This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record.
1. Trial at a Glance
RTOG 0417 was a phase 2, single-group treatment trial of bevacizumab, cisplatin, external beam radiation therapy, and internal radiation therapy in patients with previously untreated locally advanced cervical cancer. The registry reports an enrollment of 60 participants and one treatment arm.
| Feature | RTOG 0417 |
|---|---|
| Phase | Phase 2 |
| Status | Completed |
| Design | Single-group |
| Allocation | NA |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 60 |
| Number of arms | 1 |
| Lead sponsor | National Cancer Institute (NCI) |
| Sponsor type | NIH |
| Start date | 2006-08-11 |
| Primary completion date | 2010-06-04 |
| ClinicalTrials.gov | NCT00369122 |
2. Clinical Question
The clinical question was whether a treatment regimen combining bevacizumab, radiation therapy, and cisplatin could be evaluated for treatment-related adverse events in patients with previously untreated locally advanced cervical cancer.
Population
Patients with previously untreated cervical cancer represented by the registry conditions, including cervical adenocarcinoma, cervical adenosquamous carcinoma, cervical squamous cell carcinoma not otherwise specified, and specified stage IB, IIA, IIB, and III cervical cancer categories.
Intervention
Bevacizumab, cisplatin, external beam radiation therapy, and internal radiation therapy.
Comparator
There was no comparator arm. The registry identifies the study as a single-group design with one treatment arm.
Primary question
Whether the observed treatment-related serious adverse events and adverse events within the first 90 days remained within the prespecified tolerability framework described in the registry.
3. Trial Design
Radiation Therapy, Bevacizumab, Cisplatin
- Bevacizumab
- Cisplatin
- External beam radiation therapy
- Internal radiation therapy
Non-randomized treatment study
- Phase 2
- Single-group design
- No masking
- Allocation listed as NA
- Primary purpose: treatment
A single-group design changes the statistical question substantially compared with a randomized controlled trial. There is no concurrent control group against which to estimate a between-treatment difference. Instead, the registry describes a prespecified tolerability framework based on the number of serious adverse events and adverse events observed during the first 90 days from treatment start.
4. Endpoints
The registry lists one primary endpoint. It measures treatment-related serious adverse events and adverse events using CTCAE version 3.0 criteria during the first 90 days after treatment begins.
| Endpoint | Registry definition | Time frame |
|---|---|---|
| Number of Subjects With Treatment-related Serious Adverse Events (SAEs) and Adverse Events (AEs) as Assessed by CTCAE v. 3.0 Criteria Within the First 90 Days From Treatment Start | Adverse events graded using CTCAE v3.0. Treatment-related SAEs include Grade ≥4 vaginal bleeding, Grade ≥4 thrombotic event, Grade ≥3 arterial event, gastrointestinal bleeding, bowel/bladder perforation, and any Grade 5 treatment-related AE. Treatment-related AEs include all SAEs and specified Grade 3–4 gastrointestinal, hematologic, renal, cardiac, pulmonary, hepatic, and neurologic events, together with specified persistent nausea, vomiting, diarrhea, neutropenia, leukopenia, and febrile neutropenia criteria. | From start of treatment to 90 days. |
CTCAE severity framework
The registry describes CTCAE v3.0 grades as a general severity framework: Grade 1 is mild, Grade 2 moderate, Grade 3 severe, Grade 4 life-threatening or disabling, and Grade 5 death related to an adverse event.
The endpoint therefore combines a broad adverse-event concept with more narrowly defined treatment-related serious events. This distinction is important statistically because the endpoint is not simply a count of every adverse event experienced by every participant.
5. Primary Result
The registry reports 13/59 affected by serious adverse events in the treatment arm, expressed as affected participants over participants at risk. The posted statistical analysis describes a decision rule based on the number of serious adverse events and adverse events rather than reporting a conventional treatment-effect estimate, confidence interval, or p-value.
