A Phase 2b, Study of Linsitinib in Subjects With Active, Moderate to Severe Thyroid Eye Disease (TED): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Phase 2b, Study of Linsitinib in Subjects With Active, Moderate to Severe Thyroid Eye Disease (TED) (NCT05276063), a randomized phase 2, phase 3 study. The primary outcome is Percentage of Subjects Who Are Proptosis Responders at Week 24.

2. Clinical Question

The study asks whether participants with active, moderate to severe thyroid eye disease assigned to linsitinib had different Week 24 outcomes from participants assigned to placebo. The primary outcome is the percentage classified as proptosis responders. The registry record defines a response as a reduction from baseline of at least 2 mm in the primary study eye, without an increase of at least 2 mm in the contralateral non-study eye.

The question is about a specified eye measurement and a defined response threshold. It is not a direct summary of every aspect of thyroid eye disease or of all outcomes participants might experience. The registry also lists overall response, a clinical activity score outcome, and change in a quality-of-life questionnaire score as secondary outcomes.

3. Trial Design

The registry record describes NCT05276063 as a randomized, double-masked, parallel-group, placebo-controlled study. Its title identifies it as a Phase 2b study, while the recorded phase categories are PHASE2 and PHASE3. The record does not explain this difference in phase labeling.

The listed enrollment is 90. The posted primary-outcome results show 30 participants in each of the placebo BID, 75 mg BID, and 150 mg BID groups. The study’s brief summary describes oral administration twice daily and an evaluation at 24 weeks. The registry lists the status as completed and gives 2022-07-01 as the start date and 2024-10-15 as the primary completion date.

Randomization is intended to make assigned groups comparable on average, while double masking is intended to limit the influence of treatment knowledge on participants’ experiences and outcome assessment. Those are general design purposes; the registry record does not describe the randomization procedure, who was masked, whether masking was maintained, or any stratification factors. It also does not report the sequence-generation or allocation-concealment details.

The three-arm structure allows each active dose to be compared with placebo. The calculated tables and plots display the posted group outcomes and comparisons; this discussion interprets those displays rather than reproducing them. The primary outcome is described as using an Intent to Treat population, but the registry record does not give a detailed account of how that population was defined or how participants with missing Week 24 measurements were handled. The trial’s original design assumptions, including its own sample-size justification, 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: A total of 138 participants were screened; 90 subjects were consented and randomized after review of inclusion exclusion criteria, while 48 were considered screen failures

Overall Study

Analysis population for the primary outcome, as defined in the registry: Intent to Treat

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.

6. Primary Endpoint

The registry lists the primary outcome as Percentage of Subjects Who Are Proptosis Responders at Week 24, measured over 24 weeks.

The primary endpoint is the percentage of participants who are proptosis responders at 24 weeks. A participant qualifies when the primary study eye has a reduction from baseline of at least 2 mm and the contralateral non-study eye has not deteriorated by an increase of at least 2 mm. The endpoint therefore combines improvement in the study eye with a safeguard against a specified degree of worsening in the other eye.

Proptosis is the forward displacement of the eye, and the endpoint expresses the result as a binary outcome: each participant either meets the response definition or does not. The percentage summarizes how often that defined response occurred in each treatment group. This can make a clinical measurement easier to interpret at the group level, but it does not show the full range of individual changes or describe every dimension of thyroid eye disease.

The response threshold is important to interpretation. The endpoint is not simply any reduction in proptosis; it requires the stated minimum change in one eye and no threshold-level increase in the other. The registry record identifies the Intent to Treat population for this outcome and gives the timeframe as 24 weeks. It does not report further details about measurement procedures, assessment timing within the visit window, or handling of missing measurements. These details matter when considering how consistently the endpoint could be assessed across participants.

7. Results Posted in the Registry

• 75 mg BID: 40.0% (12/30) versus 20.0% (6/30) on control; risk difference 20.0 percentage points (95% CI -2.6 to 42.6), risk ratio 2.00 (0.86–4.63); unadjusted two-proportion z-test p = 0.091.
• 150 mg BID: 53.3% (16/30) versus 20.0% (6/30) on control; risk difference 33.3 percentage points (95% CI 10.5 to 56.2), risk ratio 2.67 (1.21–5.88); unadjusted two-proportion z-test p = 0.007.

