Multicenter Study of Rociletinib Administered to Patients With Previously Treated Mutant EGFR Non-small Cell Lung Cancer: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Multicenter Study of Rociletinib Administered to Patients With Previously Treated Mutant EGFR Non-small Cell Lung Cancer (NCT02147990), a randomized phase 2 study. The primary outcome is Objective Response Rate (ORR) According to RECIST Version 1.1 as Determined by Investigator Assessment.

2. Clinical Question

3. Trial Design

The registry describes TIGER-2 as a Phase 2, open-label, multicenter study of oral rociletinib as second-line EGFR-directed treatment for patients with mutant EGFR non-small cell lung cancer. It lists an enrollment of 318 and a status of terminated. The recorded study start is 2014-06-16, and the primary completion date is 2019-07-30. The posted primary endpoint timeframe extends from Cycle 1 Day 1 to end of treatment, up to approximately 57 months.

The record lists allocation as randomized, masking as none, and the model as single group. It also describes three treatment groups: rociletinib 625 mg twice daily in T790M-positive patients, rociletinib 500 mg twice daily in T790M-positive patients, and rociletinib 500 mg twice daily in T790M-negative patients. The record does not explain how randomization relates to this single-group classification or how participants were assigned among the groups. The randomization method, allocation concealment, and any stratification factors are not reported in the registry record.

Open-label means that the study was not masked according to the registry. The primary response endpoint was assessed by investigators, so knowledge of treatment assignment could be relevant to assessment; the registry does not describe additional protections against assessment bias. The treatment descriptions specify continuous dosing in 28-day cycles. The study was described as open-ended for participants, with treatment continuing until the study doctor determined it was no longer beneficial.

The primary endpoint population is not simply everyone enrolled. It includes those who received at least one dose, had a measurable baseline lesion, and had at least one post-baseline tumor assessment. How many enrolled participants were excluded from that endpoint population, and why, are not reported in the registry record.

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

Analysis population for the primary outcome, as defined in the registry: Investigator Tumor Evaluable Population - defined as all patients who received at least 1 dose of rociletinib, have at least 1 measureable tumor lesion at baseline, and have at least 1 post-baseline tumor assessment as determined by the investigator.

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

6. Primary Endpoint

The registry lists the primary outcome as Objective Response Rate (ORR) According to RECIST Version 1.1 as Determined by Investigator Assessment, measured over Cycle 1 Day 1 to End of Treatment, up to approximately 57 months..

The primary endpoint is objective response rate (ORR), defined as the percentage of patients whose best overall confirmed response was complete response (CR) or partial response (PR), recorded from treatment start until disease progression. For patients who continued treatment after progression, the first progression date was used in the analysis. Investigators assessed response using RECIST version 1.1.

Under the stated RECIST definitions, CR requires disappearance of all target lesions; pathological lymph nodes, whether target or non-target, must have a short axis < 10 mm. PR requires at least a 30% decrease in the sum of the longest diameters of target lesions, using the baseline sum as reference. The endpoint therefore classifies whether a confirmed reduction met a prespecified response category, rather than measuring a continuous change in tumor size.

General statistical context: response rate is useful for describing the proportion of an assessed population that meets a defined tumor-response threshold. It is more direct than a broad disease-control measure, which can include outcomes other than partial or complete response. ORR does not, by itself, describe response duration, time to progression, survival, symptoms, or quality of life. Those are distinct outcomes, several of which were also listed as secondary measures in the registry.

The eligible population requires a post-baseline tumor assessment. This makes the endpoint interpretable among patients with evaluable assessments, but the result may not describe every enrolled participant if some lacked the required assessments. The registry’s definition identifies the analysis population; reasons for missing assessments and their handling are not reported in the registry record.

7. Results Posted in the Registry

• Rociletinib 500 mg BID T790M+: 34.0% (33/97) versus 34.6% (53/153) on control; risk difference -0.6 percentage points (95% CI -12.7 to 11.5), risk ratio 0.98 (0.69–1.40); unadjusted two-proportion z-test p = 0.920.
• Rociletinib 500 mg BID T790M-: 18.0% (11/61) versus 34.6% (53/153) on control; risk difference -16.6 percentage points (95% CI -28.9 to -4.4), risk ratio 0.52 (0.29–0.93); unadjusted two-proportion z-test p = 0.017.

