SEARCH

SEARCH BY CITATION

Keywords:

  • disease progression date;
  • progression-free survival;
  • tumor assessment schedule;
  • advanced lung cancer;
  • clinical trials

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

BACKGROUND:

In patients with advanced lung cancer, overall survival is largely influenced by progression status. Because progression-free survival (PFS)-based endpoints are controversial, the authors evaluated the impact of the progression date (PD) determination approach on PFS estimates.

METHODS:

Individual patient data from 21 trials (14 North Central Cancer Treatment Group trials and 7 Southwest Oncology Group trials) were used. The reported PD (RPD) was defined as either the radiographic scan date or the clinical deterioration date. PD was determined using Method 1 (M1), the RPD; M2, 1 day after the last progression-free scan; M3, midpoint between the last progression-free scan and the RPD; and M4, an interval-censoring approach. PFS was estimated using Kaplan-Meier (M1-M3), and maximum-likelihood (M4) methods. Simulation studies were performed to understand the impact of the length of time elapsed between the last progression-free scan and the PD on time-to-progression estimates.

RESULTS:

PFS estimates using the RPD were the highest, and M2 was the most conservative. M3 and M4 were similar because the majority of progressions occurred during treatment (ie, frequent disease assessments). M3 was influenced less by the length of the assessment schedules (percentage difference from the true time-to-progression, <1.5%) compared with M1 (11% to 30%) and M2 (−8% to −29%). The overall study conclusion was unaffected by the method used for randomized trials.

CONCLUSIONS:

The magnitude of difference in the PFS estimates was large enough to alter trial conclusions in patients with advanced lung cancer. The results indicate that standards for PD determination, the use of sensitivity analyses, and randomized trials are critical when designing trials and reporting efficacy using PFS-based endpoints. Cancer 2012. © 2012 American Cancer Society.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

Progression-free survival (PFS) is a common outcome measure for phase 2 oncology trials for various diseases, including nonsmall cell lung cancer (NSCLC) and small cell lung cancer (SCLC). There are various reasons why PFS is an important measure. PFS is a measure of both efficacy and tumor growth associated with initial therapy. In addition, PFS is unaffected by subsequent treatment upon disease progression.1-4 In a disease with a poor prognosis like advanced NSCLC or SCLC, the true endpoint of overall survival probably is correlated with the progression status of the disease.1 In addition, with molecularly targeted therapies, patients typically experience stable disease rather than tumor shrinkage.5 PFS includes patients who achieve stable disease for an extended time in addition to those who achieve complete or partial responses.1

PFS-based endpoints in phase 2 trials remain controversial. There are ongoing discussions regarding whether PFS can fully capture the potential benefit of a treatment or whether a regimen could be approved based on PFS analysis.6-8 Moreover, the accuracy and validity of PFS as an endpoint is impacted by several in-born factors, including ascertainment bias in an open-label trial; timing and scheduling issues, such as imbalance in assessment schedules across different arms, treatment holidays, missed assessments; and the development of disease progression in the middle of a long disease evaluation interval.9-11 With the advent of the Response Evaluation Criteria in Solid Tumors (RECIST),12, 13 ascertainment bias has been reduced or minimized. However, there is no well accepted standard for addressing timing and scheduling issues within and across disease types.

Panageas et al10 elegantly discussed problems related to disease assessment scheduling and progression date (PD) determination. Tumor assessment is usually scheduled/repeated after a fixed number of treatment cycles, eg, every other cycle (usually every 6 or 8 weeks, depending on cycle length), while patients are on active treatment. Following this paradigm, disease progression can occur anytime between a progression-free scan and the following scan indicating progression, a situation called interval censoring. A common practice is to use the first documented PD as a surrogate for the true PD, which likely yields an overestimation of the median PFS associated with the study regimen. In extreme cases, this magnitude of difference in PFS estimate may alter trial conclusions (ie, resulting in false-positive conclusions). In the study by Panageas et al,10 a close relation between cycle length and PD declaration also was observed, which highlights the biased inference in practice. This is particularly of concern in single-arm, phase 2 trials in which the trial design is based on historic data instead of concurrent controls.

