The TNM system for the classification of malignant tumors was developed by Pierre Denoix, a surgeon at the Institute Gustave Roussy in Paris, and published in a series of articles between 1943 and 1952.⁶ The following year, 1953, this system was adopted by the recently formed UICC Committee on Tumour Nomenclature and Statistics under the auspices of the League of Nations. The UICC process at this time developed proposals by consensus between experts in the field. These proposals were disseminated in a series of 23 brochures or fascicles published between 1960 and 1967, which covered the TNM classification of cancers in 23 sites, lung being included in the brochure published in 1966. Subsequently, the recommendations were brought together in the 1st edition of the UICC TNM Classification of Malignant Tumours published in 1968 in a compact Livre de Poche format.⁷ The proposals were for “trial” from 1967 to 1971. At that time, there was insufficient information on lung cancer to merit a section of its own, and it was listed under “other sites.” The T descriptors included T0 for cases in which there was no evidence of the primary tumor, T1 for tumors confined to a segment or segmental bronchus, T2 where there was lobar involvement, T3 if there was involvement of more than one lobe, and T4 for tumors extending beyond the lung. The N descriptors were NX, N0, or N1, this last category being applied if intrathoracic nodes were involved. Intrathoracic nodes were described as hilar or peripheral with no mention of nodes in the mediastinum. The M1 descriptor was subdivided into M1a in which there was a pleural effusion with malignant cells present, M1b cases with palpable cervical nodes, and M1c for cases in which other distant sites were involved. Stage groupings were not proposed at that time.

The AJCC, formed in 1959 as the American Joint Committee (AJC) for Cancer Staging and End Results Reporting, developed a separate and distinctive process in which “Task Forces” were set up to gather data, which were used to inform its proposals. There was clearly a possibility that these two organizations would make different, and possibly conflicting, recommendations to the cancer community. Therefore, at a series of meetings between the UICC and the AJC, a rapprochement was reached, which ensured that these two organizations would not produce further recommendations without consultation between themselves and other National TNM Committees and International nongovernmental professional organizations.

In 1973, the Task Force on Lung Cancer of the AJC accepted proposals from Dr. Mountain, Dr. Carr, and Dr. Anderson for “A Clinical Staging System for Lung Cancer.”⁸ This was based on data from 2155 cases of lung cancer, of which 1712 were cases of non-small cell lung cancer (NSCLC), diagnosed at least 4 years before analysis. The majority of the T, N, and M descriptors in use today were introduced at that time, including the impact on T category of such features as the 3-cm cutoff between T1 and T2 tumors, the bronchoscopic extent of disease, the extent of atelectasis/consolidation of the lung parenchyma, and the invasion of chest wall, diaphragm, or mediastinum. The T categories 0 to 3 were retained but T4 was dropped, N categories 0 to 1 were retained but N2 was added to address the issue of mediastinal node involvement, and M categories 0 to 1 were retained. Pleural effusions were removed from the M1 category and became a T3 descriptor.
The resultant TNM subsets were grouped into stages I to III. Four of the possible eighteen TNM subsets had too few cases for analysis and seven others contained less than 100 cases. Survival curves showed distinct differences between prognosis in overall T, N, and M categories and the three-stage groupings to 5 years and beyond. A table showed the differing survival at 12 and 18 months for those TNM subsets for which data was available. No assessment of statistical significance was presented, and there was no validation of the individual descriptors. These proposals were incorporated in the 2nd edition of the UICC TNM Classification of Malignant Tumours published in 1975⁹ and the 1st edition of the Manual for Staging of Cancer published by the AJC in 1977.¹⁰

The 3rd edition of the UICC manual, published in 1978¹¹ and revised in 1982, further divided stage I into Ia and Ib (note that at that time, stage subgroups were lowercase) and established stage IV for cases with M1 disease. The “x” descriptor, erratically applied to some categories in earlier editions, was, for the first time, introduced as an option in all three categories of T, N, and M.

