Study Comparing Veliparib Plus Carboplatin and Paclitaxel Versus Investigator's Choice of Standard Chemotherapy in Adults Receiving First Cytotoxic Chemotherapy for Metastatic or Advanced Non-Squamous Non-Small Cell Lung Cancer (NSCLC) and Who Are Current or Former Smokers: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Study Comparing Veliparib Plus Carboplatin and Paclitaxel Versus Investigator's Choice of Standard Chemotherapy in Adults Receiving First Cytotoxic Chemotherapy for Metastatic or Advanced Non-Squamous Non-Small Cell Lung Cancer (NSCLC) and Who Are Current or Former Smokers (NCT02264990), a randomized phase 3 study. The primary outcome is Overall Survival (OS) in the Lung Subtype Panel Positive Subgroup.

2. Clinical Question

This was a comparison of treatment strategies in a defined clinical setting, not a comparison of veliparib alone with no treatment. The intervention combined veliparib with two chemotherapy drugs, while the comparator allowed investigator-selected standard chemotherapy options. The primary population was narrower than the full trial population: the endpoint analysis concerned LSP+ participants in the intention-to-treat population. The registry record does not report what the Lung Subtype Panel tested or how its positive status was determined.

The central statistical question is whether the survival data distinguish the groups with enough precision to support a difference in the specified subgroup. The reported median survival estimates describe the time by which the Kaplan-Meier survival estimate reaches one half; they do not describe every participant’s survival time. The hazard ratio summarizes a comparison over follow-up under the log-rank analysis, while its confidence interval conveys uncertainty around that estimate.

3. Trial Design

The registry describes a randomized, open-label, multicenter, parallel-group Phase 3 trial. It enrolled 595 participants and is recorded as completed. Participants were adults with metastatic or advanced non-squamous non-small cell lung cancer who were current or former smokers and receiving first cytotoxic chemotherapy. Random allocation supports a comparative design, but open-label conduct means participants and clinical teams were not masked to assigned treatment.

The intervention was veliparib with carboplatin and paclitaxel. The comparator was investigator’s choice of standard doublet chemotherapy. The registry lists carboplatin with paclitaxel and platinum-based combinations with pemetrexed among the comparator options; the arm description is truncated, so the complete choice structure is not reported in the registry record. The intervention description also mentions optional maintenance pemetrexed after the chemotherapy cycles, but the associated details are incomplete.

The primary endpoint was overall survival in the LSP+ subgroup of the intention-to-treat population. Secondary outcomes included progression-free survival and objective response rate in that subgroup, along with overall survival, progression-free survival, and objective response rate in all participants. The posted primary analysis used Kaplan-Meier estimation for medians and a log-rank method for the hazard ratio. The registry record does not report the original sample-size justification, the full statistical analysis plan, or whether any interim analyses were planned or conducted.

The primary analysis included 40 participants in each treatment group, while the all-participant survival analysis reports 297 for investigator’s choice and 298 for the veliparib combination. These are endpoint-specific analysis populations, not a replacement for the registry’s enrollment figure. The trial’s open-label design, the subgroup restriction for the primary endpoint, and the use of several secondary outcomes all matter when judging how broadly the estimates can be interpreted.

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 were randomized in a 1:1 ratio to veliparib in combination with carboplatin and paclitaxel (C/P) or investigator's choice of platinum doublet chemotherapy. Randomization was stratified by smoking status (current versus former), by the investigators' preferred platinum doublet therapy (carboplatin/paclitaxel versus cisplatin/pemetrexed versus carboplatin/pemetrexed), by gender (male versus female) and by Eastern Cooperative Oncology Group (ECOG) performance status (0 versus 1).

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Participants in the ITT population who were LSP positive (LSP+)

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: The intention-to-treat (ITT) population included all randomized participants.

6. Primary Endpoint

The registry lists the primary outcome as Overall Survival (OS) in the Lung Subtype Panel Positive Subgroup, measured over From randomization up to the data cut-off date of 15 July 2019; median follow-up time was 44.5 and 45.3 months in LSP+ participants for the investigator's choice chemotherapy and veliparib + C/P arms, respectively..

