Effect of Tumor Treating Fields (TTFields, 150 kHz) as Front-Line Treatment of Locally-advanced Pancreatic Adenocarcinoma Concomitant With Gemcitabine and Nab-paclitaxel (PANOVA-3): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Effect of Tumor Treating Fields (TTFields, 150 kHz) as Front-Line Treatment of Locally-advanced Pancreatic Adenocarcinoma Concomitant With Gemcitabine and Nab-paclitaxel (PANOVA-3) (NCT03377491), a randomized phase 3 study. The primary outcome is Overall Survival of Subjects Treated With TTFields Concomitant With Gemcitabine and Nab-paclitaxel vs Chemotherapy Alone.

2. Clinical Question

The central statistical question is whether the time-to-event pattern differed between randomized groups. The posted primary comparison is a hazard ratio, with a 95% confidence interval and a Log Rank test result. These summaries describe the relative event pattern over follow-up; they do not by themselves explain why any difference occurred or establish how the result applies to people outside the studied population.

3. Trial Design

PANOVA-3 is identified in the registry as a phase III, randomized, open-label, parallel-group study. It evaluated TTFields at 150 kHz with gemcitabine and nab-paclitaxel as front-line treatment for locally advanced pancreatic adenocarcinoma. The intervention arm received TTFields using the NovoTTF-200T System together with chemotherapy. The comparator arm received best standard of care with gemcitabine and nab-paclitaxel. The registry lists enrollment as 571.

The posted primary outcome is overall survival. Its timeframe runs from randomization until death from any cause or last known alive. The arm-level median overall survival estimates are 16.2 months for TTFields plus standard of care and 14.16 months for standard of care, with respective 95% confidence intervals of 15.01 to 17.97 months and 12.81 to 15.44 months. The sponsor-reported comparison is a Log Rank analysis with a hazard ratio of 0.82, a 95% confidence interval from 0.68 to 0.99, and p = 0.039.

Masking is recorded as none, so the study was open-label. This matters because knowledge of treatment assignment can affect some assessments or care processes, even though overall survival is based on death from any cause. The registry does not report further details about allocation procedures, stratification factors, follow-up schedules, treatment adherence, or the analysis population. Those aspects should not be inferred from the randomized, parallel-group designation alone.

The record gives a start date of 2018-05-10, a primary completion date of 2024-10-16, and a completed status. These record dates describe the study timeline; they do not supply additional information about the conduct or analysis of the trial.

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:

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: There may be more than 1 lesion present in a participant

6. Primary Endpoint

The registry lists the primary outcome as Overall Survival of Subjects Treated With TTFields Concomitant With Gemcitabine and Nab-paclitaxel vs Chemotherapy Alone, measured over From randomization until death from any cause or last known alive..

Overall survival is the time from randomization until death from any cause, with last known alive used as the endpoint for participants not known to have died by their last follow-up. It captures survival regardless of the specific cause of death, rather than only deaths attributed to the cancer or treatment. The registry identifies it as the primary outcome and measures it in months.

Time-to-event endpoints are useful when participants can have different lengths of follow-up and when some have not experienced the event by the time their outcome is assessed. A participant who is last known alive contributes survival information up to that point without being treated as if a death had occurred. This is generally handled through censoring methods. The registry states the endpoint timeframe but does not report the detailed censoring rules, follow-up procedures, or analysis population used for the sponsor’s primary analysis.

The registry reports the median overall survival for each arm and a hazard ratio for the randomized comparison. The median describes the time at which the estimated survival curve reaches the halfway point; it does not describe every participant’s experience or the full shape of the survival curve. The hazard ratio summarizes the relative rate of the event over the analyzed follow-up under the model or test framework used. Neither statistic gives an individual survival prediction. The endpoint is used to compare survival experience between randomized groups, while the confidence interval communicates uncertainty in the estimated comparison.

