This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical trial results on this page are restricted to the results posted in the ClinicalTrials.gov record; calculations added by Clinical Biostats are labelled as such.
1. Trial at a Glance
This page covers the registry results for A Study Evaluating The Efficacy and Safety of Neoadjuvant Immunotherapy Combinations in Patients With Surgically Resectable Hepatocellular Carcinoma (NCT05908786), a randomized phase 1, phase 2 study. The primary outcome is Major Pathologic Response (MPR) Rate.
| Feature | Registry value |
|---|---|
| ClinicalTrials.gov identifier | NCT05908786 |
| Phase | Phase 1, Phase 2 |
| Condition | Carcinoma, Hepatocellular |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | None (open label) |
| Enrollment | 62 |
| Arms in the primary outcome | 2 |
| Lead sponsor | Hoffmann-La Roche |
| Results posted | Yes |
2. Clinical Question
Population
Participants with surgically resectable hepatocellular carcinoma were eligible for this neoadjuvant study. The primary outcome used the efficacy-evaluable population: participants who received at least one dose of treatment for their assigned regimen. The posted primary-outcome results include 30 participants in each of the two reported cohorts.
Intervention
The intervention of interest in the reported comparison was atezolizumab plus bevacizumab plus tiragolumab before surgery. The study also included a cohort receiving atezolizumab plus bevacizumab. A separate platform arm evaluated tobemstomig plus bevacizumab, but its enrollment was closed and the posted outcome results do not include its efficacy findings.
Comparator
The comparator for the reported primary comparison was atezolizumab plus bevacizumab. Treatment was given before surgery, for up to three cycles or until surgery or unacceptable toxicity, whichever occurred first. The registry describes randomization, but the randomization ratio and allocation procedures are not reported in the registry record.
Primary question
At the time of surgery, up to 15 weeks, what percentage of efficacy-evaluable participants in each treatment cohort achieved major pathologic response? MPR meant no more than 10% residual viable tumor in the primary tumor bed, assessed by a central pathology laboratory. The reported comparison estimates the difference between the two treatment cohorts.
3. Trial Design
This was a Phase Ib/II, multicenter, randomized, open-label, parallel-group platform study in participants with surgically resectable hepatocellular carcinoma. Its stated purpose was to evaluate the efficacy and safety of neoadjuvant immunotherapy combinations. Because the study was open-label, masking was not used. The registry describes a flexible platform that could open treatment arms as agents became available, close arms with minimal clinical activity or unacceptable toxicity, or modify the participant population.
The registry lists atezolizumab plus bevacizumab, atezolizumab plus bevacizumab plus tiragolumab, and tobemstomig plus bevacizumab as treatment combinations. Participants in each described arm could receive up to three cycles before surgery, stopping earlier for surgery or unacceptable toxicity. Enrollment in the tobemstomig plus bevacizumab arm is listed as closed. The reported primary efficacy results compare the two atezolizumab-based cohorts; efficacy results for the closed arm are not reported in the registry record.
Total enrollment is listed as 62. The primary efficacy results use 30 participants in each reported cohort, while the safety figures use denominators of 31 for atezolizumab plus bevacizumab and 30 for the three-drug combination. These are outcome-specific analysis populations and should not be assumed to represent identical sets of participants. The registry defines the primary efficacy-evaluable population as everyone who received at least one dose of study treatment for the assigned regimen.
The registry lists the study status as terminated. The reason for termination, the operational details of randomization, any allocation ratio, and any changes to the planned study are not reported in the registry record. The registry lists a start date of 2023-12-05 and a primary completion date of 2025-08-27; those dates do not, by themselves, explain the termination or establish why enrollment ended.
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 planned tobemstomig + bevacizumab arm was never opened for enrolment. Participants had the option to receive adjuvant atezolizumab + bevacizumab until protocol version 4, at which time this option was removed based on the updated results of IMbrave050 (NCT04102098).
Overall Study
| Stage | Atezolizumab + Bevacizumab | Atezolizumab + Bevacizumab + Tiragolumab |
|---|---|---|
| STARTED | 32 | 30 |
| Safety-evaluable (SE) Population | 31 | 30 |
| COMPLETED | 0 | 0 |
| NOT COMPLETED | 32 | 30 |
Reasons for not completing this period:
| Reason | Atezolizumab + Bevacizumab | Atezolizumab + Bevacizumab + Tiragolumab |
|---|---|---|
| Death | 2 | 3 |
| Physician Decision | 1 | 2 |
| Study Terminated by Sponsor | 27 | 25 |
| Withdrawal by Subject | 2 | 0 |
Analysis population for the primary outcome, as defined in the registry: Efficacy-evaluable population included all participants who received at least one dose of study treatment for their assigned treatment regimen.
