Breast cancer is a major and important form of malignant disease in the Western World. In North America it was the most common malignancy among women, accounting for 27 per cent of all female cancers. Eighteen per cent of all cancer deaths are due to cancer of the breast, although since the mid-1980s the mortality from lung cancer has equalled and subsequently exceeded that of breast cancer as a cause of death in women. In the United States of America 130 000 new cases of breast cancer were diagnosed in 1987 and there were about 40 000 deaths from the disease. The equivalent figures from the United Kingdom are 30 000 and 15 000, respectively.

The risk of a woman developing breast cancer is a controversial subject and may be expressed in a variety of ways. In the Western world the cumulative risk (the proportion of a fixed group of women developing breast cancer over a set period of time) is about 7 per cent up to the age of 70. Thus, one in 14 women can expect to develop breast cancer.

There is remarkable variation in the incidence of breast cancer between different countries. The rates in the United States and Canada are six times higher than those in Asia or black Africa. Rates nearly equal to those seen in North America occur in north European countries and in New Zealand. Intermediate rates occur in eastern and southern European countries, South America, and the Caribbean. Japan has a low incidence of breast cancer, although it is becoming more common.

The difference in breast cancer rates is not simply a function of genetic susceptibility. The incidence of breast cancer in black Americans parallels that of white Americans rather than that of black Africans, and the cancer incidence of offspring of migrants to the United States of America from Japan is similar to that of native Americans.

The cause of breast cancer is unknown. However, epidemiological data indicate well-defined factors that indicate an increased liability to developing the disease. Such risk factors for breast cancer fall into three main groups: genetic, endocrine, and environmental; each may be of major, intermediate, or minor importance (Table 1) 272. Many of the minor factors are the source of continued debate.

Breast cancer is 100 times more common in women than in men. In strict epidemiological terms, therefore, female sex is a major risk factor for breast cancer, although it is often forgotten as such.

As for other epithelial cancers the incidence of breast cancer increase with age. Breast carcinoma is only occasionally seen in the late teens but thereafter there is a rapid rise in age-specific rates. Up to the age of 40 the increase in age-specific cancer rate is very steep; the rate of increase then slows dramatically, although the overall breast cancer rate continues to rise until old age (Fig. 1) 820. The cumulative risk of developing breast cancer between the ages of 20 and 40 is 0.5 per cent whereas between 50 and 70 it is 5 per cent. This accounts for the fact that the majority of patients presenting with breast cancer are over the age of 50.

The development of a second breast cancer may be a clinical manifestation of multifocal origin of the first cancer or may be an entirely new cancer. The risk of developing metachronous tumours has been addressed in a number of studies which have evaluated the development of cancer in the remaining breast after partial mastectomy. The incidence of such second cancers depends on whether or not synchronous bilateral cancers are included, but there appears to be an overall increased risk of 0.75 per cent to 1 per cent per year. Thus, the relative risk of developing a second non-synchronous primary 20 years after initial diagnosis of breast cancer is 1.2 to 1.5. This risk appears greatest in young women if their initial breast cancer is diagnosed before the age of 40.

A family history breast cancer is associated with an increased risk of the disease. The risk is greatest in patients with first-degree relatives (mother or sister) affected, especially if they were under the age of 50 when the disease developed. The relative risk of developing breast cancer is 1.7 to 2.5 in women with a history of breast cancer in a first-degree relative, and 1.5 among those with an affected second-degree relative. Whether the existence of multiple family members with breast cancer or the existence of bilateral disease in a relative indicates excessive risk is unclear, but such factors are likely to be of importance.

There is probably a direct genetic factor involved in the development of breast cancer in about 5 per cent of all patients.

Nulliparity removes a protective effect against breast cancer. Single and nulliparous married women have a relative risk rate of 1.4 compared to parous women; however, this effect of parity is almost totally due to the protective influence of early age at first birth. Women who give birth to their first child before the age of 20 have a relative risk of 0.5 compared to nulliparous women; for those whose first birth occurred after the age of 30 there appeared to be virtually no protective effect, with a relative risk of 0.94. Some evidence suggests that women whose first birth is over the age of 35 may have an increased risk of breast cancer.

If the age at first birth is taken into account, subsequent pregnancies appear to have no influence on the risk of developing breast cancer. The protective effect occurs only if the pregnancy continues to full term.

Any protective effect of breast feeding is hard to assess as it is difficult to differentiate from the effects of pregnancy. However, recent data from one study indicated an independent protective effect of breast feeding, although other studies have failed to confirm this finding.

Benign disease is not usually recognized as a major risk factor, although multiple papillomatosis may be regarded as such.

Women whose menarche occurs before the age of 12, have a relative risk of 2.30 compared to those starting menstruation after this age. This decreases as the age of onset of menstruation increases. The reduction in the age of the menarche over the past 100 years, especially in the Western world, probably results from improved nutrition and general health and may be important in the demographic variations in incidence of breast cancer.

The risk of developing breast cancer also relates to the age of the menopause. The relative risk of developing breast cancer is 0.5 per cent in those who cease to menstruate before the age of 45, compared to women who continue menstruating beyond age 55. Artificial menopause by oophorectomy or irradiation also reduces the risk of breast cancer.

An increased risk of breast cancer has been demonstrated in survivors of atomic explosions, women treated by radiation for postpartum mastitis, and patients receiving multiple chest radiographs during assessment of tuberculosis. This increased risk becomes apparent after a latent period of 10 to 15 years: the effect is most obvious in women exposed to irradiation when under the age of 35; there is little increased risk in women exposed after the age of 40.

There is a strong relationship between body weight and breast cancer although this is critically dependent on age. In women under age 50 there is little correlation between risk of breast cancer and body weight. However, in the 60 to 69 age group an increase in weight from less than 60 to 70 kg or greater increases breast cancer risk to 1.8.

Severe atypia with hyperplasia is associated with a moderately increased risk of developing breast cancer.

Minor and controversial risk factors

Evidence for an association between consumption of alcohol and increased liability to breast cancer is becoming stronger, although the risk is small (1.5).

Although weight correlates with breast cancer risk the relationship between dietary factors such as fat or cholesterol intake have not been shown to be an important factor in the development of breast carcinoma.

Some 15 case-controlled studies have evaluated the risk of breast cancer in women taking the contraceptive pill. The results are conflicting and if there is a risk it is small. Those most at risk are women taking oestrogen-based oral contraceptives early in life and taking them for at least 8 to 10 years.

The data relating to cancer risk and hormone replacement therapy are conflicting. Small doses of exogenous oestrogen therapy for short periods of time in premenopausal women appear to be safe. However, when hormone replacements are taken for 8 years or longer there may be an increased risk of 1.5 to 2.0.

Some pathological entities, such as multiple papillomatosis and hyperplasia with gross atypia, are certainly associated with an increased risk of breast cancer (3.0). The risk is lower with lesser degrees of atypia. Patients with recurrent macroscopic apocrine cysts may also have a slightly increased risk of breast cancer.

The relationship between benign breast disease and risk of cancer is bedevilled by ascribing cancer risk to fibroadenoma and ‘fibrocystic change’. There is no increased cancer risk for these two entities.

