This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Trial-specific numerical results on this page are restricted to the ClinicalTrials.gov record.
1. Trial at a Glance
SHAPE was a randomized, parallel-group, unmasked treatment trial comparing radical hysterectomy plus pelvic lymph node dissection with simple hysterectomy plus pelvic lymph node dissection in cervical cancer. The registered primary endpoint was the pelvic recurrence rate at 3 years, analyzed under a non-inferiority hypothesis.
| Feature | SHAPE |
|---|---|
| Trial name | SHAPE |
| NCT identifier | NCT01658930 |
| Condition | Cervical Cancer |
| Design | Randomized, parallel |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 700 |
| Lead sponsor | Canadian Cancer Trials Group |
| Sponsor type | Network |
| Status | Completed |
| Start | 2012-12-10 |
| Primary completion | 2023-03-11 |
| Results posted | Yes |
| Outcome measures posted | 5 |
| Statistical analyses posted | 5 |
2. Clinical Question
The central statistical question was whether simple hysterectomy plus pelvic lymph node dissection could be considered non-inferior to radical hysterectomy plus pelvic lymph node dissection with respect to the pelvic recurrence rate at 3 years.
Population
Patients enrolled in SHAPE for the condition recorded as cervical cancer. The ClinicalTrials.gov record identifies the brief trial title as “Radical Versus Simple Hysterectomy and Pelvic Node Dissection With Low-risk Early Stage Cervical Cancer.”
Intervention
Simple hysterectomy plus pelvic lymph node dissection.
Comparator
Radical hysterectomy plus pelvic lymph node dissection.
Primary question
Is the difference in 3-year pelvic recurrence rate sufficiently small that simple hysterectomy is not inferior to radical hysterectomy under the prespecified 4% margin?
3. Trial Design
Radical Hysterectomy
- Radical hysterectomy
- Pelvic lymph node dissection
Simple Hysterectomy
- Simple hysterectomy
- Pelvic lymph node dissection
The registry identifies the trial as randomized, parallel, unmasked, and intended for treatment. The ClinicalTrials.gov record does not report a stratification scheme, crossover, factorial structure, interim analysis, Bayesian component, or missing-data/imputation method.
4. Endpoints
| Endpoint | Time frame | Type | Analysis information posted |
|---|---|---|---|
| Pelvic Recurrence Rate at 3 Years | 3 years | Time-to-event primary endpoint | Yes; formal statistical analysis posted |
| Pelvic Relapse-free Survival | 3 years | Time-to-event | Yes |
| Extra-pelvic Relapse-free Survival | 3 years | Time-to-event | Yes |
| Relapse-free Survival | 3 years | Time-to-event | Yes |
| Overall Survival | 3 years | Time-to-event | Yes |
Primary endpoint definition
The registered primary endpoint was Pelvic Recurrence Rate at 3 Years. The registry definition states that the rate was estimated by 1 minus the Kaplan-Meier estimate for the probability of pelvic relapse-free survival (PRFS) at 3 years. PRFS was defined as the time from randomization to the time when a recurrence within the pelvic field was first documented. The registry definition further states that patients with a relapse outside the pelvic field documented or who died before documentation of a pelvic relapse were handled in the PRFS definition, although the registry-reported text ends before the full censoring wording is reproduced.
This construction is important statistically: the reported pelvic recurrence rate is derived from a time-to-event estimate rather than from a simple proportion of patients observed to have pelvic recurrence by the 3-year landmark.
5. Primary Result: Pelvic Recurrence Rate at 3 Years
The registry reports a formal non-inferiority analysis for the primary endpoint among all patients randomized. The groups compared were radical hysterectomy and simple hysterectomy, with the posted effect measure being the mean difference in final values.
