Trial of Atezolizumab and Bevacizumab With CHS-388 or Placebo in Patients With Hepatocellular Carcinoma: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Trial of Atezolizumab and Bevacizumab With CHS-388 or Placebo in Patients With Hepatocellular Carcinoma (NCT05359861), a randomized phase 2 study. The primary outcome is Number of Participants With Treatment-emergent Adverse Events (TEAEs).

2. Clinical Question

3. Trial Design

This was a Phase 2 trial with a planned open-label lead-in followed by a randomized phase. The registry describes the overall allocation as randomized, the model as parallel, and the planned masking as quadruple. Those design labels describe the planned comparison; the brief summary states that only the lead-in phase was completed. The lead-in was open label, so its safety and tolerability assessment was not a masked comparison against placebo.

The planned lead-in enrolled a minimum of 6 patients and up to 30 patients to receive SRF388 with atezolizumab and bevacizumab. The randomized phase was intended to compare that combination with placebo plus atezolizumab and bevacizumab. The registry reports enrollment of 30. It does not report that the planned randomized phase was completed or provide results by randomized arm.

The trial status is terminated. The record lists a study start of 2022-04-12 and primary completion of 2025-06-30. The reason for termination is not reported in the registry record. These dates and the termination status do not explain why the randomized phase did not proceed, and the record does not report a revised design or a replacement analysis plan.

The distinction between planned design and completed conduct is essential when interpreting the results. Randomization can support a fair treatment comparison when participants are assigned to study arms and outcomes are analyzed accordingly. In this record, however, no primary outcome analysis is reported for two arms. The lead-in can describe observed events in the treatment group, but it cannot by itself establish how those observations compare with placebo plus the background regimen.

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: This study was planned to include a lead-in period followed by a randomized, placebo-controlled period. Only the lead-in period was completed before study termination.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: The Safety Analysis Set included all participants who received any amount of study drug.

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 included all participants who received any amount of study drug.

6. Primary Endpoint

The registry lists the primary outcome as Number of Participants With Treatment-emergent Adverse Events (TEAEs), measured over Up to approximately 37 months.

The registered primary endpoint is the number of participants with treatment-emergent adverse events (TEAEs). A TEAE is defined as an adverse event that emerged or worsened from the first dose of study drug through 30 days after the last dose. The primary outcome is therefore a participant count: it concerns whether participants experienced an event during the specified observation period, rather than the total number of event episodes.

Safety endpoints of this kind are used to describe the burden of adverse events during treatment and follow-up. Counting participants can help convey how many people experienced at least one event, while event listings can describe the types of events reported. A count alone does not convey severity, duration, recurrence, or whether an event was caused by a study drug. Those questions require additional definitions and results.

The registry says the Safety Analysis Set included all participants who received any amount of study drug. It also says that a summary of other non-serious adverse events and all serious adverse events, regardless of causality, is located in the reported adverse-event information. The available primary outcome information does not include a completed two-arm analysis. Therefore, the endpoint definition is interpretable, but the record does not establish a comparative TEAE result for the planned randomized regimens.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does a randomized comparison contribute?

In general, random assignment is used to make treatment groups comparable with respect to both measured and unmeasured characteristics, on average. A parallel-group design assigns participants to one regimen or another for comparison. Masking is intended to limit the influence of treatment knowledge on care, reporting, and assessment. These are design principles, not proof that a completed comparison occurred. In this trial, the registry describes randomization and quadruple masking for the planned phase, but says only the open-label lead-in was completed.

Because the lead-in was open label, participants and study personnel were not described as masked during that phase. Observations from it can be useful for describing preliminary safety and tolerability, but they do not provide the same protection against bias as a randomized, masked comparison. This distinction matters especially for outcomes that depend on recognition, reporting, or clinical judgment.

What does a participant count capture?

In general, a participant-count endpoint records how many people experienced an event at least once during the defined observation period. It differs from an event count, which can include multiple events for one participant. The registered primary outcome is the number of participants with TEAEs, and the registry defines the time window from the first dose through 30 days after the last dose. That definition sets the event window, but does not by itself specify how missing follow-up, events after the window, or repeated events were handled in an analysis.

A count is descriptive unless it is placed in an appropriate comparison or analysis framework. For a randomized trial, a treatment effect estimate would typically compare outcome measures across arms and express uncertainty, often with a confidence interval. In general, a confidence interval indicates the range of effect values reasonably compatible with the observed data under the statistical model; a wider interval reflects less precision. No primary between-arm estimate or interval is reported in the registry record, so there is no comparative precision to assess.

Why do analysis populations matter?

