Japanese Journal of Clinical Oncology 30:534-541 (2000)
© 2000 Foundation for Promotion of Cancer Research
Prognostic Factors in Advanced Non-small Cell Lung Cancer: Elevated Serum Levels of Neuron Specific Enolase Indicate Poor Prognosis
Department of Internal Medicine II, Okayama University Medical School, Okayama, Japan
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONPATIENTS AND METHODSRESULTSDISCUSSIONAcknowledgmentREFERENCES |
|---|
Background: Non-small cell lung cancer (NSCLC) is resistant to chemotherapy and prognosis of advanced NSCLC patients is considered to be dependent on various prognostic factors.
Methods: We analyzed prognostic factors in patients with advanced NSCLC who had been enrolled in clinical trials conducted by the Okayama Lung Cancer Study Group between 1978 and 1992 using two kinds of multivariate analysis, Cox’s multivariate analysis and recursive partitioning and amalgamation (RPA) analysis.
Results: The first analysis was performed on 261 patients using 28 variables. Performance status (PS), clinical stage, liver metastasis or serum albumin level was an independent prognostic factor by Cox’s analysis. In the second analysis performed on 128 patients having data on neuron specific enolase (NSE), NSE was the most important prognostic factor. Using the RPA method, three subgroups with significantly different survival potentials were defined. Among them, patients with normal serum NSE levels and good PS were found to obtain a markedly favorable prognosis [median survival time (MST) 22.1 months, 3-year survival rate 42.9%], whereas the survival of patients with elevated serum NSE levels and bone metastasis was extremely short (MST 4.7 months, 3-year survival rate 0%).
Conclusions: These results indicate that analysis of prognostic factors including serum levels of NSE is useful for predicting the survival of patients with advanced NSCLC.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONPATIENTS AND METHODSRESULTSDISCUSSIONAcknowledgmentREFERENCES |
|---|
Non-small cell lung cancer (NSCLC) is generally resistant to chemotherapy or radiotherapy. Recent meta-analyses have verified a small improvement in the survival of patients with advanced NSCLC treated with cisplatin (CDDP)-based chemotherapy (1–3). However, the feasibility and efficacy of chemotherapy for advanced NSCLC patients have not been established, because of its severe toxicity and low cost effectiveness for a small survival advantage. The prognosis of advanced NSCLC patients is variable and may be dependent on different prognostic factors, including biological characteristics of cancer cells. Therefore, the analysis of these prognostic factors may define a subgroup of NSCLC patients with a similar survival potential. The clinical stage of NSCLC is determined based on the TNM classification (4), but it is not satisfactory for predicting an accurate prognosis of advanced NSCLC patients receiving chemotherapy. Accordingly, the development of a new staging system combined with available prognostic factors and the introduction of innovative treatment for the poor prognosis subgroup are desirable. The objectives of the present study were to analyze the relative contribution of pretreatment variables by univariate analysis, to determine which prognostic variables contribute independently to treatment outcome by a Cox’s regression analysis (5) and to define the patient subgroups with similar survival potentials by a recursive partitioning and amalgamation (RPA) method (6).
|
PATIENTS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONPATIENTS AND METHODSRESULTSDISCUSSIONAcknowledgmentREFERENCES |
|---|
Study Design and Eligibility Criteria
The present study was a retrospective analysis of clinical records from 261 patients with unresectable NSCLC who entered into four consecutive clinical trials of chemotherapy (two phase II and two phase III trials) conducted by the Okayama Lung Cancer Study Group between 1978 and 1992 (Table 1) (7–9). The common eligibility criteria for these trials was as follows: (1) histologically or cytologically confirmed inoperable NSCLC, (2) no prior chemotherapy, radiotherapy or surgery, (3) age 75 years or less, (4) ECOG performance status (PS) of 0, 1, 2 or 3 (10), (5) presence of measurable or evaluable disease, (6) adequate renal function (normal serum creatinine level), (7) adequate hepatic function [aspartate aminotransferase (AST), alanine aminotransferase (ALT) less than or equal to twice the upper normal limit], (8) normal bone marrow function [leukocyte counts (WBC)
4000/mm3, platelet counts (PL) 100 000/mm3], (9) no concomitant malignancies, (10) no symptomatic brain metastasis and (11) acquisition of verbal informed consent.