Serious adverse events
Affected / at risk in the treatment arm
The registry's posted analysis specifies rejection if there were ≥6 patients with SAEs or ≥22 patients with AEs.
| Reported quantity | Registry value | Statistical role |
|---|---|---|
| Participants enrolled | 60 | Total enrollment |
| Serious adverse events, affected / at risk | 13/59 | Safety result reported by arm |
| SAE rejection threshold | ≥6 pts with SAEs | Prespecified decision boundary |
| AE rejection threshold | ≥22 pts with AEs | Prespecified decision boundary |
| Target tolerable SAE rate | 5% | Design assumption described in the registry |
| Target tolerable AE rate | 35% | Design assumption described in the registry |
| Excessive SAE rate | ≥20% | Design assumption described in the registry |
| Excessive AE rate | ≥55% | Design assumption described in the registry |
| Alpha | 0.05 | Design characteristic |
| Power | 90% | Design characteristic |
The value 13/59 is a reported safety count with its corresponding at-risk denominator. It tells the reader how many participants were recorded as affected among those at risk in the treatment arm; it is not a treatment-effect estimate because the trial had no comparator group.
It also does not mean that 13 participants experienced every event included in the primary endpoint, nor does it establish that the treatment caused every adverse event represented by the count. The registry separately defines treatment-related SAEs and treatment-related AEs according to specific CTCAE v3.0 criteria.
No confidence interval or p-value is reported with the 13/59 figure in the posted statistical analysis. A confidence interval would describe statistical uncertainty around an estimated quantity, while a p-value would address a prespecified hypothesis under a defined statistical model; neither is posted on ClinicalTrials.gov for this reported count.
The most important statistical feature is therefore the prespecified decision framework: the analysis compares observed event counts with thresholds rather than comparing two randomized treatment groups.
6. Statistical Methodology
Single-group safety design
Because RTOG 0417 has one treatment arm, the primary statistical analysis is based on the observed number of qualifying adverse events rather than a treatment-versus-control contrast. The registry describes a tolerability framework in which an SAE rate of 5% and an AE rate of 35% were considered tolerable, while an SAE rate of ≥20% and an AE rate of ≥55% were considered excessive.
Count-based decision rule
The registry translates the tolerability assumptions into operational decision boundaries. If there were ≥6 patients with SAEs or ≥22 patients with AEs, the treatment would be rejected under the stated design.
The rule is based on observed event counts. It is different from a conventional two-arm hypothesis test because there is no randomized comparator group.
Alpha and power
The registry states that this study design provides alpha of 0.05 and power of 90%. In a design of this type, alpha represents the prespecified type I error framework for the decision procedure, while power describes the design's ability to identify an event-rate situation defined as excessive under the specified assumptions.
Power is a property of the planned design under particular assumptions. It should not be interpreted as the probability that the observed treatment was safe, effective, or beneficial after the study has been completed.
Eligible-patient analysis population
The posted statistical analysis specifies eligible patients who began study treatment as the analysis population. This differs from simply using the enrollment number as the denominator for every analysis.
Adverse-event classification
The primary endpoint uses CTCAE v3.0 grading and additional treatment-related definitions. The statistical analysis therefore depends on clinical classification before the event count can be compared with the prespecified boundaries.
7. Statistical Methods Explained
Why is there no treatment-versus-control hazard ratio?
Hazard ratios are generally used to compare time-to-event outcomes between groups. RTOG 0417 is a single-group study, so there is no randomized comparator against which to estimate a relative hazard of adverse events.
Why are event-count thresholds used?
A count threshold provides a direct operational rule for a single-group safety design. Instead of asking whether one treatment performs differently from another, the design asks whether the number of qualifying adverse events crosses a prespecified boundary associated with excessive toxicity.
What does the 13/59 value represent?
It represents 13 affected participants among 59 participants at risk for the reported serious adverse-event measure. The numerator is a count and the denominator identifies the corresponding at-risk population. It is not a comparative effect measure.
Why are the tolerable and excessive rates important?
The registry describes 5% as the SAE rate considered tolerable and ≥20% as excessive. For adverse events, the corresponding values are 35% and ≥55%. These assumptions define the statistical framework from which the count-based rejection boundaries are derived.
What does alpha of 0.05 mean here?
Alpha is the type I error level specified for the study design. In this context, it belongs to the prespecified decision procedure rather than to a conventional p-value comparing two treatment groups.
What does power of 90% mean?
Power describes the planned ability of the statistical design to distinguish the specified excessive-event scenario under its assumptions. It is not the probability that the observed treatment is safe and is not a post-study probability assigned to the treatment itself.
