If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
The Impact of Additional Prognostic Factors on Survival and their Relationship with the Anatomical Extent of Disease Expressed by the 6th Edition of the TNM Classification of Malignant Tumors and the Proposals for the 7th Edition
Address for correspondence: Jean-Paul Sculier, Department of Critical Care and Thoracic Oncology, Institut Jules Bordet, Université Libre de Bruxelles (ULB); 1, rue Héger-Bordet, B – 1000 Bruxelles, Belgium
To identify, in the international staging database of the International Association for the Study of Lung Cancer, those prognostic factors that were significant and independent of clinical stage.
Material and Methods
From the data submitted to the staging data base concerning 100,869 patients, cases were selected for which all the following variables were available: clinical stage, age, gender, performance status (PS), and histologic cell types. For non-small cell lung cancer (NSCLC), 12,428 patients were assessable, and for SCLC, 6609 patients were available for this study. Methods used were Cox regression analyses and recursive partitioning and amalgamation analyses.
Results
PS appeared to be a very important prognostic factor for survival in addition to clinical stage. Age and gender were other independent significant variables; For NSCLC and SCLC separately, recursive partitioning and amalgamation allowed the identification of four groups of patients with differing prognoses. In advanced NSCLC (stage IIIB / IV), some routine laboratory tests (mainly white blood cells and hypercalcaemia) were also found to be significant prognostic variables. In SCLC, albumin was an independent biologic prognostic factor.
Conclusion
In addition to stage, PS and, to a lesser extent, age and gender seem to be important prognostic factors for survival in lung cancer. Although this data was obtained from the largest series ever used for such an analysis in lung cancer, these prognostic factors and models require confirmation in the prospective study already planned by the International Association for the Study of Lung Cancer Lung Cancer Staging Project.
The International Association for the Study of Lung Cancer (IASLC) International Staging Committee (ISC) has submitted proposals for revision of the tumor, node, metastasis (TNM) descriptors
The IASLC Lung Cancer Staging Project: proposals for the revision of the T descriptors in the forthcoming (seventh) edition of the TNM classification for lung cancer.
The IASLC Lung Cancer Staging Project: proposals for the revision of the N descriptors in the forthcoming seventh edition of the TNM classification for lung cancer.
The IASLC Lung Cancer Staging Project: proposals for revision of the M descriptors in the forthcoming (seventh) edition of the TNM classification of lung cancer.
The IASLC Lung Cancer Staging Project: proposals for the revision of the TNM stage groupings in the forthcoming (seventh) edition of the TNM Classification of malignant tumours.
for lung cancer in the forthcoming (7th) edition of the International Union Against Cancer and American Joint Committee on Cancer TNM Classification of Malignant Tumors. The suggestions are based on survival analyses that show better discrimination is achieved by these changes to the TNM descriptors and stage groupings. These proposals were developed using a very large database that was specifically collected from individual databases for that purpose.
The prognostic factors subcommittee of the ISC analyzed in the retrospective ISC database the role of additional prognostic factors for survival, whether related to the tumor or patient characteristics.
METHODS
We describe here the methods specifically used for the purpose of the present study, which is the identification of significant independent prognostic factors for survival in addition to the anatomic extent of disease, expressed by TNM. The general methodology of the IASLC Lung Cancer Staging Project and the major proposals have already been published.
The IASLC Lung Cancer Staging Project: proposals for the revision of the T descriptors in the forthcoming (seventh) edition of the TNM classification for lung cancer.
The IASLC Lung Cancer Staging Project: proposals for the revision of the N descriptors in the forthcoming seventh edition of the TNM classification for lung cancer.
The IASLC Lung Cancer Staging Project: proposals for revision of the M descriptors in the forthcoming (seventh) edition of the TNM classification of lung cancer.
The IASLC Lung Cancer Staging Project: proposals for the revision of the TNM stage groupings in the forthcoming (seventh) edition of the TNM Classification of malignant tumours.
The IASLC Lung Cancer Staging Project: validation of the proposals for revision of the T, N, and M descriptors and consequent stage groupings in the forthcoming (seventh) edition of the TNM classification of malignant tumours.
The total number of patients submitted to the staging data basis was 100,869 of whom 81,015 remained eligible for analysis after exclusion of cases outside the study period (1990–2000), those with unknown cell type, those not newly diagnosed at the point of entry and those with inadequate information on stage, treatment or follow-up. Of the eligible patients there were 67,725 cases of non-small cell lung cancer (NSCLC) and 13,290 with SCLC.
Among the potentially useful prognostic variables for lung cancer survival (Table 1), data for many were not available in the IASLC staging project database, such as those related to tumor biology or the role of fluorodeoxyglucose position emission tomography (FDG-PET) scanning. For this reason, we had to restrict our analysis to those variables for which we had enough information in a significant number of patients. These included clinical stage, expressed as TNM for NSCLC and limited disease (LD) versus extensive disease (ED) for SCLC (defined by the local institution), age, gender, performance status (PS), and histologic cell type. For NSCLC, 12,428 patients with stage I–IV were assessable and for SCLC, 6609 (3739 with ED and 2870 with LD). For smaller subsets of SCLC and advanced stage NSCLC, we had laboratory values: serum calcium, serum albumin, serum sodium, hemoglobin, and white blood cell count (WBC).
