A Study of Pembrolizumab and Bavituximab in Patients With Advanced Hepatocellular Carcinoma: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study of Pembrolizumab and Bavituximab in Patients With Advanced Hepatocellular Carcinoma (NCT03519997), a na phase 2 study. The primary outcome is Overall Response Rate.

2. Clinical Question

This study asked whether tumor responses would be observed among patients treated with the combination in this clinical setting. It was a phase II therapeutic study conducted at multiple sites, with an enrollment of 35 recorded in the registry. Because it used a single group rather than random assignment to treatment and control, its primary question is descriptive: what response outcome was recorded for evaluable treated patients? It cannot, by its design alone, establish how that outcome compares with what would have occurred under another treatment or no study treatment.

The clinical population matters to interpretation. The registry describes advanced HCC not amenable to surgical or locoregional therapies and excludes prior systemic therapy for advanced HCC. The findings, if reported, would therefore be most directly relevant to that described setting, rather than to all people with HCC or to patients with earlier-stage disease. The record does not report a more detailed primary-endpoint population or the number of evaluable patients.

3. Trial Design

The registry identifies this as a phase II, non-randomized, open-label, multi-site therapeutic study of pembrolizumab and bavituximab. Its design is listed as single-group, with no masking. The listed study group received pembrolizumab 200 mg IV every 3 weeks and bavituximab 3mg/kg IV weekly. The registry records an enrollment of 35 and a completed status.

Single-group studies can describe outcomes among participants who receive a regimen, but they do not create a randomized comparison. The allocation field is recorded as NA, and the registry explicitly identifies the study as non-randomized. Accordingly, there is no like-for-like control group within this study to separate treatment-associated outcomes from the natural course of disease, participant characteristics, or other influences. Open-label status also means masking was not used; the record does not describe any blinded assessment procedure.

The registry records the study start as 2018-04-26 and primary completion as 2023-04-21. It describes the intended population as patients with locally advanced or metastatic HCC not amenable to surgery or locoregional therapies, and states that prior systemic therapy for advanced HCC was not permitted. The study was framed as first-line treatment for advanced HCC.

The primary outcome was overall response rate according to RECIST 1.1, with a stated timeframe of 36 months. Overall survival and adverse events according to CTCAE are listed as secondary outcomes. The record does not report a protocol-level statistical plan, a sample-size justification, interim analyses, or a detailed account of the analysis population. It also does not provide a primary response estimate. Thus the design and outcome definitions can be described, but an efficacy result cannot be reconstructed from the registry information.

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

Reasons for not completing this period:

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

6. Primary Endpoint

The registry lists the primary outcome as Overall Response Rate, measured over 36 months.

The primary endpoint is overall response rate (ORR). The registry defines it as the number of complete or partial responses according to RECIST 1.1 divided by the number of evaluable patients treated with pembrolizumab and bavituximab. The outcome measure is listed as a count of participants, while the clinical definition describes a response proportion. The stated timeframe is 36 months.

In general, response endpoints classify whether a participant’s disease meets prespecified criteria for response. RECIST 1.1 provides standardized criteria for categorizing tumor response, which can make response assessment more consistent across participants and study sites. A response endpoint can be useful in a phase II study because it describes the frequency of observed tumor shrinkage meeting those criteria. It is not, by itself, a comparison with another treatment, and it does not answer every question about the duration of response, survival, symptoms, or harms.

The registry record does not report how many complete or partial responses occurred, the number of evaluable participants, or the resulting ORR. It also does not report a confidence interval for ORR. The recorded enrollment should not be substituted for the evaluable denominator: the endpoint definition specifically refers to evaluable patients treated, and that count is not reported in the registry record. Without the numerator and denominator, neither the response proportion nor its precision can be assessed from the record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry identifies a single-group study and names overall response rate as the primary endpoint. It does not report a statistical analysis for that endpoint. The explanations below describe general statistical concepts relevant to interpreting a response endpoint and a single-group trial; they are not claims about analyses performed in this study.

