|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Journal of Clinical Oncology, Vol 23, No 1 (January 1), 2005: pp. 184-189 © 2005 American Society of Clinical Oncology. DOI: 10.1200/JCO.2005.07.050 Statistical Validation of the EORTC Prognostic Model for Malignant Pleural Mesothelioma Based on Three Consecutive Phase II TrialsFrom the Lung Cancer and Mesothelioma Unit, Department of Medical Oncology, and the Institute of Cell and Molecular Science—Pathology, St Bartholomew’s Hospital; London Lung Cancer Group, London, United Kingdom Address reprint requests to Dean A. Fennell, MD, West Smithfield, London EC1A 7BE, United Kingdom; e-mail: d.a.fennell{at}qmul.ac.uk
PURPOSE: Malignant pleural mesothelioma (MPM) carries a poor prognosis due to chemoresistance. The European Organisation for Research and Treatment of Cancer (EORTC) prognostic model was reported to predict survival in MPM. Our retrospective analysis set out to test the validity of the model as a prognostic tool in patients treated in three phase II trials at St Bartholomew’s Hospital (London, United Kingdom) between 1999 and 2003. PATIENTS AND METHODS: A total of 145 patients were treated in three phase II trials; vinorelbine (VIN; 70 patients), vinorelbine/oxaliplatin (VO; 26 patients), and irinotecan/cisplatin/mitomycin C (IPM; 49 patients). Two subgroups, high-risk and low-risk, were defined by EORTC prognostic score (EPS). EPS was determined by a five-parameter model incorporating age, sex, histology, probability of diagnosis, and leukocyte count. An EPS cutoff of less than 1.27 (low risk) or more than 1.27 (high risk) was used to stratify Kaplan-Meier survival curves. Each of the EPS variables exhibited either trends or significant stratification of overall survival (OS). RESULTS: Multivariate analysis confirmed leukocyte count, Eastern Cooperative Oncology Group performance status, and sarcomatous histology as independent prognostic variables. EPS stratified OS in both individual and pooled trial datasets. No association between objective tumor response and EPS classification was identified by multinomial logistic regression. EPS stratified progression-free survival for the VO and IPM cohorts, but not for VIN. CONCLUSION: This study validates the EPS system as a robust tool for stratifying small trials into low- and high-risk subgroups. EPS should facilitate patient selection and analysis in randomized clinical trials.
Malignant pleural mesothelioma (MPM) is a rapidly fatal cancer that is increasing in incidence1 and is causally associated with asbestos exposure. Chemotherapy is the mainstay of treatment, and a standard of therapy does not yet exist. Poor prognosis is invariably associated with de novo resistance resulting in short-lived objective tumor responses in a minority of patients.2,3 Patients with MPM vary with regard to overall survival. Dividing patients into low- and high-risk subgroups may facilitate the design of clinical trials and the interpretation of MPM clinical trial data. Several studies4-13 have reported prognostic factors for MPM, based on small trials. Such studies have been criticized for statistical weakness based on the instability associated with Cox multivariate modeling. The European Organisation for Research and Treatment of Cancer (EORTC) reported the derivation of a prognostic score which stratifies patients with MPM into low- and poor-prognosis groups.14 Retrospective analysis of survival curves from 181 patients treated in five EORTC phase II trials was used to determine the EORTC prognostic score (EPS) for mesothelioma. Univariate analysis of 13 variables showed that poor prognosis was most strongly associated with poor performance status, high WBC, male sex, probable diagnosis of MPM, and sarcomatous histology. Cox multivariate regression retained these variables, and the EPS was calculated as the sum of maximum likelihood parameter estimates for the five prognostic variables retained in the multivariate model. A cutoff of more than 1.27 (poor prognosis) corresponds to zero, one, or two poor prognostic factors. The aim of this study was to statistically test the validity of the EPS as a predictive variable for prognosis in a retrospective series of 145 patients who were treated at St Bartholomew’s Hospital (London, United Kingdom) in three consecutive phase II clinical trials between 1999 and 2003. The impact of EPS on objective tumor response and progression-free survival was also determined in order to gain insight into the differential behavior of MPM between the respective high- and low-risk subgroups.