8. Safety Results
The registry reports serious adverse events by arm as 13/59 for the treatment arm consisting of radiation therapy, bevacizumab, and cisplatin. No separate numerical count for the broader adverse-event category is included in the posted statistical analysis.
| Safety measure | Treatment arm |
|---|---|
| Radiation Therapy, Bevacizumab, Cisplatin | 13/59 serious adverse events, affected / at risk |
| Broader adverse-event count | Not numerically reported in the posted statistical analysis |
The distinction between serious adverse events and the broader adverse-event endpoint is central to interpreting the safety analysis. The registry defines treatment-related AEs to include all SAEs plus specified Grade 3–4 toxicities and other specified clinical criteria. Consequently, the SAE count alone should not be treated as the complete primary endpoint.
9. Limitations
- No randomized comparator: the single-group design does not provide a concurrent control group for estimating a comparative treatment effect.
- Safety-focused endpoint: the primary endpoint concerns treatment-related adverse events during the first 90 days rather than a time-to-event efficacy outcome.
- Composite event definitions: treatment-related SAEs and AEs are defined through CTCAE v3.0 grades and specific clinical criteria, so interpretation depends on consistent event classification.
- Different analysis and enrollment populations: the posted statistical analysis uses eligible patients who began study treatment, while the registry enrollment is 60 participants.
- Incomplete numerical detail for the broader endpoint: the registry reports the serious-adverse-event count but does not provide a separate numerical AE count in the posted statistical analysis.
- Count thresholds are design-specific: the ≥6 SAE and ≥22 AE boundaries should be interpreted within the tolerability assumptions specified for this study rather than as universal safety thresholds.
- Limited inferential measures: the posted analysis does not report a confidence interval or p-value for the primary safety result.
- Ineligible participants and endpoint reporting: the registry states that ineligible patients are included in adverse-event reporting on ClinicalTrials.gov but not in endpoint results or publications, per the protocol.
10. Why This Trial Matters Statistically
RTOG 0417 is useful statistically because it illustrates an important clinical-trial design that is different from the familiar randomized two-arm model. Its primary statistical problem is not estimation of a hazard ratio or a treatment difference. Instead, the study uses a prespecified safety boundary to determine whether the observed number of qualifying adverse events is compatible with the tolerability assumptions defined for the regimen.
| Concept | How it appears in RTOG 0417 |
|---|---|
| Single-group design | One treatment arm with no randomized comparator |
| Safety endpoint | Treatment-related SAEs and AEs within the first 90 days |
| CTCAE grading | Adverse events assessed using CTCAE v3.0 |
| Count-based decision rule | Rejection boundaries of ≥6 patients with SAEs or ≥22 patients with AEs |
| Tolerability assumptions | SAE rate of 5% and AE rate of 35% considered tolerable |
| Excessive-event assumptions | SAE rate of ≥20% and AE rate of ≥55% considered excessive |
| Alpha | 0.05 |
| Power | 90% |
| Analysis population | Eligible patients who began study treatment |
| Safety reporting | 13/59 serious adverse events reported for the treatment arm |
11. Understanding the Decision Boundary
The most instructive statistical feature of this trial is the difference between a rate assumption and a decision boundary. The registry describes 5% as a tolerable SAE rate and ≥20% as excessive. It then expresses the design operationally as a threshold of ≥6 patients with SAEs.
Rate assumption
A rate such as 5% is a design assumption about the level of adverse events considered tolerable under the statistical framework.
Observed count
The study accumulates actual qualifying adverse-event counts among the specified analysis population.
Decision boundary
The registry specifies ≥6 patients with SAEs or ≥22 patients with AEs as the threshold for rejecting the treatment.
Statistical conclusion
The observed count must be interpreted against the prespecified design rule rather than against an unrelated p-value or comparative effect measure.
This framework demonstrates why sample-size planning and endpoint definition are inseparable. The event thresholds are meaningful only because they were established within a design that specified tolerable and excessive event-rate assumptions together with alpha and power.