8. Secondary Efficacy Endpoints

The registry also posts results for secondary outcomes. For each outcome below, the endpoint type, the test applied to the posted summary statistics, the result and its interpretation are given in turn. Secondary endpoints are usually not adjusted for multiple testing, so they support rather than confirm the primary finding.

Percentage of Subjects Who Are Overall Responders at Week 24

• Endpoint type: Binary: the percentage of participants with the event (counts rebuilt from the posted percentages and group sizes).
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • 75 mg BID: 36.7% (11/30) versus 20.0% (6/30) on control; risk difference 16.7 percentage points (95% CI -5.7 to 39.1), risk ratio 1.83 (0.78–4.32); unadjusted two-proportion z-test p = 0.152.
  • 150 mg BID: 50.0% (15/30) versus 20.0% (6/30) on control; risk difference 30.0 percentage points (95% CI 7.1 to 52.9), risk ratio 2.50 (1.12–5.56); unadjusted two-proportion z-test p = 0.015.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between 75 mg BID and the control. The interval gives the range of effects that remain compatible with the data.
  • The 95% confidence interval excludes no difference, so the proportion with the event is statistically higher for 150 mg BID than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Percentage of Subjects With a CAS Value of 0 or 1 at Week 24

• Endpoint type: Binary: the percentage of participants with the event (counts rebuilt from the posted percentages and group sizes).
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • 75 mg BID: 46.7% (14/30) versus 36.7% (11/30) on control; risk difference 10.0 percentage points (95% CI -14.8 to 34.8), risk ratio 1.27 (0.69–2.33); unadjusted two-proportion z-test p = 0.432.
  • 150 mg BID: 56.7% (17/30) versus 36.7% (11/30) on control; risk difference 20.0 percentage points (95% CI -4.7 to 44.7), risk ratio 1.55 (0.88–2.72); unadjusted two-proportion z-test p = 0.121.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between 75 mg BID and the control. The interval gives the range of effects that remain compatible with the data.
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between 150 mg BID and the control. The interval gives the range of effects that remain compatible with the data.

Change From Baseline in the Graves' Ophthalmopathy Quality of Life (GO-QoL) Questionnaire Overall Score to Week 24.

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard errors.
• Result:
  • 75 mg BID: mean 6.39 versus 0.49 on control; difference 5.90 (95% CI -1.49 to 13.29), standardized effect d = 0.40; z-test p = 0.118.
  • 150 mg BID: mean 11.26 versus 0.49 on control; difference 10.78 (95% CI 2.78 to 18.77), standardized effect d = 0.68; z-test p = 0.008.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between 75 mg BID and the control. The interval gives the range of effects that remain compatible with the data.
  • The 95% confidence interval excludes no difference, so the mean is statistically higher for 150 mg BID than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

The secondary outcomes extend the assessment beyond the primary proptosis responder definition. They include overall response, a clinical activity score of 0 or 1, and change in the Graves' Ophthalmopathy Quality of Life questionnaire overall score. Together, they cover additional response criteria, disease activity and reported quality of life. They are complementary measures, not interchangeable versions of the primary endpoint.

Across the posted secondary results, the active-arm point estimates generally lie in a direction that is numerically higher than placebo for the listed outcomes. However, the comparisons vary in precision and in the strength of evidence shown by their reported intervals and p-values. The primary result therefore sits within a broader set of measures that is not uniformly conclusive. The secondary findings can add context, but they do not independently settle the primary question or establish consistency across every outcome.

The registry reports binary secondary comparisons using Cochran-Mantel-Haenszel analyses and reports the quality-of-life comparison using a Mixed Models Analysis. The model details and the variables used for any stratification are not reported in the registry record. Simple unadjusted re-calculations are also displayed in the calculated tables; where those differ from the sponsor analyses, the sponsor’s reported analyses should take precedence.

