AZD9773 Dose Escalation Study: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for AZD9773 Dose Escalation Study (NCT00615017), a randomized phase 2 study. The primary outcome is Change From Baseline in Creatinine Values.

2. Clinical Question

The registry record describes a study of single and repeated intravenous infusions of AZD9773, also called CytoFab, in adults with severe sepsis. Its central clinical question was whether escalating AZD9773 regimens could be assessed for safety and tolerability while also characterizing pharmacokinetics and pharmacodynamics. The primary outcome was change in creatinine from baseline to Day 28; the record also lists mortality, SOFA scores and several measures of drug concentration and disposition.

The question is exploratory in character because the study escalated doses and included different single- and multiple-infusion schedules. The registry record does not report a primary hypothesis comparing one particular AZD9773 regimen with placebo. It therefore supports describing the group summaries, but does not establish a definitive estimate of a common treatment effect across the dose cohorts.

3. Trial Design

The trial is registered as Phase 2, randomized and completed. It enrolled 70 participants and lists severe sepsis as the condition. The study began in 2008-01, with primary completion in 2009-07. The registry identifies the study as multicenter in its brief description. It lists triple masking, while the official title and brief description call the study double-blind. The registry record does not explain the difference in masking labels.

The treatment structure comprised five AZD9773 cohorts and a placebo group. Cohorts 1 and 2 received a single infusion of 50 units/kg and 250 units/kg, respectively. Cohort 3 received a loading infusion of 250 units/kg followed by 9 maintenance doses of 50 units/kg q12hrs. Cohort 4 received a loading infusion of 500 units/kg followed by 9 maintenance doses of 100 units/kg q12hrs. Cohort 5 received a loading infusion of 750 units/kg followed by 9 maintenance doses of 250 units/kg q12hrs. The placebo group received placebo.

There is a structural ambiguity in the registry: allocation is listed as randomized, but the model is listed as single group, despite the multiple dose cohorts and placebo group. These labels do not describe the same simple parallel-group structure, and the registry record does not resolve the discrepancy. The posted outcome summaries are separated by cohort, so they should be read as dose-specific descriptions rather than collapsed into one AZD9773 result.

The registry does not report the randomization procedure, allocation concealment, dose-escalation decision rules, or a statistical analysis plan. Those details should not be inferred from the trial title or from the sequence of dose regimens. The reported enrollment is not the analysis count for every outcome: evaluable numbers differ across outcome summaries.

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: 70 adult patients with severe sepsis, clinical evidence of infection requiring parenteral antibiotics, meeting systemic inflammatory response syndrome criteria, and with cardiovascular/respiratory dysfunction were randomised to receive AZD9773 (2 single dose and 3 multiple dose cohorts) in a double-blind, placebo controlled, dose escalation study

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Participants analyzed relates to the number of evaluable patients at the specified time point.

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 Creatinine Values, measured over End of study (Day 28).

The primary endpoint is the change in creatinine from baseline to follow-up at Day 28. The registry defines the change as the Day 28 mean minus the baseline mean, with baseline measured pre-infusion and follow-up described as 28 days after the start of study drug administration. Creatinine is a laboratory measure used in assessing kidney function; tracking its change can describe whether the group mean moved upward or downward over the stated follow-up interval. That general explanation does not establish that a particular change represents a clinical benefit or harm for an individual.

The registry specifies a mean and a full range for each group. The mean describes the group’s average change, while the full range gives the lowest and highest observed changes in the analyzed group. A range is not a confidence interval: it describes the spread of observed values, not the uncertainty around the estimated mean. It can also be strongly affected by extreme observations, which matters when group sizes are small.

The outcome population is described as the safety analysis set, meaning patients who started study drug infusion. The record further says that the reported number analyzed relates to evaluable patients at the specified time point. Thus the posted group counts for this endpoint need not match enrollment or the counts for other outcomes. The registry record does not report a primary between-group estimate, confidence interval, or hypothesis test for this endpoint.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

Several different statistical ideas are relevant to interpreting these results. The registry reports descriptive summaries for the primary outcome, while the recalculated secondary comparisons use posted counts. Those two forms of evidence answer different questions and should not be conflated.