Difference in 3-year pelvic recurrence rates
90% two-sided CI: -0.0162 to 0.0232
Hypothesis type: Non-inferiority
Prespecified margin of inferiority: 4%
| Primary analysis feature | Reported value |
|---|---|
| Outcome measure | Pelvic Recurrence Rate at 3 Years |
| Time frame | 3 years |
| Analysis population | All patients randomized |
| Groups compared | Radical Hysterectomy vs Simple Hysterectomy |
| Effect measure | Mean Difference (Final Values) |
| Estimate | 0.0035 |
| Confidence interval | 90%, two-sided |
| CI lower | -0.0162 |
| CI upper | 0.0232 |
| Non-inferiority margin | 4% |
The primary estimate of 0.0035 is the reported difference between the simple and radical hysterectomy groups on the 3-year pelvic recurrence-rate scale. Because the registry describes the comparison as simple versus radical, the positive sign indicates a higher estimated recurrence-rate difference for simple hysterectomy on that stated comparison scale.
The 90% confidence interval of -0.0162 to 0.0232 describes the statistical uncertainty around that estimated difference. It includes zero, so the interval is compatible with differences in either direction. For a non-inferiority question, however, the central issue is not simply whether the interval includes zero. The relevant question is whether the upper confidence bound remains within the prespecified margin of inferiority.
The reported non-inferiority margin was 4%. The upper confidence bound of 0.0232 is below that margin when expressed on the same difference scale. Thus, the statistical logic of a non-inferiority assessment is based on the relationship between the confidence bound and the margin, rather than on whether the estimate is exactly zero.
The registry does not post a p-value for this analysis. A p-value would also not measure the size of the treatment difference. The effect estimate and confidence interval provide the information needed to describe the magnitude and precision of the observed comparison, while the non-inferiority margin supplies the prespecified threshold for the trial's hypothesis.
6. Secondary Endpoint Results
Four secondary time-to-event analyses are posted in the ClinicalTrials.gov record. Each analysis uses all patients randomized and compares radical hysterectomy with simple hysterectomy. The registry reports hazard ratios but does not report the statistical method used to generate them.
| Secondary endpoint | HR | 95% CI | Hypothesis type | Registry direction |
|---|---|---|---|---|
| Pelvic Relapse-free Survival | 1.12 | 0.47–2.67 | Superiority | Simple to radical |
| Extra-pelvic Relapse-free Survival | 3.82 | 0.79–18.4 | Superiority | Simple to radical |
| Relapse-free Survival | 1.54 | 0.69–3.45 | Superiority | Simple to radical |
| Overall Survival | 1.09 | 0.38–3.14 | Superiority | Simple to radical |
Pelvic Relapse-free Survival
Hazard ratio
95% CI: 0.47–2.67
Hazard ratio of simple to radical hysterectomy
A hazard ratio of 1.12 means the estimated instantaneous event rate for the simple-hysterectomy group was 12% higher than the radical-hysterectomy group under the reported hazard-ratio interpretation. The confidence interval is wide and includes 1, so the estimate does not by itself establish a difference in pelvic relapse-free survival.
Extra-pelvic Relapse-free Survival
Hazard ratio
95% CI: 0.79–18.4
Hazard ratio for the simple hysterectomy group relative to the radical hysterectomy group
The point estimate of 3.82 indicates a substantially higher estimated instantaneous event rate in the simple-hysterectomy group under the reported model direction. The very wide 0.79–18.4 confidence interval demonstrates substantial uncertainty. Because the interval includes 1, the point estimate should not be treated as evidence by itself that the two groups differ in extra-pelvic relapse-free survival.
Relapse-free Survival
Hazard ratio
95% CI: 0.69–3.45
Hazard ratio for the simple hysterectomy group relative to the radical hysterectomy group
The reported hazard ratio of 1.54 corresponds to an estimated instantaneous event rate 54% higher for simple versus radical hysterectomy under the stated comparison direction. The 0.69–3.45 confidence interval is broad and includes 1, so considerable uncertainty remains around the point estimate.
Overall Survival
Hazard ratio
95% CI: 0.38–3.14
Hazard ratio of simple to radical hysterectomy
The overall-survival hazard ratio of 1.09 is close to 1, indicating little difference in the estimated instantaneous death rate on the point-estimate scale. The 0.38–3.14 confidence interval is wide and includes values compatible with lower or higher hazards in the simple-hysterectomy group.
