FOLFOX6m Plus SIR-Spheres Microspheres vs FOLFOX6m Alone in Patients With Liver Mets From Primary Colorectal Cancer: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for FOLFOX6m Plus SIR-Spheres Microspheres vs FOLFOX6m Alone in Patients With Liver Mets From Primary Colorectal Cancer (NCT01721954), a randomized phase 3 study. The primary outcome is Overall Survival (OS).

2. Clinical Question

3. Trial Design

This was a randomized, parallel-group, phase 3 study of first-line treatment for patients with non-resectable liver metastases from primary colorectal carcinoma. The registry reports enrollment of 209 and identifies Sirtex Medical as the sponsor. The study was completed. Its start date is 2013-05-01 and its primary completion date is 2016-12-23.

The experimental regimen combined FOLFOX6m chemotherapy with SIR-Spheres microspheres. The comparator was FOLFOX6m alone. The arm descriptions allow bevacizumab in either group when it was part of local standard care, at the investigator’s discretion. Thus, the contrast is between assignment to the two treatment strategies, not necessarily between receipt and non-receipt of bevacizumab.

The study was open label: masking is listed as none. The registry identifies a randomized parallel design but does not report the randomization sequence method, allocation concealment procedures, or any measures taken to limit the effects of knowing treatment assignment. Those details are not reported in the registry record.

Overall survival was the primary outcome in the ITT population. Progression-free survival was posted as a secondary outcome. The primary endpoint was measured from randomization until death from any cause, with assessment stated as up to 3 yrs 8 months. The record reports medians, confidence intervals for the medians, and log-rank hazard ratios with p-values. It does not report the full statistical analysis plan, the trial’s original sample-size justification, or the details of follow-up and censoring procedures.

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.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: ITT 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.

6. Primary Endpoint

The registry lists the primary outcome as Overall Survival (OS), measured over From date of randomization until the date of death from any cause assessed up 3 yrs 8 months.

Overall survival is the time from randomization until death from any cause. It is a time-to-event endpoint: each participant contributes information about how long they were followed without the event, or until death was recorded. The registry identifies the ITT population for this outcome and states that it was assessed up to 3 yrs 8 months.

The median is the point in the estimated survival distribution at which half of the population is expected to have experienced the event and half to remain event-free. It describes the middle of the observed time-to-event distribution rather than the experience of every participant. The posted median was 25.9 months for mFOLFOX6 Plus SIRT, with a 95% confidence interval of 23.1 to 29.0 months. For mFOLFOX6 Alone, the posted median was 25.0 months, with a 95% confidence interval of 22.1 to 28.5 months.

Overall survival is commonly used in clinical trials because death from any cause is a direct, clinically important event and does not depend on defining or detecting disease progression. The comparison also uses a hazard ratio, which summarizes the relative rate of the event over follow-up under the model and analysis used. A hazard ratio does not itself describe the absolute time until death or prove that the relative rate was constant over time. The registry does not report checks of that assumption.

7. Results Posted in the Registry

• mFOLFOX6 Plus SIRT vs mFOLFOX6 Alone: hazard ratio 0.93, reported p = 0.430 (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

• 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:
  • mFOLFOX6 Plus SIRT vs mFOLFOX6 Alone: hazard ratio 0.75, reported p < 0.050 (as reported in the registry).
• Interpretation:
  • The posted data give a point estimate for mFOLFOX6 Plus SIRT vs mFOLFOX6 Alone but no confidence interval, so the precision of the comparison cannot be judged.

Progression-free survival is the one posted secondary endpoint. Its reported analysis gives a hazard ratio of 0.75 with p<0.05, whereas the primary overall-survival comparison has p=0.43. The direction and statistical evidence therefore differ across the posted endpoints. This pattern can occur because endpoints capture different events and time points, and because the analyses have different observed event histories.

The secondary finding does not override or substitute for the primary endpoint. Overall survival addresses time until death from any cause; progression-free survival addresses time until documented progression or death, whichever comes first. A result on one endpoint does not necessarily imply a corresponding result on the other.

