A Study to Evaluate the Efficacy and Safety of Trastuzumab Emtansine in Combination With Atezolizumab or Atezolizumab-Placebo in Participants With Human Epidermal Growth Factor-2 (HER2) Positive Locally Advanced or Metastatic Breast Cancer (BC) Who Received Prior Trastuzumab and Taxane Based Therapy: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study to Evaluate the Efficacy and Safety of Trastuzumab Emtansine in Combination With Atezolizumab or Atezolizumab-Placebo in Participants With Human Epidermal Growth Factor-2 (HER2) Positive Locally Advanced or Metastatic Breast Cancer (BC) Who Received Prior Trastuzumab and Taxane Based Therapy (NCT02924883), a randomized phase 2 study. The primary outcome is Progression-Free Survival (PFS) as Determined by Investigator's Tumor Assessment Using Response Evaluation Criteria in Solid Tumors (RECIST) Version 1.1 (v1.1).

2. Clinical Question

The trial asked whether adding atezolizumab to trastuzumab emtansine was associated with a different progression-free survival pattern than giving trastuzumab emtansine with placebo. The question concerned participants with HER2-positive locally advanced or metastatic breast cancer who had received prior trastuzumab and taxane-based therapy, either alone or in combination, and/or had progressed within 6 months after completing adjuvant therapy.

Participants were assigned to trastuzumab emtansine plus atezolizumab or trastuzumab emtansine plus placebo. Because both groups received trastuzumab emtansine, the randomized contrast concerns the addition of atezolizumab rather than trastuzumab emtansine versus no treatment. The primary outcome was progression-free survival assessed by investigators using RECIST v1.1. It measured the time from randomization until the first recorded disease progression or death from any cause, whichever occurred first.

The registry record reports median progression-free survival of 6.8 months for trastuzumab emtansine plus placebo and 8.2 months for trastuzumab emtansine plus atezolizumab. It also reports a hazard ratio of 0.82, with a 95% confidence interval from 0.55 to 1.23 and p = 0.333. The confidence interval includes 1, and the p-value does not provide strong evidence against the no-difference reference under the reported test. These results therefore describe the observed comparison but do not establish a precise treatment contrast. The registry record does not report the original design assumptions or a multiplicity plan.

3. Trial Design

This was a randomized, multicenter, double-blind, placebo-controlled Phase II study with a parallel-group design. The registry record lists enrollment of 202 participants and describes an intent-to-treat population that included all randomized participants, analyzed according to the treatment assigned at randomization. The primary progression-free survival results include 69 participants in the trastuzumab emtansine-plus-placebo group and 133 in the trastuzumab emtansine-plus-atezolizumab group.

The comparator group received trastuzumab emtansine with placebo matched to atezolizumab. The other group received atezolizumab or matching placebo followed by trastuzumab emtansine; its assigned treatment was trastuzumab emtansine plus atezolizumab. The registry describes trastuzumab emtansine as an intravenous infusion at 3.6 mg/kg on Day 1 of each 21-day cycle. It describes atezolizumab as 1200 mg by intravenous infusion. The treatment descriptions state that therapy continued until disease progression, unmanageable toxicity, or study termination, although the intervention description for the atezolizumab group is truncated in the registry record.

The study began on 2016-09-26, reached primary completion on 2017-12-11, and is listed as completed. The primary endpoint timeframe is baseline up to approximately 15 months. The registry lists overall survival among the secondary outcomes, with follow-up described as extending to study completion or death, whichever occurred first, approximately 40 months. The original recruitment target, allocation ratio rationale, and justification for the final enrollment are not reported in the registry record.

Blinding and placebo control are relevant because they can reduce the influence of treatment expectations on participant care and outcome assessment. Randomization supports a comparison based on assigned groups, while the intent-to-treat definition preserves that assignment for the primary population. The registry record does not describe the randomization sequence or the detailed procedures used to maintain masking.

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: The intent-to-treat (ITT) population included all randomized participants grouped according to the treatment assigned at randomization.

