Investigating Safety, Tolerability and Efficacy of AZD5363 When Combined With Paclitaxel in Breast Cancer Patients: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Investigating Safety, Tolerability and Efficacy of AZD5363 When Combined With Paclitaxel in Breast Cancer Patients (NCT01625286), a randomized phase 1, phase 2 study. The primary outcome is Dose-limiting Toxicity (DLT) Events - Part A.

2. Clinical Question

The distinction between the two parts is central to reading the results. The primary endpoint is a short-term safety measure in Part A, not a comparative efficacy endpoint in the randomized Part B population. The registry record explicitly states that dose-limiting toxicity was not assessed for Part B participants. Consequently, the primary endpoint cannot answer whether the randomized Part B regimens differ in tumour response or other later outcomes.

The Part B comparison focuses on the addition of AZD5363 to paclitaxel, with placebo plus paclitaxel as the control. The stated population is ER-positive advanced or metastatic disease, and the official title notes PIK3CA mutation-status stratification. The registry record does not report results separately by mutation status, so the posted aggregate outcomes do not establish what the pattern was within that subgroup.

3. Trial Design

This is a Phase I/II study with a safety run-in and a placebo-controlled randomized expansion. The registry lists randomized allocation, quadruple masking, and a parallel-group model. The study is marked completed, with enrollment recorded as 148. Its stated purpose included investigating safety and efficacy at different AZD5363 doses and schedules with weekly paclitaxel, followed by evaluation of a selected combination against paclitaxel plus placebo in ER-positive advanced or metastatic breast cancer.

Part A included several AZD5363 dose and schedule groups. Its primary endpoint counted dose-limiting toxicity events during the Cycle 1 evaluation period, up to 28 days. Part B compared AZD5363 plus paclitaxel with placebo plus paclitaxel. This two-part structure means the Part A groups and the Part B randomized arms answer different questions: the former concern early safety across dose schedules, while the latter provide the placebo-controlled comparison.

The Part B outcomes include tumour-size change, objective response measures, duration and durability of response, and overall survival. The registry record describes RECIST tumour assessments every 12 weeks for several response outcomes. It does not report a detailed statistical analysis plan for all outcomes, the randomization procedure, the masking roles, or the intended analysis population for the Part B comparisons.

The registry lists a start date of 2012-10-03 and a primary completion date of 2017-01-28. These dates describe the recorded study timeline; they do not establish how long each participant was followed. The record includes different outcome-specific timeframes, but the follow-up and analysis details needed to interpret every outcome are not reported in the registry record.

The safety run-in and randomized expansion should not be treated as one homogeneous comparison. Part A dose groups were not the randomized placebo-controlled contrast described for Part B, and small group sizes make their outcome summaries particularly imprecise. Part B is the relevant setting for interpreting AZD5363 versus placebo alongside paclitaxel, while the primary endpoint belongs to Part A.

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.

Only the first 6 of 7 groups are shown.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: All Part A patients who either completed the DLT evaluation period with at least 80% of specified dose (of AZD5363 or paxlitaxel) or who experienced a DLT. DLT events were not assessed for Part B participants.

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 patients with data who were included in the full analysis set for Part A and the intention to treat analysis set in Part B'

Only the first 6 of 8 groups are shown.

6. Primary Endpoint

The registry lists the primary outcome as Dose-limiting Toxicity (DLT) Events - Part A, measured over During Part A DLT evaluation period (Cycle 1, up to 28 days).

The primary endpoint is the number of Part A participants with a dose-limiting toxicity during the Cycle 1 DLT evaluation period, which lasted up to 28 days. A qualifying event was an adverse event or laboratory abnormality considered related to study drug, beginning during the evaluation period and judged dose limiting. The endpoint was expressed as a participant count, not as a continuous measurement.

