Phase III Study of Trifluridine/Tipiracil With and Without Bevacizumab in Refractory Metastatic Colorectal Cancer Patients: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Phase III Study of Trifluridine/Tipiracil With and Without Bevacizumab in Refractory Metastatic Colorectal Cancer Patients (NCT04737187), a randomized phase 3 study. The primary outcome is Overall Survival (OS).

2. Clinical Question

Overall survival is a time-to-event outcome: it accounts for both whether death occurred and how long participants were followed before death or the end of observation. The posted results report a median overall survival of 10.78 months for the combination group and 7.46 months for the monotherapy group. The reported hazard ratio was 0.61, with a 95% confidence interval of 0.49 to 0.77 and p < 0.001. These figures describe the trial’s observed comparison; they do not by themselves establish what would happen for every person with this condition.

3. Trial Design

NCT04737187, the SUNLIGHT Study, was a randomized, open-label, parallel-group phase III study comparing trifluridine/tipiracil plus bevacizumab with trifluridine/tipiracil alone in refractory metastatic colorectal cancer. The registry reports 492 enrolled participants and a completed study. The recorded start date was 2020-11-25, and the primary completion date was 2022-07-19.

Both groups received trifluridine/tipiracil; bevacizumab was added in the combination group. The registry describes oral trifluridine/tipiracil dosing and a treatment cycle, but the arm-description text is truncated, so the complete regimen details are not reported in the registry record. The registry identifies the study as controlled, randomized and international. It does not describe masking, which is recorded as none.

The primary endpoint was overall survival, defined from randomization to death from any cause or the data cut-off, whichever came first. The stated maximum duration for this endpoint was up to 20 months. The primary analysis used the Kaplan-Meier method and was performed in the FAS population. The registry record does not report the original sample-size justification, the allocation ratio as a separate protocol detail, or the full statistical analysis plan.

The posted outcomes also include progression-free survival, response, disease control, adverse events and quality-of-life measures. While the registry lists these additional measures, detailed results are not present for every listed outcome in the record. The open-label design is relevant when interpreting outcomes that may depend on assessments or reports, while overall survival is based on death from any cause. Randomization supports a comparison between assigned groups, but interpretation remains tied to the trial population, follow-up and analyses actually reported.

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: Randomization (1:1 ratio): stratified by geographic region (North America, European Union, Rest of the World), time since 1st metastasis diagnosis (less than \[\<\] 18 months, greater than or equal to \[\>=\] 18 months) and rat sarcoma virus status (wild-type, mutant). Primary analysis of overall survival as primary outcome measure was completed with cutoff date of July 19, 2022; no efficacy analysis was conducted after this cutoff and participants were continued to be followed for safety analysis.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Analysis was performed on FAS population.

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: Full Analysis Set (FAS): all participants to whom a therapeutic unit was randomly assigned using Interactive Web Response System (IWRS). Participants were analyzed in the arm they were assigned by randomization.

6. Primary Endpoint

The registry lists the primary outcome as Overall Survival (OS), measured over From date of randomization to the death due to any cause or cut-off date, whichever comes first (maximum duration: up to 20 months).

Overall survival measures the time from randomization until death from any cause. Participants who had not died by the data cut-off were followed only until that cut-off for this analysis. The registry states a maximum duration of up to 20 months. Because the endpoint uses death from any cause, it does not depend on a judgement about whether a death was caused by the cancer or by treatment.

The analysis used the Kaplan-Meier method, which estimates survival over time while accounting for participants whose observed follow-up ends before the event. The median is the time at which the estimated survival curve reaches the halfway point. It is a summary of the distribution of observed survival times, not a prediction of an individual participant’s survival duration. The reported medians were 10.78 months for trifluridine/tipiracil plus bevacizumab and 7.46 months for trifluridine/tipiracil alone.

Overall survival is commonly used in trials because it is a direct, clinically meaningful time-to-event outcome. Here, the registry also reports a hazard ratio to compare the groups across follow-up, alongside confidence intervals and a stratified log-rank test. These measures complement the medians: the median summarizes each group, while the hazard ratio summarizes the relative event pattern over the observed period under the analysis used. Neither summary conveys every detail of the survival curves.

7. Results Posted in the Registry

• Trifluridine/Tipiracil + Bevacizumab vs Trifluridine/Tipiracil: hazard ratio 0.61 (95% CI 0.49–0.77), reported p < 0.001 (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.