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 Progression-Free Survival (PFS) as Determined by Investigator's Tumor Assessment Using Response Evaluation Criteria in Solid Tumors (RECIST) Version 1.1 (v1.1), measured over Baseline up to approximately 15 months.

Progression-free survival is a time-to-event outcome. In this trial, it was defined as the time from randomization until the first occurrence of disease progression or death from any cause, whichever occurred first. Progression was based on investigator assessments using RECIST v1.1 and required at least a 20% increase in the sum of target-lesion diameters with an absolute increase of at least 5 mm, or the appearance of one or more new lesions.

This endpoint combines progression and death into a single event definition. It can therefore describe how long participants remain alive without a recorded progression under the specified assessment process. The use of a time-to-event measure retains information about when events occur and can account for participants whose event has not been observed by the end of follow-up through censoring methods. The registry identifies the endpoint definition and the Log Rank method for the reported comparison, but detailed censoring rules and the handling of missed or delayed assessments are not reported in the registry record.

Median progression-free survival is the time by which half of the estimated event-time distribution has experienced the event, under the analysis method. It is a summary of the distribution, not a statement that each participant experienced that duration. A median alone does not capture the whole pattern of event timing, which is why the reported hazard ratio and its confidence interval also matter. A hazard ratio summarizes a relative event rate over follow-up under the model or analysis interpretation used; it is not a direct comparison of median times and does not by itself describe absolute time gained.

Progression-free survival is commonly used in cancer trials because it incorporates disease progression as well as death and can be observed before overall survival is fully characterized. That general rationale does not establish that it is a substitute for overall survival in every setting. The registry record does not report the rationale used to select this outcome as primary.

7. Results Posted in the Registry

• Trastuzumab Emtansine + Placebo vs Trastuzumab Emtansine + Atezolizumab: hazard ratio 0.82 (95% CI 0.55–1.23), reported p = 0.333 (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.

Overall Survival (OS)

• 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 Emtansine + Placebo vs Trastuzumab Emtansine + Atezolizumab: hazard ratio 0.74 (95% CI 0.42–1.30), reported p = 0.293 (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 Emtansine + Placebo vs Trastuzumab Emtansine + Atezolizumab and the control. The interval gives the range of effects that remain compatible with the data.

Percentage of Participants With Objective Response (OR) as Determined by Investigator's Tumor Assessment Using RECIST v1.1

• 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 Emtansine + Atezolizumab: 45.5% (60/132) versus 43.5% (30/69) on control; risk difference 2.0 percentage points (95% CI -12.5 to 16.4), risk ratio 1.05 (0.75–1.45); unadjusted two-proportion z-test p = 0.789.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Trastuzumab Emtansine + Atezolizumab and the control. The interval gives the range of effects that remain compatible with the data.

Duration of OR as Determined by Investigator's Tumor Assessment Using RECIST v1.1

• 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 Emtansine + Placebo vs Trastuzumab Emtansine + Atezolizumab: hazard ratio 1.26 (95% CI 0.52–3.03), reported p = 0.610 (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 Emtansine + Placebo vs Trastuzumab Emtansine + Atezolizumab and the control. The interval gives the range of effects that remain compatible with the data.

The secondary outcomes broaden the evidence beyond progression-free survival, but they address distinct questions. Overall survival concerns time to death; objective response concerns the proportion meeting response criteria; duration of response concerns the time a response persists. The registry also lists pharmacokinetic concentrations and anti-therapeutic antibodies. Those measures are not substitutes for clinical time-to-event or response outcomes.

Across the clinical secondary outcomes, the registry presents estimates with substantial uncertainty and does not show a uniformly conclusive pattern that would resolve the primary result. In particular, the posted analyses do not turn the uncertain primary comparison into a definite overall treatment conclusion. Some secondary comparisons are reported as time-to-event analyses, while the response outcome is a percentage. Differences in endpoint definition and analysis scale mean that the results cannot simply be combined as if they measured one common effect.

