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 Effect of Pembrolizumab (Keytruda®) on Biomarkers in Early Breast Cancer. (NCT03197389), a non-randomized early/phase 1 study. The primary outcome is PD-1 Expression.
| Feature | Registry value |
|---|---|
| ClinicalTrials.gov identifier | NCT03197389 |
| Phase | EARLY/Phase 1 |
| Condition | Breast Cancer, Triple Negative Breast Cancer, Hormone Receptor Negative Neoplasm |
| Allocation | Non-Randomized |
| Design model | Single Group |
| Masking | None (open label) |
| Enrollment | 54 |
| Arms in the primary outcome | n/a |
| Lead sponsor | Universitaire Ziekenhuizen KU Leuven |
| Results posted | Yes |
2. Clinical Question
Population
The prospective study concerned people with early breast cancer in several clinical and tumour subgroups. Cohorts A1, A2 and A3 covered triple negative, ER/PR negative and Her2 positive, and ER positive tumours, respectively. Cohorts B1, B2 and B3 covered corresponding tumour groups after neoadjuvant chemotherapy, where imaging showed residual tumour. The registry enrollment was 54, but the number whose biomarker results could be analyzed is not reported in the registry record.
Intervention
The planned intervention was pembrolizumab (Keytruda®), described as a single intravenous injection at 200 mg 10 +/- 4 days before surgery. The prospective study was open-label. The registry record also describes a separate retrospective analysis of PD-L1 expression in ER/PR negative breast tumours.
Comparator
There was no randomized control group. The registry classifies the prospective study as non-randomized and single-group. Its cohorts distinguish tumour characteristics and treatment history; they should not be read as randomized treatment-versus-control groups. A comparator for the primary biomarker endpoint is not reported in the registry record.
Primary question
The registered primary endpoint asks about PD-1 expression after a single dose of pembrolizumab. Expression was assessed semi-quantitatively on stromal tumour-infiltrating lymphocytes and reported as a percentage, with the median and full range specified. The numerical endpoint result is not reported in the registry record, so the direction and magnitude of any observed biomarker change cannot be described from that record.
3. Trial Design
The registry describes a study with two parts: a retrospective analysis and a prospective clinical study. The retrospective part, identified as S58910, was intended to examine PD-L1 expression in ER/PR negative breast tumours and its association with tumour-infiltrating lymphocytes, proliferation, apoptosis expression and clinical outcome, including distant metastases. Those aims are distinct from the prospective study’s registered primary endpoint of PD-1 expression after pembrolizumab.
The prospective component is described in the study summary as a Phase 0, single-centre, open-label, non-randomized study. The registry phase classification is EARLY PHASE1. These are the descriptions recorded for the study; the record does not explain the terminology difference. The allocation is listed as non-randomized and the model as single-group. There is no randomized control arm described.
The prospective cohorts distinguish participants having upfront surgery from those who had received neoadjuvant chemotherapy and had residual tumour visible on imaging. Within each pathway, cohort labels identify triple negative, ER/PR negative and Her2 positive, or ER positive tumour groups. Cohort B eligibility is described as requiring an imaging-estimated residual tumour size of at least 10 mm. These cohort distinctions do not create a randomized comparison, and the record does not report a planned statistical contrast between them.
The planned intervention was one intravenous pembrolizumab injection at 200 mg 10 +/- 4 days before surgery. The registry lists enrollment of 54, study start on 2018-01-18, primary completion on 2020-03-31 and status as COMPLETED. These record-level details do not establish how many participants contributed evaluable endpoint data. The analyzed population count is not reported in the registry record.
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 | Cohort A1 | Cohort A2 | Cohort A3 | Cohort B1 | Cohort B2 | Cohort B3 |
|---|---|---|---|---|---|---|
| STARTED | 16 | 5 | 18 | 9 | 1 | 5 |
| COMPLETED | 16 | 5 | 18 | 9 | 1 | 5 |
| NOT COMPLETED | 0 | 0 | 0 | 0 | 0 | 0 |
Analysis population for the primary outcome, as defined in the registry: The overall number of participants analyzed is not consistent with the numbers provided in any of the rows in the participant flow module because it was not always possible to interpret all the immunohistochemical stainings of all patients due to bad quality of the study samples.
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.