|
Chemotherapy
The COMP regimen consisted of cyclophosphamide (270 mg/m2 i.v. days 1–5), vincristine (1.4 mg/m2 i.v. day 1), methotrexate (6.5 mg/m2 i.m. days 1–5) and procarbazine (65 mg/m2 p.o. days 1–5). The cycles were repeated every 4 weeks until progression (7). In a randomized trial comparing MVB with PVB, mitomycin (10 mg/m2 i.v. day 1) or cisplatin (CDDP) (100 mg/m2 i.v. day 1) was given with vindesine (VDS) (3 mg/m2 i.v. days 1 and 8) and bleomycin (3.4 mg/m2 s.c. days 1–5) (8). The VP regimen consisted of VDS (3 mg/m2 i.v. days 1 and 8) and CDDP (20 mg/m2 i.v. days 1–5), and ifosfamide (1300 mg/m2 i.v. days 1–5) was added to VP in the VIP regimen (9).
Staging Criteria
The staging procedures included physical examination, a complete blood cell count, a standard blood chemistry profile, a chest radiograph, computed tomographic (CT) scans of the chest and abdomen, magnetic resonance imaging (MRI) or enhanced CT of the brain, radionuclide bone scan and fiber-optic bronchoscopy. Clinical stage was determined according to the international staging criteria (4). Response to chemotherapy was assessed according to the standard WHO criteria (11).
Pretreatment Prognostic Factors
In this analysis, pretreatment factors available for all patients were as follows: age (65 vs <65 years), gender, PS on ECOG scale (0 vs 1–3) (10), histology (squamous cell carcinoma vs others), body weight loss (10% vs <10%), smoking history (yes vs no), clinical stage (III vs IV), presence or absence of bone, lung, liver, brain or bone marrow metastasis and number of metastatic sites (0–2 vs 3). Laboratory values were also available for the majority of patients: WBC counts (9000 vs <9000/mm3), hemoglobin concentrations (12 vs <12 g/dl), platelet counts (15 x 104 vs <15 x 104/mm3), albumin levels (3.5 vs <3.5 g/dl), bilirubin levels (1.0 vs <1.0 mg/dl), AST levels (40 vs <40 u/l), alkaline phosphatase (ALP) levels (high vs normal), cholinesterase levels (low vs normal), lactate dehydrogenase (LDH) levels (high vs normal), cholesterol (Col) levels (125 vs <125 mg/dl), calcium levels (10 vs <10 mg/dl), urea nitrogen levels (20 vs <20 mg/dl), CRP levels (0.3 vs <0.3), erythrocyte sedimentation rates (15 vs <15 mm/h) and carcinoembryonic antigen (CEA) levels (5 vs <5 ng/ml). Serum concentrations of neuron specific enolase (NSE) (10 vs <10 ng/ml), which have been measured since 1986, were available only for 128 patients treated with the VP or VIP regimen.