Secondary endpoints are often considered supportive or exploratory, especially when many outcomes are examined. Testing several outcomes creates more opportunities for apparently notable findings even if there is no underlying effect. Unless the analysis accounts for that multiplicity, individual p-values can overstate the strength of evidence when interpreted as confirmatory. The registry record does not report a multiplicity adjustment or a hierarchy governing secondary hypotheses.

Accordingly, secondary findings should be read in context: they can help describe the overall evidence and suggest questions for future research, but they do not independently establish benefit or negate the uncertainty in the primary outcome. A complete assessment would also depend on prespecified analysis details, missing-data handling, and the relationship among endpoints, several of which are not reported in the registry record. The registry lists 9 secondary outcomes, while only some secondary results are posted in the record.

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

• Overall Survival (OS)
• Percentage of Participants With Objective Response (OR) as Determined by Investigator's Tumor Assessment Using RECIST v1.1
• Duration of OR as Determined by Investigator's Tumor Assessment Using RECIST v1.1
• Maximum Serum Concentration (Cmax) of Trastuzumab Emtansine
• Cmax of Deacetyl Mercapto 1-Oxopropyl Maytansine (DM1)
• Cmax of Total Trastuzumab
• Cmax of Atezolizumab
• Percentage of Participants With Anti-therapeutic Antibodies (ATAs) to Atezolizumab

9. Statistical Interpretation

10. Statistical Methods Explained

The primary outcome is time-to-event, so the registry reports medians, confidence intervals, and a hazard ratio rather than only a count of participants with an event. These summaries address different aspects of the outcome. The median describes a point in the event-time distribution. The hazard ratio expresses a relative comparison of event rates as defined by the analysis. The confidence interval conveys the uncertainty around that estimate.

What does the Log Rank method compare?

In general, the Log Rank method compares the event experience of groups across follow-up, taking account of the timing of events and censoring. Censoring occurs when the event time is not observed for a participant during the available follow-up; the participant contributes information up to the point at which follow-up ends. This method is often used for time-to-event comparisons. The registry identifies Log Rank as the method for the reported comparison, but does not report further implementation details, censoring rules, or checks of the method's assumptions.

The registry reports the hazard ratio as 0.82, with a 95% confidence interval from 0.55 to 1.23, and p = 0.333 for the primary endpoint. The comparison is listed as trastuzumab emtansine plus placebo versus trastuzumab emtansine plus atezolizumab. A hazard ratio's direction depends on which group is in the numerator or reference position, so the contrast label should be retained when describing it. The interval includes 1, the value corresponding to equal relative hazards in the usual interpretation. The estimate is not a direct ratio of the posted medians.

How should medians be understood?

In general, a median event time is the time at which the estimated proportion remaining event-free reaches one half. It describes the center of the estimated distribution, not every participant's experience, and it does not show how the groups compare at every time point. The registry reports median progression-free survival of 6.8 months in the placebo-combination group and 8.2 months in the atezolizumab-combination group, each with a reported confidence interval. The intervals are 4.0 to 11.1 months and 5.8 to 10.7 months, respectively. Overlap between separate median intervals is not, by itself, a formal test of the treatment contrast; the reported comparison uses the Log Rank method and hazard ratio.

What does a confidence interval convey?

In general, a confidence interval is a range of values compatible with the data and the statistical procedure under its assumptions. It is not a guarantee that the true effect lies in that particular interval, and it does not quantify the probability that a treatment is effective. Wider intervals indicate less precision than narrower intervals on the same scale. For a hazard ratio, an interval containing 1 includes the no-relative-difference value. The primary interval spans values below and above that reference, indicating that the data do not pin down the direction and size of the relative comparison with precision.

How should the p-value be read?

In general, a p-value measures how unusual data at least as discrepant as those observed would be under a specified null model and analysis procedure. It is not the probability that the null hypothesis is true, nor is it the probability that the result arose by chance. The reported p-value for the primary comparison is 0.333. It does not provide strong evidence against the no-difference reference under the reported Log Rank test. It should be interpreted alongside the estimate, interval, outcome definition, and study context, rather than as a stand-alone verdict.

