A Study of Lomvastomig (RO7121661) and Tobemstomig (RO7247669) Compared With Nivolumab in Participants With Advanced or Metastatic Squamous Cell Carcinoma of the Esophagus: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study of Lomvastomig (RO7121661) and Tobemstomig (RO7247669) Compared With Nivolumab in Participants With Advanced or Metastatic Squamous Cell Carcinoma of the Esophagus (NCT04785820), a randomized phase 2 study. The primary outcome is Overall Survival (OS).

2. Clinical Question

3. Trial Design

This was a randomized, blinded, active-controlled, parallel-group Phase II study. The registry lists quadruple masking and a global, multicenter design. Its stated enrollment was 190 participants, and its status is completed. The start date was 2021-06-25 and the primary completion date was 2025-01-30.

The original plan was to randomize participants in a 1:1:1 ratio to lomvastomig, tobemstomig, or nivolumab. After approval of protocol amendment version 3, recruitment into the lomvastomig arm stopped. The registry says the decision was based on strategic considerations, not emerging safety or efficacy data, and that the benefit/risk assessment for lomvastomig remained unchanged. Thereafter, randomization was to tobemstomig or nivolumab in a 1:1 ratio. The registry does not report the dates of the amendment or the detailed timing of recruitment into each arm.

The primary endpoint was overall survival in the intention-to-treat population, which included all randomized participants. The registry reports arm-level median survival estimates and a Log Rank comparison of nivolumab with tobemstomig. The posted primary results include 81 participants for nivolumab, 27 for lomvastomig, and 82 for tobemstomig. Those arm counts describe the posted outcome results; they should not be read as proof that each arm had the same follow-up or the same amount of outcome information.

The record lists 37 outcome measures and 4 analyses. It identifies adverse events, response, disease control, duration of response, progression-free survival, and quality-of-life outcomes among the secondary outcomes. The original statistical design details, including the trial's own sample-size justification and any interim-analysis plan, 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: Participants were randomized in a 1:1:1 ratio to receive nivolumab, tobemstomig (RO7247669), or lomvastomig (RO7121661). From Protocol Version 3 onwards, recruitment into the lomvastomig arm was stopped and participants were randomized in a 1:1 ratio to receive either nivolumab or tobemstomig. Two participants did not receive any study treatment and were excluded from the safety analysis.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: ITT population included all the randomized participants.

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: Intent-to-treat (ITT) population included all the randomized participants.

6. Primary Endpoint

The registry lists the primary outcome as Overall Survival (OS), measured over From randomization to death (up to approximately 38.7 months).

Overall survival is the time from randomization to death from any cause. It is a time-to-event endpoint: each participant contributes information about how long they were followed and whether death was observed during that period. The registry defines the primary analysis population as all randomized participants, the intention-to-treat population.

The median overall survival is the time by which the estimated proportion still alive has fallen to one half. It gives a central summary of the survival experience without requiring the survival times to follow a particular symmetric distribution. A Kaplan-Meier estimate accounts for participants whose event time is not observed during the available follow-up by using their information up to the point they are last known to be observed. This is a general explanation of the method; the record does not describe the amount or pattern of censoring in the arms.

Overall survival is used because it captures death from any cause and has an unambiguous event definition. The record gives the timeframe as from randomization to death, up to approximately 38.7 months. The posted median estimates and hazard ratio therefore describe survival during the observed study period, not a guarantee about later outcomes. The registry does not report a longer-term follow-up summary beyond that timeframe.

The sponsor also reports a Log Rank comparison expressed as a hazard ratio. A hazard ratio describes a relative comparison of event rates over follow-up under the model's interpretation; it is not a difference in median survival and does not directly state how many months an individual may live. The statistical interpretation must account for its confidence interval and p-value as well as the arm-level medians.

7. Results Posted in the Registry

• Nivolumab vs Tobemstomig: hazard ratio 1.05 (95% CI 0.84–1.33), reported p = 0.769 (as reported in the registry).

