A Study of Pemetrexed and Carboplatin/Cisplatin or Gemcitabine and Carboplatin/Cisplatin With or Without IMC-1121B in Participants Previously Untreated With Recurrent or Advanced Non-Small Cell Lung Cancer (NSCLC): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study of Pemetrexed and Carboplatin/Cisplatin or Gemcitabine and Carboplatin/Cisplatin With or Without IMC-1121B in Participants Previously Untreated With Recurrent or Advanced Non-Small Cell Lung Cancer (NSCLC) (NCT01160744), a randomized phase 2 study. The primary outcome is Progression-Free Survival (PFS).

2. Clinical Question

The study was designed to compare outcomes with and without IMC-1121B in first-line treatment of recurrent or advanced non-small cell lung cancer. The posted primary results include a separate comparison for each chemotherapy and histology grouping. They report median progression-free survival and sponsor-estimated hazard ratios, with confidence intervals and Log Rank p-values. These are distinct summaries: the medians describe the middle of the observed time-to-event distribution, while a hazard ratio compares event rates over follow-up under the method used for the analysis.

The registry record does not report a single pooled comparison across both histology groups. The results therefore need to be read within each regimen group; they do not establish that one chemotherapy backbone is better than another. The study was randomized and open-label, and the registry lists an enrollment of 280. Randomization supports comparison of assigned groups, but does not remove uncertainty in estimates or make the separate regimen comparisons interchangeable.

3. Trial Design

This was a randomized, open-label, multicenter, parallel-group Phase 2 study. It evaluated IMC-1121B, also known as ramucirumab, in combination with platinum-based chemotherapy against the corresponding chemotherapy without IMC-1121B. The registry lists 280 participants enrolled and describes the study as completed. The recorded study dates run from 2010-09 to primary completion in 2014-01.

The four treatment groups were organized by histology and chemotherapy backbone. For non-squamous disease, the groups received pemetrexed with carboplatin or cisplatin, with or without IMC-1121B. For squamous disease, the groups received gemcitabine with carboplatin or cisplatin, with or without IMC-1121B. The registry descriptions specify carboplatin AUC 6 with pemetrexed and AUC 5 with gemcitabine; cisplatin was an alternative in each regimen description. The record does not describe how the choice between carboplatin and cisplatin was made.

The primary endpoint results list 71 participants in the pemetrexed chemotherapy group and 69 in its IMC-1121B combination group. The gemcitabine chemotherapy group lists 69 participants, and its combination group lists 71. These are the group sizes posted for the primary outcome, while the intent-to-treat definition includes all randomized participants. Safety summaries list the numbers at risk separately for each safety group.

The treatment groups were not masked. That matters most for outcomes involving judgment or reporting, although progression was defined using RECIST v1.1 and the record gives specific censoring rules. The registry record does not report the original sample-size justification, randomization procedure details, allocation concealment, treatment duration, or interim-analysis plan. Those details should not be inferred from the fact that the study was randomized or from the posted enrollment.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Before assignment: Participants who died (any cause) or had disease progression were considered to be study completers.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Intent-to-Treat (ITT) Population: All randomized participants. Participants censored: Pem+Carb/Cis=14, Ram+Pem+Carb/Cis=13, Gem+Carb/Cis=14, Ram+Gem+Carb/Cis=18.

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: All randomized participants.

6. Primary Endpoint

The registry lists the primary outcome as Progression-Free Survival (PFS), measured over Randomization to PD or death (up to 24 months).

Progression-free survival (PFS) measures the time from randomization until the first objective disease progression or death from any cause, whichever occurs first. The registry defines progression using Response Evaluation Criteria in Solid Tumors, Version 1.1 (RECIST v1.1). The stated timeframe is randomization to progression or death, up to 24 months.

PFS is a time-to-event endpoint: it uses both whether an event occurred and when it occurred. A participant who remains alive without documented progression contributes follow-up until the last objective tumor assessment and is then censored. The record also specifies censoring at the last radiographic assessment for participants lost to follow-up without progression, and at randomization when no baseline or post-baseline radiologic assessment was available.

