A Phase 2 Trial of the Safety and Efficacy of Bardoxolone Methyl in Patients With Rare Chronic Kidney Diseases - PHOENIX: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Phase 2 Trial of the Safety and Efficacy of Bardoxolone Methyl in Patients With Rare Chronic Kidney Diseases - PHOENIX (NCT03366337), a non-randomized phase 2 study. The primary outcome is Change From Baseline in Estimated Glomerular Filtration Rate (eGFR) at Week 12.

2. Clinical Question

PHOENIX was a Phase 2 study of safety, tolerability, and efficacy in patients with these rare chronic kidney diseases. The central efficacy question was not whether bardoxolone methyl outperformed a concurrent control, because the record does not describe a control group. Rather, the primary outcome was the change in eGFR from baseline during treatment, assessed separately for each disease cohort.

This distinction shapes what the results can establish. A within-cohort change describes the observed course of eGFR among participants receiving the study drug, but without a concurrent randomized comparison it cannot isolate the treatment effect from other explanations for change. The record identifies Week 12 as the primary endpoint time and lists the analysis population as all enrolled patients. It does not report the cohort-specific eGFR estimates, standard errors, or confidence intervals. Consequently, the magnitude and precision of the primary outcome cannot be described from the registry information available.

3. Trial Design

PHOENIX was a completed, multicenter Phase 2 trial sponsored by Biogen. The registry records an enrollment of 103 and gives a start date of 2017-12-26 and a primary completion date of 2019-01-02. It describes a parallel model, nonrandomized allocation, and no masking. Participants were grouped into disease-specific cohorts for autosomal dominant polycystic kidney disease (ADPKD), IgA nephropathy (IgAN), type 1 diabetes-associated CKD (T1D), and focal segmental glomerulosclerosis (FSGS).

Each cohort received bardoxolone methyl capsules. The listed regimen began at 5 mg daily, with titration up to a maximum of 20 mg or 30 mg daily according to UACR category. No placebo or active comparator arm is described. The cohort labels therefore represent different underlying diseases, not alternative study treatments suitable for a direct treatment-versus-control comparison.

The brief summary describes a shared assessment schedule: in-person assessments during treatment at Weeks 1, 2, 4, 6, 8, and 12; telephone contacts on Days 3, 10, 21, 31, 38, and 45; and an in-person follow-up at Week 16, four weeks after treatment ended. The primary outcome was change from baseline in eGFR at Week 12.

There is a design-description inconsistency in the record. Its allocation field says nonrandomized, while the brief summary says participants would be randomized on Day 1. The record does not resolve this discrepancy. The nonrandomized allocation designation also appears alongside an explicit caution that the cohorts are not a like-for-like control comparison. The details of any allocation procedure 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.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Intent-to-treat population (all enrolled patients)

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: All participants who received any amount of study drug

6. Primary Endpoint

The registry lists the primary outcome as Change From Baseline in Estimated Glomerular Filtration Rate (eGFR) at Week 12, measured over 12 weeks after participant receives the first dose.

The primary endpoint was change from baseline in estimated glomerular filtration rate (eGFR) at Week 12, measured in mL/min/1.73 m^2. eGFR is an estimate of kidney filtration based on blood/serum assessment. The registry description states that higher eGFR values represent better kidney function and lower values represent poorer kidney function. A change from baseline expresses how the measured value at the specified follow-up differs from the participant’s starting value.

Change in eGFR is useful as a continuous outcome because it retains information about the size and direction of measured change, rather than reducing the result to a simple event count. Continuous outcomes are commonly summarized using a mean or model-based mean, together with a measure of uncertainty. The registry identifies the reported parameter as a least-squares mean and the dispersion measure as a standard error. These summaries would describe the cohort’s estimated average change and the precision of that estimate, if numerical results were available.

The Week 12 timing places the endpoint during the treatment period described in the record. However, the numerical cohort estimates, standard errors, and confidence intervals are not reported in the registry record. The endpoint definition alone therefore does not establish the size, direction, or precision of any observed cohort change, nor does a change from baseline by itself establish a causal treatment effect.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry identifies the primary outcome as a continuous measure and names the analysis parameter as a least-squares mean, with standard error as the dispersion measure. A continuous outcome can take a range of numerical values; a mean summarizes its central tendency. A least-squares mean is generally a model-based estimate of an average, often adjusted for terms included in a statistical model. Which terms were included, how the model was specified, and what assumptions were checked are not reported in the registry record.