The histological assessment of breast cancer is of paramount importance in establishing the diagnosis of the tumour. It also helps determine the patient's prognosis and allows a greater understanding of the biology of the disease in any one particular case. Even relatively recently pathologists were satisfied in ascribing the term ‘breast carcinoma’ to all breast tumour specimens; the complexity of the subject is such that a more precise histological diagnosis is now required.

There are many methods of pathologically classifying breast cancer; most are based on whether the tumour is invasive or non-invasive and whether it is derived from the duct system or the lobule (Table 2) 273. It should however, should be remembered that most tumours arise from the terminal ductules. This fact also explains the common occurrence of mixed tumours with both lobular and ductal components existing within the same breast cancer.

This is the most common form of breast cancer accounting for 85 to 90 per cent of all cases. It can conveniently be subdivided into in situ and invasive types.

Ductal carcinoma in situ, or intraductal carcinoma, is a preinvasive form of breast cancer. It is characterized by a proliferation of malignant breast epithelial cells, is confined to the duct system, and does not invade the basement membrane or surrounding tissues. In the unscreened population, ductal carcinoma in situ accounts for less than 5 per cent of all cases of breast cancer; in screening programmes the incidence increases to 15 to 20 per cent because it is associated with mammographically visible microcalcification. Postmortem studies show that the incidence of ductal carcinoma in situ is about 6 per cent in apparently normal breasts, but is as high as 40 per cent in mastectomy specimens associated with invasive cancer.

The most important aspect of ductal carcinoma is its malignant potential. The little that is known of the natural history of this entity comes from retrospective evidence of treating ductal carcinoma in situ by local excision alone. Such studies demonstrate a 30 to 50 per cent of ipsilateral invasive cancer, usually in the same quadrant, after an interval of some 10 to 15 years. The risk of developing invasive cancer depends on the extent of ductal carcinoma in situ.

There are five main histological types of intraductal carcinoma. The ducts, the terminal duct lobular units, and the lobules may be filled with malignant cells (solid type), the lesions may undergo central necrosis (comedo type), or they may have a sieve-like appearance (cribriform type). They may produce papillary projections (papillary or micropapillary types) or they may cling to the duct wall (clinging type). Comedo and papillary forms (Fig. 2) 821 are the most common and are both associated with multicentric disease; comedo carcinoma is the most likely to become invasive and has the greatest expression of both DNA aneuploidy and of the C- erb2 oncogene.

Once intraductal carcinoma has invaded the basement membrane of the duct it has demonstrated the most important prerequisite of all malignant tumours—the ability to infiltrate into surrounding tissue. The term ‘invasive ductal cancer’ therefore refers to the majority of tumours arising from the duct system. A large number of different morphological types of invasive duct cancer is apparent to the pathologist. Some have prognostic importance. However, they can be basically subdivided into those with specific histological features, and those in which there is no characteristic microscopic appearance. The majority of invasive ductal cancers fall into the latter group. These are described as infiltrating ductal carcinoma, not otherwise specified (NOS).

This is essentially a diagnosis for exclusion, but it accounts for about 65 per cent of all invasive mammary cancers. No specific gross or microscopic features allow recognition of this type of cancer: it is simply the lack of specific and consistent histological features that characterize lobular carcinoma and other special types of ductal cancer (Fig. 3) 822. In a system of histological grading, infiltrating ductal carcinoma, not otherwise specified, would usually be classified as intermediate or high grade and, within the spectrum of invasive breast cancer, it has a relatively poor prognosis. However, other features such as tumour size and axillary nodal involvement are of more prognostic importance than histological features alone.

Medullary carcinoma is a well recognized entity for about 6 per cent of all invasive mammary cancers (Fig. 4) 823. Macroscopically they tend to be soft, well circumscribed, and have a uniform consistency. They have a smooth contour and may even appear to have a pseudocapsule. Their most obvious histological characteristic is a prominent lymphocytic infiltration, a feature not seen in ductal carcinoma (NOS).

Medullary carcinoma is less frequently associated with lymph node metastases than other types of breast cancer and is therefore associated with a better prognosis.

This type of breast cancer is uncommon, accounting for only about 3 per cent of all infiltrating cancers of the breast (Fig. 5) 824. Tubular carcinoma is well differentiated and previously had been often inappropriately classified in the not otherwise specified group.

Most tubular cancers are small, being about 1cm in diameter. They are hard and on section have a radial appearance. Histologically they form tubular structures with an open central space lined by a single layer of epithelium.

Recent interest in tubular carcinoma has been heightened by the impact of the breast screening programme. Not only is tubular carcinoma more frequently seen in screened individuals than in the symptomatic population but other abnormalities detected on mammography can cause diagnostic confusion. Of particular importance is the differential diagnosis between tubular cancer, sclerosing adenosis, and radial scar.

Patients with tubular carcinoma have a good prognosis and a 10-year survival rate in excess of 75 per cent. Nodal metastases, even when present, tend to be limited to one or two nodes only. It is uncommon for pure, well differentiated tubular carcinomas to metastasize to distant sites.

It is unclear whether special types of tumour such as tubular carcinoma dedifferentiate into more aggressive types of cancer as they progress (phenotypic drift). Such questions are of paramount importance when assessing the value of breast screening programmes.

Mucinous carcinoma (Fig. 6) 825 is characterized by a small rounded tumour which histologically demonstrates pools of mucous material in which are aggregates of cancer cells. This tumour accounts for about 2 or 3 per cent of all invasive cancer, but it has been associated with a survival rate which is appreciably better than that of ductal carcinoma (NOS). These tumours said to occur in older women and, like tubular carcinoma, rarely metastasize to lymph nodes.

Unlike other subtypes of ductal carcinoma, papillary carcinoma can be easily subdivided into in-situ and invasive types. The non-invasive papillary carcinoma has the fronded structure of a papilloma. It is usually well delineated from surrounding breast tissue by a fibrous covering, hence the essentially inaccurate term of intracystic papillary carcinoma. As long as there is no extracapsular evidence of in-situ carcinoma simple excision of these ‘intracystic papillary carcinomas’ will suffice.

Invasive papillary carcinoma to account for only about 2 per cent of all breast cancers. Microscopically, they are usually well circumscribed and histologically demonstrate papillary formation. The prognosis is generally good.

This has only recently been recognized as a special type of invasive ductal carcinoma. It accounts for about 3 per cent of all breast cancers and may be more common in the screened population. It classically presents with a firm mass, but microscopy reveals characteristic stromal invasion with a cribriform pattern. They tend to be well differentiated and have a good prognosis.

The presence of signet ring cells in breast cancer is a rare but important phenomenon. Most of these tumours are diffuse and poorly defined. Histologically, for reasons which are unclear, they are often associated with lobular carcinoma and are characterized by signet ring cells with a peripherally placed nucleus and marked cytoplasmic vacuolization. Their prognosis is poor; they selectively metastasize to the peritoneal cavity and it becomes difficult to differentiate metastatic signet cell carcinoma from those originating in the gastrointestinal tract.

Clear cell tumours also have a poor prognosis. Their appearance is characteristic on haematoxylin and eosin staining; further selective stains demonstrate either a lipid-rich or glycogen-rich pattern.

