A Phase II Trial of Valproic Acid in Patients With Advanced Thyroid Cancers of Follicular Cell Origin: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Phase II Trial of Valproic Acid in Patients With Advanced Thyroid Cancers of Follicular Cell Origin (NCT01182285), a non-randomized phase 2 study. The primary outcome is RAI (Radioactive Iodine) Uptake and Tg (Thyroglobulin) Level Compared Pre and Post- Valproic Treatment.

2. Clinical Question

3. Trial Design

This was a phase II, non-randomized, unmasked, parallel-design study of valproic acid in advanced thyroid cancers of follicular origin. The registry lists thirteen enrolled participants and identifies the National Cancer Institute as the sponsor. The study is marked completed. Its recorded start date is 2010-09-24 and its primary completion date is 2015-09-28.

The brief summary describes participants continuing standard thyroid hormone suppression therapy while receiving valproic acid. The stated initial treatment period was ten weeks, with clinic visits, blood sampling, and participant diaries for doses and side effects. The primary outcome timeframe specifies assessment from study entry to after ten weeks for phase 1, and from ten to sixteen weeks of treatment for phase 2. The listed treatment arms include a phase 1 radioiodine-resistant group and two phase 2 schedules. Details in the arm descriptions are incomplete, so the full treatment paths cannot be reconstructed from the registry record.

The registry reports that eight of the thirteen phase 1 participants moved on to the phase 2 schedule 2 portion. It also states that thyroglobulin data from the University of California San Francisco were unavailable for five participants, leaving three analyzed in the phase 2 portion for that data. These figures describe different stages and data availability; they should not be collapsed into a single analysis population.

Allocation was not randomized and masking was absent. No randomized control comparison is available to separate treatment effects from natural variation, selection, or other changes over time. The registry record does not report the rationale for the enrollment target, a formal statistical analysis plan, or interim analyses. The record lists four outcome measures, including best overall response and NIS expression, but the corresponding analyzed results are not reported in the registry record.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Before assignment: Five and eight participants were enrolled from University of California San Francisco (UCSF) and the National Institutes of Health (NIH), respectively.

First Intervention Week 1-10

Reasons for not completing this period:

Second Intervention Week 11-17

Third Intervention Week 11-52

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: 13 participants were enrolled in phase 1 and 8/13 (5 from University of California San Francisco (UCSF) moved on from phase 1 to the phase 2 schedule 2 portion. However, Tg data from UCSF is unavailable for 5 of the participant, thus only 3 were analyzed in the phase 2 portion.

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

Population: Five and eight participants were enrolled from University of California San Francisco (UCSF), and the National Institutes of Health, respectively.

6. Primary Endpoint

The registry lists the primary outcome as RAI (Radioactive Iodine) Uptake and Tg (Thyroglobulin) Level Compared Pre and Post- Valproic Treatment, measured over Entry to study and after 10 weeks of treatment for Phase 1, and 10 weeks of treatment to 16 weeks of treatment for phase 2..

The primary endpoint combines information about radioactive iodine uptake, thyroglobulin (Tg) or thyroglobulin-antibody (Tg-Ab) levels, and tumor lesions. It classifies participants as having complete response, partial response, stable disease, or progressive disease. For phase 1, the stated comparison begins at study entry and continues to the assessment after ten weeks; the phase 2 timeframe runs from ten to sixteen weeks of treatment.

The rules distinguish response categories using multiple kinds of evidence. Complete response requires increased radioiodine uptake after treatment, a low or decreased marker level under the stated thresholds, and disappearance of all lesions at the later assessment. Partial response can be assigned through specified changes in uptake or marker levels together with a decrease in a target lesion. Stable disease requires no change in uptake and marker levels and no significant lesion change. Progressive disease is defined using increases in tumor mass, markers, or target-lesion size. These rules are intended to capture more than a single laboratory or imaging feature.