Progression Free Survival (PFS)

• Endpoint type: Time to event: compared as a hazard ratio.
• Test: Hazard ratio as reported by the sponsor (Stratified log-rank test); it was not recalculated.
• Result:
  • Trifluridine/Tipiracil + Bevacizumab vs Trifluridine/Tipiracil: hazard ratio 0.44 (95% CI 0.36–0.54), reported p < 0.001 (as reported in the registry).
• Interpretation:
  • The 95% confidence interval excludes no difference (a ratio of 1), so the hazard of the event is statistically lower for Trifluridine/Tipiracil + Bevacizumab vs Trifluridine/Tipiracil than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Overall 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:
  • Trifluridine/Tipiracil: 1.2% (3/246) versus 6.1% (15/246) on control; risk difference -4.9 percentage points (95% CI -8.2 to -1.6), risk ratio 0.20 (0.06–0.68); unadjusted two-proportion z-test p = 0.004.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the proportion with the event is statistically lower for Trifluridine/Tipiracil than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Percentage of Participants With Disease Control

• 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:
  • Trifluridine/Tipiracil: 41.9% (103/246) versus 69.5% (171/246) on control; risk difference -27.6 percentage points (95% CI -36.1 to -19.2), risk ratio 0.60 (0.51–0.71); unadjusted two-proportion z-test p = <0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the proportion with the event is statistically lower for Trifluridine/Tipiracil than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

The secondary outcomes span time to progression or death, response and disease control, adverse events, and patient-reported quality-of-life measures. The posted progression-free survival comparison is directionally consistent with the overall-survival finding: the reported hazard ratio is below 1 for the combination group compared with monotherapy. The posted response and disease-control summaries also show different observed results between groups. Taken together, these endpoints provide context across disease control and survival, rather than a replacement for the primary overall-survival analysis.

These outcomes are not interchangeable. Progression-free survival concerns radiological tumour progression or death, whereas overall survival concerns death from any cause. Response and disease control are assessed as categorical outcomes within their stated assessment periods. Quality-of-life measures address participant-reported health and deterioration, and adverse-event outcomes concern safety. A coherent pattern across different outcomes can be informative, but each endpoint has its own definition, measurement process and uncertainty.

The registry lists several additional secondary measures, including survival probabilities at specified times and quality-of-life deterioration or utility measures. Detailed results for all listed measures are not reported in the registry record. The record also does not report a multiplicity adjustment or hierarchy for interpreting secondary comparisons. When many endpoints are assessed, the chance of at least one apparently notable result can increase unless the analysis plan accounts for multiple testing. Therefore, unadjusted secondary findings generally carry less confirmatory weight than the designated primary result.

Some posted secondary comparisons are recalculated from group summaries rather than presented as sponsor-reported analyses. Such calculations can describe the posted data but do not establish that the same analysis was prespecified or adjusted for other comparisons. Interpret secondary outcomes as supporting context, with attention to whether the result was sponsor-reported, the endpoint’s definition, and the absence of reported multiplicity details.

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

• Progression Free Survival (PFS)
• Probability of Participants With Progression Free Survival at 3, 6, 9 and 12 Months
• Overall Response Rate (ORR)
• Percentage of Participants With Disease Control
• Number of Participants With Treatment-emergent Adverse Events (TEAE) and Treatment-emergent Serious Adverse Events (TESAEs)
• European Organization for the Research and Treatment of Cancer Quality of Life Questionnaire Core 30 (EORTC QLQ-C30): Time to Definitive Deterioration of >=10 Points in Sub-scale Scores - Kaplan-Meier Analysis
• European Organization for the Research and Treatment of Cancer Quality of Life Questionnaire Core 30 (EORTC QLQ-C30): Time to Definitive Deterioration of >=10 Points in the Global Health Status Score - Kaplan-Meier Analysis
• Change From Baseline in European Quality of Life Group Questionnaire With 5 Dimensions and 5 Levels Per Dimension (EQ-5D-5L): Health State Utility Index Value at Specified Timepoints

9. Statistical Interpretation

10. Statistical Methods Explained

The primary outcome is time-to-event data: each participant contributes information about the time from randomization until death or the end of observed follow-up. Some participants may have no recorded death by the data cut-off. Their information is censored at the last time represented in the analysis rather than treated as if they had died at that time. Censoring allows the analysis to use the available follow-up, but interpretation relies on assumptions about how censoring relates to the outcome.

