A Study of Nivolumab in Relapsed/Refractory Primary Central Nervous System Lymphoma (PCNSL) and Relapsed/Refractory Primary Testicular Lymphoma (PTL): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study of Nivolumab in Relapsed/Refractory Primary Central Nervous System Lymphoma (PCNSL) and Relapsed/Refractory Primary Testicular Lymphoma (PTL) (NCT02857426), a non-randomized phase 2 study. The primary outcome is BICR-Assessed Objective Response Rate (ORR).

2. Clinical Question

The study’s stated purpose was to determine whether nivolumab was effective in relapsed or refractory PCNSL and PTL. Its primary outcome was the blinded independent central review (BICR)-assessed objective response rate (ORR), reported separately for the two populations. The response criteria differed by disease: the endpoint description names the IPCG Criteria for PCNSL and Lugano 2014 response evaluation for PTL.

This design addresses whether objective responses were observed among treated participants in each cohort. It does not make a randomized comparison between nivolumab and another treatment or a control condition. As a result, the clinical question is descriptive within each cohort rather than a question about the relative effect of nivolumab against a concurrent comparator. The registry record does not report the cohort-specific response estimates, so the size and precision of any observed response cannot be assessed from the efficacy information recorded there.

3. Trial Design

This was a phase 2, open-label, non-randomized, single-group study with two disease cohorts: relapsed or refractory PCNSL and relapsed or refractory PTL. Nivolumab was the intervention. The registry records enrollment of 66 and lists the study status as completed. The recorded study start was 2016-10-21, and primary completion was 2019-06-11.

Because the study was open-label and had no masking, treatment assignment was not concealed from participants or investigators in the way it may be in a masked trial. The primary outcome was nevertheless assessed by blinded independent central review. That distinction matters: independent blinded assessment can reduce some opportunities for assessment bias, but it does not create a randomized control group or mask treatment during care.

The two cohorts represent different lymphoma populations, and the endpoint definition specifies a separate response evaluation approach for each. PCNSL responses were based on the IPCG Criteria, while PTL responses were based on Lugano 2014 response evaluation. The primary endpoint description says the percentage was calculated within each cohort among treated participants. The registry record does not report the treated population count for either cohort alongside the response results.

Several secondary outcomes are listed: BICR-assessed progression-free survival, investigator-assessed ORR, investigator-assessed duration of response, and overall survival. The record lists five outcome measures overall, but no analysis records. No randomized treatment contrast is available from this design. Any comparison between the two disease cohorts would also require care, because they are different populations and the study does not identify one as a control for the other. The registry record does not report further details about the statistical plan, sample-size justification, interim analyses, or handling of protocol deviations.

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: 47 PCNSL participants treated. 19 PTL participants treated.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: All treated PCNSL and PTL participants

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 BICR-Assessed Objective Response Rate (ORR), measured over Up to approximately 51 months.

The primary endpoint is BICR-assessed objective response rate (ORR), expressed as the percentage of treated participants in each disease cohort whose best overall response was a confirmed complete response or partial response. The endpoint description specifies that PCNSL responses use the IPCG Criteria and PTL responses use Lugano 2014 response evaluation. The stated timeframe is up to approximately 51 months.

ORR is a binary-outcome summary: each participant is classified according to whether the endpoint response definition was met, and the cohort result expresses the share meeting that definition. In general, an objective response endpoint can describe whether measurable tumor response was recorded under prespecified criteria. BICR means that review is conducted by a blinded independent central review process, rather than relying only on the treating investigator’s assessment. The registry record does not provide detail on how disagreements or unevaluable assessments were handled.

