| Japanese Journal of Clinical Oncology | Pages |
Introduction
Subjects And Methods
Patient Eligibility
Clinical Assessment
Treatment Protocol
Statistical Analysis
Results
Univariate Analysis
RPA Analysis
Discussion
Acknowledgements
References
Prognostic Factors and Prognostic Staging System for Small Cell Lung Cancer
Treatment results for combined chemotherapy in small cell lung cancer (SCLC) have reached a plateau during the last decade. Although most patients initially respond to therapy, median survival has not yet reached 1 year, and the 3-year survival rate is still approximately 15-20% among the large patient population (1 ). The main reason for the failure to achieve long-term survival rates is the development of drug resistance in the majority of patients. Many attempts have been made to overcome the development of drug resistance, including the use of alternating chemotherapy. We conducted a randomized study by the Lung Cancer Chemotherapy Study Group (LCCSG) in Japan Clinical Oncology Group (JCOG 8502) comparing two standard chemotherapy regimens with a regimen consisting of the same two regimens given on an alternating basis in patients with SCLC between April 1985 and May 1988 (2 ). Our results indicate that the alternating chemotherapy regimen attained more favorable results than the standard chemotherapy regimens: cyclophosphamide, adriamycin and vincristine (CAV), and cisplatin and etoposide (PE), although the differences were not dramatic.The present analysis was therefore performed to assess the prognostic factors other than treatment, and the predictors of long-term (3-year) survival, through multivariate analysis, and to define patient subgroups showing significantly different survivals using recursive partitioning and amalgamation (RPA) analysis.
Between April 1985 and May 1988, a total of 300 patients with SCLC were accrued by five hospitals across Japan which were members of LCCSG. Eligibility criteria for entry into this study were as follows: 1, histologically or cytologically proven SCLC; 2, no prior therapy; 3, measurable or evaluable disease; 4, a performance status of 0-3 on the Eastern Cooperative Oncology Group (ECOG) scale; 5, age <75 years; 6, adequate bone marrow reserve with a leukocyte count of >4000/mm3 and a platelet count of >100 000/mm3; 7; adequate liver function with bilirubin <1.5 mg/dl and levels of alkaline phosphatase (AlkP), asparate aminotransferase (AST), and alanine aminotransferase (ALT) no greater than twice the normal upper limit; 8, adequate renal function with serum creatinine <1.5 mg/dl and blood urea nitrogen <25 mg/dl; and 9, written informed consent from the patients.
Of these 300 patients, 14 (4.6%) were excluded for the following reasons. There were three cancellations due to patient refusal before the start of chemotherapy. Eleven patients were ineligible: four had histologies other than SCLC, three were >= 75 years of age, one had congestive heart failure and one had a simultaneous squamous cell carcinoma of the lung. Two treatment violations occurred in the CAV-PE arm. There were, therefore, 286 patients (95%) eligible for the study. Of these 286, 97 were allocated to the CAV arm, 97 to the PE arm and 92 to the CAV/PE arm.
All patients had a complete history, physical examination, evaluation of performance status and investigations to define the extent of their disease, which included routine hematology and biochemistry, chest X-ray, bronchoscopy, computed tomography or ultrasound examination of the abdomen, radionuclide bone scan and bone marrow aspiration cytology and/or biopsy. Routine biochemistry included AST, ALT, AlkP, LDH and measurement of serum electrolytes (sodium, potassium, chloride), serum creatinine and blood urea nitrogen.
Limited disease (LD) was defined as that confined to one thorax, including the bilateral mediastinal and supraclavicular nodes; any involvement beyond these confines was defined as extensive disease (ED). Patients with pleural effusion were included in the ED category.
