A Study of Trastuzumab Emtansine (T-DM1) Plus Pertuzumab/Pertuzumab Placebo Versus Trastuzumab [Herceptin] Plus a Taxane in Participants With Metastatic Breast Cancer (MARIANNE): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study of Trastuzumab Emtansine (T-DM1) Plus Pertuzumab/Pertuzumab Placebo Versus Trastuzumab [Herceptin] Plus a Taxane in Participants With Metastatic Breast Cancer (MARIANNE) (NCT01120184), a randomized phase 3 study. The primary outcome is Percentage of Participants With Death or Disease Progression According to Independent Review Facility (IRF) Assessment.

2. Clinical Question

3. Trial Design

MARIANNE (NCT01120184) was a randomized, parallel, three-arm, multicenter, phase III study. The registry reports an enrollment of 1095 and a completed status. The study began on 2010-07-31, with primary completion on 2014-09-30. Its stated purpose was to evaluate efficacy and safety as first-line treatment in participants with HER2-positive progressive or recurrent locally advanced or previously untreated metastatic breast cancer.

The three regimens were trastuzumab plus docetaxel or paclitaxel, T-DM1 plus pertuzumab, and T-DM1 plus pertuzumab-placebo. The brief summary describes the trastuzumab-plus-taxane arm as open-label and the other arms as blinded; the registry also lists triple masking. Further operational detail about masking is not reported in the registry record, so the labels should be read as registry descriptions rather than a complete account of who was masked to what.

The primary endpoint concerned death or disease progression assessed by an independent review facility, through the clinical cutoff of 16-Sept-2014, over a timeframe described as up to 48 months from randomization. Overall survival was assessed to a later clinical cutoff of 15-May-2016, over a timeframe described as up to 70 months. Assessment schedules and additional endpoint details are stated in the registry record, but the original statistical design assumptions are not reported in the registry record.

The posted primary outcome is expressed as a percentage of participants with the event, with an ITT Population identified for analysis. Group results are also posted for investigator-assessed progression, death before clinical cutoff, and overall survival. Those outcomes help describe the pattern across endpoints, but they do not substitute for the prespecified primary outcome or establish a broader treatment conclusion on their own.

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.

Population: Intent-to-Treat (ITT) Population: All participants randomized in the study.

6. Primary Endpoint

The registry lists the primary outcome as Percentage of Participants With Death or Disease Progression According to Independent Review Facility (IRF) Assessment, measured over Up to 48 months from randomization until clinical cutoff of 16-Sept-2014 (at Screening, every 9 weeks for 81 weeks, then every 12 weeks thereafter and/or up to 42 days after last dose).

The primary endpoint is the percentage of participants who died or experienced disease progression according to independent review facility assessment. An event could therefore be death or radiographic progression; the endpoint is not a measure of tumor shrinkage alone. Tumors were assessed using RECIST version 1.1, based on radiographic images submitted to the independent review facility. Progression was defined as a greater than or equal to 20 percent and 5-millimeter increase in the sum of target-lesion diameters, relative to the smallest sum recorded during the study, or the appearance of new lesions. A confirmatory assessment was included 4 to 6 weeks after study drug discontinuation.

This kind of combined endpoint is used to capture whether participants have died or had disease worsening during the assessment period. It can summarize clinically relevant events in a single proportion, but it does not show how much each component contributed. The posted percentage is calculated from participants with an event and the number analyzed. The registry identifies the ITT Population and gives a timeframe of up to 48 months from randomization until the clinical cutoff of 16-Sept-2014. The record does not report component-specific primary results, so the death and progression contributions cannot be separated from this endpoint result.

7. Results Posted in the Registry

• Trastuzumab + Taxane: 63.3% (231/365) versus 64.3% (236/367) on control; risk difference -1.0 percentage points (95% CI -8.0 to 5.9), risk ratio 0.98 (0.88–1.10); unadjusted two-proportion z-test p = 0.775.
• Trastuzumab Emtansine + Pertuzumab: 59.8% (217/363) versus 64.3% (236/367) on control; risk difference -4.5 percentage points (95% CI -11.6 to 2.5), risk ratio 0.93 (0.83–1.04); unadjusted two-proportion z-test p = 0.208.