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.

Disease Control Rate (DCR) by RECIST v1.1 as Determined by Investigator Assessment

• Endpoint type: Binary: the percentage of participants with the event (counts rebuilt from the posted percentages and group sizes).
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Rociletinib 500 mg BID T790M+: 76.3% (74/97) versus 67.3% (103/153) on control; risk difference 9.0 percentage points (95% CI -2.3 to 20.2), risk ratio 1.13 (0.97–1.33); unadjusted two-proportion z-test p = 0.129.
  • Rociletinib 500 mg BID T790M-: 59.0% (36/61) versus 67.3% (103/153) on control; risk difference -8.3 percentage points (95% CI -22.7 to 6.1), risk ratio 0.88 (0.69–1.11); unadjusted two-proportion z-test p = 0.250.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Rociletinib 500 mg BID T790M+ 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 Rociletinib 500 mg BID T790M- and the control. The interval gives the range of effects that remain compatible with the data.

The registry lists 18 posted secondary outcomes, spanning duration of response, disease control, progression-free survival, overall survival, and changes in quality-of-life and symptom measures. Only the disease control rate (DCR) results and their unadjusted group comparisons are included in the numerical results. Numerical findings for the other listed secondary outcomes are not reported in the registry record.

As a broad pattern, the posted DCR estimates do not mirror a single, consistent separation across the three groups. The comparisons do not show clear evidence of a difference between the two T790M-positive dose groups or between the reference group and the T790M-negative group under the unadjusted tests. DCR includes disease control beyond complete or partial response, so its pattern can differ from ORR without being contradictory. The primary endpoint asks whether confirmed tumor shrinkage met response criteria; disease control is a broader category.

These secondary outcomes add context to the primary response question, but they answer different questions. Duration of response and progression-free survival concern time, overall survival concerns survival, and quality-of-life measures concern patient-reported domains. A response-rate result cannot substitute for those outcomes, and the registry’s endpoint list alone does not establish their results.

Unadjusted secondary findings carry less weight because the record lists multiple endpoints and comparisons but does not report a multiplicity strategy or a formal analysis hierarchy. When many tests are considered, the chance of at least one apparently notable result can increase even if there is no underlying difference. A p-value from an individual unadjusted comparison should therefore be viewed as exploratory unless supported by a prespecified plan and appropriate control of multiple testing. The registry reports no analyses in its formal analyses listing, so the secondary estimates should be interpreted as descriptive evidence rather than as a confirmatory set of claims.

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

• Duration of Response (DOR) in T790M Positive Patients According to RECIST Version 1.1 as Determined by Investigator Assessment
• Disease Control Rate (DCR) by RECIST v1.1 as Determined by Investigator Assessment
• Progression-free Survival (PFS) in T790M Positive Patients by RECIST v1.1 as Determined by Investigator Assessment
• Overall Survival (OS) Determined by Investigator Assessment
• Change From Baseline to Cycles 5, 10 and End of Treatment (EOT) in European Organization for Research and Treatment of Cancer Quality of Life Questionnaire for Patients With Cancer (EORTC QLQ-C30) Global Health Status Quality of Life Scale
• Change From Baseline to Cycles 5, 10 and End of Treatment (EOT) in Dermatology Life Quality Index (DLQI)
• Change From Baseline to Cycles 5, 10 and End of Treatment (EOT) in EORTC QLQ-LC-13 Alopecia Scale
• Change From Baseline to Cycles 5, 10 and End of Treatment (EOT) in EORTC QLQ-LC-13 Coughing Scale

9. Statistical Interpretation

10. Statistical Methods Explained

Endpoint type. ORR is a binary outcome at the participant level: each assessed participant either meets the confirmed complete- or partial-response definition or does not. The reported group result is the percentage of participants meeting that definition. This differs from a continuous outcome such as change in tumor measurement and from a time-to-event outcome such as progression-free survival.

