Revisiting Stage IIIB and IV Non-small Cell Lung Cancer. Analysis of the Surveillance, Epidemiology, and End Results Data

Transcription

1 CHEST Revisiting Stage IIIB and IV Non-small Cell Lung Cancer Analysis of the Surveillance, Epidemiology, and End Results Data William N. William, Jr, MD; Heather Y. Lin, PhD; J. Jack Lee, PhD; Scott M. Lippman, MD; Jack A. Roth, MD; and Edward S. Kim, MD Original Research LUNG CANCER Background: The purpose of this population-based study is to provide a detailed analysis of survival outcome of patients with stage IIIB and IV non-small cell lung cancer (NSCLC) enrolled in the Surveillance, Epidemiology and End Results (SEER) program. Methods: We retrieved, from the SEER database, data on demographics, disease extension (size, extent of primary tumor, and nodal status), histology, primary treatment, and survival time of NSCLC cases diagnosed between 1998 and 2003 (n 138,063). Cases were reclassified into separate T4 (satellite, invasive, or pleural effusion) and M1 (ipsilateral, contralateral, or distant) categories based on the extent of the primary tumor and the location of metastatic disease. Univariate and multivariate analyses were performed to assess the effects of each variable on survival. Results: For stage IIIB NSCLC, T4 satellite had the best prognosis (comparable to T2 lesions), followed by T4 invasive and T4 pleural effusion. For stage IV, M1 ipsilateral had the best prognosis, followed by M1 contralateral and M1 distant. Nodal status remained a powerful determinant of survival, particularly for patients with T4 satellite, T4 invasive, M1 ipsilateral, and, to a less extent, M1 contralateral. Other prognostic variables were identified within each subgroup. Conclusions: In this report, we present the most comprehensive analysis performed to date of patients with stage IIIB and IV NSCLC enrolled in the SEER program. The survival trends observed here suggest that T4 satellite lung cancer cases should be redefined as T2b, and not T3 as recently proposed for the upcoming TNM classification, seventh edition. (CHEST 2009; 136: ) Abbreviations: AJCC American Joint Committee on Cancer; CI confidence interval; HR hazard ratio; IASLC International Association for the Study of Lung Cancer; NOS not otherwise specified; NSCLC non-small cell lung cancer; SEER Surveillance, Epidemiology, and End Results; UICC Union Internacional Contra la Cancrum Stage IIIB and IV non-small cell lung cancer (NSCLC) account for approximately 46% of newly diagnosed cases of this disease. 1 Despite the common classification, stage IIIB NSCLC comprises a heterogeneous group of patients who are often treated with different modalities. Similarly, although patients with stage IV NSCLC are usually considered incurable, there is a small subgroup of patients (ie, oligometastatic disease confined to the lung) who might experience longer term disease-free survival after definitive surgical resection. Although recognized by most clinicians, the prognostic differences in selected subgroups of patients within stage IIIB and IV NSCLC have only been explored in small, mostly single-institution studies. 2 8 Even the cohort For editorial comment see page 660 of patients utilized as the basis for the validation of the current (sixth) edition of the Union Internacional Contra la Cancrum (UICC)/American Joint Committee on Cancer (AJCC) TNM staging system contained only 1,030 patients with stage IIIB and 1,427 patients with stage IV disease, 1 and therefore CHEST / 136 / 3/ SEPTEMBER,

2 had limited power to accurately evaluate selected subgroups of patients with possible favorable T4 characteristics or oligometastatic disease. Recognizing the limitations of the population used to define the current staging system, the International Association for the Study of Lung Cancer (IASLC) organized a task force 9 to collect a large multinational database of patients with NSCLC, with the goal of refining the TNM classification. As a result, in 2007, the first proposal for a new (seventh edition) TNM classification system was published, 10 based on information derived from that defined cohort (which contained 2,570 stage IIIB and 3,213 stage IV cases). Among other changes, this proposal included a redefinition of satellite nodules within the same lobe (formerly T4) to T3, malignant pleural effusion (formerly T4) to M1a, ipsilateral nodules in different lobes (formerly M1) to T4, and contralateral nodules and no extrathoracic disease (formerly M1) to M1a This new TNM classification was corroborated by an analysis of the Surveillance, Epidemiology, and End Results (SEER) database performed by the IASLC International Staging Committee. 