In general, time-to-event endpoints are useful when participants have different follow-up times or when some have not experienced the event by the time of analysis. Censoring allows their observed follow-up to contribute without treating an unobserved later outcome as if it were known. It relies on assumptions about the censoring process; the registry record does not report diagnostic checks of those assumptions.

The posted summaries include median PFS and sponsor-reported hazard ratios. A median describes the time by which half of the estimated event-time distribution has experienced an event; it does not describe every participant’s course. The hazard ratio summarizes relative event rates over follow-up under the analysis method. Neither summary, by itself, conveys all details of the survival curves or the timing of any differences.

7. Results Posted in the Registry

• Pem + Carb or Cis (Non-Squamous) vs Ram + Pem + Carb or Cis (Non-Squamous): hazard ratio 0.75 (95% CI 0.55–1.03), reported p = 0.132 (as reported in the registry).
• Gem + Carb or Cis (Squamous) vs Ram + Gem + Carb or Cis (Squamous): hazard ratio 0.88 (95% CI 0.64–1.22), reported p = 0.521 (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.

Percentage of Participants With Best Overall Response of Complete Response (CR) or Partial Response (PR) [Objective Response Rate (ORR)]

• Endpoint type: Binary: the percentage of participants with the event (counts rebuilt from the posted percentages and group sizes).
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Ram + Pem + Carb or Cis (Non-Squamous): 49.3% (34/69) versus 38.0% (27/71) on control; risk difference 11.2 percentage points (95% CI -5.1 to 27.6), risk ratio 1.30 (0.88–1.90); unadjusted two-proportion z-test p = 0.180.
  • Gem + Carb or Cis (Squamous): 24.6% (17/69) versus 38.0% (27/71) on control; risk difference -13.4 percentage points (95% CI -28.6 to 1.8), risk ratio 0.65 (0.39–1.08); unadjusted two-proportion z-test p = 0.088.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Ram + Pem + Carb or Cis (Non-Squamous) 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 Gem + Carb or Cis (Squamous) and the control. The interval gives the range of effects that remain compatible with the data.

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:
  • Pem + Carb or Cis (Non-Squamous) vs Ram + Pem + Carb or Cis (Non-Squamous): hazard ratio 1.03 (95% CI 0.74–1.42), reported p = 0.892 (as reported in the registry).
  • Gem + Carb or Cis (Squamous) vs Ram + Gem + Carb or Cis (Squamous): hazard ratio 0.93 (95% CI 0.68–1.27), reported p = 0.685 (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 Pem + Carb or Cis (Non-Squamous) vs Ram + Pem + Carb or Cis (Non-Squamous) 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 Gem + Carb or Cis (Squamous) vs Ram + Gem + Carb or Cis (Squamous) and the control. The interval gives the range of effects that remain compatible with the data.

The secondary outcomes add several perspectives beyond the primary PFS endpoint. Objective response rate captures the proportion with a complete or partial response, while duration of response focuses on how long responses last. Overall survival measures time to death from any cause. The safety-related outcome records treatment-emergent adverse events and deaths. These outcomes have different definitions, populations, and follow-up requirements, so a pattern across them should not be reduced to a single success-or-failure label.

The response-rate results provide information about tumor response, but response is not equivalent to remaining progression-free or living longer. Overall survival is a distinct time-to-event outcome and can be influenced by events after progression and by subsequent care, details of which are not reported in the registry record. Duration of response applies to participants who achieved a response; it therefore answers a conditional question and should not be interpreted as an outcome for every randomized participant.

Across the posted secondary endpoints, the numerical pattern is not uniformly aligned with the primary time-to-event comparisons. That contrast is a reason to examine endpoint definitions and uncertainty rather than select whichever result appears most favorable. A secondary endpoint can be informative even when its result differs from the primary endpoint, but it does not automatically overturn the primary finding.