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.

Objective Response Rate (ORR)

• 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:
  • Lomvastomig: 3.7% (1/27) versus 8.6% (7/81) on control; risk difference -4.9 percentage points (95% CI -14.3 to 4.5), risk ratio 0.43 (0.06–3.33); unadjusted two-proportion z-test p = 0.396.
  • Tobemstomig: 9.8% (8/82) versus 8.6% (7/81) on control; risk difference 1.1 percentage points (95% CI -7.8 to 10.0), risk ratio 1.13 (0.43–2.97); unadjusted two-proportion z-test p = 0.806.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Lomvastomig 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 Tobemstomig and the control. The interval gives the range of effects that remain compatible with the data.

Disease Control Rate (DCR)

• 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:
  • Lomvastomig: 29.6% (8/27) versus 43.2% (35/81) on control; risk difference -13.6 percentage points (95% CI -33.9 to 6.7), risk ratio 0.69 (0.36–1.29); unadjusted two-proportion z-test p = 0.212.
  • Tobemstomig: 41.5% (34/82) versus 43.2% (35/81) on control; risk difference -1.7 percentage points (95% CI -16.9 to 13.4), risk ratio 0.96 (0.67–1.37); unadjusted two-proportion z-test p = 0.821.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Lomvastomig 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 Tobemstomig and the control. The interval gives the range of effects that remain compatible with the data.

Progression-free Survival (PFS)

• Endpoint type: Time to event: compared as a hazard ratio.
• Test: Hazard ratio as reported by the sponsor (Log Rank); it was not recalculated.
• Result:
  • Nivolumab vs Tobemstomig: hazard ratio 0.89 (95% CI 0.71–1.11), reported p = 0.494 (as reported in the registry).
• Interpretation:
  • The 95% confidence interval includes no difference (a ratio of 1), so these data do not show a statistically significant difference between Nivolumab vs Tobemstomig and the control. The interval gives the range of effects that remain compatible with the data.

The secondary outcomes broaden the description beyond overall survival. The registry lists response, disease control, duration of response, progression-free survival, adverse events, and quality-of-life measures covering general health, functioning, and dysphagia. They address different aspects of the study and use different outcome definitions and statistical scales. They should not be treated as interchangeable evidence about a single effect.

Across the reported secondary comparisons between nivolumab and tobemstomig, objective response and progression-free survival do not show a clear statistical separation in the sponsor-reported results. The posted arm-level response and disease-control summaries also have overlapping uncertainty. This overall pattern is broadly consistent with the primary result's lack of a statistically distinguishable difference in the reported comparison. It does not establish that the regimens have the same effects, and secondary outcomes do not substitute for the primary endpoint.

The registry lists 36 posted secondary outcome results, but the results for many specifically named measures, including duration of response and the quality-of-life outcomes, are not reported in the registry record. Detailed interpretation of those outcomes is therefore not possible from the recorded results. The record identifies 37 outcome measures in total, but listing an outcome is not the same as reporting a result for it.

Secondary analyses are generally less definitive when many outcomes or comparisons are examined and no multiplicity adjustment or hierarchy is reported. If many tests are performed, the chance of obtaining at least one apparently notable result by chance can increase. This is a general statistical concern, not a claim that any particular result is false. The registry does not report how multiplicity was addressed, so isolated secondary p-values should carry less weight than the prespecified primary analysis and should be interpreted as supportive or exploratory rather than conclusive.