13 The recent IASLC proposal for TNM revisions is a commendable effort. However, a full description of the SEER database analysis performed by the IASLC task force has not been published. 13 We believe that this information would be extremely useful for the acceptance of the new staging system and could assist clinicians in estimating prognosis and determining treatment options. Therefore, the purpose of this study was as follows: (1) to present a thorough survival analysis of patients with stage IIIB and IV NSCLC enrolled in the SEER program, with a comprehensive description of the criteria for case selection and statistical methodology utilized; (2) to independently validate the TNM classification system proposed by the IASLC task force for this subgroup of patients; and (3) to provide detailed information on the survival outcome of patients with less common presentations of stage IIIB and IV Manuscript received December 16, 2008; revision accepted February 9, Affiliations: From the Departments of Thoracic/Head & Neck Medical Oncology (Drs. William, Lippman, and Kim), Biostatistics (Drs. Lin and Lee), and Thoracic and Cardiovascular Surgery (Dr. Roth), The University of Texas MD Anderson Cancer Center, Houston, TX.. Reproduction of this article is prohibited without written permission from the American College of Chest Physicians ( misc/reprints.xhtml). Correspondence to: Edward S. Kim, MD, Department of Thoracic/Head & Neck Medical Oncology, The University of Texas MD Anderson Cancer Center, 1515 Holcombe Blvd, Box 432, Houston, TX 77030; edkim@mdanderson.org DOI: /chest NSCLC (ie, satellite nodules within the same lobe, ipsilateral nodules in different lobes, and contralateral nodules), especially as it pertains to the associations between the extent of the primary tumor and other prognostic factors of interest (eg, nodal spread and initial treatment ). Materials and Methods This was a retrospective, population-based study encompassing cases registered in the SEER program and made publicly available through online access. Individual patient data were retrieved with the use of the Surveillance Research Program, National Cancer Institute SEER*Stat software ( version 6.3.6, accessed on February 13, The authors had no access to the identity of the patients, and therefore informed consent from the study participants was not deemed necessary. Criteria for Selection of Cases and Data Collection The criteria for selection of cases and data collection are described in online supplemental material A. Stage Grouping and Definition of Subgroups Within Stage IIIB and IV Stage grouping was performed according to the UICC/AJCC TNM staging system, sixth edition. 14 Patients with stage IIIB and IV disease were then subdivided into the subgroups described in Table 1. Statistical Analysis The distribution of overall survival (defined as the time from diagnosis to death) was estimated by the Kaplan-Meier method. The log-rank test was performed to evaluate differences in survival between groups. The Cox proportional hazards model was applied to study the effect of multiple covariates on overall survival. The proportional hazards assumption was examined to study the good- Table 1 Definition of Stage IIIB and IV Subgroups Subgroups T4 satellite T4 invasive T4 pleural effusion M1 ipsilateral M1 contralateral M1 distal Description Separate tumor nodules in the same lobe Extension to the carina, trachea, esophagus, mediastinum, extrapulmonary, major blood vessels (pulmonary artery or vein, superior vena cava, aorta, azygos vein), or nerves (recurrent laryngeal, vagus, phrenic, cervical sympathetic), or extension to the heart or visceral pericardium Malignant pleural effusion or pleural effusion not otherwise specified Separate tumor nodules in different lobes confined to one lung Tumor nodules in both lungs Extrapulmonary metastases or distant lymph node metastases 702 Original Research