Calculating the EPS One hundred forty-five patients were enrolled onto three single-center clinical trials at St Bartholomew’s Hospital between 1999 and 2003. Based on their order of enrollment, patients received either vinorelbine (VIN),15 VIN and oxaliplatin (VO), or irinotecan/cisplatin/mitomycin C (IPM).16,17 Patient characteristics are summarized in Table 1. EPS was calculated for each patient as the conditional sum of five constants; each constant is retained in the EPS formula based on the presence of a poor prognostic factor. The formula for calculation of the EPS is:
One hundred thirty-four patients were assessable for EPS, which was calculated to allow subgrouping of patients into low-risk (EPS < 1.27) and high-risk (EPS > 1.27) groups. The cutoff point of 1.27 represents the largest sum that can be calculated from a minimum of two constants in the EPS formula.
Survival Analysis
Multinomial Logistic Regression
EPS Predicts Survival in the Individual Phase II Trials EPS was determined for patients treated in the VO, IPM, and VIN trials. Using EPS, patients were assigned into subgroups: low-risk subgroup, EPS less than 1.27; high-risk subgroup, EPS more than 1.27. Kaplan-Meier curves were stratified according to EPS and compared using the log-rank (LR) test. Results are summarized in Figures 1 and 2, and Tables 2 and 3.
OS for patients treated with VO was 10.4 months from diagnosis (95% CI, 8.0 to 12.8), and 8.8 months from first treatment (95% CI, 6.6 to 11.0). Despite the small sample sizes, a trend toward worse survival from diagnosis was observed in the high-risk VO subgroup (OS, 10.4 months; 95% CI, 7.7 to 13.9), compared with the low-risk subgroup (OS, 11.3 months; 95% CI, 8.3 to 14.3; LR = 1.0; P = .3; Fig 1A). The trend toward worse OS from first treatment in the VO trial was 8.4 months (95% CI, 3.3 to 13.4) versus 7.2 months (95% CI, 6.6 to 7.8; LR = 0.07; P = .8), in the low- versus high-risk subgroups, respectively. In the IPM trial, OS was 16.6 months (95% CI, 10.4 to 22.8) from diagnosis, and 10.1 months (95% CI, 6.6 to 13.6) from first treatment. The EPS yielded statistically significant separation of the survival curves for patients receiving IPM chemotherapy. OS from diagnosis for the low-risk IPM subgroup was 19.2 months (95% CI, 8.3 to 13.6), and 10.8 months (95% CI, 13.1 to 25.3) for the high-risk subgroup (Fig 1B). From first treatment with IPM, OS was 12.7 months (95% CI, 5.4 to 21.1) for the low-risk subgroup, compared with 8.9 months (95% CI, 5.3 to 12.5) for the high-risk subgroup; LR was 4.5, P < .01. For the VIN trial, OS was 13.1 months (95% CI, 10.4 to 13.1) from diagnosis, and 9.9 months (95% CI, 7.1 to 12.7) from first treatment. Kaplan and Meier survival curves were stratified by EPS; from diagnosis OS for low-risk patients was 19.1 months (95% CI, 14.6 to 23.8), compared with 9.9 months for high-risk patients (95% CI, 8.5 to 11.3; LR = 13.4; P < .01). OS from first treatment in the VIN trial, stratified by EPS, was 11.7 months (95% CI, 4.2 to 19.4) for the low-risk subgroup, compared with 7.3 months (95% CI, 5.0 to 9.5) for the high-risk subgroup (LR = 10.3; P < .01).
EPS Predicts Survival for the Pooled Trials To confirm a correlation between EPS cutoff and OS, the distribution of EPS scores was divided into thirds, using the 33rd percentile (EPS = 1.15), and 66th percentile (EPS = 1.75). This frequency distribution was normal, (Figs 3A and B). Kaplan and Meier curves plotted for each subgroup (EPS < 1.15, 1.15 < EPS < 1.75, and EPS > 1.75), showed a trend toward worse survival with increasing EPS (LR = 16.4; P = .0003; Fig 3C).
