Atezolizumab and Bevacizumab in Combination With TACE for Patients With BCLC B HCC: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Atezolizumab and Bevacizumab in Combination With TACE for Patients With BCLC B HCC (NCT05776875), a na phase 2 study. The primary outcome is Grade 3 or Higher Treatment Related Adverse Events.

2. Clinical Question

The investigators’ stated rationale was to examine a combination of local and systemic treatment in BCLC B hepatocellular carcinoma. The registry summary describes the proposed biological rationale as the possibility that local tumor ablation could release tumor-associated antigens and potentially increase immune response. This is the rationale for asking the question, not a finding from this study. The available record does not establish whether the combination changed tumor response, progression, or safety compared with another approach.

3. Trial Design

This is a phase 2, single-group study sponsored by Yale University. The registry lists an enrollment of 3 and records allocation as not applicable and masking as none. The study title and intervention description identify atezolizumab and bevacizumab in combination with transarterial chemoembolization for patients with BCLC B hepatocellular carcinoma. The arm is described as a single-arm pilot or feasibility study.

The registry lists a start date of 2023-06-07, a primary completion date of 2024-12-05, and a status of terminated. The primary endpoint timeframe is up to 18 months. The primary endpoint description explicitly notes early termination and says the reported endpoint quantity is the number of participants who experienced any grade 3 or higher event. The reason for termination, the timing of termination relative to each participant’s follow-up, and the extent of follow-up completed are not reported in the registry record.

The record lists the treatment components but does not provide a complete treatment schedule in the arm description. It refers to a screening period, a treatment period, and a treatment discontinuation visit; it also starts to describe dosing every 21 days, but the dosing description is incomplete. The planned treatment duration, TACE schedule, and rules for modifying or stopping treatment are not reported in the registry record.

Secondary outcomes listed include response rates in solid tumors and hepatocellular carcinoma, time to progression, time to TACE progression, and time to untreatable progression. The registry lists six outcome measures, but it does not provide analysis entries for them. No randomized comparison is possible from this design. The tables, plots, and sample-size section should therefore be read as descriptions or planning illustrations, not as results from a controlled comparison.

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.

Overall Study

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.

6. Primary Endpoint

The registry lists the primary outcome as Grade 3 or Higher Treatment Related Adverse Events, measured over up to 18 months.

The primary endpoint is titled Grade 3 or Higher Treatment Related Adverse Events and uses participants as its unit. The registry describes safety and tolerability assessment using the rate of grade 3 or higher adverse events, graded with CTCAE version 5, over up to 18 months. It also says that, because the study terminated early, the reported quantity is the number of participants who experienced any grade 3 or higher event. This wording shifts attention from a planned rate-based assessment to a participant count and does not specify in the available record whether the reported events were treatment-related.

In general, an adverse-event endpoint is used to summarize harms observed during a study. A severity threshold can focus attention on events judged to be more serious in intensity, while attribution to treatment distinguishes events considered related to the study treatment from events occurring for other reasons. These are different features: severity and attribution should not be treated as interchangeable.

The registry record does not report the primary endpoint count itself. The safety summary does report counts for serious adverse events and other adverse events, but those categories do not by themselves establish the number of participants with grade 3 or higher events. They also do not supply the treatment-related attribution required by the endpoint title. Accordingly, the available figures cannot be substituted for the primary endpoint result.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry identifies the study as phase 2 and single group, with allocation not applicable and masking none. These design features determine what the statistical summaries can establish. In general, a single-group study describes outcomes among participants who received the study regimen. Without a concurrent comparator, it cannot separate outcomes associated with treatment from outcomes that might also occur under another treatment or without the study regimen. The design can still document feasibility and observed events, but those descriptive aims are different from estimating a comparative effect.

What does a participant count show?

The primary endpoint uses participants as its unit and is described as a count of participants with grade 3 or higher treatment-related adverse events. A participant count records how many people meet a definition, rather than the total number of event episodes. In general, repeated events in the same participant and a single event in a participant are not equivalent when counting people with at least one event. The registry’s wording refers to participants who experienced any grade 3 or higher event after early termination. It does not provide the corresponding value in the record.

The safety summary separately reports participants at risk and participants affected for serious adverse events and other adverse events. In general, the number at risk identifies the group used for that safety summary, while the affected count describes participants with a recorded event in the category. These summaries do not automatically answer a separate endpoint that applies a severity threshold and treatment-related attribution. Event category, severity, and attribution are distinct pieces of information; interpreting one as a substitute for another would require details not reported.

What does the Bayesian approach mean?

The endpoint description says the rate was to be assessed using a Bayesian Predictive Probability approach. In general, Bayesian methods combine observed data with a prior distribution to represent uncertainty about an unknown quantity. Predictive probability methods can assess the probability of a future or target outcome under a specified model and decision framework. Their interpretation depends on the model, prior choices, target rate, and decision thresholds. None of those details, nor a Bayesian analysis result, is reported in the registry record. The mention of the approach alone is not enough to reconstruct the planned analysis or its conclusion.