7. Results Posted in the Registry

• TTFields + Standard of Care vs Standard of Care: hazard ratio 0.82 (95% CI 0.68–0.99), reported p = 0.039 (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 of Subjects Treated With TTFields Concomitant With Gemcitabine and Nab-paclitaxel vs Chemotherapy Alone

• 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:
  • TTFields + Standard of Care vs Standard of Care: hazard ratio 0.85 (95% CI 0.68–1.05), reported p = 0.137 (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 TTFields + Standard of Care vs Standard of Care and the control. The interval gives the range of effects that remain compatible with the data.

Local Progression-free Survival of Subjects Treated With TTFields Concomitant With Gemcitabine and Nab-paclitaxel vs Chemotherapy Alone

• 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:
  • TTFields + Standard of Care vs Standard of Care: hazard ratio 0.84 (95% CI 0.67–1.06), reported p = 0.151 (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 TTFields + Standard of Care vs Standard of Care and the control. The interval gives the range of effects that remain compatible with the data.

Objective Response Rate of Subjects Treated With TTFields Concomitant With Gemcitabine and Nab-paclitaxel vs Chemotherapy Alone

• Endpoint type: Binary: the number of participants with the event, compared as proportions.
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Standard of Care: 30.0% (73/243) versus 36.1% (88/244) on control; risk difference -6.0 percentage points (95% CI -14.4 to 2.3), risk ratio 0.83 (0.65–1.07); unadjusted two-proportion z-test p = 0.158.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Standard of Care and the control. The interval gives the range of effects that remain compatible with the data.

The secondary endpoints cover several different dimensions: progression-free survival, local progression-free survival, objective response, one-year survival, quality of life, pain-free survival, puncture-free survival, and resectability rate. They are not interchangeable measures. Some describe time until an event, some describe response or resectability, and others concern patient experience or a specified survival landmark.

The quantitative secondary results available in the registry record include progression-free survival, local progression-free survival, and objective response. Their directions are broadly consistent with the direction of the primary hazard-ratio estimate, but their confidence intervals and p-values do not provide the same strength of statistical evidence as the primary result. They should be treated as supportive context rather than independent confirmation of the primary conclusion. Detailed numeric results and interpretations for those individual endpoints are stated with their respective outcome information and are not repeated.

Numeric results for the remaining named secondary outcomes are not reported in the registry record described by these data. That absence prevents a complete assessment of consistency across the full set of secondary outcomes. It is not evidence that those outcomes had no effect or were not measured.

Multiplicity matters because testing several outcomes creates more opportunities for chance findings. Unless the analysis plan controls the overall false-positive risk across outcomes, a nominal p-value for an individual secondary endpoint can be less persuasive than a primary result specified for the main question. The registry record does not report a multiplicity adjustment or a hierarchy for secondary testing. Secondary estimates may still be informative, but their interpretation should account for the number and range of outcomes examined and for the uncertainty in each estimate.

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

• Progression-free Survival of Subjects Treated With TTFields Concomitant With Gemcitabine and Nab-paclitaxel vs Chemotherapy Alone
• Local Progression-free Survival of Subjects Treated With TTFields Concomitant With Gemcitabine and Nab-paclitaxel vs Chemotherapy Alone
• Objective Response Rate of Subjects Treated With TTFields Concomitant With Gemcitabine and Nab-paclitaxel vs Chemotherapy Alone
• One-year Survival Rate of Subjects Treated With TTFields Concomitant With Gemcitabine and Nab-paclitaxel vs Chemotherapy Alone
• Quality of Life of Subjects Treated With TTFields Concomitant With Gemcitabine and Nab-paclitaxel vs Chemotherapy Alone
• Pain-free Survival of Subjects Treated With TTFields Concomitant With Gemcitabine and Nab-paclitaxel vs Chemotherapy Alone
• Puncture-free Survival of Subjects Treated With TTFields Concomitant With Gemcitabine and Nab-paclitaxel vs Chemotherapy Alone
• Resectability Rate of Subjects Treated With TTFields Concomitant With Gemcitabine and Nab-paclitaxel vs Chemotherapy Alone

9. Statistical Interpretation

10. Statistical Methods Explained

The primary outcome is time-to-event data: each participant contributes information about time from randomization until death or the last time known alive. Participants who have not had a recorded death by the end of their available follow-up are not counted as deaths; their information is censored at the relevant observation point. This approach allows analyses to use follow-up that differs among participants. The exact censoring rules and follow-up procedures are not reported in the registry record.