Why can over-dispersion matter for other outcome types?

For count outcomes, over-dispersion means the variation in the counts is greater than expected under a simple model. In general, ignoring it can make uncertainty estimates too small and produce overly confident inferences. The primary outcome is time-to-event, not a count endpoint, so over-dispersion is not the central issue in interpreting its reported Log Rank result. This general point is relevant when evaluating methods for other kinds of clinical outcomes, but the registry record does not describe an over-dispersion analysis for this trial.

The primary population is explicitly defined as intent-to-treat: all randomized participants grouped according to assigned treatment. This preserves the randomized assignment for the primary analysis. It does not, on its own, explain how incomplete follow-up was handled or how participants with missing assessments contributed to the analysis. The registry record does not report those details, nor does it report whether a per-protocol or other analysis population was used for additional primary analyses.

The registry also lists an objective-response outcome as a percentage and response duration as a time-to-event outcome. These measures answer different questions from progression-free survival. A response proportion describes whether an objective response was recorded under the assessment criteria; response duration concerns how long a recorded response lasted. Overall survival concerns time until death. These outcomes can complement the primary endpoint but cannot be treated as interchangeable, and their analysis methods and multiplicity handling are not fully described in the registry record.

Statistical methods not stated in the registry record should not be inferred from the reported estimates. The original sample-size justification, assumptions about event accrual, interim analyses, detailed missing-data procedures, and prespecified multiplicity strategy are not reported in the registry record. The sample-size figures associated with the posted results are planning illustrations based on observed effects, not evidence of the trial's original design or an analysis of achieved power.

11. Confidence Intervals and Precision

Precision is reflected in how narrowly an estimate is bounded by its confidence interval. For the primary hazard ratio, the registry reports 0.82 with a 95% confidence interval from 0.55 to 1.23. The interval extends to values below and above 1, so the posted result allows for different relative event-rate patterns. It does not isolate a single direction or a narrowly estimated magnitude. The p-value, 0.333, is consistent with the lack of strong evidence against the no-difference reference under the reported Log Rank test.

The median estimates also have uncertainty. For trastuzumab emtansine plus placebo, the median is 6.8 months with a 95% confidence interval from 4.0 to 11.1 months. For trastuzumab emtansine plus atezolizumab, the median is 8.2 months with a 95% confidence interval from 5.8 to 10.7 months. These intervals communicate uncertainty in each group’s median estimate. They do not directly provide the confidence interval for a difference between the medians.

Confidence intervals are generally more informative than a binary significant-or-not label because they display a range of estimates consistent with the data under the analysis. They should not be interpreted as a probability statement about the true effect. Precision also depends on the amount and timing of event information, follow-up, and analysis assumptions. The number of primary events and detailed follow-up distribution are not reported in the registry record, limiting further assessment of why the interval has its reported width.

The uncertainty matters for interpretation: a point estimate alone can appear more conclusive than the full range supports. Conversely, an interval that includes the no-difference reference does not prove equivalence. No equivalence margin or noninferiority framework is reported in the registry record, so the results should not be described as establishing that the groups are equivalent.

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 describe hypothetical sample sizes under stated assumptions; they are not a reconstruction of the trial's original design. For time-to-event comparisons, the figures are total event counts. For the other outcomes, they are participants per arm. The 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 time-to-event calculations use the Schoenfeld approach.

The 80% power column describes the planned number of events or participants needed for an analysis to have an 80% chance of detecting the specified effect if that effect were the true one, under the planning assumptions. The 90% power column describes the corresponding target for a 90% chance. These are planning concepts, not guarantees about a future study's result. The figures are conditional on the assumed effect size and statistical framework.

Rows referring to 100% of the observed effect use the full observed effect as the planning target; rows for 75% or 50% use smaller target effects. Smaller effects are harder to distinguish from random variation, so the illustrations require more participants or events as the target effect becomes smaller. This relationship is a general feature of statistical planning, not evidence that the effect will be that size in another study.