When the length of the surveillance intervals varies, the validity of comparisons of PFS across multiple, single-arm, phase 2 trials within a disease group, or across different treatment arms within 1 randomized trial, is questionable. In a randomized trial, because of treatment delay from adverse events, for example, PD may be reported later in the more toxic arm, even if the true PFS is identical between treatment arms.11 In addition, with evaluations performed between scheduled assessments (out of concern for lack of efficacy or signs of clinical progression), patients on the arm with inferior efficacy may have more frequent tumor measurements compared with those who are receiving treatment with superior efficacy, the so-called evaluation bias. Consequently, the efficacy of the superior arm is somewhat overestimated.11 Simulation studies also have provided evidence that variations in the timing of tumor assessment can significantly bias trial conclusions based on PFS analyses and yield misleading results.9, 10

The objective of this study was to systematically assess the impact of PD determination on PFS estimates in phase 2 trials of advanced NSCLC and extensive-stage SCLC. On the basis of individual patient data from the North Central Cancer Treatment Group (NCCTG) and the Southwest Oncology Group (SWOG) advanced lung cancer trials and simulation studies, we compared PFS estimates using different approaches for determining PD, and we assessed the impact of the length of disease assessment schedules on the time-to-progression estimates.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

Trial Identification

An initial analysis was performed on individual patient data from consecutive NCCTG phase 2 trials, including 10 NSCLC trials and 4 SCLC trials.14-26 We subsequently validated the findings using individual patient data from SWOG phase 2 trials, including 4 NSCLC trials and 3 SCLC trials.27-33 All trials that were included in this study were activated in the year 2000 or beyond and used RECIST12 for tumor assessment. To be eligible, a patient must have had a baseline scan and either had a follow-up scan or died before the first scheduled postbaseline scan.

Statistical Analysis

In this study, we explored 4 approaches for PD determination, specifically: Method 1 (M1), the reported PD (RPD), which was defined as either the scan date for radiographic progression or the clinical deterioration date (determined by the treating physician) for nonradiographic progression; M2, 1 day after the last progression-free scan date; M3, the midpoint between the last progression-free scan date and the RPD; M4, the multiple-imputation method based on a nonparametric approach for interval censored data.

For M1, M2, and M3, PFS was defined as the time from registration to the earlier of disease progression or death from any cause, and it was estimated using Kaplan-Meier survival analysis. For the patients who remained alive with no documented disease progression, PFS was censored as of the last radiographic scan or the clinical assessment date that documented no disease progression. The expectation-maximization (EM) iterative convex minorant algorithm (ICM) approach (EM-ICM) was used for M4 to compute the nonparametric maximum-likelihood estimator (NPMLE).34 The EM algorithm is an iterative procedure to compute the NPMLE of the distribution function when missing data are present (in this instance, interval censored data) in which the missing data are estimated using the conditional expectation given the observed data. The ICM algorithm is another iterative procedure to compute the NPMLE of interval censored data in which only the diagonal elements of the Hessian matrix are used. In the hybrid method of EM-ICM, EM and ICM algorithm steps alternate. Specifically, the ICM step searches for the NPMLE in the self-consistent estimate set defined by the EM step until convergence is met. Furthermore, to evaluate the impact of differential progression assessment dates across arms on randomized trials, we applied the 4 methods (M1-M4) to 4 randomized trials.15, 24, 26, 30 PFS estimates between the arms in a trial were compared using the log-rank test for M1, M2, and M3 and using the generalized log-rank test for M4.35

Simulation studies explored the impact of time elapsed between the last progression-free scan date and the RPD. Because death is an unambiguous endpoint, our simulation study only considered the time-to-progression (TTP), defined as the time from the date of study entry to disease progression as specified by M1, M2, and M3. The simulations assumed an exponential distribution for the TTP distribution in which the rate parameter was calculated based on the observed median TTP of NCCTG NSCLC data (considered the true TTP) with a uniform censoring distribution (range, 2-60 months). One thousand samples of 50 observations and 100 observations were generated using various tumor assessment schedules (every 4 weeks, 6 weeks, 8 weeks, 10 weeks, and 12 weeks) for the simulated progression times. M1, M2, and M3 were then applied, and the percentage difference was calculated for each scenario as follows: (true median TTP − average of the estimated median TTP) divided by the true median TTP. Because of the nonparametric nature of M4 and the parametric nature of the simulated data, M4 was not considered in the simulations. All analyses were conducted using SAS statistical software (version 9.2; SAS Institute, Inc., Cary, NC)36 and R (version 2.13.0; R Project for Statistical Computing, Vienna, Austria).37

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

Data Description

North Central Cancer Treatment Group data

The NCCTG data were frozen in November 2009. In total, 660 patients with NSCLC and 116 patients with SCLC were included. All trials were negative for the protocol-defined primary endpoint. Disease assessment schedules followed patients either every 6 weeks (5 NSCLC trials and 3 SCLC trials) or every 8 weeks (5 NSCLC trials and 1 SCLC trial) during treatment. After treatment, patients were followed every 3 months (3 NSCLC trials and 1 SCLC trial), 3 months for a specified duration followed by 6 months thereafter (2 NSCLC trials and 2 SCLC trials), or every 6 months (5 NSCLC trial and 1 SCLC trial) (for trial characteristics, see Table 1).

