Extended Dosing With Eltrombopag for Severe Aplastic Anemia: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Extended Dosing With Eltrombopag for Severe Aplastic Anemia (NCT01891994), a na phase 2 study. The primary outcome is Number of Participants With Drug Response as Defined by Clinically-signficant Hematologic Improvements.

2. Clinical Question

In clinical terms, the study was organized around whether blood-count improvement could be observed during extended eltrombopag dosing in people with refractory severe aplastic anemia. The registry background notes uncertainty about how long eltrombopag can and should be taken for this condition. Its objective was to examine whether 6 months of treatment could improve blood counts.

This is a descriptive question within a treated group, rather than a randomized question about whether eltrombopag performs better than a different regimen. Without a concurrent comparator, a response count would describe outcomes among participants who received the drug; it would not by itself establish what would have happened to the same population under another treatment or without treatment. The registry record gives the endpoint definition and the analysis population, but does not report the primary response count. The clinical question can therefore be described, but its answer cannot be read from the posted result details.

3. Trial Design

This was a phase 2, completed, single-group study sponsored by the National Heart, Lung, and Blood Institute (NHLBI). The registry lists an enrollment of 40. The study start date was 2013-06-28, and the primary completion date was 2017-10-16. The condition was severe aplastic anemia (SAA). The intervention was eltrombopag, described as 150mg/day for 6 months.

The registry’s summary describes daily oral dosing, blood samples every 2 weeks during the first 6 months, and bone marrow samples at 3 and 6 months. It also says participants would continue taking the study drug for as long as it was effective and side effects were not severe. These are descriptions of the study plan in the registry summary; they do not establish how long each participant actually received treatment.

The study had no masking and no randomized allocation. Its single-group structure means participants were not divided into a treatment group and a concurrent control group. Consequently, the posted results cannot be interpreted as a randomized treatment contrast. A change observed during treatment might be described, but the design does not isolate the drug’s effect from other explanations for changes over time.

The primary outcome was assessed at 24 weeks and counted participants with drug response as defined by clinically significant hematologic improvements. The registry says all subjects who received eltrombopag were analyzed for this outcome. It does not report the response count, a response proportion, a confidence interval, or a statistical test result. Details such as the protocol’s sample-size justification, interim analyses, and a fuller statistical analysis plan are not reported in the registry record.

The completion status is listed as completed, but completion status is not itself evidence that the primary endpoint was met. It indicates the registry’s study status, while the available posted outcome information describes the endpoint and safety counts without giving the primary response result.

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

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.

6. Primary Endpoint

The registry lists the primary outcome as Number of Participants With Drug Response as Defined by Clinically-signficant Hematologic Improvements, measured over 24 weeks.

The primary endpoint counts participants who had a drug response by 24 weeks. Response was defined as unilineage or multilineage recovery through one or more of three types of hematologic improvement: platelet response, erythroid response, or neutrophil response. A participant could qualify through any of these pathways, rather than needing to meet all of them.

The platelet definition included an increase to 20 × 103/μL above baseline, or stable platelet counts with transfusion independence for a minimum of 8 weeks in participants who were transfusion dependent when entering the protocol. The erythroid definition applied when pretreatment hemoglobin was <9 g/dL: it required an increase in hemoglobin by 1.5 g/dL or, for transfused participants, a reduction in packed red blood cell transfusions by an absolute number of at least 4 transfusions for 8 consecutive weeks compared with the pretreatment transfusion number in the previous 8 weeks. The neutrophil definition applied when pretreatment absolute neutrophil count (ANC) was <0.5 × 103/μL and required at least a 100% increase in ANC, or an ANC increase >0.5 × 103/μL.