8. Secondary Efficacy Endpoints

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

Percentage of Participants Who Died Prior to Clinical Cutoff

• 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:
  • Trastuzumab + Taxane: 46.3% (169/365) versus 47.7% (175/367) on control; risk difference -1.4 percentage points (95% CI -8.6 to 5.8), risk ratio 0.97 (0.83–1.13); unadjusted two-proportion z-test p = 0.708.
  • Trastuzumab Emtansine + Pertuzumab: 46.3% (168/363) versus 47.7% (175/367) on control; risk difference -1.4 percentage points (95% CI -8.6 to 5.8), risk ratio 0.97 (0.83–1.13); unadjusted two-proportion z-test p = 0.704.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Trastuzumab + Taxane 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 Trastuzumab Emtansine + Pertuzumab and the control. The interval gives the range of effects that remain compatible with the data.

Overall Survival (OS) at Clinical Cutoff

• 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:
  • Trastuzumab + Taxane vs Trastuzumab Emtansine + Placebo: hazard ratio 0.93 (95% CI 0.73–1.20), reported p = 0.657 (as reported in the registry).
  • Trastuzumab + Taxane vs Trastuzumab Emtansine + Pertuzumab: hazard ratio 0.86 (95% CI 0.67–1.11), reported p = 0.569 (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 Trastuzumab + Taxane vs Trastuzumab Emtansine + Placebo and the control. The interval gives the range of effects that remain compatible with the data.
  • The 95% confidence interval includes no difference (a ratio of 1), so these data do not show a statistically significant difference between Trastuzumab + Taxane vs Trastuzumab Emtansine + Pertuzumab and the control. The interval gives the range of effects that remain compatible with the data.

Percentage of Participants With Death or Disease Progression According to Investigator Assessment

• 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:
  • Trastuzumab + Taxane: 72.1% (263/365) versus 70.3% (258/367) on control; risk difference 1.8 percentage points (95% CI -4.8 to 8.3), risk ratio 1.02 (0.93–1.12); unadjusted two-proportion z-test p = 0.600.
  • Trastuzumab Emtansine + Pertuzumab: 67.5% (245/363) versus 70.3% (258/367) on control; risk difference -2.8 percentage points (95% CI -9.5 to 3.9), risk ratio 0.96 (0.87–1.06); unadjusted two-proportion z-test p = 0.413.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Trastuzumab + Taxane 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 Trastuzumab Emtansine + Pertuzumab and the control. The interval gives the range of effects that remain compatible with the data.

Across the secondary outcomes, the posted results do not present a consistent pattern of a clearly separated treatment effect. The death-before-cutoff percentages are similar across the three arms, while investigator-assessed death or progression percentages also sit in a relatively close range. Overall-survival medians and reported hazard ratios likewise do not show a clear separation in the stated comparisons. These secondary outcomes provide context for the primary finding, but they do not override it: they use distinct outcome definitions, assessment methods, follow-up periods, and, for survival, time-to-event analyses.

Concordance across related measures can be informative, but results are not interchangeable. Independent review and investigator assessment may differ, and a death endpoint does not include progression in the same way as the primary composite. Overall survival includes follow-up to a later cutoff than the primary endpoint. The registry record also lists other secondary outcomes, but does not post comparative analyses for all of them in the details. The number of posted outcome measures is 37 secondary measures, with 39 outcome measures and 26 analyses listed in the record; those counts do not establish how many hypotheses were formally tested or how they were ordered.

Unadjusted secondary results carry less weight because examining multiple outcomes and comparisons increases the chance of finding an apparently notable result by chance. In general, multiplicity procedures can help control that risk, but whether such procedures were specified or used for these outcomes is not reported in the registry record. Secondary findings are best treated as supporting or exploratory context unless the analysis plan and multiplicity handling are available. The results should be interpreted with their estimates, precision, and analysis methods, rather than by selecting an isolated percentage or p-value.