After staging and stratification by the treatment center, patients were assigned randomly to receive CAV, PE, or alternating CAV with PE (CAV-PE). The CAV regimen consisted of cyclophosphamide at a dose of 800 mg/m2 given i.v. on day 1, adriamycin at 50 mg/m2 i.v. on day 1 and vincristine at 1.4 mg/m2 (maximum 2.0 mg) i.v. on day 1. The PE regimen consisted of cisplatin at 80 mg/m2 i.v. on day 1 and etoposide at 100 mg/m2 i.v. on days 1, 3 and 5. The treatments were repeated every 3-4 weeks. Cisplatin was given with adequate prehydration, posthydration, diuretics and antiemetic agents.
If patients on the CAV or PE regimen did not respond after two cycles of the initial regimen, they were crossed over to the other regimen. Restaging was carried out at the completion of four cycles of chemotherapy to evaluate the response. Thoracic irradiation after restaging was planned for all patients with limited disease. Thoracic irradiation consisted of one dose fraction of 2 Gy daily, 5 days per week for 4-5 weeks (total: 40-50 Gy). Chemotherapy for responding patients in each regimen was continued for 1 year from the start of the initial treatment. Subsequent treatment at the time of disease progression was at the discretion of the investigator. Prophylactic cranial irradiation was not scheduled in this study. Actual data concerning response, toxicity and dose intensity in each arm have already been reported (2 ).
Thirteen pretreatment potential prognosis variables were chosen for analysis. Each variable was used to divide the patients into two categories in order to determine its prognostic significance. All these variables are listed with their categories in Table 1 .
The cut-offs used for hematological and biochemical variables were the normal limits of the variable. Survival curves were calculated from the start of chemotherapy by the Kaplan and Meier method (3) and tests of significance were based on the log-rank test (4). Multivariate analyses including all variables listed in Table 1 were performed with the Cox proportional hazards model (5) to adjust for the prognostic variables in comparing the overall survival. We applied a Bonferroni correction of the standard significance value of 5% (0.05 divided by the number of variables used) (6). When 13 pretreatment variables were used for the analysis, a P-value <0.0039 (0.05/13) was considered significant. When treatment regimen was added to these, a P-value <0.0036 (0.05/14) was significant. A P-value <0.05 but more than the level described above was defined as marginally significant. Long-term (3-year) survivorship was investigated by logistic regression analysis. SAS computer program was used. As the classification approach, a second multivariate analysis was performed using RPA (7). The purpose of this was to obtain a classification rule that uses pretreatment variables to class patients with similar prognosis. Briefly, the entire patient population was partitioned into two subclasses according to the variable producing the most significant survival difference. Each of these subclasses was again partitioned into subclasses in the same manner. The partitioning process was stopped when no available variable produced a significant survival difference in a given subclass or when the size of a given subclass was too small. The next step was an amalgamation process, joining subclasses of patients who did not differ significantly from each other with regard to survival. This latter step of amalgamation produced the final prognostic classes. One month was calculated as 365/12 days.
Table 1
The present analysis was performed on a population of 286 eligible patients. The effect on survival of each of the 13 pretreatment variables and treatment regimens was investigated by univariate analysis (Table 1). The data for hemoglobin, WBC, serum AlkP and serum sodium were not available in some patients. PS, disease extent, number of metastatic sites, serum albumin, serum LDH, serum ALT, serum alkaline phosphatase (AlkP), serum Na, CEA and chemotherapy regimen (CAV vs CAV-PE) were the variables significantly related to survival.Multivariate Analysis
Multivariate analysis of 13 pretreatment variables and treatment regimen using Cox's proportional hazard model showed that poor PS (2-3) (P = 0.0001) and abnormally elevated serum LDH (P = 0.0001) and AlkP (P = 0.0002) were independently significant adverse prognostic factors. Variables such as extensive disease, abnormally high CEA and number of metastatic sites ( >= 2) were marginally significant (Table 2a); only varibles with P-values <0.05 are shown in Tables 2, 3 and 4. In the Tables, a relative risk value of >1.0 denotes an increased risk (e.g. 1.96 = 96% increase in risk of succumbing to SCLC). When pretreatment variables alone were analysed, the above three variables (PS, LDH, AlkP) and the extent of disease were significant prognostic factors. CEA, number of metastastatic sites and gender (female having better survival) were marginally significant (Table 2 ). In stepwise analysis, PS,LDH and extent of disease were the significant independent prognostic factors.