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: Intent-to-treat (ITT) population included all randomized participants.
| Characteristic | Unit | Atezolizumab + Bevacizumab | Atezolizumab + Bevacizumab + Tiragolumab | Total |
|---|---|---|---|---|
| Participants analysed | 32 | 30 | 62 | |
| Age, Continuous | years | 62.97 (SD 11.34) | 64.13 (SD 11.58) | 63.53 (SD 11.37) |
| Sex: Female, Male: Female | Participants | 8 | 10 | 18 |
| Sex: Female, Male: Male | Participants | 24 | 20 | 44 |
| Ethnicity (NIH/OMB): Hispanic or Latino | Participants | 4 | 1 | 5 |
| Ethnicity (NIH/OMB): Not Hispanic or Latino | Participants | 19 | 22 | 41 |
| Ethnicity (NIH/OMB): Unknown or Not Reported | Participants | 9 | 7 | 16 |
| Race (NIH/OMB): American Indian or Alaska Native | Participants | 0 | 0 | 0 |
| Race (NIH/OMB): Asian | Participants | 10 | 11 | 21 |
| Race (NIH/OMB): Native Hawaiian or Other Pacific Islander | Participants | 0 | 0 | 0 |
| Race (NIH/OMB): Black or African American | Participants | 1 | 4 | 5 |
| Race (NIH/OMB): White | Participants | 16 | 11 | 27 |
| Race (NIH/OMB): More than one race | Participants | 0 | 0 | 0 |
| Race (NIH/OMB): Unknown or Not Reported | Participants | 5 | 4 | 9 |
6. Primary Endpoint
The registry lists the primary outcome as Major Pathologic Response (MPR) Rate, measured over At the time of surgery (up to 15 weeks).
| Element | Registry entry |
|---|---|
| Result type | Number |
| Measure of spread | 95% Confidence Interval |
| Unit | percentage of participants |
| Endpoint classification | Binary Endpoint |
The primary endpoint was the major pathologic response rate, expressed as the percentage of participants achieving MPR in each treatment cohort. MPR was defined as no more than 10% residual viable tumor in the primary tumor bed at surgical resection, based on assessment by a central pathology laboratory. The endpoint was assessed at surgery, up to 15 weeks. Participants who did not proceed to surgery were counted as nonresponders for this endpoint.
Pathologic response is used in a neoadjuvant study to describe what remains of the tumor in tissue removed at surgery after treatment given beforehand. Unlike an imaging-based response measure, this endpoint relies on examination of the resected tumor. Central pathology assessment provides a common basis for classifying response across participants, although the registry does not report further details about the pathology procedures or how assessments were adjudicated.
The endpoint is a proportion rather than a time-to-event measure. It answers how many participants in the efficacy-evaluable population met the stated tissue-based response definition, not how long participants lived without recurrence or how long they survived. Those longer-term outcomes are listed among the secondary outcomes, but their numerical results are not reported in the registry record. MPR is therefore a specific measure of response at surgery, and should not be treated as interchangeable with relapse-free survival, event-free survival, or overall survival.
The definition also makes the treatment path relevant to interpretation: a participant who received treatment but did not reach surgery was classified as a nonresponder, rather than being omitted from the MPR denominator. The registry does not report the number of participants who did not proceed to surgery, so the contribution of that rule to the observed percentages cannot be assessed separately.
7. Results Posted in the Registry
- Atezolizumab + Bevacizumab + Tiragolumab: 3.3% (1/30) versus 0.0% (0/30) on control; risk difference 3.3 percentage points (95% CI -3.1 to 9.8); unadjusted two-proportion z-test p = 0.313.