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

• Percentage of Participants Who Died Prior to Clinical Cutoff
• Overall Survival (OS) at Clinical Cutoff
• Percentage of Participants With Death or Disease Progression According to Investigator Assessment
• PFS According to Investigator Assessment
• Percentage of Participants Experiencing Treatment Failure
• Time to Treatment Failure (TTF)
• One-Year Survival Rate
• Percentage of Participants With Grade ≥3 Adverse Events

9. Statistical Interpretation

10. Statistical Methods Explained

The primary outcome is a binary event measure: each analyzed participant either had death or disease progression during the defined assessment period, or did not. The registry reports the percentage with an event in each arm. In general, comparing two such percentages can be expressed as a risk difference: the event proportion in one arm minus the event proportion in the other. A risk difference of zero would mean equal observed proportions; a negative value means the first arm in the stated comparison had a lower observed proportion. The interpretation depends on which arm is listed first, so the comparison labels matter.

What does the confidence interval describe?

In general, a confidence interval describes uncertainty around an estimate under the assumptions of the statistical method. It is not a range containing a fixed proportion of individual participant outcomes, nor is it a guarantee that the true effect lies inside the interval. For the primary recalculations, the intervals include zero, meaning the observed data do not sharply resolve whether the compared event proportions differ. The width also shows that estimates remain compatible with a range of possible differences. These are unadjusted recalculations from the posted summary statistics; they should not be read as replacements for a sponsor analysis. The registry record does not report a sponsor primary-endpoint analysis.

How should p-values be read?

In general, a p-value indicates how surprising results at least as inconsistent with a specified null hypothesis would be if that null hypothesis and the statistical model held. It is not the probability that the null hypothesis is true, and it does not measure the size or clinical importance of an effect. The primary comparison p-values are unadjusted recalculations. They do not account for the full collection of endpoints or comparisons, and the registry record does not report whether a multiplicity adjustment was planned or applied to the primary analyses. Consequently, the p-values should be read alongside the estimates, their confidence intervals, and the endpoint context.

Why are overall-survival estimates different?

Overall survival is a time-to-event outcome, so the record reports medians and hazard ratios rather than only a proportion at a single time. In general, a median survival time is the time by which half of the relevant population is estimated to have experienced the event, under the analysis used. A hazard ratio compares event rates over follow-up under a time-to-event model; it is not a ratio of median times. A hazard ratio of 1 corresponds to no relative difference in the modeled event rate. The reported confidence intervals for both hazard ratios include 1, and the reported analyses use the log-rank method. The registry does not provide additional detail about the model assumptions or censoring procedures.

The median overall-survival interval for the T-DM1-plus-pertuzumab arm has no posted upper limit. That upper limit is not reported in the registry record, and it should not be inferred from the point estimate or the other arms. The hazard-ratio analyses are sponsor-reported results, with 97.5% confidence intervals and p-values stated in the record. Those reported analyses take precedence over any simple recalculation from summary medians.

What does independent review add?

Independent review facility assessment applies a review process to the submitted radiographic images, rather than relying only on investigator assessment. The registry also reports investigator-assessed progression as a secondary outcome. Differences between assessment methods can arise because readers may apply criteria or judgments differently, and because the analysis definitions and available assessments can differ. The record does not give a detailed adjudication process, reader agreement results, or the number of assessments that differed. In general, a blinded independent review can help limit some assessment bias, but it does not remove other sources of uncertainty.

The primary progression definition follows RECIST version 1.1 and specifies both an increase in target-lesion diameters and the appearance of new lesions as criteria. The combined primary endpoint also includes death. Because only the combined event percentage is posted for the primary outcome, it is not possible to determine from that result how many events were deaths versus disease progressions. The registry identifies the ITT Population, but does not report detailed rules for handling missing assessments, withdrawal, or other intercurrent events.

What does recalculated mean?

For binary outcomes, the posted counts and denominators allow simple unadjusted comparisons to be recalculated from summary statistics. Such calculations do not recreate any stratified, adjusted, or otherwise model-based sponsor analysis unless those methods are explicitly reported. The registry record does not report a sponsor analysis for the primary endpoint, so the displayed recalculated estimates provide a transparent comparison of posted summaries, not a claim about the trial's complete analysis plan. For overall survival, the record does report sponsor analyses; those analyses are the appropriate reported inferential results to cite.

General statistical interpretation also depends on study design, outcome definitions, follow-up, and analysis populations. Randomization supports comparison of the assigned arms, but does not by itself answer questions about adherence, missing outcome data, or generalizability. Details of those matters are not reported in the registry record. Therefore, estimates should be understood as descriptions of the posted trial analyses and not as proof of a universal effect.