What does the hazard ratio summarize?

A hazard is the event rate at a particular point in follow-up among people who have not yet had the event. The hazard ratio compares those rates between groups over time. The reported value of 0.82 is the sponsor’s hazard ratio for TTFields plus standard of care versus standard of care. Under the usual interpretation, a value below 1 indicates a lower estimated event hazard in the first group over the analyzed period. A hazard ratio is not a ratio of median survival times, a difference in months, or a statement about the probability that a particular participant will survive.

Interpreting one hazard ratio as a compact summary depends on how the relative hazards behave over time. If that relationship changes substantially, one number may not capture the full pattern. The registry record reports the Log Rank method and hazard ratio but does not report whether the proportional-hazards assumption was assessed or whether alternative time-varying summaries were used. The survival curves and their shapes are not described in the registry entries.

What does the Log Rank test do?

The Log Rank test compares observed and expected event patterns between groups across follow-up. It is commonly used for time-to-event comparisons and uses information from event times rather than only comparing the group medians. The registry identifies Log Rank as the method for the primary hazard-ratio comparison. It reports p = 0.039. A p-value describes how unusual a result at least this discordant with the null comparison would be under the test’s assumptions; it is not the probability that the null hypothesis is true, nor does it quantify the size or clinical importance of the effect.

The record also reports a confidence interval for the hazard ratio. A confidence interval expresses statistical uncertainty around the estimated comparison and indicates which values are compatible with the data under the method’s assumptions. It should be read together with the estimate and test result, not as a range containing every plausible individual effect with certainty.

How are the other endpoint types analyzed?

Progression-free survival and local progression-free survival are also time-to-event outcomes. Their event definitions differ: progression-free survival includes radiologic disease progression per RECIST v1.1 or death, whichever occurs first; local progression-free survival concerns local progression per RECIST v1.1 in the absence of distant metastasis or death, whichever occurs first. The registry reports Log Rank hazard ratios for these comparisons. Their estimates are secondary results and should not be conflated with the primary overall-survival analysis.

Objective response is recorded as a count of participants. Counts and group denominators describe how many participants met the response definition among those included in the reported outcome. The sponsor’s posted comparison is identified as a one-sided t-test. A separate unadjusted recalculation is not the sponsor’s analysis and uses a different statistical approach; the sponsor result remains the reported trial analysis. The registry does not provide the response criteria details beyond the stated RECIST v1.1 timeframe, nor does it explain why a one-sided test was selected.

Over-dispersion is a general statistical concern when observed variability exceeds what a model assumes. It can make standard errors too small if ignored. The registry entries do not identify an over-dispersed count model for the primary endpoint; overall survival is a time-to-event outcome and the reported method is Log Rank. No claim about additional model adjustments is warranted.

Randomization supports a comparison of assigned groups by reducing systematic differences on average, but it does not guarantee that every measured or unmeasured characteristic is balanced. The record identifies randomization and parallel groups, but does not report allocation concealment, stratification, covariate adjustment, or the primary analysis population. These missing methodological details limit a full audit of the analysis.

11. Confidence Intervals and Precision

The primary hazard-ratio estimate is 0.82, with a 95% confidence interval from 0.68 to 0.99. The interval gives a range of relative hazard values compatible with the observed data under the analysis method and its assumptions. It is not an interval for the individual survival time of a participant, and it does not mean that 95% of individual outcomes fall within those limits.