Table 1. Trial Characteristics
TrialaNo. of PatientsYear ClosedStudy AgentsDisease Assessment Schedule
During TreatmentPost-Treatment
  • Abbreviations: G-CSF, granulocyte-colony–stimulating factor; NSCLC, nonsmall cell lung cancer; SCLC, small cell lung cancer; wk, weeks; mo, months.

  • a

    The trials described in this table are reported in the literature.14-33

NSCLC     
 N0022462002Oral vinorelbineEvery 8 wkEvery 3 mo
 N00261322003Gemcitabine, pemetrexed disodiumEvery 6 wkEvery 6 mo
 N0222472006Paclitaxel, carboplatin, gefitinibEvery 8 wkEvery 3 mo
 N0323522006TemsirolimusEvery 8 wkEvery 6 mo
 N0326252006SorafenibEvery 8 wkEvery 6 mo
 N0426482007Pemetrexed disodium, bevacizumabEvery 6 wkEvery 6 mo
 N0528932008Cedirinib, gemcitabine, carboplatinEvery 6 wkEvery 6 mo
 N06261102010Pemetrexed, sorafenibEvery 6 wkEvery 3 mo, then 6 mo
 N0821652010Pemextrexed, carboplatin, bevacizumabEvery 6 wkEvery 3 mo, then 6 mo
 N9921422005Carboplatin, paclitaxelEvery 8 wkEvery 3 mo
 S0027572003Vinorelbine, docetaxelEvery 10 wkEvery 3 mo, then 6 mo
 S0339532004Gemcitabine, carboplatin, bortezomibEvery 6 wkEvery 6 mo
 S0341552005ErlotinibEvery 6 wkEvery 3 mo, then 6 mo
 S03421322005Carboplatin, paclitaxel, cetuximabEvery 6 wkEvery 3 mo, then 6 mo
SCLC     
 N0027262003Topotecan, carboplatin, G-CSFEvery 6 wkEvery 3 mo, then 6 mo
 N0124212004Imatinib mesylateEvery 8 wkEvery 3 mo, then 6 mo
 N0423462008Pemetrexed disodium, carboplatinEvery 6 wkEvery 6 mo
 N0621232008SaracatinibEvery 6 wkEvery 3 mo
 S0119312003Gemcitabine, irinotecanEvery 6 wkEvery 6 mo
 S0327422004BortezomibEvery 6 wkEvery 3 mo
 S0435582007SorafenibEvery 8 wkEvery 3 mo

Among the patients with NSCLC and SCLC, 23% and 7%, respectively, remained alive at the time of the current analysis with a median follow-up of 1 year and 1.5 years, respectively; and 67% and 66% of patients, respectively, experienced disease progression during treatment. Among the patients with NSCLC and SCLC who had disease progression, the median time from the last progression-free scan to reported progression was 1.4 months for both groups during treatment and 3 months for both groups during follow-up (for a follow-up and progression summary, see Table 2).

Table 2. Follow-Up and Progression Status Summary
VariableNCCTGSWOG
NSCLC, N = 660SCLC, N = 116NSCLC, N = 297SCLC, N = 131
  • Abbreviations: NCCTG, North Central Cancer Treatment Group; NSCLC, nonsmall cell lung cancer; SCLC, small cell lung cancer; SWOG, Southwest Oncology Group; y, years; mo, months.

  • a

    This was reported for patients who progressed on NCCTG trials (NSCLC, N = 520; SCLC, N = 104) and on SWOG trials (NSCLC, N = 123; SCLC, N = 50).

Alive, %22.66.97.13.8
Follow-up for patients who remained alive: Median [range], y1.0 [0.0-6.0]1.5 [0.1-6.3]3.7 [0.8-5.2]2.2 [0.2-2.8]
Progression occurrences: No. of patients (%)a    
 On treatment346 (66.5)69 (66.3)175 (78.1)93 (86.9)
 Long-term follow-up174 (33.5)35 (33.7)49 (21.9)14 (13.1)
Time from last progression-free scan to disease progression: Median [Range], moa
 During treatment1.4 [0.1-4.1]1.4 [0.3-3.1]1.3 [0.03-13.1]1.1 [0.0-2.1]
 During follow-up3.0 [0.0-48.3]3.0 [0.1-14.1]1.7 [0.2-24.3]1.3 [0.4-5.3]
Southwest Oncology Group data

The SWOG data were frozen in December 2010. In total, 297 patients with NSCLC and 131 patients with SCLC were included. Disease assessment schedules of the SWOG trials were similar to those for the NCCTG trials: every 6 weeks (3 NSCLC trials and 2 SCLC trials), 8 weeks (1 SCLC trial), or 10 weeks (1 NSCLC trial) during treatment and every 3 months (2 SCLC trials), 3 months for a specified amount of time followed by 6 months thereafter (3 NSCLC trials), or every 6 months (1 NSCLC trial and 1 SCLC trial) during post-treatment.