Using several measures may be useful in a disease where uptake, biochemical markers, and visible lesions do not necessarily change together. Statistically, however, a multi-part definition needs transparent application: the criteria determine how conflicting signals are classified and which observations are required. The registry gives the rules but does not report how many participants met each category, how missing components were handled, or the observed endpoint results. The clinical interpretation of the endpoint therefore cannot be assessed from the posted results alone.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does a response count represent?

The primary outcome is expressed as a count of participants classified by response category. A count is a direct description of how many people met a defined rule; it is not by itself a comparison of treatment effects. To interpret such a count, readers need the analysis population, the number assessed, the response classification rules, and information about missing observations. The registry gives the response rules and some population details, but it does not report primary response counts or a full analysis population for each assessment.

General statistical context: when a binary outcome is summarized as a count, an accompanying proportion can help describe how common the event was in the analyzed group. A confidence interval can communicate the uncertainty around that proportion. Neither a count nor a proportion establishes causality in a non-randomized study. The registry record does not report a confidence interval or an inferential analysis for the primary endpoint.

Why does the response definition matter?

The response categories use multiple types of evidence: uptake, marker levels, and lesion measurements. This is a multi-component classification. The exact rules are consequential because they determine whether evidence from one measure is sufficient, whether multiple conditions must be met, and how changes at different assessment times contribute. For example, the registry specifies that complete response requires a set of conditions, whereas partial response can be assigned using alternative stated changes alongside a lesion criterion. Stable and progressive disease also have explicit definitions.

General statistical context: a composite or multi-component endpoint can summarize several aspects of disease status, but it can also hide variation among components. A category may reflect a change in a marker, an imaging feature, or both, depending on its rules. Reporting each component and the final classification helps readers understand what drives the result. Component-level results are not reported in the registry record.

What can a non-randomized design establish?

Allocation is listed as non-randomized, masking as none, and the design model as parallel. The registry explicitly warns that the arms are not a like-for-like control comparison. Thus, the treatment schedules should not be interpreted as randomized groups. The record does not report a separate untreated or standard-treatment control group suitable for estimating a comparative treatment effect.

General statistical context: random allocation helps make treatment groups comparable on average at the outset, including on factors that are not measured. Without randomization, differences in participant characteristics or clinical decisions can influence which schedule a participant receives. An unmasked design can also affect assessment or reporting, especially when outcomes involve judgment. These are general design considerations, not evidence that a particular bias occurred in this study. The registry record does not describe adjustment methods that would address potential differences between groups.

How should the missing biomarker data be understood?

The record states that Tg data were unavailable for five participants from the University of California San Francisco and that three participants were analyzed in the phase 2 portion for that data. Missing values can reduce the information available for an endpoint and may affect interpretation if the likelihood of missingness is related to the outcome or participant characteristics. The registry does not report why these values were unavailable, whether the same participants lacked other measures, or what method was used to handle missing data.

General statistical context: an analysis that includes only participants with an observed measurement can be misleading if those participants differ systematically from those without measurements. Different missing-data assumptions can lead to different estimates. Clear reporting typically distinguishes enrolled participants, participants who reached an assessment, and participants included in each analysis. The registry provides some of these distinctions for the phase 2 Tg data, but does not provide a complete accounting across outcomes.

What do the planning illustrations mean?

The sample-size illustrations are planning calculations based on the observed effect, rather than the trial's own design justification. Their stated assumptions are a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation; the effect to detect is defined as a stated fraction of the observed effect. For time-to-event outcomes, the illustrations use total events, while other outcomes are expressed as participants per arm.

General statistical context: a planning calculation asks what sample size might be needed under specified assumptions for a future study. It is not a measure of the probability that this completed study detected an effect, and it is not observed power. In a study without a randomized, like-for-like comparison, planning around an observed effect also depends on how that effect was estimated. The registry record does not report a sample-size justification for the trial itself. No planning figures are repeated in this explanation.

What can be concluded from the posted analyses?