The registry defines the assessed population as Part A patients who either completed the evaluation period while receiving at least 80% of the specified AZD5363 or paclitaxel dose, or experienced a dose-limiting toxicity. Participants in Part B were not assessed for this endpoint. This eligibility definition matters because the count is not necessarily a simple tally among everyone enrolled in Part A; the record describes a specific evaluation population.

In an early-phase study, dose-limiting toxicity is used to characterize short-term tolerability of dose and schedule combinations and to inform selection of a regimen for further study. It is not a comprehensive measure of all adverse events over the full treatment period, nor does it by itself establish comparative safety between the Part B randomized arms. The registry record does not report a formal statistical comparison for the primary endpoint.

The posted Part A counts are 0 of 11 for Schedule 1 560 mg bd, 2 of 6 for Schedule 1 640 mg bd, 0 of 5 for Schedule 2 360 mg bd, 0 of 7 for Schedule 2 400 mg bd, and 2 of 6 for Schedule 2 480 mg bd. These are descriptive counts from the stated endpoint population. With no reported inferential analysis, they should be read as observations within the dose-finding component rather than as a ranked or definitive comparison of schedules.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

Statistical methods answer different questions depending on the outcome type and study component. The primary endpoint is a count of participants with a dose-limiting toxicity, whereas tumour-size change is continuous and response is binary. A count, a mean difference, and a risk difference have different interpretations and should not be compared as if they were the same measure.

How should the primary count be read?

The primary endpoint reports participants with an event in each Part A group. The group count and the number assessed give context for the observed event count, but small groups make any single participant's outcome influential. A zero count means no qualifying event was posted for that group under the endpoint definition; it does not mean that no other adverse events occurred, or that later risks were absent. The endpoint covers the Cycle 1 evaluation period and is restricted to the defined Part A population.

The record does not report a formal inferential comparison among the Part A dose and schedule groups. Without such a reported analysis, the counts should not be treated as evidence that one regimen is safer than another. Differences between group counts can reflect the small group sizes as well as underlying event patterns. No statistical model, interval estimate, or adjustment method for the primary endpoint is reported in the registry record.

What does a mean difference describe?

Change in tumour size is reported as a mean percentage change from baseline. A negative mean indicates that the measured tumour size was lower than baseline on average; a more negative value represents a larger average decrease. The Part B summary compares the mean in the AZD5363 arm with the mean in the placebo arm. A mean difference is the arithmetic contrast between those group means, expressed in the outcome's units. It describes an average, not the experience of every participant, and it can conceal variation in individual outcomes.

The registry posts standard deviations, which describe spread around each group mean. Greater spread means individual measurements vary more widely. Standard deviations are not confidence intervals and do not directly quantify the precision of the estimated difference. The simple unadjusted recalculation uses the posted group summaries. It does not account for baseline covariates or other possible features of the trial's analysis model.

The registry separately reports an ANCOVA analysis for Part B tumour-size change, labelled as a mean difference in final values. ANCOVA is a general method that compares groups while accounting for one or more covariates, often including baseline outcome values. The registry gives an 80% confidence interval for this reported analysis, but the exact covariates and model details are not reported in the registry record. The reported ANCOVA result and the unadjusted recalculation are therefore not interchangeable: they may differ because they use different estimands, adjustments, or analytic details.

How should binary outcomes be interpreted?

Objective response is reported as a percentage of participants, and a risk difference expresses the difference between the two groups' observed response proportions. A positive difference in the Part B comparison means the observed proportion was higher in the AZD5363 arm; a negative difference means it was lower. The point estimate is uncertain, so it should be considered with its interval. A confidence interval describes a range of estimates compatible with the data and method under repeated sampling assumptions. A wide interval indicates limited precision; an interval that includes no difference does not establish that the treatments are equivalent.

The unadjusted p-value measures how inconsistent the data are with a specified null hypothesis under the calculation's assumptions. It is not the probability that the null hypothesis is true, nor a measure of clinical importance. Because the registry does not provide a full analysis plan for these comparisons, the posted unadjusted p-values should not be treated as definitive confirmatory tests. They do not account for the fact that multiple outcomes and comparisons are reported.