3 A T4 satellite C N0 ( E / N = 431 / 1118 ) N1 ( E / N = 125 / 234 ) N2 ( E / N = 531 / 805 ) N3 ( E / N = 86 / 128 ) ness of fit, and no violations were detected. Survival curves are presented in color in online supplemental material B (Figs 1 6), and in black and white within this article (Figs 1-4). Interactions between T4 subtypes (in patients with stage IIIB), and between M1 subtypes (in patients with stage IV disease) and other variables collected were initially evaluated, and in both cases, interactions were detected. This prompted subsequent univariate and multivariate analysis of the effect of these variables on overall survival within the stage IIIB category, stage IV B T4 pleural effusion T4 invasive N0 ( E / N = 2673 / 3308 ) N1 ( E / N = 343 / 413 ) N2 ( E / N = 4660 / 5426 ) N3 ( E / N = 685 / 788 ) N0 ( E / N = 1514 / 2235 ) N1 ( E / N = 407 / 597 ) N2 ( E / N = 3598 / 4803 ) N3 ( E / N = 579 / 772 ) Figure 1. Overall survival curves evaluating the effect of N status within each subgroup of patients with stage IIIB non-small cell lung cancer defined by T4 satellite (A), T4 invasive (B) and T4 pleural effusion (C). A B Ispilateral Contralateral N0 ( E / N = 610 / 1066 ) N1 ( E / N = 152 / 244 ) N2 ( E / N = 1107 / 1470 ) N3 ( E / N = 180 / 239 ) N0 ( E / N = 1045 / 1454 ) N1 ( E / N = 174 / 245 ) N2 ( E / N = 2302 / 2862 ) N3 ( E / N = 663 / 821 ) Figure 3. Overall survival curves evaluating the effect of N status within each subgroup of patients with stage IV non-small cell lung cancer defined by M1 ipsilateral (A) and M1 contralateral (B). category, and within each subgroup defined by T4 subtype and M1 subtype. In the multivariate model, the initial treatment (ie, surgery,, none, or both) was included as a variable but not the type of surgery. For each subgroup of Ipsilateral (E/N = 2049/3019) Contralateral (E/N = 4184/5382) Distal (E/N = 51414/58748) Figure 2. Overall survival curves in patients with stage IV non-small cell lung cancer subdivided into M1 ipsilateral, M1 contralateral, M1 distant. CHEST / 136 / 3/ SEPTEMBER,

4 interest, hazard ratios (HRs) and 95% confidence intervals (CIs) from multivariate analysis are presented in Tables 2-6, as well as median survival time, 2-year, and 5-year survival rates (Tables 7 11 in online supplemental material C). The results of the univariate analysis largely coincided with multivariate analysis and are not presented in this article (except for the type of surgery variable, which could not be fitted in the model for multivariate analysis; in this case, the results of univariate analysis are depicted in Tables 2 6, and in Tables 7 11 in online supplemental material C). Because there were 35% and 41% of patients, respectively, with missing tumor size values for stages IIIB and IV, we repeated all multivariate analyses excluding tumor size as a variable in the models. The results remained similar, and thus only the analysis including tumor size is presented. Two-sided p values 5 for main effects and interaction effects were considered statistically significant. Statistical analyses were conducted using two statistical software packages (SAS, version 9.1; SAS Institute; Cary, NC; and S-PLUS; Insightful Corp; Seattle, WA). Results Patient Characteristics T2b ( E / N = 2351 / 4255 ) Modified T3 ( E / N = 5321 / 8213 ) T4 Satellite ( E / N = 1197 / 2315 ) Figure 4. Overall survival in patients with non-small cell lung cancer defined as T2b (UICC/AJCC 6th edition T2 5cm and 7 cm), modified T3 (UICC/AJCC 6th edition T2 7 cm plus UICC/AJCC 6th edition T3), or T4 satellite (UICC/AJCC 6th edition T4 defined by separate tumor nodules in the same lobe). Based on the patient selection criteria described earlier, 138,063 cases of NSCLC stages I to IV were identified. Of these, 22,091 cases (16%) were classified as stage IIIB and 67,149 cases (49%) were classified as stage IV. Survival Outcome in Patients With Stage IIIB: On multivariate analysis including all patients with stage IIIB disease, age below the median (vs above the median), female gender (vs male gender), Asian Table 2 of Patients With Stage IIIB NSCLC Defined by T4 Satellite Younger than median 1,186 Reference Reference Older than median 1, ( ) 01 Female 1,096 Reference Reference Male 1, (4 1.35) 1 White 1,931 Reference Reference Black (4 1.27) 0.77 Asian ( ) 0.74 Other (5 2.38) 4 Adeno 843 Reference Reference Squamous cell (7 1.51) 1 NOS ( ) 0.16 Large cell ( ) 1 Bronchioloalveolar (0.5 8) 01 Smaller than median 1,495 Reference Reference Larger than median ( ) 01 N0 1,118 Reference Reference N ( ) 01 N ( ) 01 N ( ) 01 No surgery and no 542 Reference Reference Surgery alone ( ) 01 Radiotherapy alone (0.57 ) ( ) 01 No surgery 1,132 Reference Reference Pneumonectomy (2 0.75) 01 Lobectomy (6 0.34) 01 Less than lobectomy 195 ( ) 01 *Univariate analysis. race (vs white race), bronchoalveolar histology (vs adeno), tumor size below the median (vs above the median), T1 (vs T2 to T4), T4 satellite and T4 invasive (vs T4 pleural effusion), N0 (vs N2 and N3), surgery,, or both (vs no surgery and no ) were all associated with better survival outcome (data not shown). As previously observed in chemotherapy clinical trials involving patients with advanced disease, 15 squamous cell histology was associated with decreased sur- 704 Original Research

5 Table 3 of Patients With Stage IIIB NSCLC Defined by T4 Invasive Table 4 of Patients With Stage IIIB NSCLC Defined by T4 Pleural Effusion Younger than median 4,913 Reference Reference Older than median 3, ( ) 01 Female 3,279 Reference Reference Male 5,128 7 (1 1.14) 3 White 6,940 Reference Reference Black 1, (3 1.24) 1 Asian ( ) 4 Other ( ) 6 Adeno 2,009 Reference Reference Squamous cell 3, (8 1.27) 01 NOS 2, (2 1.21) 2 Large cell (2 1.3) 3 Bronchioloalveolar (2 1.57) 4 Smaller than median 2,256 Reference Reference Larger than median 3, ( ) 01 N0 2,235 Reference Reference N (8 1.41) 01 N2 4, ( ) 01 N ( ) 01 No surgery and no 1,970 Reference Reference Surgery alone (0.31 1) 01 Radiotherapy alone 4, (9 0.56) (8 0.37) 01 No surgery 7,058 Reference Reference Pneumonectomy ( 0.52) 01 Lobectomy (0.37 7) 01 Less than lobectomy (1 4) 01 *Univariate analysis. Younger than median 4,077 Reference Reference Older than median 5, ( ) 01 Female 4,208 Reference Reference Male 5, (0.93 6) 0.78 White 8,035 Reference Reference Black 1, (8 7) 0.54 Asian (8 7) 01 Other ( ) 0.36 Adeno 3,989 Reference Reference Squamous cell 2,014 8 ( ) 8 NOS 3, ( ) 01 Large cell ( ) 0.12 Bronchioloalveolar ( ) 0 Smaller than median 2,651 Reference Reference Larger than median 2, ( ) 01 N0 3,308 Reference Reference N (8 1.46) 01 N2 5, ( ) 01 N ( ) 01 No surgery and no 6,307 Reference Reference Surgery alone (9 2) 01 Radiotherapy alone 2,898 3 (0.59 7) (2 1) 01 No surgery 9,455 Reference Reference Pneumonectomy (6 0.5) 01 Lobectomy ( 0.3) 01 Less than lobectomy (5 0.94) 1 *Univariate analysis. vival (data not shown). The effect of T4 subtypes on survival is illustrated by the Kaplan-Meier curves depicted in Figure 1 in online supplemental material B, with a 5-year survival rate of 24% for T4 satellite, 9 to 14% for T13N3M0, 9% for T4 invasive, and only 3% for T4 pleural effusion. Subsequently, we performed multivariate analysis within three subgroups of patients with stage IIIB NSCLC (ie, T4 satellite, T4 invasive, and T4 pleural effusion) [Tables 2 4, and Tables 7 9 in online supplemental material C]. Smaller tumor size, age less than the median, and use of, surgery, or both were associated with better survival across all subgroups. In all subgroups, advanced nodal stage was significantly associated with poor outcome (Fig 2 in online supplemental material B; Fig 1 in this article). In regard to the type of surgical treatment, analysis concerning T4 invasive and T4 satellite subgroups are clinically more meaningful, given that these are the patients that would generally CHEST / 136 / 3/ SEPTEMBER,

6 Table 5 of Patients With Stage IV NSCLC Defined by M1 Ipsilateral Table 6 of Patients With Stage IV NSCLC Defined by M1 Contralateral Younger than median 1,370 Reference Reference (69 yr) Older than median 1, ( ) 01 (69 yr) Female 1,402 Reference Reference Male 1, (9 1.34) 01 White 2,481 Reference Reference Black (7 1.2) 0 Asian ( ) 0.54 Other 16 6 ( ) 8 Adeno 1,153 Reference Reference Squamous cell ( ) 0.14 Not otherwise (6 1.39) 01 specified Large cell ( ) 0.14 Bronchioloalveolar ( ) 1 Smaller than median 1,361 Reference Reference (4.0 cm) Larger than median 1, ( ) 01 (4.0 cm) N0 1,066 Reference Reference N (3 1.54) 2 N2 1, ( ) 01 N ( ) 01 No surgery and no 1,284 Reference Reference Surgery alone ( ) 01 Radiotherapy alone (0.54 9) (0.39 1) 01 No surgery 2,146 Reference Reference Pneumonectomy (0.39 ) 01 Lobectomy (0.3 1) 