The event-based totals should not be read as participant counts, since participants may be followed for different lengths of time and not all will experience the event during follow-up. Conversely, the per-arm figures for non-time-to-event outcomes refer to participants assigned to each group. The registry record's actual enrollment is 202, but the planning illustrations are based on observed effects rather than the original recruitment plan. These figures are not observed power.

The original protocol's sample-size assumptions and justification are not reported in the registry record. The planning figures therefore answer a different question from whether the original trial met its design targets. They illustrate what a future planning exercise would imply under the listed assumptions and should not be used to infer why enrollment or allocation took the form it did.

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 = 0.820, so ln(HR) = -0.1985.
2. d = 4 × (1.96 + 0.842)² / [ln(0.820)]² = 797.2, rounded up to 798 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 <- 0.82                              # 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(2924883)
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 = 798 (80% power), 1,068 (90% power)
 75% of observed effect: events = 1,418 (80% power), 1,898 (90% power)
 50% of observed effect: events = 3,189 (80% power), 4,269 (90% power)
Simulated power with 798 events: 0.800 (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 1,209 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 202 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 <- 6.8          # control-arm median, months (converted from the posted unit)
hr          <- 0.82          # 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: 794         p (1-sided)   0.0001
  Events: 360        ~HR at bound   0.6762
    Month: 16    P(Cross) if HR=1   0.0001
              P(Cross) if HR=0.82   0.0338
    IA 2: 67%                   Z   2.5114
      N: 1210         p (1-sided)   0.0060
  Events: 720        ~HR at bound   0.8292
    Month: 25    P(Cross) if HR=1   0.0060
              P(Cross) if HR=0.82   0.5603
        Final                   Z   1.9930
      N: 1210         p (1-sided)   0.0231
 Events: 1079        ~HR at bound   0.8857
    Month: 60    P(Cross) if HR=1   0.0250
              P(Cross) if HR=0.82   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 a sponsor analysis for the primary time-to-event comparison using the Log Rank method. Its reported hazard ratio is 0.82, with a 95% confidence interval from 0.55 to 1.23 and p = 0.333. Because the primary outcome is time-to-event, the displayed interpretation uses the reported hazard ratio rather than attempting to reconstruct it from the two median values. Medians alone do not contain the event-time information needed to reproduce a hazard ratio.

The recalculation convention for time-to-event results is to use the sponsor's reported hazard ratio. The registry record indicates no discrepancy between the posted analysis and the available re-calculated presentation. There is therefore no alternative summary-statistic calculation that should displace the reported analysis. If a discrepancy were present, the sponsor's own analysis would take precedence, but no such discrepancy is indicated for this result.

The sponsor-reported estimate remains uncertain: its confidence interval includes 1, and the reported p-value does not provide strong evidence against the no-difference reference under the Log Rank test. This interpretation is not changed by relabeling or re-expressing the result. A recalculation from summary data, where available for other outcome types, would be an unadjusted calculation and would not replace a prespecified analysis. The original statistical analysis plan and details beyond the reported method are not reported in the registry record.

16. Multiplicity, Missing Data and Analysis Populations

Multiplicity arises when several outcomes, comparisons, time points, or analyses are tested. In general, the more opportunities there are to find an apparently unusual result, the greater the chance that at least one result will look persuasive by chance alone. Statistical plans can address this through prespecified testing sequences, adjusted significance thresholds, or other procedures. The registry record does not report a multiplicity adjustment, a hierarchy for testing the secondary outcomes, or whether any findings were designated exploratory.

The primary progression-free survival comparison is clearly identified, and the registry reports the intent-to-treat population for that endpoint. The stated population includes all randomized participants analyzed according to assigned treatment. This is a useful safeguard against changing group membership based on post-randomization events. However, the registry record does not report the detailed analysis plan, including whether other analysis populations were defined or how deviations from assigned treatment were treated in additional analyses.