Secondary outcomes are often interpreted with less weight than a prespecified primary outcome, particularly when multiple outcomes or analyses are examined. Testing multiple hypotheses increases the chance that at least one result appears statistically notable by chance if no adjustment is made. The registry does not report whether a multiplicity procedure was used, how the secondary endpoint fit into a testing hierarchy, or the complete set of planned analyses. Accordingly, the secondary result is best treated as supportive or exploratory evidence rather than a standalone basis for a definitive claim.

The registry reports a hazard-ratio confidence interval for neither the primary nor secondary comparison. The secondary p-value therefore does not show the range of relative effects compatible with the estimate. More complete interpretation would require the prespecified analysis plan, details of endpoint assessment and censoring, and confidence intervals for the relative-effect estimates; those details are not reported in the registry record.

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

• Progression-free Survival

9. Statistical Interpretation

10. Statistical Methods Explained

The main analysis is a time-to-event comparison. Unlike an analysis that uses only whether an event occurred, a time-to-event analysis incorporates the timing of events and can also account for participants whose event time is not observed during the available follow-up. The registry reports a log-rank method for the comparisons and a hazard ratio as the effect measure. It does not give full details of how censored observations were handled.

What does a median survival time describe?

A median survival time is a point on the estimated time-to-event distribution. It is not an average, and it is not the time at which every participant experiences the event. In this study, the posted overall-survival medians are 25.9 months and 25.0 months for the two groups. Each median has a 95% confidence interval. The intervals describe uncertainty around the median estimates; they are not ranges containing the outcomes of all participants.

Median summaries are easy to communicate, but they compress the follow-up pattern into one number per group. They do not show when the curves separate, whether they cross, or whether any difference is concentrated in part of follow-up. The record does not give a hazard-ratio confidence interval or the information needed to reconstruct the entire survival distribution from the posted summaries.

What does the hazard ratio express?

A hazard is the instantaneous event rate among participants who have not yet experienced the event, considered over time. A hazard ratio compares the hazards between groups. The reported overall-survival hazard ratio is 0.93 for mFOLFOX6 Plus SIRT versus mFOLFOX6 Alone. In general, a value below 1 indicates a lower estimated hazard in the first-listed group relative to the comparison group, while a value above 1 indicates a higher estimated hazard. This is a relative measure, not an absolute difference in months or a direct probability of survival.

Interpreting a single hazard ratio as a constant summary is most straightforward when the relative hazards are reasonably stable over follow-up. If that condition does not hold, a single value can obscure variation over time. The registry record does not report a proportional-hazards assessment, a time-varying effect analysis, or survival-curve details that would allow readers to assess this issue.

What does the log-rank test add?

The log-rank test compares the observed event experience across groups over follow-up, taking account of event timing. The reported p-value for overall survival is 0.43. A p-value describes how compatible the observed data are with a specified no-difference hypothesis under the test’s assumptions; it is not the probability that the hypothesis is true, and it does not measure the size or clinical importance of an effect. The record does not report a confidence interval for the hazard ratio, so the p-value cannot replace a direct display of uncertainty around that relative-effect estimate.

Random allocation supports a comparison of assigned treatment strategies by reducing systematic differences between groups at baseline, on average. It does not ensure that every feature is balanced in the realized groups, nor does it prevent problems arising from missing data, non-adherence, or outcome measurement. The primary outcome is described as ITT, which generally means analysis according to randomized assignment, but the registry does not give detailed rules for the analysis population or deviations from assigned treatment.

Because the study was unmasked, participants and study personnel could know treatment assignment. Death is less dependent on subjective outcome interpretation than some outcomes, but knowledge of assignment can still affect care, follow-up, or other processes. The record does not report procedures for blinded endpoint adjudication or the extent of any differential follow-up.

Statistical methods are only one part of interpretation. The estimated effects, their uncertainty, the conduct of follow-up, and the choice of endpoint all matter. The registry reports the primary log-rank result and the group medians, but the trial’s complete analysis plan, handling of missing observations, and original design assumptions are not reported in the registry record.

