Evaluation of Dose Response Relationship, Safety and Efficacy of GSK1278863 in Hemodialysis-dependent Subjects With Chronic Kidney Disease Associated Anemia: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Evaluation of Dose Response Relationship, Safety and Efficacy of GSK1278863 in Hemodialysis-dependent Subjects With Chronic Kidney Disease Associated Anemia (NCT01977482), a randomized phase 2 study. The primary outcome is Change From Baseline in Hemoglobin (Hgb) at Week 4.

2. Clinical Question

The study was Phase IIB, randomized, triple-masked, parallel-group, and multicenter. Its stated purpose was to inform selection of starting doses and dose-adjustment regimens for Phase 3 trials. That aim makes the dose-response estimates central to interpreting the study, while the later outcomes address hemoglobin maintenance and safety during continued, adjusted treatment.

3. Trial Design

This was a Phase IIB randomized, triple-masked, parallel-group, multicenter study with active control. Participants with hemodialysis-dependent anemia associated with chronic kidney disease were switching from a stable dose of recombinant human erythropoietin. The registry reports enrollment of 216. The assigned GSK1278863 doses were 4 mg, 6 mg, 8 mg, 10 mg, and 12 mg; a control group received matching placebo initially.

During the first 4 weeks, participants took their assigned blinded oral treatment once daily. The control group received placebo during this period. From Week 4, study medication continued once daily, with dose adjustment based on hemoglobin levels every 4 weeks through Week 24. In the control group, recombinant human erythropoietin was administered from Week 4 and adjusted using hemoglobin levels. Thus, the early dose-response comparison concerns the assigned initial treatment, whereas later treatment included dose adjustment and an active comparator.

The primary endpoint was change from baseline in hemoglobin at Week 4. Baseline was defined as the average of measurements at Week -4, Week -2, and Day 1; the endpoint change was Week 4 hemoglobin minus that baseline value. The registry says the dose-response dataset included all non-missing data collected through Week 4, with a specified imputation for participants who had a Week 2 measurement but lacked a Week 4 measurement.

The registry lists 18 outcome measures and 13 analyses. It also lists secondary outcomes relating to Week 24 hemoglobin, time in and outside the target range, stopping criteria, EPO, VEGF, pharmacokinetic parameters, and hepcidin. The original sample-size justification, detailed randomization procedure, and full statistical analysis plan are 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.

Before assignment: A total of 216 participants with a stable hemoglobin (Hgb) between 9.0-11.5 grams (g)/deciliter (dL) (France sites only: 10.0-11.5 g/dL) were randomized.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: ITT Population. Only participants with data available at specific time point 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.

Only the first 6 of 7 groups are shown.

6. Primary Endpoint

The registry lists the primary outcome as Change From Baseline in Hemoglobin (Hgb) at Week 4, measured over Baseline (Week -4, Week-2 and Day 1) and Week 4.

The primary endpoint measures the change in hemoglobin concentration from baseline to Week 4. Hemoglobin was recorded in g/dL. Baseline was the average of the screening values at Week -4, Week -2, and Day 1, and change was defined as the Week 4 value minus this baseline. A positive change therefore represents an increase relative to the participant’s baseline measurement, while a negative change represents a decrease.

This endpoint was used to characterize the early dose-response relationship after participants switched from stable recombinant human erythropoietin treatment. The registry describes a four-parameter Emax model for that purpose. Such a model is a general way to represent a response that changes with dose and may approach a plateau, rather than assuming that each increase in dose produces the same increment in response. It allows the analysis to estimate features of the response curve, including its placebo reference, upper response, dose associated with an intermediate response, and slope.

The endpoint is not the same as hemoglobin status later in the study. From Week 4, dose adjustment was based on hemoglobin levels, and the control group began recombinant human erythropoietin. Week 24 measures therefore reflect a later treatment period with adjustment, while the primary endpoint is the early change through Week 4. The registry’s primary analysis population is the ITT Population, restricted at each time point to participants with data available for that time point.

