A Pilot Study of the Thrombopoietin-Receptor Agonist Eltrombopag in Refractory Aplastic Anemia Patients: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Pilot Study of the Thrombopoietin-Receptor Agonist Eltrombopag in Refractory Aplastic Anemia Patients (NCT00922883), a na phase 2 study. The primary outcome is The Portion of Drug Responders as Defined by Hematologic Improvements.

2. Clinical Question

3. Trial Design

This was a Phase 2, non-randomized, single-group pilot study of eltrombopag in aplastic anemia patients with immunosuppressive-therapy-refractory thrombocytopenia. The registry lists enrollment of 43 and identifies the National Heart, Lung, and Blood Institute as sponsor. The study started on 2009-05-29, reached primary completion on 2013-05-07, and is listed as completed.

Participants received oral eltrombopag. The registry describes a starting dose of 50 mg/day and increases up to 150 mg/day, with the arm description specifying dose increases of 25 mg every 2 weeks if the platelet count had not increased by 20 × 103/µL. The brief summary describes dose adjustment toward maintaining a stable platelet count while maximizing tolerability. The response assessment timeframe was 12-16 weeks.

The study had no randomized allocation, masking, or concurrent control group. This matters for interpretation: the design can describe outcomes among treated participants, but it cannot isolate treatment effects from changes that might have occurred over time or from other influences. The registry record does not report a statistical analysis plan, sample-size justification, interim analysis approach, or further design details. Those features should not be inferred from the Phase 2 label or the pilot designation.

The record states that all subjects who received eltrombopag were analyzed for the primary outcome. It does not give the number who received treatment or a primary response count in the posted outcome information represented by these registry entries. The listed enrollment and the stated analysis population are distinct record details and should not be treated as evidence of a particular response rate.

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: 44 patients were consented, but 1 patient was removed before treatment with the investigational agent was initiated because an additional review of baseline bone marrow revealed that the patient was not eligible, because the patient had changes in the bone marrow inconsistent with a diagnosis of aplastic anemia.

Overall Study

Analysis population for the primary outcome, as defined in the registry: All subjects who received Eltrombopag were analyzed.

5. Baseline Characteristics

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

Population: Note that baseline information was available on the 44 patients consented and enrolled, but one patient was removed from protocol before initiation of the investigative agent, and thus adverse events and results information refers only to 43 patients

6. Primary Endpoint

The registry lists the primary outcome as The Portion of Drug Responders as Defined by Hematologic Improvements, measured over 12-16 weeks.

The primary endpoint is the portion of drug responders defined by hematologic improvements, assessed at 12-16 weeks. The record defines response as unilineage or multilineage recovery meeting at least one of the platelet, erythroid, or neutrophil criteria. In practical terms, the endpoint is intended to capture improvement in one or more blood-cell lineages rather than requiring improvement in all of them.

The platelet criterion includes an increase to 20 × 103/μL above baseline, or stable platelet counts with transfusion independence for a minimum of 8 weeks among participants who were transfusion dependent when entering the protocol. The erythroid criterion applies when pretreatment hemoglobin was <9 g/dL and allows a hemoglobin increase of 1.5 g/dL or, for transfused participants, a reduction of at least 4 packed red blood cell transfusions over 8 consecutive weeks compared with the pretreatment transfusion number in the previous 8 weeks. The neutrophil criterion applies when pretreatment ANC was <0.5 × 103/μL and is defined by at least a 100% ANC increase or an ANC increase >0.5 × 103/μL.

A binary responder endpoint can make a clinically interpretable summary of whether participants cross prespecified thresholds. However, the registry record does not report how many participants met the endpoint, so the observed response portion is not reported in the registry record. The endpoint description also mentions toxicity measured using Common Terminology Criteria for Adverse Events, but its role in determining responder status is not further specified.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

Statistical interpretation begins by identifying the design and the outcome being estimated. This study is listed as single-group, non-randomized, and unmasked. Its primary question concerns the portion of treated participants classified as responders at 12-16 weeks. The outcome is binary at the participant level: a participant either meets the response definition or does not. The registry record gives the criteria but does not give the count meeting them.