11. Confidence Intervals and Precision

The overall-survival median estimates are 25.9 months for mFOLFOX6 Plus SIRT and 25.0 months for mFOLFOX6 Alone. The 95% confidence intervals are 23.1 to 29.0 months and 22.1 to 28.5 months, respectively. These intervals convey uncertainty about the estimated medians. They are not individual-level ranges, and they do not directly provide a confidence interval for the between-group hazard ratio.

The reported hazard ratio is 0.93, but its confidence interval is not reported in the registry record. Without that interval, readers cannot use the registry result to see the range of relative effects compatible with the fitted estimate at the stated confidence level. The p-value, 0.43, indicates that the log-rank analysis did not produce strong evidence against the no-difference hypothesis; it does not show that the estimate is precise or that meaningful effects have been ruled out.

Overlapping confidence intervals for the separate medians are not, by themselves, a formal test of the between-group effect. The log-rank comparison is the reported inferential test. The medians and their intervals provide useful descriptive context, but they answer a different question from the hazard ratio.

Precision also depends on the number and timing of observed events and on follow-up. The record gives the study enrollment and group outcome denominators, but does not provide the number of deaths used in the primary analysis, the distribution of follow-up time, or the hazard-ratio confidence interval. This limits how fully the primary effect’s uncertainty can be judged from the registry result.

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 expressed as total events for the time-to-event outcomes, not participants per arm. For overall survival, the displayed scenarios use different fractions of the observed effect and give event targets for 80% and 90% power. These are planning illustrations, not descriptions of the number of events observed in this trial.

The stated planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, a normal approximation, and an effect to detect set to the stated fraction of the observed effect. The original trial’s own design assumptions and sample-size justification are not reported in the registry record.

In plain language, the 80% power column describes a planning target that would yield an 80% chance of detecting the specified effect under the stated assumptions, if that effect were present and the assumptions held. The 90% power column represents a higher planned chance of detection under those same conditions. Neither column is a statement about the probability that the observed result is true.

The event targets increase as the assumed effect becomes smaller: a more modest difference is harder to distinguish from random variation, so more event information is needed. For time-to-event outcomes, that information is represented by events rather than a fixed number of participants alone. The number of participants required in practice would also depend on event occurrence and follow-up, which these planning figures do not specify. This is not observed power.

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 = 0.930, so ln(HR) = -0.0726.
2. d = 4 × (1.96 + 0.842)² / [ln(0.930)]² = 5,961.4, rounded up to 5,962 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 <- 0.93                              # 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(1721954)
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 = 5,962 (80% power), 7,981 (90% power)
 75% of observed effect: events = 10,598 (80% power), 14,188 (90% power)
 50% of observed effect: events = 23,846 (80% power), 31,923 (90% power)
Simulated power with 5,962 events: 0.803 (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 13,482 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 209 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 <- 25          # control-arm median, months (converted from the posted unit)
hr          <- 0.93          # 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: 12022         p (1-sided)   0.0001
 Events: 2692        ~HR at bound   0.8667
    Month: 21    P(Cross) if HR=1   0.0001
              P(Cross) if HR=0.93   0.0338
    IA 2: 67%                   Z   2.5114
     N: 13482         p (1-sided)   0.0060
 Events: 5383        ~HR at bound   0.9338
    Month: 35    P(Cross) if HR=1   0.0060
              P(Cross) if HR=0.93   0.5603
        Final                   Z   1.9930
     N: 13482         p (1-sided)   0.0231
 Events: 8075        ~HR at bound   0.9566
    Month: 60    P(Cross) if HR=1   0.0250
              P(Cross) if HR=0.93   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 registry reports the sponsor’s overall-survival analysis as a log-rank hazard ratio of 0.93 with p=0.43. The posted group medians and their confidence intervals provide descriptive context for that comparison, while the hazard ratio and log-rank test summarize the time-to-event analysis.