Statistical Methods
All data concerning survival were updated on December 31, 1997. Differences of variables between treatment subgroups were evaluated by a 
2 test. Survival curves were generated with the Kaplan–Meier method (12) and compared by the log-rank test (13) and generalized Wilcoxon test (14). The Cox regression model was also used to determine the variables associated with better survival in a backward stepwise fashion (5). To complement the Cox model and to define prognostic subgroups with similar survival, a second multivariate analysis was performed using the RPA method (6). This method, consisting of two processes, is represented as a regression tree. Initially, the entire group of patients is partitioned into two subgroups according to the variable that produces the most significant survival difference. This variable is determined by Cox’s multivariate analysis. Each subgroup is again partitioned into two subgroups in the same manner. The partitioning process is stopped when no variable produces a further significant difference in survival between given subgroups; these subgroups are designated the terminal subgroup. An amalgamation process, the second component of RPA analysis, combines the terminal subgroups of patients whose survival does not significantly differ from each other by a log-rank test. This last step produces the final prognostic subgroups. The SPSS statistical software package (SPSS, Chicago, IL) was used to perform the above analyses.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONPATIENTS AND METHODSRESULTSDISCUSSIONAcknowledgmentREFERENCES |
|---|
Patients Characteristics and Treatment Outcome
Table 1 summarizes the results of four clinical trials conducted by the Okayama Lung Cancer Study Group. From 1975 to 1982, a phase II study of four-drug combination chemotherapy (COMP) was conducted (7). Between 1982 and 1986, patients were enrolled in a randomized study comparing two kinds of three-drug combination, MVB and PVB (8). Between 1986 and 1987, a pilot phase II study of three-drug combination (VIP) (9) was conducted, then patients were enrolled in a randomized study comparing VIP with a two-drug combination (VP) between 1987 and 1992 (9). During that period, both response rate and median survival time (MST) appeared to be gradually improved. Although CDDP-containing regimens failed to produce an improvement in survival compared with the other regimens (MST: 9.3 months for PVB/VIP/VP vs 7.4 months for COMP/MVB, p = 0.526), the patients treated with the recent protocols had a longer survival than those receiving the former regimens (MST: 9.1 months for VIP/VP vs 8.0 months for COMP/MVB/PVB, p = 0.015).
Table 2 lists the patients’ characteristics according to the chemotherapy regimens. Of 261 patients enrolled in these studies, 192 (74%) were male,188 (72%) had a PS of 0 or 1 and 81 (31%) were under 65 years old. Patients receiving the COMP regimen included more patients in early stage (IIIA, p = 0.0116) and with poor PS (2–3, p = 0.001) than those receiving the other regimens. There were no other significant differences in the patients’ characteristics according to chemotherapy regimens.
|
Univariate Analysis of Pretreatment Prognostic Variables
Univariate analysis of prognostic factors was performed for all of the patients (Tables 3 and 4). Statistically significant favorable factors in clinical features were age <65 years, female gender, histology of squamous cell carcinoma, PS 0, body weight loss <10%, stage III, absence of bone, liver or bone marrow metastasis and number of metastatic sites 0–2. Among laboratory values, WBC
9000/mm3, Hb 12 g/dl, CRP <0.3, ESR 15 mm/h, albumin 3.5 g/dl, normal ALP, normal CHE, normal LDH, total CHO 125 mg/dl, Ca 10 mg/dl and NSE 10 ng/ml were significantly favorable factors.
|
|
Multivariate Analyses of Prognostic Variables
The results of two types of Cox’s models are summarized in Table 5. The first model was applied to 221 patients having complete data for significant variables proven by univariate analysis except for NSE. The most significant factor was PS followed by stage, presence of liver metastasis and serum level of albumin. In the second analysis that was performed on 128 patients having NSE data, the serum level of NSE was the most significant prognostic factor.
|
RPA Analysis
Two types of analysis were performed. The results of the first analysis conducted for all patients are shown in Fig. 1. In this analysis, the first and most significant variable that split the entire population was PS (0 vs 1–3). There was no subsequent significant variable for the subgroup with PS 0, which was then designated terminal subgroup I. In the subgroup with PS 1–3, the second important variable was stage. In a subgroup with PS 1–3 and stage III disease, no additional significant variable was observed and this subgroup was designated terminal subgroup II. Using similar analyses, patients with PS 1–3 and stage IV disease were divided into two subgroups according to serum albumin levels. Among patients having serum albumin levels
3.5 g/dl, a subgroup of patients without liver metastasis was designated terminal subgroup III and that with liver metastasis terminal subgroup IV. Finally, a subgroup of patients with serum albumin levels of <3.5 g/dl was designated terminal subgroup V. The terminal subgroups I–V included 42, 81, 79, 10 and 27 patients, respectively. The insets in Fig. 1 show the results of a statistical comparison of survival among the five terminal subgroups or combinations of terminal subgroups, as part of the amalgamation algorithm for forming final subgroups with similar survival. Groups II and III and groups IV and V were each combined into one subgroup, because they had no significantly different survival. Finally, three subgroups based on significant differences in survival were determined: good-risk group (terminal subgroup I), intermediate-risk group (terminal subgroups II and III) and poor-risk group (terminal subgroups IV and V). The MST and 1-year survival rate were 13.5 months and 59.5% in the good-risk group, 9.3 months and 38.8% in the intermediate-risk group and 4.2 months and 5.4% in the poor-risk group, respectively. The differences in survival among these three groups of patients were highly significant (Fig. 2, p < 0.0001).