What do the planning illustrations mean?

The sample-size planning illustrations are based on observed effects and assume a two-sided significance level of 0.05, equal allocation, and a normal approximation. The displayed scenarios consider planning for 100%, 75%, and 50% of the observed effect. The resulting values are participants per arm for 80% power or participants per arm for 90% power, for comparisons where such illustrations are available.

These are planning illustrations, not observed power and not the original trial's design justification. A calculation based on an observed effect is especially uncertain because the effect itself is estimated from limited data; it can be optimistic or unstable. Smaller target effects require larger planned samples under the stated framework. This general relationship does not establish what sample size the investigators originally intended. The original sample-size justification and its assumptions are not reported in the registry record.

The record also notes that some binary comparisons have missing or non-estimable values for a two-arm comparison. Those cases do not support a meaningful comparative recalculation. A blank or unavailable comparison should not be interpreted as a zero effect or as evidence that the groups were alike.

Why can over-dispersion matter?

For binary outcomes, a simple model may assume that responses follow a common probability within each group. In practice, participants can differ in ways that make outcomes more variable than that model expects. This is called over-dispersion. When present, a model that ignores it can understate uncertainty and make intervals too narrow. This is a general statistical concern; the registry record does not report an over-dispersion assessment or a model for the posted binary outcomes.

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("Part A Schedule 1 560 mg bd", "Part A Schedule 1 640 mg bd", "Part A Schedule 2 360 mg bd", "Part A Schedule 2 400 mg bd", "Part A Schedule 2 480 mg bd")
n      <- c(11, 6, 5, 7, 6)
events <- c(0, 2, 0, 0, 2)
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  Part A Schedule 1 560 mg bd  11  0  0.0  0.0  28.5  -0.0  25.9  0.2  26.5
ARM  Part A Schedule 1 640 mg bd  6  2  33.3  4.3  77.7  9.7  70.0  9.9  71.0
ARM  Part A Schedule 2 360 mg bd  5  0  0.0  0.0  52.2  0.0  43.4  0.4  45.9
ARM  Part A Schedule 2 400 mg bd  7  0  0.0  0.0  41.0  0.0  35.4  0.3  36.9
ARM  Part A Schedule 2 480 mg bd  6  2  33.3  4.3  77.7  9.7  70.0  9.9  71.0

11. Multiplicity, Missing Data and Analysis Populations

Multiplicity arises when a study evaluates several outcomes, comparisons, timepoints, or subgroups. Even when each analysis is considered separately, looking across many analyses increases the chance of observing a result that appears noteworthy by chance. Common approaches include adjusting significance thresholds or defining an ordered testing strategy, but the registry record does not report a multiplicity adjustment or a testing hierarchy for the secondary endpoints.

The reported unadjusted p-values should consequently be read as nominal results. They do not show that the study-wide chance of a false-positive conclusion was controlled. This matters because the record lists several secondary outcomes, and some have multiple assessment contexts. The official title also describes stratification by PIK3CA mutation status, but subgroup results and any subgroup analysis plan are not reported in the registry record.

Missing data and analysis populations are also important. The primary endpoint population is defined by completion of the DLT period at the specified dose level or experience of a DLT. For secondary outcomes, the posted denominators vary by outcome and arm. The registry record does not report a complete account of why outcome data are unavailable, how missing values were handled, or whether any imputation was used. It also does not report the full Part B analysis population definition, such as whether all randomized participants were included in each analysis.

Different denominators across outcomes can arise for several reasons, including outcome eligibility or availability of an assessment, but the reason for each difference is not reported in the registry record. It would therefore be inappropriate to assume that missing observations were random, that all randomized participants contributed to every result, or that the posted percentages use the same population throughout.

