Comparing Treatment Outcomes: The Origins of the ‘Staging’ Concept
The early 1920s laid the groundwork for a number of schemes, which used the anatomical spread of cervical cancer as the basis for classification. German-born Henry Schmitz, Professor of Obstetrics and Gynecology at Loyola university in chicago, was the first to suggest a system designed to mimic the ‘natural history’ of the disease.111 Schmitz proposed to divide the malignancies of the cervix into four groups representing the progressive growth of the tumour. in Group P1 the malignancy was localized within the cervix and it did not exceed one centimetre in diameter. Group P2 indicated a growth that had extended to the periphery of the cervix in a longitudinal or transversal direction. The uterus was less mobile owing to the decreased elasticity of the paracervical tissues. Group P3 meant that either one or both parametria or the regional lymph nodes had been invaded. Group P4 included tumours with absolute fixation of the uterus.112
Schmitz’s classification proved popular in the united States. it was in use at the Mayo Clinic from 1923, and after that it was rapidly accepted at a number of other centres.113 In the meantime the classification of cervical cancer had become a subject of frequent discussion at the meetings of the German Gynaecological Society. At the Society’s Heidelberg Congress in 1923, Winter, Doderlein and others proposed a new scheme which classified cervical tumours into four groups (operable, borderline, inoperable and hopeless) according to the extent of the growth and the mobility of the uterus.114 The problem with this system was that it was still based on criteria of ‘operability’. As pointed out earlier, these differed between surgeons, and they were not applicable to radiotherapy.
In 1928 the Cancer Commission of the League of Nations set up a Radiological Subcommission to investigate the radium therapy of cancer, with special reference to cancer of the uterus.115 One of the purposes of the Subcommission was to collate statistical data on the results of treatment. The desire for uniform statistics led to the formation of a small ‘classification committee’ consisting of epidemiologist Janet Lane-Claypon, physicist Friedrich Voltz, and radiation expert Antoine Lacassagne. Under the chairmanship of James Heyman, the committee produced a system to enable inter-institutional comparison of results. The scheme, later known as the League of Nations Classification for Cervical Cancer, maintained the four-group subdivision common to other classification systems, but it avoided using subjective criteria of
‘operability’ as the differentiating characteristic. It allocated cases to four different groups, based on the anatomical spread of the growth, and the mobility of the uterus.116 Members of the committee were guided by four main considerations. First, the definitions of the different stage- groups should be as simple and precise as possible. Second, the rules for allocating cases should be easily interpreted. Third, one stage should be sufficiently differentiated from the other by characteristics easily recognized on clinical examination. Fourth, each group should contain a number of cases large enough to be of statistical value.
One of the problems with inter-institutional comparisons was that many clinics did not receive random, unselected samples of patients. This meant that assessing results in terms of ‘overall’ cure rates (that is to say, the proportion of ‘cured’ patients as a percentage of either (a) the total number examined with a view to treatment, whether they are treated or not; or (b) the total number actually treated) might produce a misleading picture of the efficacy of different therapies. The committee thus decided to use ‘stage-rates’, defined as the percentage of women with a certain stage of carcinoma who were still alive five years after treatment.117
Members of the Radiological Subcommission hoped that their classification system would meet with international agreement, but its acceptance and widespread use were slow to materialize. The main problem in the early 1930s was that most of the North American clinics were reluctant to give up the Schmitz system. Efforts to publicize the League of Nations’ scheme resulted in the publication of an annual report, presenting overall five-year survival rates by stage.118 The first Annual Report was published in 1937 with Heyman as Editor in Chief (he was to fill this role until 1956, when he was succeeded by his former pupil and close friend Hans-Ludwig Kottmeier). Six participating European centres had agreed to supply the data: the Centre for Tumours at Brussels University; the Liverpool Radium Institute; the Marie Curie Hospital in London; the Radium Centre for Carcinoma of the Uterus of the London County Council; the Paris Radium Institute; and the Radiumhemmet. The release of the first report was followed by the publication of an Atlas illustrating the division of cervical cancer into four stages. Written by Heyman in collaboration with M. Strandquist, the Atlas was styled as a vademecum that the busy practitioner could carry in his or her pocket for quick reference. It contained definitions, staging diagrams and descriptive text in English, French and German.119