The SWOG data also demonstrated good maturity. Among patients with NSCLC and SCLC, compared with NCCTG data, 7% and 4% of patients, respectively, remained alive at the time of the current analysis with a median follow-up of 3.7 years and 2.2 years, respectively; and 78% and 87% of patients, respectively, reported progression during treatment. Among the patients with NSCLC and SCLC who had disease progression, the median time from the last progression-free scan to progression was 1.3 months and 1.1 months, respectively, during treatment and 1.7 months and 1.3 months, respectively, after treatment. Follow-up and progression status are summarized in Table 2.

Overall Progression-Free Survival Estimates

North Central Cancer Treatment Group data

The median PFS across all NSCLC trials was 4.3 months (M1), 1.8 months (M2), 3.3 months (M3), and 3.45/3.52 months (M4; lower/upper limit). The median PFS across all SCLC trials was 2.7 months (M1), 0.03 months (M2), 1.8 months (M3), and 1.45/1.45 months (M4). For M2 compared with M1, the percentage difference in the PFS estimates among patients with NSCLC and SCLC was 52% and 98%, respectively; likewise, for M3 compared with M1, the percentage difference was 29% and 46%, respectively (for a summary of PFS estimates using the 4 methods, see Table 3). For both NSCLC and SCLC, PFS estimates using the RPD were the highest (as expected), with M2 the most conservative. M3 and M4 yielded similar results, because most disease progression occurred during treatment, when frequent disease assessments were performed.

Table 3. Overall Progression-Free Survival Estimates by Method
Progression Date Determination MethodNCCTGSWOG
NSCLC, N = 660SCLC, N = 116NSCLC, N = 297SCLC, N = 131
  1. Abbreviations: CI, confidence interval; NA, not applicable; NCCTG, North Central Cancer Treatment Group; NSCLC, nonsmall cell lung cancer; PFS, progression-free survival; SCLC, small cell lung cancer; SWOG, Southwest Oncology Group; mo, months; y, years.

Method 1    
 Median PFS [95% CI], mo4.3 [3.8-4.7]2.7 [1.6-4.2]3.1 [2.5-4.1]1.3 [1.2-1.6]
 6-Mo estimate [95% CI], %36.5 [32.9-40.4]21.5 [15.1-30.6]31 [26.1-36.7]10.7 [6.5-17.5]
 1-Y estimate [95% CI], %12.3 [9.9-15.3]3.8 [1.4-9.8]12.5 [9.2-16.8]3.8 [1.6-9]
Method 2    
 Median PFS [95% CI], mo1.8 [1.6-2.0]0.03 [0.03-1.4]1.5 [1.2-1.7]0.03 [NA]
 6-Mo estimate [95% CI], %17.1 [14.4-20.3]4.7 [2.0-11]17.2 [13.4-22]5.3 [2.6-11]
 1-Y estimate [95% CI], %7.9 [5.9-10.6]1.3 [0.2-8.1]6.4 [4.1-9.9]3.1 [1.2-8]
Method 3    
 Median PFS [95% CI], mo3.3 [2.7-3.6]1.8 [0.8-3.3]2.2 [1.6-2.8]0.7 [0.6-0.8]
 6-Mo estimate [95% CI], %28.1 [24.7-31.8]13.6 [8.6-21.7]23.6 [19.1-29.1]6.7 [3.4-13.1]
 1-Y estimate [95% CI], %8.9 [6.8-11.7]1.0 [0.1-6.9]8.6 [5.8-12.6]3.4 [1.3-8.8]
Method 4    
 Lower/upper, mo3.45/3.521.45/1.452.79/2.860.03/0.03
Southwest Oncology Group data

The median PFS across all NSCLC trials was 3.1 months (M1), 1.5 months (M2), 2.2 months (M3), and 2.79/2.86 months (M4). The median PFS across all SCLC trials was 1.3 months (M1), 0.03 months (M2), 0.7 months (M3), and 0.03/0.03 months (M4). The percentage difference in the PFS estimates among patients with NSCLC and SCLC was 58% and 99%, respectively, for M2 compared with M1; for M3 compared with M1, the percentage difference was 23% and 33%, respectively (for more detailed summaries, see Table 3). Conclusions from the SWOG data regarding differences in PFS estimates are similar to what was observed in the NCCTG data reported above.

Results of PFS estimates by trial, although not presented, were similar to the results of the overall PFS estimates. The percentage difference in the PFS estimates ranged from as low as 25% to as high as 99% among NSCLC trials (NCCTG and SWOG combined) and ranged from 43% to 98% among SCLC trials for M2 compared with M1; similarly, for M3 compared with M1, the percentage difference in the PFS estimates were between 15% and 47% among NSCLC trials and between 19% and 50% among SCLC trials.