7. Safety Results
The ClinicalTrials.gov record provides serious adverse event counts by arm in the form of affected participants divided by participants at risk.
| Safety measure | Radical Hysterectomy | Simple Hysterectomy |
|---|---|---|
| Serious adverse events | 0/344 | 0/338 |
On the ClinicalTrials.gov record, no serious adverse events were reported among the listed participants at risk in either arm. The denominators are 344 and 338, respectively. The ClinicalTrials.gov record does not provide a broader adverse-event table, adverse-event grades, treatment-emergent adverse events, or individual event categories.
8. Statistical Methodology
Kaplan-Meier estimation
The primary endpoint is constructed from the Kaplan-Meier estimate of pelvic relapse-free survival at 3 years. This is a standard time-to-event approach that accounts for the timing of events and censoring rather than treating every participant as if their entire follow-up were identical.
Here, di represents events occurring at an event time and ni represents the number at risk immediately before that time. The registry's primary endpoint then uses 1 minus the PRFS estimate at 3 years.
Time-to-event analysis
Pelvic recurrence, pelvic relapse-free survival, extra-pelvic relapse-free survival, relapse-free survival, and overall survival are all time-related outcomes. For these endpoints, the timing of an event matters. A participant followed for a shorter period without an event does not contribute the same amount of information as a participant observed for the full follow-up period without an event.
Hazard ratio
The secondary analyses report hazard ratios. A hazard ratio compares the estimated instantaneous event rates between groups over the analyzed follow-up. A value of 1 represents equal hazards on the model's relative scale; values above 1 indicate a higher estimated hazard in the numerator group, while values below 1 indicate a lower estimated hazard.
The registry-reported analysis notes specify the simple-to-radical direction for the posted hazard ratios. This direction is essential when interpreting whether a value above or below 1 favors one group on the particular endpoint.
Non-inferiority framework
The primary analysis is explicitly classified as non-inferiority, with a reported margin of inferiority of 4% for the difference in 3-year pelvic recurrence rates between the simple and radical hysterectomy groups.
Non-inferiority is different from a conventional superiority test. The objective is not necessarily to show that the two treatments have exactly equal outcomes. Instead, the design asks whether the new or less intensive strategy can be ruled out as being worse than the comparator by more than a clinically prespecified amount.
For the posted SHAPE primary result, the upper confidence bound is 0.0232 and the reported margin is 4%. The statistical interpretation therefore centers on whether the upper bound remains within the allowable inferiority threshold.
Analysis population
The primary analysis explicitly uses all patients randomized. This is important because randomization defines the treatment groups for the primary comparison. The ClinicalTrials.gov record does not provide a separate per-protocol analysis or an as-treated primary efficacy analysis.
Statistical method reporting
The ClinicalTrials.gov record identifies the effect measures and estimates, but does not name the statistical method for any of the five posted analyses. Consequently, this page explains the statistical concepts that correspond to the reported endpoints without attributing an unreported specific regression model, test statistic, or estimation procedure to the registry.
9. Statistical Methods Explained
Why is the primary endpoint based on Kaplan-Meier estimation?
Pelvic recurrence is a time-to-event outcome. Participants can have different lengths of follow-up, and some may not experience a pelvic recurrence during their observed follow-up. Kaplan-Meier estimation allows the analysis to use the timing of observed recurrences while accounting for censoring. SHAPE then expresses the 3-year pelvic recurrence rate as one minus the estimated probability of remaining pelvic relapse-free at 3 years.
Why is non-inferiority judged against a margin?
A non-inferiority trial needs a threshold that defines how much worse the new strategy could be while still satisfying the trial's non-inferiority criterion. SHAPE reports a 4% margin of inferiority for the difference in 3-year pelvic recurrence rates. The confidence interval is therefore evaluated against that threshold rather than against zero alone.
