This page separates registry-reported trial information from statistical interpretation. The ClinicalTrials.gov record contains posted outcome measures but no formal statistical-analysis results.
1. Trial at a Glance
FORWARD I was a randomized, parallel-group, open-label phase 3 oncology trial with 366 enrolled participants. It compared mirvetuximab soravtansine with investigator's choice of chemotherapy in women with folate receptor alpha-positive advanced epithelial ovarian cancer, primary peritoneal carcinoma, or fallopian tube cancer.
| Feature | FORWARD I |
|---|---|
| Trial name | FORWARD I |
| Phase | Phase 3 |
| Condition | Epithelial ovarian cancer; primary peritoneal carcinoma; fallopian tube cancer; ovarian cancer |
| Population | Women with folate receptor (FR) alpha-positive advanced epithelial ovarian cancer (EOC), primary peritoneal or fallopian tube cancer |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 366.0 |
| Primary endpoints | 2 |
| Endpoint types | Binary; Time-to-event |
| Outcome measures posted | 11 |
| Results posted | Yes |
| Status | Completed |
| Start | 2016-03-02 |
| Primary completion | 2019-01 |
| Lead sponsor | ImmunoGen, Inc. |
| Sponsor type | Industry |
2. Clinical Question
The registered study asks a comparative treatment question: in women with FR alpha-positive advanced epithelial ovarian cancer, primary peritoneal cancer, or fallopian tube cancer, how does mirvetuximab soravtansine arm and 62.9 weeks for IC chemotherapy arm) compare with investigator's choice of chemotherapy on the registered primary endpoints?
Population
Women with folate receptor alpha-positive advanced epithelial ovarian cancer, primary peritoneal cancer, or fallopian tube cancer.
Intervention
Mirvetuximab soravtansine.
Comparator
Investigator's choice of chemotherapy. The registered interventions identify paclitaxel, pegylated liposomal doxorubicin, and topotecan as chemotherapy drugs.
Primary question
How do the randomized treatment groups compare on progression-free survival as assessed by BIRC according to RECIST Version 1.1, including the prespecified population with high folate receptor alpha level?
3. Trial Design
Mirvetuximab soravtansine arm and 62.9 weeks for IC chemotherapy arm)
- Study drug: mirvetuximab soravtansine arm and 62.9 weeks for IC chemotherapy arm)
- Randomized treatment arm
- Serious adverse events: 67/243
Investigator's choice chemotherapy
- Comparator arm designated as investigator's choice of chemotherapy
- Registered chemotherapy interventions include paclitaxel
- Registered chemotherapy interventions include pegylated liposomal doxorubicin
- Registered chemotherapy interventions include topotecan
- Serious adverse events: 31/109
4. Trial Timeline
Trial start
The registered study start date was March 2, 2016.
Primary completion
The registered primary completion date was January 2019.
Registry status
The trial is listed as completed, with results posted and 11 outcome measures reported in the registry.
5. Primary Endpoints
| Registered primary endpoint | Time frame | Endpoint type |
|---|---|---|
| Progression-Free Survival (PFS), as Assessed by BIRC Per RECIST Version 1.1 in All Participants Randomized to the Study | From the date of randomization until the time of death or PD (maximum exposure: 86.9 weeks for mirvetuximab soravtansine arm and 62.9 weeks for IC chemotherapy arm) | Time-to-event |
| PFS, as Assessed by BIRC Per RECIST Version 1.1 in Participants With High Folate Receptor Alpha Level (≥ 75% of Tumor Staining) | From the date of randomization until the time of death or PD (maximum exposure: 86.9 weeks for mirvetuximab soravtansine arm and 62.9 weeks for IC chemotherapy arm) | Binary |
Endpoint definition in the registry
The registry defines PFS as the time from randomization until PD or death, whichever occurred first, estimated using the Kaplan-Meier method. The registry's registry-reported definition describes progressive disease using RECIST Version 1.1, including at least a 20% increase in the sum of the longest diameters of target lesions from the nadir, together with an absolute increase of at least 5 mm, with additional RECIST progression criteria.
6. Planned Analysis
The ClinicalTrials.gov record contains posted outcome measures, but no formal statistical analyses. Consequently, there are no treatment-effect estimates, confidence intervals, or p-values in the ClinicalTrials.gov record that can be attributed to the formal statistical comparison of either primary endpoint.