These secondary results carry less weight as confirmatory evidence when their multiplicity strategy is unknown. When several outcomes and treatment comparisons are examined, the chance of at least one apparently notable result can increase. That general issue does not make secondary outcomes uninformative, but it means isolated p-values should be treated cautiously unless the analysis plan explains how the family of tests was handled. The registry record does not report a multiplicity adjustment or a hierarchy for the secondary outcomes.

The secondary outcomes are best read as supportive context: they broaden the clinical questions examined and may suggest whether the primary pattern appears alongside other measures. The registry record does not report enough about prespecification, missing-data handling or analysis details to determine how confirmatory each result was intended to be.

Secondary outcomes registered for this trial, for reference (only those above were recalculated):

• Percentage of Subjects Who Are Overall Responders at Week 24
• Percentage of Subjects With a CAS Value of 0 or 1 at Week 24
• Change From Baseline in the Graves' Ophthalmopathy Quality of Life (GO-QoL) Questionnaire Overall Score to Week 24.

9. Statistical Interpretation

10. Statistical Methods Explained

The primary endpoint is binary: participants either meet the proptosis responder definition or do not. The registry presents the outcome as a percentage of participants and reports comparisons as risk differences. A risk difference expresses the contrast between the event proportions in two groups on the proportion scale. A value above zero indicates a higher response proportion in the active group than in the placebo group under the stated comparison; a value below zero indicates a lower response proportion. This is a general explanation of the measure, not an additional trial result.

What the reported analysis estimates

The registry lists a Cochran-Mantel-Haenszel method for the primary comparisons. In general, this method estimates an association or contrast across strata and can account for stratification in the analysis. The registry record does not identify the strata or explain how they were defined, so it is not possible to describe what factors the reported analysis accounts for. Its reported risk differences and confidence intervals are the sponsor’s estimates for the comparisons with placebo.

A confidence interval gives a range of values compatible with the data and the analysis, with a stated confidence level. Here the reported primary intervals are 95% confidence intervals. General statistical interpretation does not treat the interval as a probability statement about the particular true value after the data have been observed. Rather, the interval communicates estimation uncertainty under the method and its assumptions. Whether the interval crosses zero is relevant to a test of no risk difference, but it does not by itself determine clinical importance.

Why the re-calculation differs

The page also includes simple, unadjusted re-calculations based on posted summary statistics. These are not the same as the sponsor’s Cochran-Mantel-Haenszel analyses. The unadjusted calculations compare the posted group summaries directly and do not reproduce any adjustment or stratification used in the sponsor’s method. The registry record flags a discrepancy between these estimates and the sponsor’s reported analyses. For that reason, the sponsor’s own analysis takes precedence; the re-calculation is supplementary context rather than a replacement result.

For the primary endpoint, the re-calculated risk differences are 0.2 for 75 mg BID versus placebo and 0.333 for 150 mg BID versus placebo. The corresponding unadjusted p-values are 0.091 and 0.007. They are not identical to the sponsor’s reported estimates or p-values. Comparing the two sets illustrates why analysis methods and analysis populations matter: results computed from group summaries need not match a method that accounts for the structure used in the reported analysis.

What a p-value says

A p-value describes how surprising data at least as incompatible with a specified null hypothesis would be if that hypothesis and the test assumptions held. It is not the probability that the null hypothesis is true, the probability that a result occurred by chance, or a measure of effect size. The registry identifies the sponsor’s analyses as superiority comparisons. The record does not report the complete statistical analysis plan, including all test details or any prespecified decision rule, so interpretation should remain tied to the analyses actually posted.

For the primary comparison at 150 mg BID, the reported confidence interval is entirely above zero and its reported p-value is 0.01. For the 75 mg BID comparison, the reported interval includes zero and the reported p-value is 0.090. These observations describe the posted analyses; they do not establish that a dose has or lacks an effect in every relevant sense. The point estimate, interval, study design and analysis context should be considered together.