12. Primary Endpoint: What Is Being Counted?
The primary endpoint is broader than serious adverse events alone. The registry defines treatment-related SAEs through specific high-grade or clinically important events, while treatment-related AEs include all SAEs and additional Grade 3–4 toxicities and specified persistent or severe clinical events.
| Category | Examples from the registry definition |
|---|---|
| Treatment-related SAEs | Grade ≥4 vaginal bleeding; Grade ≥4 thrombotic event; Grade ≥3 arterial event; gastrointestinal bleeding; bowel/bladder perforation; any Grade 5 treatment-related AE |
| Treatment-related AEs | All SAEs; specified Grade 3–4 nausea, vomiting, or diarrhea; specified prolonged neutropenia or leukopenia; febrile neutropenia; Grade 3–4 hematologic toxicity; Grade 3–4 GI, renal, cardiac, pulmonary, hepatic, or neurologic AEs |
This structure illustrates a common statistical principle in safety analysis: an endpoint definition must be established before counting events. A simple total of all reported symptoms would answer a different question from the prespecified treatment-related AE endpoint.
13. Analysis Population and Denominators
The registry identifies the primary statistical analysis population as eligible patients who began study treatment. The trial enrolled 60 participants, while the reported serious-adverse-event figure uses 59 as its at-risk denominator.
A trial's enrollment number does not automatically become the denominator for every endpoint. Eligibility, treatment initiation, follow-up, and endpoint-specific rules can determine the population used for a particular analysis.
This distinction is particularly important for safety reporting. The registry explicitly states that ineligible patients are included in adverse-event reporting on ClinicalTrials.gov but are not included in endpoint results or publications, according to the protocol.
14. Alpha, Power, and the Meaning of a Statistical Decision
The registry states that the study design provides alpha of 0.05 and power of 90%. These quantities describe the operating characteristics of the planned statistical design.
Alpha = 0.05
This specifies the type I error framework for the decision procedure. It is not itself a measure of the magnitude or clinical importance of an adverse-event burden.
Power = 90%
This describes the planned ability to detect the specified excessive-event scenario under the design assumptions. It is not a probability that the treatment is safe.
For a single-group safety study, the interpretation of the design therefore begins with the prespecified event-rate assumptions and decision boundaries. It does not begin with a comparison of treatment-group means, a hazard ratio, or a two-sided p-value.
15. Results in Statistical Context
The reported 13/59 serious-adverse-event result should be read together with the study's decision framework. The registry specifies that ≥6 patients with SAEs or ≥22 patients with AEs would lead to rejection of the treatment under the stated design.
The reported count is directly interpretable as a safety observation: 13 affected participants were reported among 59 participants at risk for the serious-adverse-event measure.
However, a count alone is not equivalent to an effect estimate. There is no second randomized group from which to calculate a treatment difference or comparative ratio. The registry also does not provide a confidence interval or p-value for this result.
The prespecified thresholds provide the statistical context. The relevant question is whether the observed qualifying-event counts satisfy the study's predefined tolerability decision rule, not whether the count produces a conventional between-group significance test.
16. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
The study uses a single-group, count-based safety design. Its posted statistical analysis specifies tolerability assumptions, rejection thresholds, alpha of 0.05, and power of 90%, with eligible treated patients as the analysis population.
Clinical interpretation
The treatment regimen consisted of bevacizumab, cisplatin, external beam radiation therapy, and internal radiation therapy. The registry reports 13/59 serious adverse events for the treatment arm during the study's safety assessment framework.
These two perspectives should remain distinct. The statistical design determines how safety observations are evaluated; the clinical meaning depends on the nature, severity, attribution, and timing of the adverse events defined by the protocol.
17. What This Analysis Does Not Show
Because the study is single-group, the registry record does not provide a randomized treatment comparison from which to infer a relative treatment effect. The posted primary statistical analysis is focused on safety-event thresholds rather than comparative efficacy.