General: confidence intervals. A confidence interval expresses the uncertainty associated with an estimate under a statistical procedure. For a response percentage, it gives a range of values reasonably compatible with the data and method used. A narrower interval indicates greater precision; a wider interval indicates less precision. It does not mean that each value in the interval is equally likely, and it does not guarantee that the underlying response percentage lies within the interval.

Group comparisons. The posted recalculated comparisons are unadjusted risk-difference comparisons, using the 625 mg twice-daily T790M-positive group as the reference. A risk difference is the response proportion in one group minus the response proportion in the reference group; negative values indicate a lower observed proportion in the first group. The comparison also reports a confidence interval and an unadjusted p-value. The 500 mg T790M-positive comparison has p = 0.92, while the 500 mg T790M-negative comparison has p = 0.017. These p-values describe compatibility with a no-difference null model under the stated unadjusted calculation; they do not measure effect size or the probability that the null hypothesis is true.

The 500 mg T790M-positive risk-difference interval includes zero, consistent with the absence of a clear separation in this unadjusted comparison. The T790M-negative comparison’s interval does not include zero. Such findings do not remove concerns about group comparability, selection, or multiple testing. In particular, T790M-positive and T790M-negative groups are defined by different biomarker status, so the comparison cannot isolate the causal contribution of that status from other differences between participants.

General: statistical significance and clinical importance. A p-value is not a measure of whether a difference matters clinically. A small p-value can occur with a modest difference, while a potentially meaningful difference can be estimated imprecisely. Interpretation should consider the estimated response rates, the confidence intervals, the study design, the endpoint’s relevance, and the analysis plan. The registry does not report a prespecified threshold for a clinically important ORR difference.

Planning illustrations. The sample-size illustrations use the observed effect as a starting point and consider 100%, 75%, and 50% of that effect. The stated assumptions are a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. These are planning calculations based on the observed effect, not the trial’s own documented design assumptions. For binary outcomes, the figures are participants per arm; the time-to-event convention of total events does not apply to this ORR outcome.

General: power and sample size. Power is the probability, under specified assumptions, of detecting an effect of a specified size using a chosen test. An 80% power calculation and a 90% power calculation describe different target probabilities of detecting that assumed effect. They are not probabilities that the observed result is correct. If the effect to detect is smaller, more participants are generally required to distinguish it from random variation while holding the other assumptions fixed. The original study’s sample-size justification, target effect, and design assumptions are not reported in the registry record.

Analysis populations and assessment. The primary endpoint population requires at least one dose, a measurable baseline lesion, and at least one post-baseline assessment as determined by the investigator. This is an evaluable population rather than an explicitly stated intention-to-treat population. The handling of participants who were enrolled but did not satisfy those criteria is not reported in the registry record. Nor does the record state whether missing response assessments were counted as nonresponse, imputed, or addressed by another method.

Multiplicity and analysis specification. The registry lists several secondary outcomes and reports no formal analyses for them in the analysis listing. It does not describe a testing hierarchy, multiplicity correction, interim analyses, or a prespecified statistical analysis plan. Without those details, individual unadjusted p-values should not be treated as confirmatory across a collection of endpoints and group comparisons. The sponsor’s analysis does not take precedence over the recalculated results because the discrepancy flag is false; the recalculated comparisons remain simple, unadjusted analyses based on posted summary statistics.

11. Confidence Intervals and Precision

The primary ORR estimates have different levels of precision across the three groups. For 625 mg twice daily in T790M-positive patients, the estimate is 34.6% and the 95% confidence interval is 27.1% to 42.7%, based on 153 participants. For 500 mg twice daily in T790M-positive patients, the estimate is 34.0% and the interval is 24.7% to 44.3%, based on 97 participants. For 500 mg twice daily in T790M-negative patients, the estimate is 18.0% and the interval is 9.4% to 30.0%, based on 61 participants.

General statistical context: uncertainty in a percentage estimate depends in part on the number of assessed participants and the observed proportion. Smaller groups commonly produce less precise estimates, although the exact interval also depends on the data and interval method. The T790M-negative group’s interval is broad relative to its point estimate, so a range of response percentages remains compatible with the posted data. The intervals do not convey uncertainty about biases arising from eligibility, missing assessments, open-label assessment, or differences among groups.