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

• Number of Participants With Adverse Events (AEs)
• Objective Response Rate (ORR)
• Disease Control Rate (DCR)
• Duration of Response (DoR)
• Progression-free Survival (PFS)
• Percentage of Participants Reporting Clinically Meaningful Improvement in GHS/QoL, Emotional and Social Functioning as Measured by the European Organisation for Research and Treatment of Cancer (EORTC) Quality-of-life Questionnaire (QLQ)-C30
• Percentage of Participants Reporting Clinically Meaningful Improvement in GHS/QoL, Emotional and Social Functioning, as Measured by the EORTC QLQ - Item Library 97 (IL97)
• Percentage of Participants Reporting a Clinically Meaningful Improvement in Dysphagia, as Measured by the EORTC QLQ for Oesophageal Cancer-18 (OES-18)

9. Statistical Interpretation

10. Statistical Methods Explained

Overall survival is analyzed as a time-to-event outcome. The registry says that Kaplan-Meier methods were used to estimate median overall survival and that the 80% confidence interval for the median was computed using the Brookmeyer and Crowley method. These methods accommodate incomplete observation times by using the time participants were last known to be observed, rather than treating every participant without an observed death as though death had occurred at the end of follow-up. This is a general explanation; details about censoring rules and the amount of censoring are not reported in the registry record.

What does a Kaplan-Meier median describe?

The Kaplan-Meier curve estimates the proportion of a group that remains free of the event over time. For overall survival, the event is death from any cause. The median is the time at which the estimated survival curve reaches one half, when that point can be estimated. It is a summary of the curve, not a full description of its shape. Two groups can have similar medians while differing at other points in follow-up, or have different medians while their uncertainty remains substantial. The registry record does not post the underlying participant-level times in the registry record described for this analysis.

How should the hazard ratio be understood?

The sponsor reports a hazard ratio for nivolumab versus tobemstomig using a Log Rank method. In general, a hazard ratio compares the rates at which events occur over follow-up, conditional on participants having remained under observation and event-free until a given time. A value near 1 indicates little estimated relative separation under that comparison. It is not a ratio of median survival times and is not an absolute difference in months. Interpretation as a single summary across time is most straightforward when the relative hazards are reasonably stable; whether that condition was assessed or met is not reported in the registry record.

The Log Rank test compares the observed timing of events between groups over follow-up. It is designed for time-to-event data and uses more information than a simple comparison of the two median values alone. The sponsor-reported p-value is 0.769. A p-value is a measure of how compatible the observed data are with a specified null hypothesis under the test's assumptions; it is not the probability that the null hypothesis is true. The registry lists the analysis as a superiority analysis but does not report a noninferiority or equivalence margin.

What does the confidence interval add?

A confidence interval expresses uncertainty around an estimate under the method used. The primary median intervals are 80% confidence intervals, as is the interval for the sponsor-reported hazard ratio. They are not 95% intervals. A wider interval indicates less precision, while a narrower one indicates more precision under the same kind of analysis. The interval for the hazard ratio includes 1, so the posted result is compatible with no relative difference as well as with effects in either direction. It does not establish equivalence.

The registry reports the primary population as the intention-to-treat population, including all randomized participants. This population definition preserves the randomized comparison as far as the reported analysis permits. The record does not give further details on how participants with missing outcome information were handled, whether any participants were excluded from the posted analyses, or how deviations from assigned treatment were addressed.

For binary secondary outcomes, such as objective response rate and disease control rate, a percentage describes the proportion meeting the stated outcome definition. The registry reports sponsor analyses for objective response rate using the Cochran-Mantel-Haenszel method, including an odds ratio and a difference in overall response rates. That method can account for stratification factors, but the stratification factors used are not reported in the registry record. The record does not report a sponsor analysis for disease control rate. A simple unadjusted recalculation from arm-level summaries is a different analysis from a sponsor's stratified analysis, and the estimates need not match exactly.

For progression-free survival, the registry posts median estimates and a sponsor-reported hazard ratio from a Log Rank method. Progression-free survival is a time-to-event endpoint, but its event definition is not described in the registry record for this analysis. The registry record does not report additional details about the endpoint's censoring rules or the proportional-hazards assessment.

