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 Pembrolizumab in Advanced Hepatocellular Carcinoma as Second-line Treatment After Failure of Sorafenib (NCT03163992), a na phase 2 study. The primary outcome is Overall Response Rate.
| Feature | Registry value |
|---|---|
| ClinicalTrials.gov identifier | NCT03163992 |
| Phase | Phase 2 |
| Condition | Hepatocellular Carcinoma |
| Allocation | Not applicable (single group) |
| Design model | Single Group |
| Masking | None (open label) |
| Enrollment | 60 |
| Arms in the primary outcome | n/a |
| Lead sponsor | Samsung Medical Center |
| Results posted | Yes |
2. Clinical Question
Population
People with advanced hepatocellular carcinoma after failure of sorafenib were the intended population. The trial summary says an evaluable specimen for PD-L1 status had to be available and confirmed before enrollment, while enrollment was to proceed without regard to PD-L1 expression status. The registry record does not give further eligibility criteria or describe the enrolled participants’ characteristics.
Intervention
The intervention was pembrolizumab, listed as 200 mg every 3 weeks. The registry describes the study as a single-arm, single-center, open-label trial. It does not report whether participants received other treatments alongside pembrolizumab.
Comparator
There was no concurrent control group: the study model is single group, and allocation is listed as not applicable. The record therefore does not offer a randomized comparison with another treatment, placebo, or a usual-care group. Comparisons across PD-L1 status are mentioned in the endpoint description, but no separate arms or subgroup results are reported.
Primary question
The stated primary endpoint was overall response rate (ORR): the proportion of participants with a complete or partial response, excluding stable disease. The registry gives assessment by imaging and names RECIST criteria, although its descriptions differ on whether the version was RECIST 1.1 or RECIST v1.0. No observed ORR, response count, confidence interval, or statistical test is reported in the registry record. The study’s stated question can therefore be described, but its answer cannot be determined from the posted results.
3. Trial Design
This was a phase 2, single-group study of pembrolizumab for advanced hepatocellular carcinoma after failure of sorafenib. The registry identifies Samsung Medical Center as sponsor and describes the trial as single-center and open-label. Masking is listed as none, and allocation is not applicable. With no concurrent control group, the design does not support a randomized treatment comparison.
The planned enrollment is listed as 60, and the summary describes approximately 60 subjects. That figure is an enrollment target in the record; it does not establish how many participants enrolled, received treatment, completed follow-up, or contributed to the primary endpoint analysis. Those counts are not reported in the registry record.
The intervention is pembrolizumab (MK-3475), 200 mg every 3 weeks. The summary states that participants would be evaluated every 6 weeks, with a margin of 7 days, after the first drug administration during the first 12 months, and every 12 weeks, with the same margin, thereafter until progression was documented by radiologic imaging. The record does not report how closely the observed visits followed this schedule.
The trial is marked completed. Its listed start date is 2017-12-26 and its primary completion date is 2021-05-17. The primary endpoint timeframe is described as through study completion, an average of 24 months. The registry does not provide a detailed protocol, recruitment flow, follow-up distribution, or a separate account of the analysis population. It lists progression-free survival and overall survival as secondary outcomes, but does not report their results.
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
| Stage | Pembrolizumab |
|---|---|
| STARTED | 60 |
| COMPLETED | 1 |
| NOT COMPLETED | 59 |
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.