Risk difference: Atezolizumab + Bevacizumab + Tiragolumab
95% CI: -3.1 to 9.8 percentage points
Recalculated from the posted summary statistics
8. Secondary Efficacy Endpoints
The registry also posts results for secondary outcomes. For each outcome below, the endpoint type, the test applied to the posted summary statistics, the result and its interpretation are given in turn. Secondary endpoints are usually not adjusted for multiple testing, so they support rather than confirm the primary finding.
| Secondary outcome | Type | Effect measure | Estimate (95% CI) | Planning size, 80% power |
|---|---|---|---|---|
| Objective Response Rate (ORR), as Assessed by the Investigator According to RECIST v1.1 | Binary Endpoint | Risk difference | +0.0 pts (-15.2 to 15.2) | n/a |
| ORR, as Assessed by the Investigator According to Hepatocellular Carcinoma-Specific Modified Response Evaluation Criteria in Solid Tumors (HCC mRECIST) | Binary Endpoint | Risk difference | -6.7 pts (-25.5 to 12.1) | 490 per arm |
| Percentage of Participants Downstaged to Within Milan Criteria (for Participants Beyond Criteria at Randomization) | Binary Endpoint | Risk difference | +6.2 pts (-5.6 to 18.1) | n/a |
Objective Response Rate (ORR), as Assessed by the Investigator According to RECIST v1.1
- Endpoint type: Binary: the percentage of participants with the event (counts rebuilt from the posted percentages and group sizes).
- Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
- Result:
- Atezolizumab + Bevacizumab + Tiragolumab: 10.0% (3/30) versus 10.0% (3/30) on control; risk difference 0.0 percentage points (95% CI -15.2 to 15.2), risk ratio 1.00 (0.22–4.56); unadjusted two-proportion z-test p = 1.000.
- Interpretation:
- The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Atezolizumab + Bevacizumab + Tiragolumab and the control. The interval gives the range of effects that remain compatible with the data.
ORR, as Assessed by the Investigator According to Hepatocellular Carcinoma-Specific Modified Response Evaluation Criteria in Solid Tumors (HCC mRECIST)
- Endpoint type: Binary: the percentage of participants with the event (counts rebuilt from the posted percentages and group sizes).
- Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
- Result:
- Atezolizumab + Bevacizumab + Tiragolumab: 13.3% (4/30) versus 20.0% (6/30) on control; risk difference -6.7 percentage points (95% CI -25.5 to 12.1), risk ratio 0.67 (0.21–2.13); unadjusted two-proportion z-test p = 0.488.
- Interpretation:
- The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Atezolizumab + Bevacizumab + Tiragolumab and the control. The interval gives the range of effects that remain compatible with the data.
Percentage of Participants Downstaged to Within Milan Criteria (for Participants Beyond Criteria at Randomization)
- Endpoint type: Binary: the percentage of participants with the event (counts rebuilt from the posted percentages and group sizes).
- Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
- Result:
- Atezolizumab + Bevacizumab + Tiragolumab: 6.2% (1/16) versus 0.0% (0/18) on control; risk difference 6.2 percentage points (95% CI -5.6 to 18.1); unadjusted two-proportion z-test p = 0.282.
- Interpretation:
- The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Atezolizumab + Bevacizumab + Tiragolumab and the control. The interval gives the range of effects that remain compatible with the data.
The secondary outcomes address several distinct questions: radiographic response before surgery, downstaging among participants beyond Milan criteria at randomization, pathologic response, recurrence, events, and survival. These outcomes are related to the primary endpoint in that they describe tumor response or subsequent clinical course, but they are not interchangeable. MPR is a tissue-based assessment at surgery, whereas ORR is an investigator assessment before surgery and time-to-event outcomes concern events over follow-up.
Among the numeric secondary comparisons reported, the direction is not uniformly aligned. The RECIST v1.1 ORR estimates are the same in the two reported cohorts, while the HCC mRECIST and downstaging point estimates differ in direction from one another. Their confidence intervals include zero, so these comparisons leave substantial uncertainty about the size and direction of any cohort differences. This mixed pattern does not provide a consistent secondary-outcome signal that resolves the uncertainty in the primary MPR comparison.
The registry names pCR, relapse-free survival, event-free survival, overall survival, survival rates at specified follow-up points, and additional response outcomes. Numerical results for those outcomes are not reported in the registry record. Consequently, they cannot be used to corroborate or contradict the MPR finding based on the registry entries reported.
Secondary results generally carry less confirmatory weight when many outcomes are evaluated without a reported multiplicity strategy. Testing or estimating multiple outcomes increases the chance that some results will appear notable by chance, and the record does not report a procedure to address that concern. The secondary comparisons are therefore best treated as exploratory descriptions. They can identify questions for further study, but should not be used alone to claim a definitive treatment difference or to generalize the primary result to outcomes that were not numerically reported.