The primary group confidence intervals overlap. That visual fact should not replace the direct comparison: the unadjusted comparison of the two T790M-positive groups has p = 0.92, while the comparison of the T790M-negative group with the reference has p = 0.017. The corresponding comparison intervals are reported for risk differences, and their relationship to zero is more directly relevant to a no-difference comparison than overlap between separate group intervals.

Precision is also distinct from clinical importance. The confidence intervals describe statistical uncertainty under the calculation used; they do not define which response rate would be meaningful to patients or clinicians. A clinically meaningful threshold is not reported in the registry record. The endpoint-specific intervals also do not establish how durable responses were, since duration of response is a separate outcome.

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 sample-size illustrations are planning calculations, not counts of participants who entered the study. For this binary response outcome, the estimates are participants per arm, not total events. The assumptions listed for the calculations are a two-sided significance level of 0.05, equal allocation between two arms, a normal approximation, and an effect to detect set to the stated fraction of the observed effect. The illustrations show planning at 100%, 75%, and 50% of the observed effect, with separate targets for 80% and 90% power.

In plain language, the 80% power column gives the number of participants per arm that the calculation estimates would be needed to have an 80% chance of detecting the assumed effect under those assumptions. The 90% power column gives the corresponding number per arm for a 90% target chance. These are hypothetical planning targets for a two-arm comparison; they do not describe the actual enrollment or group sizes in this study. The registry’s original design assumptions and sample-size justification are not reported in the registry record.

The rows that use a smaller fraction of the observed effect require more participants per arm. This is expected because a smaller difference is harder to distinguish from random variation when other assumptions remain fixed. The number also depends on the observed response rates used to form the planning illustration. Since the estimates are based on the observed effect, they can be unstable and should not be treated as a general recommendation for a future study without a separately justified target effect and design.

These figures are not observed power. Observed power calculated from the effect seen in the study would not add useful information about the result’s uncertainty; the confidence intervals and the actual group sizes are more informative for reading the posted findings.

13. How the Sample Size Is Derived

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

Formula

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

Worked calculation for this trial

1. Control proportion p₀ = 0.346; treatment proportion p₁ = 0.340; average p̄ = 0.343.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.6715 = 1.3161.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.6715 = 0.5651.
4. n = (1.3161 + 0.5651)² / (-0.0062)² = 92,093.1, rounded up to 92,094 participants per arm for 80% power.

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

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
p0 <- 0.34640522875817                          # control proportion
p1 <- 0.34020618556701                          # treatment proportion
n_per_arm <- function(power, frac = 1) {
  q1 <- p0 + (p1 - p0) * frac               # smaller effect = fraction of the observed difference
  pbar <- (p0 + q1) / 2
  z_b <- qnorm(power)
  ceiling(round((z_a * sqrt(2 * pbar * (1 - pbar)) + z_b * sqrt(p0 * (1 - p0) + q1 * (1 - q1)))^2 / (q1 - p0)^2, 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

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

# Simulation: how often does the z-test reject at the 80%-power sample size?
set.seed(2147990)
n    <- n_per_arm(0.80)
reps <- 20000
x0 <- rbinom(reps, n, p0); x1 <- rbinom(reps, n, p1)
ph <- (x0 + x1) / (2 * n)
z  <- ((x1 - x0) / n) / sqrt(ph * (1 - ph) * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(z) > z_a, na.rm = TRUE), reps))

Output from running the code

100% of observed effect: n per arm = 92,094 (80% power), 123,287 (90% power)
 75% of observed effect: n per arm = 163,898 (80% power), 219,413 (90% power)
 50% of observed effect: n per arm = 369,166 (80% power), 494,208 (90% power)
power.prop.test: 92093.1 
Simulated power at n = 92,094 per arm: 0.800 (20000 replicates)

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

14. Multiplicity, Missing Data and Analysis Populations

The registry lists a primary ORR endpoint and multiple secondary endpoints, including disease control, response duration, progression-free survival, overall survival, and quality-of-life measures. There are also comparisons among the three treatment groups. With multiple outcomes and comparisons, repeated testing can increase the chance of observing a nominally unusual result by chance. The registry record does not report whether tests were arranged in a hierarchy, whether any multiplicity adjustment was planned, or whether a particular secondary outcome was intended to support a confirmatory conclusion.