| Characteristic | Unit | Pembrolizumab |
|---|---|---|
| Participants analysed | 60 | |
| Age, Categorical: <=18 years | Participants | 0 |
| Age, Categorical: Between 18 and 65 years | Participants | 28 |
| Age, Categorical: >=65 years | Participants | 32 |
| Age, Continuous | years | 60 (37 to 84) |
| Sex: Female, Male: Female | Participants | 12 |
| Sex: Female, Male: Male | Participants | 48 |
| Ethnicity (NIH/OMB): Hispanic or Latino | Participants | 0 |
| Ethnicity (NIH/OMB): Not Hispanic or Latino | Participants | 60 |
| Ethnicity (NIH/OMB): Unknown or Not Reported | Participants | 0 |
| Race (NIH/OMB): American Indian or Alaska Native | Participants | 0 |
| Race (NIH/OMB): Asian | Participants | 60 |
| Race (NIH/OMB): Native Hawaiian or Other Pacific Islander | Participants | 0 |
| Race (NIH/OMB): Black or African American | Participants | 0 |
| Race (NIH/OMB): White | Participants | 0 |
| Race (NIH/OMB): More than one race | Participants | 0 |
| Race (NIH/OMB): Unknown or Not Reported | Participants | 0 |
| Region of Enrollment: South Korea | Participants | 60 |
6. Primary Endpoint
The registry lists the primary outcome as Overall Response Rate, measured over through study completion, an average of 24 months.
| Element | Registry entry |
|---|---|
| Result type | Count Of Participants |
| Measure of spread | not applicable |
| Unit | Participants |
| Endpoint classification | Binary Endpoint |
The primary endpoint is overall response rate (ORR), defined in the registry as the proportion of participants whose best response is a partial or complete response. Stable disease is not counted as a response. In general, this endpoint focuses on whether measurable tumors shrink or disappear according to prespecified imaging criteria; it does not capture every aspect of disease course or patient experience.
The trial summary identifies RECIST 1.1, while the primary outcome description says RECIST v1.0. The record does not explain this difference, so the applicable version cannot be resolved from the registry entry alone. The description refers to CT or MRI assessment and specifies complete response as disappearance of all target lesions and partial response as a decrease in the sum of the longest diameters of target lesions meeting the stated threshold. The endpoint description also states that overall response combines complete and partial response.
ORR is used in general as a direct measure of tumor response, expressed as a proportion rather than as time until an event. To interpret an estimate, readers would need the number of participants assessed, the number who responded, the response criteria and assessment rules, and a measure of uncertainty such as a confidence interval. Those result details are not reported in the registry record. The record names ORR as the primary endpoint but does not give an observed response estimate.
7. Why No Re-analysis Is Shown
8. Statistical Interpretation
The registry record establishes the trial question and design, but it does not provide the primary efficacy result. The primary endpoint is ORR, yet no response count, response proportion, confidence interval, p-value, or analysis result is reported. Consequently, the posted information does not establish the observed response rate or how precisely it was estimated. No conclusion about the effect of pembrolizumab can be drawn from the endpoint definition alone.
The planned enrollment is 60, but that is not an observed analysis denominator. The registry does not state how many participants were enrolled, evaluable for response, or included in an analysis. Without those details, even a response count would not be enough to interpret an ORR fully: the denominator and rules for handling participants without an assessment would also matter.
The single-group design means the study has no concurrent randomized control arm. Any observed response proportion would describe outcomes among participants treated in this trial, subject to the study’s eligibility, follow-up, and assessment procedures. It would not by itself isolate the treatment’s effect from other explanations, such as participant selection, disease course, or assessment-related factors. The record does not describe an adjusted or comparative analysis.
The endpoint description contains an unresolved difference in the RECIST version: the summary names RECIST 1.1 and the primary outcome description names RECIST v1.0. These are the registry’s own descriptions; the record does not state which version governed the final endpoint assessment. Readers should not assume a resolution or substitute one definition for the other.
The registry lists progression-free survival and overall survival as secondary outcomes, but gives no estimates, event counts, time-to-event analyses, or uncertainty measures for them. The presence of named secondary outcomes is not equivalent to reporting their results. Likewise, an adverse-event listing does not fill the gap in the efficacy result.
There are no analyses listed in the record. It does not state whether any interim analyses occurred, which statistical tests were used, whether response assessment was independent or investigator-based, or how missing assessments were handled. These details are not reported in the registry record. Sample-size illustrations, if considered alongside the posted material, are planning scenarios and not evidence of an observed treatment effect or of the precision achieved in this trial.