What does a single-group result describe?

In general, a single-group study can describe outcomes observed in participants receiving an intervention. It lacks a concurrent randomized group that could show what happened over the same period without the intervention or under a different treatment. As a result, an observed change cannot automatically be attributed to treatment. The distinction is between describing what occurred in the study group and estimating a comparative causal effect. This record does not include a control group, so it cannot supply a randomized treatment-versus-control contrast.

Randomization, when used, helps balance known and unknown participant characteristics across groups on average. Masking can reduce certain biases in treatment delivery or outcome assessment. Here, allocation is recorded as not applicable and masking as none. These design features do not make the observations unusable, but they define the limits of what the trial can establish. The registry record does not report additional procedures that might address these limitations.

How should the primary endpoint be read?

The responder definition is composite: improvement in any one or more of the specified lineages may qualify, subject to the applicable criteria. This allows an outcome to reflect different hematologic responses. It also means that the overall responder category can combine participants whose improvements differ in type. In general, composite outcomes are most interpretable when their components and decision rules are clear. The registry describes several criteria, including baseline-dependent thresholds and transfusion-related conditions, but does not report component-specific results. It also mentions toxicity measured using Common Terminology Criteria for Adverse Events without explaining its precise role in the responder classification.

The endpoint is a count of participants meeting criteria, rather than a continuous measure of change. A count can be converted into a proportion when both the count and relevant analysis denominator are available. Neither a response count nor a response proportion is reported in the registry entries for this analysis, so no such estimate should be inferred. The stated primary population is all subjects who received eltrombopag; the record does not supply the responder count needed to summarize that population’s outcome.

What do confidence intervals and tests add?

In general, a confidence interval expresses uncertainty around an estimate, accounting for the information in the observed data and the chosen statistical method. A narrower interval usually indicates greater precision, while a wider interval indicates less precision. An interval does not remove design limitations, and it cannot resolve the absence of a control group. The registry record does not report a confidence interval or a statistical test for the primary response outcome. A p-value, if one were available, would address compatibility with a specified statistical model and null hypothesis; it would not alone convey the size or clinical meaning of an outcome.

For a binary endpoint, precision depends on the observed response count and the analysis population. Since the response count is not reported, neither an uncertainty interval nor a recalculation from the reported summary statistics is possible. The registry entries also do not include analyses comparing outcomes across subgroups or adjustment for other variables.

Why does over-dispersion matter in general?

Over-dispersion is a general concern for count outcomes when the variability in the data exceeds what a chosen model assumes. If a model underestimates variability, its uncertainty estimates may be too small and its inferential results too confident. This trial’s primary endpoint is described as a participant count, but the registry entries do not report a fitted count model, dispersion estimate, or adjusted analysis. It would not be appropriate to claim that over-dispersion was assessed or that a particular model was used.

What does “recalculated” mean?

When a result is labelled recalculated, it generally refers to an unadjusted calculation from posted summary statistics rather than the sponsor’s own analysis. Such a calculation can be useful for checking or illustrating a result, but it may not reproduce an adjusted or otherwise prespecified analysis. In this record, there is no primary response result or analysis from which to describe a recalculation. The registry entries do not identify a discrepancy for the primary outcome, and no recalculated response estimate is available.

Any sample-size illustration based on an observed effect would be a planning illustration, not observed power and not part of the trial’s reported design rationale. General sample-size calculations describe assumptions about a future study, such as the target effect and error rates. They do not establish the probability that this completed study detected an effect. The registry record does not report its own sample-size justification or statistical design assumptions, so those should not be imputed.

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("Eltrombopag")
n      <- c(43)
events <- c(17)
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  Eltrombopag  43  17  39.5  25.0  55.6  26.4  54.4  26.3  54.5

11. Multiplicity, Missing Data and Analysis Populations

The primary endpoint permits response through platelet, erythroid, or neutrophil improvement. These are multiple components within a composite endpoint, but the registry entries do not report separate component analyses or explain whether any additional outcomes were tested as formal hypotheses. The outcome-measure count alone does not establish the number of statistical tests or whether a multiplicity adjustment was planned. The statistical plan and any multiplicity strategy are not reported in the registry record.

