This page separates information reported in the ClinicalTrials.gov record from statistical interpretation. The registry identifies results as posted, but the ClinicalTrials.gov record contains no formal statistical analysis records. Accordingly, numerical treatment-effect estimates, confidence intervals, and p-values are not reported here.
1. Trial at a Glance
STORM was a completed phase 2 oncology trial in multiple myeloma evaluating selinexor and dexamethasone. The ClinicalTrials.gov record describes a non-randomized, parallel design with no masking and a treatment purpose, enrolling 202 participants across two parts.
| Feature | STORM |
|---|---|
| Trial name | STORM |
| NCT identifier | NCT02336815 |
| Brief title | Selinexor Treatment of Refractory Myeloma |
| Condition | Multiple Myeloma |
| Phase | Phase 2 |
| Status | COMPLETED |
| Allocation | NON_RANDOMIZED |
| Design model | PARALLEL |
| Masking | NONE |
| Primary purpose | TREATMENT |
| Enrollment | 202 |
| Interventions | Selinexor; Dexamethasone |
| Lead sponsor | Karyopharm Therapeutics Inc |
| Sponsor type | INDUSTRY |
2. Clinical Question
The registered primary question centers on the proportion of participants achieving an overall response according to International Myeloma Working Group criteria, with assessment performed by an Independent Review Committee.
Population
Participants enrolled in the STORM phase 2 trial for multiple myeloma.
Intervention
The ClinicalTrials.gov record identifies selinexor and dexamethasone as the study interventions.
Comparator
The trial is registered as non-randomized. The ClinicalTrials.gov record does not identify a randomized comparator regimen.
Primary question
What percentage of participants experienced an overall response according to the registered IMWG-based definition and independent review?
3. Trial Design
The distinction between a randomized and non-randomized design is central to interpreting STORM statistically. Randomization creates a mechanism for balancing both measured and unmeasured prognostic factors in expectation. Because this registry identifies STORM as non-randomized, an observed response proportion is principally a description of the enrolled population rather than a randomized estimate of a treatment effect against a concurrent control group.
4. Study Parts and Safety Population
The ClinicalTrials.gov record reports two parts and provide serious adverse-event counts separately for each part. They do not provide the treatment assignment structure or intervention-specific enrollment counts for the two parts, so the safety figures should be retained in their registry form rather than re-labeled as randomized treatment-arm comparisons.
| Study component | Serious adverse events | At risk |
|---|---|---|
| Part 1 | 45 | 79 |
| Part 2 | 78 | 123 |
5. Primary Endpoint
| Endpoint | Registered definition | Endpoint type |
|---|---|---|
| Part 2: Percentage of Participants With Overall Response Rate (ORR) Per International Myeloma Working Group (IMWG) as Assessed by an Independent Review Committee (IRC) | IRC assessed ORR per 2016 IMWG criteria. The registry definition states that participants were classified according to response categories including Partial Response (PR), defined as greater than or equal to 50% reduction of serum M-Protein and reduction in 24-hour urinary M-protein by greater than or equal to 90% or to less than 200 mg per 24 hours, with additional response definitions continuing in the registry record. | Binary |
The registered time frame is Baseline until disease progression/discontinuation from the study, or death, whichever occurred first, with the registry text ending before the maximum duration is specified.
6. Results
Results are posted for the primary endpoint in the ClinicalTrials.gov record. However, no formal statistical analyses were posted. No numerical estimate, confidence interval, p-value, or formal comparative statistical analysis for the primary endpoint is provided in the ClinicalTrials.gov record.
Primary endpoint: IRC-assessed ORR
Formal statistical analysis records reported: none
No ORR estimate, confidence interval, or p-value is included in the ClinicalTrials.gov record.
The primary endpoint is a binary response outcome: each participant is classified according to whether the registered response criteria were met. The natural descriptive quantity is therefore the proportion of participants with an overall response.
Because STORM is registered as non-randomized, an observed ORR would not by itself establish that selinexor and dexamethasone caused the observed response rate. Without a randomized comparator, differences from historical response experience can be affected by patient selection, eligibility criteria, disease characteristics, assessment procedures, and other differences between populations.
A confidence interval around an ORR would quantify statistical precision around the estimated response proportion. It would not describe the range of responses that an individual participant could experience, and it would not by itself remove the design limitations created by non-randomized allocation.
A p-value, if one were reported against a prespecified reference value, would quantify evidence against that statistical null hypothesis under the specified analysis. It would not measure the size or clinical importance of the response rate.
7. What Analysis Would Normally Be Used for a Binary ORR Endpoint?
For a binary endpoint such as ORR, the first statistical quantity is the observed response proportion:
A confidence interval for p can then be calculated using an appropriate binomial method. If a prespecified benchmark response rate exists, a one-sample hypothesis test can compare the observed response proportion with that benchmark.
For a randomized two-group trial, investigators could compare response proportions between treatment groups using a method appropriate to the analysis plan, such as a test of two proportions or a regression model for a binary outcome. That framework is different from STORM's registered non-randomized design.
What an ORR estimate tells you
It describes the proportion of participants meeting the prespecified response criteria within the defined assessment framework.