Secondary outcomes registered for this trial, for reference (only those above were recalculated):
- Pathologic Complete Response (pCR) Rate
- Relapse-free Survival (RFS), as Assessed by the Investigator According to European Association for the Study of the Liver (EASL) and/or Response Evaluation Criteria in Solid Tumors, Version 1.1 (RECIST v1.1)
- Event-free Survival (EFS), as Assessed by the Investigator According to EASL and RECIST v1.1
- Overall Survival (OS)
- OS Rate at 6 Months, 12 Months, and 18 Months
- Objective Response Rate (ORR), as Assessed by the Investigator According to RECIST v1.1
- ORR, as Assessed by the Investigator According to Hepatocellular Carcinoma-Specific Modified Response Evaluation Criteria in Solid Tumors (HCC mRECIST)
- Percentage of Participants Downstaged to Within Milan Criteria (for Participants Beyond Criteria at Randomization)
9. Statistical Interpretation
For atezolizumab plus bevacizumab, the reported MPR rate was 0.0%, with a 95% confidence interval from 0.0% to 11.57%. For atezolizumab plus bevacizumab plus tiragolumab, the rate was 3.3%, with a 95% confidence interval from 0.08% to 17.22%. These are estimates for the efficacy-evaluable population at surgery, not estimates of a population-wide treatment effect established with high precision.
The reported risk-difference estimate for the three-drug cohort relative to atezolizumab plus bevacizumab was 0.033, with a 95% confidence interval from -0.031 to 0.098. The interval includes zero, so the posted estimate is compatible with no difference as well as with differences in either direction under the interval's statistical assumptions. The unadjusted p-value was 0.313. That value is not the probability that the treatments are equivalent or that the null hypothesis is true; it describes how unusual the observed data would be under a specified null model, subject to the analysis assumptions.
The point estimates are numerically different, but the confidence intervals are broad relative to the small observed response percentages. A small number of responses can produce substantial uncertainty in a proportion. Thus the results do not establish a precise separation between the cohorts. Nor does the absence of a clear separation establish that the regimens have the same effect: a study can fail to distinguish effects because its estimates are uncertain.
The comparison is between treatment cohorts in a randomized platform study, but the registry record does not report detailed randomization procedures, a formal hypothesis-testing plan, or adjusted analyses. No sponsor-reported analysis entries are listed for the primary endpoint. The estimates should therefore be read as posted cohort summaries and an unadjusted comparison, rather than as a fully documented confirmatory analysis.
These results address MPR at surgery only. They do not by themselves answer whether participants had longer relapse-free or event-free survival, or whether overall survival differed. Such outcomes are listed among the study's secondary outcomes, but their numerical results are not reported in the registry record. Similarly, the posted MPR result cannot establish the effect of the closed tobemstomig plus bevacizumab arm, for which efficacy estimates are not reported.
Interpretation also depends on the endpoint's defined population and handling of surgery. The efficacy-evaluable population included participants who received at least one dose, and those who did not proceed to surgery were counted as nonresponders. This rule is explicit, but the registry does not give the number not reaching surgery or describe the reasons. Readers therefore cannot determine how much the nonresponder classification affected each cohort's result. The appropriate conclusion is limited to the reported response estimates and their uncertainty, without extending them to unreported outcomes or treatment arms.
10. Statistical Methods Explained
What does a response percentage describe?
A response percentage is the proportion of participants in the stated analysis population who meet the endpoint definition. Here, MPR required no more than 10% residual viable tumor in the primary tumor bed at surgery. The registry reports percentages rounded off. The denominator is the efficacy-evaluable population, defined as participants who received at least one dose for the assigned regimen. Participants who did not proceed to surgery were treated as nonresponders for MPR, so the endpoint definition incorporates both response assessment and the rule for participants without a surgical specimen.
What does the confidence interval add?
A confidence interval conveys uncertainty around an estimated response percentage or difference. In general, it describes a range of values that are compatible with the data under the interval procedure and its assumptions. A wide interval indicates that the estimate is not very precise; a narrow interval indicates greater precision. The registry specifies 95% confidence intervals for MPR but does not report the method used to construct them. The interval should therefore be interpreted as reported, without assuming a particular calculation method.
For the primary comparison, the registry reports a risk difference of 0.033 and a 95% confidence interval from -0.031 to 0.098. A risk difference expresses the difference in response proportions between cohorts on the probability scale. Because its interval includes zero, the results do not pin down the direction of a difference. The cohort-specific intervals likewise indicate that the estimates are uncertain. Overlap or non-overlap of separate cohort intervals is not a substitute for the interval calculated directly for the contrast; the direct risk-difference interval is the relevant reported estimate for that comparison.