TABLE 1Potentially Useful Prognostic Variables for Lung Cancer Survival
Tumor characteristics: localization of metastatic sites: brain, liver, adrenals, bone, lung; number of metastatic sites; pleural effusion; type of lesions
(assessable, measurable); tumor size and volume; histology: non-small cell lung cancer (NSCLC) or small cell lung cancer (SCLC); squamous cell
carcinoma versus adenocarcinoma versus large cell carcinoma…; neuroendocrine tumors; tumor differentiation and grade; lymphatic and blood vessel
Survival was measured from the date of entry (date of diagnosis for registries, date of registration for trials) for clinically staged data and was calculated by the Kaplan-Meier method. Prognostic groups were assessed by Cox regression analysis, using the SAS system for windows version 9.0 PHREG procedure.
Recursive partitioning and amalgamation (RPA) analyses were performed to generate tree-based models by stage plus the key prognostic factors; age, gender, PS, and, for NSCLC only, cell type. The analyses were performed on a randomly selected training set comprising two-thirds of the available cases, reserving the remaining cases for later validation. Separate training and validation sets were established for NSCLC and SCLC. The random selection processes were stratified by type of database submission (clinical trials, registries, consortia/surgical series) and time period of case entry (1990–1995 versus 1995–2000).
The RPA analysis generated tree-based models for the survival data using logrank test statistics for recursive partitioning, for selection of the important groupings and bootstrap resampling to correct for the adaptive nature of the splitting algorithm. The terminal nodes were then grouped according to similar hazards and the newly formed groups were in each case evaluated using the remaining one-third of the data.
Laboratory values were first checked for consistency of the units used by laboratories across the contributing databases, and queries or corrections were made where necessary. When viewing survival prognosis for each laboratory measure as a continuous variable in a running logrank test, we found that the best split point for the data for prognosis was consistently at the defined upper or lower limits of normal. Therefore each result was defined dichotomously based on the known normal range for each laboratory measurement. Using these cutpoints, the laboratory measurements were analyzed for prognosis univariately using Kaplan Meier, and multivariately in conjunction with other factors with Cox proportional hazards regression.
Because of the multiple tests performed on the database, the level of significance for p was chosen as <0.01.
RESULTS
Non-small Cell Lung Cancer
Analyses of Survival Using General Characteristics
Cox Models
The following variables were considered: clinical TNM stage as proposed by the IASLC Lung Cancer Project, age, gender, histologic cell type (adenocarcinoma versus squamous cell carcinoma versus other types), and performance status (PS) using the Zubrod scale.
The results of the Cox proportional hazards regression model performed on the whole population (12,426 patients) is shown in Table 2. Older age, more advanced stage, male gender, poorer PS and nonsquamous cell histology were found to be significantly associated with decreased survival. The R2 value for the full model with clinical stage (as proposed by IASLC) and the full set of prognostic factors is 36.2. For a model with just clinical stage and without the other factors, the R2 value is 32.9. Thus, removing the prognostic factors from the model does not substantially effect the hazard ratios on the clinical stage parameters.
TABLE 2Multivariate Analysis of Prognostic Factors for Survival in NSCLC, Using General Characteristic Variables (Clinical TNM Stage as Proposed by IASLC Staging Project, Age, Gender, PS, Histological Cell Type)
Variable
n/N (%)
HR (95% CI)
P
Stage II
1531/12426 (12%)
1.80 (1.65, 1.97)
<0.001
Stage IIIA
2048/12426 (16%)
2.71 (2.49, 2.95)
<0.001
Stage IIIB/IV
7280/12426 (59%)
5.34 (4.95, 5.76)
<0.001
Age (continuous)
N = 12426
1.01 (1.00, 1.01)
<0.001
Squamous cell type
5304/12426 (43%)
0.93 (0.89, 0.97)
<0.001
Male gender
9764/12426 (79%)
1.17 (1.11, 1.23)
<0.001
PS 1 (vs. 0)
6294/12426 (51%)
1.38 (1.32, 1.44)
<0.001
PS 2 (vs. 0)
1423/12426 (11%)
2.09 (1.95, 2.23)
<0.001
PS 3–4 (vs. 0)
579/12426 (5%)
3.48 (3.17, 3.83)
<0.001
R2 = 36.2%.
NSCLC, non-small cell lung cancer; IASLC, International Association for the Study of Lung Cancer; TNM, tumor, node, metastasis; PS, performance status; HR, hazard ratio; 95% CI, 95% confidence interval; p value from Wald χ2 Test in Cox Regression; R2: R-squared using method by O’Quigley and Xu.
The same prognostic factors were found to be important when the multivariate analysis was performed using the 6th edition of the TNM Classification of Malignant Tumors
TABLE 3Multivariate Analysis of Prognostic Factors for Survival in NSCLC, Using the 6th Edition of TNM (R2 = 35.4%)
Variable
n/N (%)
HR (95% CI)
P
Age
N = 12426
1.01 (1.01, 1.01)
<0.001
6th Ed TNM II
1119/12426 (9%)
1.86 (1.70, 2.03)
<0.001
6th Ed TNM IIIA
1925/12426 (15%)
2.54 (2.35, 2.75)
<0.001
6th Ed TNM IIIB & IV
7401/12426 (60%)
4.79 (4.48, 5.12)
<0.001
Male
9764/12426 (79%)
1.16 (1.11, 1.22)
<0.001
PS 1
6294/12426 (51%)
1.38 (1.32, 1.45)
<0.001
PS 2
1423/12426 (11%)
2.11 (1.97, 2.25)
<0.001
PS 3–4
579/12426 (5%)
3.50 (3.18, 3.85)
<0.001
Squamous cell
5304/12426 (43%)
0.92 (0.88, 0.96)
<0.001
NSCLC, non-small cell lung cancer; TNM, tumor, node, metastasis; PS, performance status; HR, hazard ratio; 95% CI, 95% confidence interval; p value from Wald χ2 Test in Cox Regression; R2: R-squared using method by O’Quigley and Xu.