The individual effects of each EPS variable on OS was then determined. WBC showed a trend in OS from first treatment (LR = 2.7; P = .09). ECOG performance status1,2 was highly predictive for OS (LR = 21.6; P < .0001); OS for the low-risk subgroup, 12.7 months (95% CI, 8.6 to 16.8); versus 8.1 months for the high-risk subgroup (95% CI, 5.9 to 11.4). Sarcomatous histology also stratified survival (LR = 7.35; P = .007) with OS of 10 months (95% CI, 8.5 to 11.5) in the low-risk subgroup, versus 8.6 months in the high-risk subgroup (95% CI, 5.9 to 11.4). Sex did not stratify OS from first treatment, (LR = 0.02; P = .9), and the insufficient number of patient cases with probable histology precluded analysis. International Mesothelioma Interest Group (IMIG) stage, which is not an EPS variable, stratified OS (LR = 8.5; P = .04), whereas age above the median (age > 62 years), did not stratify OS (LR = 0.5; P = .49). In a Cox multivariate analysis using the EPS variables to model survival with a backward conditional stepwise method, WBC, performance status, and sarcomatous histology were retained as variables in the model, but sex was not. The impact of the sum of any EPS variable (between 0 and 4) on survival was examined. Median OS was reduced with the successive addition of any one variable, as shown in Table 4.
EPS Does Not Predict Objective Tumor Response in MPM MPM is characterized by significant resistance to chemotherapy. Differences in the chemoresponsiveness of tumors could account for the observed separation of low- and high-risk subgroups by the EPS. This hypothesis was therefore tested, by multinomial logistic regression, using partial remission, stable disease, and progressive disease as the categoric dependent variables, and EPS as covariate. EPS did not predict response to chemotherapy (  2 for model 2.56; P = .28). Sarcomatoid histology, which stratified OS, did not predict response when used as a covariate in multinomial regression analysis.
EPS Does Not Predict PFS Across All Trials
This study provides strong evidence to support the use of the EPS as a statistically valid method to predict survival in patients with MPM. The EPS is robust and can stratify small trials. EPS is simple to derive and interpret. This may be considered an advantage over the Cancer and Leukemia Group B (CALGB) prognostic scoring system,9 which though statistically sound in its derivation, yields six strata for risk, which demands a large sample size to ensure statistically significant subgroup separation. The presence of sarcomatous histology in the EPS suggests that differences in the pathological subtype can influence differential outcome in the EPS subgroups. The sarcomatous mesothelioma subtype is associated with lower response to chemotherapy compared with the epitheloid subtype.11 Multinomial regression, however, failed to demonstrate a difference in objective response rates between patients with an EPS of more than 1.27 versus an EPS of less than 1.27 for patients with sarcomatoid histology, suggesting that chemoresistance does not account for differences in survival between subgroups, per se. Pooled analysis of PFS in patients treated in the IPM and VO trials showed significant stratification into high- and low-risk subgroups by the EPS. This result suggests that durability of response affects OS more than objective tumor response, at least in these two clinical trials. The finding that the EPS does not stratify PFS for patients treated with vinorelbine may relate to the way in which treatment was administered. Vinorelbine was given weekly, until disease progression. The rate of progression was not determined by EPS. Analysis of overall survival stratified by tumor response in the VIN trial shows that patients experiencing partial response and stable disease during therapy have the same prognosis (Fennell et al, unpublished observations). This is not the case for patients treated with IPM or VO. This finding suggests a predominantly cytostatic activity of vinorelbine, in which tumor shrinkage does not confer a significant advantage over tumor stabilization. This is also supported by the fact that patients treated with more than 12 injections of vinorelbine had a trend to longer OS (Fennell et al, unpublished observations). The EPS does, however, effectively predict OS in vinorelbine-treated patients, and this implies that high-risk patients’ tumors progress at a faster rate from the time of progression onward. The rate of survival from diagnosis for low-risk patients treated with VIN and IPM was more than 19 months; however, it is uncertain what the survival of untreated patients in this subgroup would be. The British Thoracic Society MSO1 trial,18 which is currently enrolling patients, is a randomized phase III study comparing two chemotherapy regimens, vinorelbine alone15and the combination of mitomycin C, vinblastine, and cisplatin,19 and active supportive care.18 It will be of considerable interest to determine the survival of patients receiving active supportive care with respect to the EPS, to better understand the impact of chemotherapy in this subgroup. In summary, the EPS is a valuable tool for predicting outcome in patients with MPM. EPS is useful for interpreting the benefit of chemotherapy in clinical trials, and may provide a basis for designing trials in which patients may be stratified to receive tailored therapy, and improve overall prognosis.
The authors indicated no potential conflicts of interest.
We thank Richard Stephens, of the MRC Clinical Trials Unit, for reading the manuscript, and helpful comments. This study was conducted on behalf of the London Lung Cancer Group.
