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
I-SPY 1 was a phase 2, single-group diagnostic study evaluating MRI/MRS in women receiving chemotherapy for stage III breast cancer. The registry lists disease-free three-year survival as the primary endpoint, measured at 3 years.
| Feature | I-SPY 1 |
|---|---|
| Acronym | I-SPY 1 |
| ClinicalTrials.gov identifier | NCT00043017 |
| Title | Magnetic Resonance Imaging in Women Receiving Chemotherapy for Stage III Breast Cancer |
| Condition | Breast Cancer |
| Phase | Phase 2 |
| Design model | Single-group |
| Allocation | Not applicable |
| Masking | None |
| Primary purpose | Diagnostic |
| Enrollment | 356 |
| Number of arms | 1 |
| Status | Completed |
| Start date | May 2002 |
| Primary completion date | May 2005 |
| Lead sponsor | American College of Radiology Imaging Network |
| Sponsor type | Network |
2. Clinical Question
The registry describes a study of MRI/MRS in women receiving chemotherapy for stage III breast cancer. The registered primary outcome is disease-free three-year survival, with the endpoint measured at 3 years.
Population
Women receiving chemotherapy for stage III breast cancer.
Intervention
MRI/MRS, with gadopentetate dimeglumine listed in the registry as an intervention.
Comparator
The registry describes a single-group design, so there is no randomized comparator arm.
Primary question
How does disease-free three-year survival perform in this study population receiving chemotherapy, with MRI/MRS incorporated into the study?
3. Trial Design
| Study element | Registry description |
|---|---|
| Enrollment | 356 |
| Number of arms | 1 |
| Allocation | NA |
| Design model | SINGLE_GROUP |
| Masking | NONE |
| Primary purpose | DIAGNOSTIC |
| Interventions | MRI/MRS (procedure); gadopentetate dimeglumine (radiation) |
4. Endpoints
| Endpoint | Registry definition | Time frame |
|---|---|---|
| Disease-free three-year survival | Disease-free three-year survival | 3 years |
The registry identifies one primary endpoint. No additional primary endpoint is listed in the record summarized here.
5. Statistical Methodology
Disease-free survival as a time-to-event endpoint
Disease-free three-year survival is a time-to-event endpoint. Rather than recording only whether an event occurred, survival analysis uses the timing of the event and can also accommodate patients whose event status is not observed through the full follow-up period.
For a study reporting disease-free survival at a fixed time point, the natural statistical target is the estimated probability of remaining disease-free through that time. A survival-function representation is:
Here, T represents the time to the event defining failure of disease-free survival, and S(t) represents the probability of remaining event-free beyond time t.
Kaplan-Meier estimation
A typical analysis of disease-free survival uses the Kaplan-Meier estimator. It estimates the survival curve from observed event times while retaining information from participants who are censored before experiencing the event.
At a particular event time, the Kaplan-Meier estimate updates according to the number of events and the number of participants still at risk. Conceptually:
di is the number of events at time ti, while ni is the number at risk immediately before that event time.
Fixed-time survival
Because the registered endpoint is specifically defined at 3 years, a central summary would be the estimated disease-free survival probability at 3 years. A confidence interval would normally accompany that estimate to quantify statistical uncertainty.
Censoring
Survival analysis is useful when not every participant has a fully observed event time. A participant who remains disease-free through the last observed follow-up can contribute information until that point and then be censored. The validity of the resulting estimate depends on the assumptions governing censoring and follow-up.
6. Statistical Methods Explained
Why is disease-free survival analyzed as a time-to-event outcome?
Because the endpoint concerns whether participants remain disease-free through a specified period, the timing of disease recurrence or another event contributing to the endpoint contains information. Time-to-event methods use that timing rather than reducing every participant to a simple yes/no classification without regard to follow-up duration.
Why would Kaplan-Meier estimation be appropriate?
Kaplan-Meier estimation is designed for time-to-event data in which some participants may be censored. It produces an estimated survival function that can be evaluated at clinically relevant times, including the registered 3-year time point.
What does disease-free survival at 3 years mean?
It is the estimated probability of remaining disease-free through 3 years under the endpoint definition used by the study. It is a population-level probability, not a guarantee that a particular participant will or will not remain disease-free.