This study included both LD and ED patients. When survival for 143 patients with LD was investigated separately with respect to each variable, serum LDH (P = 0.0001) and PS (P = 0.0013) were shown to be of independent prognostic significance for the attainment of survival, and AlkP (P = 0.0148) was marginal (Table 3 a). In stepwise analysis, treatment regimen as well AlkP was also a marginally significant factor, favoring alternating regimen. Even when pretreatment variables alone were analysed, these significant prognostic factors were not changed (Table 3 b).
Table 2. Cox's Regression Analysis of all 278 patients
(b)
For survival of the 135 patients with ED, PS (P = 0.0050) and LDH (P = 0.0063) were independent prognostic factors. AlkP and sex were marginal. Treatment regimen was not a prognostic factor (Table 3 c). When pretreatment variables alone were examined, LDH and PS also were significant factors, and AlkP, CEA, sex, serum sodium and serum albumin were marginally significant factors (Table 3 d). In stepwise analysis, the significant variables were not changed.
Thirty-three patients (11.5%) survived for at least 3 years from the start of chemotherapy, of whom 28 had LD and 5 had ED. Three-year disease-free survival rates among LD and ED patients were 19.2% (28/146) and 3.5% (5/142) respectively. Multivariate analysis of long-term (3-year) survivors among LD patients showed that PS (P = 0.029) and WBC (P = 0.044) were independent prognostic factors. The patients with WBC >= 10 * 103/[mu]l had a better prognosis. In stepwise analysis, independent factors were LDH (P = 0.029), PS (P = 0.041) and treatment regimen (P = 0.043) favoring alternating regimen over others. There were too few long-term (3-year) survivors among patients with ED to make a similar multivariate analysis meaningful.
Table 3 Cox's Regression Analysis according to disease extent
(b) (c)
(d)
Table 4. Multiple variate analysis by logistic model
For RPA analysis, all prognostic variables except for treatment regimens examined in the Cox's proportional hazard model were used. Fig. 1 shows the tree that resulted from this analysis of 284 patients with SCLC. Serum LDH was the variable that split the initial population into the 147 patients with normal LDH and 137 patients with abnormally elevated LDH. Subsequent partitioning of internal nodes in the former group was determined by the covariates of PS and serum sodium. In the latter group, PS alone defined the splits. Five terminal nodes were made in this recursive partition tree. Following the amalgamation process, three prognostic classes were finally formed. The median survival rates of the patients were 16.0 mo, 9.4 mo and 6.6 mo for classes I-III respectively (P = 0.0001 by generalized Wilcoxon test, P < 0.0001 by log-rank test). The most favorable class (I) was defined by knowledge of LDH, PS and serum sodium levels; class (III) with the poorest survival was defined by LDH and PS. The intermediate prognostic class (II) was also defined by LDH, PS and serum sodium levels. Of these factors, there were significant correlations between LDH and PS (P = 0.0001) and between serum sodium levels and PS (P = 0.0025), but not between serum sodium levels and LDH (P = 0.2074). When this analysis was performed on LD patients alone, only two classes were defined by LDH and PS: the median survival of one favorable class (normal serum LDH and PS 0-1) of 77 patients was 18.1 mo and that of the other poor class (abnormal LDH or normal LDH and PS 2-3) of 70 patients was 9.9 mo. The difference in survival had P < 0.0001. In ED alone, three classes were defined by LDH, number of metastases and PS: The median survival of the most favorable class (normal LDH, number of metastases 0-1) of 42 patients was 11.8 mo, that of the poorest class (abnormal LDH, PS 2-3) of 50 patients was 6.6 mo, and that of the intermediate class (other combinations of LDH, number of metastases and PS) of 45 patients was 8.3 mo. The difference had also P < 0.0001. Survival and response in all patients plotted by this analysis are shown in Fig. 2 and Table 5 . The 2-, 3- and 5-year survival rates for patients were 31%, 22% and 19% in the favorable class (I), 6%, 5% and 2% in the intermidiate class (II) and 2%, 0%, and 0% in the poorest class (III) respectively.