Missing outcome data can affect interpretation, especially for time-to-event outcomes where follow-up may end before progression or death is observed. General methods account for such participants through censoring, but valid interpretation depends on the censoring and follow-up mechanisms and on the analysis assumptions. The registry record does not report numbers of participants censored for the primary outcome, reasons for censoring, missing assessment patterns, or sensitivity analyses addressing incomplete follow-up. It also does not report how missing data were handled for the response outcome.

The registry specifies that progression was assessed by investigators using RECIST v1.1, but does not report details such as assessment schedules, independent review, or adjudication procedures. Nor are interim analyses, stopping rules, or the original statistical analysis plan reported. Their absence from the registry record should not be taken as evidence that such procedures were or were not used. It means the posted information is insufficient to evaluate them.

These reporting gaps do not invalidate the results that are posted, but they limit how completely a reader can assess robustness and confirmatory status. The primary estimate and uncertainty can be interpreted as reported; stronger claims about analysis choices, missingness, or family-wise error control would require information not reported in the registry record.

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: pyrexia (7.5%), pneumonia (3.0%), abdominal pain (3.0%), sepsis (3.0%), urinary tract infection (3.0%).

The registry lists serious adverse events and other adverse events by treatment group. For serious adverse events, the reported affected counts are 52 of 133 participants in the trastuzumab emtansine-plus-atezolizumab group and 16 of 67 in the trastuzumab emtansine-plus-placebo group. For other adverse events, the reported counts are 130 of 133 and 62 of 67, respectively. The serious-event denominators differ from the primary outcome group sizes; the reason for that difference is not reported in the registry record.

The record also lists deaths of 42 and 22 in the respective groups. It does not establish from those counts alone whether deaths were caused by treatment, related to disease, or occurred within a particular risk window. The definition and attribution of these deaths are not reported in the registry record.

The most frequently listed serious-event terms include pyrexia at 7.5%, and pneumonia, abdominal pain, sepsis, and urinary tract infection at 3.0% each; vomiting is listed at 2.3%. These entries describe reported terms and percentages, not a complete safety profile. Comparisons of adverse-event counts should account for group sizes, follow-up duration, event definitions, and attribution procedures, which are not fully described in the registry record. These figures are descriptive and do not by themselves establish that a treatment caused an event or indicate an individual's risk.

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

The result means that the registry record reports a primary progression-free survival comparison between randomized groups, with a hazard ratio of 0.82 and a 95% confidence interval from 0.55 to 1.23. The interval includes the no-relative-difference reference, and the reported p-value is 0.333. The posted evidence does not precisely determine the comparative effect under the reported analysis.

The medians describe the observed progression-free survival summaries for each group, but they should not be interpreted as durations guaranteed for an individual participant. Nor do they by themselves establish that adding atezolizumab changed the course of disease. The data do not prove that the groups are identical, and the analysis is not described as an equivalence or noninferiority study. The distinction between a result that does not demonstrate a clear difference and a result that proves no difference is important.

The endpoint reflects investigator-assessed progression or death according to the registry's definition. It does not directly measure every aspect of health, quality of life, toxicity, or long-term survival. Secondary outcomes add other perspectives, but they have different meanings and do not eliminate the uncertainty in the primary endpoint. Safety counts describe observed events and do not determine causation.

The record also leaves important statistical details unreported, including original sample-size justification, multiplicity procedures, and details of missing outcome handling. Those gaps limit assessment of design intentions and robustness, but they do not justify filling in assumptions. The most defensible reading is restricted to the posted estimates, their uncertainty, the stated outcome definitions, and the listed analysis method.

These results apply to the randomized population described in this study and should not automatically be generalized to people with different disease characteristics or treatment histories. They are a statistical description of a particular trial comparison, not medical advice and not a recommendation about treatment.

19. Why This Trial Matters Statistically

This trial is statistically informative because it makes a randomized, double-blind comparison of two regimens that share trastuzumab emtansine and differ by atezolizumab versus placebo. That design gives a direct framework for describing the effect associated with the added treatment component, while blinding and placebo control can reduce some sources of assessment and expectation bias. The intent-to-treat definition retains participants according to randomized assignment for the primary analysis.