The record also says the study terminated early and that the endpoint report is the number of participants with any grade 3 or higher event. This description does not reveal whether the stated Bayesian procedure was completed, how early termination affected the analysis, or whether any planned decision criterion was met. Those details are not reported in the registry record. A reader should distinguish the named method in the endpoint description from an available, reproducible analysis result.

How should uncertainty be understood?

In general, a confidence interval gives a range of values compatible with data and a stated statistical procedure; wider intervals indicate less precision. A Bayesian credible interval has a related but different interpretation, based on a posterior distribution. Neither type of interval is reported for the primary endpoint in the information. The safety counts are descriptive counts and do not, on their own, quantify uncertainty around a treatment effect.

In general, over-dispersion occurs when observed variability is greater than a model assumes. It can make standard errors and uncertainty estimates too small if ignored. This concept matters in some count and rate analyses, but the registry record does not report an over-dispersion assessment or a model specification for this trial. It would be inappropriate to assert that over-dispersion was present or that a particular correction was used.

The sample-size section is explicitly a planning illustration, not observed power. In general, power is a design property under specified assumptions, whereas an observed power calculation based on an estimated effect does not replace an interval or a direct account of uncertainty. The section describes assumptions including a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. Those are assumptions for an illustrative calculation; they do not describe this single-group trial’s allocation, which is recorded as not applicable. The planning tables should not be read as evidence that the study enrolled enough participants to answer its clinical question.

Finally, the registry lists secondary outcomes but no analysis entries. In general, time-to-event outcomes require information about event definitions, censoring, follow-up, and the analysis method. A list of outcome names does not supply those details or provide estimates. The same distinction applies to response outcomes: a named response measure is not itself a reported response result. The registry record does not report analysis populations, missing-data methods, multiplicity procedures, or the detailed statistical analysis plan.

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("Atezolizumab and Bevacizumab in Combination With T")
n      <- c(3)
events <- c(1)
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  Atezolizumab and Bevacizumab in Combination With T  3  1  33.3  0.8  90.6  6.1  79.2  6.8  80.6

11. Multiplicity, Missing Data and Analysis Populations

The registry lists several secondary outcomes: response rate in solid tumors, response rate in hepatocellular carcinoma, time to progression, time to TACE progression, and time to untreatable progression. It also lists six outcome measures overall. However, the record contains no analysis entries and does not report estimates for the secondary outcomes. Their presence in the outcome list should not be mistaken for evidence that each was measured, analyzed, or reported successfully.

When a study assesses multiple outcomes, multiplicity can arise because several statistical questions are considered. In general, a prespecified multiplicity procedure can control how uncertainty is interpreted across those questions. The registry record does not report whether multiplicity was addressed, which outcomes were treated as primary or secondary in the analysis plan beyond the listed primary outcome, or whether any adjustment was planned. It is therefore not possible to determine how multiple outcome tests, if any, were handled.

Missing-data methods and analysis populations are also not reported in the registry record. The record does not say how participants with incomplete follow-up, discontinuation, or unavailable outcome measurements were handled. Although early termination is noted, the amount of follow-up and completeness of outcome data are not described. The number enrolled cannot establish how many contributed to each endpoint.

In general, these details matter because different analysis populations or missing-data assumptions can lead to different summaries. They also matter for judging whether reported results follow a prespecified plan. Here, no detailed statistical analysis plan, imputation method, or population definition is available in the record. The reader should treat these aspects as unknown rather than infer that a particular approach was used.

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 non-serious adverse event terms

The posted safety summary records 3 participants at risk for serious adverse events, with 0 participants affected, and 3 at risk for other adverse events, with 3 affected. It also records 0 deaths. These are counts for the categories shown in the registry safety summary, not a full listing of individual events.

In general, an at-risk count is the denominator for the category, while an affected count is the number of participants recorded with an event in that category. These figures do not describe the number of event episodes, their timing, duration, severity grades, or causal assessments. The registry record does not list serious adverse event terms. It does not provide the details needed to identify which events were grade 3 or higher or considered treatment-related.

The primary endpoint is defined around grade 3 or higher treatment-related adverse events, and its description says that after early termination the reported quantity is the number of participants with any grade 3 or higher event. That primary endpoint count is not reported in the registry record. The other-adverse-event count cannot be assumed to equal it, since the category label does not establish the relevant grade or attribution. The safety figures are therefore useful as a limited description of posted categories, but they do not answer the primary endpoint question.

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

The record establishes that a phase 2, single-group study of atezolizumab and bevacizumab with TACE was registered for patients with BCLC B hepatocellular carcinoma, and that the registry lists enrollment of 3 and a terminated status. It also documents safety-category counts: 0 participants with serious adverse events among 3 at risk, 3 participants with other adverse events among 3 at risk, and 0 deaths. These are the principal posted numerical safety observations.