Outcomes for Randomized Trials

For randomized trials (3 NCCTG trials and 1 SWOG trial),15, 24, 26, 30 the 4 methods resulted in the same overall conclusions, and the comparisons across arms in each trial were consistent across the 4 methods (for detailed summaries, see Table 4).

Table 4. Progression-Free Survival Comparisons for Randomized Trials
TrialaProgression Date Determination MethodMethod 4: Lower/Upper, mo
Median PFS, mo
Method 1Method 2Method 3
  • Abbreviations: CI, confidence interval; HR, hazard ratio; PFS, progression-free survival.

  • a

    Trials 1 through 4 were reported by Adjei 2010,15 Dy 2010,24 Molina 2011,26 and Herbst 2010,30 respectively.

  • b

    Trials 1 and 4 did not have control arms. For the purpose of the current analysis, Arm C in Trial 1 and Arm B in Trial 4 were used as reference for HR calculation (see Adjei 201015 and Herbst 201030).

  • c

    This arm was used as the reference for HR calculation.

  • d

    P value for arm comparison (log-rank test).

  • e

    P value for arm comparison (generalized log-rank test).

Trial 1b    
 Arm A4.62.93.73.94/4.24
 Arm B3.61.62.62.46/3.02
 Arm Cc4.21.63.44.14/4.20
 HR [95% CI]    
  A vs C1.10 [0.74-1.63]0.97 [0.65-1.42]1.03 [0.70-1.52]
  B vs C1.27 [0.80-2.02]1.32 [0.83-2.08]1.39 [0.87-2.22]
 Pd.59.24.32.67
Trial 2    
 Arm A5.73.14.64.47/4.60
 Arm Bc4.91.62.94.63/4.99
 HR [95% CI]0.76 [0.47-1.22]0.74 [0.46-1.19]0.73 [0.45-1.17]
 Pd.25.19.18.25
Trial 3    
 Arm A3.11.32.12.60/2.60
 Arm Bc3.81.53.43.48/4.11
 HR [95% CI]0.84 [0.56-1.25]0.86 [0.58-1.29]0.87 [0.58-1.30]
 Pd.38.40.50.67
Trial 4b    
 Arm A6.74.05.54.07/4.11
 Arm Bc5.23.34.23.25/3.84
 HR [95% CI]0.82 [0.53-1.26]0.84 [0.55-1.30]0.83 [0.54-1.28]
 Pe.43.35.39.77

Simulation Results

Eleven percent of patients with NSCLC and 7% of patients with SCLC died without progression on NCCTG trials; and 24% of patients with NSCLC and 16% of patients with SCLC died without progression on trials. Thus, death without progression in both NCCTG and SWOG data represented relatively low percentages. Therefore, our simulated TTP data represent the majority of patients in our trials: patients who remained alive with disease progression (both during treatment and during follow-up) and those who progressed at the time of death. The median TTP for NCCTG NSCLC patients was 4.3 months, which was used as the true median TTP for the simulation study.

M1 (RPD) consistently overestimated the true median TTP across assessment schedules, and M2 (in which the PD was 1 day after the last progression-free scan) consistently underestimated the true median TTP. M3 (midpoint) provided TTP estimates closest to the true median TTP, as reflected by the <1.5% difference in the average median TTP estimates from the true median TTP of 4.3 months. This index also increased with increase in the length of the disease assessment schedule for M1 and M2. The length of the disease assessment schedule did not impact the results of M3 (for details of the simulation results from samples of 50, see Table 5). The results were similar for trials with samples of 100 (data not shown).

Table 5. Simulation Results of the Time-to-Progression Estimates (N = 50; 1000 Simulation Runs)
Disease Assessment Schedule, wkAverage TTP Estimate (% Difference From the True Median TTP), mo
Method 1Method 2Method 3
  1. Abbreviations: TTP, time-to-progression.

45.1 (10.9)4.3 (−8.3)4.6 (1.1)
65.3 (15.5)4.0 (−12.8)4.6 (0.3)
85.6 (21.1)3.8 (−18)4.7 (1.2)
105.8 (26.3)3.5 (−23.6)4.6 (0.3)
126.0 (30.2)3.3 (−28.5)4.6 (−0.2)

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

In this study, we systematically studied the impact of the approach used to declare the progression date on PFS estimates in phase 2 trials of advanced NSCLC and extensive-stage SCLC. Although NCCTG and SWOG trials had different data collection schedules for monitoring disease progression, the results are strikingly consistent. PFS estimates using the RPD were highest because of the length of the assessment interval. M2 (1 day after the last progression-free scan) was the most conservative. M3 (midpoint between the last progression-free scan and the RPD) and M4 (nonparametric interval censoring) yielded similar results, lying between the estimates using M1 and M2, because most disease progression occurred during treatment, when frequent disease assessments were performed (every 4 weeks, 6 weeks, 8 weeks, or 10 weeks). Analysis of randomized trials revealed that the trial conclusions remained unaffected by the method of PD determination. The simulation study also confirmed these findings. Although, in reality, the true median PFS is unknown, our simulation results provide reasonably convincing evidence that the traditional PFS analysis approach of using the RPD may yield misleading results (overestimation), especially in single-arm trials.