| Question | What the registry record establishes |
|---|---|
| Was there a randomized comparator? | No. The design is single-group with one arm. |
| Can a treatment-versus-control hazard ratio be estimated? | No randomized comparator is present for such a comparison. |
| What was the primary endpoint? | Treatment-related SAEs and AEs within the first 90 days from treatment start. |
| What safety count is reported? | 13/59 serious adverse events for the treatment arm. |
| Is a confidence interval reported for that count? | No. |
| Is a p-value reported for that count? | No. |
| What decision thresholds are specified? | ≥6 patients with SAEs or ≥22 patients with AEs. |
18. Important Statistical Distinctions
| Distinction | Why it matters |
|---|---|
| Single-group vs randomized | A single-group study cannot directly estimate a randomized treatment effect against a concurrent control. |
| Count vs rate | The reported 13/59 is a count with an at-risk denominator, whereas the tolerability assumptions are expressed as event rates. |
| SAE vs AE | SAEs are one component of the primary endpoint; the broader treatment-related AE definition includes additional specified events. |
| Enrollment vs analysis population | The enrolled population is 60, while the posted primary analysis population is eligible patients who began treatment and the reported SAE denominator is 59. |
| Alpha vs p-value | Alpha is a design parameter controlling type I error; a p-value is an observed-data quantity generated by a specified statistical test. |
| Power vs probability of safety | Power is a property of a design under specified assumptions, not the probability that a treatment is safe after observing the data. |
19. Trial Timeline
Study start
The registry lists 2006-08-11 as the study start date.
Primary completion
The registry lists 2010-06-04 as the primary completion date.
Registry status
The ClinicalTrials.gov record identifies RTOG 0417 as completed.
20. Source and Record Context
The ClinicalTrials.gov record identifies the study as RTOG 0417, NCT00369122, a phase 2 study sponsored by the National Cancer Institute (NCI). The sponsor type is listed as NIH.
| Registry field | Reported value |
|---|---|
| NCT ID | NCT00369122 |
| Acronym | RTOG 0417 |
| Phase | 2 |
| Status | COMPLETED |
| Lead sponsor | National Cancer Institute (NCI) |
| Sponsor type | NIH |
| Enrollment | 60 |
| Number of arms | 1 |
21. Why the Safety Endpoint Requires Careful Reading
Safety endpoints can appear simple because they are often expressed as event counts, but the statistical interpretation depends on the endpoint definition, observation window, analysis population, and decision rule.
RTOG 0417 specifies all four. The observation window is from treatment start to 90 days. The analysis population is eligible patients who began study treatment. The endpoint uses CTCAE v3.0 and additional treatment-related definitions. Finally, the statistical analysis defines event-count thresholds linked to tolerability assumptions.
That combination makes the trial a useful example of how a clinical endpoint becomes a statistical decision rule. The count is not interpreted in isolation; it is evaluated within the design that established what constitutes an excessive safety signal.
22. Statistical Summary
Core statistical structure
Primary endpoint: treatment-related SAEs and AEs within 90 days from treatment start
Decision rule: ≥6 patients with SAEs or ≥22 patients with AEs
Design alpha: 0.05 · Power: 90%
The central statistical lesson is that not every phase 2 trial is designed around a conventional treatment-effect estimate. RTOG 0417 uses a single-group safety framework in which observed event counts are compared with predefined tolerability boundaries. The resulting analysis should therefore be read as a safety decision framework rather than as a randomized efficacy comparison.
23. Related Clinical Trial Results
24. Sources
- ClinicalTrials.gov: RTOG 0417 (NCT00369122).
Continue through the Clinical Biostats knowledge graph
Explore additional clinical-trial methodology, statistical concepts, and analysis workflows.
25. Record Summary
RTOG 0417 is a phase 2, single-group study of bevacizumab, cisplatin, external beam radiation therapy, and internal radiation therapy in patients with previously untreated locally advanced cervical cancer. Its primary endpoint measures treatment-related serious adverse events and adverse events assessed with CTCAE v3.0 from treatment start through 90 days.
The statistical design is based on predefined tolerability assumptions and count thresholds rather than a randomized treatment comparison. The registry describes 5% as a tolerable SAE rate and 35% as a tolerable AE rate, while ≥20% and ≥55%, respectively, are considered excessive. The operational rejection boundaries are ≥6 patients with SAEs or ≥22 patients with AEs, with alpha of 0.05 and power of 90%.
The reported serious-adverse-event result is 13/59 affected participants among those at risk in the treatment arm. The posted statistical analysis does not provide a confidence interval or p-value for this result, and the registry does not provide a separate numerical count for the broader AE component in the posted analysis. The appropriate interpretation therefore centers on the prespecified safety framework, endpoint definitions, and analysis population rather than on an unreported comparative effect measure.