Why does censoring matter?
If a participant's disease-free status is not observed for the complete follow-up period, treating that participant as though the event either definitely occurred or definitely did not occur would discard or distort information. Survival methods instead use the observed follow-up period and censor appropriately.
Can a single-group survival estimate establish a treatment effect?
Not in the same way as a randomized comparison. Without a randomized comparator, an observed disease-free survival estimate cannot isolate the causal contribution of the study intervention from other factors affecting outcome.
What would a confidence interval add?
A confidence interval around the estimated 3-year disease-free survival probability would describe statistical uncertainty associated with estimating the population survival probability from the enrolled participants. It would not describe the range of outcomes expected for individual patients.
7. Planned Analysis
The registered primary endpoint is disease-free three-year survival, with a time frame of 3 years. For an endpoint of this type, the standard statistical framework is a time-to-event analysis, commonly using Kaplan-Meier estimation to characterize the disease-free survival function and to obtain the estimated survival probability at the prespecified 3-year time point.
If event times and censoring information are available, the analysis can preserve the timing of disease-free events and the amount of follow-up contributed by participants who do not experience an event during observation. An uncertainty interval around the 3-year estimate would ordinarily accompany the point estimate.
Because the registered design is single-group, there is no randomized between-arm hazard ratio or treatment-group comparison for this endpoint. A single-group survival estimate should therefore be interpreted descriptively unless a separate prespecified external comparison or benchmark is part of the study's statistical framework.
8. Understanding the Three-Year Endpoint
What is estimated?
The probability of remaining disease-free through the registered 3-year time point.
What is not estimated?
A randomized treatment effect, because the registered study has one arm and no randomized comparator.
Why timing matters
Participants can contribute different amounts of follow-up, making time-to-event analysis more informative than a simple proportion observed at a single visit.
Why uncertainty matters
The enrolled sample provides an estimate of the underlying disease-free survival experience, so statistical uncertainty should accompany the estimate.
Three-year survival is not the same as a median
A fixed-time survival probability and a median survival measure answer different questions. The registered endpoint specifies a probability at 3 years rather than a median disease-free survival time. A median would identify the time at which the estimated survival function reaches 0.50, if that point were observed.
Three-year survival is not an individual prediction
A population-level survival estimate should not be read as a prediction that each individual participant has the same probability of remaining disease-free. Individual prognosis can vary substantially with factors that are not captured by a single aggregate survival estimate.
9. Single-Group Design and Causal Interpretation
The absence of randomization is one of the most important statistical characteristics of this study. Randomization creates comparability between treatment groups in expectation and provides the foundation for a randomized causal treatment comparison. A single-group study does not provide that same internal comparator.
| Question | Single-group implication |
|---|---|
| Is there a randomized control group? | No. The registry describes one study arm and a single-group design. |
| Can a treatment hazard ratio be estimated? | Not from the registered single-group design alone. |
| Can disease-free survival be described? | Yes. A time-to-event estimate can describe disease-free survival in the enrolled population. |
| Does the survival estimate isolate the effect of MRI/MRS? | No. A single-group estimate does not provide the randomized counterfactual required for a direct treatment-effect comparison. |
| What is the main statistical focus? | Estimating and interpreting disease-free survival over the registered 3-year time frame. |
This distinction is particularly important when interpreting diagnostic studies. An imaging procedure can be incorporated into a clinical study while the principal clinical endpoint remains a time-to-event outcome. The statistical analysis must therefore distinguish the measurement role of the imaging procedure from the outcome analysis represented by disease-free survival.
10. Endpoint Interpretation: Disease-Free Three-Year Survival
Disease-free three-year survival is a fixed-time survival probability. Conceptually, it asks: what proportion of the study population is estimated to remain disease-free through 3 years? The estimate incorporates the observed timing of events and the available follow-up.
It does not mean that every participant has the same individual probability of remaining disease-free, nor does it establish that MRI/MRS caused a particular level of disease-free survival. The latter requires a causal comparison that the registered single-group design does not provide by itself.
A confidence interval around the 3-year survival estimate would quantify uncertainty in the population-level estimate. A narrower interval generally indicates greater statistical precision, while a wider interval indicates greater uncertainty. The interval is not a range containing the outcomes of individual participants.