Analysis of prognostic factors in our randomized trial using Cox's proportional hazard model showed that PS, serum LDH, AlkP and the disease extent were the most important variables affecting survival. This is consistent with the findings of others (8 -12 ). The finding that the serum LDH level is correlated with disease extent has been reported previously (8 -11 ,13 ). These parameters may well reflect the extent of disease beyond the limits of usual staging procedures. Ganz et al. (14 ) demonstrated a significant correlation between disease activity and serum LDH in the serial monitoring of response to therapy in patients with SCLC. Thus elevated levels of serum LDH may reflect not only tumor mass but also tumor aggressiveness.
It has been recognized that a number of other factors have prognostic implications in SCLC. Increasing age (15 ,16 ), male gender (15 ,17 ), the number of metastatic sites (18 ) and abnormal levels of AlkP (9 ,12 ,19 ) CEA (20 ), serum sodium (8 ,9 ) hemoglobin concentration (8 ) and serum albumin (9 ,10 ,21 ) can all influence prognosis adversely. In our study, CEA, number of mestastatic sites and gender were additional prognostic factors, but these were marginal when we apply a Bonferroni correction of the standard significance value of 5 % (0.05 divided by the number of variables) (6 ); a P-value of <0.0036 when pretreatment variables and treatment regimen were analysed and a P-value of <0.0039 when pretreatment variables alone were done, would be considered significant. Only LDH and PS satisfy this criterion. Variables CEA, sex, serum sodium and serum albumin were marginally significant when the analysis was limited to patients with extensive disease (Table 3 d).
Table 5.
Table 6.
In this study, pretreatment variables plus treatment factor were examined separately from pretreatment factor alone. The prognostic significance of the treatment arm has already been reported by Fukuoka (2 ). After adjusting for prognostic factors, the difference between the PE and the CAV-PE arms was significant (P = 0.032), favoring the CAV-PE arm. Analysis of recent follow-up data showed that this held true only when LD patients alone were examined by Cox's proportional hazard model in the stepwise method (P = 0.0182; Table 3 a). However, we must be cautious about this result because this is a subset analysis with too few patients to have a reasonable beta or type II error. Recently, at least five randomized studies (2 ,22 -25 ) (including one by ourselves) testing alternating chemotherapy employing CAV and PE in SCLC have been published. The Canadian multicenter randomized trial (24 ) presented evidence of a significant benefit of alternating chemotherapy compared with CAV in patients with ED (median survivals 8.0 months on CAV and 9.6 months on CAV-PE, P = 0.03). In our study (2 ), survival was significantly higher for the alternating arm than for the PE arm in patients with LD (median survival 11.7 months on PE and 16.8 months on CAV-PE), but not in those with ED.
In our study, disease extent was the primary pretreatment predictor of long-term (3-year ) survivals; 19.2% of patients with LD survived 3 years, compared with 3.5% of patients with ED. Of LD patients, PS and WBC had a significant influence on the probability of 3-year disease-free survival. The greater incidence of long-term survivors with WBC >= 10 * 103/[mu]l remained unexplained. The P-value of 0.044 was marginally significant and may have been fortuitous. Østerlind et al. (26 ) and Spiegelman et al. (27 ) reported that, in multivariate regression analysis, LD and female gender were the major prognostic factors of long-term survival. A reduced death hazard in females compared to males was also observed in Ettinger's series (28 ). However, gender was not an important factor in our study nor in a large British series (19 ). Østerlind et al. reported that PS and serum LDH did not have a significant influence on long-term survival. Other series reported that serum urate concentration (26 ) and histologic subtype (29 ) were important factors for long-term survival. These discrepancies may reflect the heterogeneity of patients accrued in each clinical trial. In the stepwise method, the alternating regimen influenced 3-year survival in our study with a P-value of 0.0425. This must be interpreted carefully because it evolved as a prognostic factor only through this stepwise method. Furthermore, a P-value of 0.043 is not highly significant.