11. Confidence Intervals and Precision

The primary risk-difference confidence intervals are wide enough to include zero. For trastuzumab plus taxane compared with T-DM1 plus pertuzumab-placebo, the recalculated estimate is -0.010, with a 95% confidence interval from -0.080 to 0.059. Compared with T-DM1 plus pertuzumab, the estimate is -0.045, with a 95% confidence interval from -0.116 to 0.025. In general, an interval that crosses zero leaves uncertainty about the direction of the difference. Its endpoints indicate that the posted summaries remain compatible with a range of values, rather than establishing a single precise effect.

Arm-level percentages are rounded in the registry record, while the recalculations use the posted event counts and denominators. Small differences between displayed arm percentages and a recalculated contrast can therefore reflect rounding. The record does not post confidence intervals for each primary arm percentage, so the uncertainty information is available through the recalculated comparison intervals rather than separate arm-level intervals.

For overall survival, the reported hazard-ratio confidence intervals extend across 1: 0.73 to 1.20 for trastuzumab plus taxane compared with T-DM1 plus pertuzumab-placebo, and 0.67 to 1.11 compared with T-DM1 plus pertuzumab. The record describes these as 97.5% confidence intervals. The median estimates also have uncertainty intervals; for the T-DM1-plus-pertuzumab arm, the upper confidence limit is not reported in the registry record. Confidence intervals quantify statistical precision under the analysis assumptions; they do not address unreported design choices, missing data, or applicability to other populations.

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 original trial design. For binary endpoints, the figures are participants per arm. For the time-to-event overall-survival calculations, they are total events. The stated planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation; the effect to detect is the stated fraction of the observed effect. The original sample-size justification and design assumptions are not reported in the registry record.

The 80% power column gives the modeled number of participants per arm, or total events for the time-to-event calculations, associated with an 80% chance of detecting the specified effect under those assumptions, if that effect and model assumptions hold. The 90% power column gives the corresponding planning amount for a 90% chance. These are prospective planning quantities, not evidence about the probability that the trial result is correct.

Each row uses a different fraction of the observed effect: 100%, 75%, or 50%. The smaller the effect one wishes to detect, the more participants or events are needed under the same assumptions. The figures can become large when the observed contrast is small, but they should not be interpreted as a proposed redesign of MARIANNE. They are not observed power: observed power is not an appropriate substitute for reading the effect estimate and its confidence interval. The trial's actual enrollment and the planning illustrations answer different questions, and the registry does not report that these illustrations were used to determine the trial's original enrollment.

13. How the Sample Size Is Derived

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

Formula

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

Worked calculation for this trial

1. Control proportion p₀ = 0.643; treatment proportion p₁ = 0.633; average p̄ = 0.638.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.6797 = 1.3321.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.6796 = 0.5720.
4. n = (1.3321 + 0.5720)² / (-0.0102)² = 35,018.5, rounded up to 35,019 participants per arm for 80% power.

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

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
p0 <- 0.643051771117166                          # control proportion
p1 <- 0.632876712328767                          # treatment proportion
n_per_arm <- function(power, frac = 1) {
  q1 <- p0 + (p1 - p0) * frac               # smaller effect = fraction of the observed difference
  pbar <- (p0 + q1) / 2
  z_b <- qnorm(power)
  ceiling(round((z_a * sqrt(2 * pbar * (1 - pbar)) + z_b * sqrt(p0 * (1 - p0) + q1 * (1 - q1)))^2 / (q1 - p0)^2, 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

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

# Simulation: how often does the z-test reject at the 80%-power sample size?
set.seed(1120184)
n    <- n_per_arm(0.80)
reps <- 20000
x0 <- rbinom(reps, n, p0); x1 <- rbinom(reps, n, p1)
ph <- (x0 + x1) / (2 * n)
z  <- ((x1 - x0) / n) / sqrt(ph * (1 - ph) * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(z) > z_a, na.rm = TRUE), reps))

Output from running the code

100% of observed effect: n per arm = 35,019 (80% power), 46,880 (90% power)
 75% of observed effect: n per arm = 62,161 (80% power), 83,216 (90% power)
 50% of observed effect: n per arm = 139,648 (80% power), 186,949 (90% power)
power.prop.test: 35018.5 
Simulated power at n = 35,019 per arm: 0.802 (20000 replicates)