What does the unadjusted p-value mean?
The registry reports an unadjusted p-value of 0.313 for the primary contrast. In general, a p-value describes how compatible the observed data are with a specified null hypothesis under the assumptions of a statistical test. It does not measure the probability that the null hypothesis is true, the size of an effect, or the likelihood that the result will be reproduced. The registry does not report the test procedure or a complete formal analysis plan. The p-value is therefore best read alongside the estimated risk difference and its confidence interval, not as a standalone decision rule.
“Unadjusted” means the reported value is not described as adjusted for multiple comparisons. The registry record does not report a multiplicity procedure for the many outcomes and treatment comparisons associated with a platform study. Without a stated procedure, the p-value should not be treated as controlling the chance of false-positive findings across the full set of outcomes.
Why can small event counts produce broad intervals?
General statistical principle: estimates based on few observed events tend to have greater sampling uncertainty than estimates based on many events, especially when the outcome is a proportion near zero. The primary results show low response percentages, so the confidence intervals extend over a range of possible values. This uncertainty is visible even though the cohort denominators are the same in the primary results. The reported intervals, rather than the point estimates alone, convey how much the data constrain plausible response rates and between-cohort differences.
How should the secondary-outcome comparisons be read?
The posted secondary comparisons include binary percentages and risk differences, with confidence intervals. For example, the RECIST v1.1 ORR comparison reports 10.0% in each cohort and a risk-difference estimate of 0.0, with a 95% confidence interval from -0.152 to 0.152. The HCC mRECIST ORR comparison reports a risk-difference estimate of -0.067, with a 95% confidence interval from -0.255 to 0.121. These estimates are descriptive comparisons with substantial uncertainty; the registry lists no reported analyses for these secondary endpoints.
The downstaging outcome applies only to participants beyond Milan criteria at randomization. Its reported comparison uses 18 participants in the atezolizumab plus bevacizumab cohort and 16 in the three-drug cohort. A denominator specific to that subgroup should not be confused with the full cohort size used for MPR. The registry reports a risk-difference estimate and interval for this outcome, but does not report an adjusted analysis.
What does the planning illustration represent?
The sample-size planning illustration is based on an observed secondary effect and stated assumptions: a two-sided significance level of 0.05, equal allocation, and a normal approximation. Its rows describe planning for 100% of the observed effect, planning for 75% of the observed effect, and planning for 50% of the observed effect. For 80% power, the illustration gives 490 participants per arm for the full observed effect, 906 per arm for 75% of that effect, and 2,115 per arm for 50% of that effect. For 90% power, the corresponding figures are 655, 1,212, and 2,830 participants per arm.
These are hypothetical planning illustrations, not the study's original sample-size justification and not observed power. The assumptions that determined the original study size are not reported in the registry record. General statistical principle: power calculations depend on the effect size, outcome variability, analysis method, and design assumptions. A calculation based on an observed effect is especially sensitive to uncertainty in that effect and should not be read as evidence that the trial had a particular probability of detecting a true effect.
The registry also lists 20 outcome measures and 19 posted secondary outcomes, while reporting no analysis entries for the primary outcome and no reported analyses for the secondary comparisons described. The specific methods used to handle missing observations beyond the MPR nonresponder rule, the planned model, and any interim analyses are not reported in the registry record. No assumption about those procedures should be inferred from the reported percentages alone.
11. Confidence Intervals and Precision
The MPR estimates are imprecise. The atezolizumab plus bevacizumab cohort had a reported rate of 0.0%, with a 95% confidence interval from 0.0% to 11.57%. The atezolizumab plus bevacizumab plus tiragolumab cohort had a rate of 3.3%, with an interval from 0.08% to 17.22%. These intervals show that the data remain compatible with a range of underlying response percentages, rather than locating each cohort's response rate narrowly.
The interval for the primary risk difference, -0.031 to 0.098, includes zero and extends to estimates in either direction. This means the posted data do not precisely distinguish a difference from no difference. It also means that the estimate should not be interpreted as proof of equivalence: an equivalence conclusion would require a defined margin and an analysis designed for that question, neither of which is reported in the registry record.