The American committee, now the AJCC, did not make these changes in its 2nd edition, which was published in 1983.¹²

By 1986, Dr. Mountain had assembled a new database containing 3753 cases of lung cancer with a minimum follow-up of 2 years. The proposals from this source were accepted by the AJCC, and subsequently by the UICC and cancer committees in Germany and Japan, creating “A new International Staging System for Lung Cancer.”¹³ The recommendations were published in the 4th edition of the UICC TNM Classification of Malignant Tumours in 1987¹⁴ and in the 3rd edition of the American manual in 1988.¹⁵ Changes proposed in this edition include the addition of “visceral pleural invasion” as a T2 descriptor, the designation of superficial tumors limited to the bronchial wall as T1 irrespective of location, a recommendation that the occasional pleural effusion that was cytologically negative could be ignored in defining the T category, the reemergence of the T4 category, and the creation of an N3 category. The existing T3 descriptors were split between T3 and the new T4 category on the basis that the former would retain those descriptors that indicated that such tumors were “candidates for complete resection,” whereas the latter would be “inoperable.” The previous descriptor of mediastinal invasion was split into its component parts, with invasion of the mediastinal pleura or pericardium remaining T3, whereas invasion of the great vessels, heart, trachea, esophagus, carina, and vertebral bodies became T4 descriptors, along with the presence of a pleural effusion. The situation was confused by additional definitions of T3 and T4 given in the text. Those tumors with “limited, circumscribed extrapulmonary extension” were to be retained within the T3 category, whereas those with “extensive extrapulmonary extension” became T4. These conflicting definitions resulted in a lack of clarity as to whether tumors invading such structures as the pericardium remained T3 if there was extensive invasion and were considered inoperable or became T4, or if invasion limited to a circumscribed area of the esophagus and resected completely at surgery should be considered to be T3 or T4. Metastases to the ipsilateral mediastinal nodes and subcarinal nodes remained within the N2 category, and the new N3 category was added to accommodate metastases to the contralateral mediastinal nodes, contralateral hilum or ipsilateral, and contralateral supraclavicular or scalene lymph nodes. Additional changes in that edition include the moving of T1N1M0 cases from stage I to stage II and the division of stage III into IIIA (containing T3 and N2 cases) and IIIB (containing T4 and N3 cases). Once again, a table showed the differing survival prospects for TNM subsets, and a graph showed statistically significant survival differences between stage groupings. No validation was presented for the individual descriptors or to substantiate the movement of some into T3 and others T4.

The AJCC made no changes in the classification for lung cancer in its 4th edition published in 1992.¹⁶

At the time of the next revision in 1997, the database of Dr. Mountain has increased to include 5319 cases, all but 66 being NSCLC, 4351 cases treated at the MD Anderson Cancer Center between 1975 and 1988, and 968 cases referred there from the National Cancer Institute Cooperative Lung Cancer Study Group for confirmation of stage and histology.¹⁷ Tables showed statistically significant differences in survival as far as 5 years between clinical/evaluative cTNM categories and pathological/postsurgical pTNM categories T1N0M0 and T2N0M0 and these were divided into a new stage IA and stage IB, respectively. Similarly, T1N1M0 cases were placed in a new stage IIA, and T2N1M0 and T3N0M0 cases became stage IIB. The remaining TNM categories in stages IIIA, IIIB, and IV remained unchanged although statistically significant differences were found between some TNM categories. An additional paragraph determined that “the presence of satellite tumor(s), not lymph nodes, within the primary-tumor lobe of the lung should be classified as T4. Intrapulmonary ispilateral [sic] metastasis in a distant, that is, nonprimary lobe(s) of the lung, should be classified M1.”¹⁷ No data was presented to support these suggestions and the wording used to describe such additional pulmonary nodules was loaded to underline the apparent logic of considering some to be “satellite” lesions and, therefore, a T descriptor, whereas those in other lobes were a “metastasis” and, therefore, an M descriptor.

These recommendations were accepted by the AJCC and the UICC-TNM Prognostic Factors Project Committee and appeared in the 5th edition of their publications in 1997.¹⁸,¹⁹

There were no changes in the lung cancer classification in the 6th edition of TNM Classification of Malignant Tumours published in 2002.²⁰,²¹