|
|
The results of the second analysis conducted on patients having NSE data are shown in Fig. 3. Using similar methods, patients with serum NSE levels
10.0 ng/ml were divided into a subgroup with PS 0 (terminal subgroup I) and a subgroup with PS 1–3 (terminal subgroup II). Patients with serum NSE levels >10 ng/ml were divided into terminal subgroups III and IV according to the bone metastasis. In the amalgamation process, terminal subgroup I was determined as the good-risk group, terminal subgroups II and III the intermediate-risk group and terminal subgroup IV the poor-risk group. The MST and 1-year survival rate were 22.1 months and 85.7% in the good-risk group, 10.0 months and 42.7% in the intermediate-risk group and 4.7 months and 0% in the poor-risk group, respectively. The differences in survival among these three groups of patients were also highly significant (Fig. 4, p < 0.0001).
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONPATIENTS AND METHODSRESULTSDISCUSSIONAcknowledgmentREFERENCES |
|---|
By RPA analysis of prognostic factors in NSCLC, Albain et al. in the Southwest Oncology Group (SWOG) divided 904 extensive NSCLC patients into three subgroups with similar survival potentials using PS, Hb, age and LDH. The MST and 1-year survival rate of these three groups were 7.6 months and 27%, 5.1 months and 16% and 3.0 months and 6%, respectively (15). Paesmans et al. in the European Lung Cancer Working Party reported four subgroups with similar survival potentials among 1052 advanced NSCLC patients using disease extent, PS, gender, WBC, neutrophil count and weight loss. The MST and 1-year survival rate of these four subgroups were 60 weeks and 53%, 38 weeks and 38%, 26 weeks and 21% and 14 weeks and 10%, respectively (16). In the present study, following the first analysis of 261 NSCLC patients, three subgroups were defined based on PS, clinical stage, liver metastasis and serum albumin levels. The MST and 1-year survival rates of these three subgroups were 13.5 months and 59.5%, 9.3 months and 38.8% and 4.2 months and 5.4%, respectively. The difference in survival among the subgroups seemed to be prominent in our analysis.
In the second analysis of 128 patients having data for serum NSE levels, we also defined three subgroups based on NSE, clinical stage, serum albumin levels and bone metastasis. The difference in survival among these three subgroups in the second analysis was also significant and we considered that this classification was useful in the clinical study, because it was very simple.
The International Association for the Study of Lung Cancer (IASLC) presented a Consensus Report for prognostic factors of NSCLC in 1991 (17). In this report, clinical stage and PS were definite prognostic factors and weight loss, gender, LDH and histology (squamous vs others) were possible prognostic factors. In addition to these factors, the updated consensus report by IASLC in 1994 included laboratory values (Hb, PLT and WBC) and biological factors (dominant oncogenes, suppressor oncogenes, cell adhesion molecules, neuroendocrine markers and so on). In the latter report, RT dose and response and CDDP-based chemotherapy and response were also considered as prognostic factors (18). The present study could not confirm the effectiveness of CDDP-based chemotherapy statistically. However, the recent intensive CDDP-based chemotherapy regimens in our trials were more effective to prolong survival in NSCLC patients than the previous combination regimens (8,9).