Secretory carcinomas are rare. They are frequently small and are characterized by abundant intra-and extracellular areas of vacuolization. These tumours frequently occur in younger women and generally have a good prognosis; extremely rare breast cancers occurring in childhood are often of this type.

Inflammatory carcinoma has previously been ascribed to tumours occurring in pregnancy and lactation (mastitis carcinomatosis), although it is now apparent that it occurs in all age groups. The diagnosis of inflammatory cancer is dependent on the observation of tumour emboli in dermal lymphatics, giving rise to the red infiltrative appearance of this condition. It must be differentiated from peu d'orange which is simply due to lymphatic obstruction of the breast as a result of axillary nodal metastases from tumours of any type. Its prognosis is very poor.

Other rare types of infiltrating ductal carcinoma include those with carcinoid, adenocystic, and squamous features.

Lobular carcinoma can also be conveniently subdivided into in-situ and invasive forms, depending on whether the basement membrane of the lobule has been invaded by tumour.

Lobular carcinoma in situ, like its intraductal counterpart, is also a preinvasive form of breast cancer. The term ‘lobular carcinoma in situ’ was used in its original description by Foote and Stewart in 1941; other terms, such as intra-acinous carcinoma and lobular neoplasia, have also been widely used.

The microscopic criteria for diagnosing lobular carcinoma in situ are well defined (Fig. 7) 826. Within the lobule there must be a uniform proliferation of cells and, as a result, there should be no interstitial spaces and expansion of at least half of the acini in the lobular unit. If these criteria are not completely fulfilled then the term ‘atypical lobular hyperplasia’ is appropriate (Fig. 8) 827. The term ‘lobular neoplasia’ was used by Haagensen to describe both atypical lobular hyperplasia and lobular carcinoma in situ although its use has been criticized as it implies that atypical lobular hyperplasia is a neoplastic process.

The most important aspect of lobular carcinoma in situ is its potential for becoming invasive. The risk of invasive lobular carcinoma developing following simple biopsy for lobular carcinoma in situ is only about 25 per cent over 20 years: this is a lower risk than that of ductal carcinoma in situ. Unlike ductal carcinoma in situ, lobular carcinoma in situ rarely expresses the C- erb2 oncogene.

The biggest single study evaluating the malignant potential of lobular carcinoma in situ is that of Haagensen. As well as identifying the risk of frank malignant change, Haagensen made a number of important observations (Table 3) 274. It is now generally agreed that as the risk of invasive cancer after diagnosis of lobular carcinoma in situ (relative risk, 10) relates equally to both breasts, it should be regarded as a risk factor for tumour development rather than a direct precursor. Atypical lobular hyperplasia carries a 4-fold increased risk for development of breast cancer.

In practice, lobular carcinoma in situ is usually discovered by chance. It has no mammographic features and is usually found on biopsy undertaken for other reasons.

Invasive lobular cancer accounts for about 10 per cent of all cases of breast carcinoma, although its incidence varies quite widely. Up to 10 per cent of lobular cancers have a co-existing ductal component. Five main subtypes are described: classical, solid, alveolar, mixed, and pleomorphic. Their main pathological features are summarized in Table 4 275. The most common type, classical lobular carcinoma, is characterized by the ‘Indian filing’ of invading malignant cells, often in a ‘targetoid’ pattern (Fig. 9) 828. The ‘monotonous’ features of individual cells along with lack of a specific histologic architecture results in difficulty in grading lobular carcinoma.

Invasive lobular carcinoma is important with respect to its tendency to bilaterality and the fact that its clinical diagnosis may be difficult because of the diffuse infiltrative nature which frequently produces distortion of the breast rather than a lump. The prognosis of classic invasive lobular carcinoma is generally somewhat better than that of invasive ductal cancer. The other types of lobular carcinoma, especially its pleomorphic variant, have a worse prognosis than the classic form of this disease.

The lay public, and indeed a number of medical practitioners, believe that transition from normality to breast cancer and thence to metastatic disease is a rapid process. The reverse is, in fact, the case. The natural history of breast cancer is normally characterized by long duration, but shows extreme heterogeneity between patients. Breast cancer is one of the more slowly growing tumours which therefore renders it suitable for screening programmes. The evolution of breast cancer, especially in its preclinical phases, can be measured in years or even decades, although there are exceptions to this rule in which the disease takes on a more aggressive form.

Our understanding of the preclinical phases of breast cancer is based on indirect observations of the morphological changes from normality to hyperplasia, atypia, carcinoma in situ, and finally to invasive cancer. It should be remembered, however, that this progression is by no means inevitable and, in theory, stages may be missed out. It is also possible that a given stage may be permanent or may even regress to a more normal state. Histological characteristics may be correlated with other biological factors, such as labelling index, doubling time, and growth fractions. Serial mammography has indirectly indicated that the doubling time for human breast cancer is usually of the order of 100 to 300 days, although exceptions to this are frequently seen. If it is also assumed that if a tumour of 1cm size contains about 10&sup9; cells it would require about 30 doublings for a malignant cell to produce this volume. This does, of course, assume that there is no cell loss, that there is a linear growth rate, and that the tumour is initially derived from a single cell. Although these assumptions can be challenged, they do indicate that the preclinical phase of breast cancer would still be of the order of about 7 years and often much longer. Clinical observations of patients treated by biopsy alone for hyperplasia or atypia give some support for these figures.

The heterogeneity of breast cancer can also be appreciated from the percentage of cells that are undergoing cell division at any particular time (labelling index). Tumours with high labelling indices are associated with short doubling times and are therefore more aggressive. Overall, the median labelling index is about 3 per cent; the range extends from less than 1 per cent to approximately 40 per cent. High labelling indices are associated with poor prognostic factors such as premenopausal status and with a poor overall clinical outcome.

The natural history of clinically apparent breast cancer is also surprisingly slow. Historical observations on untreated breast cancer as recorded by the Middlesex Hospital, London, and the Connecticut Tumor Registry show an average survival of about 2.5 to 3 years without treatment. Such data can be criticized as they are historical and probably relate to a group of patients presenting with locally advanced rather than metastatic disease. They do, however, indicate the need for prolonged monitoring in patients with clinical breast cancer.

Once breast cancer is clinically apparent it will, by definition, tend to spread locally and to regional or distant sites. The extent and rate of spread varies greatly from one individual to another and is characteristic of the heterogeneity of the disease. Some tumours spread rapidly to regional nodes or distant sites while remaining small in the breast; others grow slowly to a large size without metastasizing. Clearly, those with the greatest metastatic potential have the worst prognosis.

Within the breast there are three main mechanisms of spread. The most important mechanism is local spread by direct infiltration into the surrounding parenchyma. This occurs by the ramifying projections that give the characteristic macroscopic stellate appearance of breast cancer. If uncontrolled, direct infiltration of overlying skin or the underlying fascia occurs.