11. Statistical Considerations for Censoring and Follow-Up
Three-year disease-free survival requires appropriate handling of participants whose complete three-year disease-free status is not observed. Survival methods allow these participants to contribute information through the point at which they are last known to be event-free.
The analysis distinguishes an observed event from the end of available follow-up. This distinction is fundamental to valid time-to-event estimation.
The interpretation of a Kaplan-Meier estimate depends on the assumptions behind censoring. In particular, informative loss to follow-up can create bias if the probability of being censored is related to the underlying future disease-free outcome in a way not adequately addressed by the analysis.
For a three-year endpoint, adequate follow-up is also important. Participants observed for less than the full time period can still contribute useful information to the survival curve, but they do not provide the same direct three-year observation as a participant followed through the full period without an event.
12. What Is and Is Not Identified by This Registry Record
| Statistical element | Registry record |
|---|---|
| Primary endpoint | Disease-free three-year survival |
| Primary endpoint time frame | 3 years |
| Enrollment | 356 |
| Study arms | 1 |
| Formal posted statistical analyses | None |
| Posted endpoint estimate | Not reported in the registry record summarized here |
| Posted confidence interval | Not reported in the registry record summarized here |
| Posted p-value | Not reported in the registry record summarized here |
This distinction matters because the presence of a registered endpoint does not itself constitute an observed statistical result. A prespecified endpoint defines what is intended to be measured; an analysis result requires the corresponding outcome data and statistical analysis.
13. Design Features With Statistical Consequences
Single-group allocation
There is one registered study arm, so the principal analysis is not a between-group randomized comparison.
No masking
The registry identifies the study as unmasked. Masking therefore does not provide a mechanism for reducing knowledge of assignment in the study design.
Diagnostic purpose
The registry identifies the primary purpose as diagnostic, which places the study in a different design context from a conventional randomized treatment-efficacy trial.
Three-year time frame
The primary endpoint is explicitly tied to 3 years, making the fixed-time survival estimate a central quantity of interest.
14. Statistical Interpretation vs Clinical Interpretation
Statistical interpretation
The registered primary endpoint is a time-to-event outcome assessed at a fixed 3-year time point. A Kaplan-Meier framework is a standard way to estimate such a survival probability while accounting for censoring.
Clinical interpretation
The endpoint describes disease-free status over the specified follow-up period in women receiving chemotherapy for stage III breast cancer. Clinical meaning depends on the endpoint definition, follow-up, patient population, and the broader study context.
These two interpretations should remain distinct. Statistical analysis describes how the observed time-to-event experience is estimated and how precisely it is measured. Clinical interpretation considers what that outcome means in the context of the disease and the study's diagnostic purpose.
15. Important Limitations and Interpretation Issues
- No randomized comparator: the registered single-group design does not support the same causal treatment-effect interpretation as a randomized controlled trial.
- Single-group survival estimate: disease-free survival can be estimated descriptively, but the estimate does not by itself identify what would have happened to the same participants under an alternative intervention.
- Censoring assumptions: survival estimation depends on appropriate handling of censoring and follow-up.
- Fixed-time endpoint: disease-free survival at 3 years provides information at a specified time point but does not fully describe the entire disease-free survival curve.
- No posted formal analyses: the registry record summarized here does not contain a posted statistical estimate, confidence interval, or p-value for the registered primary endpoint.
- Diagnostic context: the registry identifies the primary purpose as diagnostic, so conclusions about diagnostic procedures should not automatically be framed as randomized treatment-efficacy conclusions.
- Intervention labeling: the registry lists MRI/MRS as a procedure and gadopentetate dimeglumine as an intervention; the statistical interpretation should preserve the distinction between the registered intervention description and the clinical outcome.
