This Article Abstract Full Text (PDF) Data Supplement Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Save to my personal folders Download to citation manager Rights & Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Korn, E. L. Articles by Kirkwood, J. M. Search for Related Content PubMed PubMed Citation Articles by Korn, E. L. Articles by Kirkwood, J. M. Related Articles Related Editorial Social Bookmarking                            What's this?

Meta-Analysis of Phase II Cooperative Group Trials in Metastatic Stage IV Melanoma to Determine Progression-Free and Overall Survival Benchmarks for Future Phase II Trials

Edward L. Korn, Ping-Yu Liu, Sandra J. Lee, Judith-Anne W. Chapman, Donna Niedzwiecki, Vera J. Suman, James Moon, Vernon K. Sondak, Michael B. Atkins, Elizabeth A. Eisenhauer, Wendy Parulekar, Svetomir N. Markovic, Scott Saxman, John M. Kirkwood

From the Biometric Research Branch, Division of Cancer Treatment and Diagnosis, National Cancer Institute, Bethesda, MD; Southwest Oncology Group Statistical Office, Seattle, WA; Eastern Cooperative Oncology Group; and Department of Medicine, Beth Israel Deaconess Medical Center, Boston, MA; National Cancer Institute of Cancer Clinical Trials Group, Queen's University, Kingston, Ontario, Canada; Cancer and Leukemia Group B Statistical Center, Durham, NC; Mayo Clinic, Rochester, MN; H. Lee Moffitt Cancer Center and Research Institute, University of South Florida, Tampa, FL; Peace Corps, Washington, DC; and Department of Medicine, University of Pittsburgh School of Medicine, Pittsburgh, PA

Corresponding author: Edward L. Korn, PhD, Biometric Research Branch, EPN-8129, Division of Cancer Treatment and Diagnosis, National Cancer Institute, Bethesda, MD 20892; e-mail: korne{at}ctep.nci.nih.gov

	ABSTRACT

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION Authors' Disclosures of... Author Contributions Appendix REFERENCES

 
Purpose Objective tumor response rates observed in phase II trials for metastatic melanoma have historically not provided a reliable indicator of meaningful survival benefits. To facilitate using overall survival (OS) or progression-free survival (PFS) as an endpoint for future phase II trials, we evaluated historical data from cooperative group phase II trials to attempt to develop benchmarks for OS and PFS as reference points for future phase II trials.

Patients and Methods Individual-level and trial-level data were obtained for patients enrolled onto 42 phase II trials (70 trial arms) that completed accrual in the years 1975 through 2005 and conducted by Southwest Oncology Group, Eastern Cooperative Oncology Group, Cancer and Leukemia Group B, North Central Cancer Treatment Group, and the Clinical Trials Group of the National Cancer Institute of Canada. Univariate and multivariate analyses were performed to identify prognostic variables, and between-trial(-arm) variability in 1-year OS rates and 6-month PFS rates were examined.

Results Statistically significant individual-level and trial-level prognostic factors found in a multivariate survival analysis for OS were performance status, presence of visceral disease, sex, and whether the trial excluded patients with brain metastases. Performance status, sex, and age were statistically significant prognostic factors for PFS. Controlling for these prognostic variables essentially eliminated between-trial variability in 1-year OS rates but not in 6-month PFS rates.

Conclusion Benchmarks are provided for 1-year OS or OS curves that make use of the distribution of prognostic factors of the patients in the phase II trial. A similar benchmark for 6-month PFS is provided, but its use is more problematic because of residual between-trial variation in this endpoint.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION Authors' Disclosures of... Author Contributions Appendix REFERENCES

 
New agents are needed for the treatment of metastatic melanoma because no evidence of survival prolongation with existing therapy has been established. Phase II clinical trials offer a means to screen therapies for further testing in the phase III setting. However, phase II designs require benchmarks for deciding whether a new therapy is worth pursuing. Historically, phase II trials have tended to use objective response rate (tumor shrinkage) as a benchmark. However, the few therapies that have demonstrated promising response rates in patients with metastatic stage IV melanoma have not had meaningful effects on survival. In addition, tumor shrinkage as a benchmark may not be appropriate for targeted molecular and immunologic therapies that could offer survival benefits for patients manifesting only disease stabilization. Both of these considerations suggest that it might be more appropriate to use an overall survival (OS) or progression-free survival (PFS) endpoint as a benchmark for future phase II trials.¹ It may be possible that such benchmarks derived from historical data will allow more effective selection of new regimens for phase III testing on the basis of results obtained in future cooperative group phase II trials (avoiding the cost and time of concurrent control arms as a reference). The development of these benchmarks has been investigated by performing a meta-analysis of previously collected data from metastatic melanoma phase II trials performed by the participating cooperative groups.