7. Results Posted in the Registry

• GSK1278863 4 mg: mean -0.24 versus -0.64 on control; difference 0.40 (95% CI -0.01 to 0.81), standardized effect d = 0.44; z-test p = 0.053.
• GSK1278863 6 mg: mean 0.08 versus -0.64 on control; difference 0.72 (95% CI 0.28 to 1.16), standardized effect d = 0.72; z-test p = 0.001.
• GSK1278863 8 mg: mean 0.42 versus -0.64 on control; difference 1.06 (95% CI 0.70 to 1.42), standardized effect d = 1.30; z-test p = <0.001.
• GSK1278863 10 mg: mean 0.64 versus -0.64 on control; difference 1.28 (95% CI 0.83 to 1.73), standardized effect d = 1.25; z-test p = <0.001.

8. Secondary Efficacy Endpoints

The registry also posts results for secondary outcomes. For each outcome below, the endpoint type, the test applied to the posted summary statistics, the result and its interpretation are given in turn. Secondary endpoints are usually not adjusted for multiple testing, so they support rather than confirm the primary finding.

Hgb Concentration at Week 24

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard deviations.
• Result:
  • GSK1278863 4 mg: mean 10.29 versus 10.56 on control; difference -0.27 (95% CI -0.83 to 0.29), standardized effect d = -0.29; z-test p = 0.343.
  • GSK1278863 6 mg: mean 10.54 versus 10.56 on control; difference -0.02 (95% CI -0.56 to 0.52), standardized effect d = -0.02; z-test p = 0.943.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between GSK1278863 4 mg and the control. The interval gives the range of effects that remain compatible with the data.
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between GSK1278863 6 mg and the control. The interval gives the range of effects that remain compatible with the data.

Number of Participants With Hgb in the Target Range at Week 24

• Endpoint type: Binary: the number of participants with the event, compared as proportions.
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • GSK1278863 4 mg: 72.2% (13/18) versus 58.3% (14/24) on control; risk difference 13.9 percentage points (95% CI -14.7 to 42.5), risk ratio 1.24 (0.79–1.93); unadjusted two-proportion z-test p = 0.353.
  • GSK1278863 6 mg: 59.3% (16/27) versus 58.3% (14/24) on control; risk difference 0.9 percentage points (95% CI -26.1 to 28.0), risk ratio 1.02 (0.64–1.61); unadjusted two-proportion z-test p = 0.947.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between GSK1278863 4 mg and the control. The interval gives the range of effects that remain compatible with the data.
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between GSK1278863 6 mg and the control. The interval gives the range of effects that remain compatible with the data.

Maximum Observed Change From Baseline in Erythropoietin (EPO)

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard deviations.
• Result:
  • GSK1278863 4 mg: mean 48.86 versus 1,946 on control; difference -1,898 (95% CI -4,660 to 865.01), standardized effect d = -0.32; z-test p = 0.178.
  • GSK1278863 6 mg: mean 36.63 versus 1,946 on control; difference -1,910 (95% CI -4,672 to 852.55), standardized effect d = -0.32; z-test p = 0.175.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between GSK1278863 4 mg and the control. The interval gives the range of effects that remain compatible with the data.
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between GSK1278863 6 mg and the control. The interval gives the range of effects that remain compatible with the data.

The secondary outcomes address different aspects of treatment beyond the early Week 4 dose-response endpoint. They include Week 24 hemoglobin concentration, hemoglobin target-range measures, stopping criteria, EPO and VEGF changes, pharmacokinetic parameters, and hepcidin change. These measures can add context about later treatment and biological or safety-related outcomes, but they are not interchangeable with the primary endpoint.

The posted Week 24 hemoglobin summaries are descriptive group means and standard deviations, while the target-range outcome reports counts of participants in the range. The available group sizes for these later outcomes differ from the Week 4 analysis sizes. These summaries therefore describe participants with data for the particular endpoint and time point; they should not be treated as a direct continuation of the Week 4 comparison without accounting for the later dose-adjustment period and the change in control treatment.