01 Less than lobectomy ( ) 01 *Univariate analysis Younger than median 2,323 Reference Reference (69 yr) Older than median 3, ( ) 01 (69 yr) Female 2,438 Reference Reference Male 2, (5 1.23) 01 White 4,317 Reference Reference Black (2 1.3) 2 Asian (0.73 1) 5 Other 24 1 (2 1.56) 0.52 Adeno 1,804 Reference Reference Squamous cell 1,225 9 ( ) 0.12 NOS 1, (4 1.27) 01 Large cell ( ) 0.38 Bronchioloalveolar (0.54 5) 01 Smaller than median 1,784 Reference Reference (4.0 cm) Larger than median 1, ( ) 01 (4.0 cm) N0 1,454 Reference Reference N ( ) 4 N2 2, ( ) 01 N ( ) 01 No surgery and no 3,436 Reference Reference Surgery alone ( ) 01 Radiotherapy alone 1, (0.73 7) ( ) 01 No surgery 5,027 Reference Reference Pneumonectomy 18 6 ( ) 0.16 Lobectomy (7 5) 01 Less than lobectomy ( ) 01 *Univariate analysis. be considered potential candidates for resection. For T4 satellite, lobectomy was associated with the best outcome, followed by less than lobectomy and pneumonectomy (Table 2). For T4 invasive, lobectomy was associated with the best outcome, followed by pneumonectomy and less than lobectomy (Table 3). Pneumonectomy was associated with a significant survival benefit when compared to no surgery, in both the T4 satellite and T4 invasive subgroups. Survival Outcome in Patients With Stage IV: On multivariate analysis including all stage IV patients, age below the median (vs above the median), female gender (vs male gender), Asian race (vs white race), adeno and bronchoalveolar histology (vs squamous cell, large cell, and not otherwise specified [NOS]), tumor size below the median (vs above the median), N0 (vs N1 3), M1 ipsilateral and M1 706 Original Research

7 contralateral (vs M1 distant), surgery, or both (vs no surgery and no ) were associated with better survival outcome (data not shown). In Figure 3 in online supplemental material B, and Figure 2 in this article, we present the Kaplan-Meier survival curves according to the subtype of M1. The 5-year overall survival rate for M1 ipsilateral disease was 8%, compared to 4% for M1 contralateral and 1% for M1 distant. We also conducted multivariate analysis within each subgroup of stage IV disease (ie, M1 ipsilateral and M1 contralateral) [Tables 5 and 6 in this article, and Tables 10 and 11 in online supplemental material C]. For patients with M1 ipsilateral, N0 and N1 status conferred better survival as compared to N2 and N3. For patients with M1 contralateral, there was also a statistically significant improvement in survival of N0 to 1 vs N2 to 3 stages, albeit less prominent (Fig 4 in online supplemental material B, and Fig 3 in this article). Regarding treatment, patients in the M1 ipsilateral group or M1 contralateral group had a statistically significant survival advantage if they received and/or surgery (compared to no treatment) [Tables 5 and 6 in this article, and Tables 10 and 11 in online supplemental material C]. In both groups, lobectomy was associated with a better survival outcome, followed by less than lobectomy. Pneumonectomy was still associated with a survival advantage (compared to no treatment) in patients with M1 ipsilateral disease. This was not evident for patients with M1 contralateral disease, but the number of patients was too small (n 18) to allow meaningful comparisons (Tables 5 and 6 in this article). Contrasting Overall Survival of Subgroups Within Stage IIIB and IV With Stages Defined by the UICC/AJCC TNM Classification, Sixth Edition Since the newly defined subgroups of T4 and M1 present distinct survival outcomes, especially when analyzed in conjunction with the nodal stage, we plotted the survival curves of T4 satellite, T4 invasive, T4 pleural effusion, M1 ipsilateral, M1 contralateral, and M1 distal according to nodal status, and contrasted them with the survival curves of stages I to IV defined by the current UICC/AJCC TNM staging system, sixth edition (Fig 5 in online supplemental material B). In this exploratory analysis, the curve for T4 satellite N0, N1, and N2 to 3 were superimposable with the curves for stage IB/ IIA, IIB, and IIIA/IIIB disease, respectively (Fig 5, B, in online supplemental material B). Of note, stages IB/IIA, IIB, and IIIA/IIIB include patients with T2N0, T2N1, and T2N2/T2N3 tumors, respectively. These results indicate that T4 satellite lesions might have a prognosis comparable to UICC/AJCC, sixth edition, T2 lesions, and not T3 lesions (as in the new TNM classification proposed by the IASLC). 