For interpretation, the response percentage needs to be considered alongside its confidence interval and the number of treated participants assessed. A confidence interval describes statistical uncertainty around an estimate; it does not account for every source of bias or make a single-arm result equivalent to a controlled comparison. Although the endpoint definition names a 95% confidence interval as the dispersion measure, the registry record does not report the cohort-specific response estimates or their intervals. Therefore, neither the observed response frequency nor its precision can be described from the recorded primary outcome information.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary measure is a percentage: the number of treated participants who met the confirmed-response definition relative to the treated participants in the relevant cohort. In general, a percentage for a binary outcome summarizes how often a defined event was recorded in a group. It is interpretable only alongside a clear event definition and a clear denominator. Here, the registry defines the response event and says the calculation is within each treated cohort, but it does not report the actual cohort-specific response percentages or denominators.

What does a confidence interval convey?

The endpoint lists a 95% confidence interval. In general, a confidence interval represents uncertainty in an estimate under the assumptions of the statistical method used. A narrower interval indicates greater statistical precision than a wider interval, but neither width nor location can be assessed without the actual estimate and interval. An interval also does not resolve systematic problems such as selection bias, outcome misclassification, or lack of a control group. The registry record does not identify the interval-construction method or report the intervals for the primary outcome.

What does BICR contribute?

BICR refers to blinded independent central review. In general, central review can apply response criteria independently of the treating team, and blinding can reduce the influence of knowledge about treatment on assessment. The primary endpoint specifies BICR assessment for ORR in both cohorts. The record does not describe the review procedures in further detail, so it is not possible to assess matters such as the handling of discordant assessments, missing scans, or confirmation rules beyond the endpoint’s stated confirmed-response definition.

Why do the response criteria matter?

The response event is not simply an unstructured clinical judgment: the registry names specific criteria. The PCNSL cohort uses the IPCG Criteria, and the PTL cohort uses Lugano 2014 response evaluation. In general, consistent criteria make response classifications more reproducible, while disease-specific criteria recognize that assessment may differ across conditions. Because the criteria differ, each cohort’s response estimate should be interpreted according to its own definition. A direct numerical comparison between cohorts would not automatically be a like-for-like comparison.

What can a single-group design estimate?

A single-group study can estimate the observed outcome frequency among treated participants, subject to the study’s enrollment, follow-up, assessment, and analysis procedures. In general, a randomized concurrent control helps distinguish treatment-associated outcomes from outcomes that might occur under another condition. This study is recorded as non-randomized, with a single-group model and no masking. It therefore does not provide a randomized estimate of a treatment contrast. Differences between outcomes in this study and outcomes from another study would not be randomized comparisons and could reflect differences in populations, criteria, follow-up, or other factors.

How should cohort sizes and enrollment be read?

The registry records enrollment of 66, while the primary endpoint is defined among treated participants within each cohort. Enrollment and the analysis denominator are not necessarily interchangeable. The record does not provide the treated denominator for either primary endpoint estimate, nor does it report how many participants had an evaluable response assessment. No group totals or endpoint denominators should be inferred from the overall enrollment figure or from safety counts.

What is known about the statistical plan?

The registry record lists the endpoint and its confidence-interval type, but does not report a detailed statistical analysis plan. It does not report the specific interval method, a sample-size justification, planned interim analyses, or methods for missing primary outcome data. It also lists no analysis records. These omissions limit assessment of whether the recorded endpoint definition and the analysis performed fully match a prespecified plan. They do not establish that a particular method was or was not used; the method details are not reported in the registry record.

How should sample-size illustrations be understood?

Any sample-size illustrations associated with the posted results are planning calculations based on the observed effect, not evidence of the trial’s original design or observed power. In general, such calculations describe what sample size might be needed under stated assumptions, including the target effect and allocation. They do not show that a study had a given probability of detecting its result after the fact. The listed assumptions describe a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation, with the target effect expressed as a fraction of the observed effect. Those assumptions do not describe this trial’s actual single-group, two-cohort design, and the registry record does not report the trial’s own sample-size justification.

For binary endpoints, the event frequency and sample size affect the precision of an estimate. In general, sparse events or a small denominator can yield substantial uncertainty. This principle explains why endpoint counts and intervals are important, but the absent primary response values mean that precision cannot be assessed for this study from the registry record.