	PATIENTS AND METHODS

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION Authors' Disclosures of... Author Contributions Appendix REFERENCES

 
Individual and Trial-Level Data
Individual patient data from participants in all metastatic melanoma phase II trials (including randomized phase II trials) that completed accrual in the years 1975 through 2005 were requested. Trials were identified both by a literature search and by searching records in the statistical offices of the participating groups. Individual-level variables considered were sex, age, Eastern Cooperative Oncology Group performance status (PS), presence of visceral metastases (VISC), serum lactate dehydrogenase (LDH) level, OS, and PFS. OS was measured from the time of registration to death (or censoring); PFS was measured from the time of registration to progression (typically, a 25% to 50% increase in the sum of bidimensional products of all measurable lesions or appearance of new lesions) or death from any cause (or censoring). Trial-level variables considered were exclusion of patients with brain metastases (BRAIN-METS), exclusion of patients with liver metastases, exclusion of patients with visceral metastases, previous treatment for metastatic disease, and the year during which accrual was completed. For the purposes of this study, all treatments were considered inactive.

Statistical Methods
OS and PFS distributions were estimated using Kaplan-Meier plots. Prognostic variables for OS and PFS were assessed using likelihood-ratio tests from proportional hazards modeling. Prognostic variables for event rates (1-year OS rate and 6-month PFS rate) were assessed using logistic regression models; the small number of observations that were censored before the relevant time point (eg, 1 year) were omitted. The reported P values for statistical significance are all two-sided and unadjusted for multiple comparisons, with P .05 taken as being statistically significant.

Exploration of between trial-arm variability in event rates was examined by two methods (detailed in Appendix A, online only). In one method, the event rate for each of the treatment arms was compared with the overall event rate to look for outliers. The second method used a logistic-normal model, where the variance of the normal effect is an estimate of the underlying between trial-arm variability not accounted for by differing distributions of prognostic variables. If this variability is large, then comparisons to a historical control rate are problematic² and may lead to too many false-positive trials (Appendix A).

All analyses were performed using SAS (Version 9.1; SAS Institute, Cary, NC).

	RESULTS

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION Authors' Disclosures of... Author Contributions Appendix REFERENCES

 
Individual-level data were obtained from 42 trials (70 trial arms) involving 2,100 patients (Appendix B, online only). The years of closure to accrual ranged from 1977 through 2005. LDH level was available for only 136 patients and was not considered further. Two trial-level variables were not considered because all the trials included patients with visceral disease, and only two trials (n = 55) excluded patients with liver metastases.

Prognostic Variables for OS
For the 2,100 patients, 73 patients had censored observations for OS (either lost to follow-up or still alive at the time of this data collection), 25 patients of whom had censored observations before 365 days. The median survival time was 6.2 months (95% CI, 5.9 months to 6.5 months), with 25.5% (95% CI, 23.6% to 27.4%) alive at 1 year (Fig 1A). In the univariate analyses, PS was the most important individual-level prognostic variable (Fig 1B), followed by presence of visceral disease (Fig 1C) and sex (Fig 1D); age was not prognostic (Fig 1E). Whether the trial excluded patients with brain metastases (Fig 1F) and the year during which the trial closed (Fig 1G) were important trial-level prognostic variables; whether the trial excluded patients who had previous treatment was not prognostic (Fig 1H). Details of the univariate analyses are given in Appendix Table A1, online only.

View larger version (22K): [in this window] [in a new window] [PowerPoint Slide for Teaching]   Fig 1. Overall survival (A) categorized by (B) performance status, (C) presence/absence of visceral disease, (D) sex, (E) age, (F) whether the trial excluded/allowed patients with brain metastases, (G) year trial closed, and (H) whether the trial excluded/allowed previous treatment.