Panageas et al10 demonstrated the bias associated with PFS analyses using simulations with PDs marked by lower limit, midpoint, and upper limit of interval censored data. These methods are similar to our approach, except that the lower limit was defined as the date of the last progression-free scan (whereas we used 1 day after the last progression-free scan). Regardless, their study reported that, compared with the prespecified true median PFS, the PFS estimates using the Kaplan-Meier method based on the 3 approaches described above demonstrated a certain pattern: the upper limit-based method consistently overestimates the true median PFS, and the lower limit consistently underestimates it. Our findings, thus, are in line with those of Panageas et al.10

Several possible solutions to the timing and scheduling problem when PFS is used as the primary endpoint have been proposed. Whether in single-arm or randomized trials, 1 simple way to reduce PFS assessment bias introduced by timing and scheduling is to carefully consider the relation between evaluation frequencies and median PFS. Panageas et al10 recommended the interval censoring approach as the solution for PFS analysis. Our results also highlight the need for appropriate interval censoring survival analysis. Although advances in the methodology and the availability of easy-to-use statistical software have simplified the application of interval censoring in survival analysis,36 we note that interpreting the interval censoring results and the associated survival curves remains largely uncommon in the oncology literature. In addition, based on our current analysis, the midpoint approach provides PFS estimates similar to those provided with the interval censoring approach. Therefore, the midpoint approach could be considered a reasonable alternative solution. Once the PD is imputed by using the midpoint approach, it is fairly easy to apply an accepted time-to-event method, such as Kaplan-Meier analysis, with straightforward interpretations. Furthermore, sensitivity analyses for evaluating the robustness of conclusions based on PFS in oncology clinical trials have been recommended by the US Food and Drug Administration38 and by Bhattacharya et al.11

Another approach is to use a binary endpoint for diseases that have a short median PFS, such as advanced NSCLC and SCLC (for example, the 6-month PFS rate). This also was recommend by Panageas et al10 and Friedlin et al.9 With careful attention to trial design, a tumor assessment date could coincide with the primary endpoint (the PFS assessing date; for example, 6 months for the 6-month PFS rate). This is especially important for patients who have not progressed before 6 months, because all patients would be evaluated by the fixed time point of 6 months, which increases the homogeneity of results across trials. Other proposals include a placebo-controlled, double-blind design (when feasible) to limit the evaluation time bias between arms in randomized trials.9 Independent central review for progression-based endpoints also has been proposed and implemented.39 In addition, because the percentage change in tumor burden is less subject to the issues of timing and scheduling, combining it with the PFS endpoint may be another direction worth exploring.40

In conclusion, when PFS is used as the primary endpoint for phase 2 trials in advanced lung cancer, consideration should be taken to minimize the bias caused by timing and scheduling of disease progression assessment. Appropriate statistical analysis, including an interval censoring approach and sensitivity analyses, should be performed to reduce potential biases. The interpretation of trial outcomes based on PFS merits caution. Standards for PD determination, use of sensitivity analyses, and randomized trials are critical when designing trials and reporting efficacy using PFS-based endpoints.

FUNDING SOURCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

This work was supported in part by Public Health Service grants CA-25,224 and CA-15,083 from the National Cancer Institute, US. Department of Health and Human Services.

