PACE: Complete Statistical Analysis of Ponatinib in CML and Ph+ ALL

1. Trial at a Glance

PACE was a completed phase 2, non-randomized, single-group treatment trial evaluating ponatinib in chronic myeloid leukemia and Ph+ acute lymphoblastic leukemia. The registry reports 449 participants, 7 arms, and 14 posted outcome measures.

2. Clinical Question

The primary clinical-statistical question in PACE is not a randomized comparison between two treatment groups. Instead, the registered primary outcomes ask what percentage of participants in specified CML and Ph+ ALL populations achieved prespecified hematologic or cytogenetic responses after initiation of ponatinib.

3. Trial Design

The study began on 2010-09-30 and reached primary completion on 2018-12-20. The registry lists 449.0 enrolled participants and 7 arms. Because the design is single-group and non-randomized, the statistical interpretation differs fundamentally from a conventional parallel-group randomized trial.

4. Study Cohorts and Safety Population Information

The ClinicalTrials.gov record reports serious adverse events by arm/cohort. These figures are useful for describing the safety burden represented in the posted results, but they are not efficacy comparisons and should not be converted into treatment-effect estimates.

The table reports the number affected and the corresponding number at risk exactly as reported in the registry. It does not provide enough information to establish whether the cohorts were prospectively powered for comparisons with one another, and the ClinicalTrials.gov record does not report formal statistical comparisons of serious adverse-event rates.

5. Primary Endpoints

The registry identifies 3 primary endpoints, all classified as binary endpoints in the ClinicalTrials.gov record. Their time frames and definitions are important because a binary response endpoint depends not only on the response criterion but also on the assessment window.

MCyR definition

The registry defines MCyR as the percentage of participants with complete cytogenetic response (CCyR) or partial cytogenetic response (PCyR). Cytogenetic response is the percentage of Philadelphia chromosome positive (Ph+) metaphases in bone marrow (BM). MCyR is defined as CCyR or PCyR, where CCyR means no Ph+ cells and PCyR means 1 to 35% Ph+ cells.

MaHR in AP-CML

The registry defines MaHR as the percentage of participants with complete hematologic response (CHR) or no evidence of leukemia (NEL). The registry definition for CHR reports white blood cells (WBC)≤institutional upper limit of normal, absolute neutrophil count (ANC)≥1000/mm³, platelets≥100,000/mm³, no blasts or promyelocytes in peripheral blood, BM blasts ≤5%, <5% myelocytes plus metamyelocytes in peripheral blood.

MaHR in BP-CML/Ph+ ALL

The registry defines MaHR as the percentage of participants with CHR or NEL. The registry definition for CHR reports WBC≤institutional upper limit of normal, ANC≥1000/mm³, platelets ≥100,000/mm³, no blasts or promyelocytes in peripheral blood, BM blasts ≤5%, <5% myelocytes plus metamyelocytes in peripheral blood, basophils in peripheral blood <5%, no extramedullary involvement.

6. Results Status

ClinicalTrials.gov reports that results have been posted and that 14 outcome measures are available. However, the ClinicalTrials.gov record contains no formal statistical analyses for the three registered primary endpoints.

This distinction matters. A posted percentage can describe how frequently an endpoint occurred, but it is not itself a comparative effect estimate. For a non-randomized single-group study, the central inferential question is also different from that of a randomized trial: the observed response proportion describes the treated population, while causal attribution to the intervention requires an appropriate counterfactual comparison.

7. Planned Analysis

Because the ClinicalTrials.gov record contains posted outcome measures but no formal statistical analyses, the appropriate approach is to distinguish what the registry measures from how a binary endpoint of this type would ordinarily be analyzed.

For this trial, the registered endpoint definitions also make the time window part of the estimand. MCyR is assessed up to 12 months after initiation of study treatment, whereas the two MaHR endpoints are assessed up to 6 months. A response percentage without its endpoint definition and assessment window would therefore be incomplete.

8. Statistical Methodology

Binary response analysis

All three registered primary endpoints are classified as binary in the ClinicalTrials.gov record. Binary endpoints reduce each participant's outcome to one of two categories for the specified endpoint definition, such as responder versus non-responder.

Confidence intervals for a proportion

A confidence interval around a response proportion provides a measure of statistical precision. For a binary endpoint, binomial methods are commonly used. The choice between exact and approximate interval procedures should follow the prespecified statistical analysis plan when available.

One-sample hypothesis testing

If a single-group trial has a prespecified historical benchmark, a one-sample binomial test can be used to compare the observed response proportion with that benchmark. Such a comparison requires a justified reference value and a prespecified hypothesis. The registry-reported PACE data do not provide a historical response threshold or a formal hypothesis-testing result.

Why the analysis population matters

The denominator is critical for binary response endpoints. An analysis using all enrolled participants can differ from one using an evaluable response population. The ClinicalTrials.gov record identifies total enrollment as 449.0 but do not provide the primary-endpoint responder counts, denominators, or formal analysis-population definitions for the three endpoints.

Time-windowed response

The endpoint time frame is part of the statistical definition. The CP-CML MCyR endpoint uses a window of up to 12 months after initiation of study treatment, while the AP-CML and BP-CML/Ph+ ALL MaHR endpoints use windows of up to 6 months.

9. Statistical Methods Explained

What makes these endpoints binary?

Each primary endpoint asks whether a participant satisfies a prespecified response definition within a specified time frame. The resulting classification is binary: the participant meets the response criterion or does not meet it.

Why is MCyR different from MaHR?