What an ORR estimate does not tell you
It does not by itself establish causality, durability of benefit, or comparative treatment efficacy.
What a confidence interval tells you
It describes statistical uncertainty around an estimated response proportion under the selected interval method.
What a p-value tells you
It quantifies evidence against a specified null hypothesis; it is not a measure of effect size.
8. Statistical Methodology
Binary endpoint analysis
The primary endpoint is registered as binary. Binary outcomes reduce each participant's endpoint status to one of two outcome categories for the purpose of the primary analysis. For ORR, those categories are response versus no response under the prespecified response-assessment rules.
Independent review
The registered endpoint specifies assessment by an Independent Review Committee (IRC). Independent assessment can reduce the possibility that investigators' knowledge of treatment or other clinical information influences response classification. The ClinicalTrials.gov record does not specify the statistical adjudication procedure used by the IRC.
Time-to-event follow-up in the endpoint definition
Although the primary endpoint itself is binary, the registry defines its assessment window from baseline until disease progression, discontinuation, or death, whichever occurs first. This makes the timing of assessment relevant to determining whether a participant has an opportunity to meet the response criteria.
Non-randomized analysis
Because allocation is registered as non-randomized, a simple response proportion is primarily descriptive. Statistical adjustment can sometimes be used in non-randomized research, but the ClinicalTrials.gov record does not identify an adjustment model, covariates, propensity method, or other causal-inference procedure.
An observed response proportion describes outcomes among the participants studied. A causal treatment-effect estimate requires additional design or modeling assumptions, particularly when there is no randomized concurrent comparator.
9. Statistical Methods Explained
Why is ORR a binary endpoint?
Overall response rate is based on whether an individual participant meets the prespecified response criteria. The final endpoint therefore has a yes/no structure even though the underlying clinical assessments used to determine response can involve multiple measurements.
Why use an Independent Review Committee?
The registry specifies IRC assessment for the primary endpoint. Independent review provides a standardized assessment mechanism separate from the treating investigator's assessment. This is particularly relevant when tumor response classification determines the endpoint.
How would an ORR normally be summarized?
The basic summary is the number of responders divided by the number of participants included in the relevant analysis population. The result is expressed as a proportion or percentage, accompanied by an appropriate confidence interval when statistical precision is being reported.
Why does non-randomization matter?
Without random assignment, treatment exposure can be associated with baseline characteristics or clinical-selection factors. Consequently, an observed response rate cannot automatically be interpreted as the response rate that would have occurred under a different treatment in the same participants.
What does a confidence interval around ORR mean?
A confidence interval quantifies uncertainty arising from sampling and the selected statistical model or interval procedure. A narrower interval generally indicates greater statistical precision than a wider interval, but neither interval width nor point estimate addresses systematic bias caused by non-randomized treatment allocation.
Why is the assessment time frame important?
The registered endpoint is assessed from baseline until progression, discontinuation, or death. Participants therefore have an assessment history that can end for different reasons. The definition of the analysis population and rules for participants without an evaluable response are important components of any final ORR analysis, but those rules are not reported in the available statistical-analysis data.
10. Analysis Population and Missing Data
The ClinicalTrials.gov record does not provide a statistical-analysis record specifying the primary analysis population, handling of participants who discontinue before response assessment, treatment of unevaluable participants, or missing-data rules.
| Analysis issue | What the ClinicalTrials.gov record establishes | Statistical implication |
|---|---|---|
| Primary endpoint | Binary IRC-assessed ORR | Response proportions are the natural descriptive measure |
| Allocation | Non-randomized | Observed response cannot automatically be interpreted causally |
| Missing response assessment | Not specified in the ClinicalTrials.gov record | Denominator and missing-data handling can affect ORR estimation |
| Analysis population | Not specified in the ClinicalTrials.gov record | Exact responder proportion cannot be reconstructed from the ClinicalTrials.gov record |
| Imputation | Not specified | No imputation method should be inferred |
11. Multiplicity, Interim Analysis, and Bayesian Methods
The ClinicalTrials.gov record does not report a multiplicity-adjustment procedure, alpha-spending strategy, interim-analysis framework, or Bayesian analysis method.
| Design topic | Information in the ClinicalTrials.gov record |
|---|---|
| Multiplicity | No procedure reported |
| Interim analysis | No statistical interim-analysis method reported |
| Alpha spending | Not reported |
| Bayesian methods | Not reported |
| Non-inferiority margin | Not applicable to the registry-reported primary endpoint description |
| Crossover | Not reported |
| Factorial design | Not reported |
| Missing-data imputation | Not reported |
These omissions should not be filled with assumptions. In a statistical review, an unreported method is different from a method that was not used.
12. Safety Results
The ClinicalTrials.gov record provides serious adverse-event counts separately for Part 1 and Part 2. They do not provide a statistical comparison of serious adverse events between interventions.
| Study part | Affected by serious adverse events | At risk |
|---|---|---|
| Part 1 | 45 | 79 |
| Part 2 | 78 | 123 |
These figures indicate the number of participants affected by serious adverse events and the corresponding number at risk as reported in the ClinicalTrials.gov record. Because the registry information does not identify these counts as randomized treatment-group comparisons, they should not be interpreted as evidence that one intervention has a higher or lower serious-adverse-event rate than another.