The analysis listing reports no formal analyses. A detailed statistical analysis plan, interim analysis procedures, rules for stopping, and the original sample-size justification are not reported in the registry record. The trial’s termination status is listed, but the reason for termination and any relation to efficacy, safety, recruitment, or other considerations are not reported. These absences limit the ability to distinguish prospectively planned analyses from analyses selected after results became available.

The primary endpoint population consists of patients with a dose, a measurable baseline lesion, and at least one post-baseline tumor assessment. The registry does not provide the number excluded from this population or describe the characteristics of participants without evaluable assessments. Missing-data rules for the primary or secondary outcomes are not reported in the registry record. In particular, it is not stated how participants who discontinued early, had no post-baseline scan, or had an unconfirmed response were treated in the analyses.

Those details matter because exclusions or missing assessments can affect the population represented by an estimate. The open-label design and investigator assessment are also relevant to interpretation, but the registry does not describe blinding of image review or other assessment safeguards. The record identifies an investigator tumor-evaluable population; it does not report a separate intention-to-treat analysis for the primary endpoint. These limitations do not invalidate the posted estimates, but they leave important aspects of the analysis and its robustness unclear.

15. 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: malignant neoplasm progression (14.9%), hyperglycaemia (6.5%), cataract (6.3%), pneumonia (5.0%), vomiting (4.8%).

The registry reports serious adverse events for each treatment group using a group at-risk count and the number affected. Serious adverse events were reported for 77 of 154 patients in the 625 mg twice-daily T790M-positive group, 48 of 100 in the 500 mg twice-daily T790M-positive group, and 34 of 63 in the 500 mg twice-daily T790M-negative group. These counts describe participants with serious adverse events, not the total number of serious events.

The registry also reports other adverse events for 153 of 154, 100 of 100, and 63 of 63 participants in the respective groups. Death counts are 27, 7, and 9, respectively. The record does not establish from these counts alone whether deaths were caused by treatment, cancer progression, or another cause. The serious adverse-event terms reported most often include malignant neoplasm progression at 14.9%, hyperglycaemia at 6.5%, cataract at 6.3%, pneumonia at 5.0%, vomiting at 4.8%, and dyspnoea at 4.8%.

These are descriptive safety counts. Differences between groups should be interpreted cautiously because the groups vary in size and T790M status, and the record does not report an adjusted comparative safety analysis. The figures do not by themselves establish causality or provide a complete account of event timing, duration, severity grades, or management.

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

The posted results describe investigator-assessed tumor response among patients who met the tumor-evaluable population definition. The response percentages are estimates for the three rociletinib groups, with uncertainty expressed by 95% confidence intervals. The unadjusted comparison between T790M-positive groups does not show a clear difference in ORR. The comparison of the T790M-negative group with the reference T790M-positive group shows a nominal difference in the posted unadjusted calculation.

These findings do not establish that one dose is superior to another, or that T790M status itself caused the observed group difference. The record’s classification of allocation and model does not clarify assignment, and no non-rociletinib comparison group is described. The groups defined by T790M status are not interchangeable populations. A between-group contrast can reflect differences in patient characteristics or assessment, in addition to the feature being compared.

ORR captures confirmed tumor shrinkage meeting RECIST criteria. It does not establish how long a response lasted, whether disease control persisted, whether survival changed, or whether quality of life improved. Although those outcomes appear among the listed secondary measures, most of their numerical results are not reported in the registry record. The posted endpoint also represents participants with a post-baseline assessment, not necessarily every enrolled participant.

General statistical context: an estimate is not a guarantee for an individual patient, and a confidence interval is not a prediction interval for individual outcomes. Statistical evidence about a group-level response percentage also does not by itself settle whether that response is clinically important. The registry does not report a clinical-importance threshold for ORR.