In general, over-dispersion means observed variability is greater than expected under a model's assumed variance. When present and not accounted for, it can make standard errors and confidence intervals too small and p-values too optimistic. The registry record does not report a dispersion assessment for the primary time-to-event analysis, so over-dispersion should not be asserted as a feature of these results. More broadly, the statistical plan, model diagnostics, adjustment factors beyond those identifiable from the reported method, and handling of multiple secondary hypotheses are not reported in the registry record.

11. Confidence Intervals and Precision

Precision is conveyed by the confidence intervals around the estimates, not by the point estimates alone. The primary results use 80% confidence intervals. For nivolumab, the median overall survival estimate was 8.08 months, with an interval from 6.74 to 9.0 months. For tobemstomig, it was 6.67 months, with an interval from 5.42 to 8.71 months. For lomvastomig, it was 4.76 months, with an interval from 2.83 to 5.82 months. These intervals show that the true median values are not estimated with exact precision; the statistical method gives a range of values compatible with the observed data at the stated confidence level.

The hazard ratio for nivolumab versus tobemstomig was 1.05, and its 80% confidence interval extended from 0.84 to 1.33. Because that interval includes 1, it does not exclude no relative difference. It also spans values on either side of 1, so the data do not tightly determine the direction or size of a relative effect. The interval should not be read as a range containing the true value with certainty, nor as a probability distribution for the effect.

An 80% confidence interval is less conservative than a 95% interval, and the confidence level matters when comparing estimates across studies or interpreting uncertainty. The registry specifies 80% intervals for these results; a 95% interval is not reported. The confidence intervals are also not a substitute for the analysis method or for an assessment of whether its assumptions are reasonable.

Arm sizes and event information affect precision, but the registry does not provide all details needed to quantify their separate contributions. Lomvastomig had 27 participants in the posted primary results, compared with 81 for nivolumab and 82 for tobemstomig. This context is relevant when considering the descriptive estimates, but it does not justify a new calculation of comparative precision. The sponsor's interval for the primary hazard ratio is the posted inferential comparison.

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 planning figures are illustrations based on observed effects, not the study's original sample-size design. For binary outcomes, the figures are participants per arm. For time-to-event outcomes, they are total events across the comparison. The overall-survival and progression-free-survival illustrations therefore count events, while the response and disease-control illustrations count participants in each arm.

The stated assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is set at the stated fraction of the observed effect. The original trial's sample-size justification and its design assumptions are not reported in the registry record; these planning assumptions apply only to the retrospective illustrations.

For each effect scenario, the 80% power column gives the number of participants per arm, or total events for time-to-event outcomes, that the illustration estimates would be needed to have an 80% chance of detecting that specified effect under the stated assumptions if it were the true effect. The 90% power column makes the same kind of projection for a 90% chance. Power is not the probability that a treatment works, and the figures do not describe what the completed trial actually achieved.

Rows for 100%, 75%, and 50% of the observed effect show progressively smaller effects to detect. Smaller effects generally require more information to distinguish from no effect, so their planning requirements are larger. When an observed effect is close to no difference, the resulting illustrations can become very large. The figures are planning illustrations and are not observed power.

These calculations are sensitive to the assumptions and the observed effect used as their starting point. They do not account for every possible feature of a future design, such as unequal allocation, loss to follow-up, or adjustments not listed among the assumptions. Such design details are not reported in the registry record. The illustrations should therefore be read as conditional statistical examples, not as recommendations or as evidence that the original trial was adequately or inadequately sized.

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

d is the total number of events needed (Schoenfeld's formula) with 1:1 allocation and proportional hazards.