What does a least-squares mean represent?

In general, a least-squares mean is an estimated mean derived from a statistical model, rather than necessarily the simple arithmetic average of observed values. It can be useful when an analysis accounts for specified features of the data. Its interpretation depends on the model and the population to which the estimate applies. Since the model details and cohort-specific estimates are not reported in the registry record, the meaning of any adjustment cannot be assessed for PHOENIX. The registry’s label alone does not indicate which baseline factors, if any, were included.

How should the standard error be read?

A standard error describes the sampling uncertainty of an estimated quantity, such as a mean. It is not the same as the standard deviation, which describes variability among individual observations. A smaller standard error generally indicates a more precise estimate, while a larger standard error indicates less precision. The standard error can be used as an input to construct a confidence interval under appropriate assumptions, but the record does not report numerical standard errors or confidence intervals for the primary endpoint. Without those values, precision cannot be quantified.

A confidence interval, in general, gives a range of values compatible with the data under the statistical model and its assumptions. Its width conveys information about precision. A confidence interval is not a guarantee that the true value lies in the interval, and it does not remove bias from a nonrandomized design. Even a precise estimate of within-cohort change would not by itself establish how much of that change was caused by treatment when no comparable control group is available.

Why does the allocation structure matter?

Randomization is a method for assigning participants to groups so that, on average, the groups are comparable in measured and unmeasured characteristics. PHOENIX is designated nonrandomized in the registry, and the disease cohorts differ in their underlying conditions. Thus the cohorts are not interchangeable treatment groups, and contrasts between their outcomes would not estimate a treatment effect. The record’s brief summary also refers to randomization on Day 1, which conflicts with the nonrandomized allocation designation. The allocation procedure is not reported in a way that resolves this inconsistency.

Because the study is open-label, masking is listed as none. In general, awareness of treatment can influence participant behavior, care, reporting, or outcome assessment, depending on the outcome and study procedures. eGFR is a laboratory-derived measure, but the record does not provide details about laboratory procedures, blinded assessment, or how potential measurement variation was handled. Those details are not reported in the registry record.

What can be inferred from planning calculations?

The record describes a planning illustration based on an observed effect, using a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. It characterizes the effect to detect as a stated fraction of the observed effect. The illustration is not the trial’s own design justification, and planning sample sizes are not observed power. The numerical observed effect and planning values are not reported in the information available for this analysis. In general, such calculations describe hypothetical future-study requirements under stated assumptions; they do not show that the completed study had a particular probability of finding an effect.

Sample-size calculations depend on the outcome variability, target effect, significance level, desired power, allocation, and analysis assumptions. If those inputs are uncertain or do not match the eventual study, the resulting planning estimates may not translate directly to a later trial. The registry does not report the original sample-size justification or its assumptions, so the planning illustration should not be mistaken for the trial’s prespecified design rationale.

Why is the outcome’s distribution relevant?

For a continuous endpoint, an analysis may rely on assumptions about the distribution of measurements or model residuals. General statistical practice includes checking whether these assumptions are plausible and considering methods robust to departures where needed. The registry record does not state which diagnostics or alternative methods were used. Over-dispersion is chiefly a concern for count data when observed variability exceeds the model’s expectation; it is not directly the same issue as variability in a continuous eGFR outcome. No model details are available to assess the handling of distributional concerns in this trial.

Finally, the statistical interpretation depends on who was included in the analysis and how missing measurements were handled. The primary population is identified as intent-to-treat, defined in the record as all enrolled patients. The methods for accounting for unavailable Week 12 measurements, if any, are not reported in the registry record. Consequently, neither the extent nor the potential influence of missing outcome data can be assessed.

10. Confidence Intervals for Each Arm

No two-arm comparison is shown for this trial, but each arm's mean can be given with a t-based 95% confidence interval computed from the posted mean, spread and number of participants.