Overall survival is the time from randomization until death. In this trial, it was estimated with the Kaplan-Meier method; participants recorded as alive at the data cut-off were censored at the date they were last known to be alive. Censoring allows participants with incomplete follow-up at the cut-off to contribute information up to their last known survival time rather than requiring a death to have been observed.

Overall survival is commonly used in clinical trials because it is a direct time-to-event outcome. It does not depend on a radiographic judgement of tumor response, although the quality of the estimate still depends on follow-up, event ascertainment, and assumptions about censoring. The registry reports the primary endpoint in months and gives a median and 95% confidence interval for each group.

The primary endpoint was assessed in LSP+ participants in the intention-to-treat population, rather than in all enrolled participants. Median follow-up in that subgroup was 44.5 months for investigator’s choice and 45.3 months for veliparib plus carboplatin and paclitaxel. The registry record states that survival follow-up extended from randomization to the data cut-off date of 15 July 2019.

A median is a summary of the estimated survival curve, not a mean and not a statement that each participant lived for that length of time. The hazard ratio provides a different summary, comparing event rates over the observed follow-up under the reported analysis. Neither summary alone describes the full shape of the survival experience.

7. Results Posted in the Registry

• Investigator's Choice Chemotherapy vs Veliparib + Carboplatin + Paclitaxel: hazard ratio 0.64 (95% CI 0.40–1.05), reported p = 0.113 (as reported in the registry).

8. Secondary Efficacy Endpoints

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

Progression Free Survival (PFS) in the Lung Subtype Panel Positive Subgroup

• 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:
  • Investigator's Choice Chemotherapy vs Veliparib + Carboplatin + Paclitaxel: hazard ratio 0.65 (95% CI 0.39–1.08), reported p = 0.260 (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 Investigator's Choice Chemotherapy vs Veliparib + Carboplatin + Paclitaxel and the control. The interval gives the range of effects that remain compatible with the data.

Objective Response Rate (ORR) in the Lung Subtype Panel Positive Subgroup

• 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:
  • Veliparib + Carboplatin + Paclitaxel: 22.5% (9/40) versus 30.0% (12/40) on control; risk difference -7.5 percentage points (95% CI -26.7 to 11.7), risk ratio 0.75 (0.36–1.58); unadjusted two-proportion z-test p = 0.446.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Veliparib + Carboplatin + Paclitaxel and the control. The interval gives the range of effects that remain compatible with the data.

Overall Survival in All Participants

• 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:
  • Investigator's Choice Chemotherapy vs Veliparib + Carboplatin + Paclitaxel: hazard ratio 0.99 (95% CI 0.83–1.18), reported p = 0.846 (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 Investigator's Choice Chemotherapy vs Veliparib + Carboplatin + Paclitaxel and the control. The interval gives the range of effects that remain compatible with the data.

The secondary outcomes cover both the LSP+ subgroup and all participants, and include progression-free survival, objective response rate, and overall survival. Their results do not form a single uniform pattern across endpoints: time-to-event estimates and response proportions do not point in the same direction throughout, and the all-participant survival summary is not identical to the subgroup result. This variation is a reason to read each result in its own population and with its own uncertainty, rather than to treat the secondary outcomes as one combined finding.

The secondary analyses add context to the primary overall-survival comparison, but they do not overturn or confirm it by themselves. Progression-free survival concerns a different event definition from death, while objective response rate is a proportion based on tumor response assessment. Their measurement schedules and follow-up differ, and the LSP+ analyses apply to a smaller population than the all-participant analyses.

The registry reports odds-ratio analysis for the LSP+ response outcome and log-rank hazard-ratio analyses for survival outcomes. Those measures answer different statistical questions and should not be compared as though they were on a common scale. The subgroup and all-participant analyses also address different target populations.

General: Multiple secondary outcomes create a multiplicity concern: if many hypotheses are tested, some results may appear notable by chance even when there is no underlying effect. An unadjusted secondary result carries less confirmatory weight because it does not account for the family of other tests. The registry record does not report whether multiplicity adjustment was used. Accordingly, secondary outcomes are best treated as supportive context, not as independent proof of a treatment effect or as a substitute for the primary result.

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

• Progression Free Survival (PFS) in the Lung Subtype Panel Positive Subgroup
• Objective Response Rate (ORR) in the Lung Subtype Panel Positive Subgroup
• Overall Survival in All Participants
• Progression Free Survival (PFS) in All Participants
• Objective Response Rate (ORR) in All Participants

9. Statistical Interpretation

10. Statistical Methods Explained

What does the median overall survival describe?