Worked calculation for this trial

1. Hazard ratio HR = 1.050, so ln(HR) = 0.0488.
2. d = 4 × (1.96 + 0.842)² / [ln(1.050)]² = 13,188.7, rounded up to 13,189 events in total 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
hr <- 1.05                              # observed hazard ratio
events_needed <- function(power, frac = 1) ceiling(round(4 * (z_a + qnorm(power))^2 / log(hr^frac)^2, 6))   # Schoenfeld, 1:1 allocation
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: events = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(events_needed(0.80, f), big.mark = ","), format(events_needed(0.90, f), big.mark = ",")))

# Simulation: given d total events under proportional hazards, events in the treatment arm are Binomial(d, hr / (1 + hr)).
# The score (log-rank type) statistic is (D1 - d / 2) / sqrt(d / 4).
set.seed(4785820)
d    <- events_needed(0.80)
reps <- 20000
D1 <- rbinom(reps, d, hr / (1 + hr))
z  <- (D1 - d / 2) / sqrt(d / 4)
cat(sprintf("Simulated power with %s events: %.3f (%d replicates)\n", format(d, big.mark = ","), mean(abs(z) > z_a), reps))

Output from running the code

100% of observed effect: events = 13,189 (80% power), 17,656 (90% power)
 75% of observed effect: events = 23,447 (80% power), 31,389 (90% power)
 50% of observed effect: events = 52,755 (80% power), 70,624 (90% power)
Simulated power with 13,189 events: 0.798 (20000 replicates)

The simulation uses the fact that, with d events in total and proportional hazards, the number of events in the treatment arm is binomial. It is an approximation that conditions on the total number of events, and it should land close to the target power.

14. Analyses Reported by the Sponsor

The sponsor's primary analysis compares nivolumab with tobemstomig using a Log Rank method. It reports a hazard ratio of 1.05, an 80% confidence interval from 0.84 to 1.33, and p = 0.769. These are the sponsor's reported inferential results and are the appropriate results to prioritize when describing the registry's primary comparison.

Recalculated comparisons based on posted summary statistics are simple and unadjusted. They can be useful as a check or as an illustration of what the arm-level summaries imply, but they do not reproduce an analysis that uses participant-level time-to-event information. For the primary endpoint, the registry provides the sponsor's Log Rank comparison, and no separate unadjusted recalculation should replace it. The record does not flag a discrepancy between the sponsor results and recalculated results.

For objective response rate, the sponsor reports a Cochran-Mantel-Haenszel analysis for nivolumab versus tobemstomig, with an odds ratio of 1.12, an 80% confidence interval from 0.55 to 2.27, and p = 0.837. The sponsor also reports a difference in overall response rates of 1.11, with an 80% confidence interval from -4.69 to 6.91 and p = 0.837. Unadjusted calculations from the arm summaries use a different approach and may differ because they do not reproduce the sponsor's stratified method. Where sponsor analyses are reported, they take precedence over such simple recalculations.

The registry does not report a sponsor analysis for disease control rate. A recalculation for that endpoint should not be presented as though it were the sponsor's confirmatory analysis. The distinction is between a posted sponsor analysis and a supplementary calculation from summary data, not between two equally authoritative versions of the same analysis.

15. Multiplicity, Missing Data and Analysis Populations

Multiplicity arises when a study examines several outcomes, comparisons, or analysis options. Even when each statistical test is conducted at a conventional significance threshold, a larger collection of tests can increase the chance of at least one result appearing statistically notable by chance. This general principle is relevant because the registry identifies 37 outcome measures and 4 analyses. The registry record does not report a multiplicity-control strategy, a hierarchy of secondary outcomes, or which analyses were designated confirmatory beyond identifying overall survival as the primary outcome.

Without that information, secondary p-values should not be read as if they were a single prespecified test. The record does not report whether family-wise error, false discovery, or another approach was used. It also does not report whether the sponsor interpreted secondary analyses as exploratory. The absence of those details in the registry record does not prove that no approach existed in study documents; it means the registry does not establish how multiplicity was handled.

The primary population is defined as all randomized participants, which is the intention-to-treat population. This is an important feature because retaining participants according to randomization helps preserve the comparison created by random assignment. The record does not describe the number of participants with missing survival information, how missing data were handled for binary outcomes, the handling of withdrawal, or whether any analysis population differed from the stated intention-to-treat group.