The registry record does not identify a separate recalculated overall-survival estimate that differs from the sponsor-reported result. The sponsor’s analysis is therefore the relevant reported inferential result, and there is no reported discrepancy to reconcile. The confidence interval for the hazard ratio is not reported, so the registry does not provide a direct interval-based comparison of the relative-effect estimate’s precision.

The planning illustrations use the observed effect as an input for hypothetical future-study requirements. They are not a reanalysis of the trial’s observed outcome and do not replace the sponsor’s reported analysis. Their purpose is to show how planning targets change under different assumed effect sizes and power levels, not to revise the primary trial conclusion.

16. Multiplicity, Missing Data and Analysis Populations

The registry reports overall survival as the primary outcome and progression-free survival as a secondary outcome. It also reports analyses for these outcomes, but does not state whether the hypotheses were tested under a prespecified hierarchy or whether an adjustment for multiple comparisons was applied. When several outcomes or analyses are considered, the chance of at least one apparently positive result can increase. Without the analysis plan, the reported secondary p-value cannot be interpreted as though it were necessarily the only additional test conducted.

The primary outcome population is identified as ITT. The record does not explain the exact operational definition of that population, how participants with missing endpoint information were handled, or whether any participants were excluded from the primary analysis. It also does not describe censoring rules, loss to follow-up, or sensitivity analyses for missing data. These details are important in time-to-event studies because incomplete follow-up can affect the amount and pattern of information contributing to the estimate.

The registry identifies the study as randomized and open label, but does not report the randomization procedure, allocation concealment, or blinded outcome assessment. These omissions do not establish that the procedures were inadequate; they mean that readers cannot assess them from the registry record.

Other statistical details are also not reported in the registry record, including the original sample-size rationale, interim-analysis plans, the full statistical analysis plan, and checks of assumptions underlying the hazard-ratio analysis. The posted results can be described as reported, but the missing information limits independent assessment of whether all analyses followed prespecified procedures and how robust the findings are to alternative reasonable methods.

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: febrile neutropenia (10.9%), abdominal pain (9.8%), pneumonia (6.5%), diarrhoea (5.4%), pulmonary embolism (5.4%).

The registry lists serious adverse event information for mFOLFOX6 Plus SIRT and mFOLFOX6 Alone, with separate counts for those at risk and those affected. For serious adverse events, the record lists 92 at risk and 47 affected in the mFOLFOX6 Plus SIRT group, and 112 at risk and 47 affected in the mFOLFOX6 Alone group. It also lists 58 and 64 deaths, respectively. The registry provides a separate set of at-risk and affected counts for other adverse events.

The most frequent serious adverse event terms listed include febrile neutropenia at 10.9%, abdominal pain at 9.8%, pneumonia at 6.5%, diarrhoea at 5.4%, pulmonary embolism at 5.4%, and dehydration at 4.3%. The registry excerpt does not make clear the denominator and event-collection details for each listed percentage, or provide a complete account of severity, timing, duration, or attribution.

These figures are safety descriptions, not a randomized statistical comparison with uncertainty estimates. They should not be treated as proof that event risks are equal or different between groups. The registry does not report a comparative safety analysis or enough detail to assess how adverse events were collected and classified.

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

The registry result answers a specific statistical question: how overall survival was compared between randomized assignment to mFOLFOX6m plus SIR-Spheres and assignment to mFOLFOX6m alone in this study population. The reported medians are close in numerical value, and the log-rank comparison gives a hazard ratio of 0.93 with p=0.43. The analysis does not provide strong statistical evidence of a difference in overall survival.

That interpretation is not the same as proving that the strategies have identical effects. A non-significant result can arise when the true effect is small, when information is limited, or through random variation. The absent hazard-ratio confidence interval makes it especially difficult to judge which relative effects remain compatible with the primary result. Equivalence or non-inferiority would require appropriate design criteria and analyses; the registry does not report such a conclusion.

The posted secondary progression-free survival result has a different pattern of statistical evidence. Because it is secondary and the record does not report multiplicity handling, it should not be treated as a replacement for the primary overall-survival result. Different endpoints also capture different aspects of the disease course, so they need not yield matching statistical findings.