General: For a time-to-event outcome, the Kaplan-Meier method estimates the proportion of participants who remain event-free over time while accounting for censoring. The median is the time at which the estimated curve reaches 50% survival, if the curve reaches that point. It is a feature of the estimated curve, not an average individual survival time. Here, the event is death, and the registry says participants still alive at the cut-off were censored at their last known date alive.

The reported medians are 9.2 months for investigator’s choice and 11.2 months for the veliparib combination. Each median is accompanied by a 95% confidence interval. The intervals communicate uncertainty in the estimated median; they are not ranges containing 95% of individual participants’ survival times.

How should the hazard ratio be read?

General: A hazard ratio compares the instantaneous event rates between groups over follow-up, subject to the assumptions of the model and the way the contrast is defined. A value of 1 represents no relative difference in the compared hazards. A value below or above 1 indicates a relative direction only after the numerator and denominator groups are identified. The registry names the contrast as investigator’s choice chemotherapy versus veliparib plus carboplatin and paclitaxel.

The reported hazard ratio is 0.644, with a 95% confidence interval of 0.396 to 1.048. The interval crosses 1, so it includes the no-difference value. The p-value is 0.113. The result should therefore be read as an estimate with appreciable uncertainty, rather than as a precise estimate of a treatment difference.

What does the log-rank method contribute?

General: The log-rank test compares time-to-event experience between groups across the observed follow-up, accounting for when events occur and for censoring. The reported hazard ratio gives a relative summary of the comparison, while the log-rank p-value assesses how compatible the observed data are with a null comparison under the test’s assumptions. These summaries answer related but not identical questions. The registry identifies the method as Log Rank and reports the hazard ratio, confidence interval, and p-value.

The record does not report a detailed model specification, checks of proportional hazards, or the full handling of censoring beyond the description of censoring for participants still alive at the data cut-off. It is therefore not possible to assess those aspects from the registry record alone.

Why are the confidence interval and p-value both useful?

General: A confidence interval shows a range of effect values that are reasonably compatible with the observed data under the analysis method. It helps readers judge both the direction and precision of an estimate. A p-value measures compatibility with a specified null hypothesis under a statistical model; it is not the probability that the null hypothesis is true, nor the probability that the result occurred by chance.

For the primary hazard ratio, the interval includes 1 and the p-value is 0.113. Those features indicate that the result does not rule out no relative difference at the conventional 0.05 level. They do not prove that the groups are the same. The interval also extends to values on either side of 1, so the estimate remains compatible with more than one interpretation of the relative contrast.

What does the subgroup analysis mean?

The primary endpoint is reported in LSP+ participants in the intention-to-treat population, with 40 participants in each group. A subgroup result applies directly to that defined group, not automatically to all randomized participants. A smaller analysis population generally yields less information than a larger one, although the actual precision also depends on the number and timing of events. The primary event count is not reported in the registry record.

General: Subgroup analyses can be informative when the subgroup is defined and analyzed appropriately, but they can be vulnerable to imprecision and to multiple testing if many subgroups or endpoints are considered. The registry record does not report whether LSP+ was a prespecified subgroup, how many other subgroup analyses were planned, or whether multiplicity procedures were used.

How should the secondary endpoints be considered?

Progression-free survival and objective response rate were also reported for LSP+ participants, with additional analyses of overall survival, progression-free survival, and objective response rate in all participants. These endpoints differ in what they measure and in their analysis populations. Their estimates provide supporting context but are not interchangeable with the primary survival endpoint.

General: When several outcomes are tested, the chance of seeing at least one apparently notable result can increase unless the analysis plan accounts for multiplicity. An unadjusted p-value does not account for the number of other tests. The registry record does not report whether the secondary analyses were adjusted for multiplicity, so they should be interpreted as exploratory or supportive rather than treated as independent confirmatory evidence.

11. Confidence Intervals and Precision

The primary subgroup estimate is based on 40 participants in each group. The reported median overall survival estimates are 9.2 months and 11.2 months, with 95% confidence intervals of 5.1 to 11.7 months and 7.5 to 15.8 months, respectively. The intervals are broad, indicating that the median estimates are not highly precise. They also overlap, although overlap of separate median confidence intervals is not itself a formal test of the between-group contrast.