For the overall PFS endpoint, the registry explicitly states that PFS is estimated using the Kaplan-Meier method. This is appropriate for a time-to-event endpoint because participants may be followed for different lengths of time and may be censored before experiencing progression or death.
A conventional randomized analysis of this type would compare the time-to-event distributions between treatment groups and estimate a relative treatment effect, commonly with a hazard ratio and confidence interval. However, because no such statistical analysis is posted on ClinicalTrials.gov for FORWARD I, this page does not assign a hazard ratio, confidence interval, or p-value to the trial.
The high-FR-alpha endpoint is registered separately for participants with ≥ 75% of tumor staining. Although the registry classifies this endpoint as binary in the ClinicalTrials.gov record, its wording still describes PFS as time from randomization until death or progression and states that Kaplan-Meier estimation is used. The distinction is important: the registry's endpoint classification and its substantive definition should both be preserved rather than silently replacing one with the other.
7. Statistical Methodology
Randomization
Randomization is the central design feature for the comparative question. By assigning participants to treatment groups rather than allowing treatment choice to determine assignment, randomization provides the framework for comparing outcomes while reducing systematic differences in measured and unmeasured baseline characteristics in expectation.
For an analysis of an endpoint measured after randomization, the randomized groups remain the primary comparison groups. This is especially important for time-to-event outcomes because follow-up, discontinuation, progression, and death can occur at different times in different participants.
Kaplan-Meier estimation
The registry specifically states that PFS was estimated using the Kaplan-Meier method. Kaplan-Meier estimation uses the observed event times and accounts for right censoring, allowing participants who have not experienced progression or death by the end of their observable follow-up to contribute information without treating them as if they experienced the event.
Here, di represents the number of events at event time ti, while ni represents the number at risk immediately before that time.
Time-to-event analysis
PFS is fundamentally a time-to-event endpoint because the outcome incorporates when progression or death occurs, not simply whether a participant eventually experiences one of those events. That distinction determines the statistical methods that are appropriate.
A participant who remains alive and progression-free through the end of observed follow-up does not ordinarily count as a simple "non-event" in the same way as a participant with complete long-term observation. Instead, the participant is censored at the relevant follow-up time, provided the censoring rules are met.
RECIST Version 1.1 assessment
The registered primary endpoint specifies assessment by a blinded independent radiology review committee (BIRC) according to RECIST Version 1.1. The purpose of an independent radiologic assessment is to provide a standardized assessment framework for tumor progression rather than relying solely on an investigator's subjective judgment.
The registry definition identifies progression using changes in the sum of the longest diameters of target lesions and includes both a relative and absolute increase criterion. This illustrates why endpoint definitions matter statistically: changing the progression rule changes which observations are classified as events and therefore changes the resulting PFS distribution.
Binary versus time-to-event classification
The ClinicalTrials.gov record classifies the two primary endpoints differently: the overall randomized PFS endpoint is identified as time-to-event, while the high-FR-alpha PFS endpoint is identified as binary. Yet the high-FR-alpha endpoint's registry-reported definition also describes time from randomization until death or progression and states that Kaplan-Meier estimation is used.
This is a useful reminder that an endpoint's registry classification should be checked against its actual definition and analysis method. A label such as "binary" does not by itself justify replacing a time-to-event analysis with a simple proportion comparison.
Independent assessment and open-label design
The trial was unmasked, meaning the registered masking field is "NONE." Because treatment assignment was not masked, knowledge of treatment could potentially affect aspects of clinical management or investigator assessment. The use of BIRC assessment for the primary PFS endpoint provides an independent radiologic assessment framework that is relevant to this concern.
8. Statistical Methods Explained
Why was randomization important?
Randomization establishes the treatment comparison before outcomes occur. In a properly conducted randomized trial, differences observed after randomization can be interpreted within a causal framework because treatment assignment, rather than prognosis or treatment preference, determines the assigned group.
Why use Kaplan-Meier analysis for PFS?
PFS incorporates the time until progression or death, and participants may have different amounts of observable follow-up. Kaplan-Meier estimation handles this right-censoring structure while retaining the timing information contained in the observed events.
What does "progression or death, whichever occurred first" mean?
It defines a composite time-to-event endpoint. The PFS clock begins at randomization and ends at the first qualifying progression or death. A participant does not need to experience both events for the endpoint to occur.
Why does RECIST Version 1.1 matter?
RECIST supplies prespecified rules for determining whether imaging findings constitute progression or response. Standardized rules reduce ambiguity in event classification. Because PFS depends on when progression is declared, the progression definition directly affects the statistical endpoint.