What does a response proportion represent?

Generally, a proportion is a numerator divided by a denominator. Here, the endpoint definition identifies the numerator as participants with complete or partial response under RECIST 1.1 and the denominator as evaluable patients treated with the combination. A response count alone would not tell the full story unless the denominator were also known. Likewise, enrollment is not necessarily the same as the endpoint denominator, because the definition specifically uses evaluable treated patients.

A response proportion is an estimate based on the participants observed in a study. If the study were repeated in a similar population, the observed proportion could vary. A confidence interval is a way to express uncertainty around an estimate: it gives a range of values compatible with the data under the method’s assumptions. A narrower interval generally indicates more precision, while a wider interval indicates less precision. The registry record does not report an ORR estimate or its confidence interval, so neither response frequency nor precision can be evaluated.

Why does the single-group design matter?

In general, random assignment helps make treatment groups comparable in expectation, including with respect to factors that may affect outcomes. A single-group study has no concurrent randomized control group to provide that comparison. Outcomes can be described in the treated group, but differences from an external expectation or historical result may reflect differences in patient eligibility, disease status, follow-up, assessment, or other features rather than treatment alone. This limitation applies even when a response measure has a standardized definition.

The registry lists allocation as NA and describes the study as non-randomized and open-label. The record does not report masking. General statistical interpretation should therefore avoid treating any within-study outcome as a randomized treatment contrast. There is no like-for-like control comparison in the recorded design.

What does a confidence interval add?

In general, a point estimate gives a single summary, while an interval conveys sampling uncertainty under a statistical model. It does not account automatically for every possible source of bias, nor does it guarantee that the true value lies within the interval. For a response proportion, uncertainty depends in part on the amount of information in the numerator and denominator. Since neither is reported for the primary endpoint, an interval cannot be interpreted from the registry entry.

A hypothesis test, when specified, addresses a particular statistical question under its assumptions. A p-value is not the probability that a treatment works, nor is it a measure of the size or clinical importance of an outcome. The registry record does not list a primary endpoint test or p-value. No statistical significance claim is warranted from the endpoint definition alone.

How should planning calculations be read?

In general, sample-size calculations describe planning scenarios under stated assumptions; they do not report observed study outcomes. The planning assumptions associated with this record describe a two-sided significance level of 0.05, equal allocation between two arms, a normal approximation, and an effect set as a stated fraction of the observed effect. Those assumptions describe a hypothetical planning framework, not the allocation used in the single-group trial.

A planning illustration is not observed power. Observed power is not a substitute for reporting an estimate and its uncertainty, and a planning calculation cannot establish that the study detected an effect. The record does not report a sample-size justification for the trial’s design. The planning assumptions should not be read as evidence that participants in this study were assigned equally between two groups; the registered design is single-group.

Why can event counts differ from risk estimates?

In general, an event count indicates how many events or affected participants were recorded, whereas a risk estimate requires a defined population at risk and a clearly specified observation period. Counts can be useful descriptions, but they do not by themselves describe timing, duration, recurrence, or comparative risk. The posted safety data give counts and terms, but the registry record does not provide a comparative safety analysis. Without a control group, the counts cannot quantify an excess risk attributable to the combination.

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("Pembro + Bavi")
n      <- c(28)
events <- c(9)
for (i in seq_along(n)) {
  ci <- binom.test(events[i], n[i])$conf.int          # Clopper-Pearson exact interval
  cr <- qbeta(c(0.025, 0.975), 1 + events[i], 1 + n[i] - events[i])   # Bayesian, uniform Beta(1,1) prior
  wi <- (function(x, n) { p <- x / n; z <- qnorm(0.975); d <- 1 + z^2 / n
    c((p + z^2 / (2 * n) - z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d, (p + z^2 / (2 * n) + z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d) })(events[i], n[i])
  cat(sprintf("ARM|%s|%d|%d|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f\n", arm[i], n[i], events[i], 100 * events[i] / n[i],
              100 * ci[1], 100 * ci[2], 100 * wi[1], 100 * wi[2], 100 * cr[1], 100 * cr[2]))
}