We applied RPA, a new statistical tool, to define subsets for the purpose of prognosis, staging and stratification. Treatment regimen was not included in this analysis because, in our multivariate analysis, treatment did not affect survival significantly when all patients were studied. While the aim of Cox's regression analysis is to identify pretreatment variables that have an impact on survival, that of RPA is to group patients with a similar prognosis. These two approaches clearly differ, and one may not necessarily be superior to the other. However, RPA analysis has some advantages over Cox's proportional hazard model (15 ). First, it is not necessary to delete all cases with missing data on any prognostic factor. This is,of course, due to data handling software rather than methodology itself. Second, grouping of patients is a natural outcome of the analysis, compared with the arbitrary approach in Cox's model. Third, exploration of interactions among the prognostic factors is automatic. Fourth, intermediate prognostic classes are obtained without using mathematical formulae. Using this method, some groups (15 ,30 ,31 ) have proposed a new staging system. Of these, serum LDH level was the only variable that significantly affected prognosis in all studies. Stage, PS and serum sodium were the second most significant variables. Abnormal serum level of sodium is known to be associated with ectopic secretion of antidiuretic hormone (32 ). The syndrome of inappropriate antidiuretic hormone secretion (SIADH) is most frequently found in SCLC patients with brain metastases (33 ). Hyponatremia may represent extensive diseases especially metastasis to the brain. Our results were compatible with those of other studies. Two reports (31 ,34 ) are in abstract form. Another two (15 ,30 ) along with our results are shown in Table 6 . In RPA, our results showed only three significant factors
(PS, serum LDH and serum sodium) and the number of subclasses was three. This may have been due to the lower number of patients studied than in the other trials, which made 4 subclasses using 614 patients in the Canadian study (30 ) and 2580 patients in the SWOG study (15 ). Albain et al. proposed 4 subclassifications of disease extent, serum LDH, age and pleural effusion. They also pointed out that serum LDH emerged as a highly significant factor. Sagman et al. also subclassified 4 groups according to PS, disease extent, the limited disease stage (presence or absence of mediastinal disease), serum LDH, white blood cell count, sex, AlkP and liver metastasis. In our patients, three classes of similar prognosis were identified: the most favorable class was defined by knowledge of LDH, PS and serum sodium levels, and the class with the poorest prognosis was defined by knowledge of LDH and PS. This classification approach can be applied to the management of individual patients as well as clinical trials. As pointed out by Østerlind et al. (31 ), the time-consuming extensive staging of a subset of patients having the poorest prognosis may not be necessary. Fifteen percent of patients in Østerlind's study and 22% in our study corresponded to this group. In our poorest survival group, the proportion of patients achieving a complete response was only 3.3 % (2/61). Therefore, the requirement for the assessment of prognosis and response and treatment planning could be simplified, patient discomfort minimized and health care costs might be reduced. The classification differs slightly from study to study, perhaps due partly to the number and geneity of the study patients involved. However, it should be noted that even when the number of patients examined was large, the maximum number of subclasses made was four. This suggests that different prognostic factors such as molecular biomarkers may need to be incorporated into this analysis. The usefulness of this new staging system needs to be determined prospectively.
In conclusion, disease extent, PS and serum LDH are the major prognostic factors in patients with SCLC, including long-term survival. Since the most significant factors are also useful for RPA, serum LDH, sodium and PS contribute to the new subclassification. This subclassification using RPA will be used for the design, implementation and interpretation of clinical studies, as well as decision-making in individual patients.
