An Expanded Access Program of Tarceva (Erlotinib) in Participants With Advanced Non-Small Cell Lung Cancer (NSCLC): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for An Expanded Access Program of Tarceva (Erlotinib) in Participants With Advanced Non-Small Cell Lung Cancer (NSCLC) (NCT00949910), a na phase 4 study. The primary outcome is Percentage of Participants With Objective Response According to Response Evaluation Criteria in Solid Tumors (RECIST).

2. Clinical Question

This is a descriptive question about responses recorded during an access program, not a randomized comparison of erlotinib against another treatment or against no treatment. Without a concurrent control group, the record cannot establish how response outcomes would have differed under another treatment strategy. It also does not report a numerical primary outcome estimate that would allow readers to assess the observed response frequency or its precision.

3. Trial Design

This was a phase 4, single-group expanded access program of erlotinib for participants with advanced NSCLC. The study is listed as completed. Enrollment is reported as 6586. The study start is recorded as 2004-11, and primary completion as 2009-04. The sponsor was Hoffmann-La Roche.

The program’s stated purpose was to provide erlotinib to participants who had received at least one course of standard chemotherapy or radiation therapy, or who were not medically suitable for either. The treatment description specifies single-agent erlotinib for inoperable, locally advanced, recurrent, or metastatic NSCLC. Treatment was to continue until unacceptable toxicity, disease progression, or withdrawal for any other reason. The record does not provide further details about dose, dose modifications, or the criteria used to determine treatment eligibility.

The design has no randomized allocation, no masking, and no like-for-like control arm. Allocation is listed as not applicable, and the study model is single group. The registry record therefore supports description of outcomes and safety observations among participants in the program, but it does not support a randomized estimate of the effect of erlotinib relative to another treatment.

The primary outcome was the percentage of participants with objective response according to RECIST. The record describes assessment at baseline, according to institutional standards during treatment for up to 3.5 years, and every 6 months thereafter. The overall stated timeframe was up to approximately 4.5 years. The primary outcome’s actual numerical result is not reported in the registry record. The record lists seven secondary outcome measures, including disease control, best overall response, disease progression or death, progression-free survival, death, and overall survival. Numerical findings for these outcomes are not reported in the registry record.

The registry record does not report a protocol-based sample-size justification, a statistical analysis plan, or details of any interim analyses. Those details cannot be inferred from the enrollment figure or from the expanded access design.

4. Participant Flow and Accounting

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

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: ITT/Safety Population.

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: Intent-to-Treat (ITT)/Safety Population: All participants who received at least one dose of erlotinib.

6. Primary Endpoint

The registry lists the primary outcome as Percentage of Participants With Objective Response According to Response Evaluation Criteria in Solid Tumors (RECIST), measured over Up to approximately 4.5 years; assessed at Baseline, according to institutional standards during treatment (up to 3.5 years), and every 6 months thereafter.

The primary endpoint is the percentage of participants with an objective response according to RECIST. Objective response was defined as a best overall response of complete response (CR) or partial response (PR) during the study. CR required disappearance of all clinical and radiographic evidence of target and non-target lesions, normal tumor markers, and absence of tumor-related symptoms. PR required a decrease of at least 30 percent in the sum of the longest diameters of target lesions relative to baseline. A response was to be confirmed at least 28 days after the initial assessment of CR or PR.

A response endpoint gives a defined way to describe tumor change during follow-up. Combining CR and PR captures participants meeting either of the specified response categories, while RECIST supplies criteria for classifying changes in lesions. The registry record states that the percentage was reported to the nearest integer. However, it does not report the resulting percentage, the number of participants meeting the response definition, or a confidence interval. The endpoint definition can therefore be understood, but the observed response frequency and its statistical precision cannot be evaluated from the registry record.

Assessments were scheduled at baseline, according to institutional standards during treatment for up to 3.5 years, and every 6 months thereafter. Because assessment during treatment followed institutional standards, the record does not establish a uniform assessment interval for every participant. Timing and completeness of assessment can affect which response is recorded as best overall response. The registry record does not report further details about how missing or unevaluable assessments were handled.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry record gives an outcome definition and assessment schedule, but it does not report the statistical methods used to analyze the primary endpoint. It lists no analysis records. The discussion below distinguishes general statistical concepts from methods confirmed for this study; the latter are not reported in the registry record.