What does the Kaplan-Meier method estimate?

The Kaplan-Meier method estimates the probability of remaining event-free over time from the observed event and censoring times. The overall-survival median is the time at which that estimated probability reaches the halfway point. A median does not show every feature of the survival distribution: two groups can have the same median but different patterns at other times. The registry gives group medians and confidence intervals, but does not provide a numerical survival estimate at every time point in the primary results.

How should the hazard ratio be read?

The sponsor-reported hazard ratio compares the event rates between the groups over follow-up within the stratified time-to-event analysis. The reported overall-survival estimate is 0.61 for trifluridine/tipiracil plus bevacizumab versus trifluridine/tipiracil. A value below 1 indicates a lower estimated hazard in the first-named group under this comparison. It is not a ratio of median survival times and does not mean that each participant’s risk changed by the same amount. Interpreting a single hazard ratio as a concise summary is most straightforward when the relative hazard is reasonably stable over time; the record does not report a diagnostic assessment of that condition.

What does the stratified log-rank test do?

The stratified log-rank test compares the time-to-event experience of the randomized groups while accounting for the strata used in the test. The registry identifies this as the method for the reported hazard-ratio analysis and gives p < 0.001 for overall survival. The registry record does not identify the stratification factors, so their clinical or statistical role cannot be described from this record. The p-value assesses compatibility of the observed data with the test’s null hypothesis, under the method’s assumptions; it does not quantify the probability that the treatment effect is real.

The registry also states that the primary analysis used the FAS population. That population label is recorded, but its exact operational definition, handling of participants with incomplete data, and any exclusions from the analysis are not reported in the registry record. Randomized assignment provides the basis for the between-group comparison, while details of how all post-randomization events were handled are not available beyond the stated treatment-policy estimand.

Why use confidence intervals as well as a p-value?

A confidence interval expresses uncertainty around an estimated effect under the statistical model and procedure used. For the overall-survival hazard ratio, the 95% confidence interval is 0.49 to 0.77. This interval communicates a range of effect estimates compatible with the analysis at the stated confidence level; it is not a range containing a fixed proportion of individual outcomes. It adds information about precision that a p-value alone does not convey.

For the group medians, the registry reports 95% confidence intervals as well: 9.36 to 11.83 months for the combination group and 6.34 to 8.57 months for monotherapy. These intervals describe uncertainty in the median estimates, not the range of survival times for individual participants. The reported methods and intervals should be read as properties of this analysis, rather than as guarantees about a future study.

Why can a time-to-event analysis be more informative than a simple event count?

A simple count of deaths would ignore when events occurred and how long each participant was observed. Time-to-event methods retain timing information and can use follow-up from participants who have not had the event by the cut-off, through censoring. This is useful when participants have different observed follow-up. It also means the result depends on how event times and censoring are recorded and on the assumptions used by the method. The registry gives the endpoint definition and method, but detailed censoring rules and diagnostics are not reported in the registry record.

For secondary binary outcomes, such as overall response rate and disease control, the registry records percentages and confidence intervals. These are different from survival analyses: they classify whether an outcome was recorded within the stated assessment period rather than comparing time until an event. An unadjusted recalculation of a binary comparison from posted summary data is not a substitute for a sponsor-reported inferential analysis, particularly where the registry does not post a corresponding analysis. Different outcome types answer different questions and require methods suited to their data structure.

Finally, the registry reports randomized allocation and an open-label design. Randomization helps balance known and unknown participant characteristics across groups on average, but does not remove uncertainty from sampling or guarantee balance in every characteristic. Lack of masking can affect assessments or reporting for outcomes that involve judgement, though it does not change the definition of overall survival. The record does not report further details of the statistical analysis plan, including diagnostic checks or the handling of every intercurrent event.

11. Confidence Intervals and Precision

The overall-survival median estimates were 10.78 months, with a 95% confidence interval of 9.36 to 11.83, for trifluridine/tipiracil plus bevacizumab; and 7.46 months, with a 95% confidence interval of 6.34 to 8.57, for trifluridine/tipiracil. The intervals express uncertainty in the group median estimates. They are not individual survival ranges and do not show how outcomes vary among participants.