Confidence intervals are useful because they put uncertainty alongside the point estimate, but they do not account for every limitation. They do not resolve questions about selection into the efficacy-evaluable population, the nonresponder rule for participants who did not undergo surgery, or unreported analysis choices. The registry does not state the interval-construction method, so further claims about its exact coverage properties cannot be made from the record.
Other posted binary comparisons also have broad intervals. For instance, the RECIST v1.1 ORR risk-difference interval ranges from -0.152 to 0.152, and the HCC mRECIST ORR interval ranges from -0.255 to 0.121. These intervals include zero and leave considerable uncertainty about the direction and size of the contrasts. The downstaging comparison is based on smaller subgroup denominators than the main MPR comparison, which is relevant when considering its precision; the registry does not report an adjustment that would make these outcomes more precise.
12. Multiplicity, Missing Data and Analysis Populations
The registry lists 20 outcome measures, including the primary MPR endpoint and a range of secondary efficacy and safety outcomes. A multiplicity plan would explain how the analysis accounts for the number of outcomes, comparisons, or treatment arms when interpreting statistical evidence. Such a plan is not reported in the registry record. The reported p-value is explicitly unadjusted, and no adjustment method or hierarchy of hypotheses is reported. Thus, the p-value should not be interpreted as if it controlled false-positive risk across all study questions.
The registry reports no analysis entries for the primary result and no reported analyses for the secondary comparisons described. It does provide cohort percentages, confidence intervals, and unadjusted comparisons, but does not state a complete statistical analysis plan, a model for the primary analysis, or whether any interim analyses were conducted. Those details are not reported in the registry record, and their absence should not be filled in by assumption.
The stated efficacy-evaluable population includes participants who received at least one dose of assigned treatment. For MPR, participants who did not proceed to surgery were counted as nonresponders. This rule gives a specified outcome classification for that circumstance, but the number of participants who did not undergo surgery and the reasons are not reported. The handling of missing data for other endpoints, withdrawals, and incomplete assessments is also not reported in the registry record. It is therefore not possible to assess whether missingness differed across cohorts or how it might affect the secondary estimates.
The registry's total enrollment is 62, while the primary efficacy estimates use 30 participants per reported cohort. Safety denominators are 31 and 30. The record does not explain the differences between these outcome-specific denominators and total enrollment, nor does it provide a participant flow accounting for all randomized participants. Those population details matter when interpreting which participants contribute to each result. No claim about an intention-to-treat analysis, exclusions, or imputation should be inferred from the available descriptions.
13. Safety Results
The registry reports serious adverse events as affected participants divided by participants at risk.
| Arm | Participants with serious adverse events | Participants at risk | Percentage |
|---|---|---|---|
| Atezolizumab + Bevacizumab | 14 | 31 | 45.2% |
| Atezolizumab + Bevacizumab + Tiragolumab | 10 | 30 | 33.3% |
Deaths
| Arm | Deaths (participants affected of at risk) |
|---|---|
| Atezolizumab + Bevacizumab | 2 of 31 |
| Atezolizumab + Bevacizumab + Tiragolumab | 3 of 30 |
Other (non-serious) adverse events
Participants with at least one non-serious event at or above the registry's 5% reporting threshold.