The upper limit is close to 1, the value corresponding to equal hazards in the two groups. The interval therefore indicates uncertainty about the size of the relative difference, even though the sponsor-reported interval excludes 1 and the reported p-value is 0.039. The point estimate alone would suggest greater precision than the interval supports. The lower limit also shows that effects more pronounced than the central estimate remain compatible with the data under the reported analysis.

The arm-specific median estimates have their own 95% confidence intervals: 15.01 to 17.97 months for TTFields plus standard of care and 12.81 to 15.44 months for standard of care. These intervals quantify uncertainty in each arm’s median estimate. Their overlap is not a substitute for a formal test of the between-group contrast; the hazard-ratio analysis is the reported comparison for the primary endpoint.

Confidence intervals describe sampling uncertainty under a statistical procedure. They do not account automatically for possible bias from missing data, departures from model assumptions, outcome assessment practices, or selective reporting. Those issues require design and analysis details that are not reported in the registry record. Precision should thus be considered together with the endpoint definition, analysis method, and completeness of the available information.

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 assumptions. For time-to-event outcomes, the figures represent total events needed, not participants per arm. For the objective-response comparison, the figures are participants per arm. The time-to-event calculations use the Schoenfeld approach; the other calculations use a normal approximation.

The stated planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and an effect to detect set to the stated fraction of the observed effect. The registry does not report the trial’s own original sample-size justification or design assumptions. The planning values should therefore not be presented as a reconstruction of the study’s original target or as proof that the trial had a particular probability of detecting its observed result.

The 80% power column represents the planned number of events or participants, under the stated assumptions, for a design that would have an 80% chance of detecting the specified effect if that effect were true. The 90% power column represents the corresponding planning target for a higher chance of detection. These are prospective design concepts, not probabilities that can be assigned to the trial after its results are known. This is not observed power.

Rows use different fractions of the observed effect to illustrate how planning changes when the effect to detect is smaller. A smaller target effect is harder to distinguish from random variation and therefore requires more events or participants under the same assumptions. For survival outcomes, accruing events is central to the calculation; the number of participants needed in practice also depends on follow-up and event occurrence. For the response outcome, the planning values are expressed directly as participants per arm. The figures are illustrations for interpreting design requirements, not claims about what was observed in PANOVA-3.

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(3377491)
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: 0.807 (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 1,437 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 571 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 <- 14.16          # 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: 1156         p (1-sided)   0.0001
  Events: 360        ~HR at bound   0.6761
    Month: 19    P(Cross) if HR=1   0.0001
              P(Cross) if HR=0.82   0.0338
    IA 2: 67%                   Z   2.5114
      N: 1438         p (1-sided)   0.0060
  Events: 719        ~HR at bound   0.8292
    Month: 31    P(Cross) if HR=1   0.0060
              P(Cross) if HR=0.82   0.5603
        Final                   Z   1.9930
      N: 1438         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

For overall survival, the registry reports the sponsor’s Log Rank analysis as a hazard ratio of 0.82, a 95% confidence interval from 0.68 to 0.99, and p = 0.039. The interpretation of the primary result should rely on this reported time-to-event analysis. The record does not identify a separate recalculated hazard ratio for this endpoint, and the time-to-event summaries use the sponsor-reported hazard ratios.

For objective response, the sponsor reports a one-sided t-test with p = 0.094. The simple unadjusted recalculation from the posted summary counts gives p = 0.158. These are not the same analysis: the sponsor’s stated test and the unadjusted recalculation use different methods. The recalculated result is a descriptive check, not a replacement for the sponsor’s analysis or a corrected trial result. The registry discrepancy flag is false, and there is no reported discrepancy requiring the sponsor result to be superseded.

Where a sponsor analysis is reported, it is important to distinguish it from any calculation based only on arm-level summary statistics. A summary-data recalculation may not reproduce adjustments or procedures used in the full trial analysis. The record does not report further details about the sponsor’s analysis populations or adjustments, so the comparison between approaches should remain limited to the stated methods and results.