R code and output

arm <- c("Bardoxolone Methyl - ADPKD", "Bardoxolone Methyl - IgAN", "Bardoxolone Methyl - T1D", "Bardoxolone Methyl - FSGS")
n   <- c(31, 26, 28, 18)
m   <- c(9.31, 8, 5.46, 7.83)
sp  <- c(1.3743, 1.57, 2.2792, 2.216)
sd  <- sp * sqrt(n)      # standard errors converted to SDs
for (i in seq_along(n)) {
  se <- sd[i] / sqrt(n[i]); h <- qt(0.975, n[i] - 1) * se
  cat(sprintf("CARM|%s|%d|%.3f|%.3f|%.3f|%.3f\n", arm[i], n[i], m[i], sd[i], m[i] - h, m[i] + h))
}

CARM  Bardoxolone Methyl - ADPKD  31  9.310  7.652  6.503  12.117
CARM  Bardoxolone Methyl - IgAN  26  8.000  8.005  4.767  11.233
CARM  Bardoxolone Methyl - T1D  28  5.460  12.060  0.783  10.137
CARM  Bardoxolone Methyl - FSGS  18  7.830  9.402  3.155  12.505

11. Multiplicity, Missing Data and Analysis Populations

The registry lists one outcome measure and four analyses, corresponding to the four disease-specific cohorts. Reporting separate cohort analyses can be clinically and statistically useful because the conditions differ, but it also means that the results should be interpreted as a set rather than as a single isolated test. When multiple hypotheses are examined, the chance of at least one apparently unusual result can increase. General methods for addressing multiplicity include prespecifying a primary comparison, adjusting significance thresholds, or presenting results as exploratory. Whether any such approach was used is not reported in the registry record.

The record does not report a statistical analysis plan, a multiplicity adjustment, or an inferential hierarchy across the four cohorts. It also does not report numerical efficacy estimates or p-values in the information, so the role of statistical significance cannot be assessed. The cohort-specific analyses should not be treated as independent confirmatory tests without information about their prespecification and error-control strategy.

The primary population is described as intent-to-treat, meaning all enrolled patients according to the registry’s definition. The record does not explain whether every enrolled participant contributed a Week 12 eGFR measurement, how participants with missing measurements were handled, or whether any alternative analysis populations were used. It also does not state the amount or pattern of missing outcome data. Missingness can affect interpretation if participants without follow-up measurements differ from those with observed measurements; the possible direction or size of that effect cannot be established without further information.

Interim analyses, stopping rules, and procedures for handling protocol deviations are not reported in the registry record. Nor are details available on covariate adjustment, model diagnostics, or sensitivity analyses. These omissions limit assessment of how robust the primary findings would be to alternative analytic choices. The absence of those details should be distinguished from evidence that no such procedures occurred; the registry record simply does not report 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 2% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: cholecystitis (5.6%), histoplasmosis disseminated (3.2%), sepsis (3.2%), renal failure acute (3.2%), hypertensive crisis (3.2%).

The safety record presents serious adverse events separately for each disease cohort. In ADPKD, 2 of 31 participants at risk were recorded as affected by serious adverse events. In IgAN, 0 of 26 were affected; in T1D, 0 of 28 were affected; and in FSGS, 1 of 18 was affected. These are cohort-specific counts. They should not be added together to create a combined estimate, and they do not provide a comparison with an untreated or control group.

The registry also lists other adverse events: 30 of 31 participants in ADPKD, 22 of 26 in IgAN, 25 of 28 in T1D, and 16 of 18 in FSGS. Deaths are recorded as 0 in each cohort. Serious events and other events are distinct registry categories, so the figures should not be treated as interchangeable. The record lists cholecystitis at 5.6% and disseminated histoplasmosis, sepsis, acute renal failure, and hypertensive crisis at 3.2% each among the top serious adverse event terms.

These counts describe events recorded during the study; they do not establish that the study drug caused an event or that the event rate would differ from a comparison group. Cohort sizes are limited, and there is no like-for-like control comparison described. The registry record does not report further details needed to assess event timing, duration, relatedness, or severity beyond the serious-event classification.

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

The registry establishes that PHOENIX was a Phase 2 study of bardoxolone methyl in four disease-specific cohorts, with change from baseline in eGFR at Week 12 identified as the primary outcome. It also provides the outcome’s unit and identifies a least-squares mean and standard error as the intended summary measures. However, cohort-specific numerical efficacy estimates and uncertainty intervals are not reported in the registry record. Readers therefore cannot determine from this record how much eGFR changed in each cohort or how precisely the change was estimated.

Even if within-cohort changes were available, they would not by themselves demonstrate a causal treatment effect. The record identifies nonrandomized allocation, no masking, and no like-for-like control comparison. Change from baseline can reflect treatment as well as the underlying disease course, measurement variation, and other factors. The design description’s reference to randomization in the brief summary conflicts with the nonrandomized allocation designation; the registry does not resolve that inconsistency.