ARM  Pembro + Bavi  28  9  32.1  15.9  52.4  17.9  50.7  17.9  50.8

11. Multiplicity, Missing Data and Analysis Populations

The registry lists overall response rate as the primary outcome and overall survival and adverse events according to CTCAE as secondary outcomes. It also records 3 outcome measures and no analyses. The record does not report a statistical analysis plan describing how these outcomes were analyzed, how many hypotheses were tested, or whether any adjustment for multiple comparisons was planned.

In general, when several outcomes or statistical tests are examined, the chance of at least one apparently notable result can increase. Multiplicity procedures are methods for controlling or describing that issue. Their relevance depends on the prespecified hypotheses and analysis strategy. Since those details are not reported in the registry record, it is not possible to determine how multiplicity was handled or whether formal hypothesis testing was intended.

The record also does not report missing-data methods, the number of patients excluded from the primary analysis, or the definition of the evaluable population beyond the endpoint’s reference to evaluable patients treated. That matters because the ORR denominator is not necessarily identical to enrollment. The registry does not report the number of evaluable patients, the number with missing response assessments, or how any such cases were treated analytically.

It is therefore not possible to assess whether missing outcome information could have influenced the response estimate, or whether the analysis followed an intention-to-treat, per-protocol, or other specified approach. Those analysis-population details are not reported in the registry record. The absence of these details should be treated as a reporting limitation, not as evidence that missing data or multiplicity were handled in a particular way.

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: hepatic failure (5.7%), acute kidney insufficiency (5.7%), weakness (5.7%), sepsis (5.7%), colitis (2.9%).

The registry reports serious adverse events affecting 7 of 35 participants at risk in the Pembro + Bavi group. It reports other adverse events affecting 33 of 35 at risk and lists 14 deaths. The top serious adverse event terms and their recorded percentages are hepatic failure, 5.7%; acute kidney insufficiency, 5.7%; weakness, 5.7%; sepsis, 5.7%; colitis, 2.9%; and cough, 2.9%.

These entries are descriptive safety figures. They should not be interpreted as proof that the combination caused an event, and the single-group design provides no randomized safety comparator. The record does not report event timing, duration, recurrence, attribution, or a detailed account of exposure and follow-up. It also does not state whether participants could be represented under more than one serious-event term. The listed term percentages should therefore be read as recorded values, not combined or treated as mutually exclusive categories.

Serious adverse events and other adverse events are separate registry groupings; a participant may not be assumed to fit only one category. The counts and percentages describe the posted safety record, but by themselves do not establish comparative safety or explain the listed deaths. The registry record does not report a causal analysis of deaths.

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

This trial contributes a description of a phase II, open-label, single-group study of pembrolizumab plus bavituximab in advanced HCC. Its primary endpoint was overall response rate under RECIST 1.1 among evaluable treated patients, with a stated timeframe of 36 months. The registry record also includes safety counts and selected serious adverse event terms.

The primary response result is not reported in the registry record. Without the response count, evaluable denominator, and an uncertainty estimate, readers cannot determine the observed response frequency or its precision. The study’s completed status and enrollment entry do not substitute for those missing endpoint results. Nor can the response result be inferred from the safety information.

Even if the response result were available, the single-group, non-randomized design would support a description of outcomes among the studied participants rather than a randomized comparison against another treatment. The study cannot by itself show whether the combination produced better outcomes than a control, because no concurrent control arm is recorded. Differences from other studies could also reflect differences in populations, methods, assessment, or follow-up.

The safety counts are also descriptive. They do not establish that the combination caused the reported events, and the registry record does not supply a comparative safety estimate or a causal analysis of the deaths listed. General conclusions about benefit or harm would require more complete outcome reporting and an appropriate comparative design or other evidence suited to the question.