Authors’ disclosures of potential conflicts of interest are found at the end of this article.
1. Peto J, Decarli A, La Vecchia C, et al: The European mesothelioma epidemic. Br J Cancer 79:666-672, 1999[CrossRef][Medline] 2. Alberts AS, Falkson G, Goedhals L, et al: Malignant pleural mesothelioma: A disease unaffected by current therapeutic maneuvers. J Clin Oncol 6:527-535, 1988[Abstract] 3. Fennell DA, Rudd RM: Defective core-apoptosis signalling in diffuse malignant pleural mesothelioma: Opportunities for effective drug development. Lancet Oncol 5:354-362, 2004[CrossRef][Medline] 4. Antman K, Shemin R, Ryan L, et al: Malignant mesothelioma: Prognostic variables in a registry of 180 patients, the Dana-Farber Cancer Institute and Brigham and Women’s Hospital experience over two decades, 1965-1985. J Clin Oncol 6:147-153, 1988[Abstract]
5. Chailleux E, Dabouis G, Pioche D, et al: Prognostic factors in diffuse malignant pleural mesothelioma: A study of 167 patients. Chest 93:159-162, 1988 6. Ruffie P, Feld R, Minkin S, et al: Diffuse malignant mesothelioma of the pleura in Ontario and Quebec: A retrospective study of 332 patients. J Clin Oncol 7:1157-1168, 1989[Abstract] 7. Boutin C, Rey F, Gouvernet J, et al: Thoracoscopy in pleural malignant mesothelioma: A prospective study of 188 consecutive patients. Part 2: Prognosis and staging. Cancer 72:394-404, 1993[CrossRef][Medline] 8. Fusco V, Ardizzoni A, Merlo F, et al: Malignant pleural mesothelioma: Multivariate analysis of prognostic factors on 113 patients. Anticancer Res 13:683-689, 1993[Medline]
9. Herndon JE, Green MR, Chahinian AP, et al: Factors predictive of survival among 337 patients with mesothelioma treated between 1984 and 1994 by the Cancer and Leukemia Group B. Chest 113:723-731, 1998
10. Edwards JG, Abrams KR, Leverment JN, et al: Prognostic factors for malignant mesothelioma in 142 patients: Validation of CALGB and EORTC prognostic scoring systems. Thorax 55:731-735, 2000 11. Ceresoli GL, Locati LD, Ferreri AJ, et al: Therapeutic outcome according to histologic subtype in 121 patients with malignant pleural mesothelioma. Lung Cancer 34:279-287, 2001[CrossRef][Medline] 12. Metintas M, Metintas S, Ucgun I, et al: Prognostic factors in diffuse malignant pleural mesothelioma: Effects of pretreatment clinical and laboratory characteristics. Respir Med 95:829-835, 2001[CrossRef][Medline] 13. Steele JP: Prognostic factors in mesothelioma. Semin Oncol 29:36-40, 2002
14. Curran D, Sahmoud T, Therasse P, et al: Prognostic factors in patients with pleural mesothelioma: The European Organization for Research and Treatment of Cancer experience. J Clin Oncol 16:145-152, 1998
15. Steele JP, Shamash J, Evans MT, et al: Phase II study of vinorelbine in patients with malignant pleural mesothelioma. J Clin Oncol 18:3912-3917, 2000 16. Fennell D, Steele J, Shamash J, et al: Second line therapy of malignant pleural mesothelioma with irinotecan, cisplatin, and mitomycin C (IPM): A phase II study. Lung Cancer 41:S221, 2003 (suppl 2, abstr P-518) 17. Fennell D, Steele J, Shamash J, et al: A phase II study of irinotecan, cisplatin, and mitomycin C (IPM) in malignant pleural mesothelioma. Lung Cancer 41:S221, 2003 (suppl 2, abstr P-517)
18. Muers MF, Rudd RM, O’Brien ME, et al: BTS randomised feasibility study of active symptom control with or without chemotherapy in malignant pleural mesothelioma: ISRCTN 54469112. Thorax 59:144-148, 2004
19. Middleton GW, Smith IE, O’Brien ME, et al: Good symptom relief with palliative MVP (mitomycin-C, vinblastine and cisplatin) chemotherapy in malignant mesothelioma. Ann Oncol 9:269-273, 1998 Submitted July 21, 2004; accepted October 5, 2004.
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2005 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