What is the role of a blinded independent review committee?
A BIRC provides an independent assessment of radiologic outcomes. In an open-label trial, independent review is particularly relevant because investigators know treatment assignment. The BIRC does not change the randomized allocation; it supplies an additional standardized assessment of the imaging-based endpoint.
A hazard ratio requires event and follow-up information together with a specified statistical model or comparison procedure. A p-value requires a defined hypothesis test and its corresponding sampling distribution or testing framework. Enrollment counts and serious-adverse-event counts cannot be transformed into a valid PFS treatment-effect estimate.
9. High Folate Receptor Alpha Endpoint
The second registered primary endpoint focuses on participants with high folate receptor alpha level (≥ 75% of tumor staining). This creates a prespecified population within the broader randomized trial for a separate primary endpoint.
| Feature | High-FR-alpha primary endpoint |
|---|---|
| Population definition | Participants with high folate receptor alpha level (≥ 75% of tumor staining) |
| Endpoint | PFS, as assessed by BIRC per RECIST Version 1.1 |
| Time frame | From the date of randomization until the time of death or PD |
| Maximum exposure stated | 86.9 weeks for mirvetuximab soravtansine arm and 62.9 weeks for IC chemotherapy arm) |
| Registry classification | Binary |
| Posted statistical analysis reported | None |
Statistically, this endpoint raises an important population-definition issue. The analysis population is narrower than the full set of randomized participants, so its precision depends on how many randomized participants satisfy the ≥ 75% tumor-staining criterion and how many progression or death events occur within that population. Those counts are not reported in the ClinicalTrials.gov record.
The high-FR-alpha endpoint should be treated as a distinct prespecified analysis population rather than as an informal subgroup selected after seeing the results. The registry identifies the threshold explicitly as ≥ 75% of tumor staining.
Because no formal analysis record is reported, there is no basis here for concluding how the randomized treatment groups compared within this population.
10. Safety Results
The ClinicalTrials.gov record reports serious adverse events by treatment arm. The denominator is explicitly described as the number at risk for the serious-adverse-event summary.
| Safety measure | Mirvetuximab soravtansine arm and 62.9 weeks for IC chemotherapy arm) | Investigator's choice chemotherapy |
|---|---|---|
| Serious adverse events | 67/243 | 31/109 |
The counts show the number affected and the number at risk, but the ClinicalTrials.gov record does not provide a formal statistical comparison of serious adverse events. The visual proportions above are descriptive calculations of the registry-reported counts and should not be interpreted as a reported inferential result.
11. Endpoint Types and Their Statistical Implications
| Endpoint characteristic | Statistical implication |
|---|---|
| PFS | Requires methods that account for event timing and censoring. |
| Death or progression | Defines a composite event in which the first qualifying event determines PFS. |
| RECIST Version 1.1 | Provides prespecified rules for classifying radiologic progression. |
| BIRC assessment | Provides an independent radiologic assessment framework. |
| High FR-alpha population | Creates a separate prespecified analysis population defined by ≥ 75% tumor staining. |
| Binary registry classification | Requires reconciliation with the registry-reported time-to-event definition and Kaplan-Meier statement. |
The statistical structure is therefore more informative than simply labeling FORWARD I as a "positive" or "negative" trial. The design contains two registered primary endpoints, but the ClinicalTrials.gov record does not provide the numerical comparisons needed to characterize their observed treatment effects.
12. What a Formal PFS Comparison Would Normally Report
For an endpoint such as PFS, a complete statistical result would normally communicate several complementary pieces of information rather than relying on a single number.
Kaplan-Meier estimate
Shows the estimated probability of remaining progression-free over time and incorporates censoring.
Hazard ratio
Provides a model-based relative comparison of the instantaneous event rate between randomized treatment groups when a suitable time-to-event model is used.
Confidence interval
Quantifies statistical uncertainty around an estimated treatment effect under the specified analysis framework.
P-value
Addresses evidence against a prespecified null hypothesis under the chosen statistical test; it is not a measure of effect size.
The absence of a statistical-analysis entry is therefore itself an important property of this record.
13. How to Interpret a Hypothetical Hazard Ratio
If a time-to-event analysis reports a hazard ratio below 1 for mirvetuximab soravtansine versus investigator's choice chemotherapy, that would indicate a lower estimated instantaneous rate of progression or death in the mirvetuximab group under the fitted model.