The serious adverse event figures provide descriptive safety information for the participants in each cohort. They do not show how those counts compare with a control group, and they do not establish causation. Zero recorded deaths in each cohort is a description of the posted results, not proof that there was no risk of death or that the treatment is safe in other settings.

Accordingly, the record supports an account of the study’s objective, endpoint, cohort structure, and recorded adverse event counts. It does not support a conclusion about the size of the primary efficacy result, the precision of that result, or a causal benefit-risk comparison. Those questions require numerical outcome reporting and an appropriate comparative design, along with fuller information about the statistical analyses and missing data.

14. Why This Trial Matters Statistically

PHOENIX raises several important statistical issues that often arise when studying rare chronic diseases. Separate cohorts allow a study to address conditions with different clinical characteristics, but they also mean that evidence is divided across disease groups rather than centered on one shared treatment comparison. The registry lists four cohort analyses and one primary outcome measure. This structure makes clear why results should be interpreted within their intended disease cohorts and why a comparison between the cohorts would not be equivalent to a randomized treatment contrast.

The endpoint is continuous, which can preserve more information about the degree of change than a binary outcome. The registry specifies a model-based mean and standard error, indicating an intended focus on average change and its precision. Yet numerical estimates are not reported in the registry record, so the outcome’s magnitude and precision cannot be evaluated. This illustrates the difference between a well-defined endpoint and a fully interpretable result: defining what was measured does not convey what the data showed.

The nonrandomized allocation designation is also consequential. A within-group before-and-after analysis can describe change over time, but causal attribution is difficult without a concurrent comparison group. The record’s conflicting mention of randomization makes transparent reporting of allocation methods especially important for readers assessing possible bias.

Finally, safety counts are most informative when their population, denominator, event definition, and comparative context are clear. The registry provides cohort-level counts and selected serious-event terms, but no like-for-like control comparison is described. PHOENIX therefore illustrates the value of separating descriptive evidence from causal evidence, and of reporting estimates with uncertainty alongside methods for missing data, multiplicity, and allocation.

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 most important trial-specific limitation is the absence of cohort-specific numerical primary outcome results in the registry record. Although change from baseline in eGFR at Week 12 is defined, no estimates, standard errors, or confidence intervals are reported in the information available. This prevents assessment of the observed direction, magnitude, or precision of the primary endpoint.

The record identifies the study as nonrandomized and does not describe a concurrent control group. Accordingly, a change over time cannot be interpreted as a controlled treatment effect. The brief summary’s statement that participants would be randomized on Day 1 conflicts with the nonrandomized allocation field, and the record does not clarify how assignment actually occurred. The study is also listed as open-label with no masking, so the record does not support an assessment of whether awareness could have influenced study conduct or outcome ascertainment.

The primary endpoint’s model details are not reported. The record does not specify covariates, diagnostic checks, sensitivity analyses, or whether the least-squares mean was adjusted for particular factors. It also does not describe how missing Week 12 measurements or protocol deviations were handled, despite identifying the intent-to-treat population as all enrolled patients. The extent and pattern of missing data are not reported in the registry record.

Four disease-specific analyses raise questions about multiplicity and the status of each analysis, but a testing strategy or adjustment is not reported. The serious adverse event counts are descriptive and lack a like-for-like control comparison; additional event context such as timing and relatedness is not reported in the registry record. These limitations constrain conclusions about both efficacy and comparative safety.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: A Phase 2 Trial of the Safety and Efficacy of Bardoxolone Methyl in Patients With Rare Chronic Kidney Diseases - PHOENIX — NCT03366337.
• Registry record last updated 2025-06-03; status last verified 2025-05.
• Results first posted 2022-01-18.
• 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

PHOENIX was a completed, multicenter, open-label Phase 2 trial sponsored by Biogen. The registry records 103 enrolled participants and a parallel, nonrandomized design. It studied bardoxolone methyl in separate cohorts with ADPKD, IgAN, T1D-associated CKD, and FSGS. The listed primary outcome was change from baseline in eGFR at Week 12, summarized as a least-squares mean with standard error.

The registry reports cohort-specific serious adverse event and other adverse event counts, as well as zero recorded deaths in each cohort. It does not report numerical cohort-specific eGFR changes or their uncertainty in the information available. The record also contains inconsistent descriptions of allocation: nonrandomized in the design field and randomized in the brief summary. Missing-data procedures, model details, and multiplicity methods are not reported in the registry record. The absence of a like-for-like control comparison limits causal interpretation of changes and safety counts.