Other outcome models

The registry uses Cochran-Mantel-Haenszel analyses for the posted binary secondary comparisons and a Mixed Models Analysis for the quality-of-life outcome. A mixed model is a general approach that can represent relationships among repeated or otherwise related observations, depending on the model specification. The quality-of-life comparison is reported as a mean difference in final values, with a 95% confidence interval. The registry does not give the model specification, covariance structure, or other details needed to reconstruct that analysis.

The calculated re-calculations for secondary outcomes are also simple, unadjusted comparisons. They should not be confused with the sponsor’s reported secondary analyses. Where the record flags the overall discrepancy, the sponsor’s reported analysis is the primary reference for interpreting the posted comparisons. In general, models can differ because of adjustment, outcome definition, included observations or other analytic choices; the registry record does not identify which of these explains each difference.

Multiplicity and interpretation

Testing several doses and several outcomes creates multiple opportunities to observe a nominally notable result. General statistical practice may address this with a prespecified multiplicity strategy, but the registry record does not report such a strategy. The posted p-values should therefore be read as the values for the individual analyses shown, not as proof that family-wise error was controlled across all comparisons. The record also does not report interim analyses or stopping rules.

The primary outcome is described for the Intent to Treat population, which generally aims to analyze participants according to randomized assignment. The detailed definition and the treatment of missing outcomes are not reported in the registry record. Those missing details limit the ability to assess whether the reported estimates preserve the intended randomized comparison and how sensitive they might be to assumptions about unobserved outcomes.

11. Confidence Intervals and Precision

The sponsor’s 95% confidence interval for 150 mg BID versus placebo on the primary endpoint extends from 0.098 to 0.593 around a reported risk difference of 0.345. It stays above zero, while still spanning a range of possible contrast sizes. The interval supports the direction of the reported estimate under the sponsor’s analysis, but its breadth indicates that the exact size is not pinned down precisely.

For 75 mg BID versus placebo, the reported 95% confidence interval extends from -0.040 to 0.475 around a risk difference of 0.217. It includes zero as well as positive values. Consequently, these results are less conclusive about the direction and size of the contrast than an interval entirely on one side of zero would be. This does not show that there is no difference; it shows that the estimate is uncertain enough to include that possibility under the reported analysis.

Confidence intervals are useful because they show more than a single point estimate. In general, a wider interval reflects less precision, and an interval’s compatibility with zero is relevant to assessing evidence against a no-difference hypothesis. Neither feature answers whether a difference is important in clinical terms. The registry record does not state a clinically important threshold for the proptosis response risk difference.

The trial’s posted groups are modest in size, and the calculated tables include intervals for the re-calculated comparisons as well. Those unadjusted intervals are not substitutes for the sponsor’s intervals, given the discrepancy. Precision should be interpreted alongside the reported method, endpoint definition, analysis population and missing-data handling. Details about missing observations and the original design rationale are not reported in the registry record, so the precision displayed cannot resolve those uncertainties.

12. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The sample-size table and curve are planning illustrations derived from the observed effect, not the trial’s own original design calculation. For these outcomes, the entries refer to participants per arm, not total events. The stated planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. Each row considers an effect to detect equal to the stated fraction of the observed effect.

The 80% power column shows the planned number of participants per arm that would give an 80% chance of detecting the specified effect under those assumptions, if the assumptions and effect size used for planning held. The 90% power column describes the corresponding planning number for a 90% chance. These are design calculations, not guarantees about the result of a future study. The original trial’s sample-size justification and planning assumptions are not reported in the registry record.

Rows that plan for a smaller fraction of the observed effect generally require more participants per arm. This is because a smaller difference between groups is harder to distinguish from sampling variability, all else held equal. The table is therefore useful for seeing how strongly a planning requirement depends on the assumed effect size. It should not be read as evidence that the observed study had adequate or inadequate power, and it is not observed power.

The illustrations compare two arms at a time and use equal allocation between those arms. They do not account for every possible design feature or analysis choice in a future trial. In particular, the registry record does not report the original study’s statistical plan or sample-size rationale, so these illustrations cannot be used to infer what guided enrollment in this study. The tables and curve are planning aids for understanding how sample requirements change with the assumed effect, rather than a retrospective score of the study’s success.