11 To further explore this observation and place it in perspective according to the newly IASLC-defined T2b and T3 categories, 11 we plotted the survival curves of our cohort of patients who had been reclassified as having T2b lesions, modified T3 or T4 satellite (Fig 6 in online supplemental material B, and Fig 4 in this article). In this reclassification, T2b was defined as T2 (according to the UICC/AJCC, sixth edition) with size 5cmand 7 cm. Modified T3 lesion was defined as T3 (according to the UICC/AJCC, sixth edition), plus T2 (according to the UICC/AJCC, sixth edition) with size 7 cm. T4 satellite was defined as the presence of separate tumor nodules in the same lobe. Our results demonstrate that the T4 satellite group survival curve overlapped with the T2b group survival curve, and not with the modified T3 curve. The HR for death was (95% CI, to 7; p 0.966) for T4 satellite vs T2b, and (95% CI, 88 to 0.780; p.0001) for T4 satellite vs modified T3. Discussion In this article, we present the most comprehensive analysis performed to date of patients with stage IIIB and IV NSCLC enrolled in the SEER program. The main conclusions of this study were as follows: (1) that patients in the stage IIIB T4 satellite subgroup presented with the best prognosis, with overall survival comparable to those with T2 lesions. The survival of patients with T4 pleural effusion was comparable to that of those with stage IV disease; (2) that patients in the stage IV M1 ipsilateral subgroup presented with the best prognosis. Patients in the M1 contralateral subgroup had worse survival than those in the M1 ipsilateral subgroup, but still better when compared to those in the M1 distant subgroup; and (3) that nodal status remained a strong determinant of survival, particularly for patients with T4 satellite, T4 invasive, M1 ipsilateral, and, to a less extent, M1 contralateral lesions. The survival trends reported here are in accordance with the recent IASLC proposal 10 for the new (seventh edition) TNM classification system, based on a multinational cohort of patients. Besides independently validating the previous IASLC analysis, we demonstrate in this article for the first time the association of other variables (ie, age, gender, size of tumor, nodal status, histology, race, and initial treatment ) with outcome, as determined by multivariate analysis within each subgroup of patients with stage IIIB and IV NSCLC. Most provocwww.chestjournal.org CHEST / 136 / 3/ SEPTEMBER,

8 ative and relevant to clinical practice is the description of the prognostic effect of lymph node status in patients with T4 satellite and M1 ipsilateral tumors, for whom a more aggressive form of definitive treatment might be indicated in the presence of N0 to 1 disease. The survival curves in this setting were comparable to the survival curves of patients with stage IB/IIA (for T4 satellite N0 lesions), stage IIB (for T4 satellite N1 lesions), and stage IIB-IIIA (for M1 ipsilateral N0 to 1 lesions) of the TNM classification system, sixth edition. Of note, IB-IIIA and T4 satellite N0 to 1 patients are currently considered for surgical resection, and perhaps this should also include M1 ipsilateral N0 to 1 patients. This is further illustrated by the fact that patients with T4 satellite and M1 ipsilateral had a HR for death of 2 and 2, respectively, when treated with surgery (compared to no surgery/no ) on multivariate analysis. A fraction of patients with T4 satellite, T4 invasive, and M1 ipsilateral lesions with advanced nodal involvement (N2 to N3) may still experience reasonable 2-year survival rates and should be referred to a multidisciplinary team for consideration for multi therapy as well. Of note, our univariate and multivariate analyses demonstrated a benefit from surgery (even pneumonectomy) and/or in patients with T4 invasive, T4 satellite, and M1 ipsilateral lesions (compared to no surgery/no ). Nonetheless, the effects of the initial treatment on survival presented in this article should be viewed cautiously, given that, in essence, this constitutes a retrospective study, and selection bias toward initial surgery and/or radiation could have occurred. Another important limitation of this study is