The registry defines the Safety Analysis Set as participants who received any amount of study drug. In general, a safety population is selected to capture participants who were exposed to treatment and could therefore contribute safety information. Different outcomes can use different analysis populations. For example, efficacy analyses may define eligibility or evaluability differently, but those definitions must be stated rather than assumed. The efficacy analysis population is not reported in the registry record.

For a randomized comparison, an analysis that follows original assignment can preserve the advantages of randomization, while other analysis sets may answer different questions and can be more vulnerable to selection differences. The record does not report which analysis approach was planned for comparative outcomes. Since the randomized phase was not completed, a reader cannot evaluate how assignment, treatment exposure, or deviations from treatment were handled in an efficacy analysis.

What can the sample-size assumptions establish?

In general, sample-size planning uses assumptions about the outcome, the effect worth detecting, the allocation, and the chosen error rate. The registry’s planning description specifies a two-sided significance level of 0.05, equal allocation, and a normal approximation; it also describes illustrations based on fractions of an observed effect. Such calculations are planning illustrations, not evidence that a study had a particular probability of detecting an effect after the fact. They are not observed power.

A planning calculation depends on its input effect and other assumptions. The registry record does not report the trial’s sample-size justification or provide a completed primary outcome comparison from which an observed two-arm effect could be interpreted. No planning output should therefore be treated as a result of this study or as confirmation that the planned randomized phase had adequate precision. The reason for the study’s termination and any changes to its planned enrollment are also not reported in the registry record.

What do the secondary outcomes add?

The registry lists progression-free survival under RECIST v1.1 and HCC mRECIST, as well as response rate, duration of response, and disease control rate under those criteria. In general, response criteria provide standardized rules for classifying changes in disease measurements. Time-to-event outcomes such as progression-free survival require a definition of the event and methods for handling participants who have not had the event by the analysis cutoff. The record lists these measures but does not report their results or the detailed analysis methods used for them.

Different endpoint definitions can yield different summaries because they classify outcomes differently. A reader should not assume that two criteria produce interchangeable results, or that planned secondary outcomes were analyzed, when the registry has no reported analyses. The available record supports a description of intended outcomes and the limited safety information, not an inferred efficacy analysis.

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("Casdozokitug + Atezolizumab + Bevacizumab")
n      <- c(30)
events <- c(30)
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  Casdozokitug + Atezolizumab + Bevacizumab  30  30  100.0  88.4  100.0  88.6  100.0  88.8  99.9

11. Multiplicity, Missing Data and Analysis Populations

The registry lists multiple secondary outcomes, including progression-free survival, objective response rate, duration of response, and disease control rate, each with RECIST v1.1 and HCC mRECIST versions. Testing or estimating several outcomes raises a general multiplicity issue: the more comparisons made, the greater the chance that some results may appear notable by chance. Statistical plans may address this with prespecified testing sequences or adjustments, but the registry record does not report a multiplicity procedure or a detailed statistical analysis plan. No multiplicity-adjusted result is reported.

Missing-data handling is also not reported in the registry record. This matters because loss of outcome information can affect estimates, particularly when the reason for missingness is related to prognosis, treatment, or adverse events. The record does not specify how missing endpoint observations, incomplete follow-up, or participants who stopped treatment were handled for the planned efficacy outcomes.

The stated Safety Analysis Set includes participants who received any amount of study drug. The registry does not report an efficacy analysis population or further analysis-population rules for the planned randomized comparison. As only the lead-in was completed, the reported safety information should not be mistaken for an intention-to-treat comparison between randomized arms. The registry record does not report interim analyses or the statistical plan for such analyses.

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 0% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: sepsis (6.7%), colitis (6.7%), disease progression (6.7%), anorectal infection (3.3%), cellulitis (3.3%).

The listed safety group is casdozokitug with atezolizumab and bevacizumab. The registry reports 30 participants at risk for serious adverse events, of whom 15 were affected. For other adverse events, it reports 30 at risk and 30 affected. The listed number of deaths is 13. These counts should be read as descriptive information for this group, not as a comparison with placebo plus atezolizumab and bevacizumab.

The registry lists sepsis, colitis, and disease progression at 6.7% each among the leading serious-event terms. Anorectal infection, cellulitis, and Escherichia bacteraemia are listed at 3.3% each. The record’s primary endpoint description defines TEAEs by timing, while the adverse-event information also describes serious events and other non-serious events. These categories should not be treated as interchangeable endpoints.

Counts and percentages do not establish that a study drug caused an event. The registry description says serious adverse events were summarized regardless of causality. The record does not provide a randomized comparator result, and it does not report enough detail to interpret duration, recurrence, or the relationship between individual events and treatment. The death count should not be interpreted as a treatment-attributable count.