The primary endpoint is a time-to-event measure, which requires interpretation of both the event-time summary and the relative comparison. The medians and hazard ratio communicate different features of the data, and the confidence interval shows why the point estimate should not be treated as decisive. The reported interval spans the no-difference reference, illustrating the value of presenting uncertainty rather than relying on a single estimate or a threshold label.

The trial also illustrates why secondary outcomes and safety counts need careful context. Outcomes differ in their definitions and analysis scales; examining many outcomes without a reported multiplicity strategy weakens the confirmatory weight of individual findings. Safety counts are important descriptive information but do not establish causality. The record's limited reporting on missing data, censoring, interim analyses, and original design assumptions further shows why registry results should be interpreted with attention to what is documented and what remains unknown.

Finally, the sample-size illustrations demonstrate how planning targets depend on the assumed effect and power goal. They are distinct from the original design rationale and from observed trial evidence. Separating planning calculations from trial results, and sponsor analyses from simple re-calculations, helps prevent apparent precision or power from being mistaken for a demonstrated treatment effect.

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.

Several limitations specific to this record constrain interpretation. The primary result is imprecise, with a hazard-ratio confidence interval that includes 1. The registry reports the medians and comparison but does not provide the primary event count or the detailed survival estimates over follow-up. The relative comparison therefore cannot be independently evaluated across the full event-time pattern from the registry information alone.

The registry identifies an intent-to-treat population, but does not report detailed censoring rules, missing assessment patterns, sensitivity analyses, or reasons for loss of follow-up. These details would help assess whether incomplete observation could influence the time-to-event estimates. The record also does not report the original sample-size justification, interim analyses, or complete statistical analysis plan. These omissions prevent evaluation of whether all analyses and decision rules followed prespecified procedures.

Multiplicity handling is not reported despite the number of secondary outcomes. This limits the weight that can be placed on isolated secondary findings as confirmatory evidence. Some secondary outcomes have results that are not fully reported, and overall survival medians are not reported in the registry record. The objective-response analysis includes 132 participants in the atezolizumab group, whereas the primary outcome lists 133; the reason for this difference is not reported in the registry record.

The safety record provides counts and selected serious-event terms, but does not report comprehensive event definitions, attribution, duration, or follow-up exposure details. The death counts alone do not establish cause or treatment relationship. In addition, the treatment description for the atezolizumab arm is truncated in the registry record. These limitations concern the information available for this interpretation and should not be taken as proof that a procedure or analysis was absent from the full study materials.

21. Related Tutorials

22. Related Statistical Calculators

23. Sources

• ClinicalTrials.gov: A Study to Evaluate the Efficacy and Safety of Trastuzumab Emtansine in Combination With Atezolizumab or Atezolizumab-Placebo in Participants With Human Epidermal Growth Factor-2 (HER2) Positive Locally Advanced or Metastatic Breast Cancer (BC) Who Received Prior Trastuzumab and Taxane Based Therapy — NCT02924883.
• Registry record last updated 2021-02-17; status last verified 2021-01.
• Results first posted 2019-02-12.
• 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

NCT02924883 is a completed, randomized, double-blind, placebo-controlled Phase II study of trastuzumab emtansine with atezolizumab or placebo in participants with HER2-positive locally advanced or metastatic breast cancer after prior trastuzumab and taxane-based therapy. The registry lists enrollment of 202. The primary endpoint was investigator-assessed progression-free survival using RECIST v1.1.

The registry reports median progression-free survival of 6.8 months with trastuzumab emtansine plus placebo and 8.2 months with trastuzumab emtansine plus atezolizumab. The reported hazard ratio is 0.82, with a 95% confidence interval from 0.55 to 1.23 and p = 0.333. The interval includes the no-relative-difference reference, so the comparison remains uncertain.

Secondary outcomes include overall survival, objective response, response duration, pharmacokinetic measures, and anti-therapeutic antibodies. Safety results are descriptive. Multiplicity procedures, detailed missing-data methods, and the original sample-size rationale are not reported in the registry record.