What does the primary estimate of 0.0035 mean?
The estimate is the reported mean difference in final pelvic recurrence-rate values between the simple and radical hysterectomy groups, using the direction specified by the registry analysis. It is an estimate of the group-level difference; it does not mean that every individual participant experienced a recurrence difference of exactly 0.0035.
What does the 90% confidence interval tell us?
The interval of -0.0162 to 0.0232 quantifies uncertainty around the primary estimate under the analysis framework. It includes both negative and positive values, meaning the observed data are compatible with differences in either direction. In the non-inferiority framework, the upper bound is particularly important because it represents the largest difference supported by the specified confidence level in the direction of inferiority.
What does a hazard ratio of 3.82 mean?
The extra-pelvic relapse-free survival analysis reports a hazard ratio of 3.82 for simple relative to radical hysterectomy. On the reported direction, this corresponds to an estimated instantaneous event rate 3.82 times that of the radical-hysterectomy group. It does not mean that 3.82 times as many patients necessarily experienced the event, nor does it describe an absolute difference in 3-year event probability.
Why does the confidence interval matter more than the point estimate alone?
A point estimate such as 3.82 is only one estimate of the underlying treatment contrast. Its associated confidence interval, 0.79–18.4, is very wide. The interval therefore communicates that the data provide substantially less precision than the point estimate alone might suggest. The same principle applies to the other secondary hazard ratios.
10. Primary vs Secondary Statistical Questions
| Endpoint | Role | Statistical question | Reported effect |
|---|---|---|---|
| Pelvic Recurrence Rate at 3 Years | Primary | Does the difference remain within the 4% non-inferiority margin? | Mean difference 0.0035; 90% CI -0.0162 to 0.0232 |
| Pelvic Relapse-free Survival | Secondary | How do time-to-event hazards compare? | HR 1.12; 95% CI 0.47–2.67 |
| Extra-pelvic Relapse-free Survival | Secondary | How do time-to-event hazards compare? | HR 3.82; 95% CI 0.79–18.4 |
| Relapse-free Survival | Secondary | How do overall relapse-free hazards compare? | HR 1.54; 95% CI 0.69–3.45 |
| Overall Survival | Secondary | How do mortality hazards compare? | HR 1.09; 95% CI 0.38–3.14 |
The primary endpoint and secondary endpoints answer different questions. The primary endpoint is a non-inferiority question about pelvic recurrence at a fixed 3-year time point, constructed from a time-to-event estimate. The secondary hazard ratios instead summarize relative event rates over time. These measures should not be treated as interchangeable.
11. Confidence Intervals and Effect Size
Primary endpoint
The primary estimate is 0.0035 with a 90% confidence interval from -0.0162 to 0.0232. The interval provides a range of plausible values for the difference under the specified analysis framework.
Pelvic relapse-free survival
The HR is 1.12 with a 95% CI of 0.47–2.67. The interval spans 1 and is substantially wider than the point estimate alone.
Extra-pelvic relapse-free survival
The HR is 3.82 with a 95% CI of 0.79–18.4. The width of the interval indicates substantial uncertainty around the point estimate.
Overall survival
The HR is 1.09 with a 95% CI of 0.38–3.14. The point estimate is close to 1, but the interval remains wide.
A confidence interval should not be interpreted as a range containing the true effect with a fixed probability after the study has been completed. It is better understood as an interval generated by a statistical procedure that reflects sampling uncertainty under its assumptions.
12. What the Hazard Ratios Do — and Do Not — Mean
The pelvic relapse-free survival HR of 1.12 means that the estimated instantaneous event rate for simple hysterectomy was 12% higher than that for radical hysterectomy under the registry's stated simple-to-radical comparison.
It does not mean that 12% more participants experienced pelvic relapse, nor does it imply a 12% difference in the 3-year pelvic recurrence rate.
The extra-pelvic relapse-free survival HR of 3.82 is a relative time-to-event measure. It does not mean that the simple-hysterectomy group had 3.82 times the absolute probability of an extra-pelvic relapse at 3 years.