The secondary results do not simply confirm or overturn the primary dose-response result. They address later outcomes under treatment adjustments, and their pattern is not a single common effect measure. The registry also lists additional outcomes for which specific numerical results are not reported in the registry record. Secondary analyses are usually interpreted with more caution when many outcomes or comparisons are examined, especially when the record does not state a multiplicity strategy. Unadjusted findings can be useful descriptively and for generating questions, but carry less weight as confirmatory evidence than a clearly prespecified primary analysis. The full hierarchy and multiplicity plan are not reported in the registry record.

Secondary outcomes registered for this trial, for reference (only those above were recalculated):

• Hgb Concentration at Week 24
• Percentage of Time Within, Below, and Above Hgb Target Range Between Weeks 20 and 24
• Number of Participants With Hgb in the Target Range at Week 24
• Number of Participants Reaching Pre-defined Hgb Stopping Criteria
• Maximum Observed Change From Baseline in Erythropoietin (EPO)
• Maximum Observed Percent Change From Baseline in Vascular Endothelial Growth Factor (VEGF)
• Population Plasma PK Parameters of GSK1278863 and Metabolites
• Percent Change From Baseline in Hepcidin at Week 24

9. Statistical Interpretation

10. Statistical Methods Explained

General statistical explanation: A continuous endpoint such as hemoglobin change can be described by a group mean and standard deviation. The mean describes the center of the observed group values; the standard deviation describes how dispersed those values are around the mean. These measures answer different questions. A mean difference compares two group averages, whereas a standardized effect size expresses a difference relative to variability. Neither measure alone describes the full uncertainty in an estimate.

How was dose response modeled?

The registry states that a four-parameter Emax model was used to model the Week 4 dose-response relationship. In general, an Emax curve represents a dose-associated response that can rise and approach a plateau. The registry defines E0 as the expected baseline-to-Week 4 hemoglobin change for a participant receiving placebo and experiencing the average baseline hemoglobin observed in the study. Emax is the expected change for a participant receiving the highest dose beyond which the model expects no further increase in response. ED50 is the dose attaining an intermediate response, and Gamma is the slope parameter. Alpha is the coefficient for the centered-baseline covariate.

The registry reports estimates for E0, Emax, ED50, Gamma, variance, and Alpha, along with 95% confidence intervals. It also reports derived dose estimates, including the minimally effective dose, target dose, and doses associated with specified changes. Those estimates depend on the fitted model and its definitions. They should be interpreted as model-based estimates, not as separately randomized comparisons at every possible dose. The registry does not report a p-value for these model parameters.

What do the pairwise recalculations represent?

The recalculated dose-versus-control estimates are simple unadjusted comparisons based on posted group summary statistics. For the continuous Week 4 endpoint, the mean difference is the dose-group mean minus the control-group mean. Its confidence interval expresses uncertainty around that estimated difference under the recalculation. The unadjusted p-value assesses compatibility with a no-difference reference under the test used for that recalculation; it does not account for testing multiple doses. These estimates are not adjusted for baseline or other covariates in the way that a fitted dose-response model may be.

The recalculations use the posted standard deviations and group sizes. Their assumptions do not reproduce the sponsor’s full model, imputation details, or any adjustments that may have been used for the sponsor-reported analysis. Since the registry reports no discrepancy between the sponsor analysis and recalculation, the recalculations can be read as a separate, transparent summary of pairwise contrasts, while the sponsor’s model remains the analysis specifically described for dose response.

Why does dispersion matter?

General statistical explanation: Greater variation within groups makes a group mean less precise when the number of observations is held constant. Standard deviations help show this variation, while confidence intervals combine the observed variation with the amount of available data. Where group sizes differ, the precision of comparisons can also differ. This is relevant to the 12 mg group, whose Week 4 analysis size is 19, compared with 39 or 40 in several other groups.