Only the first 6 of 7 groups are shown.
| Characteristic | Unit | Cohort A1 | Cohort A2 | Cohort A3 | Cohort B1 | Cohort B2 | Cohort B3 |
|---|---|---|---|---|---|---|---|
| Participants analysed | 16 | 5 | 18 | 9 | 1 | 5 | |
| Age, Categorical: <=18 years | Participants | 0 | 0 | 0 | 0 | 0 | 0 |
| Age, Categorical: Between 18 and 65 years | Participants | 10 | 4 | 15 | 7 | 1 | 3 |
| Age, Categorical: >=65 years | Participants | 6 | 1 | 3 | 2 | 0 | 2 |
| Sex: Female, Male: Female | Participants | 16 | 5 | 18 | 9 | 1 | 5 |
| Sex: Female, Male: Male | Participants | 0 | 0 | 0 | 0 | 0 | 0 |
| Race (NIH/OMB): American Indian or Alaska Native | Participants | 0 | 0 | 0 | 0 | 0 | 0 |
| Race (NIH/OMB): Asian | Participants | 0 | 0 | 0 | 0 | 0 | 0 |
| Race (NIH/OMB): Native Hawaiian or Other Pacific Islander | Participants | 0 | 0 | 0 | 0 | 0 | 0 |
| Race (NIH/OMB): Black or African American | Participants | 0 | 0 | 0 | 0 | 0 | 0 |
| Race (NIH/OMB): White | Participants | 16 | 5 | 18 | 9 | 1 | 5 |
| Race (NIH/OMB): More than one race | Participants | 0 | 0 | 0 | 0 | 0 | 0 |
| Race (NIH/OMB): Unknown or Not Reported | Participants | 0 | 0 | 0 | 0 | 0 | 0 |
| Region of Enrollment: Belgium | participants | 16 | 5 | 18 | 9 | 1 | 5 |
6. Primary Endpoint
The registry lists the primary outcome as PD-1 Expression, measured over 2 years.
| Element | Registry entry |
|---|---|
| Result type | Median |
| Measure of spread | Full Range |
| Unit | percentage |
| Endpoint classification | not reduced to a single comparison |
The primary endpoint is PD-1 expression, recorded as a percentage and summarized using the median and full range. The description states that expression was assessed semi-quantitatively on stromal tumour-infiltrating lymphocytes (sTIL) after a single pembrolizumab dose. Immunohistochemical staining used 5-μm thick sections and an automatic immunostainer; the endpoint description names a monoclonal antibody for PD1 and identifies the clone used.
As general statistical context, a median describes the central position of observed values without relying on an assumption that the measurements follow a symmetric distribution. A full range communicates the lowest and highest observed values, but it does not show how values are distributed between those extremes or how precisely the median is estimated. The registry does not give the endpoint’s numerical median, range, or a confidence interval.
Biomarker endpoints can be used to characterize a biological measurement in tissue. They do not, by themselves, establish a clinical benefit or predict an individual’s outcome. The registry lists a timeframe of 2 years, but does not clarify the timing of the tissue assessment relative to dosing beyond describing PD-1 expression after a single dose. The operational definition of the percentage, including its denominator, is not reported in the registry record.
7. Why No Re-analysis Is Shown
8. Statistical Interpretation
The registry record identifies PD-1 expression as the primary endpoint and specifies a median and full-range summary. It does not report the numerical result. Consequently, there is no recorded median or range in this information to interpret, and no estimate of the size or direction of a biomarker change can be stated. The absence of a numerical endpoint result also means that precision cannot be assessed from the registry record.
The study’s non-randomized, single-group design is important when interpreting any observed biomarker values. Without a randomized control group, a before-and-after difference or a post-dose measurement would not by itself isolate the effect of pembrolizumab from other explanations. The registry describes an endpoint after a dose, but does not report a comparator for that endpoint or a pre-dose PD-1 measurement. A causal treatment effect is therefore not established by the design description alone.
The cohort labels reflect differences in tumour subtype and treatment history. Cohort A refers to participants scheduled for upfront surgery, while cohort B refers to participants who had received neoadjuvant chemotherapy and had residual tumour visible on imaging. Results across these groups, if reported elsewhere, would need to be interpreted in light of those differences. They are not like-for-like randomized groups, and the registry does not describe an adjusted comparison across cohorts.
The registry reports enrollment of 54, but also notes that the number analyzed for the primary endpoint does not agree with participant-flow counts because some immunohistochemical stainings could not be interpreted due to poor sample quality. The number of evaluable samples is not reported in the registry record. Without that denominator, the extent of missing endpoint data and its possible influence on the observed summary cannot be assessed.