The overall-survival HR of 1.09 is close to 1 on the relative hazard scale. The associated 95% CI of 0.38–3.14 is much wider, emphasizing that the point estimate alone is not an adequate description of the uncertainty.
13. Non-Inferiority: The Key Statistical Issue
The most important methodological feature of SHAPE in the ClinicalTrials.gov record is the non-inferiority framework for the primary endpoint. This changes how the primary confidence interval should be read.
| Quantity | Reported value | Role in interpretation |
|---|---|---|
| Primary estimate | 0.0035 | Observed difference on the registry's reported scale |
| 90% CI lower bound | -0.0162 | Lower end of uncertainty interval |
| 90% CI upper bound | 0.0232 | Critical bound for inferiority assessment |
| Non-inferiority margin | 4% | Maximum prespecified inferiority threshold |
| Hypothesis type | Non-inferiority | Primary inferential framework |
If the upper confidence bound remains below the prespecified inferiority margin, the data satisfy the central numerical criterion used in a conventional confidence-interval approach to non-inferiority. The SHAPE ClinicalTrials.gov record reports an upper bound of 0.0232 against a margin of 4%.
This is different from saying that the two operations are proven identical. Non-inferiority allows a prespecified amount of difference. The scientific interpretation therefore depends on the appropriateness of the margin as well as the observed estimate and confidence interval.
14. Multiplicity, Interim Analysis, and Other Design Topics
The ClinicalTrials.gov record supports a discussion of the primary non-inferiority hypothesis and the four posted secondary analyses. They do not report an interim analysis plan, alpha-spending method, multiplicity adjustment, stratification factors, crossover, factorial structure, Bayesian methods, or missing-data/imputation method.
| Design topic | Information in the ClinicalTrials.gov record |
|---|---|
| Non-inferiority margin | 4% margin of inferiority for the difference in 3-year pelvic recurrence rates |
| Crossover | Not reported |
| Factorial design | Not reported; design model is parallel |
| Multiplicity adjustment | Not reported |
| Interim analysis | Not reported |
| Missing-data/imputation method | Not reported |
| Stratification | Not reported |
| Bayesian methods | Not reported |
The absence of a reported method in the registry analysis should not be treated as evidence that the method was not used in the full protocol or statistical analysis plan. It means only that the ClinicalTrials.gov record does not specify it.
15. Analysis Population and Censoring
The primary analysis population is explicitly identified as all patients randomized. The secondary analyses also use all patients randomized.
This is particularly relevant for time-to-event outcomes because participants can be censored. Censoring means that the event of interest has not been observed by the last point at which the participant contributes follow-up information, or that the participant otherwise enters the analysis's censoring mechanism. Kaplan-Meier estimation is designed to incorporate such incomplete follow-up rather than treating censored participants as if they had experienced the event.
The primary and secondary analyses use all randomized patients, preserving the treatment assignment defined at randomization for the reported efficacy comparisons.
16. Limitations
- Registry method detail: the ClinicalTrials.gov record does not name the statistical method for any of the five analyses.
- Secondary-event precision: the secondary hazard-ratio confidence intervals are broad, particularly for extra-pelvic relapse-free survival.
- Hazard-ratio interpretation: hazard ratios are relative time-to-event measures and should not be substituted for absolute 3-year risks or risk differences.
- Proportional-hazards assumption: when a Cox-type hazard ratio is used, interpretation of a single HR as a stable relative effect depends on the underlying modeling assumptions. The ClinicalTrials.gov record does not report a formal assessment of proportional hazards.
- Non-inferiority depends on the margin: the statistical conclusion is tied to the prespecified 4% margin. The ClinicalTrials.gov record does not provide the clinical or statistical derivation of that margin.
- Multiplicity: the ClinicalTrials.gov record does not report how the primary and secondary hypotheses were ordered or whether multiplicity adjustments were applied.