A second mode of local spread is by direct infiltration along ducts, although it is unclear whether this represents actual tumour growth or whether it reflects a field change of pre-existing in-situ disease. The presence of widespread in-situ cancer explains the phenomenon of multifocality in patients with breast cancer. Multifocality, or cancer within the same quadrant as the primary tumour, should be distinguished from multicentricity, which is defined as the presence of carcinoma outside the quadrant containing the primary tumour. In one study, 43 per cent of 246 patients with invasive breast cancer less than 4cm in diameter demonstrated multifocality. In 43 per cent of these patients affected areas were more than 2cm from the primary tumour; in 9 per cent additional foci of cancer were found more than 4cm away. Such multifocal tumours are confined to the duct system in 66 per cent of patients, but the remainder demonstrate invasion into surrounding tissues. The incidence and extent of multifocality depends on the size of the primary tumour. These findings have obvious implications when treating breast cancer by conservative surgery.

Local lymphatic and vascular spread within the breast are also of prognostic importance. Of particular importance are those lymphatic pathways extending into the pectoral fascia and sub-areolar regions.

The regional spread of breast cancer is defined as that to the axillary, internal mammary, and supraclavicular nodes.

The axillary nodes represent the most important site of regional spread from breast cancer. Spread to axillary nodes is the most important prognostic indicator of breast cancer: approximately 45 per cent of all patients have nodal disease at presentation. The likelihood of axillary nodal spread is a function of the size of the breast primary, but even when correction is made for tumour size, disease in the axillary nodes remains the most important prognostic variable. For tumours less than 2cm in diameter, the incidence of axillary nodal spread is less than 20 per cent. The incidence of axillary nodal disease is 35 per cent in those with tumours 2 to 5cm in diameter, and 50 per cent of patients with tumours greater than 5cm in size.

The clinical assessment of axillary nodes is notoriously unreliable. About 30 per cent of palpable and apparently diseased nodes are found to be histologically free of metastases; conversely, up to 30 per cent of apparently normal axillary nodes demonstrate histological evidence of metastatic disease.

The extent of nodal disease reported depends to a great extent on the pathological evaluation of the specimens. Not only is there a great variation from one centre to another in the surgical removal of nodes at axillary sampling or clearance, but there is also a marked difference between pathologists in retrieving and examining all the nodes provided in the surgical specimen. Techniques such as clearing the axillary fat with xylene to increase nodal yield and more thorough sectioning of the nodes themselves increase the positivity rate. However, few pathologists have adopted these methods.

The relationship between axillary nodal spread and prognosis depends on three factors: the number of nodes affected, the level of axillary nodal disease, and the extent of disease within the nodes themselves. Accurate appraisal of the nodes is therefore dependent on close liaison between the surgeon and pathologist with respect to orientation of the specimen at the time of axillary exploration.

Survival rates based on the number of diseased lymph nodes show remarkable unity between studies. Results assessed by the American College of Surgeons have shown that the number of nodes affected by cancer provides valuable prognostic information. Documentation of nodal status as negative or positive for metastatic disease yields only qualitative data and fails to provide the prognostic information that can be obtained from full analysis of the number of nodes affected at axillary clearance. The number of negative nodes recovered on axillary clearance is of no prognostic importance, although some authorities claim that to ensure that the axilla is clear of metastases at least four, and possibly up to 10, negative nodes must be isolated. After 5 years women treated by mastectomy have an 82 per cent relative survival rate if no lymph nodes are involved and a 60 per cent survival rate if one or two lymph nodes are involved at the time of initial treatment. Women with five or six nodes affected have a 47 per cent survival rate; this is reduced to only 31 per cent if 11 or 12 nodes are involved. Women with more than 20 nodes involved have only an 8 per cent survival rate at 5 years.

Thus, according to studies made by the National Surgical Adjuvant Breast Project (NSABP), within the group of women with positive nodes, it is apparent that if one to three nodes are affected by the tumour the prognosis, while not good, can be relatively optimistic. However, women with four to 12 positive lymph nodes should be given a more guarded prognosis, and those with more than 13 axillary nodes affected have a decidedly poor outcome.

As previously described, the axillary nodes are conveniently divided into three groups depending on their relationship to the pectoralis minor muscle. Prognosis relates to the level of axillary node affected, although this is a less powerful factor than the total number of nodes affected by the tumour. Thus, Shottenfeld et al. showed that patients with disease in level I nodes had a 5-year survival rate of 65 per cent, those with disease at level II had a 5-year survival rate of 31 per cent, and no patients with disease affecting level III nodes survived for 5 years.

Patients with disease in level III nodes usually have a large number of affected lymph nodes. So-called skip metastases, in which metastatic deposits are found at levels II or III, but not at level I are found in only 2 per cent of patients. Thus, a dissection limited to level I alone is effective in determining the presence or absence of nodal disease but will tend to underestimate the degree of spread in many patients and will not provide accurate prognostic information.

The extent of disease within individual axillary nodes may also be of importance although, as is the case for level of involvement, there is a greater correlation with the number of nodes involved. Deposits less than 2 mm in size (micrometastases) can be detected only by histological methods. Such deposits usually occur in one node only; occasionally they are found in two, but their presence in a third is exceptional in the absence of macroscopic metastases. The prognostic significance of occult micrometastases is the subject of wide debate. They are probably far more common than is suggested by routine pathological examination of axillary nodes. The likelihood of detecting a micrometastasis during routine single section of an axillary node is about 6 per cent; this increases to 36 per cent if six equally spaced sections are taken through each lymph node.

The work involved in detecting micrometastases is extensive, and the question still remains as to their overall prognostic importance. It is likely that the presence of a single micrometastasis implies a slightly worse prognosis than for node negative disease, but improved survival compared to patients with a single pathologically obvious macrometastasis. This prognostic importance, however, is only applicable after at least 10 years; in practice the prognosis for patients with a single micrometastasis can be regarded as similar to those with node-negative disease.

In contradistinction to micrometastases, Haagensen and Stout down-staged their patients if nodal metastases were greater than 2.5cm in diameter. Spread of metastatic disease outside the confines of the axillary nodes into surrounding fat is also a bad prognostic sign.

The internal mammary chain of lymph nodes lies at the anterior end of the intercostal spaces adjacent to the internal thoracic artery. Metastases in the internal mammary nodes are more commonly associated with medial or periareolar tumours and the overall incidence of such involvement is as high as 20 per cent. Disease affecting the internal mammary lymph nodes is rare in the absence of axillary nodal spread; only 8 per cent of patients have such disease if the axilla is clear.

Metastatic disease in the internal mammary nodes alone has the same prognostic implication as axillary nodal disease. However, if both the internal mammary and axillary nodes are affected the outlook is very poor, with a 25 per cent 10-year survival rate.

Because of the poor prognosis of patients with disease of the internal mammary nodes, various attempts at treatment have been adopted, including both surgical excision (in extended radical mastectomy) and radiation therapy. In some studies this aggressive approach has been associated with an improvement in survival in women with such metastatic disease. However, the routine excision of internal mammary nodes in all patients with breast cancer has not been associated with an overall improvement in survival.

It is unclear whether intramammary node disease, like supraclavicular node spread, should be regarded as regional or metastatic disease.

Metastatic disease in the supraclavicular nodes implies extensive involvement of the internal mammary or axillary nodes. The frequent association with widespread metastatic disease means that supraclavicular nodal disease is associated with a poor prognosis.

