肺癌分期（2009）

DOI 10.1378/chest.08-0978 2009;136;260-271Chest Frank C. Detterbeck, Daniel J. Boffa and Lynn T. Tanoue The New Lung Cancer Staging System http://www.chestjournal.org/content/136/1/260.full.html and services can be found online on the World Wide Web at: The online version of this article, along with updated information ) ISSN:0012-3692http://www.chestjournal.org/site/misc/reprints.xhtml( of the copyright holder. may be reproduced or distributed without the prior written permission Northbrook IL 60062. All rights reserved. No part of this article or PDF by the American College of Chest Physicians, 3300 Dundee Road, 2007Physicians. It has been published monthly since 1935. Copyright CHEST is the official journal of the American College of Chest Copyright © 2009 American College of Chest Physicians at 3rd Military Medical University on July 13, 2009www.chestjournal.orgDownloaded from The New Lung Cancer Staging System Frank C. Detterbeck, MD, FCCP; Daniel J. Boffa, MD; and Lynn T. Tanoue, MD, FCCP The International Association for the Study of Lung Cancer (IASLC) has conducted an extensive initiative to inform the revision of the lung cancer staging system. This involved development of an international database along with extensive analysis of a large population of patients and their prognoses. This article reviews the recommendations of the IASLC International Staging Committee for the definitions for the TNM descriptors and the stage grouping in the new non-small cell lung cancer staging system. (CHEST 2009; 136:260–271) Abbreviations: AJCC � American Joint Committee on Cancer; IASLC � International Association for the Study of Lung Cancer; NSCLC � non-small cell lung cancer; SEER � Surveillance, Epidemiology, and End Results; UICC � Union Internationale Contre le Cancer D efinition of the stage is an essential part of theapproach to patients with cancer, and it has led to the development of a universally accepted stage classification systems for most tumors. The Union Internationale Contre le Cancer (UICC) and the American Joint Committee on Cancer (AJCC) serve as the official bodies that define, periodically review, and refine the stage classification systems. The 6th edition of the staging system was published in 2002,1 and the 7th edition will be published in 2009. In preparation for this, much work has been done by the International Association for the Study of Lung Cancer (IASLC) to recommend changes that are based on a large international database and are backed by careful validation and statistical analysis. This article reviews the development and final rec- ommendations of the IASLC International Staging Committee for non-small cell lung cancer (NSCLC), which have been accepted by the UICC and AJCC for the new edition. For editorial comment see page 6 The TNM staging system for lung cancer dates back to an initial proposal by Dr. Clifton Mountain that was adopted by the AJCC in 1973 and by the UICC in 1974. The original system was based heavily on intuition with limited corroboration from a database of 2,155 patients from the MD Anderson Cancer Center in Houston, TX. Subsequent revi- sions of the TNM staging system continued to be based on this database, which grew to include 5,319 cases at the time the lung cancer staging system was last revised in 1997. The limitations of this system are that it was based on what was essentially a single institution series, included a limited number of patients (so that many subgroups were quite small), spanned a long time frame, and was weighted some- what toward surgically treated patients by the nature of the database. Nevertheless, this early work laid a significant foundation and defined a staging system that has held up very well even in comparison with the new IASLC staging system. Fundamentals of the UICC/AJCC Staging System The NSCLC stage classification is based on the TNM system, which is used for most cancers. The T descriptor defines the extent of the primary tumor, From Thoracic Surgery (Drs. Detterbeck and Boffa), Section of Pulmonary and Critical Care Medicine (Dr. Tanoue), Yale University School of Medicine, New Haven, CT. 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. Manuscript received April 11, 2008; revision accepted February 11, 2009. Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (www.chestjournal. org/site/misc/reprints.xhtml). Correspondence to: Frank C. Detterbeck, MD, FCCP, Professor and Section Chief, Thoracic Surgery, Yale University School of Medicine, 330 Cedar St, BB 205, New Haven, CT 06520-8062; e-mail: frank.detterbeck@yale.edu DOI: 10.1378/chest.08-0978 CHEST Special Feature 260 Special Feature Copyright © 2009 American College of Chest Physicians at 3rd Military Medical University on July 13, 2009www.chestjournal.orgDownloaded from the N descriptor the extent of involvement of re- gional lymph nodes, and the M descriptor the extent of spread to distant sites. The staging system is based solely on the anatomic extent of disease. Other factors, such as clinical symptoms or molecular bio- logical characterization of the tumor, have not been included. Increasing T status reflects tumors that are larger or invasive into more peripheral (ie, visceral pleura, chest wall) or more central structures (ie, lobar or mainstem bronchus, mediastinum). In lung cancer, nodal staging depends on the location of involved nodes (as opposed to the number of nodes). The M descriptor defines the presence or absence of more distant metastatic disease. The method of staging has a major impact on the prognostic implications of the stage classification, a fact that is well recognized by the UICC and AJCC as shown in Table 1. The two most commonly encountered types of stage assessment are clinical staging (the stage determined using all information available prior to any treatment) and pathologic staging (determined after a resection has been car- ried out). The extent of clinical staging can vary from a clinical evaluation alone (history and physical ex- amination) to extensive imaging (CT/PET scans) or invasive staging techniques. It must be emphasized that a surgical staging procedure (such as mediasti- noscopy) is still part of clinical staging because surgical resection as a treatment has not taken place. Clinical stage is denoted by the prefix “c” and pathologic stage by the prefix “p.” The UICC also defines a classification system for the presence or absence of residual tumor after treatment, as shown in Table 2. Typically this is applied to describe the completeness of a surgical resection. Development and Methodology of the IASLC Staging System A proposal to develop an international effort to inform a future revision of the TNM staging classi- fication for lung cancer originated in 1996 at a workshop sponsored by IASLC. An international committee was established and work began in 1999. An unrestricted grant from Eli Lilly and Company enabled the establishment of a database (Eli Lilly and Company played no role in the data collection, analysis, or recommendation development process). The database was developed in cooperation with Cancer Research and Biostatistics (Seattle, WA), which is an independent scientific foundation and the statistical center for the Southwest Oncology Group. Data elements and definitions were finalized in October 2002. Data were collected from multiple sources and sites around the globe. Committees were formed to analyze the data and develop recom- mendations (including validation and methodology, T, N, M descriptors, nodal chart, prognostic factors, and small cell lung cancer). The initial recommen- dations were revised and approved by the full IASLC International Staging Committee. These proposals were published in a series of detailed articles in the Journal of Thoracic Oncology in 2007.2–6 The pro- posed staging recommendations were presented to the AJCC and UICC in 2007, were approved by these bodies, and are slated to be published in the 7th edition of the UICC Staging Manual in 2009. At the time the database was closed to additional entries, 100,869 cases had been submitted from 45 sources in 20 countries.7 The final data set involved 81,015 cases after exclusion of ineligible cases (due to incomplete information in 42%, outside the 1990 to 2000 time frame in 28%, unknown histology in 13%, incomplete survival data in 6%, recurrent cases in 6%, and ineligible histologic types in 6% [carci- noid, sarcoma, and others]).5,7 Small cell lung cancer accounted for 16% and NSCLC for 84% of cases. Only the NSCLC cases were used to derive the T, N, M descriptors and stage groupings reviewed in this article. SCLC and carcinoid tumor staging are ad- dressed in separate publications.8,9 The database included cases from four continents (the proportion of NSCLC cases is as follows: Europe, 58%; North America, 21%; Asia, 14%; and Australia, 7%). Sub- mitted cases came from series (40%), registries (30%), and clinical trials (30%). Treatment involved Table 1—Types of Staging Assessments Prefix Name Definition c Clinical Prior to initiation of any treatment, using any and all information available (eg, including mediastinoscopy) p Pathologic After resection, based on pathologic assessment y Restaging After part or all of the treatment has been given r Recurrence Stage at time of a recurrence a Autopsy Stage as determined by autopsy Table 2—Residual Tumor After Treatment Symbol Name Definition R0 No residual No identifiable tumor remaining; negative surgical margins R1 Microscopic residual Microscopically positive margins but no visible tumor remaining R2 Gross residual Gross (visible or palpable) tumor remaining www.chestjournal.org CHEST / 136 / 1 / JULY, 2009 261 Copyright © 2009 American College of Chest Physicians at 3rd Military Medical University on July 13, 