MCyR is based on cytogenetic findings, specifically the percentage of Ph+ metaphases in bone marrow. MaHR is based on hematologic and disease-status criteria such as WBC, ANC, platelet count, blasts, and extramedullary involvement. A statistical analysis should preserve these endpoint definitions rather than combining them into a generic response measure.

Why does the denominator matter?

A response percentage is conditional on the population used as its denominator. If the number of enrolled participants, treated participants, and evaluable participants differ, the resulting percentages answer different questions. The ClinicalTrials.gov record reports enrollment of 449.0 but do not supply the endpoint-specific denominators.

Why would a confidence interval be useful?

A confidence interval communicates precision around the estimated response proportion. A point estimate alone does not show how much statistical uncertainty accompanies that estimate. For a binary endpoint, binomial confidence-interval methods are natural candidates, with the exact method determined by the prespecified analysis plan.

Why is a p-value not automatically appropriate?

For a single-group binary endpoint, a p-value requires a null hypothesis, typically involving a specified reference response rate. Without a justified benchmark, simply testing whether the observed response differs from zero would not answer a clinically meaningful comparative question. The registry-reported PACE data do not provide such a benchmark or a formal p-value.

Why can this not be interpreted like a randomized trial?

Randomization creates a framework for comparing outcomes between treatment assignments. PACE is identified in the ClinicalTrials.gov record as non-randomized and single-group. Therefore, an observed response proportion does not by itself estimate the causal difference between ponatinib and another treatment.

10. Primary Endpoint Interpretation

11. Safety Results

The ClinicalTrials.gov record reports serious adverse events by arm/cohort as affected participants over the number at risk. These are descriptive safety data. They are not presented as a formal comparative safety analysis because the registry design is non-randomized and single-group and no statistical safety analyses are reported.

The affected/at-risk counts vary substantially across the listed cohorts. That variation is important when reading the table: a count from a cohort with 18 participants at risk is not statistically interchangeable with a count from a cohort with 203 participants at risk.

12. Multiplicity and Multiple Primary Endpoints

The registry identifies 3 primary endpoints. They cover different disease populations and different response constructs, with one CP-CML endpoint assessed up to 12 months and two endpoints assessed up to 6 months.

When several endpoints are designated primary, the statistical plan should specify how success is defined across them and whether multiplicity adjustment is required. The registry-reported PACE data identify the three endpoints but do not provide an alpha-allocation scheme, multiplicity procedure, or formal statistical-analysis results.

13. Interim Analysis, Missing Data, and Other Design Features

The ClinicalTrials.gov record does not report a prespecified interim-analysis method, alpha-spending procedure, multiplicity adjustment, missing-data or imputation method, stratification scheme, non-inferiority margin, crossover design, factorial structure, or Bayesian analysis.

These omissions should not be interpreted as evidence that a method was absent from the full protocol or statistical analysis plan. They mean only that the ClinicalTrials.gov record does not provide that information.

14. Trial Timeline

15. Understanding Response Proportions in a Single-Group Trial

16. Clinical Interpretation vs Statistical Interpretation

17. Important Limitations and Interpretation Issues

• No randomized comparator: the registry-reported design is non-randomized and single-group, so causal treatment comparisons are not available from the design alone.
• No formal statistical analyses reported: no formal statistical analyses were posted despite results being posted.
• Endpoint-specific populations: the three primary endpoints apply to different disease populations, so they should not be treated as three measurements of one common population.
• Different time frames: MCyR is assessed up to 12 months after initiation of study treatment, while both MaHR endpoints are assessed up to 6 months.
• Different biological constructs: MCyR is a cytogenetic endpoint, whereas MaHR is a hematologic response endpoint.
• Denominator uncertainty: total enrollment is 449.0, but the ClinicalTrials.gov record does not provide endpoint-specific analysis denominators or responder counts.
• Multiplicity: three primary endpoints are registered, but the ClinicalTrials.gov record does not describe the formal type I error strategy.
• Safety interpretation: serious adverse-event counts are reported by cohort, but no formal comparative safety analysis is reported.
• Cohort size: the reported safety cohorts have different numbers at risk, which affects the precision of their descriptive estimates.
• Registry scope: absence of a method in the ClinicalTrials.gov record does not prove that it was absent from the full protocol or statistical analysis plan.

18. Why This Trial Matters Statistically

PACE is a useful teaching case because it illustrates a statistical situation that differs sharply from the randomized survival-trial framework often encountered in clinical-trial reporting. The key inferential object is a set of binary response proportions defined within specific disease populations and time windows.

19. Related Tutorials

Learn more about the statistical methods and concepts relevant to this trial:

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: PACE, NCT01207440.
• PubMed: PMID 29567798.
• PubMed: PMID 36054756.
• PubMed: PMID 34991679.
• PubMed: PMID 24180494.
• PubMed: PMID 22825216.

22. Record Summary

PACE is a completed phase 2, non-randomized, single-group treatment trial of ponatinib in chronic myeloid leukemia and Ph+ acute lymphoblastic leukemia. The registry identifies 449.0 enrolled participants, 7 arms, 3 binary primary endpoints, and 14 posted outcome measures. Its primary endpoints are defined around MCyR in CP-CML and MaHR in AP-CML and BP-CML/Ph+ ALL, with assessment windows of up to 12 months for MCyR and up to 6 months for the MaHR endpoints.

The central statistical lesson is that an observed binary response proportion is not automatically a treatment effect. In a non-randomized single-group design, the response proportion describes the study population under the registered endpoint definition. Formal comparative inference requires an appropriate comparator or prespecified reference value. The registry-reported PACE data contain no formal statistical analyses, so the page deliberately distinguishes the posted outcome framework from numerical inferential results that are not present in the source data.