What does the primary mean describe?

A group mean is the average of the individual changes recorded for participants in that group. A negative mean change indicates that the Day 28 group mean was lower than its baseline mean under the registry’s stated subtraction convention; a positive mean indicates the reverse. The value is not a description of every participant, nor does a mean alone reveal the distribution of individual responses. In this study, the posted group sizes are small, and the full ranges are wide in some cohorts. That combination makes it important to look beyond the mean when assessing how representative it may be of the group.

The reported full range runs from the smallest to the largest observed change. It is a descriptive measure of spread. It is not a measure of how precisely the mean is estimated, and it does not say how likely the true group mean is to fall within those endpoints. A confidence interval serves a different purpose: it gives a range of values compatible with the data under a specified statistical method and assumptions. The primary endpoint results do not include a confidence interval for the mean or for a contrast between groups.

What does a risk difference mean?

A risk difference compares the observed event proportions in two groups by subtracting one proportion from the other. A value above zero means the event proportion is higher in the first group as defined for that comparison; a value below zero means it is lower. For the mortality comparisons, the recalculated estimates compare each named AZD9773 cohort with placebo. A risk difference is expressed on an absolute scale, unlike a risk ratio, which compares event proportions multiplicatively. The registry record does not identify a sponsor-reported analysis for these comparisons.

A confidence interval around a risk difference communicates uncertainty about the estimated contrast. When an interval includes zero, the observed data are compatible with no difference under the interval’s method, as well as with differences in either direction covered by the interval. In the mortality comparisons for cohorts 3 and 4, the reported intervals include zero. Their breadth also means the estimates are not precise. This is not the same as demonstrating equivalence or proving that the groups have identical outcomes.

How should p-values be read?

A p-value describes how unusual data at least as incompatible with a specified null hypothesis would be if that hypothesis and the analysis assumptions were true. It is not the probability that the null hypothesis is true, nor is it a measure of clinical importance. The mortality p-values are unadjusted recalculations. Because the registry reports no sponsor analyses for these comparisons, they should be understood as descriptive exploratory calculations rather than as a substitute for a stated trial analysis plan.

For the TNF-alpha limit-of-quantification outcome, the recalculated comparison for cohort 3 has p < 0.001. The posted count is 5 of 10 in cohort 3 and 0 of 22 in placebo. The small p-value reflects the separation in those observed counts under the calculation used; it does not resolve whether the endpoint was prespecified for confirmatory testing, how missing measurements were handled, or how testing across multiple outcomes and cohorts was addressed. The registry record does not report those details.

What do the planning illustrations mean?

The planning illustrations use an observed effect as a starting point to show how many participants per arm for 80% power or participants per arm for 90% power might be needed under stated assumptions. They use a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation; the effect to detect is a stated fraction of the observed effect. These are planning illustrations, not the trial’s original sample-size justification and not observed power. They are particularly sensitive to an unstable observed effect from small groups. The registry record does not report the original design assumptions or sample-size justification.

Observed power is not a useful substitute for an interval estimate: it is largely a restatement of the observed test result under a chosen effect, and it does not show the range of effects compatible with the data. A planning calculation instead asks what sample size might be needed under specified assumptions for a future study. It does not establish that the assumed effect is real, clinically important, or reproducible.

Why does over-dispersion matter in general?

In general, over-dispersion occurs when outcomes vary more than a statistical model assumes. A model that understates variability can produce intervals that are too narrow and tests that are too optimistic. Clinical outcomes in severe illness can vary for many reasons, so analysts often consider whether the chosen model fits the data. The registry record does not report a model-based primary analysis, a dispersion assessment, or a method for addressing over-dispersion in these results. This general issue therefore cannot be used to revise the posted estimates, but it is relevant when designing or evaluating a later analysis.

10. Multiplicity, Missing Data and Analysis Populations

The registry lists 24 outcome measures and identifies 12 secondary outcomes as posted, but it does not report a statistical analysis plan or a multiplicity strategy. Multiplicity arises when several outcomes, groups, time points, or comparisons are examined: even if each individual test uses a conventional threshold, testing many hypotheses increases the chance that at least one result appears statistically notable by chance. A prespecified hierarchy, adjustment procedure, or clear designation of exploratory analyses can help readers judge the strength of evidence. None is described in the registry record, so the unadjusted secondary p-values should not be treated as confirmatory findings.