This study was supported by Grants-in-Aid for Cancer Research (59 S-1 and 62 S-1) and a Comprehensive 10-Year Strategy for Cancer Control from the Ministry of Health and Welfare. We thank the following doctors who contributed to the study: Masunari Yamamoto, Nagahisa Kodama, Kaoru Kubota, and Mitsumasa Ogawara, of the National Kinki Central Hospital for Chest Diseases, Osaka; Shunichi Negoro, Minoru Takada, Noriyuki Masuda, Kaoro Matsui, Youko Kusunoki, Nobuhide Takifuji, Shinei Ryu, Shinzo Kudo, and Masayuki Nishioka, of Osaka Prefectural Habikino Hospital; Yuichiro Ohe, Kenji Eguchi, Tetsu Shinkai, Yasutsuna Sasaki, Sunao Egawa, and Ryosuke Ono, of the National Cancer Center Hospital, Tokyo; Tetsuro Kodama, Akira Hayashibe, Mitsuteru Hino, Naotake Nukariya, Yuji Inoue, Hiroko Sasaki, Takahiro Hamada, and Mizuyoshi Sakura, of the National Matsudo Hospital, Chiba; Shinichiro Nakamura, Takahiro Sakura, Hiroshi Nishio, Takeshi Horai, and Toshihiro Inoue, of the Center for Adult Disease, Osaka. We also gratefully acknowledge the excellent statistical assistance of Mr. Yasuo Uchida and Toshiyuki Ijima, and manuscript preperation of Miss.Yuki Hirochi and Erina Hatashita.
Variable
Category
No. of
patientsMST
(mo)[chi]2
P-value
(logrank)
PS
0+1
194
12.8
47.57
0.0001
2+3
92
7.8
Disease extent
LD
147
14.4
36.6
0.0001
ED
139
8.8
Age
<66
173
11.6
0.3
0.5864
>= 66
113
10.1
Sex
Male
233
10.7
0.03
0.8621
Female
53
11.1
No. of metastatic sites
0+1
242
11.6
37.15
0.0001
>= 2
44
7.4
Hemoglobin
>= 12
231
11.1
3.14
0.0762
(g/dl)
<12
54
9.9
WBC count
<10
33
9.3
1.66
0.1979
(*103/[mu]l)
>= 10
252
11.2
Serum Albumin
>= 3.5
85
11.3
5.02
0.0251
(g/dl)
<3.5
201
9.1
Serum LDH
Normal
147
14.3
44.22
0.0001
Elevated
137
8.0
Serum ALT
Normal
228
11.4
3.86
0.0496
Elevated
57
9.3
Serum AlkP
Normal
231
11.6
14.2
0.0002
Elevated
54
8.5
Serum Na
>= 136
245
11.4
7.22
0.0072
(mmol/l)
<136
40
9.1
Serum CEA
Normal
164
11.6
3.84
0.05
Elevated
115
9.9
Regimen
CAV
97
10
PE
97
10.1
CAV/PE
92
12
Comparison of regimens
CAV vs PE
0.12
0.7222
CAV vs CAV-PE
5.20
0.0226
PE vs CAV-PE
3.69
0.0581
RESULTS
Univariate Analysis
Variable
RR
95%CI
P-value
LDH
1.96
1.48-2.59
0.0001
PS
2.02
1.47-2.78
0.0001
AlkP
1.70
1.21-2.38
0.0002
Extent
1.54
1.14-2.09
0.0046
CEA
1.34
1.04-1.73
0.0260
No. meta.
1.57
1.04-2.38
0.0332
Stepwise method
Variable
RR
95%CI
P-value
PS
2.03
1.53-2.70
0.0001
LDH
1.91
1.46-2.49
0.0001
Extent
1.71
1.30-2.25
0.0001
AlkP
1.59
1.16-2.18
0.0040
Sex
0.68
0.48-0.96
0.0297
CEA
1.31
1.02-1.69
0.0354
Variable
RR
95%CI
P-value
LDH
1.93
1.47-2.55
0.0001
PS
2.01
1.46-2.76
0.0001
AlkP
1.73
1.24-2.41
0.0013
Extent
1.62
1.20-2.17
0.0015
CEA
1.33
1.03-1.72
0.0275
No. meta.