Those observations do not establish the treatment-related grade 3 or higher event count named in the primary endpoint. That value is not reported in the registry record. The safety categories are not interchangeable with the primary endpoint definition, and the aggregate counts do not provide event-level severity or attribution. A reader cannot determine from them how many participants met the primary endpoint.

The record also does not provide the listed secondary outcome analyses. It therefore does not show response rates, time-to-progression results, or estimates for the other progression outcomes. With no randomized comparator, even reported outcomes would describe this study group rather than establish a contrast against another regimen. The record cannot support a causal conclusion about what would have happened to the same participants under a different treatment.

In general, very small studies can be useful for early descriptive and feasibility questions, but their observations have limited precision and broad generalizability. Here, the enrollment of 3 offers little basis for characterizing variation across patients, and the lack of endpoint analyses leaves the main safety result unresolved. The posted information does not prove that treatment is effective, ineffective, safe, or unsafe. It records the design and selected safety counts, while leaving the primary endpoint result and secondary analyses unreported.

14. Why This Trial Matters Statistically

This trial is statistically informative as an example of how design, reporting, and interpretation constrain one another. A single-group pilot or feasibility study can describe what was observed among participants who received a regimen. It cannot provide a randomized estimate of how those outcomes compare with a different treatment. Keeping that distinction clear prevents descriptive counts from being presented as comparative evidence.

The record also illustrates why endpoint definitions matter. The primary endpoint is framed around grade 3 or higher treatment-related adverse events, while the early-termination description refers to participants with any grade 3 or higher event. The safety summary separately presents serious and other adverse events. Severity, seriousness, and treatment attribution are related but distinct concepts. Without an endpoint value and supporting event details, the reader cannot reconcile these summaries or determine how the primary definition was applied.

Early termination makes transparency about follow-up and analysis especially important. The registry acknowledges early termination, but does not report the reason, participant-level follow-up, the primary endpoint count, or secondary analyses. These omissions limit assessment of both completeness and precision. They also mean that planning illustrations cannot take the place of observed trial results.

More generally, this record underscores that a list of outcomes and a named statistical method do not constitute a reported analysis. The Bayesian Predictive Probability approach is named, but its model, assumptions, decision criteria, and result are not reported. Likewise, secondary outcomes are listed without estimates. For statisticians and clinical researchers, the key lesson is to separate what the registry records, what the design could estimate, and what remains unknown. That separation is essential when interpreting small, terminated studies.

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 main limitation is the very small enrollment: the registry records 3 participants. Such a small group provides a narrow description and makes any aggregate count highly sensitive to the experience of an individual participant. The single-group design adds a separate limitation: there is no concurrent comparator, so the study cannot estimate a randomized treatment contrast.

The registry status is terminated, and the primary endpoint description refers to early termination. The reason for termination, the amount of follow-up achieved, and the impact of early termination on outcome availability are not reported in the registry record. The primary endpoint count is not reported, even though the record describes what quantity was reported after termination. The secondary outcomes are listed but no analyses are reported.

The safety summary is limited to aggregate category counts. It does not give event terms, timing, duration, grade, or treatment attribution. No serious adverse event terms are listed. The aggregate other-adverse-event count does not reveal whether those events met the primary endpoint’s grade threshold or were considered treatment-related. The reported 0 deaths likewise does not provide broader information about follow-up or other outcomes.

Other design and analysis details are also missing. A complete intervention schedule is not available in the arm description; the dosing text is incomplete. The registry does not report a detailed statistical analysis plan, sample-size justification, analysis populations, missing-data methods, multiplicity approach, or analysis results. These limitations prevent independent assessment of the planned analysis and mean that the posted record supports only a constrained description of the study, not a comprehensive evaluation of its outcomes.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Atezolizumab and Bevacizumab in Combination With TACE for Patients With BCLC B HCC — NCT05776875.
• Registry record last updated 2025-01-15; status last verified 2024-12.
• Results first posted 2025-01-15.
• 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

The registry record describes a phase 2, single-group pilot or feasibility study of atezolizumab and bevacizumab in combination with TACE for patients with BCLC B hepatocellular carcinoma. Yale University is listed as sponsor. Enrollment is 3; allocation is not applicable, masking is none, and status is terminated. The listed start date is 2023-06-07, and the primary completion date is 2024-12-05.

The primary outcome concerns grade 3 or higher treatment-related adverse events over up to 18 months. Its description notes early termination, but the endpoint count is not reported in the registry record. The safety summary reports 0 serious adverse events among 3 at risk, 3 participants with other adverse events among 3 at risk, and 0 deaths. Several secondary outcomes are listed, but no analysis entries are reported. The record does not provide a randomized comparison or enough information to assess the primary endpoint result, the secondary outcomes, or the full statistical analysis.