When prognostic factors were analyzed by clinical stage, as proposed by the IASLC staging project (Table 4), histology cell type was a significant prognostic factor for survival only in patients with stage IIIA, whereas PS, gender, and age were significant in all stages, but with a lower limit for age in advanced stages.
TABLE 4Multivariate Analysis of Prognostic Factors for Survival Within Clinical Stages in NSCLC (as proposed by the IASLC for the 7th Ed of TNM), Using General Characteristic Variables (Age, Gender, PS, Histological Cell Type)
Variable
n/N (%)
HR (95% CI)
P
Stage I–II
Male
2628/3098 (85%)
1.22 (1.06, 1.40)
0.006
Squamous
1747/3098 (56%)
0.99 (0.90, 1.09)
0.840
Age
N = 3098
1.02 (1.01, 1.02)
<0.001
PS 1
1852/3098 (60%)
1.32 (1.19, 1.46)
<0.001
PS 2
117/3098 (4%)
2.60 (2.07, 3.27)
<0.001
PS 3–4
25/3098 (1%)
7.19 (4.64, 11.14)
<0.001
Stage IIIA
Age
N = 2048
1.01 (1.00, 1.01)
0.002
Male
1635/2048 (80%)
1.21 (1.07, 1.38)
0.003
Squamous
1072/2048 (52%)
0.86 (0.78, 0.95)
0.003
PS 1
871/2048 (43%)
1.33 (1.20, 1.48)
<0.001
PS 2
155/2048 (8%)
1.93 (1.60, 2.33)
<0.001
PS 3–4
37/2048 (2%)
3.94 (2.74, 5.69)
<0.001
Stage IIIB/IV
Age
N = 7280
1.00 (1.00, 1.01)
0.011
Squamous
2485/7280 (34%)
0.95 (0.90, 1.00)
0.042
Male
5501/7280 (76%)
1.15 (1.09, 1.22)
<0.001
PS 1
3571/7280 (49%)
1.41 (1.34, 1.50)
<0.001
PS 2
1151/7280 (16%)
2.13 (1.97, 2.30)
<0.001
PS 3–4
517/7280 (7%)
3.45 (3.11, 3.83)
<0.001
NSCLC, non-small cell lung cancer; IASLC, International Association for the Study of Lung Cancer; TNM, tumor, node, metastasis; PS, performance status; HR, hazard ratio; 95% CI, 95% confidence interval; p value from Wald χ2 Test in Cox Regression.
A recursive positioning and amalgamation analysis was performed on the patient population, randomly divided for that purpose in to a learning set (66% = 8199 cases) and a validation set (34% = 4227 cases). The following factors were used to obtain the tree shown in Figure 1: PS (ordered), IASLC stage (ordered with IIIB-IV combined), age (continuous), squamous cell type (yes/no), and gender. The resulting, pruned survival tree showed significant split points for stage, age and PS; cell type did not appear as important as those variables. The numbers at the terminal nodes (Figure 1) represent the parameter estimates (log-hazard ratio) and the number of cases for each terminal node group. The reference group with best prognosis is circled in blue at the far left of the figure. The colored circles encompass those groups with similar prognoses. Four groups with significantly different prognosis were identified: Group 1 (blue) with stage IA–IIA (any age and any PS); Group 2 (red) with stage IIB/IIIA and PS 0–1 (any age); Group 3 (green) with stage IIB/IIIA and PS 2 (any age) or with stage IIIB/IV and PS 0 (any age) or with stage IIIB/IV, age <81 years and PS 1; Group 4 (orange) with stage IIB/IIIA and PS 3–4 (any age) or with stage IIIB/IV and PS 2–4 (any age) or with stage IIIB/IV, PS 1 and age >80 years. The resulting amalgamated categories were applied to a survival analysis on the validation set of patients (Figure 2). Median survival times were, respectively, 53 months for group 1, 16 months for group 2, 8 months for group 3, and 3 months for group 4.
FIGURE 1Survival tree of the recursive partitioning and amalgamation analysis, performed on a learning set of 8199 non-small cell lung cancer cases.
Analyses of Survival Using Laboratory Values in Advanced Stage IIIB/IV
A total of 7280 cases with advanced NSCLC in the database had data on at least one of the following laboratory values in addition to the other prognostic factors (age, gender, PS): calcium (1316 cases), albumin (1887 cases), sodium (1708 cases), hemoglobin (1564 cases), and WBC (2126 cases). A Cox model was performed with each individual laboratory value and the other prognostic factors (Table 5). The laboratory variables in advanced NSCLC seemed to be strong prognostic factors in a magnitude similar to PS, whereas age and gender were weaker. In 537 patients, data was available on all of the five laboratory values. A multivariate model (Table 6) identified as strong significant prognostic factors (p < 0.001) PS and WBC, followed by calcium (p = 0.0077), albumin (p = 0.013), and age ≥75 years (p = 0.0415).