These gaps limit how firmly the secondary results can be interpreted. A complete assessment would need the prespecified analysis plan, outcome-specific analysis populations, missing-data rules, and multiplicity strategy. In their absence, the registry summaries support descriptive reading and cautious comparison, not a reconstruction of the investigators' full inferential framework.

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 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: pyrexia (16.7%), drug hypersensitivity (16.7%), lower respiratory tract infection bacterial (16.7%), pneumonia (16.7%), urinary tract infection (16.7%).

The safety record separates serious adverse events, other adverse events, and deaths by study group. For serious adverse events, the posted figures give the number at risk and the number affected in each group. In Part A, these figures include 3 of 12 for Schedule 1 560 mg bd, 3 of 8 for Schedule 1 640 mg bd, 0 of 5 for Schedule 2 360 mg bd, 1 of 7 for Schedule 2 400 mg bd, and 4 of 6 for Schedule 2 480 mg bd. In Part B, the figures are 13 of 54 for AZD5363 and 8 of 55 for placebo.

These counts are descriptive. They do not establish that the treatment caused an event, and the registry summary does not provide a comparative statistical analysis of serious adverse events. Exposure duration, event timing, recurrent events, and the detailed event-level data needed to make more specific comparisons are not reported in the registry record. The death counts are listed separately: the Part A groups report 0, 0, 1, 1, and 0, while Part B reports 13 for AZD5363 and 15 for placebo. These counts alone do not establish cause or comparative risk.

The registry lists serious-event terms including pyrexia, drug hypersensitivity, bacterial lower respiratory tract infection, pneumonia, urinary tract infection, and renal failure, each with a posted value of 16.7%. These term-level percentages should be read as registry summaries, not as a complete description of all serious events or as proof that any event was treatment-related.

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

The posted findings describe what was recorded in this study, but they do not provide a complete answer to whether adding AZD5363 changes outcomes in the intended population. The primary endpoint is a short-term Part A safety count. It is useful for describing dose-limiting events during the first evaluation period, but it is not a randomized comparison of safety or efficacy in Part B. The Part B outcomes provide the relevant placebo-controlled summaries, yet the available comparisons are limited by the number of participants, the uncertainty around estimates, and incomplete reporting of analysis details.

A difference between observed group values is not automatically evidence of a reliable or clinically important effect. Estimates vary from sample to sample, and confidence intervals express that uncertainty under their calculation assumptions. An interval spanning no difference means the data do not precisely distinguish a difference from no difference by that method; it does not prove the treatments are equal. Conversely, a point estimate favoring one arm does not establish that the difference is real or meaningful.

The reported ANCOVA result for tumour-size change is an adjusted analysis with an 80% confidence interval, while the accompanying recalculated comparison uses posted summary statistics and is unadjusted. These results answer related but not identical statistical questions. The registry does not report the full ANCOVA model or a complete prespecified analysis plan, so readers should not treat the two estimates as if they were competing versions of one fully documented analysis.

Secondary outcomes can add context, but multiple unadjusted analyses are exploratory when the multiplicity plan is not reported. The lack of posted results for some listed outcomes cannot be interpreted as evidence of no effect. Likewise, the registry's aggregate outcomes do not establish what happened in the PIK3CA mutation subgroup, despite the official title's reference to stratification.

Nothing in these summaries supports individual treatment decisions. A complete interpretation would require full trial reporting, including the prespecified analysis plan, participant flow, outcome-specific missingness, and detailed safety and efficacy analyses. The results should be understood as registry-level summaries with the limitations described, rather than as a complete clinical evidence assessment.

14. Why This Trial Matters Statistically

This study is statistically instructive because its two-part structure links early dose and schedule evaluation with a later randomized comparison, while assigning different questions to each part. The primary endpoint is a Part A safety count, whereas the placebo-controlled Part B comparison focuses on tumour outcomes. Keeping these components separate prevents a common interpretive error: treating a safety run-in endpoint as though it were the primary efficacy comparison in the randomized expansion.