Multiple outcomes and comparisons create opportunities for chance findings. Unless the analysis plan controls the overall false-positive risk or identifies particular secondary outcomes as confirmatory, unadjusted secondary p-values carry less evidential weight than a prespecified primary analysis. The registry record does not report a multiplicity strategy or which secondary analyses were intended to be confirmatory. The secondary results are therefore best read as supportive and exploratory context, not as a set of independent definitive tests.

The record lists seven outcome measures and ten analyses overall, while four secondary outcomes are posted. That breadth makes prespecification and multiplicity handling especially relevant. The available information does not establish whether every planned outcome is posted or how missing secondary results were handled.

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

• Percentage of Participants With Best Overall Response of Complete Response (CR) or Partial Response (PR) [Objective Response Rate (ORR)]
• Overall Survival (OS)
• Duration of Response (DOR)
• Number of Participants Reporting Treatment-Emergent Adverse Events (TEAEs) and Who Died

9. Statistical Interpretation

10. Statistical Methods Explained

The primary endpoint is a time-to-event outcome, so its analysis differs from a simple comparison of the number of participants with an event at a fixed visit. Participants can be followed for different lengths of time, and some leave follow-up without a recorded progression or death. The methods must account for observed time at risk and censoring rather than classify every participant only as an event or non-event.

What does a median time-to-event value describe?

The median PFS is the estimated point at which half of the relevant time-to-event distribution has experienced progression or death. It is a compact description of the distribution, not an average and not a guarantee that an individual participant will have that amount of time without an event. Its interpretation depends on the endpoint definition and follow-up. The posted median values are accompanied by 90% confidence intervals, which communicate uncertainty about the estimated median.

What does a hazard ratio summarize?

A hazard is the rate at which an event occurs among participants who have not yet experienced it, at a given time. A hazard ratio compares those rates between groups under the fitted or test-based analysis. In general, a hazard ratio near the no-difference reference is compatible with similar event rates; estimates farther from that reference indicate a stronger relative contrast, but uncertainty still matters. The registry reports hazard ratios with 90% confidence intervals and Log Rank tests. The direction depends on which group is in the numerator of the comparison, so the estimate should be read with its named treatment groups and the sponsor’s reported analysis.

The Log Rank method compares time-to-event experience while accounting for censoring. In general, it is most straightforward to interpret as a comparison of survival distributions when the relative hazards are reasonably stable over time. The registry record does not report whether that condition was checked, nor does it provide a proportional-hazards assessment. A single hazard ratio can be less informative if the relative event pattern changes over follow-up.

How should confidence intervals and p-values be read together?

A confidence interval gives a range of estimates compatible with the observed data under the method’s assumptions and expresses statistical precision. Wider intervals indicate less precise estimates. A p-value measures how compatible the data are with a specified null model under the test assumptions; it is not the probability that the null hypothesis is true, nor does it measure the size or clinical importance of an effect. The posted primary p-values are 0.132 and 0.521. The registry record does not state the decision threshold or whether testing was one- or two-sided.

The reported intervals for both primary hazard ratios cover the no-difference reference. This means the data do not sharply distinguish the estimated comparisons from that reference at the interval level. It does not prove that groups were identical, and it does not rule out effects in either direction. A non-small p-value likewise should not be interpreted as evidence that the null model is correct.

Why does censoring matter?

Censoring means the exact event time is not observed for a participant during the relevant follow-up, even though some event-free follow-up may be available. In general, a time-to-event analysis can use the observed time up to censoring without assigning an event after that point. Valid interpretation depends on censoring not being related to unobserved event risk in a way the analysis fails to handle. The registry gives the censoring rules and counts for PFS, but does not report additional analyses of censoring patterns or their assumptions.

Participants alive without progression were censored at their last objective tumor assessment. Participants lost to follow-up without progression were censored at their last radiographic assessment. Participants without baseline or post-baseline radiologic assessments were censored at randomization. These rules define how their records contribute to the PFS analysis; they do not reveal what their outcomes would have been after censoring.