The record lists four outcome measures and zero analyses. It reports no primary outcome result counts, effect estimates, confidence intervals, or p-values. General statistical context: an estimate describes the size and direction of an observed association, while its confidence interval describes the precision of that estimate under the method used. A p-value, when reported, assesses compatibility with a specified null model; it does not measure clinical importance or prove causation. Since these analyses are not reported in the registry record, those inferential quantities cannot be evaluated for the primary outcome.

10. Multiplicity, Missing Data and Analysis Populations

The registry lists four outcome measures, including the primary response measure, best overall response, and NIS expression. It does not report analysis results for these outcomes or a plan for controlling multiplicity. When several outcomes or repeated assessments are considered, the chance of finding an apparently notable result can increase if each is tested without accounting for the wider set of analyses. This is general statistical context; the registry record does not say whether formal multiplicity adjustment was planned or used.

The primary endpoint itself draws on several components and assessment times. The record describes how response categories are defined, but does not report whether all components were available for each participant or how incomplete combinations were classified. It also does not report a prespecified hierarchy among outcomes or identify which analyses, beyond the listed primary outcome, were intended to be confirmatory.

For phase 2 Tg data, the record says that data from the University of California San Francisco were unavailable for five participants and that three were analyzed for that data. Reasons for the unavailability, any imputation, and the assumptions used for missing observations are not reported in the registry record. The record also does not provide a complete accounting of participants assessed for every endpoint or timepoint.

The analysis populations are therefore only partly described. Enrollment and movement into phase 2 are stated, but the record does not fully specify the population used for each outcome. This limits assessment of selection into analyses and makes it difficult to compare outcome summaries across phases. Statistical approaches to missingness and multiplicity are not reported in the registry record, so no claims about their handling should be inferred.

11. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: confusion (7.7%), pleural effusion (7.7%).

The registry's safety group is labeled all participants (phase 1 and phase 2 schedule 2). It reports thirteen participants at risk for serious events, with two affected, and thirteen at risk for other events, with eight affected. The death count is zero. These are participant counts in the registry summary; they are not rates per unit of follow-up time, and the record does not provide exposure-adjusted comparisons.

The top serious-event terms listed are confusion and pleural effusion, each reported as 7.7%. The registry summary does not state in this display how many episodes occurred, whether an individual could contribute more than one term, or the timing and duration of events. The terms should therefore be read as the reported safety categories, not as a complete clinical account of each event.

General statistical context: safety counts describe events observed and recorded under the study's reporting definitions. A comparison of safety risk requires comparable groups and sufficiently detailed information about exposure and follow-up. Because this study is non-randomized and no like-for-like control comparison is reported, the figures cannot establish whether an event was more or less common because of treatment. Zero deaths in the registry summary describes the posted count; it does not address other safety outcomes.

12. What the Result Does, and Does Not, Mean

The record documents a phase II, non-randomized study of valproic acid in advanced thyroid cancer of follicular origin. It specifies a response framework using radioactive iodine uptake, Tg or Tg-Ab levels, and lesion changes. It also records enrollment, movement into a later treatment schedule, a limitation in phase 2 Tg data, and selected safety counts. These details are useful for understanding what the study set out to assess and what information was available for some analyses.

The posted results do not include the primary response counts or the results for the other listed outcomes. As a result, the registry record does not show how many participants had complete response, partial response, stable disease, or progressive disease. It also does not provide estimates with confidence intervals or inferential tests for those outcomes. The absence of reported results should not be read as a zero response or as proof that no change occurred.

The study design also limits what can be concluded. Participants were not randomized, masking was absent, and the registry does not present a like-for-like control comparison. Any observed change from baseline, if reported elsewhere, would not by itself isolate the effect of valproic acid from other explanations. The partial phase 2 biomarker data further limit interpretation of that endpoint.

The safety summary reports serious and other event counts, including the listed serious-event terms, but without a comparable control group these figures do not quantify excess or reduced risk. They are descriptive registry figures and do not replace a complete safety analysis.