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

14. Multiplicity, Missing Data and Analysis Populations

The registry record lists many outcome measures and analyses, including multiple secondary endpoints and comparisons between treatment arms. Testing many hypotheses can increase the chance that at least one result appears statistically notable by chance. In general, a multiplicity plan may define which outcome is primary, how secondary outcomes are ordered, and whether significance thresholds or confidence intervals are adjusted across tests. The registry identifies a primary endpoint, but a complete multiplicity strategy, including any hierarchy or adjustment, is not reported in the registry record.

The unadjusted p-values for the primary binary comparisons are recalculated from posted summary statistics. They do not incorporate a correction for multiple comparisons. The overall-survival analyses are reported with 97.5% confidence intervals and log-rank p-values, but the reason for that confidence level and its place in any broader multiplicity plan are not reported in the registry record. Therefore, the p-values should not be interpreted as though the full family of outcomes and comparisons had been accounted for.

The primary endpoint identifies the ITT Population, but details about missing data and analysis populations are limited. The registry does not report how missing tumor assessments were handled, how participants without a confirmatory assessment were classified, or whether sensitivity analyses assessed alternative missing-data assumptions. The exact analysis population denominators can differ from enrollment and from safety denominators; the record gives endpoint-specific group sizes but does not explain every difference in the populations used.

For general interpretation, missing outcome data can matter if the chance of being observed is related to prognosis or treatment experience. The direction and size of any resulting bias cannot be determined without further information. The registry also does not report interim analyses, stopping rules, or a detailed statistical analysis plan. Those elements should not be inferred from the phase, randomization, or posted results. The available record supports description of the stated outcomes and analyses, but not reconstruction of unreported analysis decisions.

15. Safety Results

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

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 (3.7%), infusion related reaction (3.6%), anaemia (1.6%), neutropenia (1.4%), pyrexia (1.4%).

The serious adverse event figures are counts of participants with a serious adverse event among the stated number at risk in each arm. The record reports 81 among 353 for trastuzumab plus taxane, 86 among 361 for T-DM1 plus pertuzumab-placebo, and 93 among 366 for T-DM1 plus pertuzumab. These are descriptive safety counts; no comparative confidence intervals or p-values for these counts are reported in the registry record. They should not be treated as proof of a difference between arms.

The record also lists the most frequent serious adverse event terms, with febrile neutropenia at 3.7%, infusion related reaction at 3.6%, anaemia at 1.6%, neutropenia at 1.4%, pyrexia at 1.4%, and pneumonia at 1.4%. These term-specific percentages provide additional descriptive information, but the record does not give their arm-specific breakdown in the registry record. Other adverse events are listed as 342 among 353, 352 among 361, and 352 among 366, respectively. Safety death counts are not reported in the registry record.

Adverse-event summaries depend on definitions, observation periods, and the population assessed. Detailed rules for seriousness, attribution, duration, and handling of recurrent events are not reported in the registry record. Counts alone do not establish causation or determine comparative safety.

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

The posted primary results describe the proportions of participants in each arm who died or had disease progression according to independent review facility assessment. The recalculated unadjusted comparisons have confidence intervals that include zero. Thus, the registry results do not provide a clear statistical distinction in these pairwise primary comparisons. The estimates still allow a range of possible differences, so the findings should not be described as demonstrating that the regimens are equivalent.

Overall-survival results also require careful interpretation. The sponsor-reported hazard-ratio confidence intervals include 1, and the p-values are 0.657 and 0.569. These results do not establish a clear difference in the reported comparisons. The median estimates are descriptive and should not be compared as if they were hazard ratios. The upper confidence limit for the T-DM1-plus-pertuzumab median is not reported in the registry record.

Statistical uncertainty is not the same as clinical importance. The record does not state a threshold for a clinically meaningful difference, and the available results cannot determine whether a difference within the confidence intervals would matter to participants or clinicians. Nor does a non-significant result prove that there is no effect. Interpretation depends on the planned hypothesis, endpoint, analysis population, precision, and the completeness of follow-up and measurement.