16. Multiplicity, Missing Data and Analysis Populations

The registry identifies overall survival as the primary outcome and lists several secondary outcomes. It does not report whether a formal multiplicity strategy was used, such as a testing hierarchy or adjustment across multiple endpoints. Without that information, the family-wise interpretation of secondary p-values cannot be established from this record. Nominal significance for an individual secondary comparison should not automatically be treated as confirmatory evidence, especially when several outcomes are examined.

Details about missing outcome data are also not reported in the registry record. The primary outcome definition refers to the last known alive time for participants without a recorded death, but the record does not describe the extent of missing follow-up, the censoring rules, or any sensitivity analyses for loss to follow-up. For quality-of-life and other participant-reported outcomes, missingness can be especially consequential because people who complete assessments may differ from those who do not. Specific completion rates and missing-data procedures are not reported.

The registry states that allocation was randomized, but does not specify the analysis population used for the primary result. It does not report whether analyses followed intention-to-treat, per-protocol, or another population definition, or how protocol deviations were handled. Allocation concealment, stratification factors, and covariate adjustment are also not reported in the registry record.

These omissions do not invalidate the reported estimates, but they limit independent assessment of how robustly the analysis addressed multiplicity, missingness, and departures from the assigned treatment strategy. The reported primary effect and its uncertainty remain the available statistical summary; fuller interpretation would require these additional methods and participant-flow details.

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: sepsis (9.5%), cholangitis (5.8%), bile duct obstruction (5.5%), pneumonia (5.1%), anaemia (4.4%).

The registry reports serious adverse events among 147 of 274 participants at risk in the TTFields plus standard-of-care group and 131 of 273 at risk in the standard-of-care group. It also reports other adverse events among 267 of 274 and 268 of 273 participants at risk, respectively. These are counts of participants with reported events, not event totals, and they do not by themselves describe severity beyond the serious-event classification.

The posted death counts are 199 in the TTFields group and 223 in the standard-of-care group. The registry record does not identify these counts as treatment-related deaths, and they should not be interpreted that way. Overall survival counts and adverse-event reporting serve different purposes.

The listed serious-event terms and percentages include sepsis at 9.5%, cholangitis at 5.8%, bile duct obstruction at 5.5%, pneumonia at 5.1%, anaemia at 4.4%, and pyrexia at 2.9%. The denominator and arm assignment for these term-level percentages are not reported in the registry record. The figures are descriptive safety summaries; they do not establish that treatment caused an event or provide a complete comparison of event timing, duration, or severity.

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

The registry reports a primary overall-survival comparison favoring a lower estimated event hazard in the TTFields plus chemotherapy group, with a hazard ratio of 0.82 and a 95% confidence interval from 0.68 to 0.99. The sponsor-reported p-value is 0.039. These results describe the randomized comparison and its statistical uncertainty; they do not guarantee that every participant experienced the same pattern or establish a particular benefit for an individual.

The arm medians provide a complementary summary of estimated survival time, but they do not measure individual life-years gained. Nor should the difference between medians be interpreted as a causal extension of survival for every participant. A survival curve may contain information not captured by its median, and the registry entries do not describe the curve shape or provide subgroup results.

A statistical result does not by itself establish clinical importance. The registry record does not report a threshold for a clinically meaningful difference, nor does it provide enough information to assess how participants valued potential benefits relative to burdens. It also does not report the complete details of missing-data handling, treatment exposure, or the primary analysis population. Those issues matter when judging how confidently the estimate can be applied beyond the reported comparison.

The trial was open-label, so participants and clinicians were not masked to assignment. This may affect outcomes that depend on subjective reporting or clinical decisions. Overall survival is based on death from any cause, but the available record does not permit a complete assessment of follow-up practices or other possible sources of bias.