The record lists one outcome measure and one analysis. It does not report a numerical endpoint estimate, a confidence interval, a hypothesis test, or a p-value. Statistical significance and the precision of the result therefore cannot be evaluated from this record. The completed status indicates the study is recorded as completed; it does not fill in the missing endpoint result.
Overall, the registry documents a biomarker-focused, exploratory study design and an endpoint measurement approach. The posted information is not sufficient to characterize the actual PD-1 expression findings or to draw a treatment-effect conclusion. This is a statement about what the record reports, not a claim that no results exist elsewhere.
9. Statistical Methods Explained
The registry identifies PD-1 expression as the primary endpoint and specifies the median and full range as its summary statistics. It does not report the endpoint’s numerical result or describe an inferential test. The methods below explain the registered design and summaries in general terms; they should not be mistaken for a complete statistical analysis plan.
What does the median convey?
General: the median is the central observation after measurements are ordered from lowest to highest. It can be a useful summary when values are skewed or when a small number of extreme observations might pull the mean away from the typical value. It does not show the shape of the entire distribution, and it does not describe how uncertain the estimated median is. The registry does not report a confidence interval for the median.
The full range gives the minimum and maximum observed measurements. General: ranges are easy to interpret, but they depend strongly on the observations at the two extremes and do not convey how many measurements lie near either endpoint. With no reported median or range values, the primary endpoint’s observed distribution cannot be assessed from the registry record.
How was the biomarker measured?
The endpoint description says immunohistochemical stains were performed on 5-μm thick sections using an automatic immunostainer and that PD1 expression on sTIL was assessed semi-quantitatively. This describes a tissue-based biomarker assessment rather than a patient-reported or time-to-event endpoint. The record names the antibody clone and instrument, but does not provide scoring categories, reader procedures, reproducibility information, or the calculation used to express the result as a percentage. Those details are not reported in the registry record.
General: semi-quantitative assessment applies an ordered or graded interpretation to staining rather than treating the staining signal as a fully continuous direct measurement. Such measures depend on consistent specimen handling and scoring. The registry specifically notes that some stainings could not be interpreted because of poor sample quality, which means the endpoint analysis did not include every planned or available sample.
What can a single-group design establish?
The prospective study is listed as non-randomized and single-group, with no masking. General: an uncontrolled study can describe observations among participants who receive an intervention, but it cannot use random assignment to balance known and unknown prognostic factors between treatment groups. A post-dose biomarker measurement may describe the biological state at assessment. Without a reported comparator or a clearly specified pre-dose measurement, it cannot independently separate an intervention-associated change from other causes.
The cohort structure also matters. Cohorts differ by tumour characteristics and by whether participants had received neoadjuvant chemotherapy. Those factors may be associated with biomarker measurements. General: combining unlike clinical subgroups can make an overall summary difficult to interpret, while subgroup summaries may be based on limited data. The registry does not report subgroup endpoint estimates or a method for accounting for those differences.
What does the record say about analysis?
The registry lists one outcome measure and one analysis, but does not state a statistical test, model, adjustment strategy, or confidence-interval method for the primary endpoint. It also does not report a prespecified threshold for interpreting a change in PD-1 expression. These omissions prevent a detailed reconstruction of the analysis. The stated median and full range describe the intended summary, not a complete analysis plan.
General: confidence intervals express the uncertainty around an estimate under a statistical model and its assumptions. A confidence interval would add information about precision that a point estimate alone does not provide. No such interval is reported for this endpoint. P-values, if available, would address compatibility with a specified null model, not the size or clinical importance of a biomarker finding; no p-value is reported in the registry record.
General: over-dispersion is relevant when variability in data exceeds what a chosen statistical model expects, as can occur with some count outcomes. The registered primary endpoint is a percentage summarized by a median and range, and the registry does not describe a count model or an over-dispersion adjustment. There is no basis in the record to claim that this issue was assessed or that a particular model was used.
The record also contains assumptions for planning illustrations based on a stated fraction of an observed effect, with two-sided significance level 0.05, equal allocation and a normal approximation. These are planning assumptions, not evidence of the trial’s original design or observed power. No numerical planning estimates are reported in the registry record, and the actual study is not described as an equal-allocation randomized comparison. The sample-size justification and the trial’s own power calculation are not reported in the registry record.