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("PCNSL Cohort", "PTL Cohort")
n      <- c(47, 19)
events <- c(3, 5)
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  PCNSL Cohort  47  3  6.4  1.3  17.5  2.2  17.2  2.3  17.2
ARM  PTL Cohort  19  5  26.3  9.1  51.2  11.8  48.8  11.9  49.1

11. Multiplicity, Missing Data and Analysis Populations

The record lists a primary endpoint and several secondary outcomes, including progression-free survival, investigator-assessed ORR, duration of response, and overall survival. Evaluating multiple outcomes can raise multiplicity concerns: in general, when many hypotheses are tested, the chance of at least one apparently notable result can increase unless the analysis plan addresses that issue. The registry record does not report a multiplicity adjustment, a hierarchy among secondary outcomes, or other procedures for controlling repeated testing. It is therefore not possible to determine how multiplicity was handled.

The primary endpoint is defined for all treated PCNSL and PTL participants, with results reported separately by cohort. However, the registry record does not provide the cohort-specific treated denominators, response counts, estimates, or confidence intervals. It does not describe the analysis population in enough detail to establish how participants with missing or unevaluable response assessments were treated.

Missing data can affect an endpoint estimate if participants without an assessment differ from those with an assessment. In general, an analysis needs to state whether such participants are excluded, classified as nonresponders, or handled by another method. The record does not report the primary endpoint’s missing-data rules, the amount or pattern of missingness, or sensitivity analyses. It also does not provide details of any interim analyses or stopping criteria. These methods are not reported in the registry record, so no assumptions should be made about them.

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: malignant neoplasm progression (42.1%), renal failure (10.5%), seizure (8.5%), sepsis (6.4%), haemorrhage intracranial (6.4%).

The registry lists serious adverse-event counts for the PCNSL and PTL nivolumab cohorts. For PCNSL, the figures are 33 affected among 47 at risk; for PTL, 15 affected among 19 at risk. The adverse-event information also includes other-event counts and deaths, but those measures should not be treated as interchangeable with serious adverse events. The event categories and denominators need to be kept distinct when interpreting each figure.

The record lists serious adverse-event terms with percentages, including malignant neoplasm progression at 42.1%, renal failure at 10.5%, seizure at 8.5%, sepsis at 6.4%, haemorrhage intracranial at 6.4%, and anaemia at 5.3%. The record does not attach these listed percentages to a specific cohort in the information available, so they should not be assigned to one. The listed events are descriptive reports, not evidence by themselves that nivolumab caused them. The registry record does not provide the detailed definitions or attribution rules needed to interpret them as treatment-related risks. These figures also do not support a comparison against another treatment, because the trial has no concurrent control group.

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

The study was designed to estimate BICR-assessed ORR among treated participants in separate PCNSL and PTL cohorts. Its primary endpoint definition is explicit about what counts as a response and which criteria apply to each population. The registry record also identifies the study as phase 2, open-label, non-randomized, and single-group. Those details establish the design context, but they do not reveal the observed response results.

The cohort-specific primary response percentages and confidence intervals are not reported in the registry record. Consequently, the magnitude of the response estimate and its statistical precision cannot be stated from this record. Enrollment of 66 is not a substitute for the treated denominators used in the endpoint, and it does not reveal how many participants were evaluable for response. No conclusion about the frequency of response in either cohort should be inferred without those outcome values.

Even if response values were available, a single-group study would describe outcomes among participants receiving nivolumab rather than estimate a randomized difference against another treatment. General considerations such as patient selection, disease course, and assessment practices can influence observed outcomes in an uncontrolled study. BICR offers an independent assessment process for the primary endpoint, but it does not address the absence of a control group or every possible source of bias.

The safety figures are likewise descriptive. Counts and listed percentages can indicate which adverse events were recorded, but they do not alone establish causality or comparative safety. No control group is present to provide a like-for-like comparison of event frequencies.