TABLE 5Analyses Using Laboratory Values in Advanced NSCLC Disease (Stages IIIB/IV): Results of Models with Individual Laboratory Variables
Age ≥75 yr
Gender: Male
PS (ordered)
Laboratory Variable
Laboratory Variable
No. of Cases
p
HR
P
HR
P
HR
P
HR
Calcium >10.4 mg/dl
1316
0.2248
1.19
0.0356
1.16
<0.0001
1.35
0.0010
1.55
Albumin <3.2 g/dl
1887
0.0116
1.28
0.1875
1.07
<0.0001
1.47
<0.0001
1.45
Sodium <135 mmol/l
1708
0.0674
1.27
0.0016
1.21
<0.0001
1.35
<0.0001
1.35
WBC >10,000 cells/ml
2126
0.0006
1.50
0.0277
1.13
<0.0001
1.42
<0.0001
1.43
Hemoglobin <12 g/dl for females, <13 g/dl for males
TABLE 6Multivariate Model for Survival Performed in Advanced NSCLC (Stages IIIB/IV) in a Set of 537 Patients for which all 5 Laboratory Variables were Available
Variable
P
HR
Age ≥75 yr
0.0415
1.39
Male
0.4761
0.93
PS (ordered: 0, 1, 2, 3–4)
<0.0001
1.44
Calcium >10.4 mg/dl
0.0077
1.77
Albumin <32 g /dl
0.013
1.33
Sodium <135 mmol/l
0.4823
1.09
Hemoglobin < 12 g/dl for females, 13 g/dl for males
Analyses of Survival Using General Characteristics
Cox Models
The IASLC database contained data for analysis of prognostic factors for 6609 cases with SCLC. This included 2870 patients with LD and 3739 with ED. For each individual, data was available on PS, age, and gender. Two percent (122 cases) had had surgery. Table 7 shows the results of the multivariate analysis for survival on the whole population. All variables tested (PS, extent of disease, gender, and age) were found to be independent prognostic factors for survival. Table 8 shows the results of the models performed on the subsets of patients with limited or ED. The same independent prognostic factors (PS, gender, and age) were identified as in NSCLC.
TABLE 7Multivariate Analysis of Prognostic Factors for Survival in SCLC, Using General Characteristic Variables (Limited vs. Extensive Stage, Age, Gender, PS)
Variable
n/N (%)
HR (95% CI)
P
Age
N = 6609
1.01 (1.01, 1.02)
<0.001
Extensive stage (vs Lim.)
3739/6609 (57%)
2.13 (2.02, 2.25)
<0.001
Male (vs. female)
4368/6609 (66%)
1.25 (1.19, 1.32)
<0.001
PS 1
3161/6609 (48%)
1.36 (1.28, 1.44)
<0.001
PS 2
1060/6609 (16%)
1.93 (1.78, 2.09)
<0.001
PS 3–4
349/6609 (5%)
3.45 (3.05, 3.89)
<0.001
SCLC, small cell lung cancer; PS, performance status; HR, hazard ratio; 95% CI, 95% confidence interval; p value from Wald χ2 Test in Cox Regression.
TABLE 8Multivariate Analyses of Prognostic Factors for Survival in Limited and Extensive SCLC, Using General Characteristic Variables (Age, Gender, PS)
Variable
n/N (%)
HR (95% CI)
P
Limited stage
Age
N = 2870
1.01 (1.01, 1.02)
<0.001
Male
1838/2870 (64%)
1.21 (1.11, 1.32)
<0.001
PS 1
1338/2870 (47%)
1.42 (1.30, 1.55)
<0.001
PS 2
277/2870 (10%)
1.72 (1.49, 1.98)
<0.001
PS 3–4
129/2870 (4%)
3.68 (3.03, 4.47)
<0.001
Extensive stage
Age
N = 3739
1.01 (1.01, 1.02)
<0.001
Male
2530/3739 (68%)
1.28 (1.20, 1.38)
<0.001
PS 1
1823/3739 (49%)
1.32 (1.21, 1.43)
<0.001
PS 2
783/3739 (21%)
1.98 (1.79, 2.18)
<0.001
PS 3–4
220/3739 (6%)
3.32 (2.84, 3.88)
<0.001
SCLC, small cell lung cancer; PS, performance status; HR, hazard ratio; 95% CI, 95% confidence interval; p value from Wald χ2 Test in Cox Regression.
For the RPA analysis, the population was split in to a learning set (66% of the cases = 4359) and a validation set (34% = 2250). The factors entered in to this analysis were: stage (LD versus ED), PS (0–4) age and gender. The resultant tree (Figure 3) shows that stage and PS are the two most important prognostic factors. The reference group with best prognosis is circled in blue at the far left of the figure. Four groups with differing prognoses were identified: group 1 (blue) with LD, PS 0 and age <60 years or LD, PS 1–2 and age <65 years; group 2 (red) with LD, PS 1–2, age ≥65 years or female with ED, PS 0, and age <65 years; group 3 (green) female with ED, PS0 and age ≥65 years or male with ED, PS 0 or both genders with ED, PS 1, age <70 years; and group 4 (orange) with LD PS 3–4 or with ED, PS 1, age ≥70 years or with ED, PS 2–4. The resulting amalgamated categories were applied to a survival analysis in the validation set (Figure 4). Median survival times were, respectively, 17 months for group 1, 12 months for group 2, 10 months for group 3, and 6 months for group 4.