| Arm | Participants affected of at risk |
|---|---|
| Atezolizumab + Bevacizumab | 24 of 31 |
| Atezolizumab + Bevacizumab + Tiragolumab | 19 of 30 |
Most frequent serious adverse event terms
| Event term | Organ system | Atezolizumab + Bevacizumab | Atezolizumab + Bevacizumab + Tiragolumab |
|---|---|---|---|
| Abdominal pain | Gastrointestinal disorders | 0 of 31 | 2 of 30 |
| Small intestinal obstruction | Gastrointestinal disorders | 1 of 31 | 1 of 30 |
| Hepatic failure | Hepatobiliary disorders | 1 of 31 | 1 of 30 |
| Pneumonia | Infections and infestations | 1 of 31 | 1 of 30 |
| Sepsis | Infections and infestations | 1 of 31 | 1 of 30 |
| Wound dehiscence | Injury, poisoning and procedural complications | 1 of 31 | 1 of 30 |
| Pneumothorax | Respiratory, thoracic and mediastinal disorders | 1 of 31 | 1 of 30 |
| Blood loss anaemia | Blood and lymphatic system disorders | 1 of 31 | 0 of 30 |
| Febrile neutropenia | Blood and lymphatic system disorders | 1 of 31 | 0 of 30 |
| Cardiac arrest | Cardiac disorders | 1 of 31 | 0 of 30 |
Most frequent non-serious adverse event terms
| Event term | Organ system | Atezolizumab + Bevacizumab | Atezolizumab + Bevacizumab + Tiragolumab |
|---|---|---|---|
| Hypertension | Vascular disorders | 5 of 31 | 4 of 30 |
| Diarrhoea | Gastrointestinal disorders | 2 of 31 | 5 of 30 |
| Fatigue | General disorders | 3 of 31 | 4 of 30 |
| Pruritus | Skin and subcutaneous tissue disorders | 3 of 31 | 4 of 30 |
| Asthenia | General disorders | 3 of 31 | 3 of 30 |
| Pyrexia | General disorders | 4 of 31 | 2 of 30 |
| Nausea | Gastrointestinal disorders | 3 of 31 | 2 of 30 |
| Infusion related reaction | Injury, poisoning and procedural complications | 1 of 31 | 4 of 30 |
| Alanine aminotransferase increased | Investigations | 2 of 31 | 3 of 30 |
| Dysphonia | Respiratory, thoracic and mediastinal disorders | 2 of 31 | 3 of 30 |
Serious events with the highest arm-level rate: abdominal pain (6.7%), visual impairment (3.3%), small intestinal obstruction (3.3%), hepatic failure (3.3%), liver abscess (3.3%).
The reported safety figures use denominators of 31 participants for atezolizumab plus bevacizumab and 30 for atezolizumab plus bevacizumab plus tiragolumab. Serious adverse events were reported for 14 participants in the former cohort and 10 in the latter. Other adverse events were reported for 24 and 19 participants, respectively. The registry also lists 2 deaths in the first cohort and 3 in the second. These are counts from the posted safety results; they should not be treated as evidence that a particular treatment caused an event or death.
The listed serious adverse event terms include abdominal pain at 6.7%, visual impairment at 3.3%, small intestinal obstruction at 3.3%, hepatic failure at 3.3%, liver abscess at 3.3%, and pneumonia at 3.3%. The registry does not specify in the provided safety summary which cohort each listed term belongs to, so the terms should not be assigned to an individual cohort. These figures also do not establish comparative safety: the record does not report a formal between-cohort safety analysis, adjustment, or detailed event narratives.
Safety counts are most appropriately read with their denominators and in the context of the registry's reporting. The record does not report the timing, duration, severity breakdown beyond the serious-event classification, or causal attribution for each listed event. The observed counts therefore describe what was recorded, not a causal conclusion or a complete characterization of risk.
14. What the Result Does, and Does Not, Mean
The primary result describes the proportion of efficacy-evaluable participants meeting a pathology-based response definition at surgery. It indicates that MPR was uncommon in both reported cohorts in these posted results, with a small numerical difference between the cohort estimates and broad uncertainty around the contrast. It does not establish that one regimen is superior, that the regimens are equivalent, or that either treatment has a particular effect on survival.
The primary endpoint has a defined interpretation: no more than 10% residual viable tumor in the primary tumor bed at surgical resection. It does not measure tumor response by imaging, recurrence after surgery, or survival. Participants who did not reach surgery were counted as nonresponders, so the result is tied to both response among resected tumors and the specified classification of those without surgery. The number and reasons for participants not proceeding to surgery are not reported in the registry record.
The study was randomized, but the posted comparison should still be read within the limits of the record. The randomization procedures, sample-size justification, analysis plan, multiplicity handling, and missing-data methods are not reported. The platform design also allowed arms and populations to change, and efficacy results for the closed tobemstomig plus bevacizumab arm are not reported. The estimates therefore cannot be extended to that arm or used to infer its performance.
Secondary outcomes provide different perspectives but do not change the limits of the primary evidence. Some numerical response and downstaging comparisons are reported, but their intervals are broad and no adjusted analyses are listed. Numerical results for the named survival and pathologic complete response outcomes are not reported in the registry record. Those outcomes cannot be assumed to follow the pattern of MPR or the reported ORR comparisons.
The results are descriptive evidence from the reported cohorts, not a recommendation about treatment. They neither prove the absence of a meaningful difference nor demonstrate a benefit. The appropriate interpretation is that the posted data estimate MPR at surgery with substantial uncertainty, and that the broader clinical questions remain unanswered by the numerical results reported in the registry record.