Undoubtedly, the lung cancer community owes an enormous debt of gratitude to the pioneers of TNM, especially to Dr. Clifton Mountain. However, over the last decade, there has grown a feeling among lung cancer clinicians and scientists that changes were needed to the process for revision of the TNM Classification for Lung Cancer. The Mountain database, which
had been the major source of data to inform revisions of the TNM system up to and including the 6th edition had, by 1996, enlarged to include 5319 cases. Thus, it was a relatively small database, accumulated over 20 years, during which period many advances had been made in the techniques available for pretreatment staging, most noticeably the routine application of computed tomography (CT) scanning. The great majority of cases in this database had been referred for surgical treatment and had been recruited from a single center. There were understandable concerns as to whether the recommendations emanating from such a database were historically valid, globally relevant, and appropriate for evaluating treatment by nonsurgical or combined modality care. Oncologists treating small cell lung cancer (SCLC) had abandoned TNM for all, except those very limited cases in which surgery was considered and instead were using a simpler classification based on the single distinction between “limited” or “extensive” disease.²² Even when used in a surgical setting and for NSLC, the lack of validation in previous editions of the TNM Classification of Malignant Tumours had led to many of the descriptors being increasingly challenged. Data had been published suppor ting size cutoffs other than the 3-cm limit separating T1 and T2 tumors, ranging from less than 1 cm to more than 9 cm. “Irresectable” T4 tumors had been resected with good results in selected cases. The descriptors applied to cases in which there were additional tumor nodules in the lobe of the primary and other ipsilateral lobes were generally regarded as harsh. Oncologists had long treated cases with pleural effusion, the “wet” IIIB cases, with the therapeutic strategies used for patients with metastatic, stage IV disease. Clearly, if the TNM classification was to retain its central role in the day-to-day care of patients with lung cancer in an evidence-based era, its recommendations had to be intensively validated and the process for change had to be modernized to make the staging system fit for purpose. These concerns crystallized at an International Association for the Study of Lung Cancer (IASLC) workshop on “Intrathoracic Staging” held at the Brompton Hospital in London in October 1996.²³ One of the published recommendations of this meeting was “the establishment by the IASLC of a staging committee.”

At the 8th World Conference on Lung Cancer in Dublin in 1997, the board of the IASLC considered a submission from the late Dr. RJ Ginsberg and Dr. P Goldstraw to create an international staging committee. The board agreed that the IASLC, as the only global organization dedicated to the study of lung cancer, representing all clinical and research aspects of lung cancer care, had a responsibility to become involved in the revision process and to develop a new database to inform future revisions. At its next meeting in December 1998, the board agreed to provide pump-priming funds for such a project. Meetings were held in London in 1999 and 2000, during which the composition of the committee was developed to ensure speciality and geographical representation and the involvement of stakeholders such as the UICC, the AJCC, and the joint Japanese societies involved in the study of lung cancer. At the next World Conference in 2000, collaboration was established with colleagues from Cancer Research and Biostatistics (CRAB), a not-for-profit medical statistics and data management organization based in Seattle with extensive experience with multicenter data collection and analysis. At that meeting, sufficient funds were guaranteed from the pharmaceutical industry to allow a major meeting in London in 2001, to which database proprietors were invited to present an outline of the data they held. Over the 2-day workshop, data on 80,000 cases were presented from 20 databases across the globe. In was decided to estimate the budget based on the assumption that 30,000 suitable cases could be recruited and that the length of the project would be the 5-year cycle used by the UICC and AJCC at that time. Cases would be solicited from databases worldwide, treated by all modalities of care, between 1990 and 2000, a period during which there had been relative stability in staging methods. This would ensure a 5-year follow-up by the time of analysis. In collaboration with CRAB, the data fields and data dictionary were finalized. Later that year, full funding was obtained by the IASLC via a partnership agreement with the pharmaceutical industry.

Meetings continued to be held on an annual basis utilizing the world conferences, now held biennially, wherever possible.

The UICC was well aware of the need to update the revision process and, around this time, established a “TNM Process Subcommittee. Criteria were established for instituting changes to the TNM classification and for the evaluation of proposals for such changes.”²⁵ A system for continuous monitoring of the proposals in the literature, a “literature watch,” was set up. For each cancer type, a “TNM expert panel” was established to review the sifted literature and help in the evaluation of proposals for change. In May 2003, the UICC and AJCC extended the revision cycle to 7 years, which resulted in publication of the 7th edition of the TNM Classification of Malignant Tumours being deferred until 2009. The internal review processes within these organizations required that the IASLC submitted its proposals to the UICC in January 2007 and the AJCC in June 2008.