View larger version (21K): [in this window] [in a new window] [PowerPoint Slide for Teaching]   Fig 2. Progression-free survival (PFS) (A) categorized by (B) performance status, (C) presence/absence of visceral disease, (D) sex, (E) age, (F) whether the trial excluded/allowed patients with brain metastases, (G) year trial closed, and (H) whether the trial excluded/allowed previous treatment.

View larger version (12K): [in this window] [in a new window] [PowerPoint Slide for Teaching]   Fig 3. Event rates for each trial arm versus the sample size in the trial arm: (A) overall survival (OS) rates at 1 year, (B) progression-free survival (PFS) rates at 6 months. The solid lines are 95% confidence bounds. The dotted line is the overall 1-year survival rate (25%) or the overall 6-month PFS rate (15%). (A small number of plotted points have been slightly jittered to avoid complete overlap.)

Benchmarks for Future Phase II Trials
Regardless of whether previous trials showed between-trial variation in survival rates, future trials may have different rates than in the past because of patient mixes that differ in terms of prognostic variables. To address this, we consider defining the null hypothesis target for a phase II trial based on the prognostic variables recorded in the trial. Table 3 contains the relevant information for a trial using a 1-year OS rate as the endpoint. These predicted values are based on a logistic regression analysis with effects included for PS, sex, VISC, and BRAIN-METS.

What should the sample size be? If an expected 1-year OS rate for phase II trials conducted at the participating institution(s) is available, we recommend the following: Let ₀ be the expected rate. Choose the sample size (using the binomial distribution) so that a trial testing the null hypothesis that the 1-year OS rate ₀ will have 90% power to detect the alternative hypothesis that the 1-year OS rate is more than ₀+15%. If no expected rate is available, use ₀ = 35% (yielding a sample size of 72 patients). A trial with 72 patients will have 85% to 90% power to detect an increase of 15 percentage points in the 1-year OS rate over the historical control rate (with one-sided type 1 error 10%).

Alternatively, one can calculate an historical OS survival curve (Appendix C, online only), which can then be used for comparison with the observed phase II OS data on the new trial, again using a P value of less than .10 to decide whether the new regimen should be pursued. This latter approach will lead to a smaller sample size. For example, with 1 year of accrual and 1 year of follow-up, a sample size of 63 patients (instead of 72 patients) would be required to detect a hazard ratio of 1.51, which corresponds to an improvement in 1-year OS from 35% to 50%.

For 6-month PFS rates, the same approach can be used, except that the calculation of the benchmark 6-month PFS rate depends only on the PS of the patients on the trial (this being by far the most important prognostic variable). In particular, one calculates the average of the predicted values for the patients in the trial using the predicted rates of 18.0% for PS 0 patients, 12.3% for PS 1 patients, 7.4% for PS 2 patients, and 2.9% for PS 3 patients (Appendix Table A2, online only). The sample size of the trial can again be chosen to detect a 15 percentage point improvement over the historical rate ₀ of 6-month PFS. If no historical rate is available, use ₀ = 15%, yielding a sample size of 53 patients. However, because of the between-trial variability in PFS rates, the true type 1 error for phase II trials using this approach may be larger than the nominal 10%. For example, if the between-trial variance were 0.191 (Appendix Table A4, online only), then the actual type 1 error could possibly be as high as 80%, although there is the possibility of using a value larger than for the null hypothesis to lessen the type 1 error (Appendix A). We do not recommend comparisons with the whole historical control PFS curve, as assessment frequencies may unduly influence this curve.⁵

	DISCUSSION

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION Authors' Disclosures of... Author Contributions Appendix REFERENCES

 
Combinations of the prognostic variables for OS found in this study (PS, VISC, BRAIN-METS, and sex) have been noted in other studies of metastatic melanoma,^6-17 including some studies^18-21 whose trials partially overlap with the trials considered here. Additional variables not available for analysis here have been found to be prognostic for overall survival, including LDH and other laboratory biomarkers,^{11,12,16,17,19,22} number of metastatic sites,^{10,14,18,19,20} and time from diagnosis to metastases.^7,10,13,15 To our knowledge, prognostic variables for PFS have not been studied, so that the finding of the prognostic ability of PS and lack of important prognostic ability of the other variables considered is new. However, there may be prognostic variables not considered in this study, which in the future could be incorporated into the modeling. In any event, what is important for determining an historical control benchmark is not that one has controlled for all important prognostic variables, but that it is unlikely that (1) there will be a large effect of unmeasured prognostic variables when the known prognostic variables are accounted for, or (2) levels of the unmeasured variables in future trials will be different than in the historical trials.