FIGURE 3Survival tree of the recursive partitioning and amalgamation analysis, performed on a learning set of 4359 small cell lung cancer cases.
A total of 6609 cases with SCLC in the database had data on at least one of the following laboratory values in addition to the other prognostic factors (stage, age, gender, PS): calcium (1849 cases), albumin (2773 cases), sodium (2390 cases), hemoglobin (1487 cases), and WBC (1828 cases). A Cox model was performed with each individual laboratory value and the other prognostic factors (Table 9). Two laboratory variables (sodium and albumin) appeared to be very significant prognostic factors in addition to stage, PS, and gender. In 650 patients there was data available on all five laboratory values. A multivariate model (Table 10) identified albumin only, in addition to extent of disease and gender, as significant prognostic factors.
TABLE 9Analyses Using Laboratory Values in SCLC: Results of Models with Individual Laboratory Variables
Stage: Extensive
Age ≥75 yr
Gender: Male
PS (ordered)
Laboratory Variable
Laboratory Variable
No. of Cases
P
HR
P
HR
P
HR
P
HR
P
HR
Calcium >10.4 mg/dl
1849
<0.0001
1.87
0.8528
1.02
0.0008
1.19
<0.0001
1.5
0.0608
1.28
Albumin <3.2 g/dl
1887
<0.0001
1.79
0.0403
1.21
<0.0001
1.23
<0.0001
1.44
<0.0001
1.31
Sodium <135 mmol/l
2390
<0.0001
1.96
0.6003
1.07
0.0001
1.19
<0.0001
1.48
<0.0001
1.31
WBC >10, 000 cells/μl
1828
<0.0001
2.12
0.8544
1.03
<0.001
1.26
<0.0001
1.33
0.2414
1.07
Hemoglobin <12 g/dl for females, <13 g/dl for males
1487
<0.0001
2.25
0.9618
0.99
0.0011
1.21
<0.0001
1.34
0.5040
1.04
SCLC, small cell lung cancer; PS, performance status; HR, hazard ratio; WBC, white blood cells.
The IASLC staging project allowed an analysis of prognostic factors for survival in a very large population of lung cancer patients: more than 12,000 with NSCLC and more than 6000 with SCLC. In addition to clinical stage, PS appeared to be a very important prognostic factor. Age and gender were other independent significant variables. For NSCLC, histologic cell type was a significant prognostic factor in stage IIIA, with squamous cell lung cancer having a better prognosis in comparison to other cell types. For both NSCLC and SCLC, RPA allowed the identification of four groups of patients with differing prognoses. Finally, in advanced NSCLC, some biologic variables (serum calcium, albumin, and WBC) were found to be prognostic factors in addition to stage, age, and gender. In SCLC, albumin only, in addition to extent of disease and gender, was identified as a significant prognostic biologic factor. Our findings are summarized in Table 11 with a grading as following: ++++ and +++ = factors present in any model; ++ = factors significant in RPA and Cox models; + = factors significant in Cox models (or in a meta-analysis for SUVmax); +§ = biologic factors significant in Cox models not taking into consideration other biologic variables.
TABLE 11Summary of the Prognostic Factors with a Grading as Following: + + + + and + + + = factors present in any model; ++ = factors significant in RPA and Cox models; + = factors significant in Cox models (or in a meta-analysis for SUVmax); +§ = biological factors significant in Cox models not taking into consideration other biological variables
There are some methodological problems that may limit the generalization of these results. Firstly, the retrospective nature of the database, that allowed such a very high number of lung cancer patients to be collected, inevitably lead to a lot of missing data on some prognostic factors. This restricted our analysis to a subset of the population for which we had each prognostic factor: clinical stage as proposed by the IASLC staging project,
PS, gender, age, and histologic cell type. Pathologically, stage was not considered because in general we did not get much prognostic factor data, particular PS, from the surgical database and for those for which we had the data, they are nearly all either P.S. 0 or 1, which is not very interesting for prognostic analysis. For SCLC, we were unable to use TNM stage and limited our analysis to that limited versus ED as defined in the local databases. As biologic variables were mainly available in patients with advanced NSCLC (stage IIIB/IV), the analysis was restricted to those cases for which we also had the other prognostic factors (age, PS, gender). When we wished to study, in a multivariate model, the five biologic variables, data was only available in little more than 500 patients. Secondly, the local databases amalgamated in the staging project were of differing types: clinical trials, surgical series, institution series and registries. Some have been the topic of prognostic analyses that have already been published. Therefore our results are not fully independent of the data already in the literature. Thirdly, the effect of the treatment has not been taken into account because of the great heterogeneity of the various local databases. The prognostic analysis thus reflects not only the natural history of the disease but also the modalities of treatment used according to the local attitudes worldwide. Fourthly, a large number of statistical tests necessitated an increase of the threshold level of significance. Our analysis does not have the strength of a prospective study in which statistical analysis could be based upon a predefined primary end point. For all of these reasons the reported conclusions have to be considered as exploratory, requiring confirmation in a prospective validation study.
Despite these limitations, this study of prognostic factors provides important information. It confirms, in the largest series ever published, the prognostic role of gender and age in addition to PS and clinical stage, whether one uses the 6th edition of the TNM Classification of Malignant Tumors or the IASLC proposals for the forthcoming 7th edition.