It is noteworthy that the present study confirmed the prognostic significance of NSE in NSCLC. The clinical significance of neuroendocrine (NE) differentiation in NSCLC cells has been unclear in the past. Previously, several studies suggested that the presence of NE markers may be correlated with response to chemotherapy and survival (19,20). However, in our previous study, patients having elevated serum NSE levels showed a high response rate to chemotherapy, but both response duration and overall survival time in these patients were extremely short, suggesting NE differentiation in NSCLC to be a poor prognostic indicator (21). Andoh et al. also showed that a high serum NSE level was associated with significantly poor prognosis by multivariate analysis (22). On the other hand, Graziano et al. in the Cancer and Leukemia Group B (CALGB) recently reported the results of the analysis of patients with stage IIIA (N2) disease receiving induction chemotherapy followed by surgery. They showed no correlation between response to chemotherapy and the presence of NE markers and concluded that the presence of NE markers did not identify a subgroup of stage IIIA NSCLC patients having superior prognosis (23). Thus, the significance of NE differentiation in NSCLC cells has not been established so far. In the present study, response rates between patients having elevated serum NSE levels (49.4%) and those with normal serum levels (43.6%) were not significantly different. However, in the second analysis including NSE as one of the prognostic factors, NSE was the most significant factor by multivariate analysis. Furthermore, RPA analysis using NSE could classify the three subgroups having markedly different survival potential.
In the clinical trial evaluating the effect of chemotherapy, stratification or classification of the patients according to survival potential is necessary. Based on the data from our present study, advanced NSCLC patients having elevated serum NSE levels are considered to have extremely poor prognosis, if they are treated with the conventional chemotherapy. In conclusion, the present study is helpful for confirming the prognostic significance of neuroendocrine differentiation in NSCLC and discriminating a subgroup of patients with very poor prognosis among advanced NSCLC patients.
|
Acknowledgment |
|---|
|
TOPABSTRACTINTRODUCTIONPATIENTS AND METHODSRESULTSDISCUSSIONAcknowledgmentREFERENCES |
|---|
This research was partially supported by the Ministry of Education, Science, Sports and Culture, Grant-in-Aid for Scientific Research (C), No. 12670424, 2000.
|
FOOTNOTES |
|---|
+ For reprints and all correspondence: Hiroshi Ueoka, Department of Medicine II, Okayama University Medical School, 2–5–1 Shikatacho Okayama 700-8558, Japan. E-mail: hueoka@hospital.okayama-u.ac.jp
Abbreviations: NSCLC, non-small cell lung cancer; RPA, recursive partitioning and amalgamation; PS, performance status; NSE, neuron specific enolase; MST, median survival time
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONPATIENTS AND METHODSRESULTSDISCUSSIONAcknowledgmentREFERENCES |
|---|
1 Marino P, Pampallona S, Preatoni A, Cantoni A, Invernizzi F. Chemotherapy vs supportive care in advanced non-small-cell lung cancer. Results of a meta-analysis of the literature. Chest 1994;106:861–5.
2 Souquet PJ, Chauvin F, Boissel JP, Bernard JP. Meta-analysis of randomized trials of systemic chemotherapy versus supportive treatment in non-resectable non-small cell lung cancer. Lung Cancer 1995;12(Suppl 1):S147–54.
3 Non-small Cell Lung Cancer Collaborative Group. Chemotherapy in non-small cell lung cancer: a meta-analysis using updated data on individual patients from 52 randomized clinical trials. Br Med J 1995;311:899–909.
4 Mountain C. Revisions in the international system for staging lung cancer. Chest 1997;111:1710–7.
5 Cox DR. Regression models and life tables. J R Statist Soc (B) 1972;34:187–220.
6 Ciampi A, Lawless JF, Mckinney SM, Singhal K. Regression and recursive partition strategies in the analysis of medical survival data. J Clin Epidemiol 1988;41:737–48.[Web of Science][Medline]
7 Machida K. Combination chemotherapy of advanced lung cancer with cyclophosphamide, vincristine, methotrexate and procarbazine. Okayama Igakkai Zasshi 1984;96:961–72 (in Japanese).