The appropriate interpretation is therefore bounded: the registry identifies the population, regimen, design, primary endpoint definition, and selected safety results. It does not provide the primary efficacy result or the statistical analyses needed to assess its magnitude and precision. No conclusion about treatment effectiveness follows from the endpoint definition or study completion alone.

14. Why This Trial Matters Statistically

The trial is statistically informative as an example of the distinction between defining an outcome and reporting its result. The registry specifies a clinically interpretable endpoint—complete or partial response under RECIST 1.1 among evaluable treated patients—but does not report the numerator, denominator, estimate, or uncertainty interval. Readers can identify what the investigators intended to measure without being able to assess what the study found on its primary endpoint.

Its design also makes the comparison question important. A single-group study can record outcomes in a defined patient population, but it does not provide the protection against confounding that random assignment can offer. The lack of a concurrent control limits causal interpretation, including for safety. That is not a flaw in the arithmetic of an outcome; it is a limit on the kinds of conclusions the design can support.

The record’s safety entries illustrate another statistical distinction: event counts and event terms are not the same as comparative risk estimates. The serious-event and other-event counts describe what was recorded in the listed group. Without a comparator and fuller information on timing, attribution, exposure, and follow-up, these numbers cannot establish whether risks differ from those under another approach.

Finally, the registry lists no analyses and does not report a statistical plan, missing-data approach, or multiplicity strategy. Those omissions constrain independent evaluation of the primary endpoint and secondary outcomes. The study therefore matters statistically as a case in which a clearly named endpoint and a completed study record still leave key questions about the observed efficacy result and its precision unanswered.

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 specific to this trial constrain interpretation of the registry record. First, the primary endpoint is defined, but its observed result is not reported. The record gives neither the number of complete or partial responses nor the number of evaluable patients used as the denominator. It consequently provides no response rate or confidence interval for assessment.

Second, the non-randomized, single-group design includes no concurrent control group. This prevents a like-for-like randomized comparison and limits the ability to distinguish outcomes associated with treatment from outcomes related to disease course, participant characteristics, or other influences. The open-label design and lack of reported masking add to the need for caution when interpreting assessments.

Third, details needed to understand the primary analysis are not reported in the registry record. These include a statistical plan, the analysis population, missing-data methods, interim analyses, and sample-size justification. The endpoint definition refers to evaluable treated patients, but the record does not report how evaluability was determined or how missing assessments were handled.

Fourth, the safety data are limited to the posted counts and listed terms. The registry does not report a randomized safety comparison, detailed event timing or attribution, or a causal analysis of deaths. The record’s completed status and enrollment field do not resolve these limitations. These points restrict what can be concluded from the registry information; they do not establish what additional information may exist outside the registry record.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: A Study of Pembrolizumab and Bavituximab in Patients With Advanced Hepatocellular Carcinoma — NCT03519997.
• Registry record last updated 2026-07-08; status last verified 2026-06.
• Results first posted 2024-07-17.
• 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 identifies NCT03519997 as a completed phase II study of pembrolizumab and bavituximab in patients with advanced hepatocellular carcinoma. It describes a non-randomized, open-label, multi-site, single-group design and records an enrollment of 35. The listed regimen was pembrolizumab 200 mg IV every 3 weeks plus bavituximab 3mg/kg IV weekly.

The primary outcome was overall response rate according to RECIST 1.1 among evaluable treated patients, with a stated timeframe of 36 months. Overall survival and adverse events according to CTCAE are listed as secondary outcomes. The record reports selected safety counts and serious-event terms, but does not report the primary response result, its denominator, or a statistical analysis.

The study population was described as patients with locally advanced or metastatic HCC not amenable to surgical or locoregional therapies, without prior systemic therapy for advanced HCC. The registry does not report the statistical plan, missing-data approach, or primary endpoint precision.