1.54
1.03-2.32
0.0374
Sex
0.69
0.49-0.98
0.0402
(a)
Variable
RR
95%CI
P-value
LDH
2.5
1.68-3.71
0.0001
PS
2.23
1.37-3.63
0.0013
AlkP
2.01
1.15-3.53
0.0148
stepwise method
LDH
2.22
1.53-3.21
0.0001
PS
1.99
1.29-3.09
0.0020
CAV-PE
0.63
0.42-0.92
0.0182
AlkP
1.74
1.04-2.89
0.0340
Variable
RR
95%CI
P-value
LDH
2.18
1.50-3.17
0.0001
PS
2.43
1.50-3.95
0.0003
AlkP
2.03
1.19-3.49
0.0098
stepwise method
LDH
2.11
1.46-3.05
0.0001
PS
2.14
1.39-3.31
0.0006
AlkP
1.86
1.12-3.08
0.0173
Variable
RR
95%CI
P-value
PS
1.79
1.19-2.68
0.005
LDH
1.78
1.18-2.70
0.0063
AlkP
1.65
1.06-2.56
0.0268
Sex
0.62
0.40-0.97
0.0374
stepwise method
PS
1.88
1.29-2.74
0.0011
LDH
1.91
1.29-2.83
0.0013
Sex
0.54
0.35-0.83
0.0054
Variable
RR
95%CI
P-value
LDH
1.92
1.26-2.91
0.0023
PS
1.94
1.25-3.01
0.0031
AlkP
1.74
1.11-2.71
0.015
CEA
1.51
1.04-2.19
0.0309
Sex
0.62
0.40-0.98
0.0387
NA
1.8
1.03-3.16
0.0404
ALB
1.58
1.00-2.50
0.0499
Variable
RR
95%CI
P-value
PS
0.07
0.01-0.77
0.029
WBC
4.86
1.04-22.66
0.044
stepwise method
LDH
0.3
0.10-0.89
0.029
PS
0.12
0.02-0.91
0.041
CAV-PE
2.5
1.03-6.07
0.043
Class
n
MST (mo)
Survival rate (%)
Proportion of complete response
2yr
3yr
5yr
LD+ED
LD
ED
I
107
16.0
31
22
19
24.3% (26/107)
27.4% (20/73)
17.6% (6/34)
II
116
9.4
6
5
2
11.2% (13/116)
11.1% ( 7/63)
11.3% (6/53)
III
61
6.6
2
0
0
3.3% (2/61)
0.0% (0/11)
4.0% (2/50)
P = 0.0004
P = 0.0132
P = 0.1211
Authors: Sagman et al (13)
Class
n
LD/ED
LDH
LDS
WBC
Sex
AlkP
PS
LM
MST
Survival rate (%)
(*103/[mu]l)
(mo)
2 yr
5 yr
I
107
LD
1
13.6
24.8
17.6
LD
N
2
<10
F
II
188
LD
N
2
<10
M
11.3
13.9
4.7
LD
N
2
>= 10
ED
F
0-1
no
III
276
LD
E
2
8.1
3.0
0.8
ED
F
0-1
yes
ED
M
0-1
ED
N
2-3
IV
43
ED
E
2-3
5.5
0.0
0.0
Authors: Albain et al (15)
Class
n
LD/ED
LDH
Age
Pleural
MST
Survival rate (%)
(yr)
Effusion
(mo)
2 yr
I
324
LD
N
<70
no
19.0
40
II
308
LD
N
>= 70
no
12.5
20
LD
N
yes
LD
E
III
74
ED
N
10.5
10
IV
424
ED
E
6.3
2
Authors: Kawahara et al (this study)
Class
n
LDH
PS
Serum
MST
Survival rate (%)
Na
(mo)
2 yr
3 yr
5 yr
I
42
N
0-1
N
16.0
31
22
19
II
45
N
0-1
L
9.4
6
5
2
N
2-3
E
0-1
III
50
E
2-3
6.6
2
0
0
References
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: www-admin{at}oup.co.uk
Last modification: 19 May 1998
Copyright© Japanese Journal of Clinical Oncology, 1997.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||