the fact that for stages I to III and stage IV disease confined to the lung, cases were included in the analysis only if the complete extent of disease and nodal status information was available. The prognosis of these cases (especially M1 ipsilateral and M1 contralateral) may not accurately reflect the outcomes of the broader patient population because patients with better performance status (a known prognostic factor not captured by SEER) were probably more likely to be completely staged, and thus included in this analysis, due to the possibility of delivering more aggressive treatment. Despite these limitations, the revisions in TNM descriptors proposed by the IASLC can be corroborated by our independent evaluation of the cohort of patients enrolled in the SEER program. However, our findings suggest that one exception merits further consideration. T4 satellite tumors, in our analysis, behave similarly to T2b tumors, and not T3 (as proposed by the IASLC). In summary, the present study provides additional information that should be considered when adopting the proposal for the new (seventh) TNM classification system. The full description of the methodology we utilized allows for reproducibility of the results and a better judgment of their applicability by the medical community. Given the resourcefulness of the SEER database in determining survival outcome, we encourage the continuing improvement of this program and its use as a pivotal tool in reassessing and redeveloping the staging system for NSCLC and other tumor types. Acknowledgments Author contributions: Drs. William, Lin, Lee, and Kim participated in the study s conception and design, collection and assembly of data, data analysis and interpretation, and writing and final approval of the manuscript. Drs. Lippman and Roth participated in data analysis and interpretation, writing, and final approval of the manuscript. Financial/nonfinancial disclosures: The authors have reported to the ACCP that no significant conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article. References 1 Mountain CF. Revisions in the International System for Staging Lung Cancer. Chest 1997; 111: Bryant AS, Pereira SJ, Miller DL, et al. Satellite pulmonary nodule in the same lobe (T4N0) should not be staged as IIIB non-small cell lung cancer. Ann Thorac Surg 2006; 82: Port JL, Korst RJ, Lee PC, et al. Surgical resection for multifocal (T4) non-small cell lung cancer: is the T4 designation valid? Ann Thorac Surg 2007; 83: Trousse D, Barlesi F, Loundou A, et al. Synchronous multiple primary lung cancer: an increasing clinical occurrence requiring multidisciplinary management. J Thorac Cardiovasc Surg 2007; 133: Trousse D, D Journo XB, Avaro JP, et al. Multifocal T4 non-small cell lung cancer: a subset with improved prognosis. Eur J Cardiothorac Surg 2008; 33: Tsunezuka Y, Matsumoto I, Tamura M, et al. The results of therapy for bilateral multiple primary lung cancers: 30 years experience in a single centre. Eur J Surg Oncol 2004; 30: Tung YW, Hsu CP, Shai SE, et al. Surgical feasibility of ipsilateral multifocal non-small cell lung cancer in different lobes: excellent survival in node-negative subgroup. Eur J Cardiothorac Surg 2003; 24: Yano M, Arai T, Inagaki K, et al. Intrapulmonary satellite nodule of lung cancer as a T factor. Chest 1998; 114: Goldstraw P, Crowley JJ. The International Association for the Study of Lung Cancer International Staging Project on Lung Cancer. J Thorac Oncol 2006; 1: Goldstraw P, Crowley J, Chansky K, et al. 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. J Thorac Oncol 2007; 2: Original Research

9 11 Rami-Porta R, Ball D, Crowley J, et al. 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. J Thorac Oncol 2007; 2: Postmus PE, Brambilla E, Chansky K, et al. 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. J Thorac Oncol 2007; 2: Groome PA, Bolejack V, Crowley JJ, et al. 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. J Thorac Oncol 2007; 2: American Joint Committee on Cancer. AJCC cancer staging manual. 6th ed. New York, NY: Springer, Scagliotti GV, Parikh P, von Pawel J, et al. Phase III study comparing cisplatin plus gemcitabine with cisplatin plus pemetrexed in chemotherapy-naive patients with advancedstage non-small-cell lung cancer. J Clin Oncol 2008; 26: CHEST / 136 / 3/ SEPTEMBER,