No site is immune to spread of the tumour, either at the time of presentation or later in the course of the disease. The most commonly affected sites are the bone, liver, and lung, but metastases to the brain, skin, and peritoneum are by no means uncommon.

Micrometastases in the bone marrow have been recognized for many years in patients with metastatic breast cancer. It is now becoming increasingly clear that such spread also occurs in up to one-third of patients with early, operable, breast cancer. The presence of micrometastases in bone marrow may correlate with other indicators of poor prognosis such as axillary lymph node metastases, tumour size, and oestrogen receptor status, but there is no clear evidence that they indicate a poor prognosis. In short, their relevance is unclear.

The majority of women presenting with breast cancer complain of a lump. Classically this is hard, painless, immobile, and demonstrates a degree of fixity to surrounding tissues, overlying skin, or the underlying pectoral muscle. The lump may be visible on inspection and may cause distortion of the breast on elevation of the arms or when tensing the pectoral muscles. Many tumours fail to demonstrate these classical characteristics, however. Not all breast cancers are hard: they are often described as being only firm. The degree of mobility can also vary: small tubular cancers can have a degree of mobility almost characteristic of a fibroadenoma. Finally, not all cancers are painless and occasionally patients describe the disconcerting symptoms of increased discomfort in the lump prior to menstruation. It is, therefore, advisable to undertake histological and cytological examination of any discrete lump in a woman over the age of 25. A diagnosis of a fibroadenoma or ‘cystic change’ without such confirmation is extremely dangerous.

Up to 15 per cent of women with breast cancer present with a more diffuse process within the breast, in which a lump is not necessarily the presenting feature. This is particularly common in younger patients with lobular carcinoma. The diffuse nature of the tumour produces distortion, puckering, and the eventual feeling of heaviness as the tumour extends through the breast. Later there may be nipple retraction and discomfort. A similar process occurs in patients who describe a thickening in the breast. Such symptoms can be difficult to differentiate clinically from those of benign breast disease.

Changes in the skin may be the sole presenting symptom; alternatively, they may be associated with other features of the cancer. Puckering may be pronounced and is often associated with the rather scirrhotic tumours of the elderly. However, this clinical sign is commonly observed in younger patients. Peu d'orange is a feature of advanced cancer. The cause of this characteristic clinical sign is oedema of the skin: this is not due to direct infiltration of the skin by tumour but represents lymphatic obstruction of the breast as a result of axillary metastasis. It must be differentiated from inflammatory breast cancer, in which cutaneous lymphatics contain tumour emboli. In advanced, untreated cases the skin may be broken, and tumour ulcerates through the skin, with associated haemorrhage and odour. Surrounding skin nodules are not uncommon in seriously neglected cases.

Presentation with changes at the nipple are not uncommon, but these primary tumours may be difficult to diagnose as they are often small and may be easily missed by mammography. Nipple distortion and inversion are not uncommon presenting features in patients with breast cancer. A unifocal or blood-stained nipple discharge is also indicative of malignant disease, often an intraductal carcinoma developing with the major duct system beneath the nipple. The discharge associated with intraductal cancer is characteristically watery and blood-stained, in contrast to the milk discharge of galactorrhoea and multifocal, tenacious, and coloured discharge of duct ectasia.

Paget's disease (Fig. 10) 829 is usually a presenting feature of breast cancer in the elderly, although it is occasionally seen in other age groups. It usually represents an underlying intraductal carcinoma which may be quite extensive in the breast and which may also exhibit an invasive component. The origin of the characteristic, large, clear Paget cells remains an enigma. Whether they represent migration of tumour cells from the primary cancer or originate from the skin of the nipple itself remains unclear.

Finally, it should be remembered that breast cancer frequently presents with metastatic disease. This may be either to regional lymph nodes, such as those in the axilla or supraclavicular fossa, or to distant sites such as bone, lung, liver, or brain.

There is no indication for a purely clinical diagnosis of breast cancer. The medicolegal consequences of an unnecessary mastectomy are such that there should be no doubt as to the diagnosis before embarking on definitive surgery.

Confirmatory diagnosis is by pathology and radiology, the former being by far the most important.

This method has the advantages of being performed as an outpatient procedure and producing almost immediate results. The technique of aspiration is not difficult after an initial learning period, and it is relatively atraumatic. Its disadvantage is that it requires expert and specialized pathological interpretation. The false-negative rate is about 15 per cent, and false positive results occur in about 2 per cent of cases. False-negative results are due either to sampling errors as a result of missing the tumour at the time of aspiration or from failing to aspirate a particularly scirrhous acellular carcinoma. False-positives are uncommon but may be due to confusion with a hypercellular fibroadenoma or the effects of undisclosed hormone therapy, pregnancy, or lactation on normal breast tissue.

A variety of instruments can be used to provide a core of tissue in an outpatient procedure. This technique has to be performed under local anaesthetic, but it has the advantage that it produces a histological rather than a cytological specimen. The author has found the technique somewhat traumatic and liable to miss small scirrhotic tumours, although others have had great success with this method. As the core biopsy produces a histological rather than a cytological diagnosis, in-situ disease can be differentiated from invasive disease. Greater appraisal of the grade and type of tumour is also possible, although sampling errors are likely with such a small specimen.

Biopsy is required in patients who clinically are suspected to have a cancer yet in whom fine needle aspiration cytology or core biopsy fails to demonstrate malignant disease. It has the disadvantage of requiring hospital admission, although the majority of patients can be treated and discharged the same day. Its advantage is that it provides a definitive method of proving or excluding malignant disease. Open biopsy can occasionally be performed under local anaesthetic, but it is performed more easily under general anaesthesia.

Frozen section evaluation of an excised specimen at the time of definitive surgery has become less common with the more widespread use of fine needle aspiration cytology. There is a false-positive rate of about 1 to 2 per cent: sclerosing adenosis and nipple adenoma can cause particular confusion with tumour.

A further reason for the decline in popularity of frozen section techniques is the very reasonable change in the surgical approach to women with breast cancer. Modern practice should avoid the outmoded approach of performing mastectomy on the basis of frozen section while the patient is anaesthetized and unable to discuss various treatment options determined by pathological findings.

The last decade has witnessed a huge improvement in imaging methods for breast disease, and a high level of sensitivity and specificity is now available using modern mammographic techniques. The classic features of breast cancer on mammography are tissue asymmetry, mass effect, microcalcification, skin thickening, and nipple inversion (Fig. 12) 832. One or more of these features may be present, the most reliable being a combination of mass effect with localized microcalcification.

Other methods of imaging such as ultrasonography, xerography, and thermography have also been successfully used in the diagnosis of breast cancer. However, for symptomatic masses which are amenable to fine needle aspiration cytology, good quality mammography alone should suffice as a diagnostic method.

Preoperative mammography is not only of diagnostic importance but also detects potential multicentricity and acts as a base-line for radiological evaluation of the breast in the years after initial diagnosis.