15. Why This Trial Matters Statistically
This trial raises several statistical issues common to early-phase, randomized platform studies. It combines a response endpoint assessed at surgery with a treatment pathway in which surgery may occur before the maximum number of planned cycles. The endpoint definition also classifies participants who do not proceed to surgery as nonresponders. That makes the analysis rule clinically and statistically consequential: the outcome is not simply a pathology result among those operated on, but a cohort-level response measure under a defined treatment and surgery pathway.
The low MPR percentages make uncertainty especially important. A point estimate alone can make cohorts appear clearly different or similar, while the confidence intervals reveal that the data support a much wider range of plausible values. The risk-difference interval includes zero and spans both negative and positive estimates. Reading the estimate together with its interval avoids treating a small numerical difference as a settled conclusion.
The platform setting adds interpretive complexity. Arms may open or close and the participant population may be modified; the registry describes this flexibility but does not report the specific operational changes or an analysis plan for adapting comparisons. A clear account of concurrent controls, allocation, analysis populations, and multiplicity would help readers assess how the platform structure affects inference. Those details are not reported in the registry record.
Finally, the posted secondary outcomes range from imaging response to longer-term outcomes. Their relevance depends on the question being asked, and they do not all measure the same underlying construct. The absence of reported numerical results for several named outcomes, together with unadjusted secondary comparisons, reinforces the need to distinguish exploratory patterns from confirmatory evidence. This is why estimates, intervals, populations, and analysis plans matter as much as the nominal endpoint result.
16. 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.
- Participant counts were reconstructed from posted percentages and arm sizes, so small rounding differences are possible.
The primary efficacy evidence is limited to two reported treatment cohorts and an efficacy-evaluable population of 30 participants per cohort. Although total enrollment is 62, the registry does not provide a participant flow accounting for all enrollees or explain why the primary efficacy denominators differ from total enrollment. Safety denominators are 31 and 30, further underscoring that the contributing populations vary by outcome.
The registry defines non-surgery as a nonresponse for MPR but does not report how many participants did not proceed to surgery or why. The effect of this rule on the cohort percentages cannot be separated from the pathology response results. Methods for missing data on other outcomes are not reported. The registry also does not provide detailed randomization procedures, the original sample-size justification, a statistical analysis plan, or the methods used to construct the reported confidence intervals.
The study is listed as terminated, but the reason for termination is not reported in the registry record. The platform description permits arms to open or close and populations to change, but the record does not describe what changes occurred. The tobemstomig plus bevacizumab arm is listed as enrollment closed, without efficacy estimates in the posted results. Consequently, the reported findings cannot be generalized to that cohort.
Several secondary outcomes are listed without numerical results, including pCR and survival outcomes. The numeric secondary comparisons that are reported are unadjusted, and no multiplicity strategy is described. These limitations constrain both the strength of secondary inferences and the ability to connect MPR to longer-term outcomes. The record also provides limited detail about safety event timing, attribution, and cohort assignment for the listed serious adverse event terms.
17. Related Tutorials
18. Related Statistical Calculators
19. Sources
- ClinicalTrials.gov: A Study Evaluating The Efficacy and Safety of Neoadjuvant Immunotherapy Combinations in Patients With Surgically Resectable Hepatocellular Carcinoma — NCT05908786.
- Registry record last updated 2026-09-15; status last verified 2026-08.
- Results first posted 2026-09-15.
- 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.
Continue through the Clinical Biostats statistical pathway
Use this trial's endpoint as a starting point for deeper study of effect measures, confidence intervals, and sample-size planning.
20. Record Summary
NCT05908786 is a Phase Ib/II, randomized, open-label, multicenter, parallel-group platform study of neoadjuvant immunotherapy combinations in surgically resectable hepatocellular carcinoma. Total enrollment is listed as 62, and study status is terminated. The registry describes treatment with up to three cycles before surgery or until unacceptable toxicity, whichever occurs first.
The primary endpoint was MPR at surgery, up to 15 weeks, in participants who received at least one dose of assigned treatment. The reported rates were 0.0% for atezolizumab plus bevacizumab and 3.3% for atezolizumab plus bevacizumab plus tiragolumab. The risk-difference estimate was 0.033, with a 95% confidence interval from -0.031 to 0.098 and an unadjusted p-value of 0.313. The interval indicates substantial uncertainty around the contrast.
Some secondary response and downstaging results and safety counts are reported. Numerical results for several named secondary outcomes, the reason for termination, and important details of the statistical plan are not reported in the registry record.