The appropriate reading is therefore limited: the record gives response and safety descriptions for the enrolled groups and unadjusted comparisons between them. It does not, on its own, establish a causal benefit relative to another treatment, resolve the role of biomarker status, or provide a complete account of the trial’s prespecified analysis methods.

17. Why This Trial Matters Statistically

This study illustrates how the interpretation of a trial result depends on the design description, the analysis population, the endpoint, and the precision of estimates—not only on a point estimate or p-value. The registry labels the allocation randomized while also identifying a single-group model and listing distinct dose and biomarker groups. Because the relationship between these design fields and group assignment is not explained, the group comparisons require careful interpretation rather than an assumed randomized contrast.

The primary outcome is an investigator-assessed binary response endpoint. Its definition is specific: a confirmed complete or partial response under RECIST version 1.1. That specificity makes the endpoint interpretable, but the evaluable-population requirement means the estimate applies to participants with the required baseline and follow-up tumor assessments. Assessment processes and missing-data rules are not detailed in the record.

The comparisons also show why confidence intervals and design context belong beside p-values. The T790M-positive dose-group comparison has little evidence of separation in this unadjusted analysis, while the T790M-negative comparison has a nominally notable p-value. Neither result answers whether group differences are causal, and multiple outcomes and comparisons make unadjusted findings less conclusive. The secondary outcome list extends to survival and quality of life, but their numerical results are not reported in the registry record.

Finally, the sample-size illustrations are based on observed effects and stated planning assumptions rather than documented original design inputs. They can illustrate how smaller target effects require larger studies, but they cannot explain the trial’s original enrollment decision. Together, these features make the record useful for examining the distinction between descriptive evidence, comparative inference, and confirmatory analysis.

18. Limitations

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

Several limitations are specific to how this trial is described in the registry. The design fields are not fully aligned: allocation is listed as randomized, while the model is listed as single group and three groups are described. The method of assignment, randomization process, and the rationale for the model classification are not reported in the registry record. This uncertainty affects how strongly the group contrasts can be interpreted as randomized comparisons.

The study was open-label and the primary endpoint was investigator-assessed. The registry does not describe central blinded review, procedures to reduce assessment bias, or how disagreements in response assessment were handled. The primary analysis population requires an investigator-determined post-baseline assessment, but exclusions, missing scans, and missing-data handling are not described.

The treatment groups differ by both dose and T790M status. Comparing the two T790M-positive groups addresses different starting doses within that status, whereas comparison with the T790M-negative group also changes biomarker status. The record does not report adjusted analyses that account for baseline characteristics or explain how treatment assignment and biomarker status were handled analytically.

The trial is listed as terminated, but the reason is not reported. The record also does not report the original statistical plan, sample-size justification, interim monitoring, multiplicity strategy, or complete numerical results for the listed secondary endpoints. The available comparisons are unadjusted. These gaps leave uncertainty about the prespecified analysis and the robustness of individual findings; they do not, by themselves, show that the posted estimates are incorrect.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Multicenter Study of Rociletinib Administered to Patients With Previously Treated Mutant EGFR Non-small Cell Lung Cancer — NCT02147990.
• Registry record last updated 2020-08-12; status last verified 2020-07.
• Results first posted 2020-08-12.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

22. Record Summary

NCT02147990 is a Phase 2, open-label, multicenter study of oral rociletinib in previously treated mutant EGFR non-small cell lung cancer. The registry lists enrollment of 318, randomized allocation, no masking, a single-group model, and terminated status. Three groups are described by dose and T790M status.

The primary endpoint is investigator-assessed ORR by RECIST version 1.1 in a tumor-evaluable population. Posted ORR estimates are 34.6% for 625 mg twice daily in T790M-positive patients, 34.0% for 500 mg twice daily in T790M-positive patients, and 18.0% for 500 mg twice daily in T790M-negative patients. Unadjusted comparisons show no clear separation between the T790M-positive dose groups and a nominal difference between the reference group and the T790M-negative group.

The registry lists 18 secondary outcomes, but numerical results for most are not reported in the registry record. The analysis plan, missing-data methods, multiplicity approach, and reason for termination are also not reported.