Similar general prognostic factors were found to be important for NSCLC and SCLC. The RPA allowed the construction of trees (FIGURE 1, FIGURE 3), which may help in the management of lung cancer patients. For NSCLC, the model developed included all clinical stages, which is a new finding. Indeed the published survival models have so far been limited to retrospective analyses of data bases of clinical trials performed by cooperative groups in advanced disease.
Prognostic factors for survival in advanced non-small-cell lung cancer: univariate and multivariate analyses including recursive partitioning and amalgamation algorithms in 1,052 patients. The European Lung Cancer Working Party.
Clinical model to predict survival in chemonaive patients with advanced non-small-cell lung cancer treated with third-generation chemotherapy regimens based on eastern cooperative oncology group data.
It should be noted that the number of patients included in those studies ranged form 893 to 2631, which is far less than the 12,428 cases available for our analyses. By taking into consideration the clinical stage in a population with a much broader case-mix, stage seems to be the strongest prognostic factor. PS becomes important only in stage IIB or higher when analyzed with RPA due to the very low frequency of patients with poor PS in the early stage cases (FIGURE 1, FIGURE 2).
For SCLC, we also identified that the extent of disease at presentation, classified as ED versus LD, was the most important factor, with PS and age also being important. Gender was found to have a role in deciding prognosis in ED (FIGURE 3, FIGURE 4). Other models, performed on databases of clinical trial cases from cooperative groups, are published in the literature.
Prognostic factors for patients with small cell lung carcinoma: analysis of a series of 763 patients included in 4 consecutive prospective trials with a minimum follow-up of 5 years.
These publications also identify the extent of disease at presentation as the main factor. The number of patients included in those analyses was respectively, 614, 2580, and 763, many fewer than those used in the present study. The large size of our series allowed us to perform a validation analysis in a subset of 2250 patients, which were not used to construct the model.
Other prognostic factors were identified in this study by the analysis of data within different stages of disease. Histologic cell type is a controversial prognostic factor in NSCLC.
The data presented in this report (Table 4) revealed that cell type was only a significant prognostic factor for survival in stage IIIA in addition to PS, age and gender, when stage IIIA was defined using the IASLC proposals for the 7th edition of TNM.
The IASLC Lung Cancer Staging Project: proposals for the revision of the TNM stage groupings in the forthcoming (seventh) edition of the TNM Classification of malignant tumours.
Squamous cell cancer seemed to have a better prognosis than other histologies. This information suggests one should stratify by histology in trials that include patients with stage IIIA disease.
Routine laboratory variables have been included in many models for advanced disease published in the literature. Because of a lot of missing data, the analysis of these was restricted to assessing their value in advanced (stage IIIB/IV) NSCLC. Adequate data was available on five blood tests: calcium, albumin, sodium, WBC, and hemoglobin. All were found to be significant variables (Table 5). A model with a small number of patients was constructed, which suggested that WBC and calcium might be important prognostic factors, in accordance with other studies.
Prognostic factors for survival in advanced non-small-cell lung cancer: univariate and multivariate analyses including recursive partitioning and amalgamation algorithms in 1,052 patients. The European Lung Cancer Working Party.
The same approach in SCLC allowed identifying albumin as a significant independent factor in addition to PS and extent, which is a new information, albumin having not been so far investigated in large series.
Many potentially useful prognostic variables (Table 1) were not investigated because of missing data or missing variables, inevitable in a retrospective database. Some could not be assessed because they had only recently been suggested, such as for those related to molecular biology or to the value of PET scanning. In regard to the role of molecular or biologic markers in lung cancer, more than 5000 articles have been published, often of varying methodological quality. The best evidence available in the literature at present on this subject has come from the meta-analyses summarized in Table 12. The estimate of the prognostic value of these variables reported in these studies is limited by their use of univariate analyses. The independent role of the prognostic factors identified in this study has to be confirmed in prospective studies, using methodology such as that recently proposed by Zhu et al.
The prognostic value of the primary tumor SUV max (maximal standard uptake value) measured on FDG-PET has been assessed by a meta-analysis of the literature, undertaken by the European Lung Cancer Working Party evidence-based medicine committee for the IASLC staging project. This showed that SUV max is a strong prognostic factor for survival in a univariate analysis.
Primary tumor standardized uptake value (SUVmax) measured on fluorodeoxyglucose positron emission tomography (FDG-PET) is of prognostic value for survival in non-small cell lung cancer (NSCLC): a systematic review and meta-analysis (MA) by the European Lung Cancer Working Party for the IASLC Lung Cancer Staging Project.
This should be confirmed by a meta-analysis based on individual patient data allowing multivariate analysis to be performed which takes into account the prognostic factors identified in the present article.