For the primary comparison, the hazard ratio was 0.61, with a 95% confidence interval of 0.49 to 0.77. The interval indicates that the estimate is not infinitely precise; a range of values remains compatible with the analysis. Because the interval does not include 1, the reported estimate is consistent with a group difference under this model and comparison. This does not quantify the clinical importance of the difference or address every source of uncertainty.

The p-value, p < 0.001, addresses compatibility with the test’s null hypothesis, while the interval gives a sense of the estimate’s precision and plausible scale. Neither measure accounts for unreported aspects of trial conduct or analysis. Precision is also tied to the observed follow-up and the analysis population. The registry record does not report the detailed follow-up distribution or the number of deaths used in the primary analysis, limiting further assessment of how the interval’s precision arose.

Confidence intervals can be wide or narrow depending on the amount and pattern of information, among other factors. The reported intervals should therefore be considered alongside the study design and outcome definitions, not as stand-alone certainty statements. They describe uncertainty in estimates for the trial analysis, not certainty about effects in other populations or settings.

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 the observed effect, not the trial’s original design assumptions. The original sample-size justification is not reported in the registry record. The stated illustration 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 the stated fraction of the observed effect.

For time-to-event outcomes, the entries are total events required, not participants per arm. For other outcomes, the entries are participants per arm. The 80% power column describes the planning sample size or event count associated with an 80% chance of detecting the specified effect under the illustration’s assumptions, if that effect and assumptions held. The 90% power column gives the corresponding planning target for a 90% chance. These are design probabilities under a model, not guarantees that a particular study will reach a result.

The rows represent planning for 100%, 75% or 50% of the observed effect. A smaller effect is harder to distinguish from random variation, so the illustration requires more participants or events to achieve the same target power. The displayed values are not observed power and do not describe the probability that the trial’s result is correct. They are planning references tied to specified assumptions and should not be mistaken for the actual enrollment rationale for SUNLIGHT.

The scale differs across endpoint types: time-to-event planning uses event counts, while binary outcome planning uses participants per arm. These quantities should not be compared as if they were the same unit. The original protocol’s assumptions, including any intended effect size and other design considerations, are not reported in the registry record.

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 = 0.610, so ln(HR) = -0.4943.
2. d = 4 × (1.96 + 0.842)² / [ln(0.610)]² = 128.5, rounded up to 129 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 <- 0.61                              # 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(4737187)
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 = 129 (80% power), 173 (90% power)
 75% of observed effect: events = 229 (80% power), 306 (90% power)
 50% of observed effect: events = 514 (80% power), 689 (90% power)
Simulated power with 129 events: 0.793 (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. Group-Sequential Design Analysis

Many time-to-event trials, especially in oncology, stop or report early when interim data are overwhelming, so a single fixed-sample number understates the real design problem. The calculation below is an illustrative group-sequential design built from this trial's posted control median and hazard ratio. The registry record does not state the planned number of looks or the spending function, so the design is a planning illustration, not a reconstruction of the protocol. Assumed: one-sided α = 0.025, 90% power, three equally spaced looks on information, Lan-DeMets O'Brien-Fleming spending, 24 months of accrual, 60 months in total and a 1% monthly dropout hazard.

Under these assumptions the design needs about 203 participants in total, with the boundaries below.

Fixed design versus group-sequential alternatives

Group-sequential designs cost a small increase in maximum sample size and events compared with the fixed design, in exchange for the chance to stop early for overwhelming efficacy. O'Brien-Fleming spending keeps early boundaries very conservative and the final boundary close to the fixed-design value, whereas Pocock spending spends more alpha early.

Sensitivity to a smaller true effect

Planning for exactly the observed effect is optimistic, because observed effects in published trials tend to be somewhat larger than the true effect. The table repeats the main design (three looks, O'Brien-Fleming spending, 90% power) for smaller true hazard ratios.

For context, this trial enrolled 492 participants.

Cross-check, upper boundaries: gsDesign 3.71, 2.511, 1.993 | rpact 3.71, 2.511, 1.993. Agreement between the two packages confirms the boundary calculation.