The hazard ratio offers a direct comparative summary: 0.644, with a 95% confidence interval from 0.396 to 1.048. The interval includes 1 and extends on both sides of it. Thus the data do not pin down the relative hazard with enough precision to exclude no difference at the conventional 0.05 level. The associated p-value is 0.113.

General: A 95% confidence interval reflects the uncertainty of an estimate under the statistical method and its assumptions. It does not mean that 95% of participants’ outcomes fall inside the interval, and it does not assign a 95% probability to the specific interval containing the true effect. Wider intervals generally indicate less precision than narrower intervals, all else equal.

Precision depends on more than the number of participants. For a survival outcome, the number and timing of events, length of follow-up, and censoring also affect the information available. The registry reports median follow-up of 44.5 and 45.3 months in the two LSP+ groups, but does not report the primary subgroup event count. The confidence interval is therefore especially important for communicating the uncertainty that remains.

The primary estimate should not be treated as proof of benefit, proof of no effect, or proof of equivalence. The interval permits materially different underlying contrasts, and the trial’s subgroup estimate is less precise than the all-participant survival summary. The registry record does not report a prespecified equivalence or non-inferiority margin.

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 illustrations translate an observed effect into hypothetical sample-size requirements under stated assumptions. For time-to-event outcomes, the listed quantities are total events, not participants per arm. For other outcomes, the quantities are participants per arm. This distinction matters because event-driven planning for survival analyses depends on how many events are observed, while the binary-outcome illustration is expressed as participants assigned to each arm.

The illustrations use a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. They consider detecting the stated fraction of the observed effect. The original trial’s own sample-size assumptions and justification are not reported in the registry record. The figures are planning illustrations based on the observed effect, not a description of how the trial was originally designed.

The 80% power column represents the hypothetical number of events or participants needed for an 80% chance of detecting the assumed effect under those planning assumptions, if that effect and the other assumptions were correct. The 90% power column represents the corresponding requirement for a 90% chance. These are design concepts, not probabilities that the observed result is true.

The rows considering smaller fractions of the observed effect require more information: smaller effects are harder to distinguish from random variation, so more events or participants are generally needed. For example, planning for 75% or 50% of the observed effect requires more information than planning for the full observed effect in the posted illustrations. The values should be treated as conditional scenarios, not as recommendations or forecasts of what another trial would necessarily require.

These calculations are not observed power. Observed power is not an informative substitute for interpreting the estimate and its confidence interval; the uncertainty in the posted results is better assessed directly from those estimates.

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 = 0.644, so ln(HR) = -0.4401.
2. d = 4 × (1.96 + 0.842)² / [ln(0.644)]² = 162.1, rounded up to 163 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 <- 0.644                              # 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(2264990)
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 = 163 (80% power), 218 (90% power)
 75% of observed effect: events = 289 (80% power), 386 (90% power)
 50% of observed effect: events = 649 (80% power), 869 (90% power)
Simulated power with 163 events: 0.772 (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 267 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 595 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 <- 9.2          # control-arm median, months (converted from the posted unit)
hr          <- 0.644          # 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: 198         p (1-sided)   0.0001
  Events: 73        ~HR at bound   0.4190
   Month: 18    P(Cross) if HR=1   0.0001
             P(Cross) if HR=0.64   0.0338
   IA 2: 67%                   Z   2.5114
      N: 268         p (1-sided)   0.0060
 Events: 146        ~HR at bound   0.6594
   Month: 28    P(Cross) if HR=1   0.0060
             P(Cross) if HR=0.64   0.5603
       Final                   Z   1.9930
      N: 268         p (1-sided)   0.0231
 Events: 219        ~HR at bound   0.7635
   Month: 60    P(Cross) if HR=1   0.0250
             P(Cross) if HR=0.64   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 of the primary endpoint using the log-rank method. Its hazard ratio is 0.644, the 95% confidence interval is 0.396 to 1.048, and the p-value is 0.113. The posted medians and confidence intervals provide separate descriptive summaries of the same endpoint in the two LSP+ groups.