Multiplicity is a general issue when several outcomes or subgroup comparisons are tested: the chance of obtaining at least one apparently notable result can increase as testing expands. Methods such as adjusting significance thresholds or controlling a family-wise error rate can address that concern, but the registry entries do not say whether such methods were used. No claims about adjustment or unadjusted testing can be made for this study.

The primary analysis population is described as all subjects who received eltrombopag. The registry entries do not report participant-level missingness, withdrawals, exclusions after treatment, or how missing endpoint assessments were handled. They also do not state whether any sensitivity analyses were performed. Missing-data methods can affect a binary responder estimate, particularly when the outcome status is not observed for every participant; however, the extent and handling of missing data are not reported in the registry record.

The listed enrollment is not a substitute for an endpoint denominator. It should not be assumed that every enrolled participant contributed an observed response assessment, nor should an unreported response count or denominator be reconstructed. Without the analysis details and outcome count, the effect of missing observations and the robustness of the primary result cannot be evaluated.

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 0% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: febrile neutropenia (16.3%), sepsis (7.0%), conjunctival haemorrhage (4.7%), anaemia (2.3%), neutropenia (2.3%).

The registry lists serious adverse events affecting 16 participants in the eltrombopag group, with 43 at risk. It also lists other adverse events affecting 38 participants, with 43 at risk, and reports 4 deaths. These are counts recorded for the study group; they are not comparative estimates against an untreated or alternative-treatment group.

The most frequently listed serious adverse-event terms are febrile neutropenia at 16.3%, sepsis at 7.0%, and conjunctival haemorrhage at 4.7%. Anaemia, neutropenia, and thrombocytopenia are each listed at 2.3%. These percentages are registry-reported and are quoted as recorded. The registry entries do not give additional event-level detail, timing, severity breakdowns beyond the serious-event listing, or an attribution assessment for each event.

Serious adverse events and other adverse events are separate registry categories and should not be treated as interchangeable. The counts do not indicate how many events occurred per participant, whether the same participant experienced events in more than one category, or whether an event was caused by treatment. Because the trial has no concurrent comparator, these data describe the safety experience recorded among treated participants but do not establish relative risk or comparative safety. The registry record does not report further detail on the deaths or their relationship to treatment.

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

This study addresses a clinically important setting described in the registry: aplastic anemia with thrombocytopenia refractory to immunosuppressive therapy. It evaluates eltrombopag in a single-group pilot design and defines a hematologic response based on platelet, erythroid, or neutrophil improvement. The record also reports serious and other adverse-event counts and lists serious-event terms.

The central primary outcome result is not reported in the registry record. The response criteria alone do not reveal how many participants met them, what portion of the treated group responded, or how precise an estimate would be. Accordingly, the record as represented by these registry entries does not allow a quantitative interpretation of the primary response outcome.

Even if a response count were available, the single-group, non-randomized design would support a description of outcomes in treated participants rather than a randomized comparison. Changes during follow-up could have explanations beyond the intervention, and the record does not provide a control group to help distinguish those influences. The findings should not be framed as proving that eltrombopag caused a response or as establishing a comparative advantage.

The endpoint also combines different forms of hematologic recovery. A participant could qualify through different lineages and baseline-dependent rules, so the responder category would not mean that all participants experienced the same type or magnitude of improvement. The registry does not report component-specific response counts or further clarify the endpoint’s reference to toxicity assessment.

The safety counts provide descriptive information but do not establish causality. Serious events, other adverse events, and deaths are reported for the treated group; without a comparator or further attribution information, these counts cannot quantify the excess risk associated with treatment. The record also does not report the statistical analysis details or missing-data handling needed to assess the primary endpoint fully.

In sum, the record establishes what was studied and how response was defined, but not the primary response result. Its statistical meaning is therefore limited to the design, endpoint framework, and reported safety observations, with substantial uncertainty about efficacy estimation and no randomized comparative inference.