It would not mean that the same percentage of participants avoided progression, nor would it directly state the difference in median PFS or the difference in absolute PFS probability at a particular time.
A confidence interval would describe statistical uncertainty around the estimated hazard ratio. A narrow interval would indicate greater precision than a wide interval, all else equal. It would not describe the range of individual patient outcomes.
A p-value would quantify the evidence against a specified null hypothesis under the chosen testing framework. It would not tell the reader how large or clinically important the treatment effect is. Effect estimates and uncertainty intervals remain necessary for understanding magnitude and precision.
14. Censoring and PFS Interpretation
Because PFS is measured from randomization until progression or death, not every participant necessarily contributes an observed event. Participants can remain progression-free at their last assessment and therefore contribute censored follow-up to a Kaplan-Meier analysis.
This creates an important distinction between not having observed progression or death and being known never to experience progression or death. Kaplan-Meier methodology uses the former observation without treating it as the latter.
The first qualifying event determines the endpoint. If neither event is observed during the relevant follow-up, the participant may contribute censored information according to the prespecified analysis rules.
The quality of a PFS estimate therefore depends not only on the number of participants enrolled, but also on the number and timing of progression and death events, the follow-up process, and the prespecified censoring rules.
15. Open-Label Design and Independent Review
FORWARD I is registered with masking: none. In an open-label treatment comparison, participants and treating investigators know which intervention has been assigned.
The primary PFS endpoints are nevertheless specified as being assessed by a blinded independent radiology review committee according to RECIST Version 1.1. That distinction matters because knowledge of treatment assignment and independent radiologic review address different aspects of potential bias.
| Feature | FORWARD I | Statistical relevance |
|---|---|---|
| Masking | None | Treatment assignment was not masked. |
| Primary PFS assessment | BIRC | Independent radiologic assessment is specified. |
| Assessment framework | RECIST Version 1.1 | Provides standardized progression criteria. |
| Allocation | Randomized | Provides the principal framework for treatment-group comparison. |
16. Multiplicity and Primary Endpoints
The registry identifies 2 primary endpoints. Both concern PFS, but the second is restricted to participants with high folate receptor alpha level, defined as ≥ 75% of tumor staining.
When a study has more than one primary endpoint, the statistical interpretation depends on the prespecified testing strategy. Formal control of type I error may require a specified hierarchy, alpha allocation, or another multiplicity procedure, depending on the trial's statistical analysis plan.
The ClinicalTrials.gov record does not provide such a multiplicity strategy. Therefore, this page does not infer one and does not treat either endpoint as independently significant or nonsignificant.
| Primary endpoint | Population | Multiplicity information reported |
|---|---|---|
| Overall PFS by BIRC per RECIST Version 1.1 | All participants randomized | Not reported |
| PFS by BIRC per RECIST Version 1.1 | High FR-alpha level (≥ 75% tumor staining) | Not reported |
17. Analysis Population Considerations
The overall primary endpoint explicitly applies to all participants randomized to the study. That wording is important because it anchors the principal efficacy population to randomization rather than to treatment completion or another post-randomization subset.
The second primary endpoint is defined within a narrower population based on tumor staining. The statistical interpretation of that endpoint therefore depends on the number of randomized participants meeting the high-FR-alpha criterion and on the event and censoring information within that population. Those quantities are not contained in the ClinicalTrials.gov record.
18. Missing Data, Imputation, and Follow-Up
The ClinicalTrials.gov record does not specify an imputation strategy for missing measurements, a detailed censoring algorithm, or rules for handling missing radiologic assessments. Those details can materially affect a PFS analysis because progression status is established from longitudinal assessments.
For a time-to-event endpoint, ordinary missing-data language also needs to be distinguished from censoring. A participant who has not experienced progression or death by the last usable assessment is handled through the prespecified time-to-event and censoring framework; this is not equivalent to simply filling in a missing outcome with an imputed value.
Because the ClinicalTrials.gov record does not contain a detailed missing-data or censoring plan, no specific imputation method is attributed to FORWARD I here.
19. Non-Inferiority, Factorial, and Bayesian Methods
| Design feature | Supported by the ClinicalTrials.gov record? |
|---|---|
| Non-inferiority margin | Not reported. |
| Factorial design | Not reported; the registered design model is parallel. |
| Bayesian analysis | Not reported. |
| Interim-analysis method | Not reported. |
| Stratification factors | Not reported. |
| Multiplicity procedure | Not reported. |
This distinction is deliberate. The presence of two primary endpoints does not establish a particular multiplicity procedure, and a phase 3 randomized design does not by itself establish a specific interim-monitoring or Bayesian framework. Such features should be taken from the protocol or statistical analysis plan rather than inferred from the trial's phase.