The registry reports the Week 4 analysis as ITT, with only participants who have data at the specific time point analyzed. It also describes an imputation for participants with a Week 2 measurement but missing Week 4 hemoglobin: their Week 4 change was imputed as twice their Week 2 change. The record does not give a complete account of other missing-data handling or sensitivity analyses. Readers should therefore distinguish the stated imputation from any assumptions not documented in the registry.

How should the p-values and intervals be read?

General statistical explanation: A p-value is not the probability that a treatment effect is true, and it does not measure the size or clinical importance of an effect. A confidence interval gives a range of values compatible with the data and the analysis assumptions at the stated confidence level. Wider intervals indicate less precision than narrower intervals. An interval that includes zero for a mean difference is compatible with both a negative and a positive difference under that interval estimate; an interval entirely on one side of zero is more directionally constrained, but still does not establish clinical importance.

The pairwise p-values are marked unadjusted. The registry does not report whether or how multiplicity was handled for the primary dose-response model or secondary analyses. Multiple testing can make isolated unadjusted findings less persuasive as confirmatory evidence. The original analysis plan, including any prespecified hierarchy or multiplicity adjustment, is not reported in the registry record.

What does the target-dose model add?

The reported model includes estimates of doses associated with changes of -0.25 g/dL, 0.25 g/dL, 0.5 g/dL, 0.75 g/dL, and 1 g/dL. For example, the estimate for a change of 0.5 g/dL is 8.542 mg, with a 95% confidence interval from 6.931 to 10.523. Such quantities translate a fitted curve into dose-based summaries, but their uncertainty comes from the model and they should not be mistaken for directly observed group averages. The registry’s purpose statement says these dose-response data were intended to support dose selection and adjustment regimen decisions for later trials.

11. Confidence Intervals and Precision

The precision of the Week 4 comparisons varies by dose. For the recalculated mean differences, the confidence interval for 4 mg is -0.005 to 0.805 g/dL; for 6 mg it is 0.283 to 1.157; for 8 mg it is 0.699 to 1.421; for 10 mg it is 0.832 to 1.728; and for 12 mg it is 0.633 to 1.867. These are unadjusted intervals from the pairwise recalculations, not intervals from the sponsor’s dose-response model.

The 4 mg interval includes zero and therefore does not tightly establish the direction of that contrast under the recalculation. The intervals for the other dose comparisons lie on the positive side of zero, but they still span a range of possible mean differences. The 12 mg interval is wider than those for several other doses; its analyzed group size is 19, compared with 39 or 40 in several other groups. Unequal group size and observed variability contribute to differences in precision.

The registry also reports 95% confidence intervals for the model parameters and derived dose estimates. For instance, ED50 is estimated as 33.531 mg, with an interval from 15.566 to 48.948; the interval is broad relative to the point estimate. The estimated minimally effective dose is 0.418 mg, with an interval from 0.000 to 2.342. These are model-based estimates and their intervals reflect uncertainty in the fitted curve, not direct observations at each dose value.

General statistical explanation: A confidence interval is an estimate of uncertainty, not a range that guarantees the true value lies inside it for an individual study. Its width depends on the data and analysis assumptions. The registry does not give a clinical threshold for a meaningful difference in the primary endpoint, so the intervals alone cannot determine whether an observed difference is clinically important. They describe statistical precision, not medical benefit or harm.

12. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The sample-size illustrations are planning calculations based on the observed effect in each comparison, not the trial’s original design justification. For these continuous outcomes, the figures are participants per arm, not a total across groups. The general calculation assumptions stated by the registry are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is set at the stated fraction of the observed effect. For time-to-event outcomes, the registry says the figures would represent total events, but the primary outcome is continuous.