TABLE 12Meta-Analyses Published on the Prognostic Value of Biological or Genetic Markers for Survival in Lung Cancer
Biological Variable
Prognostic Factor
Reference
Bcl-2
Good
Martin et al., 200323
TTF1
Poor
Berghmans et al., 200624
Cox2
Poor
Mascaux et al., 200625
EGFR
Poor
Nakamura et al., 200626; Meert et al., 200227
ras
Poor
Mascaux et al., 200528; Huncharek et al., 199929
Ki67
Poor
Martin et al., 200430
HER2
Poor
Meert et al., 200331 Nakamura et al., 200532
VEGF
Poor
Delmotte et al., 200233
Microvascular density
Poor
Meert et al., 200234
p53
Poor
Steels et al., 200135 Mitsudomi et al., 200036 Huncharek et al., 200037
In conclusion, the present analysis of the database of the IASLC staging project identified important prognostic factors for survival in lung cancer patients in addition to clinical stage. Those factors were PS, age, and gender. In stage IIIA NSCLC, histologic cell type seemed to be important. In advanced NSCLC some routine laboratory tests were found to be additional, significant factors. In SCLC, albumin was also found to be an independent prognostic factor. Models were constructed using a RPA method for both NSCLC and SCLC. They allowed the identification of groups of patients with differing prognoses, taking into account clinical stage and PS and, to a lesser extent, age, and gender. Although the results reported in this study were obtained in the largest series ever used for prognostic analysis in lung cancer, the prognostic variables and models found require to be confirmed by a prospective study, such as that already planned by the IASLC Lung Cancer Staging Project. It is our hope that colleagues around the world will continue to support this initiative.
APPENDIX 1
IASLC ISC
P. Goldstraw (Chairperson), Royal Brompton Hospital, Imperial College, London, UK; H. Asamura, National Cancer Centre Hospital, Tokyo, Japan; D. Ball, Peter MacCallum Cancer Centre, East Melbourne, Australia; V. Bolejack, Cancer Research and Biostatistics, Seattle, Washington, USA; E. Brambilla, Laboratoire de Pathologie Cellulaire, Grenoble Cedex, France; P.A. Bunn, University of Colorado Health Sciences, Denver, Colorado; D. Carney, Mater Misericordiae Hospital, Dublin, Ireland; K. Chansky, Cancer Research and Biostatistics, Seattle, Washington, USA; T. Le Chevalier (resigned), Institute Gustave Roussy, Villejuif, France; J. Crowley, Cancer Research and Biostatistics, Seattle, Washington, USA; R. Ginsberg (deceased), Memorial Sloan-Kettering Cancer Center, New York, USA; D. Giroux, Cancer Research And Biostatistics, Seattle, Washington, USA; P. Groome, Queen's Cancer Research Institute, Kingston, Ontario, Canada; H.H. Hansen (retired), National University Hospital, Copenhagen, Denmark; P. Van Houtte, Institute Jules Bordet, Bruxelles, Belgium; J.-G. Im, Seoul National University Hospital, Seoul, South Korea; J.R. Jett, Mayo Clinic, Rochester, Minnesota, USA; H. Kato (retired), Tokyo Medical University, Tokyo Japan; C. Kennedy, University of Sydney, Sydney, Australia; M. Krasnik, Gentofte Hospital, Copenhagen, Denmark; J. van Meerbeeck, University Hospital, Ghent, Belgium: T. Naruke (deceased), Saiseikai Central Hospital, Tokyo, Japan; E.F. Patz, Duke University Medical Center, Durham, North Carolina, USA; P.E. Postmus, Vrije Universiteit Medical Center, Amsterdam, the Netherlands; R. Rami-Porta, Hospital Mutua de Terrassa, Terrassa, Spain; V. Rusch, Memorial Sloan-Kettering Cancer Center, New York, USA; J.P. Sculier, Institute Jules Bordet, Bruxelles, Belgium; Z. Shaikh, Royal Brompton Hospital, London, UK; F.A. Shepherd, University of Toronto, Toronto, Ontario, Canada; Y. Shimosato (retired), National Cancer Centre, Tokyo, Japan; L. Sobin, Armed Forces Institute of Pathology, Washington, DC; W. Travis, Memorial Sloan-Kettering Cancer Center, New York, USA; M. Tsuboi, Tokyo Medical University, Tokyo, Japan; R. Tsuchiya, National Cancer Centre, Tokyo, Japan; E. Vallieres, Swedish Cancer Institute, Seattle, Washington, USA; J. Vansteenkiste, Leuven Lung Cancer Group, Belgium; Yoh Watanabe (deceased), Kanazawa Medical University, Uchinada, Japan and H. Yokomise, Kagawa University, Kagawa, Japan.