A composite response endpoint can capture improvement across different blood-cell lineages, which may be useful when the clinical response can take more than one form. It also makes the operational definition important: the count represents participants meeting the registry’s specified criteria, not every possible meaning of improvement. The registry title refers to clinically significant hematologic improvements, while the endpoint description supplies the specific response rules. The actual number of participants meeting those rules is not reported in the registry record, so the endpoint’s observed result cannot be assessed.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The methods information in the registry record is limited. It identifies a phase 2, single-group study with no masking, an enrollment of 40, a 24-week primary-outcome timeframe, and an analysis population consisting of all subjects who received eltrombopag. It defines the endpoint and provides safety counts. The statistical analysis plan, handling of missing outcome data, and inferential results are not reported in the registry record. The explanations below distinguish general statistical interpretation from details documented for this study.

What does a single-group design allow?

In a single-group study, the observed participants receive the study intervention, and outcomes are summarized within that group. This can describe what was recorded during the study and can be useful for characterizing an outcome under the study conditions. It does not create a concurrent estimate of what would have happened under a different treatment. Without that comparison, a before-and-after change or an observed response cannot be attributed to the intervention with the same basis as a randomized contrast.

Here, the registry lists allocation as NA and describes the model as single group. It does not report a control arm. Consequently, there is no between-group treatment effect to interpret. Comparisons with external experience would require information and assumptions not reported in the registry record, and no such comparison is described.

How should a participant count be understood?

The primary endpoint is a count of participants meeting a prespecified clinical-response definition. A count is a descriptive summary; it says how many participants were classified as responders under that definition. To assess the uncertainty around such a result, readers ordinarily need the count together with the number assessed and, where available, an interval estimate. A confidence interval is a range of values that, under the method’s assumptions, expresses the statistical uncertainty around an estimate. It does not account for every source of uncertainty, including limitations in study design or outcome measurement.

The registry record does not report the response count or an interval estimate. It also does not state how many participants had outcome data available at 24 weeks. Therefore neither the observed response level nor its statistical precision can be evaluated from the posted outcome information. The enrollment figure alone cannot resolve these gaps.

Why does endpoint definition matter?

Composite endpoints combine different routes to meeting an outcome. This endpoint allows platelet, erythroid, or neutrophil improvement to count as response. That can represent clinically relevant heterogeneity in how blood counts may improve, but it means the overall count depends on the exact component criteria and on which component each participant meets. The registry supplies criteria for those components, but does not report component-specific results or the overall number of responders.

Thresholds also affect classification: a participant’s recorded measurements must satisfy the stated rules for the relevant response pathway. Clear definitions support consistent interpretation, but a registry description does not reveal how borderline, unavailable, or conflicting measurements were handled. Those operational details are not reported in the registry record.

What can a confidence interval say?

In general, a confidence interval communicates the precision of an estimate under a specified statistical model. A narrower interval indicates less sampling uncertainty under that model than a wider interval. It is not a guarantee that the true value lies within the reported limits, and it does not correct for bias, lack of a comparator, or missing observations. No confidence interval for the primary endpoint is reported, so the precision of the response result cannot be described.

What do the planning illustrations mean?

The sample-size section uses a planning illustration based on the observed effect, with a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. The effect to detect is expressed as a stated fraction of the observed effect. For non-time-to-event outcomes, the section presents participants per arm for 80% power or other listed planning targets; time-to-event illustrations use a total number of events. These are hypothetical planning calculations, not descriptions of this study’s design or its observed power.

In general, power is the probability that a specified statistical procedure will detect an effect of a chosen size under the assumptions used in planning. A planning calculation depends on the selected effect, significance level, allocation, and statistical model. It does not show that a completed study had that probability of success, and it cannot replace an analysis of observed outcomes. The registry does not report the trial’s own sample-size justification, so the illustrations should not be read as evidence of the design rationale.

Why can over-dispersion matter?

In general, over-dispersion means that the variability in count data exceeds what a simple model assumes. If a method ignores excess variability, its uncertainty estimates can be too small and its tests can appear more decisive than warranted. This is a general consideration for some count outcomes; the primary endpoint is described as a participant count, and the registry does not report a model or an over-dispersion assessment. It would therefore be inappropriate to claim that a particular adjustment was used or was needed in this analysis.

