry lists a Part 2 primary endpoint of ORR and multiple secondary outcomes. These include Part 1 ORR; progression-free survival, disease control rate, clinical benefit rate, duration of response, overall survival, and drug or metabolite concentration outcomes in Parts 1 and 2. When multiple outcomes or comparisons are examined, the chance of at least one apparently noteworthy result can increase unless the analysis plan addresses multiplicity. The registry record does not report a multiplicity adjustment or a hierarchy for interpreting the outcomes.

The primary analysis population is described as enrolled patients who received at least 1 dose of fruquintinib or tislelizumab. Part 1 patients were also included in relevant Part 2 cohorts based on tumor type, as described in the record. The registry does not report how missing response assessments were handled, how patients without an evaluable assessment were classified, or whether sensitivity analyses were performed. It also does not report a detailed approach to missing data for the secondary outcomes.

The record does not report a detailed statistical analysis plan, interim analyses, or the trial’s sample-size justification. It lists 0 analyses and gives no numerical ORR estimate or confidence interval. Therefore, readers cannot assess multiplicity procedures, missing-data assumptions, or the precision of the primary result from the registry record. These are absences in the record, not evidence that no analyses or procedures existed outside it.

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 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: hypertension (100.0%), constipation (9.1%), disease progression (9.1%), pain in extremity (9.1%), abdominal pain (7.7%).

The registry reports serious adverse events and other adverse events by Part 2 cohort, together with deaths. Serious adverse events affected 0 of 1 participants at risk in Cohort A, 2 of 11 in Cohort B, 1 of 1 in Cohort C, and 20 of 39 in Cohort D. Other adverse events affected 1 of 1, 11 of 11, 1 of 1, and 39 of 39 participants at risk in those cohorts, respectively. Death counts were 1, 8, 0, and 20, respectively.

These counts are descriptive, and the cohorts differ in disease and treatment-history composition. They do not establish comparative safety between treatments or cohorts. The record does not specify in these figures whether deaths were related to treatment, nor does it provide timing, exposure duration, or causes of death. The listed serious-event terms include hypertension at 100.0%, constipation at 9.1%, disease progression at 9.1%, pain in extremity at 9.1%, abdominal pain at 7.7%, and nausea at 5.1%. The registry record does not provide enough context to infer more than the listed terms and percentages.

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

The registry record establishes that this was an open-label, non-randomized Phase 1b/2 study of fruquintinib with tislelizumab, using a safety lead-in followed by tumor-specific dose expansion. It defines the Part 2 primary endpoint as investigator-assessed ORR under RECIST v1.1 and describes the analysis population. It also records adverse-event counts by expansion cohort.

The record does not report a numerical ORR result or confidence interval. It therefore does not show what response percentage was observed or how precisely it was estimated. The listed safety counts describe participants by cohort but do not show a controlled comparison, establish a treatment effect, or give a complete account of the circumstances of deaths and adverse events. No conclusion about treatment benefit or lack of benefit can be drawn from these registry entries alone.

The study’s non-randomized allocation and absence of masking matter because there is no like-for-like randomized control comparison. Its cohorts represent different cancers and, for some groups, different immuno-oncology treatment histories. Those features limit causal interpretation and make direct comparisons between cohorts particularly difficult.

The registry record also does not report details needed to evaluate multiplicity, missing outcome data, interim analyses, or the original sample-size justification. Planning illustrations are not observed power and cannot fill in an unreported clinical result. The appropriate interpretation is therefore bounded: the record describes the study design, endpoint definitions, and certain cohort-level safety figures, while its numerical primary efficacy finding and important analysis details are not reported in the registry record.

14. Why This Trial Matters Statistically

This study illustrates a common challenge in early-phase oncology research: a sequential design can begin with a safety lead-in and then examine a selected dose across disease-specific expansion cohorts. The registry identifies the dose-finding purpose of Part 1 and the intended use of the recommended Phase 2 dose in Part 2. Keeping these distinct objectives in view helps readers avoid treating a dose-selection study as though it were a randomized efficacy comparison.

It also shows why endpoint definitions matter. ORR depends on confirmed investigator-assessed responses under RECIST v1.1, with specified criteria for complete and partial response and a defined period for determining best overall response. Without the numerical estimate and confidence interval, the endpoint definition alone cannot tell readers the observed response rate or its precision.

The cohort structure underscores the importance of clinical context in statistical interpretation. Triple negative breast cancer cohorts were separated by immuno-oncology treatment history, while the other expansion cohorts covered endometrial cancer and microsatellite-stable metastatic colorectal cancer. A single pooled interpretation would obscure those differences, and comparisons across the cohorts would not be randomized comparisons.

Finally, the record demonstrates that registry entries can be informative even when results are incomplete. The design, endpoint, analysis population, and cohort-level safety entries are recorded, but the primary numerical efficacy result and several analysis details are not. Recognizing both what is documented and what remains unreported is essential to interpreting clinical-trial statistics responsibly.

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 limitations are specific to this study’s design and registry record. The trial was open-label, non-randomized, and sequential, with no randomized control group. Consequently, the record cannot support a controlled estimate of the combination’s effect against another treatment. The expansion cohorts also differed by cancer type and, for triple negative breast cancer, by immuno-oncology treatment history, limiting direct comparisons between them.

The primary endpoint was defined for Part 2, but its numerical result and 95% confidence interval are not reported in the registry record. The record lists 0 analyses. As a result, readers cannot assess the observed ORR, its precision, or how the estimate varied across the disease-specific cohorts.

The safety entries are cohort-level counts, but the record does not provide details in those entries about event timing, exposure duration, cause of death, or relationship to study treatment. The listed serious-event terms and percentages also lack enough accompanying context to interpret them as a complete safety profile. The record does not report missing-response handling, multiplicity procedures, interim analyses, or the trial’s own sample-size justification.

These limitations concern what the registry record documents and what can be inferred from the described design. They do not establish that the trial lacked additional analyses or data outside the registry record.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: A Study of Fruquintinib in Combination With Tislelizumab in Advanced Solid Tumors — NCT04577963.
• Registry record last updated 2025-07-14; status last verified 2025-07.
• Results first posted 2025-07-14.
• 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

NCT04577963 is a non-randomized, open-label Phase 1b/2 study of fruquintinib combined with tislelizumab in advanced solid tumors. Its design included a safety lead-in to determine the recommended Phase 2 dose and a dose-expansion phase across four cohorts: triple negative breast cancer with prior immuno-oncology treatment, triple negative breast cancer without such prior treatment, endometrial cancer, and microsatellite-stable metastatic colorectal cancer. The registry lists enrollment of 52, a start date of 2021-08-09, a primary completion date of 2024-06-18, and terminated status.

The Part 2 primary endpoint was investigator-assessed objective response rate under RECIST v1.1. The record lists cohort-level serious and other adverse-event counts and deaths, but no numerical ORR result or confidence interval. It also does not report a detailed statistical analysis plan, multiplicity procedures, or missing-data methods.