Once the diagnosis of breast cancer has been made certain investigations are of value in subsequent management. All patients with breast cancer should undergo chest radiography and routine evaluation of haemoglobin, full blood count, electrolytes, and liver function tests. The value of other more sophisticated tests is open to doubt. There has been a vogue for performing bone, liver, and even brain scanning in all patients for the detection of asymptomatic metastases in patients who were felt to have early operable disease. In practice, however, there is a low incidence of true positive scans in ‘operable’ cancer. The positive yield for bone scanning, in patients with apparently localized disease is about 2 per cent, and approximately 50 per cent of these cases will be false-positives as a result of previous trauma or coexisting arthritis. There is an even lower positive yield with liver and brain scanning, some studies demonstrating a 100 per cent incidence of negative results following such investigations.

Although there is some disagreement over the usefulness of routine preoperative scanning, the overall incidence of positive results is low. However, patients with clinical stage III disease and those with symptoms suggestive of metastases should undergo further investigation. Brain scanning is of little value in patients with apparently operable breast cancer and the current recommendation is that if screening investigations are to be performed they should be limited to bone scintigraphy and ultrasonography of the liver.

Staging relates to the classification of breast cancer according to the anatomical extent of disease, each stage serving to aggregate cases having an approximately similar prognosis.

In essence, staging is somewhat simplistic as every variable in breast cancer is classified under a few simple categories. There is therefore loss of precision and wide variation within the stages. There is also the question of relating clinical to pathological staging. As discussed above there is a 30 per cent error rate in clinical evaluation of axillary nodes and the assessment of tumour size is similarly inaccurate. Pathological assessment of nodes and tumour size allows a greater accuracy of staging. However, since pathological staging can only be undertaken after surgical treatment it has obvious limitations in determining management.

Many staging systems have been proposed; none has been shown to be significantly better than others, although some have the advantage of simplicity. More commonly used systems are described below.

Stage I. Tumour confined to breast. Any skin involvement covers an area less than the size of the tumour.

Stage II. Tumour confined to breast. Palpable, mobile axillary nodes.

Stage III. Tumour extends beyond the breast tissue because of skin fixation in an area greater than the size of the tumour or because of ulceration. Tumour fixity underlying fascia.

Stage IV. Fixed axillary nodes, supraclavicular nodal involvement, satellite nodules or distant metastases.

In 1954 the International Union against Cancer (Union Internationale Contre Cancere) attempted to classify breast cancer on a description based on the primary tumour (T), the regional lymph nodes (N), and distant metastases (M). This system has been modified on a number of occasions and has led to confusion and recent criticism. A simple modification is as follows:

T0—no evidence of primary tumour.

T1—tumour less than 2cm in diameter.

T2—tumour 2 to 5cm in diameter.

T3—tumour greater than 5cm in diameter.

T4—tumour fixation to chest wall or skin

N0—axillary nodes not considered to contain tumour

N1—mobile involved axillary nodes

N2—fixed axillary nodes

N3—supraclavicular lymph node involvement or arm swelling.

M0—no distant metastases.

M1—distant metastases present.

The designation ‘X’ after T, N or M indicates inability to assess the clinical status of the tumour, nodes, or metastases. The correlation of the TNM System with stage is summarized in Fig. 13 833.

The Columbia classification (Haagensen, Cooley, and Stout 1943, 1969)

Stage A—no skin oedema, ulceration, or fixation to chest wall, axillary nodes not clinically involved.

Stage B—clinically involved axillary nodes less than 2.5cm in diameter and not fixed.

Stage C—grave signs of comparatively advanced carcinoma: oedema of skin, skin ulceration, fixation to chest wall massive axillary involvement with nodes greater than 2.5cm in diameter, and axillary fixation.

Stage D—advanced carcinoma including two or more signs in Stage E, satellite skin nodules, supraclavicular nodes, inflammatory cancer, arm oedema, or distant metastases.

There has been a massive change in the treatment of breast cancer over the past 100 years. The past 20 years have witnessed fundamental changes in our approach, with decreasing reliance on radical surgical excision of the primary tumour and associated regional lymph nodes. This change of emphasis has resulted from a fundamental alteration in our philosophy of the spread of breast cancer, a greater appreciation of the systemic aspects of the disease, and increased realization that variations in local regional treatment are unlikely to alter prognosis.

Early concepts regarding the spread of breast cancer were proposed by Halsted. His opinion dictated the treatment of breast cancer for almost a century. He believed that breast cancer spread by direct permeation rather than embolization to the regional nodes. The nodes acted as a barrier to the further spread of cancer until they themselves were overladen with tumour, whereupon bloodborne metastases occurred. Such a concept of the spread of disease was the basis for performing en-bloc radical resection of the entire region in radical mastectomy and for the even more radical operations which followed that of Halsted.

A major flaw in Halsted's theory was, of course, that many deaths from metastatic disease occurred after radical surgery for small localized cancers. It thus became apparent that embolization, rather than permeation, of tumour cells is the predominant mode of simultaneous spread to both regional nodes and distant sites. It is now widely believed that spread from tumour to both lymph nodes and distant sites represents two independent but correlated processes. According to this alternative theory of breast cancer, spread to axillary nodes represents the risk of metastatic disease rather than its determinant. This theory also suggests the existence of an inherent systemic component to breast cancer and implies a limitation of local regional treatment alone on prognosis. It also suggests that some patients may well have systemic metastases at the time of clinical presentation even in the absence of axillary nodal metastasis.

The pros and cons of these two conflicting theories on breast cancer have been fiercely argued. Support for Halsted's theory is based on the fact that a significant proportion of patients with disease affecting axillary nodes are apparently cured after radical surgery. Furthermore, there is generally an orderly spread of involvement up the axillary lymph nodes, skip metastases being relatively uncommon, and there is little evidence for systemic spread accompanying nodal involvement. Free circulating tumour cells have not been demonstrated and imaging techniques demonstrate subclinical metastases in only a very small proportion of patients with apparently operable breast cancer.

The advocates of the alternative theory of breast cancer base their argument on both clinical and experimental evidence. There is no evidence to suggest that treatment of disease in axillary nodes is of any significance with respect to survival, although it does substantially improve the local control of tumour in that area. Further evidence for the alternative theory of breast cancer comes from experimental studies demonstrating failure of lymph nodes to filter tumour cells and from the existence of direct communications between afferent lymphatics and efferent veins around lymph nodes. Finally, studies of the natural history of breast cancer indicate that many tumours exist for several years before they become clinically apparent. Thus, there is ample opportunity for metastases to develop during this protracted period of occult growth. It has therefore been implied by some authorities that clinically apparent invasive breast cancer is never really curable.

The management of breast cancer requires a complex multidisciplinary approach involving surgeons, radiotherapists, medical oncologists, and pathologists, as well as other professionals such as counsellors and breast care nurses. For the sake of convenience the treatment of breast cancer will be discussed in three sections: the management of the breast itself, the axillary nodes, and potential micrometastases. However, it must be stressed that these three approaches have a final common aim—the successful treatment of patients with breast cancer. A properly co-ordinated approach by the various health professionals is required if optimal results are to be achieved.

The importance of the management of the breast itself is to provide local tumour control and reduce any potential for metastasis to either local or distant sites. If this can be achieved using a surgical technique which allows conservation of the breast there is also the added bonus of improved cosmesis and, hopefully, a reduction in psychological trauma.