The choice of the time points of 1 year for OS rates and 6 months for PFS rates were somewhat arbitrary. We wanted to choose a time that was long enough so that the rates would be clinically meaningful and relatively low (to increase statistical power), but short enough to avoid a lengthy trial. For OS, one can use an historical control OS curve (as described in Appendix C) to avoid choosing a single time point for comparison of rates. The choice of a 15 percentage point improvement in rates for sample size determination was also arbitrary and can easily be changed. However, this improvement over a 35% 1-year OS rate corresponds to a hazard ratio of 1.51. This is not an insubstantial benefit to be targeting, so one would not want to target a larger difference. On the other hand, because the decision point for declaring a treatment worthy of further study in a phase II trial is approximately one half the targeted difference, only an observed improvement of 7 to 8 percentage points over the historical control would be required for a so-called positive trial that had a targeted 15 percentage point difference. Therefore, targeting a smaller difference may not be wise, because a positive trial may then not be convincing enough evidence to proceed with further development. If one had a series of trials with effective agents in addition to the trials considered in this study, then one could choose time points and targeted differences to maximize the ability of the design to identify effective agents.^23,24

The choice of whether to use OS or PFS as the primary endpoint is not straightforward. OS has the most unequivocal relevance, although it is not necessary to show clinically relevant benefit in a phase II trial (whose raison d'être is to show sufficient activity to begin a phase III trial). In addition, 6-month PFS results are obtained 6 months earlier than 1-year OS results, and a trial with a PFS endpoint may require a smaller sample size if one is willing to hypothesize a larger treatment effect (hazard ratio) for PFS than OS. Conversely, the use of OS does allow one to compare the whole survival curve with a historical curve, which can offer some benefits in a smaller sample size. In addition, the residual trial-arm variation seen with the PFS endpoint may lead to an excessive number of false-positive phase II trials. Therefore, we recommend OS as the primary endpoint. Six-month PFS rates could be used for an early assessment of the agent in the trial, after which the assessment of mature OS data would validate the decision.

Phase II trials with objective response endpoints frequently use a two-stage design in which the trial is stopped after the first stage if a minimal number of responses is not seen.²⁵ In principle, these two-stage designs can be applied to the binary endpoints of alive/dead at 1 year or progression free at 6 months. However, this would require temporarily stopping accrual after the first stage while the survival data mature. One possibility is to have multiple trials ongoing, so that while awaiting first-stage results from one agent, one could be accruing patients to a trial of a second agent.

The historical control benchmarks for OS developed in this article allow one to perform single-arm phase II trials. An alternative strategy is to conduct a randomized phase II screening trial²⁶ in which patients are randomly assigned to the experimental or control treatment. The advantages of this approach are that there are no questions concerning the adequacy of an historical control experience, and PFS can safely be used as the endpoint (in a blinded trial). The major disadvantage is its larger required sample size. For example, to detect a hazard ratio of 1.51 (which corresponds to 1-year OS rates of 35% v 50%), 220 patients would need to be accrued (over 1 year, with 1 year of additional follow-up) to have the same type 1 and 2 errors as an historical control comparison with 63 patients (or 74 patients if only the 1-year survival rates are used and not the complete survival distributions). It is possible to reduce the size of the randomized screening trial by relaxing the type 1 and 2 errors. Another disadvantage of the randomized screening design is that patients may be reluctant about having their randomized treatment be an older control therapy. Whether a randomized screening design or an historically controlled phase II trial is used, a trial demonstrating promising activity will need to be followed by a definitive randomized phase III trial.

	Authors' Disclosures of Potential Conflicts of Interest

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION Authors' Disclosures of... Author Contributions Appendix REFERENCES

 
Although all authors completed the disclosure declaration, the following author(s) indicated a financial or other interest that is relevant to the subject matter under consideration in this article. Certain relationships marked with a "U" are those for which no compensation was received; those relationships marked with a "C" were compensated. For a detailed description of the disclosure categories, or for more information about ASCO's conflict of interest policy, please refer to the Author Disclosure Declaration and the Disclosures of Potential Conflicts of Interest section in Information for Contributors.