The 80% power column means the calculated number of participants per arm for a design that would have an 80% probability of detecting the specified effect under the listed assumptions, if that effect and those assumptions were true. The 90% power column uses a higher probability of detection and therefore generally calls for more participants per arm. The rows for 100%, 75%, and 50% of the observed effect illustrate how the planning requirement changes when the target effect is smaller than the observed estimate. Smaller effects need more participants because they are harder to distinguish from random variation.

These figures are not observed power. They are planning illustrations derived from the posted effect estimates and do not show the probability that the completed study’s result is correct. They also do not establish that the original study was designed using these assumptions. The original trial’s sample-size justification and planning assumptions are not reported in the registry record.

13. How the Sample Size Is Derived

Power is the probability that a study of a given size rejects the null hypothesis when a specific true effect exists. It depends on four things: the effect to detect, the variability of the outcome, the significance level (here a two-sided α of 0.05, so z_1−α/2 = 1.960) and the desired power (z_1−β = 0.842 for 80% and 1.282 for 90%).

Formula

n is the number of participants in each arm, Δ is the difference in means to detect and σ is the common standard deviation.

Worked calculation for this trial

1. Observed difference Δ = 0.400; pooled standard deviation σ = 0.913.
2. n = 2 × (1.96 + 0.842)² × 0.913² / 0.400² = 81.7, rounded up to 82 participants per arm for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
delta <- 0.4                      # observed difference in means
sigma <- 0.91251356154306                     # pooled standard deviation
n_per_arm <- function(power, frac = 1) {
  ceiling(round(2 * (z_a + qnorm(power))^2 * sigma^2 / (delta * frac)^2, 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

# Check against R's built-in function (t-based, so very slightly larger)
cat("power.t.test:", round(power.t.test(delta = delta, sd = sigma, power = 0.80)$n, 1), "\n")

# Simulation: exact two-sample t-test using simulated means and variances
set.seed(1977482)
n    <- n_per_arm(0.80)
reps <- 20000
m_diff <- rnorm(reps, delta, sigma * sqrt(2 / n))             # difference in sample means
s2     <- sigma^2 * rchisq(reps, 2 * (n - 1)) / (2 * (n - 1))   # pooled sample variance
tstat  <- m_diff / sqrt(s2 * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(tstat) > qt(0.975, 2 * (n - 1))), reps))

Output from running the code

100% of observed effect: n per arm = 82 (80% power), 110 (90% power)
 75% of observed effect: n per arm = 146 (80% power), 195 (90% power)
 50% of observed effect: n per arm = 327 (80% power), 438 (90% power)
power.t.test: 82.7 
Simulated power at n = 82 per arm: 0.798 (20000 replicates)

The simulation draws many sets of group means and variances at the planned sample size and counts how often the two-sample t-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

14. Analyses Reported by the Sponsor

The registry reports a four-parameter Emax model for the primary dose-response analysis and presents estimates for its parameters and derived dose quantities. This is the sponsor-reported analysis most directly aligned with the stated objective of describing dose response over the first 4 weeks. The registry does not provide a p-value for the listed model parameters.

The recalculated estimates are different in purpose: they are unadjusted, pairwise comparisons of each dose group with control using posted summary statistics. They provide straightforward mean differences, confidence intervals, and unadjusted p-values. They do not reproduce the Emax model, its baseline covariate, or its dose-response curve. The recalculations should therefore be interpreted as a supplementary summary, not as a replacement for the sponsor’s model.

The registry indicates no discrepancy between the recalculation and sponsor analysis. The sponsor’s own analysis therefore does not need to be displaced by a conflicting recalculated result. The reported model estimates and the pairwise summaries can be considered together, with attention to their different methods and questions. Neither set of results by itself establishes clinical importance; the registry does not report a threshold for a clinically meaningful Week 4 difference.