The traditional surgical treatment for breast cancer involved total removal of the breast. The various types of mastectomy describe those structures resected in association with the entire breast (Table 6) 277. Up until the early 1970s breast cancer was treated by total mastectomy, in one or other of the forms described in Table 6 277. The most commonly performed operation was the Halsted radical mastectomy, although in Europe there had been a vogue for simple mastectomy with adjuvant radiotherapy since the early 1950s.

The second half of the 20th century witnessed increasing disillusionment with radical and mutilating forms of surgery for breast cancer. As a result the trend towards breast conservation has increased since the mid-1960s, although a number of centres had adopted this approach since before the Second World War. Again there are a number of different descriptions relating to breast conservation which has caused confusion. Tumourectomy, lumpectomy, tylectomy, segmental mastectomy, and quadrantectomy are all synonymous with a therapeutic procedure in which the primary tumour is removed and the breast is preserved. Unfortunately, these terms are not precisely defined, although they imply the removal of varying amounts of normal breast tissue in association with a primary tumour. The terms ‘lumpectomy’, ‘tumourectomy’, and ‘tylectomy’ imply removal of the tumour with a minimal or no margin of normal breast tissue around it. Segmental mastectomy implies excision of the tumour with a rim of associated normal breast tissue. However, this term is also somewhat misleading as it implies that the breast is anatomically a segmental organ and that tumours occur in a localized segment. This is clearly not the case. The term ‘quadrantectomy’ denotes removal of a breast quadrant, and implies wider excision of normal breast tissue than segmental mastectomy. In practice, however, there is little distinction between these terms and although a number of authorities have recommended the adoption of a uniform nomenclature, none has found universal favour.

Once the questions regarding definition of terms and nomenclature have been addressed the simple, yet fundamentally important question which remains is whether breast conservation provides results as reliable in the treatment of breast cancer as total mastectomy. Furthermore, is there an additional benefit in terms of cosmetic and emotional adjustment? Finally, if breast conservation is justified, in which patients is this appropriate?

In terms of management of the breast the simplest approach would be to remove the tumour itself, preferably with a margin of normal tissue around it. In theory the more limited procedures of tumourectomy or lumpectomy are likely to be followed by a good cosmetic result but are more likely to be followed by local recurrence because of the likelihood of failure to excise the tumour completely. More extensive forms of conservative surgery such as quadrantectomy are more likely to provide good tumour control but are more liable to be followed by a less satisfactory cosmetic result because of the amount of breast tissue excised.

Studies evaluating simple excision of the tumour without adjuvant radiotherapy have produced somewhat disappointing results, with a local recurrence rate of approximately 30 per cent within 7 years. Local recurrence may result from positive margins of excision as a result of inadequate surgery or from development of further tumours either de novo or from pre-existing multifocal disease.

Failure of local excision alone to provide adequate tumour control in the breast led to the use of adjuvant radiotherapy to the remaining breast tissue. This combination of conservative surgery with adjuvant radiotherapy has now become a standard procedure for patients with breast cancer. A large number of studies have provided retrospective analysis of conservative treatment with surgery and irradiation for the treatment of this disease. The results have been generally encouraging with local recurrence rates of under 10 per cent, although in some centres 20 per cent of patients have developed locally recurrent disease after periods of 7 years of more.

Unfortunately, as the majority of studies relating to conservative surgery with adjuvant radiotherapy are retrospective, there are doubts about case selection. A more accurate appraisal can only be obtained from analysis of prospective randomized trials. There have been three major studies comparing conservative treatment with mastectomy which require more detailed discussion.

In the first study on breast conservation Atkins and his colleagues at Guy's Hospital, London, randomized patients over the age of 50 years with clinical stage I or II disease to either ‘extended tylectomy’ or classical radical mastectomy. Both operations were followed by axillary, supraclavicular, and internal mammary irradiation and the patients treated by tylectomy also underwent X-ray treatment to the residual breast tissue. Some patients received adjuvant thiotepa. Unfortunately, the dose of radiation used in this study was somewhat suboptimal.

After 10 years there was a significantly higher local recurrence rate in patients treated by breast conservation for both Stage I and II disease, although this had no impact on survival for patients with Stage I disease. Patients with Stage II disease treated by breast conservation had a worse prognosis.

A follow-up study from the same institution by Haywood and his colleagues randomized a further 258 clinical Stage I patients to the same regimens, but again with suboptimal radiation doses. On this occasion conservative treatment resulted not only in relatively poor local control but also decreased survival. Thus, the Guy's group provided inconsistent results following breast conservation, even in patients with Stage I disease.

A second important randomized study was that of Veronesi and his colleagues from Milan. Seven hundred and one women with tumours less than 2cm in diameter and without palpable axillary nodes were randomized to either radical mastectomy or quadrantectomy, axillary dissection, and breast irradiation. Adjuvant chemotherapy was given to those patients with histologically proven lymph node metastases. Overall 13-year survival was 69 per cent and 71 per cent respectively, in the two groups. Local recurrence rates were similar in patients treated by radical mastectomy and in those treated by breast conservation.

The third study evaluating breast conservation in the treatment of breast cancer was the protocol B–06 conducted by the National Surgical Adjuvant Breast and Bowel Project (NSABP). In this trial 1843 women were randomized to mastectomy or to segmental resection (lumpectomy), with or without radiation. All patients underwent axillary clearance, and those found to have positive nodes were given adjuvant chemotherapy. Patients found to have positive margins after breast conservation were converted to the mastectomy group—perhaps introducing some bias into the study.

After 8 years there was no significant difference in survival between patients treated with breast conservation and those with mastectomy. However, those patients treated by lumpectomy without radiation suffered a greater incidence of local recurrence in the ipsilateral breast.

Thus, two of these randomized studies clearly demonstrated that patients suitable for breast conservation have an equally good prognosis when treated by a conservative procedure with adjuvant radiotherapy rather than mastectomy. All the above studies may be also criticized in that case selection was still liable to exclude patients with more aggressive tumours, who may have not been selected for randomization. However, the other major criticism of these studies—that there has been inadequate follow-up—can now be answered, in that assessment for more than 8 years is now available.

Patient choice is important. If a patient prefers to be treated by total mastectomy rather than breast conservation then, after appropriate counselling, her wishes should be adhered to. It is widely believed that women with breast cancer prefer to be treated by partial mastectomy. This is often, but not always, the case: many women who actually have the disease feel more secure after a total mastectomy, and other prefer to avoid irradiation treatment if at all possible. One recent study from the United Kingdom showed that 50 per cent of patients suffering from breast cancer actually prefer the concept of total mastectomy to breast conservation.

The other main indications for conservative therapy relate to tumour characteristics, the position of the cancer in the breast, and the nature of the breasts themselves. Tumour size is of paramount importance; most authorities recommend mastectomy if the tumour is more than 3 or 4cm in diameter. The NSABP recommends partial mastectomy only if the tumour is less than 4cm in diameter and when there is no fixation to the underlying muscle or overlying skin. Other surgeons pursue a more conservative approach and prefer to reserve breast conservation for tumours of less than 3cm in diameter. Much will, of course, depend on the size of the breast. In small breasts a tumour 4cm in diameter may require mastectomy, and conservation with good cosmetic results may only be a reality if the tumour is 2cm or less in size. On the other hand breast conservation may be permitted for tumours greater than 4cm in diameter if the breasts are large, although it must be remembered that heavy pendulous breasts may provide poor cosmetic results after conservation because of the oedema and fibrosis which may follow irradiation.