The record also illustrates why outcome type matters. Dose-limiting toxicity is summarized as a count, tumour-size change as a mean with a standard deviation, and response outcomes as percentages. Each measure conveys different information. A mean difference describes an average change, while a risk difference compares observed proportions. Neither automatically describes how consistently individual participants responded.

The juxtaposition of an unadjusted recalculation and a reported ANCOVA analysis highlights the role of analysis choices. Adjustment for covariates can change an estimate and its precision, and confidence intervals at different levels should not be compared as though they represent the same uncertainty standard. The registry gives the ANCOVA result but not its full model specification, making clear reporting of analysis methods important for independent interpretation.

Finally, the range of secondary endpoints shows why multiplicity and missingness matter. Several related outcomes may provide useful context, but unadjusted tests across many outcomes do not automatically provide confirmatory evidence. Different denominators and absent results also complicate interpretation. The record's limitations demonstrate that a numerical result is only as interpretable as its endpoint definition, analysis population, missing-data handling, and prespecified statistical plan.

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.

Several limitations are specific to the way this trial is represented in the registry. The primary endpoint is restricted to Part A and to the DLT evaluation period in Cycle 1, up to 28 days. It does not capture the full duration of safety follow-up, and the record says DLT events were not assessed for Part B participants. The Part A groups are small, so their descriptive event counts provide limited precision and should not be used as a stable ranking of schedules.

The dose-finding groups and randomized Part B arms have different purposes and are not interchangeable. Part A includes multiple schedules and doses, while Part B compares AZD5363 with placebo alongside paclitaxel. Comparing outcomes across these components would mix non-equivalent populations and study questions. The registry does not report detailed criteria for selecting the Part B dose and schedule.

Outcome denominators differ, and the registry does not explain all reasons for missing or unavailable measurements. Detailed participant flow, outcome-specific missing-data handling, and the complete Part B analysis population are not reported in the registry record. The official title notes PIK3CA mutation-status stratification, but no subgroup result is included in the registry entries.

For the Part B tumour-size outcome, the record gives both an ANCOVA analysis and summary statistics from which an unadjusted comparison can be recalculated. The model specification and covariates for ANCOVA are not reported, and the stated confidence interval level differs from the interval accompanying the unadjusted recalculation. The statistical analysis plan, multiplicity strategy, and original sample-size justification are also not reported in the registry record.

Finally, several listed secondary endpoints lack numeric results in the registry entries. The absence of those results limits any assessment of response duration or survival patterns. These gaps mean the registry summaries should be interpreted as partial reporting rather than as a complete account of all planned analyses.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Investigating Safety, Tolerability and Efficacy of AZD5363 When Combined With Paclitaxel in Breast Cancer Patients — NCT01625286.
• Registry record last updated 2023-01-18; status last verified 2022-12.
• Results first posted 2019-04-01.
• 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

NCT01625286 is a completed Phase I/II study of AZD5363 with weekly paclitaxel in advanced or metastatic breast cancer. Part A evaluated multiple AZD5363 doses and schedules and used dose-limiting toxicity during the Cycle 1 evaluation period as its primary endpoint. Part B was a randomized placebo-controlled expansion in ER-positive advanced or metastatic breast cancer, comparing AZD5363 plus paclitaxel with placebo plus paclitaxel. The official title describes stratification by PIK3CA mutation status.

The registry records enrollment of 148, randomized allocation, quadruple masking, and a parallel-group model. The primary endpoint counts varied across Part A groups, while Part B results include tumour-size and response summaries. The record also lists serious adverse events, deaths, and additional secondary outcomes. It does not report a complete statistical analysis plan, multiplicity strategy, missing-data methods, or results for every listed secondary endpoint. The posted summaries therefore support cautious, outcome-specific interpretation rather than a single definitive conclusion.