CONFLICT OF INTEREST DISCLOSURES

The authors made no disclosures.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES
  • 1
    Mandrekar SJ, Qi Y, Hillman SL, et al. Endpoints in phase II trials for advanced non-small cell lung cancer. J Thorac Oncol. 2010; 5: 3-9.
  • 2
    Buyse M, Squifflet P, Laporte S, et al. Prediction of survival benefits from progression-free survival in patients with advanced non small cell lung cancer: evidence from a pooled analysis of 2,838 patients randomized in 7 trials [abstract]. J Clin Oncol. 2008; 26(May 20 suppl). Abstract 8019.
  • 3
    Lara PN Jr, Redman MW, Kelly K, et al. Disease control rate at 8 weeks predicts clinical benefit in advanced non-small-cell lung cancer: results from Southwest Oncology Group randomized trials. J Clin Oncol. 2008; 26: 463-467.
  • 4
    Foster NR, Qi Y, Shi Q, et al. Tumor response and progression-free survival as potential surrogate endpoints for overall survival in extensive stage small-cell lung cancer: findings on the basis of North Central Cancer Treatment Group trials. Cancer. 2011; 117: 1262-1271.
  • 5
    Gelmon KA, Eisenhauer EA, Harris AL, et al. Anticancer agents targeting signaling molecules and cancer cell environment: challenges for drug development. J Natl Cancer Inst. 1999; 91: 1281-1287.
  • 6
    Wilkerson J, Fojo T. Progression-free survival is simply a measure of a drug's effect while administered and is not a surrogate for overall survival. Cancer J. 2009; 15: 379-385.
  • 7
    Tuma R. Progression-free survival remains debatable endpoint in cancer trials. J Natl Cancer Inst. 2009; 101: 1439-1441.
  • 8
    Broglio KR, Berry DA. Detecting an overall survival benefit that is derived from progression-free survival. J Natl Cancer Inst. 2009; 101: 1642-1649.
  • 9
    Freidlin B, Korn EL, Hunsberger S, et al. Proposal for the use of progression-free survival in unblinded randomized trials. J Clin Oncol. 2007; 25: 2122-2126.
  • 10
    Panageas KS, Ben-Porat L, Dickler MN, et al. When you look matters: the effect of assessment schedule on progression-free survival. J Natl Cancer Inst. 2007; 99: 428-432.
  • 11
    Bhattacharya S, Fyfe G, Gray RJ, et al. Role of sensitivity analyses in assessing progression-free survival in late-stage oncology trials. J Clin Oncol. 2009; 27: 5958-5964.
  • 12
    Therasse P, Arbuck SG, Eisenhauer EA, et al. New guidelines to evaluate the response to treatment in solid tumors. J Natl Cancer Inst. 2000; 92: 205-216.
  • 13
    Eisenhauer EA, Therasse P, Bogaert J, et al. New Response Evaluation Criteria in Solid Tumors: revised RECIST guideline (version 1.1). Eur J Cancer. 2009; 45: 228-247.
  • 14
    Kanard A, Jatoi A, Castillo R, et al. Oral vinorelbine for the treatment of metastatic non-small cell lung cancer in elderly patients: a phase II trial of efficacy and toxicity. Lung Cancer. 2004; 43: 345-353.
  • 15
    Adjei AA, Salavaggione OE, Mandrekar SJ, et al. Correlation between polymorphisms of the reduced folate carrier gene (SLC19A1) and survival after pemetrexed-based therapy in non-small cell lung cancer: a North Central Cancer Treatment Group-based exploratory study. J Thorac Oncol. 2010; 5: 1346-1353.
  • 16
    Ma CX, Nair S, Thomas S, et al. Randomized phase II trial of 3 schedules of pemetrexed and gemcitabine as front-line therapy for advanced non-small-cell lung cancer. J Clin Oncol. 2005; 23: 5929-5937.
  • 17
    Mc Kean H, Stella PJ, Hillman SL, et al. Exploring therapeutic decisions in elderly patients with non-small cell lung cancer: results and conclusions from North Central Cancer Treatment Group study N0222. Cancer Invest. 2011; 29: 266-271.
  • 18
    Jatoi A, Hillman SL, Ziegler KL, et al. Is social support associated with improved clinical outcomes in geriatric lung cancer patients? Observations from North Central Cancer Treatment Group studies N9921 and N0222. Cancer Manag Res. 2009; 1: 61-68.
  • 19
    Dy GK, Hillman SL, Rowland KM Jr, et al. A front-line window of opportunity phase 2 study of sorafenib in patients with advanced nonsmall cell lung cancer: North Central Cancer Treatment Group Study N0326. Cancer. 2010; 116: 5686-5693.
  • 20
    Adjei AA, Mandrekar SJ, Dy GK, et al. A phase II trial of pemetrexed plus bevacizumab for second-line therapy of patients with advanced non-small cell lung cancer (NSCLC): an NCCTG and SWOG study. J Clin Oncol. 2010; 28: 614-619.
  • 21
    Chee CE, Jett JR, Bernath AM Jr, et al. Phase II trial of pemetrexed disodium and carboplatin in previously untreated extensive stage small cell lung cancer, N0423. Cancer. 2010; 116: 2382-2389.