2009www.chestjournal.orgDownloaded from surgery only in 41%, radiotherapy only in 11%, chemotherapy only in 23%, and combined modalities in the rest. Clinical staging included patients subse- quently treated with neoadjuvant (preoperative) therapy, whereas preoperatively treated patients were excluded from analyses of pathologic staging. Subsets of the NSCLC patients were defined that met the analytic requirements for each T, N, M analysis. These subsets were substantially smaller than the entire group of NSCLC patients within the IASLC database. For example, clinical and patho- logic T stage involved 5,760 and 15,234 cases, re- spectively, that wereM0 and included a complete set of cTNM or pTNM as well as sufficient T descriptor details.2 For clinical and pathologic N stage, 38,265 and 28,371 cases, respectively, were selected that were cM0 with sufficient N status assessment (any T).3 For M stage, 6,596 cases were included that were cM1 or had an additional nodule in a separate ipsilateral lobe.4 The definition of stage grouping involved 13,267 clinical stage and 16,937 pathologic stage patients.5 The major determinant chosen for development of subgroups of T, N, and M descriptors as well as the stage groupings was the overall survival, based on the best stage (pathologic, if available; otherwise clini- cal). Regrouping of stage descriptors was allowed, sacrificing backward compatibility with the previous TNM system in favor of greater simplicity. Stage grouping schemes were developed using a recursive partitioning and amalgamation algorithm, and vali- dated as described later. Care was taken to prevent a deviant subset involving unstable curves or small Table 3—Definitions for T, N, M Descriptors Descriptors Definitions Subgroups* T Primary tumor T0 No primary tumor T1 Tumor � 3 cm,† surrounded by lung or visceral pleura, not more proximal than the lobar bronchus T1a Tumor � 2 cm† T1a T1b Tumor � 2 but � 3 cm† T1b T2 Tumor � 3 but � 7 cm† or tumor with any of the following‡: Invades visceral pleura, involves main bronchus � 2 cm distal to the carina, atelectasis/obstructive pneumonia extending to hilum but not involving the entire lung T2a Tumor � 3 but � 5 cm† T2a T2b Tumor � 5 but � 7 cm† T2b T3 Tumor � 7 cm; T3�7 or directly invading chest wall, diaphragm, phrenic nerve, mediastinal pleura, or parietal pericardium; T3Inv or tumor in the main bronchus � 2 cm distal to the carina§; T3Centr or atelectasis/obstructive pneumonitis of entire lung; T3Centr or separate tumor nodules in the same lobe T3Satell T4 Tumor of any size with invasion of heart, great vessels, trachea, recurrent laryngeal nerve, esophagus, vertebral body, or carina; T4Inv or separate tumor nodules in a different ipsilateral lobe T4Ipsi Nod N Regional lymph nodes N0 No regional node metastasis N1 Metastasis in ipsilateral peribronchial and/or perihilar lymph nodes and intrapulmonary nodes, including involvement by direct extension N2 Metastasis in ipsilateral mediastinal and/or subcarinal lymph nodes N3 Metastasis in contralateral mediastinal, contralateral hilar, ipsilateral or contralateral scalene, or supraclavicular lymph nodes M Distant metastasis M0 No distant metastasis M1a Separate tumor nodules in a contralateral lobe; M1aContr Nod or tumor with pleural nodules or malignant pleural dissemination� M1aPl Dissem M1b Distant metastasis M1b Special situations TX, NX, MX T, N, or M status not able to be assessed Tis Focus of in situ cancer Tis T1§ Superficial spreading tumor of any size but confined to the wall of the trachea or mainstem bronchus T1SS *These subgroup labels are not defined in the IASLC publications2–5 but are added here to facilitate a clear discussion. †In the greatest dimension. ‡T2 tumors with these features are classified as T2a if � 5 cm. §The uncommon superficial spreading tumor in central airways is classified as T1. �Pleural effusions are excluded that are cytologically negative, nonbloody, transudative, and clinically judged not to be due to cancer. 262 Special Feature Copyright © 2009 American College of Chest Physicians at 3rd Military Medical University on July 13, 2009www.chestjournal.orgDownloaded from numbers of patients from driving a recommendation, unless the effect was consistent and externally vali- dated in a population-based series. The proposed staging revisions were both inter- nally and externally validated.6 Internal validation required that differences in outcomes among the proposed subgroups were consistently observed within each type of database and geographic region, as well as the time period (from 1990 to 1995 vs from 1995 to 2000) for stage groupings.2,5 Furthermore, for the T descriptor size categories and the stage groupings, the cases used were separated into a randomly selected training set involving approxi- mately two thirds of the cases and a validation set involving the rest.2,5,6 In each analysis this