9. Statistical Methods Explained
Statistical interpretation begins by separating the study design from the endpoint result. The design describes how participants were assigned or followed; the endpoint specifies what outcome was to be measured. Here the registry describes a single-group, open-label phase 2 trial and names ORR as its primary endpoint. Neither the design label nor the endpoint definition supplies the missing response estimate.
What does a single-group result describe?
In general, a single-group study estimates outcomes in the group observed under the study conditions. For a binary endpoint such as response, each participant is classified according to whether the endpoint occurred, and the proportion is based on the response count divided by the relevant analysis denominator. That estimate depends on how response and the denominator are defined. It does not automatically estimate a difference from an untreated or alternative-treatment group.
Because this trial has no concurrent comparator, any observed ORR would be a description of the treated cohort rather than a randomized contrast. In general, randomization helps make treatment groups comparable with respect to measured and unmeasured factors in expectation. Without it, differences between groups cannot be used to attribute an outcome to treatment unless other design and analysis methods support that interpretation. No such comparative analysis is reported in the registry record.
How should a response proportion be read?
ORR combines participants with complete and partial responses and excludes those with stable disease, according to the registry’s definition. This makes the endpoint a binary classification for each participant under the response rules. The timing and criteria for assessments matter: imaging schedules determine when response can be observed, and criteria determine how changes in lesions are classified. The record describes imaging by CT or MRI and a follow-up schedule, but does not report the actual assessment schedule followed or the detailed endpoint analysis rules.
In general, a confidence interval around a proportion communicates the uncertainty associated with estimating an underlying response probability from a finite sample. A narrower interval indicates greater precision under the method’s assumptions; a wider interval indicates less precision. It is not a range containing a fixed share of individual outcomes, nor does it establish a causal treatment effect. No ORR estimate or confidence interval is reported, so precision cannot be assessed.
What does the missing analysis information mean?
The registry lists no analyses. In general, a statistical analysis plan may specify the analysis population, endpoint rules, treatment of missing observations, and method for estimating uncertainty. These choices can affect a binary endpoint. For example, classifying participants with no adequate post-baseline assessment as nonresponders differs from excluding them from the denominator. The registry does not state which approach, if either, was used. It also does not report an analyzed population or the number of participants with assessable response data.
Open-label status means masking is listed as none. In general, knowledge of treatment assignment can influence behavior, follow-up, and subjective decisions. Imaging-based response criteria provide structured rules, but the record does not report whether assessments were blinded or centrally reviewed. It would be inappropriate to assume either practice.
What can sample-size planning show?
In general, a sample-size calculation is a planning exercise based on assumptions about the effect to be detected, the significance level, allocation, and statistical method. It can illustrate the scale of a future study under those assumptions. It is not an observed result, does not show that a planned effect occurred, and is not a substitute for an estimate with a confidence interval. The planning description associated with the posted material specifies a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation, with the effect expressed as a fraction of an observed effect. Those assumptions do not describe the actual single-group design, and no sample-size illustration should be mistaken for observed power or a result from this trial.
For general context, over-dispersion is a concern when observed variability exceeds what a simple probability model assumes. Methods that account for it may produce different uncertainty estimates than methods that assume less variability. The registry does not identify an over-dispersion method or state that over-dispersion was relevant to the ORR analysis. It does not report a specific statistical test, confidence-interval method, adjustment, or model for the primary endpoint. These methods are not reported in the registry record.
The registry names progression-free survival and overall survival as secondary outcomes. In general, time-to-event outcomes require information about event timing and censoring, which occurs when observation ends before an event is recorded. A time-to-event estimate relies on follow-up and analysis rules that differ from a simple response proportion. No event counts, time-to-event estimates, or methods are reported for these outcomes, so their statistical results cannot be interpreted.
10. Multiplicity, Missing Data and Analysis Populations
The registry names ORR as the primary outcome and progression-free survival and overall survival as secondary outcomes. It does not report an analysis plan or explain how statistical testing across these outcomes was handled. When several outcomes or subgroup comparisons are tested, the chance of finding at least one apparently notable result can increase; in general, multiplicity procedures may be used to manage that issue. The record does not state whether any adjustment or testing hierarchy was planned or used.