What do the secondary analyses add?

The registry also posts objective response rate, overall survival, and duration-of-response and safety-related outcomes. These endpoints address different questions. Response rate concerns the proportion meeting complete or partial response criteria; overall survival concerns time to death from any cause; duration of response concerns time among participants who respond; and adverse-event summaries describe reported harms. They can add context, but they do not replace the primary endpoint or make its uncertainty disappear.

For binary outcomes, a risk difference describes the difference in response proportions, while an odds ratio compares odds. These are not interchangeable measures. The sponsor’s response analyses use chi-squared tests and report odds ratios, while the displayed response proportions allow a direct descriptive view of the groups. General recalculations from summary counts can differ from the sponsor’s method or estimand; they should not be mistaken for the prespecified analysis. The record does not report whether the secondary comparisons were adjusted for multiplicity.

For all endpoints, the design and analysis details available in the registry are incomplete relative to a full statistical analysis plan. The original sample-size justification, analysis populations for every secondary outcome, handling of missing response assessments, and detailed model specifications are not reported in the registry record. Interpretation should therefore stay close to the definitions and analyses that are actually recorded.

11. Confidence Intervals and Precision

Precision concerns how much uncertainty surrounds an estimate. A confidence interval gives a range of values compatible with the data and analysis assumptions; a narrower interval indicates greater precision than a wider one. It does not mean that every value inside the interval is equally plausible, nor does it guarantee that the true effect lies in the interval in a particular analysis.

For PFS, the sponsor-reported hazard ratio in the non-squamous comparison was 0.75, with a 90% confidence interval from 0.55 to 1.03. In the squamous comparison, it was 0.88, with a 90% confidence interval from 0.64 to 1.22. Both intervals include the no-difference reference and extend across values that allow different interpretations. Thus, the point estimates alone should not be treated as precise estimates of the comparative effect.

The medians also have uncertainty intervals. For the non-squamous groups, the posted median PFS values were 5.6 months, with a 90% confidence interval from 4.0 to 5.7, and 7.2 months, with a 90% confidence interval from 5.8 to 8.4. For the squamous groups, the corresponding medians were 5.4 months, with an interval from 4.7 to 5.7, and 5.6 months, with an interval from 4.4 to 6.0. These intervals describe uncertainty in each group’s median; they are not confidence intervals for a difference between medians.

The registry reports 90% intervals for the primary summaries. A confidence level describes the long-run behavior of the interval-generating procedure under its assumptions; it is not the probability that a fixed effect is inside a particular computed interval. The record does not report the reason for using 90% intervals or the full inferential plan. The group sizes and censoring mean that the precision of estimates should be considered alongside the point estimates, rather than inferred from medians alone.

12. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The planning figures are illustrations based on the observed effect, not the trial’s original design justification. For time-to-event outcomes, the figures are total events required; they are not participants per arm. For the other outcome type, the figures 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. Each planning row considers detecting a stated fraction of the observed effect.

The 80% power column represents the planned chance of detecting the specified effect under those assumptions if that effect is real. The 90% power column represents a higher target chance of detection under the same assumptions. These are planning targets, not certainty that a study of that size will detect an effect. The calculations do not show the probability that the observed trial result is true or the chance that a treatment works.

Smaller effects are harder to distinguish from random variation, so the planning illustrations require more participants or events when the target effect is reduced. For example, the non-squamous PFS planning illustration gives 380 total events for 80% power at 100% of the observed effect, and 1,518 total events at 50% of that effect. The squamous PFS illustration gives 1,922 total events at 100% and 7,685 at 50%, for the same power target. Those are event counts for planning, not the enrollment of the posted trial.