What does a percentage endpoint describe?

In general, a percentage for a binary outcome represents the share of participants classified as having the event of interest among a defined analysis population. For this endpoint, the event is an objective response, defined as CR or PR according to RECIST. To interpret such a percentage, a reader needs both the number classified as responders and the denominator used for the analysis. The registry record states that the percentage was to be reported to the nearest integer, but it does not report the observed percentage, numerator, or endpoint denominator.

The denominator matters because participants may have different amounts of follow-up or may lack an assessable response. The record labels the primary population as the ITT/Safety Population, but does not explain the membership or analysis rules for that population. It does not report whether participants without a documented response were counted as nonresponders, omitted, or handled in another way. These details are necessary to reproduce or fully interpret a response percentage.

Why use RECIST response categories?

In general, a response classification applies consistent rules to clinical and imaging information so that changes in disease can be summarized using predefined categories. Here, CR required disappearance of clinical and radiographic evidence of target and non-target lesions, normal tumor markers, and no tumor-related symptoms. PR required at least a 30 percent decrease in the sum of longest diameters of target lesions relative to baseline. The primary endpoint combined those two categories and required confirmation at least 28 days after the initial response assessment.

Using a prespecified definition can make classification more systematic, but it does not remove all measurement or implementation differences. The record says assessments during treatment followed institutional standards. It does not report whether imaging methods, reading procedures, or assessor blinding were standardized across institutions. The study itself is listed as unmasked.

How should timing of response assessment be understood?

The stated assessment schedule began at baseline, continued according to institutional standards during treatment for up to 3.5 years, and then continued every 6 months thereafter, for an overall timeframe of up to approximately 4.5 years. In general, a best-overall-response endpoint uses assessments over time to identify the best category reached during follow-up. Participants with longer or more frequent assessment may have more opportunities for a response to be detected. The registry record does not report participant-level follow-up, assessment completeness, or how different follow-up durations were handled.

What can a confidence interval add?

In general, a confidence interval communicates uncertainty around an estimated quantity. It is not a guarantee about an individual participant’s outcome. A narrower interval usually indicates greater statistical precision under the method’s assumptions, while a wider interval indicates more uncertainty. The registry record does not report a confidence interval or the method used to estimate uncertainty for the primary response percentage. Therefore, precision cannot be assessed from the posted outcome information.

What do the design and planning assumptions imply?

The study is listed as a single-group program, with allocation not applicable and no masking. It has no concurrent randomized control arm. In general, a single-group outcome summary can describe what was observed among participants in that group, but cannot isolate a treatment effect from changes over time, participant selection, other care, or other influences. A causal comparison would require a suitable comparator and an analysis that addresses the study design.

The record’s sample-size assumptions describe a planning illustration using a two-sided significance level of 0.05, equal allocation between two arms, a normal approximation, and an effect defined as a stated fraction of an observed effect. For time-to-event outcomes, the illustration uses a total event count; for other outcomes, it uses participants per arm. These are planning assumptions, not the design of this single-group program. The record does not report a trial-specific sample-size justification, and planning illustrations are not observed power or evidence of an outcome.

General power calculations concern the chance that a planned study would detect an effect of a specified size under stated assumptions. They do not measure the probability that a result is true, and they cannot replace an estimate with its uncertainty. No numerical planning results are included in the registry record, and no new sample-size calculation can be inferred from the enrollment figure.

How should secondary outcomes be treated?

The registry record lists secondary measures covering disease control, best overall response, disease progression or death, progression-free survival, death, and overall survival. In general, analyzing multiple outcomes creates questions about multiplicity: when many hypotheses are tested, the chance of at least one apparently unusual result can increase. Adjustment methods, endpoint prioritization, and interpretation depend on the analysis plan. The registry record does not report a multiplicity strategy or numerical analyses for the listed secondary outcomes, so no claim about adjustment or statistical significance can be made.

10. Exact Estimates for Each Arm

No two-arm comparison is shown for this trial, but each arm's result can still be summarized with exact interval estimates computed from the posted counts. The Clopper-Pearson interval is exact and conservative; the Wilson interval is a common closer-to-nominal alternative; the Bayesian credible interval uses a uniform Beta(1, 1) prior. For small groups the three can differ noticeably.