Participating Institutions
O. Visser, Amsterdam Cancer Registry, Amsterdam, The Netherlands; R. Tsuchiya and T. Naruke (deceased), Japanese Joint Committee of Lung Cancer Registry; J.P. Van Meerbeeck, Flemish Lung Cancer Registry-VRGT, Brussels, Belgium; H. Bülzebruck, Thorax-klinik am Universitatsklinikum, Heidelberg, Germany; R. Allison and L. Tripcony, Queensland Radium Institute, Herston, Australia; X. Wang, D. Watson and J. Herndon, Cancer and Leukemia Group B (CALGB), USA; R.J. Stevens, Medical Research Council Clinical Trials Unit, London, England; A. Depierre, E. Quoix and Q. Tran, Intergroupe Francophone de Cancerologie Thoracique (IFCT), France; J.R. Jett and S. Mandrekar, North Central Cancer Treatment Group (NCCTG), USA; J.H. Schiller and R.J. Gray, Eastern Cooperative Oncology Group (ECOG), USA; J.L. Duque-Medina and A. Lopez- Encuentra, Bronchogenic Carcinoma Co-operative Group of the Spanish Society of Pneumology and Thoracic Surgery (GCCB-S), Spain; J.J. Crowley, Southwest Oncology Group (SWOG); J.J. Crowley and K.M.W. Pisters, Bimodality Lung Oncology Team (BLOT), USA; T.E. Strand, Cancer Registry of Norway; S. Swann and H. Choy, Radiation Therapy Oncology Group (RTOG), USA; R. Damhuis, Rotterdam Cancer Registry, The Netherlands; R. Komaki and P. K. Allen, MD Anderson Cancer Center- Radiation Therapy (MDACC-RT), Houston, Texas, USA; J.P. Sculier and M. Paesmans, European Lung Cancer Working Party (ELCWP); Y.L. Wu, Guangdong Provincial People's Hospital, Peoples Republic of China; M. Pesek and H. Krosnarova, Faculty Hospital Plzen, Czech Republic; T. Le Chevalier and A. Dunant, International Adjuvant Lung Cancer Trial (IALT), France; B. McCaughan and C. Kennedy, University of Sydney, Sydney, Australia; F. Shepherd and M. Whitehead, National Cancer Institute of Canada (NCIC); J. Jassem and W. Ryzman, Medical University of Gdansk, Poland; G.V. Scagliotti and P. Borasio, Universita’ Degli Studi di Torino, S Luigi Hospital, Orbassano, Italy; K.M. Fong and L. Passmore, Prince Charles Hospital, Brisbane, Australia; V.W. Rusch and B.J. Park, Memorial Sloan-Kettering Cancer Center, New York, USA; H.J. Baek, Korea Cancer Centre Hospital, Seoul, South Korea; R.P. Perng, Taiwan Lung Cancer Society, Taiwan; R.C. Yung, A. Gramatikova, John Hopkins University, USA; J. Vansteenkiste, Leuven Lung Cancer Group (LLCG), Belgium; C. Brambilla and M. Colonna, Grenoble University Hospital-Isere Cancer Registry, France; J. Hunt and A. Park, Western Hospital, Melbourne Australia; J.P. Sculier and T. Berghmans, Institute of Jules Bordet, Brussels, Belgium; A.K. Cangir, Ankara University School of Medicine, Ankara, Turkey; D. Subotic, Clinical Centre of Serbia, Belgrade, Serbia; R. Rosell and V. Aberola, Spanish Lung Cancer Group (SLCG), Spain; A.A. Vaporciyan and A.M. Correa, MD, Anderson Cancer Center-Thoracic and Cardiovascular Surgery (MDACC-TCVS), Houston, Texas, USA; J.P. Pignon, T. Le Chevalier and R. Komaki, Institut Gustave Roussy (IGR), Paris, France; T. Orlowski, Institute of Lung Diseases, Warsaw, Poland; D. Ball and J. Matthews, Peter MacCallum Cancer Institute, East Melbourne, Australia; M. Tsao, Princess Margaret Hospital, Toronto, Ontario, Canada; S. Darwish, Policlinic of Perugia, Italy; H.I. Pass and T. Stevens, Karmanos Cancer Institute, Wayne State University, USA; G. Wright, St Vincent's Hospital, Victoria, Australia; C. Legrand and J.P. van Meerbeeck, European Organization for Research and Treatment of Cancer (EORTC), Brussels, Belgium.
REFERENCES
Goldstraw P
Crowley J
on behalf of the IASLC International Staging Project
The International Association for the Study of Lung Cancer international staging project on lung cancer.
The IASLC Lung Cancer Staging Project: proposals for the revision of the T descriptors in the forthcoming (seventh) edition of the TNM classification for lung cancer.
The IASLC Lung Cancer Staging Project: proposals for the revision of the N descriptors in the forthcoming seventh edition of the TNM classification for lung cancer.
The IASLC Lung Cancer Staging Project: proposals for revision of the M descriptors in the forthcoming (seventh) edition of the TNM classification of lung cancer.
The IASLC Lung Cancer Staging Project: proposals for the revision of the TNM stage groupings in the forthcoming (seventh) edition of the TNM Classification of malignant tumours.
The IASLC Lung Cancer Staging Project: validation of the proposals for revision of the T, N, and M descriptors and consequent stage groupings in the forthcoming (seventh) edition of the TNM classification of malignant tumours.
Prognostic factors for survival in advanced non-small-cell lung cancer: univariate and multivariate analyses including recursive partitioning and amalgamation algorithms in 1,052 patients. The European Lung Cancer Working Party.
Clinical model to predict survival in chemonaive patients with advanced non-small-cell lung cancer treated with third-generation chemotherapy regimens based on eastern cooperative oncology group data.
Prognostic factors for patients with small cell lung carcinoma: analysis of a series of 763 patients included in 4 consecutive prospective trials with a minimum follow-up of 5 years.
Primary tumor standardized uptake value (SUVmax) measured on fluorodeoxyglucose positron emission tomography (FDG-PET) is of prognostic value for survival in non-small cell lung cancer (NSCLC): a systematic review and meta-analysis (MA) by the European Lung Cancer Working Party for the IASLC Lung Cancer Staging Project.
The project was also supported by the AJCC grant “Improving AJCC/UICC TNM Cancer Staging.”
Eli Lilly and Company provided funding to support the International Association for the Study of Lung Cancer (IASLC) Staging Committee's work to establish a database and to suggest revisions to the 6th edition of the TNM classification for Lung Cancer (staging) through a restricted grant. Lilly had no input into the committee's analysis of the data, nor in their suggestions for revisions to the staging system.
31453-2&id=gr1.jpg){kind=link}
31453-2&id=gr2.jpg){kind=link}
31453-2&id=gr3.jpg){kind=link}
31453-2&id=gr4.jpg){kind=link}