The primary outcome description refers to biomarker-negative versus advanced hepatocellular carcinoma, while the trial summary says enrollment was to proceed without regard to PD-L1 expression status and required an evaluable specimen. The record does not provide a clear account of planned subgroup analyses, subgroup sizes, or results by PD-L1 status. It is not possible to determine whether the biomarker wording describes a planned comparison, a subgroup, or another aspect of the registry entry.
The registry does not report the primary endpoint denominator, the number of participants with missing or unevaluable response assessments, or how missing observations were handled. In general, missing outcome data can affect an estimated response proportion if the likelihood of missingness is related to response or participant characteristics. Different analysis rules can also change the denominator. No specific rule can be attributed to this trial from the registry record.
Analysis populations, including any intention-to-treat or evaluable population, are not reported. The record also does not state whether protocol deviations, treatment discontinuations, or incomplete follow-up affected inclusion in the primary analysis. These omissions limit assessment of both the validity and precision of any unreported endpoint result.
11. 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 |
|---|---|---|---|
| Pembrolizumab | 8 | 60 | 13.3% |
Deaths
| Arm | Deaths (participants affected of at risk) |
|---|---|
| Pembrolizumab | 35 of 60 |
Other (non-serious) adverse events
Participants with at least one non-serious event at or above the registry's 0% reporting threshold.
| Arm | Participants affected of at risk |
|---|---|
| Pembrolizumab | 24 of 60 |
Most frequent serious adverse event terms
| Event term | Organ system | Pembrolizumab |
|---|---|---|
| fatigue | General disorders | 8 of 60 |
Most frequent non-serious adverse event terms
| Event term | Organ system | Pembrolizumab |
|---|---|---|
| elevated liver enzyme | Hepatobiliary disorders | 6 of 60 |
| pruritis | Skin and subcutaneous tissue disorders | 5 of 60 |
| hyperbilirubinemina | Hepatobiliary disorders | 5 of 60 |
| diarrhea | Gastrointestinal disorders | 4 of 60 |
| constipation | Gastrointestinal disorders | 2 of 60 |
| hyperthyroidism | Endocrine disorders | 2 of 60 |
Serious events with the highest arm-level rate: fatigue (13.3%).
The safety listing reports one pembrolizumab group, with 60 participants at risk in the serious-adverse-event field and 8 affected. The separate other-adverse-event fields list 60 at risk and 24 affected; deaths are listed as 35. These are registry counts, not a comparison between treatment groups. The record does not clarify whether the event categories overlap, how events were attributed, or whether the listed deaths were considered treatment-related.
The registry also lists fatigue at 13.3% among the top serious terms. That figure is presented as a percentage in the record; a corresponding count and further detail about the term are not given. The listing does not establish the severity, timing, duration, or cause of each event. It should not be interpreted as a comparative risk estimate, since there is no control group.
In general, safety counts require context about the population observed, the event definitions, follow-up duration, and reporting rules. The registry does not provide a detailed safety analysis, exposure-adjusted rates, participant-level event histories, or a breakdown of deaths. The counts therefore describe what the record lists but do not resolve how events related to treatment or the underlying disease.
12. What the Result Does, and Does Not, Mean
The registry record documents a completed phase 2 study that asked about pembrolizumab in advanced hepatocellular carcinoma after sorafenib failure. It describes the intended population, regimen, follow-up schedule, and ORR as the primary endpoint. It does not report the primary endpoint result. As a result, the posted record cannot show how many participants responded, what the response proportion was, or how uncertain an estimate might be.
Even if an ORR were reported, a single-group study would describe outcomes in its enrolled participants rather than provide a randomized estimate of the difference between pembrolizumab and another treatment. The design would need to be considered alongside eligibility criteria, assessment procedures, missing data, and follow-up. The registry record does not provide enough analysis detail to evaluate those features fully.