16. Why This Trial Matters Statistically
I-SPY 1 is useful as a statistical teaching case because it demonstrates that a clinical trial does not need a randomized treatment comparison to require careful survival analysis. Its registered primary endpoint is disease-free three-year survival, making the key methodological issues those surrounding time-to-event estimation, censoring, fixed-time survival probabilities, and interpretation of a single-group design.
| Concept | How it appears in I-SPY 1 |
|---|---|
| Time-to-event analysis | Disease-free three-year survival is the registered primary endpoint. |
| Fixed-time survival | The primary endpoint is assessed at 3 years. |
| Kaplan-Meier estimation | A standard framework for estimating disease-free survival while accommodating censoring. |
| Censoring | Participants without a complete observed event time can contribute information through their available follow-up. |
| Single-group design | The study has one registered arm and no randomized comparator. |
| Diagnostic purpose | The registry identifies the primary purpose as diagnostic. |
| Statistical uncertainty | A survival estimate should ordinarily be accompanied by an appropriate confidence interval. |
| Causal interpretation | Without randomization or an external comparison, an observed single-group survival estimate does not establish a randomized treatment effect. |
17. A Practical Framework for Reading the Primary Endpoint
The primary target is disease-free survival at 3 years: the probability of remaining disease-free through the prespecified time point.
Use the observed timing of disease-free events and censoring rather than treating all participants as if they had identical follow-up.
A Kaplan-Meier estimator can describe the disease-free survival curve and provide an estimate at 3 years.
Report a confidence interval appropriate to the survival estimate so that the precision of the result is visible.
Interpret the result as a single-group outcome estimate rather than as a randomized treatment comparison.
18. How to Interpret a Future Posted Estimate
If a formal three-year disease-free survival estimate were reported, the most informative presentation would combine the point estimate, its confidence interval, and the underlying survival-analysis framework.
| Reported quantity | Question it answers |
|---|---|
| Three-year disease-free survival estimate | What proportion of the population is estimated to remain disease-free through 3 years? |
| Confidence interval | How precisely has the population survival probability been estimated? |
| Kaplan-Meier curve | How does estimated disease-free survival evolve over the observed follow-up? |
| Number of events | How much event information contributes to the survival estimate? |
| Number censored | How much follow-up ends without an observed disease-free event? |
A p-value is not the primary language for describing a single-group survival probability unless a specific null hypothesis and reference value have been prespecified. The most direct descriptive question is the estimated disease-free survival probability at 3 years and its uncertainty.
19. Statistical Questions This Study Illustrates
What is the estimand?
The registered estimand is disease-free survival at 3 years rather than a treatment-group difference.
What is the analysis population?
The registry identifies enrollment of 356 participants and one study arm; a more detailed analysis-population definition is not reported here.
What is the comparator?
There is no randomized comparator arm in the registered design.
What does censoring contribute?
Censored observations retain information about the period during which participants were observed to remain disease-free.
20. Limitations of Single-Group Survival Inference
Single-group survival analysis can provide a useful description of outcomes, but its inferential scope is narrower than that of a randomized comparison. If the scientific question is whether an intervention changes disease-free survival, a comparator is needed to establish the counterfactual outcome against which the observed experience can be evaluated.
This does not make the survival endpoint statistically uninformative. Rather, it changes the question that can be answered directly. The study can characterize disease-free survival in its enrolled population; it cannot, from the single-group structure alone, separate the effect of MRI/MRS from prognosis, chemotherapy, disease characteristics, or other factors that influence disease-free survival.
21. Trial Timeline
Study start
The registered study begins in May 2002.
Enrollment and study conduct
The study enrolls 356 participants in a single-group design.
Primary completion
The registered primary completion date is May 2005.
Registry status
The ClinicalTrials.gov record identifies the study status as completed.
22. Sources
- ClinicalTrials.gov: I-SPY 1 (NCT00043017), the official registry record.
- PubMed: PubMed record.
- PubMed: PubMed record.
- PubMed: PubMed record.
- PubMed: PubMed record.
23. Record Summary
I-SPY 1 is registered as a completed phase 2, single-group diagnostic study enrolling 356 participants. Its primary endpoint is disease-free three-year survival, with a registered time frame of 3 years. The study's single-group structure is central to statistical interpretation: disease-free survival can be estimated using standard time-to-event methods, but the design does not provide a randomized treatment comparison.
For this endpoint, Kaplan-Meier estimation provides the natural framework for describing disease-free survival over time and obtaining a fixed-time estimate at 3 years. Proper interpretation also requires attention to censoring, follow-up, statistical uncertainty, and the distinction between an observed survival experience and a causal treatment effect.