20. Serious Adverse Events: Statistical Context
The reported serious-adverse-event counts provide a useful safety summary, but they illustrate why denominators matter. The registry-reported values are 67/243 for mirvetuximab soravtansine and 31/109 for investigator's choice chemotherapy.
| Arm | Affected | At risk | Descriptive affected / at-risk proportion |
|---|---|---|---|
| Mirvetuximab soravtansine arm and 62.9 weeks for IC chemotherapy arm) | 67 | 243 | 67/243 |
| Investigator's choice chemotherapy | 31 | 109 | 31/109 |
The final column deliberately retains the registry-reported fraction rather than converting it into a newly reported percentage. A formal safety comparison would require the prespecified definition of the safety population, event ascertainment rules, follow-up period, and statistical method.
21. Limitations
- No statistical-analysis records in the ClinicalTrials.gov record: although results are posted, the registry record contains no formal statistical analyses for the primary endpoints.
- No treatment-effect estimates reported: hazard ratios, confidence intervals, and p-values cannot be reconstructed from enrollment and safety counts alone.
- Open-label design: masking is listed as none, so treatment assignment was known during the trial.
- Biomarker-defined primary population: the high-FR-alpha endpoint applies to participants with ≥ 75% tumor staining, but the size of that population is not reported here.
- Missing analysis-plan details: the ClinicalTrials.gov record does not specify interim-analysis procedures, multiplicity adjustment, stratification factors, or imputation methods.
- Safety interpretation: serious-adverse-event counts are reported, but no formal comparative safety analysis is reported.
- Endpoint classification: the high-FR-alpha PFS endpoint is classified as binary in the ClinicalTrials.gov record even though its definition describes a time-to-event measure estimated using Kaplan-Meier methodology.
22. Why This Trial Matters Statistically
FORWARD I is a useful teaching example because it demonstrates how the statistical structure of a clinical trial can be analyzed even when a registry record does not provide the full numerical treatment comparison.
| Concept | How it appears in FORWARD I |
|---|---|
| Randomization | The study uses randomized allocation to 2 treatment arms. |
| Parallel design | The registered design model is parallel. |
| Open-label design | Masking is listed as none. |
| Time-to-event analysis | The primary overall PFS endpoint is classified as time-to-event. |
| Kaplan-Meier estimation | The registry states that PFS was estimated using Kaplan-Meier methodology. |
| RECIST Version 1.1 | Progression assessment is specified according to RECIST Version 1.1. |
| Independent review | The primary PFS endpoints are assessed by BIRC. |
| Biomarker-defined population | A second primary endpoint is restricted to high FR-alpha level (≥ 75% tumor staining). |
| Multiple primary endpoints | The trial has 2 registered primary endpoints. |
| Safety denominators | Serious-adverse-event summaries are reported as affected / at-risk counts. |
The broader lesson is that statistical interpretation should begin with the estimand and endpoint definition, proceed through the randomized design and analysis population, and only then interpret numerical estimates.
23. Related Tutorials
Learn more about the methods used in this trial:
24. Related Statistical Calculators
25. Sources
- ClinicalTrials.gov: FORWARD I, NCT02631876.
- Linked publication: PubMed record for PMID 33667670.
Continue through the Clinical Biostats statistical pathway
Use the trial's endpoint structure to explore randomized-trial design, survival analysis, Kaplan-Meier estimation, and related statistical methods.
26. Record Summary
FORWARD I is a randomized, parallel-group phase 3 trial with 366 participants comparing mirvetuximab soravtansine with investigator's choice of chemotherapy. Its two registered primary endpoints concern progression-free survival assessed by a blinded independent radiology review committee according to RECIST Version 1.1: one in all randomized participants and one in participants with high folate receptor alpha level (≥ 75% of tumor staining).
The registry states that PFS was defined as time from randomization until progression or death, whichever occurred first, and that Kaplan-Meier methodology was used. The ClinicalTrials.gov record also reports serious adverse events as 67/243 for mirvetuximab soravtansine and 31/109 for investigator's choice chemotherapy.
The key statistical limitation is that no formal statistical analyses were posted.