For time-to-event outcomes, participants may have incomplete follow-up at the analysis cutoff and may be censored at their last known observation. This is a general feature of survival analysis, but the registry does not give the censoring rules, counts, or patterns for this study. The amount and distribution of follow-up by arm are also not reported in the registry record for these results. Those omissions limit assessment of whether incomplete observation might have affected the comparisons.

The registry reports a completed study and a primary completion date, but it does not report interim analyses, any stopping boundary for efficacy or safety, or the detailed statistical analysis plan. No inference about those design features should be made from the posted results alone.

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: dysphagia (10.1%), pneumonia (8.5%), pneumonia aspiration (7.4%), covid-19 (4.9%), stoma site infection (3.8%).

The registry posts serious adverse event counts and counts for other adverse events by arm, with the number at risk specified for each category. For serious adverse events, the figures are 32 affected among 79 at risk for nivolumab, 10 among 27 for lomvastomig, and 39 among 82 for tobemstomig. For other adverse events, the figures are 60 among 79, 24 among 27, and 67 among 82, respectively. These are counts of participants, not a comparative statistical test.

The listed serious adverse event terms include dysphagia at 10.1%, pneumonia at 8.5%, pneumonia aspiration at 7.4%, COVID-19 at 4.9%, stoma site infection at 3.8%, and anaemia at 3.7%. The registry extract does not specify in these figures the arm-specific distribution for each listed term or the exact analytic denominator for those percentages. They should not be interpreted as comparative risks between treatments.

Deaths are also recorded in the adverse-event data: 67 for nivolumab, 24 for lomvastomig, and 68 for tobemstomig. These counts should not be equated with deaths occurring as adverse events, nor compared as rates without considering follow-up and the relevant definitions. The registry record does not provide an adjusted safety comparison or enough detail to determine the timing, duration, severity grading, or causal attribution of events. These descriptive safety figures do not by themselves establish a treatment effect.

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

The primary registry result is a comparison of overall survival between nivolumab and tobemstomig. The sponsor-reported hazard ratio is 1.05, with an 80% confidence interval from 0.84 to 1.33 and p = 0.769. The interval includes the value corresponding to no relative difference, and the p-value does not indicate a statistically distinguishable difference under the reported test. The data therefore do not establish superiority of either arm in this comparison.

A result that does not show a statistically distinguishable difference is not proof of equivalence. Equivalence requires a design with a defined margin and an analysis intended to establish that differences fall within that margin. The registry record does not report an equivalence margin or an equivalence analysis. Nor does the interval demonstrate that clinically important differences have been ruled out; it spans a range of relative effects.

The arm medians are descriptive summaries, and their differences should not be interpreted as causal effects on their own. The randomized design supports comparative interpretation, but the sponsor's reported inferential analysis is the Log Rank hazard ratio comparison. The registry does not report a sponsor-reported primary comparison involving lomvastomig. Its smaller posted arm count and the change in recruitment plan should be kept in mind when reading its separate descriptive median.

The record says lomvastomig recruitment stopped for strategic considerations and not because of emerging safety or efficacy data. This statement concerns the reason for stopping recruitment; it does not establish a favorable or unfavorable result for lomvastomig. The sponsor states that its benefit/risk assessment remained unchanged.

The findings apply to the study's advanced or metastatic esophageal squamous cell carcinoma population and the follow-up captured in the registry record. They do not directly answer questions about other populations, longer follow-up, unreported quality-of-life results, or treatment choices for an individual. No medical advice follows from these registry statistics. The appropriate reading is limited to what was measured, what was reported, and the uncertainty around those estimates.