14. Why This Trial Matters Statistically

This trial is statistically notable as a pilot study in a population described as having thrombocytopenia refractory to immunosuppressive therapy. A single-group design can provide experience with treatment delivery and document outcomes observed among treated participants. Its value for statistical interpretation depends on keeping that descriptive role distinct from the stronger causal claims that a randomized comparison can support.

The response endpoint illustrates a common challenge in clinical research: a clinically meaningful concept may be represented by a composite rule that accepts different kinds of improvement. The registry’s definition spans platelet, erythroid, and neutrophil responses, with criteria that depend on baseline status and transfusion history. Such an endpoint can capture more than one pathway to improvement, but an overall responder summary would not show which component contributed unless component results were also reported.

The missing primary response count is consequential. Without the event count, denominator for the observed endpoint, and an uncertainty estimate, readers cannot assess the magnitude or precision of the main outcome from this record. A complete endpoint definition is essential, but it is not a substitute for an outcome result. This distinction is especially important when interpreting a pilot study, whose descriptive findings may inform later research but should not be mistaken for a definitive comparative estimate.

The trial also demonstrates why safety counts require design context. The serious-event and other-event figures document events among treated participants, but a single-group record cannot show how those counts compare with what would occur under another treatment or without the intervention. Descriptive safety reporting is informative, while comparative risk remains unestimated in this design.

Finally, the record leaves important analytical details unreported, including the statistical plan, missing-data handling, and multiplicity approach. These omissions limit assessment of how results were to be analyzed and how robust any reported inference would be. The trial matters statistically because it pairs a defined response framework with a design that supports description rather than randomized comparison, while the primary response result itself is not reported in the registry record.

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 limitation for interpreting the primary outcome is that the registry record does not report the number of participants meeting the response definition. Without that count, an observed response portion and its precision cannot be assessed. The definition is available, but the corresponding primary result is not reported in the registry record.

The single-group design has no concurrent randomized control. It therefore cannot separate treatment-associated changes from other changes over time or establish a comparative treatment effect. The study is also listed as unmasked, and the registry entries do not describe any additional outcome-assessment safeguards.

The composite response definition combines platelet, erythroid, and neutrophil criteria, with eligibility rules tied to pretreatment counts and transfusion status. The registry does not report results by component, so the pattern of responses across lineages cannot be assessed. The endpoint description mentions toxicity assessment but does not specify its role in classifying responders.

The registry entries do not report participant characteristics beyond the stated clinical population, endpoint missingness, withdrawals, or the handling of missing response assessments. They also do not report a statistical analysis plan, sample-size justification, interim-analysis procedures, multiplicity strategy, or confidence interval for the primary outcome. These omissions constrain evaluation of both analysis choices and uncertainty.

Safety interpretation is limited by the lack of a comparator and by the absence of further details about event timing, recurrence, and attribution. The reported serious-event terms and counts are descriptive. They should not be treated as evidence that the intervention caused each event or as a comparative risk estimate.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: A Pilot Study of the Thrombopoietin-Receptor Agonist Eltrombopag in Refractory Aplastic Anemia Patients — NCT00922883.
• Registry record last updated 2020-10-20; status last verified 2020-09.
• Results first posted 2018-07-16.
• 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

This completed Phase 2 study evaluated oral eltrombopag in a single group of patients with aplastic anemia and immunosuppressive-therapy-refractory thrombocytopenia. The registry lists enrollment of 43, no randomization, and no masking. The primary endpoint was the portion of treated participants meeting a hematologic response definition at 12-16 weeks, based on platelet, erythroid, or neutrophil criteria.

The primary response count and response portion are not reported in the registry record. The listed safety information includes serious adverse events affecting 16 participants, other adverse events affecting 38 participants, and 4 deaths. Serious-event terms include febrile neutropenia, sepsis, conjunctival haemorrhage, anaemia, neutropenia, and thrombocytopenia. These observations are descriptive and have no concurrent comparator. The record does not report the statistical plan, missing-data handling, or further analysis details needed to assess the primary outcome’s precision.