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

The record establishes that this was a Phase 2 study intended to evaluate casdozokitug with atezolizumab and bevacizumab against placebo with atezolizumab and bevacizumab in untreated, locally advanced or metastatic HCC. It also establishes that the planned study included an open-label lead-in and a randomized phase, and that only the lead-in was completed. The posted safety information gives descriptive event counts for the listed treatment group.

It does not establish whether the combination improved or worsened efficacy outcomes relative to the comparator. It does not establish a comparative difference in TEAEs, serious adverse events, or deaths. The registry record contains no primary outcome analyses and no reported randomized arm-to-arm efficacy analyses. A planned endpoint is not a result, and an observed count in one listed group is not a treatment effect.

The safety counts can inform a descriptive account of events recorded among exposed participants, but they do not answer what would have occurred under the comparator regimen. Likewise, the absence of comparative results should not be read as evidence that the regimens are equivalent. Equivalence or noninferiority requires an appropriate design, prespecified margins where relevant, and an analysis with adequate precision; none is reported.

The registry does not report the reason for termination, the statistical plan, or details of missing-data handling. Without those details, the record cannot explain why the randomized phase was not completed or how its planned analyses would have been conducted. The most defensible conclusion is therefore narrow: the study generated a lead-in safety record, but the planned randomized comparison is not available for statistical interpretation.

14. Why This Trial Matters Statistically

This trial illustrates why study design descriptions must be distinguished from results. A record can describe randomization, masking, a parallel-group structure, and a comparator while also reporting that the randomized phase was never completed. Those planned features do not create a randomized treatment contrast when comparative data are absent.

It also highlights the difference between safety description and comparative safety inference. Counts of participants with events can describe what was recorded in an exposed group. To assess whether event frequency differs between regimens, the analysis needs results from the comparator as well as a suitable estimate and an account of uncertainty. Without those elements, observed counts should not be interpreted as evidence of relative risk.

The endpoint structure matters as well. TEAEs are defined by a specified period around treatment exposure, while serious adverse events and other non-serious events are separate reporting categories. A clear definition helps readers understand what is counted, but it cannot substitute for a completed comparative analysis. Planned efficacy measures likewise identify what the study intended to assess without demonstrating that those outcomes were measured and analyzed.

Finally, the record underscores the importance of transparent reporting about analysis populations, missing data, multiplicity, and termination. Those details affect the interpretation of estimates and the limits of conclusions. In their absence, statistical interpretation should remain closely tied to what is actually reported: a completed lead-in, descriptive safety information, and no reported primary outcome comparison.

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.

The most important trial-specific limitation is that only the open-label lead-in was completed. The intended randomized phase comparing casdozokitug with placebo, each combined with atezolizumab and bevacizumab, is not reported as completed. Consequently, the registry record does not provide a treatment contrast for the primary outcome or comparative efficacy results.

The lead-in was open label and is described as an assessment of preliminary safety and tolerability. It cannot, by itself, separate effects of the study regimen from background risks, disease-related events, or reporting and assessment influences. The available adverse-event counts are descriptive and do not establish causality. The reason for termination is not reported in the registry record.

The record also does not report a detailed statistical analysis plan, the rationale for sample size, multiplicity handling, interim analyses, or methods for missing data. It does not provide an efficacy analysis population or a completed analysis of the listed secondary endpoints. These omissions limit assessment of how the intended outcomes would have been estimated and how uncertainty would have been addressed.

The reported dates, enrollment, endpoint definitions, and adverse-event information help describe the trial, but do not resolve these limitations. In particular, the listed completion date does not indicate that the planned randomized comparison was completed. Interpretations should therefore remain limited to the design and descriptive information actually reported.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Trial of Atezolizumab and Bevacizumab With CHS-388 or Placebo in Patients With Hepatocellular Carcinoma — NCT05359861.
• Registry record last updated 2026-10-01; status last verified 2026-09.
• Results first posted 2026-10-01.
• 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

NCT05359861 is a Phase 2 trial of casdozokitug, formerly known as SRF388, with atezolizumab and bevacizumab in patients with untreated locally advanced or metastatic HCC. The planned design included an open-label lead-in followed by a randomized, quadruple-masked, parallel-group comparison against placebo with atezolizumab and bevacizumab. The registry states that only the lead-in phase was completed and lists the study status as terminated.

The primary endpoint counts participants with TEAEs from the first dose through 30 days after the last dose. The registry reports descriptive safety information for the listed treatment group, but no primary outcome analyses or comparative randomized results. The reason for termination, detailed statistical plan, and missing-data methods are not reported in the registry record. The available evidence supports description of the lead-in, not conclusions about relative efficacy or comparative safety.