Employment or Leadership Position: None Consultant or Advisory Role: Vernon K. Sondak, Schering-Plough (C), Pfizer (C), Bayer-Onyx (C); Michael B. Atkins, Novartis (C), Bayer (C), Bristol-Myers Squibb Co (C); John M. Kirkwood, Eleos, Inc (C), Intrexon, Inc (C), Antigenics, Inc (C) Stock Ownership: None Honoraria: Michael B. Atkins, Novartis; John M. Kirkwood, Pharmadura CME, Stategic Edge Research Funding: John M. Kirkwood, Pfizer, Bristol-Myers Squibb Co, AstraZeneca Expert Testimony: None Other Remuneration: None

	Author Contributions

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION Authors' Disclosures of... Author Contributions Appendix REFERENCES

 
Conception and design: Edward L. Korn, Ping-Yu Liu, Vernon K. Sondak, Michael B. Atkins, Elizabeth A. Eisenhauer, Scott Saxman, John M. Kirkwood

Provision of study materials or patients: Vernon K. Sondak, Elizabeth A. Eisenhauer, Svetomir N. Markovic, John M. Kirkwood

Collection and assembly of data: Edward L. Korn, Ping-Yu Liu, Sandra J. Lee, Judith-Anne W. Chapman, Donna Niedzwiecki, Vera J. Suman, James Moon, Vernon K. Sondak, Elizabeth A. Eisenhauer, Wendy Parulekar, Svetomir N. Markovic, Scott Saxman, John M. Kirkwood

Data analysis and interpretation: Edward L. Korn, Sandra J. Lee, Judith-Anne W. Chapman, Vera J. Suman, Vernon K. Sondak, John M. Kirkwood

Manuscript writing: Edward L. Korn, Judith-Anne W. Chapman, Vera J. Suman, Vernon K. Sondak, Michael B. Atkins, Svetomir N. Markovic, John M. Kirkwood

Final approval of manuscript: Edward L. Korn, Ping-Yu Liu, Sandra J. Lee, Judith-Anne W. Chapman, Donna Niedzwiecki, Vera J. Suman, Vernon K. Sondak, Michael B. Atkins, Elizabeth A. Eisenhauer, Wendy Parulekar, Svetomir N. Markovic, Scott Saxman, John M. Kirkwood

	Appendix

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION Authors' Disclosures of... Author Contributions Appendix REFERENCES

 
Appendix A: Methods for Exploring Between Trial-Arm Variability in Event Rates
Two methods were used. In one method, the event rate for each of the k treatment arms was compared with the overall event rate using a Bonferroni correction for the multiple comparisons. In particular, the event rate for treatment arm i (r_i) with n_i observations was considered different than the overall event rate (r) with statistical significance P < .05 if r_i was not contained in the interval:

where z_1–.025/k is the 1–.025/k quantile of a standard normal distribution and n is the total sample size.

The second method used a logistic-normal model:

where _ij is the probability that the jth person in the ith trial arm would be a survivor at the time point of interest, x_ij represents the values of prognostic variables for this person, β is vector of parameters associated with the prognostic variables, and _i is a random residual trial-arm effect with an assumed normal distribution. Regardless of whether outlying trials are identified by the first method, a large variance component associated with the residual between-trial effect will inflate the type 1 error of a historically controlled phase II trial (Table A5).

One possibility to lessen the type 1 error when there is a nonzero between-trial variance component is to set the null hypothesis rate higher than the historical control rate. For example, suppose the average historical control rate is 15% and the between-trial variance component is 0.20. Then the historical rate for any given trial could reasonably range from 8% to 27%, and the type I error of a nominal 10% type 1 error hypothesis test of 15% versus 30% could reasonably range from less than 1% to 80% (line 5, Table A5). An ad hoc approach to this problem would be to use a trial design that targeted 27% (the 95th percentile) versus 42% (27% + 15%) rates, which would require a sample size of 66 patients, to reasonably ensure that the type 1 error is less than 10%.