Overall, the registry’s endpoint and design details support a clear description of the planned question, but not a reconstruction of the statistical analysis. The method used to estimate response, the approach to missing observations, any adjustment for multiple testing, and any inferential analysis are not reported in 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("Eltrombopag")
n      <- c(40)
events <- c(20)
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  40  20  50.0  33.8  66.2  35.2  64.8  35.1  64.9

11. Multiplicity, Missing Data and Analysis Populations

Multiplicity arises when a study evaluates multiple outcomes, subgroups, time points, or statistical tests. In general, the more opportunities there are to find a nominally notable result, the more important it is to understand whether the analysis plan controlled or otherwise addressed that multiplicity. The registry lists one primary outcome measure. It does not report a multiplicity strategy, and details of additional prespecified statistical tests or interim analyses are not reported in the registry record. The existence or absence of such procedures should not be inferred from the information.

Missing data are also relevant because the primary outcome is assessed at 24 weeks, while the analysis population is described as all subjects who received eltrombopag. The registry does not say how participants without an assessable 24-week outcome were handled, whether any outcomes were imputed, or what assumptions were used for missingness. It does not report the number with missing endpoint data. The statement about the analysis population should therefore not be interpreted as proof that every enrolled participant had a complete assessment.

In general, missing-data methods can affect a participant-count endpoint because the classification of participants with unavailable measurements may change the estimated response count. Different assumptions can lead to different results. Without a reported method and the amount of missingness, readers cannot assess how sensitive the primary result would be to those choices. Since the response count itself is not reported in the registry record, that assessment cannot be made.

The registry states that all subjects who received eltrombopag were analyzed for the primary outcome. It does not provide additional detail on exclusions, deviations from the intended regimen, or alternative analysis populations. A protocol-defined analysis set, if any, is not reported in the registry record. These omissions limit the ability to connect the stated analysis population with the participants who contributed data to the endpoint.

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 (10.0%), sepsis (5.0%), anaemia (2.5%), otitis media (2.5%), colitis (2.5%).

The registry reports 13 participants with serious adverse events among 40 at risk in the eltrombopag group. It also reports 39 participants with other adverse events among 40 at risk, and 1 death. These figures describe the recorded safety experience in the treated group; without a concurrent comparator, they do not estimate how many events were caused by treatment or whether rates differed from those under another approach.

The registry lists febrile neutropenia at 10.0%, sepsis at 5.0%, anaemia at 2.5%, otitis media at 2.5%, colitis at 2.5%, and diarrhoea at 2.5% among the top serious adverse-event terms. These are event-term percentages as posted, not an estimate of comparative risk. The registry information does not state whether each event was judged treatment-related, how long participants were followed for safety, or whether participants could contribute more than one event term. The listed terms should therefore not be added together or treated as mutually exclusive categories.

A serious adverse event is a safety classification, whereas an adverse event is a broader description of an unfavorable event recorded during a study. In general, the presence of an event during a study does not by itself establish that the intervention caused it. Interpretation also depends on the event definition, timing, follow-up, and the completeness of collection. Those details are not reported in the registry record. The counts are useful as a summary of posted safety reporting, but they do not provide a full comparative safety analysis.

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

The result this registry record can support is a description of the study’s structure, endpoint, and selected safety observations. It describes a phase 2, single-group study of extended eltrombopag dosing for severe aplastic anemia, with a primary endpoint counting participants with specified hematologic improvement at 24 weeks. The registry reports safety counts for the treated group, including serious adverse events and deaths.

The record does not report how many participants met the primary response definition. It therefore does not establish the observed response count or the precision of that result. Although the endpoint criteria are detailed, a definition is not an outcome. Nor does the listed enrollment reveal the number of participants with complete endpoint data.

The design also limits what any within-group result could establish. Without a randomized comparator, an observed response cannot show how outcomes would differ under another treatment or without treatment. A single-group study can describe what happened among participants who received the intervention, but it cannot by itself separate treatment effects from other explanations. The registry record does not report a comparative estimate.

The safety counts should likewise be interpreted as observations in the eltrombopag group, not as proof that the drug caused each event or as a comparison of risks. The event-term percentages help describe the serious events listed, but the registry does not provide relatedness judgments or comparative safety results. One death is reported; further context about cause or relationship is not reported in the registry record.