  • 22
    Bryce AH, Mattar B, Hillman SL, et al. Phase II trial of oral topotecan and intravenous carboplatin with G-CSF support in previously untreated patients with extensive stage small cell lung cancer: a North Central Cancer Treatment Group Study. Am J Clin Oncol. 2010; 33: 353-357.
  • 23
    Molina JR, Mandrekar SJ, Rowland KM, et al. A phase II NCCTG “window of opportunity” front-line study of the mTOR inhibitor, CCI-779 (temsirolimus) given as a single agent in patients with advanced NSCLC [abstract]. J Thorac Oncol. 2007; 2( 8 suppl): S413.
  • 24
    Dy GK, Mandrekar SJ, Nelson GD, et al. A randomized phase II study of gemcitabine (G) and carboplatin (C) with or without cediranib (AZD2171 [CED]) as first-line therapy in advanced non-small cell lung cancer (NSCLC) [abstract]. J Clin Oncol. 2010; 28( 15 suppl). Abstract 7603.
  • 25
    Molina JR, Foster NR, Nelson GD, et al. A phase II trial of the c-Src inhibitor AZD-0530 after 4 cycles of chemotherapy for patients with extensive stage small cell lung cancer (ES-SCLC). Paper presented at: 13th World Conference on Lung Cancer; July 31 to August 4, 2009; San Francisco, Calif.
  • 26
    Molina JR, Dy GK, Foster NR, et al. A randomized phase II study of pemetrexed (PEM) with or without sorafenib (S) as second line therapy in advanced non-small cell lung cancer (NSCLC) of non-squamous histology: NCCTG N0626 study [abstract]. J Clin Oncol. 2011; 29( suppl). Abstract 7513.
  • 27
    Hesketh PJ, Chansky K, Lau DH, et al. Sequential vinorelbine and docetaxel in advanced non-small cell lung cancer patients age 70 and older and/or with a performance status of 2: a phase II trial of the Southwest Oncology Group (S0027). J Thorac Oncol. 2006; 1: 537-544.
  • 28
    Davies AM, Chansky K, Lara PN Jr, et al. Bortezomib plus gemcitabine/carboplatin as first-line treatment of advanced non-small cell lung cancer: a phase II Southwest Oncology Group Study (S0339). J Thorac Oncol. 2009; 4: 87-92.
  • 29
    Hesketh PJ, Chansky K, Wozniak AJ, et al. Southwest Oncology Group phase II trial (S0341) of erlotinib (OSI-774) in patients with advanced non-small cell lung cancer and a performance status of 2. J Thorac Oncol. 2008; 3: 1026-1031.
  • 30
    Herbst RS, Kelly K, Chansky K, et al. Phase II selection design trial of concurrent chemotherapy and cetuximab versus chemotherapy followed by cetuximab in advanced-stage non-small-cell lung cancer: Southwest Oncology Group study S0342. J Clin Oncol. 2010; 28: 4747-4754.
  • 31
    Akerley W, McCoy J, Hesketh PJ, et al. Gemcitabine and irinotecan for patients with untreated extensive stage small cell lung cancer: SWOG 0119. J Thorac Oncol. 2007; 2: 526-530.
  • 32
    Lara PN Jr, Chansky K, Davies AM, et al. Bortezomib (PS-341) in relapsed or refractory extensive stage small cell lung cancer: a Southwest Oncology Group phase II trial (S0327). J Thorac Oncol. 2006; 1: 996-1001.
  • 33
    Gitlitz BJ, Moon J, Glisson BS, et al. Sorafenib in platinum-treated patients with extensive stage small cell lung cancer: a Southwest Oncology Group (SWOG 0435) phase II trial. J Thorac Oncol. 2010; 5: 1835-1840.
  • 34
    Wellner JA, Zhan Y. A hybrid algorithm for computation of the nonparametric maximum likelihood estimator from censored data. J Am Stat Assoc. 1997; 92: 945-959.
  • 35
    Zhao Q, Sun J. Generalized log-rank test for mixed interval-censored failure time data. Stat Med. 2004; 23: 1621-1629.
  • 36
    So Y, Johnston G, Kim SH. Analyzing Interval-Censored Survival Data With SAS Software. Cary, NC: SAS Institute and the University of North Carolina-Chapel Hill; 2010. Macros available at: http://support.sas.com/kb/24980.html. Accessed January 28, 2011.
  • 37
    The R Project for Statistical Computing. R interval censoring package—Icens. Available at: http://cran.r-project.org/web/packages/Icens/index.html. Accessed May 10, 2011.
  • 38
    US Food and Drug Administration. Guidance for industry: clinical trial endpoints for the approval of cancer drugs and biologics, 2007 Available at: http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/UCM071590.pdf. Accessed March 20, 2011.
  • 39
    Amit O, Bushnell W, Dodd L, Roach N, et al. Blinded independent central review of the progression-free survival endpoint. Oncologist. 2010; 15: 492-495.
  • 40
    Fridlyand J, Kaiser LD, Fyfe G. Analysis of tumor burden versus progression-free survival for Phase II decision making. Contemp Clin Trials. 2011; 32: 446-452.