Appendix B: Description of Trials and Trial Arms

Appendix C: Constructing the Historical Control Overall Survival Distribution for a Trial
To calculate the historical control curve for a trial with n patients, for patient i let:

where β_ps = 0 if the patient has performance status of 0, β_ps = 0.436 if the patient has performance status of 1, β_ps = 0.948 if the patient has performance status of 2 or 3, β_sex = 0.248 and SEX = 1 if the patient is male and 0 if the patient is female, β_visc = 0.421 and VISC = 1 if the patient has visceral disease and 0 if not, and β_brain = 0.304 and BRAIN = 0 if patients with brain metastases are excluded from the trial and 1 if not.

The historical control curve is then given by:

where

and S₀(t) is given in the Supplementary File. At the time of analysis of the new phase II trial, the survival curve of this trial based on n patients is compared with the historical survival curve. There are various methods to obtain the statistical significance of such a comparison. One is to perform a stratified analysis treating the S_i(t) as if they were life-table estimates.²⁹

	ACKNOWLEDGMENTS

 
We thank Lesley Seymour, MD, PhD, and Wendy Walsh, BSc, for supplying NCIC Clinical Trials Group data and Michael Smylie, MD, for providing developmental input to the meta-analysis.

	NOTES

 
Authors' disclosures of potential conflicts of interest and author contributions are found at the end of this article.

	REFERENCES

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION Authors' Disclosures of... Author Contributions Appendix REFERENCES

 
1. Korn EL, Arbuck SG, Pluda JM, et al: Clinical trial designs for cytostatic agents: Are new approaches needed? J Clin Oncol 19: 265-272, 2001[Abstract/Free Full Text]

2. Taylor JMG, Braun TM, Li Z: Comparing an experimental agent to a standard agent: Relative merits of a one-am or randomized two-arm Phase II design. Clin Trials 3: 335-348, 2006[Abstract/Free Full Text]

3. Margolin KA, Liu PY, Flaherty LE, et al: Phase II study of carmustine, dacarbazine, cisplatin, and tamoxifen in advanced melanoma: A Southwest Oncology Group study. J Clin Oncol 16: 664-669, 1998[Abstract]

4. Balch CM, Buzaid AC, Soong S-J, et al: Final version of the American Joint Committee on Cancer staging system for cutaneous melanoma. J Clin Oncol 19: 3635-3648, 2001[Abstract/Free Full Text]

5. Panageas K, Ben-Porat L, Dickler MN, et al: When you look matters: The effect of assessment schedule on progression-free survival. J Natl Cancer Inst 99: 428-432, 2007[Abstract/Free Full Text]

6. Einhorn LH, Burgess MA, Vallejos C, et al: Prognostic correlations and response to treatment in advanced metastatic malignant melanoma. Cancer Res 34: 1995-2004, 1974[Abstract/Free Full Text]

7. Amer MH, Al-Sarraf M, Vaitkevicius VK: Clinical presentation, natural history and prognostic factors in advanced malignant melanoma. Surg Gynecol Obstet 149: 687-692, 1979[Medline]

8. Rampen F: Malignant melanoma: Sex differences in survival after evidence of distant metastases. Br J Cancer 42: 52-57, 1980[Medline]

9. Presant CA, Bartolucci AA: Prognostic factors in malignant melanoma: The Southeastern Cancer Study Group experience. Cancer 49: 2192-2196, 1982[CrossRef][Medline]

10. Balch CM, Soong S, Murad TM, et al: A multifactorial analysis of melanoma: IV. Prognostic factors in 200 melanoma patients with distant metastases. J Clin Oncol 1: 126-134, 1983[Abstract]

11. Heimdal K, Hannisdal E, Gundersen S: Regression analyses of prognostic factors in metastatic malignant melanoma. Eur J Cancer Clin Oncol 25: 1219-1223, 1989[CrossRef][Medline]

12. Sirott MN, Bajorin DF, Wong GYC, et al: Prognostic factors in patients with metastatic malignant melanoma. Cancer 72: 3091-3098, 1993[CrossRef][Medline]

13. Barth A, Wanek LA, Morton DL: Prognostic factors in 1,521 melanoma patients with distant metastases. J Am Coll Surg 181: 193-201, 1995[Medline]

14. Brand CU, Ellwanger U, Stroebel W, et al: Prolonged survival of 2 years or longer for patients with disseminated melanoma. Cancer 79: 2345-2353, 1997[CrossRef][Medline]