The primary endpoint is reported as a mean and full range by group, without a posted between-group test or confidence interval. The record does not specify how the analysis handled missing Day 28 creatinine measurements, whether any imputation was used, or whether sensitivity analyses were conducted. It states that the analyzed count reflects evaluable participants at the specified time point. Because the counts differ across outcomes and cohorts, the available summaries do not establish that every enrolled participant contributed to every analysis.

The safety endpoint population is described as patients who started study drug infusion, while the primary outcome description also refers to evaluable patients at the time point. The record does not provide a detailed analysis-population definition for each secondary outcome. Nor does it report whether exclusions, withdrawals, or missing measurements differed by cohort. Without those details, readers cannot determine how much missingness may have influenced the observed summaries.

The mortality comparisons presented as recalculations are unadjusted and are not listed as sponsor-reported analyses. The record does not describe an adjustment for baseline factors or a plan for comparing multiple dose regimens with placebo. It also does not report interim analyses, stopping rules, or a sample-size justification. These omissions do not prove that such procedures were absent, but they are not reported in the registry record and should not be assumed.

11. 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: sepsis (25.0%), respiratory failure (20.0%), atrial fibrillation (17.4%), peripheral ischaemia (16.7%), pneumonia (13.0%).

The adverse-event summaries report serious and other adverse events by treatment cohort. For serious adverse events, the registry gives both the number at risk and the number affected: 3 of 8 in cohort 1, 6 of 9 in cohort 2, 7 of 12 in cohort 3, 3 of 10 in cohort 4, 4 of 8 in cohort 5, and 13 of 23 in placebo. For other adverse events, the corresponding counts are 8 of 8, 8 of 9, 11 of 12, 7 of 10, 7 of 8, and 22 of 23. These are counts of participants affected, not counts of every event episode.

The listed serious-event terms include sepsis, respiratory failure, atrial fibrillation, peripheral ischaemia, pneumonia, and septic shock, with percentages reported for each term. Such events occur in a severe-sepsis trial population, so their presence does not by itself establish that study drug caused them. The registry record does not provide a causal attribution analysis in these summaries. Deaths in the adverse-event fields are null and therefore are not reported in the registry record for that safety summary. The separate 28-Day Mortality outcome is a different posted outcome and should not be substituted for the missing safety-field death counts.

These descriptive counts are not a formal comparison of safety risk. Cohort sizes are small and unequal, exposure schedules differ, and the record does not report adjusted safety analyses or confidence intervals for the event counts. The figures can describe what was recorded, but they cannot alone establish comparative safety.

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

The primary result means that the registry reports different observed average creatinine changes across the AZD9773 dose cohorts and placebo at Day 28. The reported values are based on the groups’ evaluable participants and are accompanied by full ranges. They describe the observed data; they are not a formal estimate of the effect of assigning AZD9773 rather than placebo. No primary between-group test, adjusted estimate, or confidence interval is reported in the registry record.

The result does not show that AZD9773 caused the observed changes. Randomization is recorded, but small cohort sizes, multiple regimens, and the absence of a reported primary comparative analysis limit causal interpretation of these group summaries. Baseline creatinine summaries and adjustment details are not reported in the registry record. A change score is also not the same as a comparison of follow-up values adjusted for baseline, and the record does not provide such an analysis.

The result does not establish that one regimen is superior, that dose escalation produces a consistent response, or that the observed changes predict a clinical outcome. The group means do not follow a simple ordered pattern, and the full ranges show individual variation. A group average cannot tell whether all participants changed in the same direction. The mortality and biomarker comparisons are secondary and unadjusted; they do not fill the gap in primary comparative evidence or confirm a clinical effect.