15. Multiplicity, Missing Data and Analysis Populations

The primary endpoint is identified as change from baseline in hemoglobin at Week 4, and the registry describes a four-parameter Emax model for the dose-response analysis. However, the registry does not state whether the model analysis included a formal hypothesis test or how its uncertainty was assessed beyond the reported confidence intervals. It also does not report a multiplicity adjustment for the unadjusted recalculated comparisons across the dose groups.

Multiplicity matters when several doses, endpoints, time points, or contrasts are examined. In general, repeated testing increases the chance of observing at least one apparently unusual result even when no underlying difference exists. This does not invalidate the reported estimates, but it affects how strongly isolated unadjusted p-values should be interpreted. Whether the sponsor used a prespecified testing hierarchy, adjusted significance thresholds, or another control of multiplicity is not reported in the registry record.

The primary analysis population is described as the ITT Population, with only participants having data at the particular time point analyzed. This statement does not mean that every randomized participant contributed to every outcome. At Week 4, the registry specifies that participants with a Week 2 hemoglobin measurement but no Week 4 measurement were included using an imputed change equal to twice the Week 2 change. The record does not report other missing-data assumptions, sensitivity analyses, or the reasons for missing observations.

The registry does not provide enough detail to determine how departures from the planned treatment schedule, discontinuation, or other intercurrent events were handled in the primary model. It likewise does not report a complete statistical analysis plan or the original sample-size justification. These omissions limit assessment of prespecification and robustness; they do not change the arithmetic of the posted summaries, but they constrain conclusions about how fully the results represent the intended analysis.

16. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

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: device related sepsis (5.3%), cardiac arrest (5.1%), gastroenteritis (5.1%), acute myocardial infarction (2.6%), angina pectoris (2.6%).

The registry reports participants with serious adverse events and participants with other adverse events by treatment group. The serious-event counts are 11 in control, 10 at 4 mg, 8 at 6 mg, 8 at 8 mg, 11 at 10 mg, and 1 at 12 mg. The corresponding group sizes at risk are 39, 39, 40, 39, 40, and 19. Other adverse event counts are also presented by group. These are observed participant counts, not estimates of causal treatment effects.

Serious adverse events are distinct from the broader category of other adverse events. The registry lists serious-event terms including device related sepsis, cardiac arrest, gastroenteritis, acute myocardial infarction, angina pectoris, and angina unstable, with the percentages recorded for those terms. The denominators underlying those listed term percentages are not reported in the registry record, so they should not be interpreted as rates without additional information.

Deaths are not reported in the registry record. The count differences across treatment groups cannot establish that treatment caused or prevented an event. Group sizes, exposure duration, event definitions, and clinical context matter for safety interpretation, and the registry summaries do not provide enough detail to make causal comparisons.

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

The results describe an early hemoglobin response after switching from stable recombinant human erythropoietin treatment to blinded GSK1278863 or placebo. At Week 4, the reported group means show a generally increasing pattern from control through the assigned GSK1278863 doses, with the 12 mg mean close to the 10 mg mean. The sponsor’s Emax model provides a formal dose-response description and estimates of curve features and dose-related quantities.

This does not establish that a particular dose is appropriate for an individual, that the observed response is clinically important, or that the same pattern would occur in another population. The registry does not state a threshold for a clinically meaningful primary-endpoint change. Statistical differences in group means and model-based dose estimates answer a different question from individual benefit, long-term outcomes, or treatment selection.

The primary endpoint concerns Week 4, while the trial’s later period involved dose adjustment based on hemoglobin levels and a switch from placebo to recombinant human erythropoietin in the control group. Later outcomes therefore address a different treatment context. They should not be read as simple extensions of the initial randomized dose comparison.

The reported confidence intervals show uncertainty around both pairwise estimates and model parameters. Unadjusted pairwise p-values do not account for multiplicity, and the registry does not report a complete multiplicity plan. Secondary results offer additional descriptive information but do not substitute for the primary analysis. Safety counts likewise describe events observed in the groups and cannot by themselves identify causal effects. The findings are most appropriately understood as a dose-ranging evidence base intended to inform later research, with uncertainty and reporting limitations retained.