13. How the Sample Size Is Derived

Power is the probability that a study of a given size rejects the null hypothesis when a specific true effect exists. It depends on four things: the effect to detect, the variability of the outcome, the significance level (here a two-sided α of 0.05, so z_1−α/2 = 1.960) and the desired power (z_1−β = 0.842 for 80% and 1.282 for 90%).

Formula

n is the number of participants in each arm, p₀ and p₁ are the control and treatment proportions, and p̄ is their average.

Worked calculation for this trial

1. Control proportion p₀ = 0.200; treatment proportion p₁ = 0.400; average p̄ = 0.300.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.6481 = 1.2702.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.6325 = 0.5323.
4. n = (1.2702 + 0.5323)² / (0.2000)² = 81.2, rounded up to 82 participants per arm for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
p0 <- 0.2                          # control proportion
p1 <- 0.4                          # treatment proportion
n_per_arm <- function(power, frac = 1) {
  q1 <- p0 + (p1 - p0) * frac               # smaller effect = fraction of the observed difference
  pbar <- (p0 + q1) / 2
  z_b <- qnorm(power)
  ceiling(round((z_a * sqrt(2 * pbar * (1 - pbar)) + z_b * sqrt(p0 * (1 - p0) + q1 * (1 - q1)))^2 / (q1 - p0)^2, 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

# Check against R's built-in function (unrounded n per arm)
cat("power.prop.test:", round(power.prop.test(p1 = p0, p2 = p1, power = 0.80)$n, 1), "\n")

# Simulation: how often does the z-test reject at the 80%-power sample size?
set.seed(5276063)
n    <- n_per_arm(0.80)
reps <- 20000
x0 <- rbinom(reps, n, p0); x1 <- rbinom(reps, n, p1)
ph <- (x0 + x1) / (2 * n)
z  <- ((x1 - x0) / n) / sqrt(ph * (1 - ph) * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(z) > z_a, na.rm = TRUE), reps))

Output from running the code

100% of observed effect: n per arm = 82 (80% power), 109 (90% power)
 75% of observed effect: n per arm = 138 (80% power), 185 (90% power)
 50% of observed effect: n per arm = 294 (80% power), 392 (90% power)
power.prop.test: 81.2 
Simulated power at n = 82 per arm: 0.807 (20000 replicates)

The simulation draws many pairs of event counts at the planned sample size and counts how often the two-proportion z-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

14. Analyses Reported by the Sponsor

The sponsor’s primary analyses use the Cochran-Mantel-Haenszel method and report risk differences of 0.345 for 150 mg BID versus placebo and 0.217 for 75 mg BID versus placebo. The corresponding 95% confidence intervals are 0.098 to 0.593 and -0.040 to 0.475, with reported p-values of 0.01 and 0.090.

The recalculated estimates are simple unadjusted re-calculations from posted summary statistics. They yield risk differences of 0.333 for 150 mg BID and 0.2 for 75 mg BID, with unadjusted p-values of 0.007 and 0.091. Those figures do not exactly reproduce the sponsor’s results. The registry record flags a discrepancy, so the sponsor’s own analysis takes precedence when describing the primary comparisons.

The two analyses answer closely related comparison questions but use different methods: the sponsor reports a Cochran-Mantel-Haenszel analysis, whereas the re-calculation is unadjusted. The record does not state the strata used in the sponsor’s method or give enough detail to identify the reason for each numerical difference. It would therefore be inappropriate to attribute the discrepancy to a particular adjustment, population rule or data-handling decision.

The sponsor analysis is the appropriate basis for the main reading of the posted primary result, while the recalculation helps show how a direct comparison of summary statistics can differ. Neither should be treated as a complete substitute for the statistical analysis plan, which is not reported in the registry record.

15. Multiplicity, Missing Data and Analysis Populations

The registry record lists one primary endpoint and several secondary outcomes, with multiple treatment comparisons against placebo. Repeated testing across outcomes and comparisons raises a general multiplicity issue: the more analyses are considered, the greater the chance that at least one result will appear notable even if there are no underlying group differences. An analysis plan can address this with a prespecified hierarchy or adjustment, but such a strategy is not reported in the registry record.