- Missing-data methods: the ClinicalTrials.gov record does not describe imputation or other missing-data procedures.
- Subsequent treatment: the ClinicalTrials.gov record does not provide information about treatment received after the randomized intervention, so no additional interpretation should be inferred.
- Generalizability: the ClinicalTrials.gov record identifies the condition and low-risk early stage cervical cancer in the brief title, but does not provide the full eligibility and baseline-characteristic detail needed to assess generalizability comprehensively.
17. Why This Trial Matters Statistically
SHAPE is particularly useful as a teaching case because the primary endpoint illustrates an important distinction in clinical-trial statistics: non-inferiority is not simply a test of whether two groups differ. The interpretation requires an effect estimate, an appropriate confidence interval, and a prespecified clinically meaningful margin.
| Concept | How it appears in SHAPE |
|---|---|
| Randomization | The trial uses randomized allocation with two parallel treatment arms. |
| Time-to-event endpoint | The primary pelvic recurrence rate is derived from a time-to-event PRFS analysis. |
| Kaplan-Meier estimation | The primary endpoint is defined using 1 minus the Kaplan-Meier PRFS estimate at 3 years. |
| Non-inferiority | The primary hypothesis uses a reported 4% margin of inferiority. |
| Confidence interval | The primary estimate is accompanied by a two-sided 90% confidence interval. |
| Hazard ratio | Four secondary time-to-event analyses report HRs with 95% confidence intervals. |
| Analysis population | Primary and secondary analyses use all patients randomized. |
| Absolute vs relative effects | The primary difference and secondary hazard ratios use different effect scales and answer different questions. |
| Precision | Several secondary HR confidence intervals are wide, demonstrating substantial uncertainty. |
| Method transparency | The ClinicalTrials.gov record identifies effect measures but does not name the analysis method. |
18. A Practical Reading of the SHAPE Results
A statistically careful reading starts with the primary endpoint rather than the secondary hazard ratios. The primary result reports a difference of 0.0035 and a two-sided 90% confidence interval of -0.0162 to 0.0232. The trial's non-inferiority margin is 4%. The upper confidence limit therefore remains below the stated margin.
The secondary endpoints provide additional information about different time-to-event outcomes. Their hazard ratios range from 1.09 for overall survival to 3.82 for extra-pelvic relapse-free survival, but the associated confidence intervals are wide and all four include 1. These secondary estimates therefore require substantially more caution than a point-estimate-only reading would suggest.
Another important distinction is between the primary pelvic recurrence question and the broader outcomes of relapse-free survival and overall survival. A non-inferiority result for one endpoint does not automatically establish non-inferiority for every other clinical outcome. Each endpoint has its own statistical estimand and uncertainty.
19. Related Tutorials
Learn more about the methods used in this trial:
20. Related Calculators
21. Sources
- ClinicalTrials.gov: SHAPE — NCT01658930.
- PubMed: PMID 38416430.
- PubMed: PMID 39453395.
- PubMed: PMID 39353164.
- PubMed: PMID 37650664.
Continue through Clinical Biostats
Explore deeper tutorials and statistical tools for confidence intervals, survival analysis, hazard ratios, non-inferiority designs, randomization, and time-to-event endpoints.
22. Record Summary
SHAPE provides a clear example of how a randomized clinical trial can be structured around a non-inferiority time-to-event endpoint. The primary analysis reports a difference of 0.0035 with a two-sided 90% confidence interval of -0.0162 to 0.0232, evaluated against a reported 4% margin of inferiority. Four secondary time-to-event analyses report hazard ratios ranging from 1.09 to 3.82, with confidence intervals that demonstrate substantial uncertainty around those point estimates.
The statistical lesson is broader than any single number. A non-inferiority analysis must be read against its prespecified margin; a hazard ratio must be interpreted according to its comparison direction; and a confidence interval is essential for understanding precision. The primary endpoint, secondary time-to-event outcomes, and safety counts therefore provide different pieces of the statistical evidence and should not be collapsed into a single summary measure.