For response-rate planning, the figures are per arm. The illustrations range from 305 participants per arm for 80% power at 100% of the observed non-squamous response effect to 1,199 per arm at 50% of that effect. The corresponding squamous illustration gives 188 and 793 participants per arm. These examples show how sensitive planning requirements can be to the effect chosen for detection. This is not observed power: observed power is not a useful substitute for reading the effect estimate and its uncertainty.

The registry record does not report the trial’s original sample-size justification, and these illustrations should not be used to claim that the trial was adequately or inadequately sized under its original plan.

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 = 0.750, so ln(HR) = -0.2877.
2. d = 4 × (1.96 + 0.842)² / [ln(0.750)]² = 379.4, rounded up to 380 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 <- 0.75                              # 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(1160744)
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 = 380 (80% power), 508 (90% power)
 75% of observed effect: events = 675 (80% power), 903 (90% power)
 50% of observed effect: events = 1,518 (80% power), 2,032 (90% power)
Simulated power with 380 events: 0.783 (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 565 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 280 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 <- 5.6          # control-arm median, months (converted from the posted unit)
hr          <- 0.75          # 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: 354         p (1-sided)   0.0001
 Events: 172        ~HR at bound   0.5670
   Month: 15    P(Cross) if HR=1   0.0001
             P(Cross) if HR=0.75   0.0338
   IA 2: 67%                   Z   2.5114
      N: 564         p (1-sided)   0.0060
 Events: 343        ~HR at bound   0.7622
   Month: 24    P(Cross) if HR=1   0.0060
             P(Cross) if HR=0.75   0.5603
       Final                   Z   1.9930
      N: 566         p (1-sided)   0.0231
 Events: 514        ~HR at bound   0.8386
   Month: 60    P(Cross) if HR=1   0.0250
             P(Cross) if HR=0.75   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

For the primary endpoint, the sponsor-reported analyses use Log Rank methods and report hazard ratios of 0.75 for the non-squamous comparison and 0.88 for the squamous comparison. Those match the primary estimates used in the interpretation. The sponsor also reports 90% confidence intervals and p-values for each comparison. No discrepancy between the posted primary estimates and the recalculated summaries is indicated.

For response rate, the sponsor’s analyses use chi-squared tests and report odds ratios. A simple unadjusted recalculation from the posted response counts instead expresses the contrast as a risk difference. Since odds ratios and risk differences are different measures, their numerical values should not be expected to match. For the response comparisons, the sponsor-reported odds ratios are 1.58 for the non-squamous comparison and 2.66 for the squamous comparison, with 90% confidence intervals and p-values reported in the registry.

Recalculations from summary statistics are useful as checks and as alternative descriptive summaries, but they do not supersede the sponsor’s own analysis. When a discrepancy is identified, the sponsor’s analysis takes precedence; no discrepancy is indicated for these results. The registry record does not provide enough methodological detail to reconstruct every aspect of the sponsor’s analyses, including all model specifications or any adjustment across comparisons.

16. Multiplicity, Missing Data and Analysis Populations

Multiplicity arises when several outcomes, treatment comparisons, or repeated analyses are tested. Even if every null hypothesis is true, making many tests increases the chance that at least one result will appear statistically unusual by chance. In general, a multiplicity strategy may control that risk through adjusted thresholds, hierarchical testing, or a defined distinction between confirmatory and exploratory analyses. The registry record does not report a multiplicity adjustment, testing hierarchy, or a detailed statistical analysis plan.

The study includes separate PFS and response comparisons across non-squamous and squamous regimen groups, as well as several secondary outcomes. The posted p-values should therefore be interpreted in the context of multiple comparisons rather than as isolated proof. The record does not state whether the two primary subgroup comparisons were jointly controlled, whether one was primary and the other supportive, or how the secondary tests were prioritized.

The intent-to-treat population for PFS is defined as all randomized participants. That is a useful stated analysis population, but the registry does not give a complete population definition for every secondary endpoint. It does not report how missing response assessments were handled, how missingness affected the survival analyses beyond the stated censoring rules, or whether sensitivity analyses were conducted.