Consequently, the posted p-values are best understood as belonging to the individual analyses shown. The record does not establish that they were adjusted to control error across all endpoints, doses or comparisons. This is particularly relevant when interpreting secondary outcomes, and it also matters when drawing conclusions across both active doses. The primary analysis remains the designated primary endpoint in the record, but the complete testing plan is not available.

The primary outcome is identified as being analyzed in the Intent to Treat population. In general, an Intent to Treat approach retains participants in the groups to which they were randomized, supporting the randomized comparison. The registry record does not explain the exact population definition, how post-randomization events were handled, or whether every randomized participant contributed a Week 24 outcome.

Missing-data methods are also not reported in the registry record. It is not clear how missing measurements, withdrawals, or participants without an assessable Week 24 endpoint were treated in the sponsor’s analysis. Missingness can matter if the likelihood of having an observed outcome differs across groups or is related to participants’ outcomes; this is a general statistical concern, not a claim about what occurred in this study.

The record does not report the full statistical analysis plan, interim analyses, stopping rules, or a sample-size justification. These omissions limit assessment of prespecification and of how the posted analyses relate to the study’s original design. The available results remain interpretable as posted comparisons, but these unreported details restrict a complete evaluation of confirmatory strength and potential sensitivity to analysis choices.

16. 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 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: electrocardiogram t wave abnormal (3.4%), guillain-barre syndrome (3.4%), pericardial effusion (3.2%), pleural effusion (3.2%).

The safety summary lists serious adverse events and other adverse events by group, alongside the number at risk for each category. For serious adverse events, the posted figures are 1 affected among 31 at risk in placebo, 0 among 30 at risk with 75 mg BID, and 2 among 29 at risk with 150 mg BID. The record lists 0 deaths in each group.

For other adverse events, the figures are 9 affected among 31 at risk in placebo, 14 among 30 at risk with 75 mg BID, and 18 among 29 at risk with 150 mg BID. These are group-level counts, not evidence by themselves that a particular event was caused by study treatment. The registry record does not provide event-level timing, duration, severity beyond the serious-event classification, or an assessment of causality for each participant.

The record identifies Electrocardiogram T Wave Abnormal and Guillain-Barre Syndrome among the top serious-event terms, each listed as 3.4%. Pericardial Effusion and Pleural Effusion are also listed, each as 3.2%. The registry extract does not explain the denominator or detail how the top terms were selected, so these percentages should be read as displayed rather than recalculated or compared as rates.

These safety figures are descriptive. The at-risk counts differ between groups, and the registry record does not provide enough information to assess exposure time or to make a causal comparison. The figures should not be interpreted as a complete safety profile or as medical advice.

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

The posted results describe the proportion of participants who met a specific proptosis response definition at Week 24 and the sponsor’s estimates comparing each active dose with placebo. In the sponsor’s analysis, the reported interval for 150 mg BID is above zero, whereas the interval for 75 mg BID includes zero. This describes the statistical evidence and uncertainty in these particular comparisons; it is not a universal statement about every outcome or participant.

The endpoint captures a defined change in eye position with a condition on the other eye. It does not encompass every feature of thyroid eye disease, and a group percentage does not show the distribution of individual responses. The secondary measures examine other aspects, but they also have their own definitions and analysis methods. The registry record does not report a threshold for what size of group difference would be clinically important.

The findings do not establish the long-term pattern of outcomes beyond the stated 24-week timeframe, nor do they answer questions not measured in the posted outcomes. They should not be taken as proof that every participant would experience the same result, or that a particular individual will respond in a particular way. Aggregate trial estimates summarize groups and remain subject to sampling uncertainty and design limitations.

The record is also incomplete for some statistical questions. It does not report the original sample-size rationale, the full analysis plan, multiplicity handling, or missing-data methods. A discrepancy is flagged between the sponsor’s reported results and simple unadjusted re-calculations; the sponsor’s own analysis takes precedence for interpreting the posted comparisons. These qualifications matter when weighing the findings.