These findings apply to the trial population and its specified regimens, endpoints, and observation periods. They do not establish results for other populations, treatment combinations, or clinical settings. The registry record is not a substitute for the full protocol and statistical analysis plan, which are not reported. No treatment recommendation follows from these statistical summaries.

17. Why This Trial Matters Statistically

MARIANNE is statistically informative because it directly compared three randomized treatment strategies in a phase III setting and reported both an independent-review primary endpoint and longer-term overall-survival outcomes. The primary endpoint combined death with disease progression assessed using RECIST version 1.1, while the secondary outcomes included investigator assessment and survival. Considering how endpoint definition and assessment method shape results is important when interpreting comparative trials.

The record also illustrates why an estimate and its precision should be considered together. Arm-level event percentages may look close, but the uncertainty intervals around the unadjusted contrasts extend across no difference. A single point estimate cannot show how much uncertainty remains. Likewise, overall-survival medians and hazard ratios answer related but different statistical questions, and the hazard-ratio intervals include 1.

Another statistical lesson is that a completed, randomized trial does not make every posted comparison definitive. Multiple endpoints and comparisons raise questions about multiplicity, and the registry does not report the complete plan for handling it. The same applies to missing data, interim analyses, and sample-size justification: these aspects affect interpretation but cannot be reconstructed from outcome summaries alone.

The planning illustrations emphasize the relationship between effect size and required information. Smaller effects require more participants or events to detect under the stated assumptions. Those calculations are not the original design justification and are not observed power. Overall, this trial is a useful example of reading endpoint definitions, comparisons, confidence intervals, reported analyses, and registry limitations together rather than relying on one number.

18. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.
• Participant counts were reconstructed from posted percentages and arm sizes, so small rounding differences are possible.

The registry record does not provide the original sample-size justification or enough information to reconstruct its assumptions. Planning figures based on observed effects are illustrations and cannot explain why enrollment was set at its reported level. The statistical analysis plan, including the full multiplicity strategy, is not reported in the registry record.

For the primary endpoint, the record gives event percentages and the ITT Population, but does not report sponsor inferential analyses. The comparisons described are unadjusted recalculations from posted summary statistics. The record does not provide detailed handling of missing radiographic assessments, incomplete follow-up, or participants who discontinued treatment. It also does not report the separate contributions of death and progression to the primary composite.

Some overall-survival information is incomplete: the upper confidence limit for the median in the T-DM1-plus-pertuzumab arm is not reported in the registry record. Safety results are descriptive, and comparative analyses or detailed attribution and event-handling methods are not reported. The record lists an open-label arm and blinded arms, alongside a triple-masking designation, without further operational detail. These registry-level gaps limit what can be inferred about assessment procedures and analysis decisions.

Finally, registry summaries do not provide all the detail needed to assess protocol deviations, adherence, censoring, baseline comparability, or sensitivity analyses. These items are not reported in the registry record. Their absence does not establish that they were not assessed; it means their results and methods cannot be evaluated from this record alone.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: A Study of Trastuzumab Emtansine (T-DM1) Plus Pertuzumab/Pertuzumab Placebo Versus Trastuzumab [Herceptin] Plus a Taxane in Participants With Metastatic Breast Cancer (MARIANNE) — NCT01120184.
• Registry record last updated 2024-03-04; status last verified 2024-02.
• Results first posted 2017-02-23.
• 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.

22. Record Summary

NCT01120184, MARIANNE, was a completed randomized, parallel, multicenter, phase III study sponsored by Hoffmann-La Roche. It enrolled 1095 participants with HER2-positive progressive or recurrent locally advanced or previously untreated metastatic breast cancer. The three treatment arms were trastuzumab plus a taxane, T-DM1 plus pertuzumab-placebo, and T-DM1 plus pertuzumab.

The primary outcome was the percentage with death or disease progression by independent review facility assessment, using RECIST version 1.1. The posted arm percentages were 63.3, 64.3, and 59.8, respectively. Recalculated unadjusted risk-difference confidence intervals included zero for both stated comparisons. Reported overall-survival hazard-ratio confidence intervals included 1. The record also reports secondary outcomes and descriptive adverse-event counts, but key details of multiplicity, missing-data handling, and the original statistical design are not reported in the registry record.