The registry record does not identify a conflicting recalculation for the primary result. The estimate and inferential statistics should therefore be read as the sponsor-reported analysis, rather than replaced by a separate calculation from the displayed group summaries. A hazard ratio cannot generally be reconstructed from medians alone, because it depends on the event-time data and the analysis method.

The sponsor’s reported result does not establish a statistically significant difference at the conventional two-sided 0.05 level: the p-value is 0.113 and the confidence interval includes 1. That interpretation is about the strength and precision of the statistical evidence, not a conclusion that the treatments are equivalent or that a difference is impossible. The registry record does not report the full analysis plan, so details beyond the named log-rank method and reported result are not available.

16. Multiplicity, Missing Data and Analysis Populations

The primary endpoint population is described as participants in the intention-to-treat population who were LSP+. The registry record does not define the full intention-to-treat analysis rules, explain how LSP+ status was determined, or report how missing subgroup status was handled. It also does not describe the analysis population rules for every secondary endpoint beyond identifying the LSP+ subgroup or all participants.

For overall survival, the registry states that participants alive at the data cut-off were censored at the date they were last known to be alive. That describes one aspect of missing follow-up for the primary endpoint, but does not report the extent of missing follow-up, reasons for loss to follow-up, or sensitivity analyses for censoring. The record does not report detailed missing-data methods for progression-free survival or objective response rate.

Several outcomes were analyzed, including primary and secondary measures in both the LSP+ subgroup and all participants. The registry record does not report whether the secondary analyses were adjusted for multiplicity, whether a testing hierarchy was specified, or whether other planned comparisons were omitted. Without those details, individual secondary p-values should not be interpreted as if each were the sole prespecified test.

The record also does not report the original sample-size justification, a full statistical analysis plan, or whether interim analyses were planned or conducted. These omissions limit assessment of prespecification, type I error control, and whether the final analyses followed the original plan. They do not invalidate the posted estimates, but they restrict how confidently the analytic choices can be evaluated from 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: pneumonia (6.5%), anaemia (5.2%), malignant neoplasm progression (4.8%), febrile neutropenia (4.4%), dyspnoea (2.7%).

The registry reports serious adverse events in 98 of 288 participants at risk in the investigator’s choice group and 121 of 293 participants at risk in the veliparib plus carboplatin and paclitaxel group. Other adverse events were reported in 269 of 288 and 275 of 293 participants, respectively. Deaths are listed as 255 and 250 in the two groups. These are registry counts; they should not be combined or converted into new percentages without the applicable definitions and analysis details.

The listed serious adverse event terms include pneumonia at 6.5%, anaemia at 5.2%, malignant neoplasm progression at 4.8%, febrile neutropenia at 4.4%, dyspnoea at 2.7%, and pericardial effusion at 2.4%. The registry record does not provide further context for interpreting these term-level percentages here, such as their exact denominators or whether events could be counted in more than one category.

Serious adverse events are a defined safety category and should not be conflated with all adverse events. These figures describe reported events in the trial groups; they do not by themselves establish that a treatment caused an event or permit a complete comparison of safety. The registry record does not provide the full safety analysis plan or detailed event attribution information.

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

The primary result describes overall survival in LSP+ participants enrolled in a trial of adults with metastatic or advanced non-squamous non-small cell lung cancer who were current or former smokers and receiving first cytotoxic chemotherapy. The observed medians were 9.2 months with investigator’s choice chemotherapy and 11.2 months with veliparib plus carboplatin and paclitaxel. The sponsor-reported hazard ratio was 0.644, with a 95% confidence interval of 0.396 to 1.048 and p = 0.113.

These data show a numerical difference between the median estimates, accompanied by substantial uncertainty in the comparative estimate. The confidence interval includes 1, and the p-value is above the conventional two-sided 0.05 threshold used in the planning assumptions. The result does not establish a statistically significant overall-survival difference in the primary subgroup. It also does not show that the regimens are equivalent, rule out a meaningful difference, or determine which regimen is preferable for any individual.

The result applies to the treatment strategies and population studied. The intervention was a combination regimen, and the comparator allowed investigator choice. The primary analysis was restricted to LSP+ participants, so it should not be assumed to describe the same effect in all participants or in other populations. The registry record does not report enough detail to evaluate the subgroup’s prespecification or the full analysis plan.