R code and output

suppressWarnings(suppressMessages({ library(gsDesign); library(rpact) }))
med_control <- 7.46          # control-arm median, months (converted from the posted unit)
hr          <- 0.61          # hazard ratio used for planning
lambdaC     <- log(2) / med_control
eta         <- 0.01               # dropout hazard per month (assumption)
acc_months  <- 24                 # accrual duration (assumption)
study_months<- 60                 # total study duration (assumption)

design <- function(k, sfu, beta, h = hr) gsSurv(k = k, test.type = 1, alpha = 0.025, beta = beta, sfu = sfu,
  lambdaC = lambdaC, hr = h, hr0 = 1, eta = eta, gamma = 1, R = acc_months, T = study_months, minfup = study_months - acc_months)

# Main design: 3 looks, Lan-DeMets O'Brien-Fleming spending, one-sided alpha 0.025, 90% power
x <- design(3, sfLDOF, 0.10)
cat(sprintf("N_TOTAL|%s\n", format(ceiling(sum(x$gamma * x$R)), big.mark = ",")))
for (i in 1:3) cat(sprintf("GS|%d|%.1f|%.0f|%.3f|%.5f|%.5f\n", i, x$T[i], x$n.I[i], x$upper$bound[i], pnorm(-x$upper$bound[i]), cumsum(x$upper$spend)[i]))
cat("\n"); print(gsBoundSummary(x))

# Comparison of designs
cmp <- list(list("Fixed design, one analysis", 1, sfLDOF), list("Two looks, O'Brien-Fleming", 2, sfLDOF),
            list("Three looks, O'Brien-Fleming", 3, sfLDOF), list("Three looks, Pocock", 3, sfLDPocock))
cat("\n")
for (b in c(0.20, 0.10)) for (d in cmp) {
  y <- design(d[[2]], d[[3]], b)
  cat(sprintf("DES|%s|%d|%.0f|%s|%.1f\n", d[[1]], round(100 * (1 - b)), tail(y$n.I, 1), format(ceiling(sum(y$gamma * y$R)), big.mark = ","), tail(y$T, 1)))
}

# Sensitivity: what if the true effect is smaller than the observed one? (3 looks, O'Brien-Fleming, 90% power)
cat("\n")
for (fr in c(1, 0.75, 0.5)) {
  y <- design(3, sfLDOF, 0.10, exp(log(hr) * fr))
  cat(sprintf("SENS|%.2f|%.3f|%.0f|%s\n", fr, exp(log(hr) * fr), tail(y$n.I, 1), format(ceiling(sum(y$gamma * y$R)), big.mark = ",")))
}

# Independent cross-check of the boundaries with rpact
r <- getDesignGroupSequential(kMax = 3, alpha = 0.025, beta = 0.10, sided = 1, typeOfDesign = "asOF", informationRates = x$timing)
cat(sprintf("\nCross-check, upper boundaries: gsDesign %s | rpact %s\n", paste(round(x$upper$bound, 3), collapse = ", "), paste(round(r$criticalValues, 3), collapse = ", ")))

Method: LachinFoulkes 
    Analysis               Value Efficacy
   IA 1: 33%                   Z   3.7103
      N: 142         p (1-sided)   0.0001
  Events: 58        ~HR at bound   0.3762
   Month: 17    P(Cross) if HR=1   0.0001
             P(Cross) if HR=0.61   0.0338
   IA 2: 67%                   Z   2.5114
      N: 204         p (1-sided)   0.0060
 Events: 116        ~HR at bound   0.6263
   Month: 27    P(Cross) if HR=1   0.0060
             P(Cross) if HR=0.61   0.5603
       Final                   Z   1.9930
      N: 204         p (1-sided)   0.0231
 Events: 173        ~HR at bound   0.7385
   Month: 60    P(Cross) if HR=1   0.0250
             P(Cross) if HR=0.61   0.9000



Cross-check, upper boundaries: gsDesign 3.71, 2.511, 1.993 | rpact 3.71, 2.511, 1.993

15. Analyses Reported by the Sponsor

The overall-survival comparison includes a sponsor-reported stratified log-rank analysis: hazard ratio 0.61, 95% confidence interval 0.49 to 0.77, and p < 0.001. The registry also gives the group medians and their confidence intervals. These sponsor-reported results are the primary inferential account for the endpoint.

Any recalculated comparison based on posted summary statistics is an unadjusted calculation and cannot reproduce the full participant-level, stratified analysis. The record does not identify the stratification factors or provide the underlying event-time data needed to independently reconstruct that analysis. Accordingly, the sponsor’s own reported analysis takes precedence for interpreting the primary result. The recalculated figures should be treated as supplementary checks of posted summaries, not as replacements for the reported analysis.