Nor does the absence of a clearly established difference prove that the groups are equivalent. The primary outcome lacks inferential intervals, while the available mortality intervals are broad and include zero. In general, uncertainty may remain substantial even when a result is not statistically distinguishable from no difference. Equivalence or non-inferiority requires an appropriate design, prespecified margins, and corresponding analyses, none of which is reported in the registry record.

The careful conclusion is descriptive: the posted results characterize observed changes and selected secondary outcomes in a small dose-escalation study. They do not provide a definitive estimate of AZD9773’s effect on creatinine or a basis for a treatment recommendation.

13. Why This Trial Matters Statistically

This study illustrates why trial structure matters when interpreting a result. Although the registry records randomization, the study includes five dose cohorts, different infusion schedules, and placebo, while its model is labeled single group. That combination makes a single pooled treatment contrast inappropriate without a stated rationale. Dose escalation is intended to examine different exposure regimens, but it also means that each cohort may answer a distinct descriptive question.

The primary outcome is a change score reported with a full range, not a comparative estimate with a confidence interval. This distinction is important in biostatistics: descriptive summaries can show what was observed, while inferential estimates address uncertainty about a contrast or target population. The outcome counts are smaller than overall enrollment and vary by analysis, underscoring that enrollment alone does not define the information available for each endpoint.

The record also illustrates how unadjusted secondary analyses should be handled. Mortality and a biomarker comparison can generate hypotheses, but a large number of outcomes and possible comparisons creates a multiplicity problem when no testing strategy is reported. A small p-value in one secondary comparison is not a replacement for a prespecified primary analysis, especially when analysis populations and missing-data methods are not fully described.

Finally, the planning illustrations emphasize that sample-size calculations depend on assumptions about the effect to detect, allocation, significance level, and method. Calculations anchored to an observed effect are not evidence that the effect is stable, and they are not the trial’s own design justification. Separating observed results, uncertainty, and future planning is central to responsible interpretation of small early-phase studies.

14. 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 registry record leaves several trial-specific uncertainties. It labels the design randomized but also lists a single-group model, and it gives both triple masking and double-blind descriptions. The record does not explain these differences. The dose cohorts use different regimens, including single infusions in some groups and loading plus repeated maintenance infusions in others, which complicates direct interpretation of a simple dose-response pattern.

The primary outcome is presented only as a group mean and full range. No primary treatment contrast, confidence interval, p-value, or adjustment for baseline creatinine is reported. The analysis counts are small and vary by cohort. The full ranges include extreme values, particularly in cohort 5, which may influence the mean; individual data are not reported, so the influence of particular observations cannot be assessed.

Missing-data methods and detailed analysis-population definitions are not reported. The primary outcome description refers to evaluable participants, while the safety population is defined by having started infusion. The record does not explain exclusions or missing follow-up measurements. The secondary analyses are unadjusted recalculations from posted summaries, and no multiplicity plan or sponsor-reported analyses are listed for them.

The registry record also does not report a sample-size justification, interim-analysis plan, stopping rules, or a detailed statistical analysis plan. These details should not be inferred from the study’s dose-escalation title. The listed adverse-event counts are descriptive and do not establish causality or comparative risk. Together, these limitations constrain causal, dose-response, and confirmatory conclusions from the posted results.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: AZD9773 Dose Escalation Study — NCT00615017.
• Registry record last updated 2013-08-22; status last verified 2013-07.
• Results first posted 2013-08-22.
• 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.

18. Record Summary

NCT00615017 is a completed Phase 2 randomized dose-escalation study sponsored by AstraZeneca. It enrolled 70 adults with severe sepsis and compared five AZD9773 regimens with placebo. The study began in 2008-01 and reached primary completion in 2009-07.

The primary outcome was change in creatinine from baseline to Day 28, reported as a mean and full range for each group. The registry also lists mortality, SOFA, pharmacokinetic and pharmacodynamic outcomes, and adverse events. The primary outcome is descriptive; no formal primary between-group analysis is reported. Selected secondary comparisons are unadjusted recalculations from posted counts, with no sponsor-reported analyses listed for those comparisons.

Varying analysis counts, small dose cohorts, unreported missing-data methods, and an unreported multiplicity plan limit interpretation. The findings characterize the posted observations but do not establish a definitive treatment effect.