15. Falkson CI, Falkson HC: Prognostic factors in metastatic malignant melanoma. Oncology 55: 59-64, 1998[CrossRef][Medline]

16. Keilholz U, Conradt C, Legha SS, et al: Results of interleukin-2-based treatment in advanced melanoma: A case record-based analysis of 631 patients. J Clin Oncol 16: 2921-2929, 1998[Abstract/Free Full Text]

17. Eton O, Legha SS, Moon TE, et al: Prognostic factors for survival of patients treated systemically for disseminated melanoma. J Clin Oncol 16: 1103-1111, 1998[Abstract]

18. Ryan L, Kramar A, Borden E: Prognostic factors in metastatic melanoma. Cancer 71: 2995-3005, 1993[CrossRef][Medline]

19. Manola J, Atkins M, Ibrahim J, et al: Prognostic factors in metastatic melanoma: A pooled analysis of Eastern Cooperative Oncology Group trials. J Clin Oncol 18: 3782-3793, 2000[Abstract/Free Full Text]

20. Unger JM, Flaherty LE, Liu PY, et al: Gender and other survival predictors in patients with metastatic melanoma on Southwest Oncology Group trials. Cancer 91: 1148-1155, 2001[CrossRef][Medline]

21. Balch CM, Soong S, Gershenwald JE, et al: Prognostic factors analysis of 17,600 melanoma patients: Validation of the American Joint Committee on Cancer melanoma staging system. J Clin Oncol 19: 3622-3634, 2001[Abstract/Free Full Text]

22. Deichmann M, Benner A, Bock M, et al: S100-beta, melanoma-inhibiting activity, and lactate dehydrogenase discriminate progressive from nonprogressive American Joint Committe on Cancer stage IV melanoma. J Clin Oncol 17: 1891-1896, 1999[Abstract/Free Full Text]

23. Van Glabbeke M, Verwij J, Judson I, et al: Progression-free rate as the principal end-point for phase II trials in soft-tissue sarcomas. Eur J Cancer 38: 543-549, 2002[CrossRef][Medline]

24. Francart J, Legrand C, Sylvester R, et al: Progression-free survival rate as primary endpoint for phase II cancer clinical trials: Application to mesothelioma—The EORTC Lung Cancer Group. J Clin Oncol 24: 3007-3012, 2006[Abstract/Free Full Text]

25. Simon R: Optimal two-stage designs for phase II clinical trials. Control Clin Trials 10: 1-10, 1989[Medline]

26. Rubinstein LV, Korn EL, Freidlin B, et al: Design issues of randomized phase II trials and a proposal for phase II screening trials. J Clin Oncol 23: 7199-7206, 2005[Abstract/Free Full Text]

27. Kish JA, Kopecky K, Samson MK, et al: Evaluation of fludarabine phosphate in malignant melanoma. Invest New Drugs 9: 105-108, 1991[Medline]

28. Amato DA, Bruckner H, Guerry D, et al: Phase II evaluation of dibromodulcitol and actinomycin D, hydroxyurea, and cyclophosphamide in previously untreated patients with malignant melanoma. Invest New Drugs 5: 293-297, 1987[Medline]

29. Breslow NE: Analysis of survival data under the proportional hazards model. Int Stat Rev 43: 45-58, 1975

Submitted May 25, 2007; accepted August 28, 2007.

CiteULike    Complore    Connotea    Del.icio.us    Digg    Facebook    Reddit    Technorati    Twitter    What's this?

Related Editorial

• Using Benchmarks Based on Historical Survival Rates for Screening New Therapies for Stage IV Melanoma Patients
  Phyllis A. Gimotty, DuPont Guerry, and Keith Flaherty
  JCO 2008 26: 517-518 [Full Text]

This article has been cited by other articles:

				D. R. Minor, D. Moore, C. Kim, M. Kashani-Sabet, S. S. Venna, W. Wang, P. Boasberg, and S. O'Day Prognostic Factors in Metastatic Melanoma Patients Treated with Biochemotherapy and Maintenance Immunotherapy Oncologist, October 1, 2009; 14(10): 995 - 1002. [Abstract] [Full Text] [PDF]