18. Why This Trial Matters Statistically

This trial is statistically informative because its primary aim was dose response, rather than only a single comparison between an intervention and control. Multiple assigned dose groups provide information about how the early hemoglobin outcome varies across doses. The sponsor-reported Emax model further expresses that pattern as a curve, with estimates for the placebo reference, maximum response, intermediate-response dose, and slope.

The design also illustrates why the choice of estimand and time point matters. The primary endpoint is measured at Week 4, during the initial blinded dose period. Treatment was then adjusted according to hemoglobin levels, and the control group began active recombinant human erythropoietin. A Week 24 summary consequently answers a different question from the early dose-response analysis. Keeping those periods distinct avoids treating measurements from different treatment conditions as if they were one unchanged comparison.

The posted summaries also demonstrate the role of uncertainty. Group means and standard deviations describe observed outcomes, while confidence intervals show that estimates have varying precision. The dose groups do not all have the same number of analyzed participants, and this is especially relevant to the 12 mg group. Unadjusted pairwise results are useful for describing contrasts, but multiple dose comparisons and secondary outcomes make multiplicity and prespecification important.

The registry’s stated purpose was to inform starting-dose selection and dose-adjustment regimens for later trials. That makes the fitted curve and its uncertainty relevant to future study planning. At the same time, model-based dose summaries are estimates rather than direct observations at every dose, and their interpretation depends on the model. The record supports a careful statistical description, not a clinical recommendation.

19. 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 primary outcome is an early measure at Week 4, and later treatment differed from the initial randomized period because doses were adjusted and the control group moved from placebo to recombinant human erythropoietin. This limits direct interpretation of later outcomes as continued comparisons of the original assigned treatments.

The analyzed group sizes vary across dose groups and outcomes. In particular, the Week 4 12 mg group has 19 participants, while several other Week 4 groups have 39 or 40. The registry says only participants with time-point data were analyzed and describes a specific imputation for a missing Week 4 measure when a Week 2 measure was available. Broader missing-data handling and sensitivity analyses are not reported in the registry record.

The pairwise estimates are unadjusted recalculations from posted summaries, while the primary analysis is described as a four-parameter dose-response model. They are not methodologically equivalent. The registry does not report p-values for the model parameters, a complete multiplicity strategy, the original sample-size justification, or the detailed statistical analysis plan. The listed model-derived dose estimates should therefore be interpreted with their confidence intervals and in light of the model assumptions.

Many secondary outcomes are listed, but their full numerical results are not reported in the registry record. The record also does not report denominators for the listed serious-event term percentages or a death count. These reporting gaps restrict evaluation of secondary findings and safety patterns. The registry summaries do not establish clinical importance or causal effects for individual adverse events.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Evaluation of Dose Response Relationship, Safety and Efficacy of GSK1278863 in Hemodialysis-dependent Subjects With Chronic Kidney Disease Associated Anemia — NCT01977482.
• Registry record last updated 2018-06-08; status last verified 2018-05.
• Results first posted 2017-02-14.
• 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.

23. Record Summary

The registry describes a completed Phase IIB randomized, triple-masked, parallel-group, multicenter dose-ranging study with enrollment of 216. It evaluated GSK1278863 at 4 mg, 6 mg, 8 mg, 10 mg, and 12 mg against a control that received placebo during the initial 4 weeks and recombinant human erythropoietin thereafter. The primary endpoint was change from baseline hemoglobin at Week 4 in hemodialysis-dependent participants with chronic kidney disease-associated anemia switching from stable recombinant human erythropoietin.

The posted group summaries show increasing Week 4 mean changes across most assigned doses, and the registry reports a four-parameter Emax model with parameter and dose estimates. Pairwise recalculations are unadjusted and have different purposes from the sponsor’s model. Later secondary outcomes and safety counts provide additional descriptive information, with missing analysis details and multiplicity plans not reported in the registry record.