18. Why This Trial Matters Statistically

This study illustrates why a time-to-event endpoint should not be reduced to a comparison of two medians. The registry posts medians for each arm, but the sponsor's primary comparison uses a Log Rank method and reports a hazard ratio with a confidence interval and p-value. The medians summarize the survival distributions at one point; the time-to-event comparison uses the observed timing of events over follow-up. Each answers a related but distinct statistical question.

The example also highlights the role of interval estimates. A point estimate such as a hazard ratio of 1.05 can look close to no relative difference, but the confidence interval is needed to assess precision. Here the 80% interval spans values below and above 1, leaving uncertainty about both the direction and size of a relative effect. A nonsignificant result is not the same as evidence that two treatments are equivalent.

Recruitment changed during the study: the initial plan included three arms, while later randomization continued between tobemstomig and nivolumab. The record explicitly attributes stopping recruitment into lomvastomig to strategic considerations rather than emerging safety or efficacy data. This matters statistically because arm sizes and the context of the analysis differ, while the reported primary comparison concerns the two arms that continued randomization. Interpretation should follow the comparison actually reported rather than assume a fully balanced three-arm design throughout.

Finally, the registry lists numerous secondary outcomes, but details about multiplicity, missing data, interim analyses, and the original sample-size design are not reported. That gap illustrates why a registry result is best read alongside its endpoint definitions, population, analysis method, and uncertainty, without treating every listed outcome or calculation as equally confirmatory.

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.

The registry record gives a limited account of the statistical design. The trial's original sample-size justification, design assumptions, and any planned interim analyses are not reported in the registry record. The retrospective planning illustrations are based on observed effects and should not be taken as evidence about the rationale or adequacy of the original design.

The primary endpoint is defined clearly as time from randomization to death from any cause, but the record does not report details of censoring, completeness of follow-up, or the distribution of follow-up by arm. It also does not report whether the proportional-hazards assumption was assessed for the hazard-ratio interpretation. These omissions constrain evaluation of the time-to-event analysis beyond the estimates and method listed.

The primary sponsor comparison is nivolumab versus tobemstomig. The registry does not report a sponsor comparison involving lomvastomig, whose posted primary results include 27 participants, while the other arms include 81 and 82. The registry explains that recruitment into lomvastomig stopped after a protocol amendment for strategic reasons and not because of emerging safety or efficacy data. Even with that explanation, the smaller arm contributes less information to its descriptive estimates, and the reported primary comparison should not be generalized to an unreported lomvastomig contrast.

Many secondary outcomes are listed, but results for a number of named measures are not reported in the registry record. The record also does not describe a multiplicity strategy or missing-data handling for binary and quality-of-life outcomes. Safety counts are descriptive and do not include an adjusted comparison in the registry record for this analysis. These limitations do not show that the omitted analyses were never conducted; they limit what can be concluded from the registry record itself.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: A Study of Lomvastomig (RO7121661) and Tobemstomig (RO7247669) Compared With Nivolumab in Participants With Advanced or Metastatic Squamous Cell Carcinoma of the Esophagus — NCT04785820.
• Registry record last updated 2026-03-27; status last verified 2026-03.
• Results first posted 2026-03-27.
• 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

NCT04785820 is a completed randomized, blinded, active-controlled, parallel-group Phase II study of lomvastomig and tobemstomig compared with nivolumab in advanced or metastatic esophageal squamous cell carcinoma. The primary endpoint was overall survival in the intention-to-treat population, measured from randomization to death from any cause.

The sponsor-reported comparison of nivolumab with tobemstomig used a Log Rank method and gave a hazard ratio of 1.05, an 80% confidence interval from 0.84 to 1.33, and p = 0.769. The interval includes 1; the result does not establish superiority or equivalence. Lomvastomig recruitment stopped for strategic reasons, not emerging safety or efficacy data, according to the registry.

The record lists multiple secondary and safety outcomes, but statistical-plan details including multiplicity handling, missing-data methods, interim analyses, and the original sample-size justification are not reported in the registry record. The posted results should be interpreted with those limits in mind.