				C. L.H. Snozek, D. J. O'Kane, and A. Algeciras-Schimnich Pharmacogenetics of Solid Tumors: Directed Therapy in Breast, Lung, and Colorectal Cancer: A Paper from the 2008 William Beaumont Hospital Symposium on Molecular Pathology J. Mol. Diagn., September 1, 2009; 11(5): 381 - 389. [Abstract] [Full Text] [PDF]

				J. Weber, J. A. Thompson, O. Hamid, D. Minor, A. Amin, I. Ron, R. Ridolfi, H. Assi, A. Maraveyas, D. Berman, et al. A Randomized, Double-Blind, Placebo-Controlled, Phase II Study Comparing the Tolerability and Efficacy of Ipilimumab Administered with or without Prophylactic Budesonide in Patients with Unresectable Stage III or IV Melanoma Clin. Cancer Res., September 1, 2009; 15(17): 5591 - 5598. [Abstract] [Full Text] [PDF]

				C. Robert and F. Ghiringhelli What Is the Role of Cytotoxic T Lymphocyte-Associated Antigen 4 Blockade in Patients with Metastatic Melanoma? Oncologist, August 1, 2009; 14(8): 848 - 861. [Abstract] [Full Text] [PDF]

				M. LeBlanc, C. Rankin, and J. Crowley Multiple Histology Phase II Trials Clin. Cancer Res., July 1, 2009; 15(13): 4256 - 4262. [Abstract] [Full Text] [PDF]

				L. Rubinstein, J. Crowley, P. Ivy, M. LeBlanc, and D. Sargent Randomized Phase II Designs Clin. Cancer Res., March 15, 2009; 15(6): 1883 - 1890. [Abstract] [Full Text] [PDF]

				L. H. Camacho, S. Antonia, J. Sosman, J. M. Kirkwood, T. F. Gajewski, B. Redman, D. Pavlov, C. Bulanhagui, V. A. Bozon, J. Gomez-Navarro, et al. Phase I/II Trial of Tremelimumab in Patients With Metastatic Melanoma J. Clin. Oncol., March 1, 2009; 27(7): 1075 - 1081. [Abstract] [Full Text] [PDF]

				A. Sharma, A. K. Sharma, S. V. Madhunapantula, D. Desai, S. J. Huh, P. Mosca, S. Amin, and G. P. Robertson Targeting Akt3 Signaling in Malignant Melanoma Using Isoselenocyanates Clin. Cancer Res., March 1, 2009; 15(5): 1674 - 1685. [Abstract] [Full Text] [PDF]

				M. B. Atkins, J. Hsu, S. Lee, G. I. Cohen, L. E. Flaherty, J. A. Sosman, V. K. Sondak, and J. M. Kirkwood Phase III Trial Comparing Concurrent Biochemotherapy With Cisplatin, Vinblastine, Dacarbazine, Interleukin-2, and Interferon Alfa-2b With Cisplatin, Vinblastine, and Dacarbazine Alone in Patients With Metastatic Malignant Melanoma (E3695): A Trial Coordinated by the Eastern Cooperative Oncology Group J. Clin. Oncol., December 10, 2008; 26(35): 5748 - 5754. [Abstract] [Full Text] [PDF]

				J. R. Kirshner, S. He, V. Balasubramanyam, J. Kepros, C.-Y. Yang, M. Zhang, Z. Du, J. Barsoum, and J. Bertin Elesclomol induces cancer cell apoptosis through oxidative stress Mol. Cancer Ther., August 1, 2008; 7(8): 2319 - 2327. [Abstract] [Full Text] [PDF]

				S. J. Moschos, P. M. Chaudhary, and J. M. Kirkwood Resolving "Kinks" of Chemotherapy in Melanoma J Natl Cancer Inst, June 18, 2008; 100(12): 833 - 835. [Full Text] [PDF]

				P. A. Gimotty, D. Guerry, and K. Flaherty Using Benchmarks Based on Historical Survival Rates for Screening New Therapies for Stage IV Melanoma Patients J. Clin. Oncol., February 1, 2008; 26(4): 517 - 518. [Full Text] [PDF]

This Article Abstract Full Text (PDF) Data Supplement Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Save to my personal folders Download to citation manager Rights & Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Korn, E. L. Articles by Kirkwood, J. M. Search for Related Content PubMed PubMed Citation Articles by Korn, E. L. Articles by Kirkwood, J. M. Related Articles Related Editorial Social Bookmarking                            What's this?