The stated enrollment of 60 is not a substitute for an analyzed denominator, and the registry does not report the number assessed for the primary endpoint. Nor does it report results for the named secondary outcomes. The safety counts provide additional descriptive information but do not establish causation or comparative safety.
Thus, the record supports a description of the trial’s purpose and design, but not a conclusion about the magnitude or precision of its efficacy findings. It also does not resolve which RECIST version governed the primary endpoint: the summary names RECIST 1.1, whereas the endpoint description names RECIST v1.0. Readers should treat the absent results and unresolved methodological details as unknown rather than infer them from the trial’s completion status or from its endpoint definitions.
13. Why This Trial Matters Statistically
This trial illustrates why an endpoint definition and a clinical result are separate pieces of evidence. The registry identifies ORR and explains what constitutes response, but does not report an observed ORR or its uncertainty. Without that result, readers can understand the intended measurement but cannot judge the response estimate or its precision.
It also highlights the limits of a single-group design. Such studies can describe outcomes under a treatment regimen and may be useful in clinical research, but they do not provide a concurrent randomized comparison. A response proportion alone cannot separate treatment effects from differences in who enrolled, natural disease course, or other influences. This distinction is important when reading reports that discuss outcomes without a control group.
The record contains additional statistical uncertainties: no analysis is listed, the analysis population and missing-data handling are not described, and the endpoint descriptions differ on the RECIST version. The study also names secondary outcomes without reporting their estimates. Each omission affects what can be evaluated, even though the basic trial design and intended endpoint are clear.
Finally, the enrollment target and safety counts should not be mistaken for efficacy findings. Statistical interpretation depends on the analyzed population, endpoint data, and methods, not only on whether a study is marked completed. The record’s limitations make it especially important to distinguish what the trial planned to measure from what its posted results actually report.
14. 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 trial-specific limitation for interpretation is that the registry does not report the primary ORR result or the number of participants analyzed. The planned enrollment is listed as 60, but the actual enrollment, response-evaluable population, and missing-assessment counts are not stated. This prevents evaluation of the response estimate and its precision.
The single-group design has no concurrent control arm, so it cannot provide a randomized comparison. The study is open-label, and the record does not say whether imaging assessments were reviewed independently or with masking. The assessment schedule is described in the summary, but adherence to that schedule and the duration of observed follow-up are not reported.
The registry’s descriptions of response criteria are inconsistent: the summary names RECIST 1.1, while the primary endpoint description names RECIST v1.0. The record does not explain which version was applied. It also does not report analysis rules, missing-data handling, analysis populations, statistical methods, or whether subgroup analyses were performed.
Progression-free survival and overall survival are named as secondary outcomes but have no posted results. Safety information is limited to listed counts and a serious-term percentage, without detailed event definitions, timing, attribution, or overlap information. Together, these omissions restrict assessment of efficacy, precision, subgroup findings, and the interpretation of adverse events.
15. Related Tutorials
16. Related Statistical Calculators
17. Sources
- ClinicalTrials.gov: Pembrolizumab in Advanced Hepatocellular Carcinoma as Second-line Treatment After Failure of Sorafenib — NCT03163992.
- Registry record last updated 2022-08-08; status last verified 2022-08.
- Results first posted 2022-08-04.
- 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.
18. Record Summary
The registry identifies NCT03163992 as a completed phase 2, single-group, open-label trial sponsored by Samsung Medical Center. It studied pembrolizumab 200 mg every 3 weeks in advanced hepatocellular carcinoma after failure of sorafenib. The listed enrollment is 60, and the record gives 2017-12-26 as the start date and 2021-05-17 as the primary completion date.
ORR is the primary endpoint, with progression-free survival and overall survival listed as secondary outcomes. The registry record does not report efficacy estimates, analysis methods, or analysis populations. It also gives differing RECIST version descriptions. Safety fields list adverse-event counts and fatigue at 13.3% among the top serious terms, without enough detail to establish attribution or comparative risk.