The registry reports sponsor analyses for progression-free survival as well. For response and disease control, it posts outcome percentages and summary-based comparisons but no sponsor-reported analysis in the listed results. Those recalculations therefore do not have the same status as the sponsor’s prespecified or reported inferential analyses. The registry record does not state whether additional analyses were performed but not posted.

16. Multiplicity, Missing Data and Analysis Populations

The registry identifies overall survival as the primary endpoint and reports a sponsor analysis for that comparison. It also lists multiple secondary outcomes, including progression-free survival, response, disease control, safety and quality-of-life measures. When a study examines multiple outcomes, testing each at the same nominal threshold can increase the chance of false-positive findings across the family of tests. Multiplicity procedures, such as a prespecified testing hierarchy or adjustment, can address that issue.

The registry record does not report a multiplicity strategy, a hierarchy for secondary endpoints, or whether any adjustment was applied. Consequently, the inferential weight of secondary p-values should not be assumed to match that of the primary result. This is particularly relevant for analyses that are not accompanied by a sponsor-reported test. Lack of a reported approach does not establish that no approach was used; it means the record does not provide the details needed to assess it.

The record identifies the FAS population for the primary analysis but does not define its membership or describe exclusions and missing observations in detail. It does not report the missing-data assumptions, censoring diagnostics, sensitivity analyses, or methods used to handle incomplete quality-of-life assessments. For time-to-event outcomes, participants without an event by the cut-off contribute censored follow-up, but the detailed rules and assessments are not reported in the registry record.

The primary estimand is described as assessing the effect of randomized treatments in all participants regardless of whether intercurrent events occurred, using a treatment-policy strategy. That description clarifies the intended treatment-effect question at a high level, but the record does not give operational details for every intercurrent event or explain how each was represented in the analysis. The absence of these details limits independent evaluation of estimand implementation and robustness.

These gaps matter because statistical conclusions depend not only on the reported estimate and p-value but also on which participants and observations enter each analysis, how incomplete follow-up is handled, and how multiple questions are interpreted. The registry provides useful endpoint definitions and selected results, but it does not present a complete protocol or statistical analysis plan.

17. 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: malignant neoplasm progression (4.5%), anaemia (3.7%), intestinal obstruction (3.7%), febrile neutropenia (2.4%), covid-19 (2.4%).

The registry reports participants with serious adverse events and participants with other adverse events by randomized group. Serious adverse events were recorded for 66 participants in the combination group and 79 in the monotherapy group. Other adverse events were recorded for 241 participants in each group. The record also reports deaths in each group; deaths are not, by themselves, proof that an adverse event caused the death.

The listed serious-event terms include malignant neoplasm progression at 4.5%, anaemia at 3.7%, intestinal obstruction at 3.7%, febrile neutropenia at 2.4%, COVID-19 at 2.4%, and jaundice at 2.0%. These are the posted term percentages, not a complete description of all events or their severity and timing. The registry record does not provide detailed exposure-adjusted rates or enough information to assess the comparative contribution of each treatment.

Safety counts describe events recorded during the study, not a causal comparison in isolation. Interpretation should consider how events were defined, collected and attributed, and how long participants were observed. Those details are not fully reported in the registry record. The figures should be read as descriptive trial safety information rather than as an individual risk estimate.

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

The primary analysis reports different overall-survival estimates between the randomized groups. The median was 10.78 months with trifluridine/tipiracil plus bevacizumab and 7.46 months with trifluridine/tipiracil alone. The sponsor-reported hazard ratio was 0.61, with a 95% confidence interval of 0.49 to 0.77 and p < 0.001. This describes the observed group-level comparison in the trial population under the stated analysis.

The result does not mean that every participant in the combination group lived longer than every participant in the monotherapy group. Group summaries cannot predict an individual’s outcome. Nor does the hazard ratio translate directly into a difference in months for each participant. It summarizes the relative event pattern under the time-to-event analysis, while the medians summarize separate points on each group’s estimated survival distribution.

The result also does not answer every relevant question. It does not by itself establish the balance of benefits and harms for an individual, the effect in people unlike the enrolled population, or the durability of outcomes beyond the reported follow-up. The stated maximum duration was up to 20 months, and the registry record does not provide the complete follow-up distribution or all details of treatment exposure and subsequent care.