10. Multiplicity, Missing Data and Analysis Populations
The record lists one primary outcome measure and one analysis. It does not describe a multiplicity strategy, such as adjustment for testing several hypotheses or a hierarchy for prioritizing endpoints. The retrospective component has several stated aims, but the registry does not report how analyses of those aims were handled in relation to the prospective primary endpoint. A multiplicity adjustment or a decision not to adjust is not reported in the registry record.
Missing endpoint data are directly relevant. The registry says the number analyzed for the primary endpoint is inconsistent with participant-flow counts because some immunohistochemical stainings could not be interpreted due to poor sample quality. It does not report how many samples were evaluable, how many were uninterpretable, or whether the missing measurements differed across cohorts. The reasons for poor sample quality are not further described in the registry record.
The analysis population is not defined in detail. The record does not say whether the primary analysis included all participants who received the intervention, only those with an evaluable stain, or another specified group. The method for handling missing values, and any sensitivity analysis assessing the effect of missing data, are not reported in the registry record.
General: missing measurements can affect interpretation when the probability of a measurement being unavailable is related to the underlying value or to participant characteristics. Excluding uninterpretable samples may be reasonable for a tissue assay, but the impact depends on who is excluded and why. Without counts and an analysis-population definition, the direction or size of any resulting bias cannot be assessed.
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 |
|---|---|---|---|
| Cohort A1 | 0 | 16 | 0.0% |
| Cohort A2 | 0 | 5 | 0.0% |
| Cohort A3 | 0 | 18 | 0.0% |
| Cohort B1 | 0 | 9 | 0.0% |
| Cohort B2 | 0 | 1 | 0.0% |
| Cohort B3 | 0 | 5 | 0.0% |
Deaths
| Arm | Deaths (participants affected of at risk) |
|---|---|
| Cohort A1 | 0 of 16 |
| Cohort A2 | 0 of 5 |
| Cohort A3 | 0 of 18 |
| Cohort B1 | 0 of 9 |
| Cohort B2 | 0 of 1 |
| Cohort B3 | 0 of 5 |
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 |
|---|---|
| Cohort A1 | 0 of 16 |
| Cohort A2 | 0 of 5 |
| Cohort A3 | 0 of 18 |
| Cohort B1 | 0 of 9 |
| Cohort B2 | 0 of 1 |
| Cohort B3 | 0 of 5 |
The registry’s safety figures are organized by cohort. In each listed cohort, the number at risk for serious adverse events and for other adverse events is recorded alongside the number affected. Cohort A1 lists 16 at risk in each category and 0 affected; A2 lists 5 and 0; A3 lists 18 and 0; B1 lists 9 and 0; B2 lists 1 and 0; and B3 lists 5 and 0. The death count is recorded as 0 for each cohort. No serious adverse-event terms are listed.
These are the figures in the registry record, not evidence that adverse experiences are impossible or that every possible event was captured. The record’s study summary says participants would be monitored for adverse experiences, but the safety details do not specify the observation period, event-collection procedures, severity grading, or completeness of reporting. The categories and denominators should therefore be read as registry-reported safety counts rather than as a comprehensive safety profile.
The cohorts differ in size and clinical characteristics, and the registry does not provide a randomized comparison. The counts cannot establish whether the intervention caused or prevented events. The figures also do not supply a precision estimate for event risk. No conclusion beyond the recorded counts is supported by this safety information.
12. What the Result Does, and Does Not, Mean
The study is relevant as a prospective investigation of a tissue biomarker after pembrolizumab in early breast cancer, alongside a separate retrospective analysis of PD-L1 and related tumour features. Its primary endpoint focuses on PD-1 expression on stromal tumour-infiltrating lymphocytes. The registry records the measurement approach and the intended summary statistics, making the biological question identifiable.
What the registry record does not establish is the observed result. It contains no numerical median or full range for PD-1 expression and no reported confidence interval or hypothesis-test result. Thus, the magnitude, direction and precision of the primary biomarker finding cannot be described. The missing result should not be interpreted as a finding of no change.
The design also limits the conclusions that could be drawn from an observed value. The prospective component is open-label, non-randomized and single-group. The cohorts are defined by tumour subtype and treatment history rather than random allocation. As a general principle, these features do not provide the balance or direct counterfactual comparison that randomization can support. The record does not report a control group for the primary endpoint.
PD-1 expression is a biomarker measurement, not itself a clinical outcome. Even a clearly measured biomarker difference would not automatically establish an effect on tumour control, survival, distant metastases or another patient-relevant outcome. The registry record does not report a relationship between the primary PD-1 endpoint and those outcomes.