R code and output

arm    <- c("Erlotinib")
n      <- c(6586)
events <- c(724)
for (i in seq_along(n)) {
  ci <- binom.test(events[i], n[i])$conf.int          # Clopper-Pearson exact interval
  cr <- qbeta(c(0.025, 0.975), 1 + events[i], 1 + n[i] - events[i])   # Bayesian, uniform Beta(1,1) prior
  wi <- (function(x, n) { p <- x / n; z <- qnorm(0.975); d <- 1 + z^2 / n
    c((p + z^2 / (2 * n) - z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d, (p + z^2 / (2 * n) + z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d) })(events[i], n[i])
  cat(sprintf("ARM|%s|%d|%d|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f\n", arm[i], n[i], events[i], 100 * events[i] / n[i],
              100 * ci[1], 100 * ci[2], 100 * wi[1], 100 * wi[2], 100 * cr[1], 100 * cr[2]))
}

ARM  Erlotinib  6586  724  11.0  10.2  11.8  10.3  11.8  10.3  11.8

11. Multiplicity, Missing Data and Analysis Populations

The registry record lists seven secondary outcome measures alongside the primary objective-response outcome. It does not report analysis results for those measures, an analysis plan, or a method for addressing multiplicity. In general, when several outcomes are examined, readers need to know which outcomes were primary, how tests were organized, and whether any adjustment was planned. Without that information, a collection of possible outcomes should not be treated as a set of independently confirmatory findings.

Missing-data handling is not reported in the registry record. This matters because the primary endpoint depends on tumor assessments over time, and the stated schedule includes assessments according to institutional standards during treatment. The record does not report how participants with no post-baseline assessment, incomplete follow-up, withdrawal, or unevaluable scans were handled. It also does not give the number of participants with an assessable primary outcome. No particular missing-data approach can be assumed.

The primary population is described as the ITT/Safety Population, but the registry record does not define its membership or explain whether the same population was used for efficacy and safety analyses. Enrollment of 6586 does not establish the denominator for the primary endpoint or for each safety measure. The record also does not report analysis-specific exclusions, protocol deviations, or sensitivity analyses.

Consequently, the posted record does not permit assessment of whether missing outcomes or the choice of analysis population could influence the response estimate. These are reporting gaps, not proof that the study did or did not use a particular method.

12. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

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

Serious events with the highest arm-level rate: non-small cell lung cancer (12.3%), dyspnoea (5.2%), disease progression (4.4%), pneumonia (3.3%), general physical health deterioration (3.0%).

The registry record reports serious adverse events for the erlotinib group. It lists 6586 participants at risk and 2983 affected for serious events. The most frequently listed serious-event terms are non-small cell lung cancer at 12.3%, dyspnoea at 5.2%, disease progression at 4.4%, pneumonia at 3.3%, general physical health deterioration at 3.0%, and respiratory failure at 1.7%. These are the percentages recorded for the listed terms; the registry record does not provide further event-level detail or an explanation of the denominator for each term.

For the separate other-adverse-event category, the record lists 6586 at risk and 0 affected. That entry should be read as a reported registry count for that category, not as evidence that participants had no symptoms or no adverse experiences of any kind. The record does not provide enough information to interpret the category’s scope or reporting procedures.

These figures are descriptive safety reports from a single-group program. They do not establish that erlotinib caused each event, and there is no concurrent control group for comparison. The record does not report event timing, severity grading beyond the serious-event classification, exposure duration, attribution assessments, or how events were collected. The safety counts therefore have important interpretive limits.

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

The registry record describes an expanded access program in which erlotinib was given as a single agent to participants with advanced NSCLC. It provides a definition for objective response, specifies a broad assessment timeframe, and lists serious adverse-event terms. It also records enrollment of 6586 and identifies the study as completed.

However, the numerical result for the primary response endpoint is not reported in the registry record. The record therefore does not show the percentage of participants with CR or PR, the number who met the endpoint, or an uncertainty interval. It is not possible to judge the size or precision of the observed response outcome from the endpoint definition alone. Numerical findings for the listed secondary outcomes are also not reported in the registry record.

Even if descriptive outcome estimates were available, the single-group design would limit causal interpretation. There was no randomized comparator, and allocation is listed as not applicable. The record cannot establish how outcomes would have differed under another treatment or without erlotinib. Comparisons with results from other populations or studies would require careful attention to differences in eligibility, prior treatment, assessment practices, follow-up, and outcome definitions; those comparisons are not posted by this registry entry.