Overall, the result is best understood as an estimate of a specified Week 24 group contrast, accompanied by a confidence interval and a reported method. It neither guarantees a particular outcome for an individual nor resolves all questions about safety, other disease features, or outcomes over time.

18. Why This Trial Matters Statistically

This study is statistically informative because it compares two linsitinib doses with placebo using random assignment and masking, and it reports a defined binary primary endpoint. Randomized comparisons can reduce bias from baseline differences on average, while masking can limit some influences on assessments. These are general features of the design; the registry record does not provide enough detail to assess how allocation and masking were implemented.

The three-arm layout makes it possible to inspect each dose against a shared placebo group, while also requiring careful interpretation of multiple comparisons. The primary outcome has a transparent response definition and timeframe, making the target of the analysis explicit. The reported risk differences and confidence intervals give readers information about both estimated direction and precision rather than only a statement about whether a p-value crosses a threshold.

The record also illustrates why analysis methods matter. Sponsor-reported Cochran-Mantel-Haenszel estimates differ from direct unadjusted re-calculations, and the registry flags a discrepancy. Reporting the sponsor analysis alongside the summary-statistic re-calculation makes the distinction visible, but the unreported strata and analysis-plan details prevent a full reconstruction.

Finally, the study places the primary endpoint alongside measures of overall response, clinical activity and quality of life. These can broaden interpretation, while also raising multiplicity and missing-data questions. The registry record does not resolve those issues. Thus, the trial is a useful example of reading estimates, intervals, analysis methods and reporting gaps together rather than relying on a single result in isolation.

19. 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.
• Participant counts were reconstructed from posted percentages and arm sizes, so small rounding differences are possible.

The registry record leaves several trial-specific details unclear. It lists the study title as Phase 2b but records phase categories of PHASE2 and PHASE3, without explaining the labeling. It does not report detailed eligibility criteria, so the precise boundaries of the population represented by subjects with active, moderate to severe thyroid eye disease cannot be assessed from this record alone.

The primary endpoint is defined, but measurement procedures, assessment training, visit windows and adjudication details are not reported. These details can affect how reproducibly a threshold-based response is classified. The record identifies the Intent to Treat population but does not specify its operational definition or how missing Week 24 measurements were handled.

The sponsor’s primary method is named, but its strata and the full model specification are not reported. The statistical analysis plan, multiplicity approach, interim analyses and stopping rules are also not reported in the registry record. This limits the ability to determine which analyses were prespecified and how the family of comparisons was interpreted.

A discrepancy exists between the sponsor’s analyses and the unadjusted re-calculations from summary statistics. The sponsor’s own analysis takes precedence, but the record does not explain the source of the difference. The posted safety summary is also limited: it gives counts and selected serious-event terms but not detailed event narratives, timing, exposure time or causal assessments.

Finally, the reported primary timeframe is 24 weeks. Longer-term outcomes and the original sample-size justification are not reported in the registry record. These limitations do not erase the posted estimates, but they constrain how completely the study’s methods, robustness and broader implications can be evaluated.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: A Phase 2b, Study of Linsitinib in Subjects With Active, Moderate to Severe Thyroid Eye Disease (TED) — NCT05276063.
• Registry record last updated 2026-09-23; status last verified 2026-08.
• Results first posted 2026-08-26.
• 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.

23. Record Summary

NCT05276063 is a completed randomized, double-masked, parallel-group study of oral linsitinib versus placebo in subjects with active, moderate to severe thyroid eye disease. The registry lists 90 participants and three groups: placebo BID, 75 mg BID and 150 mg BID. The primary endpoint is the percentage of Intent to Treat participants meeting the defined proptosis responder criterion at 24 weeks.

The sponsor reports Cochran-Mantel-Haenszel risk-difference comparisons with placebo. The reported 150 mg BID interval is above zero; the reported 75 mg BID interval includes zero. The registry flags a discrepancy with simple unadjusted re-calculations, so the sponsor’s own analysis takes precedence. Secondary results cover overall response, clinical activity score and GO-QoL change. The original sample-size rationale, multiplicity strategy and missing-data methods are not reported in the registry record.