The result should therefore be understood as evidence from one randomized phase III comparison in the population described by the registry, with a reported primary estimate and a confidence interval that conveys uncertainty. It is not a guarantee of outcome, a complete account of all benefits and harms, or a basis for medical advice. The registry result alone also cannot settle questions about applicability to populations or treatment settings not represented in the record.

19. Why This Trial Matters Statistically

This trial poses a clear comparative question: whether adding TTFields to gemcitabine and nab-paclitaxel changes overall survival relative to the same chemotherapy regimen. Randomization makes a between-group comparison more informative than an uncontrolled description because it is intended to reduce systematic differences in treatment assignment. The open-label design and the unreported details of allocation and analysis still matter when assessing the strength and limits of that comparison.

The primary endpoint is time-to-event rather than a simple status measured at a single visit. That distinction affects the analysis: follow-up time and censoring contribute information, while a median alone cannot summarize the full event pattern. The sponsor-reported hazard ratio and Log Rank result thus provide a different view from the arm-specific median estimates.

The result is also a useful illustration of how precision and evidence strength must be read together. The primary estimate is below 1 and its reported confidence interval ends at 0.99, while the p-value is 0.039. The interval nonetheless spans a range of effect sizes. A point estimate and p-value cannot replace consideration of that uncertainty.

Finally, the trial includes outcomes beyond overall survival, including progression, response, quality of life, and other measures. Their variety illustrates why secondary findings need careful interpretation: outcomes address distinct questions, their results are not all reported in the available record, and multiplicity can make isolated nominal findings less conclusive. Statistical reading therefore requires attention both to the main comparison and to what the registry does not report.

20. Limitations

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

The registry record gives a concise outcome summary but does not report several details needed to assess the trial analysis fully. These include eligibility criteria beyond the broad disease and treatment setting, allocation concealment, stratification, analysis-population definitions, covariate adjustment, and protocol-deviation handling. The enrollment figure is reported, but participant flow and the number included in each endpoint analysis are not comprehensively described in the registry entries.

The open-label design is a relevant limitation for outcomes that may depend on participant reports, clinician judgment, or decisions about assessment and care. The record does not report whether outcome assessors were masked. Overall survival is defined using death from any cause, but complete details about follow-up, censoring, and missing survival information are not reported.

Multiplicity procedures are not reported, and numeric results for several named secondary outcomes are not available in the registry record. That limits the ability to evaluate consistency across all planned outcomes or to distinguish confirmatory findings from exploratory ones. The objective-response sponsor test is identified as a one-sided t-test, but the reason for that choice and further test details are not stated.

The safety summary gives counts and selected event terms, but the term-level percentage denominators and arm assignments are not reported. The registry summary also does not provide the timing, duration, or full severity distribution for adverse events. These limitations constrain interpretation of the record; they do not establish that the trial’s conduct or results were invalid.

21. Related Tutorials

22. Related Statistical Calculators

23. Sources

• ClinicalTrials.gov: Effect of Tumor Treating Fields (TTFields, 150 kHz) as Front-Line Treatment of Locally-advanced Pancreatic Adenocarcinoma Concomitant With Gemcitabine and Nab-paclitaxel (PANOVA-3) — NCT03377491.
• Registry record last updated 2026-01-21; status last verified 2026-01.
• Results first posted 2026-01-21.
• 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

PANOVA-3 is a completed, randomized, open-label, parallel-group phase III study of TTFields with gemcitabine and nab-paclitaxel versus gemcitabine and nab-paclitaxel alone for front-line treatment of locally advanced pancreatic adenocarcinoma. The registry lists enrollment as 571.

Overall survival was the primary outcome, measured from randomization until death from any cause or last known alive. The sponsor reported a Log Rank hazard ratio of 0.82, a 95% confidence interval from 0.68 to 0.99, and p = 0.039. Secondary outcomes include progression, response, survival, quality of life, pain, puncture-free survival, and resectability. The registry record provides selected secondary results but does not report numeric results for all named secondary outcomes or several important analysis details.