In A Woman’s Disease: the History of Cervical Cancer, historian Ilana Lowy writes that ‘thanks to the introduction of a uniform system of classification of cervical cancers, doctors were able to compare the results of treatments of this disease’.120 In reality this comparative analysis remained problematic, because the data gathered were not recorded in a uniform manner. Despite the classification committee’s best efforts, inconsistencies in the staging process were apparent by the mid-1930s. These were partly due to the fact that the notion of ‘uterine mobility’, used to define the different stages, was too vague to serve as a criterion for classification. The other problem was that clinicians used different procedures to evaluate cases. some investigators assessed the results on the basis of the clinical stage of the disease at the first full examination; others corrected the stage when the findings at operation or necropsy showed the clinical assessment to have been incorrect. The latter approach was favoured by surgeons, who argued that it was not possible to ascertain lymphatic gland involvement (and hence the stage of the disease) by clinical examination alone. The idea of restaging after surgery may seem more rational, but it prevented a fair comparison of treatment modalities because the patients who underwent surgery were not assessed in the same way as those who received radium
In order to ensure greater accuracy and uniformity of grading, changes to the wording and definitions for the various stages of cervical cancer were made in 1937. In the revised system clinicians were instructed to stage a malignancy without reference to the degree of uterine fixation. They were also required to determine the extent of disease at examination, before any therapy could affect the tumour.122
In 1950, at the New York World Congress of the International Federation of Gynecology and Obstetrics (FIGO), further changes were applied to the 1937 classification. The most controversial of these was the introduction of Stage 0 carcinoma to designate abnormal non-invasive lesions (defined by some as ‘latent’ pre-invasive cancer). The modified staging system was nominated ‘the International classification of the Stages of Carcinoma of the Uterine Cervix’, and all the organizations concerned with the problem of cervical cancer were invited to adopt the scheme. Data from other gynaecological sites were gradually added during the 1960s and 1970s.123 In 1958 FIGO became the official patron of the Annual Report, but the collection and publication of the data remained dependent on the generous support of a number of international cancer organizations, especially the Radiumhemmet.
Two more staging systems emerged after the end of the Second World War. The French Permanent Cancer Survey, which organized the recording of tumours in France from 1943 onwards, introduced a new classification based on the spread of the disease from the initial tumour (T), to the lymph nodes (N) and finally to other organs through the process of metastasis (M).124 At the end of the 1950s this TNM system was adopted by the Union for International Cancer Control (UICC) and it is now the most widely used in the field of cancer. In the United States, the American Joint Committee on Cancer (AJCC) created its own system at about the same time. Evolving in parallel, the two systems contained significant differences, but in the early 1980s they were gradually unified.
Both FIGO and TNM systems are now used for staging of gynaecological tumours, but there are substantial differences with regard to their objectives, methods of staging used, and separate classification of components. The TNM classification is based on a dual system distinguishing between clinical (pre-treatment) classification and a pathological (post-surgical histopathological) staging. Like the FIGO system it aims to provide inter-institutional comparisons of patients and treatments, but in addition it has further objectives: help for planning therapy; estimation of prognosis; and quality assurance of clinical classification. The FIGO system on the other hand aims primarily at providing inter-institutional comparisons, and it only allows a single staging. It is now either a surgical or pathological staging in most sites, with the exception of cervical cancer, and cancer of the vagina, which are still staged by clinical methods: this means that lymph node status is not included.125 The main reason given now for maintaining a clinical staging for cervical cancer is that the vast majority of cases are seen in developing countries, where many of the diagnostic and investigatory techniques used in the developed world are not readily available. As the FIGO system aims to be a universal system, the use of a clinical staging system is considered to be more appropriate for cervical cancer.126 Different considerations apply to cancer of the vagina. Surgery has a limited role here partly because of the proximity of the bladder and rectum, partly because many cases present with disease beyond the vagina. Surgico-pathological staging is thus regarded as inappropriate in this type of cancer.127
Decades of debate and discussion have brought clinicians no closer to the goal which the original staging systems were meant to achieve, however. During the 1950s, the addition of Stage 0 carcinoma and the gradual introduction of programmes combining radiotherapy with surgery and/or chemotherapy shook the simple plan of reporting end-results, introducing statistical pitfalls that made comparison of results extremely difficult. By the early twenty-first century, gynaecologists were acknowledging defeat: ‘Experience accumulated over decades has shown that a more precise answer to the question of superiority of one to the two treatment modalities we have today will not be found’, the Argentinian gynaecologist Guillermo di Paola commented in 2001: ‘This means that in the future, staging has other purposes and this may lead to alterations in concept and in detail.’128