General statistical context: a study's ability to answer a question depends on the design, completeness of outcome data, precision of estimates, and transparency of analysis. The registry record offers endpoint definitions and selected counts, but not the outcome analyses needed to judge the magnitude or precision of response. It can therefore be read as a record of study conduct and reported information, not as a basis for a causal treatment conclusion.

13. Why This Trial Matters Statistically

This study illustrates why endpoint definitions and analysis populations matter in clinical research. Its primary response classification draws on radioiodine uptake, biochemical markers, and lesion changes at different timepoints. A reader needs to know both the decision rules and the availability of each component to understand how a participant was classified. The registry gives the rules, but not the resulting category counts or the handling of incomplete observations.

The record also makes clear that enrollment is not the same as availability for every analysis. Thirteen participants enrolled in phase 1, eight moved into the phase 2 schedule 2 portion, and only three were analyzed in the phase 2 portion for Tg data after specified values were unavailable. Keeping these stages separate is essential when interpreting any reported result and when assessing how representative an analysis might be.

Non-randomized treatment schedules can be informative for describing feasibility and observed outcomes, but they do not provide the same basis for causal comparison as randomized allocation. The absence of a like-for-like control means that a before-and-after change cannot be attributed to treatment on its own. This limitation is especially important when endpoints combine measurements that may vary for different reasons.

Finally, the record distinguishes descriptive safety counts from comparative safety evidence. Counts and listed event terms communicate what was recorded in the study group, while comparative risk requires an appropriate reference and sufficiently detailed follow-up. The entry thus highlights a broad statistical lesson: transparent definitions, complete outcome reporting, and design-appropriate comparisons are all necessary before readers can assess the size, precision, and meaning of a clinical result.

14. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The main limitation for interpretation is that the registry record does not report the primary endpoint results. It gives detailed response criteria but no counts by response category, estimates, confidence intervals, or statistical tests. Without those results, the magnitude and precision of any observed response cannot be assessed.

The non-randomized allocation and absence of masking limit causal interpretation. The registry notes that the arms are not a like-for-like control comparison, so differences between schedules cannot be treated as randomized treatment effects. The record does not report adjustment methods or a full statistical analysis plan.

The phase 2 biomarker information is incomplete: Tg data were unavailable for five participants from the University of California San Francisco, and three were analyzed for that data. The reasons for missingness and the approach to missing observations are not reported in the registry record. The record also does not fully describe analysis populations for every outcome or assessment time.

The arm descriptions are partial, which prevents a full reconstruction of treatment sequences and schedule-specific exposure. The registry lists other outcomes but does not report their analyzed results. It also does not report how multiplicity was addressed, whether interim analyses occurred, or the rationale for the enrollment target. These omissions mean the posted record supports a limited descriptive reading, not a complete evaluation of efficacy or safety.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: A Phase II Trial of Valproic Acid in Patients With Advanced Thyroid Cancers of Follicular Cell Origin — NCT01182285.
• Registry record last updated 2018-05-16; status last verified 2018-04.
• Results first posted 2016-12-14.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

18. Record Summary

The registry record describes a completed phase II, non-randomized, unmasked, parallel-design study of valproic acid for advanced thyroid cancers of follicular origin. Thirteen participants enrolled in phase 1. The primary outcome classifies response using radioactive iodine uptake, Tg or Tg-Ab levels, and lesion changes over the stated phase-specific assessment periods.

The record states that eight participants moved to the phase 2 schedule 2 portion and that only three were analyzed in the phase 2 portion for Tg data because values from five participants were unavailable. It reports no primary response counts or analyses. The safety summary lists thirteen at risk, two affected by serious events, eight affected by other events, and zero deaths; confusion and pleural effusion are each listed at 7.7% among the top serious terms.

Because the study was not randomized and the registry does not provide a like-for-like control comparison or outcome analyses, the posted information does not establish a causal treatment effect.