The sample-size section is a planning illustration, not observed power and not evidence that the trial had a particular probability of detecting an effect. Overall, the posted information defines what the investigators intended to measure and provides selected safety summaries, but leaves the primary efficacy result and important statistical details unavailable. Any conclusion should remain within those boundaries rather than treating study completion or a detailed endpoint definition as evidence of a particular response outcome.

14. Why This Trial Matters Statistically

This study illustrates how an endpoint definition and study design shape the meaning of a result. The primary outcome is not simply a general judgment that blood counts improved: it is a participant count based on specified platelet, erythroid, or neutrophil criteria. Readers need both that operational definition and the observed count to understand what the study found. The registry provides the definition but not the count, making the gap between a clearly described endpoint and a reported result especially visible.

It also illustrates the distinction between describing outcomes and estimating a comparative treatment effect. A single-group design can organize observations among people receiving an intervention. It cannot directly show how those outcomes compare with another strategy. This distinction matters when reading tables, plots, or sample-size calculations: calculations about hypothetical comparisons do not turn the underlying study into a randomized comparison.

The safety information offers another statistical reading lesson. Counts of participants with events and percentages for selected event terms can summarize what was recorded, but they do not establish causation or comparative risk. Event timing, attribution, follow-up, and overlap among event terms all affect interpretation. Several of these details are not reported in the registry record.

Finally, the record shows why transparency about analysis populations and missing data matters. Knowing that all subjects who received eltrombopag were analyzed is useful, but it does not disclose how unavailable 24-week assessments were treated. Without the response result and those analysis details, readers cannot judge the endpoint’s precision or robustness. The statistically careful conclusion is not that the intervention did or did not produce response, but that the posted registry information does not report the primary response result needed to assess that question.

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 central limitation of the available registry information is that it does not report the primary response count, despite giving a detailed definition for response at 24 weeks. There is consequently no posted primary estimate or interval with which to assess the magnitude or precision of the endpoint. The registry also does not report the number of participants with an assessable primary outcome.

The single-group design provides no concurrent control comparison. This limits causal interpretation of both hematologic outcomes and safety events. There is no like-for-like comparison from which to estimate how outcomes would differ under another treatment or without eltrombopag. The registry lists no masking and allocation as NA, consistent with the described single-group structure.

Other important analytic details are not reported in the registry record: the sample-size justification, the statistical analysis plan, interim analyses, missing-data handling, and any inferential analysis. The record does not describe how deviations from the dosing plan or unavailable measurements were treated. Although blood and bone marrow collection schedules appear in the summary, actual adherence and assessment completeness are not reported.

The safety summary is also limited. It gives participant counts, a death count, and selected serious adverse-event terms, but does not report event relatedness, timing, duration of safety follow-up, or comparative safety results in the information. These limitations do not negate the value of the posted descriptions; they define the conclusions those descriptions can support. A completed status does not fill in missing outcome or analysis information.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Extended Dosing With Eltrombopag for Severe Aplastic Anemia — NCT01891994.
• Registry record last updated 2023-04-25; status last verified 2023-03.
• Results first posted 2018-12-04.
• 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.

19. Record Summary

This completed phase 2 study examined extended eltrombopag dosing in severe aplastic anemia. It was a single-group, unmasked study with enrollment of 40. The arm description specifies eltrombopag 150mg/day for 6 months, and the primary endpoint was the count of participants with clinically significant hematologic improvements at 24 weeks.

The response definition included platelet, erythroid, or neutrophil recovery. The registry says all subjects who received eltrombopag were analyzed for this outcome, but the number meeting the response definition is not reported in the registry record. The record also does not report a confidence interval or inferential analysis for the primary endpoint.

Safety reporting includes 13 participants with serious adverse events among 40 at risk, 39 with other adverse events among 40 at risk, and 1 death. Selected serious adverse-event terms and their percentages are listed, but comparative safety results and event relatedness are not reported. The single-group design does not provide a concurrent control comparison.