The results apply to the trial’s randomized comparison and recorded follow-up, not automatically to every clinical setting or patient. The study was open label, bevacizumab use was discretionary, and important details about follow-up, missing data, analysis procedures, and baseline characteristics are not reported in the registry record. The findings should therefore be read as a report of the available trial analysis rather than a complete account of all evidence about the regimens.

Nothing in these statistics determines what treatment an individual should receive. They describe group-level estimates and their reported uncertainty, and they do not ensure a particular outcome for an individual participant.

19. Why This Trial Matters Statistically

This trial illustrates why a clinically direct endpoint and a relative-effect analysis can tell complementary parts of a statistical story. The medians provide interpretable time summaries, while the hazard ratio and log-rank test use the timing of events over follow-up. None of these statistics should be treated as a complete description on its own.

It also illustrates the distinction between absence of strong evidence for a difference and evidence that two strategies are equivalent. The primary p-value is 0.43, but the hazard-ratio confidence interval is not reported. Without an interval, readers have less direct information about the precision and range of plausible relative effects than they would with a complete estimate-and-interval report.

The contrast between the primary and secondary results reinforces the importance of prespecification and multiplicity. A secondary endpoint can provide useful information, but its evidential weight depends in part on whether it was planned, how many tests were conducted, and whether any adjustment was used. The registry does not report those details.

Finally, a randomized design is valuable but does not eliminate the need for transparent reporting. Information about randomization procedures, outcome handling, missing data, censoring, and analysis assumptions helps readers judge the credibility and portability of an estimate. The posted record gives a clear outline of the treatment comparison and main results, while leaving several such details unreported.

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 reports the primary hazard ratio and p-value but does not report a confidence interval for the hazard ratio. This limits direct assessment of the precision of the relative-effect estimate. The median confidence intervals describe uncertainty around the separate medians, not around the between-group hazard ratio.

The record does not provide the full survival curves, numbers of deaths included in the primary analysis, follow-up distribution, or a detailed account of censoring. It also does not report whether the proportional-hazards assumption was assessed. These omissions constrain interpretation of a single hazard ratio over time.

The study was open label, and the registry does not describe allocation concealment or blinded outcome assessment. The record also does not explain how missing outcomes, protocol deviations, or analysis-population details were handled, despite identifying the primary population as ITT. These are reporting limitations; they do not establish that the trial procedures were deficient.

Bevacizumab was allowed at investigator discretion where it formed part of institutional standard care. The record does not describe how often it was used in each arm, so the treatment contrast cannot be separated into effects attributable to the study regimens and any variation in bevacizumab use.

The registry does not report the original sample-size justification, interim-analysis plans, multiplicity procedures, or the full statistical analysis plan. Safety summaries also lack detail about event ascertainment and comparative analysis. These trial-specific gaps mean the posted results cannot support a complete independent appraisal of every design and analysis decision.

21. Related Tutorials

22. Related Statistical Calculators

23. Sources

• ClinicalTrials.gov: FOLFOX6m Plus SIR-Spheres Microspheres vs FOLFOX6m Alone in Patients With Liver Mets From Primary Colorectal Cancer — NCT01721954.
• Registry record last updated 2019-11-05; status last verified 2019-10.
• Results first posted 2019-10-22.
• 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

NCT01721954 is a completed randomized, open-label, parallel-group phase 3 study sponsored by Sirtex Medical. It compared first-line FOLFOX6m plus SIR-Spheres microspheres with FOLFOX6m alone in patients with non-resectable liver metastases from primary colorectal carcinoma. Bevacizumab could be used at investigator discretion where it was part of local standard care.

The primary outcome was overall survival in the ITT population. The registry reports median survival of 25.9 months and 25.0 months, with a log-rank hazard ratio of 0.93 and p=0.43. Progression-free survival was posted as a secondary outcome, with a reported hazard ratio of 0.75 and p<0.05. The registry does not report confidence intervals for either hazard ratio or several details needed for a full appraisal of the analysis, missing data, and multiplicity.