The safety figures describe events reported in the registry record, not comparative risks. The listed serious-event percentages should not be interpreted as proof that treatment caused those events. The record does not report a control group, event attribution, or enough detail about ascertainment to quantify treatment-related risk.

In general, an endpoint definition tells readers what investigators intended to measure, while an estimate and its uncertainty communicate what was observed and how precisely it was measured. Here, the former is described but the primary numerical result and its precision are not reported in the registry record. The appropriate interpretation is therefore limited to the documented design, planned outcome definitions, enrollment, and reported safety entries—not a conclusion about treatment effect.

14. Why This Trial Matters Statistically

This trial is statistically notable as a large single-group expanded access program rather than a randomized efficacy comparison. Enrollment is recorded as 6586, and the study aimed to provide erlotinib to participants with advanced NSCLC who had received standard chemotherapy or radiation therapy, or who were not medically suitable for either. Such a program can produce descriptive information about treatment use and reported outcomes in its enrolled population.

The design also illustrates why enrollment size alone does not determine how much can be concluded. Without a concurrent control group, observed outcomes cannot distinguish treatment effects from other influences on disease course or care. Without the primary numerical result, denominator, and uncertainty estimate, the response endpoint cannot be evaluated quantitatively from the registry record, despite the recorded enrollment.

The endpoint definition is detailed enough to identify the intended RECIST response categories, including confirmation timing. At the same time, assessment during treatment followed institutional standards, and the record does not report how variation in assessments or missing outcomes was handled. This highlights the importance of reporting both outcome definitions and the analysis population and methods used to turn assessments into a summary estimate.

The entry also shows the difference between outcomes being listed and results being available. Several secondary outcomes are named, but their numerical analyses are not reported in the registry record. Statistical interpretation requires results and methods, not just a list of measures. These gaps limit what can be learned from the registry entry and make clear why transparent reporting of denominators, missing-data handling, uncertainty, and analysis plans matters for clinical research.

15. 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 main limitation for interpreting this record is that the primary objective-response result is not reported in the registry record. The percentage, numerator, denominator, and confidence interval are unavailable, so neither the observed frequency nor its precision can be assessed. Numerical results for the listed secondary outcomes are also not reported.

The single-group design has no randomized comparator. It can describe outcomes among participants in the program, but it cannot provide a controlled estimate of the difference between erlotinib and another treatment. The record’s allocation field is not applicable, and the study is listed as unmasked. No comparative causal conclusion can be drawn from the design information alone.

Other important details are not reported in the registry record: eligibility criteria beyond the general treatment-history description, participant characteristics, treatment exposure and dose details, outcome-analysis denominators, missing-data handling, follow-up completeness, and the statistical analysis plan. The primary population is labeled ITT/Safety Population, but its definition is not given. Assessment during treatment followed institutional standards, and the record does not say how assessment differences across institutions were addressed.

The safety information is also limited. The record does not explain event collection, attribution to treatment, event timing, or the scope of the separate other-adverse-event category. The percentages for listed serious events are descriptive and lack a control group for comparison. The enrollment figure should not be assumed to equal the number analyzed for every endpoint or the number exposed for the same duration.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: An Expanded Access Program of Tarceva (Erlotinib) in Participants With Advanced Non-Small Cell Lung Cancer (NSCLC) — NCT00949910.
• Registry record last updated 2016-10-05; status last verified 2016-10.
• Results first posted 2016-10-04.
• 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.

19. Record Summary

This completed phase 4 study was a single-group expanded access program of single-agent erlotinib for participants with advanced NSCLC. Enrollment was 6586. The study start is recorded as 2004-11 and primary completion as 2009-04. Participants were described as having received at least one course of standard chemotherapy or radiation therapy, or as not medically suitable for either.

The primary outcome was the percentage with objective response according to RECIST, defined as CR or PR, with confirmation at least 28 days after initial assessment. The assessment timeframe was up to approximately 4.5 years. The primary numerical result and its precision are not reported in the registry record. Seven secondary outcomes are listed, with no numerical analyses reported.

There was no randomized comparator. Serious adverse-event entries include 2983 affected among 6586 at risk, with listed terms and percentages recorded in the registry. The registry record does not report enough statistical, outcome, or safety detail to support a comparative treatment-effect estimate.