Accordingly, the record supports a description of the trial’s question, design, endpoint definitions, and posted safety information. It does not support a quantified statement about primary efficacy, a treatment-effect comparison, or a causal safety conclusion. Those boundaries reflect what is and is not reported, rather than a judgment about the unreported study findings.

14. Why This Trial Matters Statistically

This trial is statistically notable for the contrast between a clearly specified response endpoint and the limited efficacy results recorded in the registry. The primary measure is a cohort-specific percentage of treated participants with a confirmed response, assessed by BICR. The record defines the response categories and identifies the criteria used for each lymphoma population. These details clarify what the intended outcome represents.

At the same time, the design determines the scope of inference. A non-randomized, single-group study can describe outcomes in treated participants, but it cannot directly estimate a randomized treatment contrast. The two disease cohorts are not interchangeable control groups: they represent different populations and use different response criteria. Keeping results separated by cohort is therefore important for interpretation.

The absence of cohort-specific response estimates and confidence intervals in the registry record also illustrates why endpoint definitions alone are insufficient for statistical interpretation. A reader needs the observed estimate, its denominator, and a measure of uncertainty to judge the frequency and precision of an outcome. The overall enrollment figure does not supply those pieces of information for the primary endpoint.

The record’s safety information presents a related lesson. Counts and event percentages are useful descriptions, but their meaning depends on the event definition, denominator, cohort attribution, and whether events are considered treatment-related. Without a concurrent control, observed safety counts do not provide a comparative estimate of risk. Taken together, the record emphasizes the distinction between what a trial is designed to measure and what can be concluded from the results actually reported in the registry.

15. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The main trial-specific limitation for interpreting efficacy is that the registry record does not report the primary ORR values or confidence intervals for either cohort. It also does not provide the cohort-specific treated denominators or response-assessment counts. As a result, the observed response frequency and precision cannot be evaluated from the recorded primary outcome information.

The study is non-randomized and single-group, with no concurrent control. Thus, even a fully reported response estimate would be descriptive of treated participants and would not, on its own, establish a causal treatment effect relative to another condition. The two cohorts also concern different diseases and use different response criteria, which makes a simple cross-cohort comparison difficult to interpret.

The record does not report the detailed statistical analysis plan, the confidence-interval construction method, missing-data rules, multiplicity procedures, interim analyses, or sample-size justification. It also does not provide enough detail to assess the treatment of unevaluable responses, protocol deviations, or the completeness of follow-up for the primary endpoint. These are reporting limitations: their absence from the registry record does not establish what was or was not done in the study.

Safety interpretation has further constraints. The listed serious adverse-event terms and percentages are not tied to a specific cohort in the available information, and the record does not supply detailed attribution or event-definition procedures. The uncontrolled design also precludes a randomized comparison of safety outcomes. These limitations apply to interpretation of the registry record and should not be mistaken for evidence that an unreported analysis or procedure was absent from the full study documentation.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: A Study of Nivolumab in Relapsed/Refractory Primary Central Nervous System Lymphoma (PCNSL) and Relapsed/Refractory Primary Testicular Lymphoma (PTL) — NCT02857426.
• Registry record last updated 2021-12-23; status last verified 2021-11.
• Results first posted 2020-10-28.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

19. Record Summary

The registry record describes a completed phase 2, open-label, non-randomized, single-group study of nivolumab in two cohorts: relapsed or refractory PCNSL and relapsed or refractory PTL. Enrollment was 66. The primary endpoint was BICR-assessed ORR, defined as the percentage of treated participants with a confirmed complete or partial response, using disease-specific response criteria.

The record lists secondary outcomes and descriptive safety figures. It does not report the cohort-specific primary response estimates, their confidence intervals, or the treated denominators used for those estimates. Therefore, the efficacy frequency and precision cannot be evaluated from the registry record. The study has no concurrent control group, so its design does not provide a randomized treatment comparison.