8 Ozawa S. A comparative study of cis-platinum and mitomycin C in patients with advanced non-small cell lung cancer. Okayama Igakkai Zasshi 1986;98:771–81 (in Japanese).
9 Ohnoshi T, Hiraki S, Ueda N, Fujii M, Machida K, Ueoka H, et al. Phase II study of ifosfamide, cisplatin and vindesine combination in advanced non-small cell lung cancer. Acta Med Okayama 1991;45:357–61.
10 Oken MM, Creech RH, Tormey DC, Horton J, Davis TE, McFadden ET, et al. Toxicity and response criteria of the Eastern Cooperative Oncology Group. Am J Clin Oncol 1982;5:649–55.[Web of Science][Medline]
11 Miller KB, Hoogstraten B, Staquet M, Winkler A. Reporting results of cancer treatment. Cancer 1981;47:207–14.[Web of Science][Medline]
12 Kaplan EL, Meier P. Nonparametric estimation for incomplete observations. J Am Statist Assoc 1958;53:457–81.[Web of Science]
13 Mantel N. Evaluation of survival data and two new rank order statistics arising in its consideration. Cancer Chemother Rep 1966;50:163–70.[Medline]
14 Gehan EA. A generalized Wilcoxon test for comparing arbitrarily singly-censored samples. Biometrica 1965;52:203–24.
15 Albain KS, Crowley JJ, LeBlanc M, Livingston RB. Survival determinants in extensive-stage non-small-cell lung cancer: the Southwest Oncology Group experience. J Clin Oncol 1991;19:1618–26.
16 Paesmans M, Sculier JP, Libert P, Bureau G, Dabouis G, Thiriaux J, et al. Prognostic factors for survival in advanced non-small-cell lung cancer: univariate and multivariate analyses including recursive partitioning and amalgamation algorithms in 1052 patients. J Clin Oncol 1995;13:1221–30.[Abstract]
17 Feld R, Arriagada R, Ball DL, Mattson K, Sorensen JB. Prognostic factors in non-small cell lung cancer: a consensus report. Lung Cancer 1991;7: 3–5.
18 Feld R, Borges M, Giner V, Ginsberg R, Harper P, Klastersky J, et al. Prognostic factors in non-small cell lung cancer. Lung Cancer 1994;11(Suppl 3):S19–23.[Web of Science][Medline]
19 Berendsen HH, de Leij L, Poppema S, Postmus PE, Boes A, Sluiter HJ, et al. Clinical characterization of non-small-cell lung cancer tumor showing neuroendocrine differentiation features. J Clin Oncol 1989;7:1614–20.[Abstract]
20 Graziano SL, Mazid R, Newman N, Tatum A, Oler A, Mortimer JA, et al. The use of neuroendocrine immunoperoxidase markers to predict chemotherapy response in patients with non-small-cell lung cancer. J Clin Oncol 1989;7:1398–406.[Abstract]
21 Shibayama T, Ohnoshi T, Ueoka H, Horiguchi T, Kodani T, Segawa Y, et al. Serum neuron specific enolase (NSE) levels in patients with non small cell lung cancer. Nippon Kyobu Shikkan Gakkai Zasshi 1992;30:1097–102 (in Japanese).
22 Andoh M, Gemma A, Takenaka K, Hisakatsu S, Yamada K, Usuki J, et al. Serum neuron specific enolase level as a prognostic factor in non-small cell lung cancer. Intern Med 1994;33:271–6.[Web of Science][Medline]
23 Graziano SL, Kern JA, Herndon JE, Tatum A, Brisson ML, Memoli V, et al. Analysis of neuroendocrine markers, HER2 and CEA before and after chemotherapy in patients with stage IIIA non-small cell lung cancer: a Cancer and Leukemia Group B study. Lung Cancer 1998;21:203–11.[Web of Science][Medline]
Received June 29, 2000; accepted October 2, 2000.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||