validation confirmed the training set results. Finally, the pro- posed descriptors were externally validated against the Surveillance, Epidemiology and End Results (SEER) database from the same time period, con- trolling for cell type, gender, age, and region. Some details of the analyses are worth noting. The size criteria for the T descriptor are quite robust and broadly applicable, having been extensively vali- dated. The optimal size cutpoints were initially de- termined using a pN0 R0 learning set (rounded to the nearest whole centimeter).2 This was then tested in a pN0 R0 validation set as well as in other node stage groups and in incompletely resected patients. The analysis was also confirmed within different histologic types of NSCLC, among different types of databases and geographic regions. In contrast, the analysis of the impact of additional nodules in the primary tumor lobe or in a different ipsilateral lobe is derived from a potentially skewed population, namely primarily resected patients from Asia and Australia. Nevertheless, the results were supported by data from other geographic regions and by SEER data.2 The N status analysis involved a large number of patients and was consistently a major prognostic Figure 1. Prognosis according to size category. A: overall survival by tumor size for patients with cT1-3N0M0 tumors using the IASLC classification. Modified from Rami-Porta et al.2 B: overall survival by tumor size for patients with pT1-3N0M0 R0 tumors using the IASLC classification. Modified from Rami- Porta et al.2 Centr � central; Inv � invasion; MST � median survival time (months); 5-year � 5-year overall survival. Figure 2. Prognosis according to additional nodules, T4 invasion, and pleural dissemination. A: overall survival for patients with cT3,4/cM1a status due to additional tumor nodules using the IASLC classification (any cN), compared with other categories of T3 and T4. Modified from Rami-Porta et al.2 B: overall survival for patients with pT3,4/pM1a status due to additional tumor nodules using the IASLC classification (any pN, any R), compared with other catego- ries of T3 and T4. Modified from Rami-Porta et al.2 Add’l � additional; Pts � patients. See the legend of Figure 1 for abbrevia- tions not used in the text. www.chestjournal.org CHEST / 136 / 1 / JULY, 2009 263 Copyright © 2009 American College of Chest Physicians at 3rd Military Medical University on July 13, 2009www.chestjournal.orgDownloaded from factor in all regions and database types. However, survival was consistently better in each N category in Asian/Australian patients, intermediate in North American, and worse in European patients.6 The M descriptor analysis was almost exclusively based on clinically staged patients for distant metastases and contralateral lung nodules, but 34% of patients with pleural dissemination and 53% of patients with additional nodules in a different ipsilateral lobe were pathologically staged (ie, resected) patients.4 T Descriptor The IASLC International Staging Committee an- alyzed the effect of tumor size in detail, in part because there were many patients in whom reliable information was available. The size threshold of 3 cm was confirmed as a significant cutpoint and retained as a definition of a T1 vs a T2 tumor. In addition, significant cutpoints were identified at 2, 5, and 7 cm. Tumors � 7 cm had survival that tracked with other definitions of T3 (ie, invasion or central loca- tion) and are therefore placed within this group. The 2-cm and 5-cm additional cutpoints were addressed by the definition of subgroups of T1 (T1a and T1b) and T2 (T2a and T2b). The survival differences between each of these size subgroups were highly statistically significant in pathologically staged pa- tients, regardless of pN status or R status. Among clinically staged patients, survival differences be- tween cT1a, cT1b, and cT2a were not consistently statistically significant, probably due to the smaller numbers of patients for whom these data were available. Table 3 summarizes the definitions of the T descriptor subgroups. The survival curves for the size categories of the T descriptor for clinically and pathologically staged patients are shown in Figure 1. The prognostic value of additional tumor nodules was also investigated. (Nodules were counted as synchronous primary cancers only if they were of different histologic types or were explicitly defined as synchronous primary cancers in the source data- base.) Patients with additional satellite nodules in the same lobe as the primary tumor had relatively good survival, similar to T3 patients, and therefore are now classified as T3 (previously classified as T4). The survival of these patients (T3Satell) was statistically significantly better than patients with T4 tumors because of invasion of major mediastinal structures (T4Inv). However, the prognosis of these patients (T4Inv)