Secondary outcomes provide additional but distinct evidence. They involve different endpoints, follow-up schedules, and analysis populations, and the registry record does not report whether multiplicity was addressed. Safety counts also require interpretation in the context of event definitions and attribution. Taken together, the posted results are a record of estimates and uncertainty, not a treatment recommendation or a definitive claim about benefit or lack of benefit.

19. Why This Trial Matters Statistically

This trial illustrates why a point estimate is not the whole statistical result. The primary subgroup medians differ numerically, but the comparative hazard-ratio confidence interval includes 1 and the p-value is 0.113. Precision and uncertainty therefore materially affect what can be concluded from the observed estimates.

It also shows why the analysis population matters. The trial enrolled 595 participants, while the primary endpoint is reported for the LSP+ subgroup, with 40 participants in each group. The all-participant survival analysis is a different population and cannot simply be substituted for the primary subgroup result. Keeping those populations distinct prevents an estimate from being generalized beyond the group it describes.

The outcome measures also answer different questions. Overall survival, progression-free survival, and objective response rate are not interchangeable, and their numerical summaries use different statistical approaches. A collection of secondary results can provide context, but multiplicity and uncertainty limit the strength of conclusions drawn from any one secondary finding.

Finally, the registry record makes some important analytic details unavailable, including the original sample-size justification, the full statistical plan, and multiplicity procedures. Recognizing what is and is not reported is part of sound statistical interpretation. The result is most useful when read with its population, endpoint definition, confidence interval, and reporting limitations in view.

20. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The primary endpoint analysis is in the LSP+ subgroup rather than the full enrolled population. The subgroup contains 40 participants in each group, and the registry record does not report the number of deaths contributing to that analysis. The primary confidence interval is consequently important, and its breadth leaves uncertainty about the comparative effect.

The trial was open-label, with no masking. Overall survival is less dependent on subjective outcome assessment than response outcomes, but open-label conduct may still matter for other aspects of care and for secondary assessments. The registry record does not report further safeguards or assessment procedures that would allow those issues to be evaluated.

The comparator was investigator’s choice of standard chemotherapy, rather than one single regimen. The arm description is incomplete in the registry record, limiting a detailed account of which comparator options were used and how often. The intervention description is also truncated after mentioning optional maintenance pemetrexed.

The record does not report the full statistical analysis plan, original sample-size justification, multiplicity adjustment, interim-analysis arrangements, detailed missing-data methods for secondary endpoints, or subgroup prespecification. Those omissions limit assessment of the analyses beyond the methods explicitly identified. The primary result is also tied to the data cut-off date of 15 July 2019; later follow-up is not represented in the posted endpoint description.

Secondary outcomes use distinct definitions and populations, and their multiplicity status is not reported. Safety figures are descriptive and do not by themselves establish causation or provide all details needed for comparative safety assessment. These limitations should temper broad interpretation without replacing the estimates that the registry reports.

21. Related Tutorials

22. Related Statistical Calculators

23. Sources

• ClinicalTrials.gov: Study Comparing Veliparib Plus Carboplatin and Paclitaxel Versus Investigator's Choice of Standard Chemotherapy in Adults Receiving First Cytotoxic Chemotherapy for Metastatic or Advanced Non-Squamous Non-Small Cell Lung Cancer (NSCLC) and Who Are Current or Former Smokers — NCT02264990.
• Registry record last updated 2021-02-26; status last verified 2021-02.
• Results first posted 2021-02-26.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

24. Record Summary

NCT02264990 is a completed randomized, open-label, multicenter, parallel-group Phase 3 trial sponsored by AbbVie. It enrolled 595 adults with metastatic or advanced non-squamous non-small cell lung cancer who were current or former smokers and receiving first cytotoxic chemotherapy.

The comparison was veliparib plus carboplatin and paclitaxel versus investigator’s choice standard chemotherapy. The primary endpoint was overall survival in LSP+ participants in the intention-to-treat population. Median overall survival was 9.2 months and 11.2 months, respectively; the reported hazard ratio was 0.644 (95% CI 0.396 to 1.048; p = 0.113).

The estimate is uncertain: its confidence interval includes 1, and the p-value does not meet the conventional 0.05 threshold. Secondary outcomes and safety findings provide additional context, but their interpretation is limited by differing endpoints and populations and by analytic details not reported in the registry record.