The record reports enrollment of 54, but notes that some stains could not be interpreted because of poor sample quality and that the analyzed number does not match participant-flow counts. Since the evaluable endpoint count and missing-data handling are not reported, the completeness and representativeness of the biomarker analysis remain uncertain.
The appropriate reading is therefore limited: this is a completed biomarker study with a registered endpoint and a described assay approach, but the registry information does not present the primary numerical result. It cannot support a conclusion about whether pembrolizumab did or did not alter PD-1 expression, or about clinical benefit or harm.
13. Why This Trial Matters Statistically
Statistically, this study illustrates the distinction between a biological question and the information needed to estimate an answer. The registry defines PD-1 expression as a percentage measured on sTIL and specifies a median and full range. Those details clarify the intended endpoint, but without the numerical summary, the result itself cannot be evaluated.
The record also highlights why analysis populations matter in biomarker studies. Tissue assays can fail to yield interpretable measurements, and the registry explicitly notes poor sample quality as a reason that analyzed counts differ from participant-flow counts. Reporting the number of evaluable samples, the number excluded and the reasons for exclusion would help readers assess how much information contributed to the endpoint and whether missingness could affect interpretation.
The non-randomized, single-group structure reinforces the need to distinguish description from causal comparison. The cohort categories capture clinical heterogeneity, but they do not serve as randomized controls. General: design features determine which statistical conclusions are credible; sophisticated summaries cannot compensate for the absence of a suitable comparison when the goal is causal attribution.
Finally, the study has a biomarker endpoint rather than a direct clinical outcome as its primary measure. This makes careful definition, assay reliability and transparent reporting especially important. The registry identifies some laboratory procedures but leaves the scoring details, analysis population and numerical endpoint estimate unreported. These gaps show how a registry can document the intended scientific question while still leaving readers unable to assess the precision or substantive meaning of the observed result.
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 most immediate limitation of the registry information is that the primary endpoint result is not reported. Although PD-1 expression is specified as a percentage with median and full-range summaries, neither the numerical summary nor a precision estimate is available in the registry record. This prevents evaluation of the biomarker distribution or the size of any observed difference.
The analyzed population is also unclear. The record acknowledges that not all immunohistochemical stains could be interpreted because of poor sample quality and that the analyzed count does not match participant-flow counts. It does not report the evaluable count, the pattern of unavailable samples, or the treatment of missing measurements in analysis.
The prospective study is non-randomized, open-label and single-group. No control group for the primary endpoint is described. The cohorts also differ by tumour characteristics and treatment history, and the registry does not report an adjusted comparison across them. These design features restrict causal interpretation of any biomarker observations.
The endpoint description gives some assay details but does not report the scoring rule, denominator for the percentage, reader blinding or reproducibility. The timeframe is listed as 2 years, while the tissue-assessment timing relative to dosing is not fully specified. These details are not reported in the registry record.
Further limitations include the absence of a reported statistical test, confidence interval, missing-data sensitivity analysis, multiplicity approach and sample-size justification. The registry’s planning assumptions do not substitute for a documented trial analysis plan. These omissions limit independent assessment of both precision and analytic choices.
15. Related Tutorials
16. Related Statistical Calculators
17. Sources
- ClinicalTrials.gov: Effect of Pembrolizumab (Keytruda®) on Biomarkers in Early Breast Cancer. — NCT03197389.
- Registry record last updated 2021-05-19; status last verified 2021-05.
- Results first posted 2021-05-19.
- 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 record describes NCT03197389, titled “Effect of Pembrolizumab (Keytruda®) on Biomarkers in Early Breast Cancer.” The sponsor is Universitaire Ziekenhuizen KU Leuven. The study is listed as EARLY PHASE1, non-randomized, open-label and single-group, with enrollment of 54 and status COMPLETED.
The study summary describes a retrospective analysis of PD-L1 in ER/PR negative breast tumours and a prospective Phase 0 study of pembrolizumab in early breast cancer. The registered primary endpoint is PD-1 expression, measured semi-quantitatively on sTIL and summarized as a percentage, median and full range. The numerical endpoint result is not reported in the registry record.
The record notes that some immunohistochemical stains were uninterpretable because of poor sample quality, leaving the analyzed count inconsistent with participant-flow counts. It reports no affected participants or deaths in the listed cohort safety figures. Those data do not constitute a randomized comparison or establish a treatment effect.