Overall survival is a direct endpoint, but its analysis still has uncertainty and depends on how participants were followed and censored. Confidence intervals describe statistical uncertainty in the estimates; they do not account for all possible limitations, such as unreported analysis details. Secondary findings can add context, but multiplicity and missing analysis-plan information mean they should not be treated as equally confirmatory.

Accordingly, the posted result supports a statistical difference between the randomized groups in this trial as reported, but it is not a guarantee, a causal explanation for every observed death, or medical advice. Interpretation should remain bounded by the study’s population, design, endpoint definitions and the information recorded in the registry.

19. Why This Trial Matters Statistically

This trial illustrates how a randomized comparison can be summarized using more than one time-to-event statistic. The medians describe the estimated survival time at the halfway point of each group’s survival distribution, while the hazard ratio compares event rates across follow-up. Reporting both helps readers avoid reducing the outcome to a single number and makes clear that a relative comparison and a time summary are related but not identical.

The confidence interval adds information that the p-value cannot provide alone. The primary hazard-ratio interval, 0.49 to 0.77, conveys uncertainty around the reported estimate, while the p-value addresses compatibility with a null hypothesis under the specified test. Neither replaces consideration of the analysis population, censoring, follow-up or the trial’s open-label design.

The record also demonstrates the distinction between primary and secondary evidence. A designated primary endpoint supports the main confirmatory question, while multiple secondary outcomes broaden the description of survival, disease status, safety and quality of life. Without a reported multiplicity strategy, secondary findings require more cautious interpretation. Summary-based recalculations can be useful as checks, but they are not interchangeable with sponsor-reported analyses using the trial’s full analysis structure.

Finally, the study highlights the importance of statistical transparency. Some essential details—including the original sample-size rationale, stratification factors, full missing-data approach and multiplicity plan—are not reported in the registry record. Such omissions do not erase the posted result, but they limit how completely independent readers can assess its design and robustness.

20. 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 does not report the original sample-size justification or the complete statistical analysis plan. It does not identify the stratification factors used for the stratified log-rank test, detail the censoring rules, or provide diagnostics for the time-to-event assumptions. These omissions limit independent reconstruction and evaluation of the primary analysis.

The study was open-label, with no masking. This is especially relevant to outcomes that depend on assessment or participant reporting, including response and quality-of-life measures. The record does not provide enough detail to determine how assessment procedures may have mitigated potential influence from awareness of assigned treatment.

The primary outcome states a maximum duration of up to 20 months, but the registry record does not give the complete follow-up distribution or detailed information about the timing of all events. The FAS population is named but not operationally defined. Missing-data handling, sensitivity analyses and detailed implementation of the treatment-policy estimand are not reported in the registry record.

Several secondary outcomes are listed, but not all have detailed posted results. The record does not report a multiplicity approach, which limits the strength of conclusions drawn from multiple secondary comparisons. Some binary outcome comparisons are available as recalculations rather than sponsor-reported analyses, and such calculations cannot reproduce participant-level analyses or account for unreported design features.

The safety information is also limited to the posted counts and selected serious-event terms. Detailed exposure, event timing and attribution information are not reported in the registry record. These limitations do not negate the reported estimates, but they constrain how broadly and precisely the findings can be interpreted.

21. Related Tutorials

22. Related Statistical Calculators

23. Sources

• ClinicalTrials.gov: Phase III Study of Trifluridine/Tipiracil With and Without Bevacizumab in Refractory Metastatic Colorectal Cancer Patients — NCT04737187.
• Registry record last updated 2024-09-24; status last verified 2024-08.
• Results first posted 2023-12-26.
• 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.

24. Record Summary

NCT04737187, the SUNLIGHT Study, was a completed randomized, open-label, phase III comparison in refractory metastatic colorectal cancer. The registry reports 492 enrolled participants and compares trifluridine/tipiracil plus bevacizumab with trifluridine/tipiracil monotherapy.

The primary endpoint was overall survival, analyzed using the Kaplan-Meier method in the FAS population. The sponsor-reported hazard ratio was 0.61, with a 95% confidence interval of 0.49 to 0.77 and p < 0.001. Median overall survival was 10.78 months in the combination group and 7.46 months in the monotherapy group.

The record also lists secondary efficacy, safety and quality-of-life outcomes. Not all listed secondary results or statistical-plan details are reported. Findings should be interpreted as trial-level statistical evidence, with attention to confidence intervals, the open-label design, multiplicity and gaps in the registry record.