For PFS, the registry does describe censoring: participants alive without progression were censored at their last objective tumor assessment; those lost to follow-up without progression at their last radiographic assessment; and participants without baseline or post-baseline radiologic assessments at randomization. The listed censoring counts are 14, 13, 14, and 18 in the four primary groups, respectively. These rules are important for understanding who contributes observed time and where outcome information ends, but the record does not report reasons for censoring in detail or analyses addressing potentially informative censoring.

The record also does not report interim analyses, stopping rules, protocol deviations, or the original sample-size justification. These omissions do not show that such procedures were absent; their details are not reported in the registry record. They limit the ability to evaluate how the final analysis fits the full planned testing strategy.

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: thrombocytopenia (11.3%), pneumonia (9.9%), dyspnoea (8.5%), nausea (7.5%), vomiting (7.5%).

The safety summaries report participants at risk, participants affected by serious adverse events, participants affected by other adverse events, and deaths in each treatment group. For serious events, the posted at-risk and affected counts are 69 and 38 for pemetrexed chemotherapy, 67 and 44 for the corresponding regimen with IMC-1121B, 63 and 29 for gemcitabine chemotherapy, and 71 and 39 for its IMC-1121B combination. These are descriptive counts; they should be read with the group-specific at-risk totals rather than compared as raw counts alone.

The registry lists thrombocytopenia at 11.3%, pneumonia at 9.9%, dyspnoea at 8.5%, nausea at 7.5%, vomiting at 7.5%, and accidental overdose at 7.5% among the top serious-event terms. The record does not give enough detail in this summary to establish the denominator or attribution for each term, nor does it show whether events overlapped within participants.

Safety-event counts are not, by themselves, proof that a treatment caused an event. They describe events recorded during the study and need interpretation alongside exposure, severity, timing, and clinical context. The registry summary does not provide a full event-by-event account or an adjusted comparison of safety risk.

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

The posted results describe outcomes observed in a randomized Phase 2 study of IMC-1121B combined with platinum-based chemotherapy compared with corresponding chemotherapy alone. For PFS, the registry reports medians, hazard ratios, confidence intervals, and Log Rank p-values separately for non-squamous and squamous regimen groups. The uncertainty intervals span the no-difference reference, so the primary estimates do not give a precise answer about the comparative effect in either group.

That conclusion is narrower than saying the treatments are the same. Failure to establish a clear difference is not proof of equivalence, and the posted results do not define a margin for equivalence or non-inferiority. The estimates remain compatible with a range of possible effects. Nor do these findings show that IMC-1121B works or does not work; they describe what the registry reports and how uncertain those estimates are.

The trial’s regimen groups also limit cross-group conclusions. The non-squamous comparison uses pemetrexed-based chemotherapy, while the squamous comparison uses gemcitabine-based chemotherapy. Differences between those groups could reflect histology, chemotherapy backbone, participant characteristics, or other factors; the registry does not report a direct randomized comparison between those backbones. The results should not be used to rank pemetrexed against gemcitabine.

Secondary outcomes and safety results add context but answer different questions and have their own uncertainties. The registry does not report a multiplicity plan, complete missing-data methods for every outcome, or the original design assumptions. Those omissions make it difficult to assess the full confirmatory status of secondary findings or reconstruct the original statistical rationale.

The appropriate statistical reading is therefore descriptive and cautious: the study reports separate treatment comparisons, but the primary effect estimates are not sufficiently precise to settle the comparative PFS question. Further conclusions about clinical value, individual treatment choices, or applicability beyond the registered population are not established by these statistics.

19. Why This Trial Matters Statistically

This trial illustrates why a treatment comparison should not be reduced to a difference between two medians. PFS is a time-to-event endpoint with censoring, and the registry reports both median times and hazard ratios. The medians provide an accessible description of the distributions, while the hazard ratios summarize relative event rates over follow-up. Their confidence intervals show that point estimates alone can give an incomplete impression of certainty.