A safety comparison would normally begin with event counts and denominators for clearly defined analysis populations. Depending on the design and endpoint definition, investigators could report risks, risk differences, relative risks, odds ratios, incidence rates, or time-to-first-event analyses. None of those formal comparative estimates is included in the ClinicalTrials.gov record.
13. Trial Timeline
Trial start
The registry identifies May 26, 2015 as the study start date.
Primary completion
The registry identifies July 26, 2019 as the primary completion date.
Current registry status in the ClinicalTrials.gov record
The trial is listed as COMPLETED, with 202 participants enrolled.
14. What the Primary Endpoint Does — and Does Not — Mean
The registered ORR endpoint is the percentage of participants meeting the specified IMWG response criteria. It is therefore a measure of how frequently the prespecified response state was observed within the analyzed population.
ORR does not directly measure how long participants remain alive or free of progression. A response endpoint and a time-to-event endpoint answer different clinical questions and should not be treated as interchangeable.
Because the registry design is non-randomized, the observed response proportion cannot by itself establish the causal effect of selinexor or dexamethasone compared with another treatment.
A very precise response estimate can still be affected by selection bias or other systematic limitations. Confidence intervals address statistical uncertainty; they do not automatically correct for non-randomized treatment allocation.
15. Limitations
- Non-randomized allocation: the registry identifies the trial as NON-RANDOMIZED, so observed response outcomes cannot automatically be interpreted as randomized treatment effects.
- No concurrent randomized comparator: the ClinicalTrials.gov record does not identify a randomized control regimen against which ORR was formally compared.
- Incomplete statistical-analysis record: no formal statistical analyses were posted despite results being posted.
- No numerical primary result in the ClinicalTrials.gov record: an ORR estimate, confidence interval, and p-value are not provided in the ClinicalTrials.gov record.
- Incomplete endpoint-definition extraction: the registry-reported IMWG definition ends before the complete response-category description is reproduced.
- Analysis population not specified: the ClinicalTrials.gov record does not identify the exact denominator used for the primary ORR analysis.
- Missing-data rules not specified: the ClinicalTrials.gov record does not describe how participants without an evaluable response were handled.
- Safety comparison not specified: serious adverse events are reported by study part, but no formal comparative safety analysis is reported.
- Limited design information: the ClinicalTrials.gov record does not report stratification factors, multiplicity procedures, interim-analysis methods, or Bayesian methods.
16. Why This Trial Matters Statistically
STORM is useful as a teaching example because it illustrates a statistical problem that differs from the more familiar randomized phase 3 setting: how to interpret a clinically important binary response endpoint when treatment allocation is non-randomized and the available registry record does not provide the underlying formal analysis.
| Concept | How it appears in STORM |
|---|---|
| Binary endpoint | Primary endpoint is IRC-assessed overall response rate. |
| Response criteria | ORR is defined according to 2016 IMWG criteria in the registry. |
| Independent review | The primary endpoint specifies assessment by an Independent Review Committee. |
| Non-randomized design | Allocation is registered as NON-RANDOMIZED. |
| Parallel design | The design model is registered as PARALLEL. |
| Statistical estimand | The primary registered quantity is the percentage of participants with overall response. |
| Confidence intervals | Would quantify precision around an estimated response proportion if reported. |
| Missing data | Denominator and handling rules are not provided in the ClinicalTrials.gov record. |
| Safety analysis | Serious adverse events are reported as affected/at-risk counts by study part. |
| Causal inference | Non-randomization limits direct causal interpretation of observed response rates. |
17. Related Tutorials
Learn more about the methods used to understand this trial:
18. Related Calculators
19. Sources
- ClinicalTrials.gov: NCT02336815 — STORM.
- PubMed: PMID 34488662.
- PubMed: PMID 34465265.
- PubMed: PMID 34404623.
- PubMed: PMID 31433920.
- PubMed: PMID 29381435.
Continue through the Clinical Biostats statistical library
Explore tutorials and calculators for binary endpoints, clinical-trial methodology, confidence intervals, and other statistical methods.
20. Record Summary
STORM is a completed phase 2, non-randomized, parallel oncology trial in multiple myeloma with 202 participants and two registered interventions: selinexor and dexamethasone. Its primary registered endpoint is the percentage of participants with an overall response according to 2016 IMWG criteria as assessed by an Independent Review Committee, measured from baseline until disease progression, discontinuation, or death, whichever occurred first.
The key statistical issue is the distinction between describing response and estimating a causal treatment effect. A binary ORR endpoint can be summarized with a response proportion and, when appropriate, a confidence interval. However, the non-randomized design limits causal interpretation, and the ClinicalTrials.gov record does not contain the formal statistical-analysis record needed to report an ORR estimate, confidence interval, or p-value.
The safety information reported with the record consists of serious-adverse-event counts of 45/79 for Part 1 and 78/123 for Part 2. Those figures should be interpreted as registry-reported affected/at-risk counts rather than as a randomized comparative safety analysis.