It also shows why subgroup and regimen structure matter. The four groups are arranged as two corresponding comparisons, each within a different histology and chemotherapy backbone. Combining them without a justified model could obscure meaningful differences in the populations or regimens. Conversely, comparing one histology group directly with the other would not be a randomized test of chemotherapy backbone.

The study is also a useful example of the distinction between sponsor analyses and unadjusted recalculations. The primary endpoint is represented by sponsor-reported Log Rank hazard ratios. For response, the sponsor reports odds ratios while a simple recalculation can express the data as risk differences. Both are statistical summaries, but they answer the contrast on different scales and should not be treated as interchangeable.

Finally, the posted secondary outcomes and multiple comparisons make analysis planning important. Without a reported testing hierarchy or multiplicity strategy, individual secondary p-values have a more limited role. The registry’s incomplete reporting of the original sample-size justification and analysis details reinforces a general principle: statistical interpretation depends not only on the result, but also on how outcomes, populations, and analyses were planned.

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 features specific to this trial constrain interpretation. It was an open-label study, so participants and investigators knew the assigned regimen. PFS was based on objective progression criteria, but the registry does not report whether tumor assessments were independently reviewed or blinded. The record gives the RECIST v1.1 definition and censoring rules, but not detailed assessment procedures or checks on censoring assumptions.

The trial has two regimen-specific comparisons rather than one pooled comparison. The non-squamous groups use pemetrexed and the squamous groups use gemcitabine, with carboplatin or cisplatin as the platinum agent. This structure means the results do not directly compare the two chemotherapy backbones, and differences between the two comparisons cannot be attributed solely to histology or to the addition of IMC-1121B.

The primary outcome group sizes are 69 to 71 participants, and PFS censoring counts differ across the groups. The posted confidence intervals convey uncertainty, but the registry does not report detailed information about follow-up distribution, reasons for censoring, or sensitivity analyses. The study’s Phase 2 designation also does not, by itself, specify the strength of confirmatory evidence intended by the design.

Important statistical details are not reported in the registry record: the original sample-size justification, full analysis plan, multiplicity strategy, interim-analysis procedures, and complete missing-data handling for each secondary outcome. The registry summary also does not provide a direct adjusted treatment estimate for every secondary endpoint. These gaps limit reconstruction of the planned analysis and make the secondary findings harder to classify as confirmatory versus exploratory.

Finally, the posted summaries are not a substitute for individual-level data or a complete clinical report. They do not permit examination of time-varying effects, baseline balance, event timing, or all safety details. Those limitations do not invalidate the reported estimates, but they narrow what can be concluded from the registry record alone.

21. Related Tutorials

22. Related Statistical Calculators

23. Sources

• ClinicalTrials.gov: A Study of Pemetrexed and Carboplatin/Cisplatin or Gemcitabine and Carboplatin/Cisplatin With or Without IMC-1121B in Participants Previously Untreated With Recurrent or Advanced Non-Small Cell Lung Cancer (NSCLC) — NCT01160744.
• Registry record last updated 2019-09-13; status last verified 2019-08.
• Results first posted 2014-12-29.
• 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

The registry describes a randomized, open-label, parallel-group Phase 2 study of IMC-1121B with platinum-based chemotherapy versus the corresponding chemotherapy alone in previously untreated participants with recurrent or advanced non-small cell lung cancer. The four treatment groups comprise pemetrexed-based regimens for non-squamous disease and gemcitabine-based regimens for squamous disease, each with or without IMC-1121B.

The primary endpoint was PFS, defined as time from randomization to objective progression or death. Sponsor-reported hazard ratios were 0.75 for the non-squamous comparison and 0.88 for the squamous comparison; both 90% confidence intervals include the no-difference reference. Secondary outcomes include response rate and overall survival, with safety summaries also posted.

The registry does not report the original sample-size justification, multiplicity strategy, or a complete statistical analysis plan. The posted results support cautious interpretation of the separate comparisons, with uncertainty retained around the primary estimates.