A further contraindication to breast conservation is multicentricity. Patients with a second tumour in the breast, whether it is clinically apparent or seen on mammography, should be considered for mastectomy if the tumours cannot be excised in continuity. Mammography is therefore an essential preoperative investigation to determine the presence of multicentric tumours.

Multifocality, as opposed to multicentricity, is present in up to 50 per cent of patients with breast cancer. Its extent depends on tumour size and it is this multifocality which may result in the unacceptable incidence of local recurrence if breast conservation is performed without adjuvant radiation therapy. It is yet to be determined whether adjuvant radiotherapy truly sterilizes multifocal disease or whether it simply delays recurrence to the very long term (10–20 years). Widespread multifocal disease certainly relates to local recurrence—its presence is a relative contraindication to breast conservation.

The question of multifocality is particularly important in patients with lobular carcinoma. It may be felt that this potentially multifocal and multicentric disease should be treated by mastectomy. However, as long as criteria relating to tumour size and type are adhered to there is no evidence that this particular type of tumour is any more likely to recur than is ductal carcinoma.

Centrally located tumours are a relative contraindication to breast conservation. Such tumours are often quite diffuse and they can be difficult to remove by segmental mastectomy. Furthermore, the end cosmetic result is often disappointing as many women perceive the nipple as being a fundamentally important part of the breast—both physiologically and psychologically. It is not uncommon for women to hold the view that if the nipple has to be removed then the treatment may as well be mastectomy. Despite this opinion many women are, in fact, pleased with the end cosmetic result, even though the nipple has to be removed in the treatment of a centrally located tumour. This is particularly true for those women with larger breasts, in whom a nipple reconstruction or prosthesis may be used.

Poor tumour differentiation is also a relative contraindication to breast conservation, although this is a somewhat empirical judgment in many cases. Moreover, an assessment of tumour grade is not possible until the cancer has actually been removed. Many poorly differentiated tumours are somewhat ill-defined and this specific clinical feature may be a further relative contraindication to breast conservation.

There have been doubts as to whether breast conservation is appropriate for patients with node-positive disease. As the presence of disease in axillary nodes indicates a higher risk of tumour dissemination there is, in fact, an argument in favour of breast conservation in these circumstances.

Despite the various contraindications to breast conservation this type of treatment has now become the management of choice for the majority of patients with breast cancer. However, there is still much variation from centre to centre, with some authorities recommending total mastectomy in up to 70 per cent of cases, whereas others recommend breast conservation for virtually all their patients.

Local recurrence after either total mastectomy or breast conservation surgery is a major clinical problem. Locally recurrent disease must be distinguished from regional recurrence; the two are often confused. Local recurrence is that seen in the breast after conservative therapy or on the chest wall after mastectomy; regional recurrence denotes recurrent disease in regional lymph nodes. Local recurrence may be either localized or widespread. Both types, but particularly the latter, may be associated with widespread metastatic disease and it is therefore mandatory to stage all patients with either local or regional recurrence for distant metastases.

In general, the more radical the treatment the lower the incidence of local recurrence either on the chest wall after mastectomy or in the remaining breast tissue following breast conservation therapy. Thus, very minimal treatment such as lumpectomy alone is associated with a higher incidence of local recurrence than is partial mastectomy or quadrantectomy, particularly if associated with adjuvant irradiation. Similarly, there is a higher incidence of local recurrence after partial than total mastectomy. The lowest local recurrence rate (2 per cent for stage I disease) has been recorded after extended radical mastectomy.

The overall incidence of local recurrence after breast conservation with adjuvant radiotherapy is about 10 per cent, although some authorities have recorded figures in excess of 20 per cent in a period of 7 years or more. Local recurrence after total mastectomy has previously occurred in about 5 to 8 per cent of patients. However, in modern practice this incidence may be even greater because of selection of large, prognostically unfavourable tumours for this type of treatment.

Recurrent disease, either in the breast after partial mastectomy or on the chest wall after total mastectomy, is more common in patients with positive axillary nodes. It has therefore been suggested that as well as giving radiotherapy to all patients after partial mastectomy, the chest wall should be irradiated after total mastectomy if disease has spread to the nodes. Meta-analysis, however, has shown no improved survival in patients given adjuvant radiotherapy, although it reduces the local recurrence rate on the chest wall. Addition of radiotherapy to mastectomy may actually increase overall mortality because of an increased liability to myocardial infarction and lung cancer following irradiation.

Tumour size has also been shown in some studies to be associated with increased incidence of local recurrence after both total mastectomy and breast conservation. Some authorities therefore recommend adjuvant irradiation after both partial and total mastectomy if the primary tumour is large. Again, there is no evidence that this improves survival, although it reduces local recurrence.

Positive surgical margins following partial mastectomy would certainly be regarded as being an important risk factor for recurrence, although not all studies have confirmed this. Although positive surgical margins may relate to inadequate surgical technique they are also likely to occur in patients with extensive in-situ disease. It is therefore likely that widespread multifocality is an important factor predisposing to local recurrence.

Other factors such as high tumour grade, lymphatic or vascular invasion, and young age are also associated with recurrent disease.

Local recurrence after breast conservation therapy can be treated by salvage mastectomy. Such treatment has been claimed to produce a 5-year survival rate of up to 50 per cent, and this may well be so for those patients with localized recurrence. Further breast conservation is sometimes feasible in these women. Unfortunately, however, up to 20 per cent of those who develop local recurrence in the residual breast tissue after breast conservation have advanced or unresectable disease. This is because local recurrence after breast conservation with adjuvant radiotherapy can occur as a ‘field-change’ throughout the irradiation field. Positive surgical margins may occur despite extensive surgical resection, musculocutaneous flaps, and skin grafts. Further local recurrence is not uncommon in these individuals, who face terminal illness with severe physical and psychological morbidity from uncontrolled and prolonged chest wall recurrence. Systemic treatment with tamoxifen or chemotherapy may alleviate symptoms but responses are often quite short-lived when recurrence is widespread.

Local recurrence after total mastectomy can present a difficult problem. Small focal areas of local recurrence can be surgically excised. If irradiation has not been previously used, X-ray therapy may prevent or delay further local recurrences. Similarly, adjuvant drug treatment with either chemotherapy or endocrine therapy may help reduce the risk of further recurrence.

Local recurrence after total mastectomy may present as a widespread area of recurrent disease. This is particularly difficult to treat. Surgical excision with associated transposed musculocutaneous flaps or skin grafting is often associated with positive margins, further local recurrence, and the rapid development of metastatic disease. Radiotherapy may be attempted if this has not already been used; chemotherapy or endocrine treatment may produce some regression of this somewhat remorseless process.

All women undergoing total mastectomy should be offered reconstruction, even if they have advanced disease. Although it was hoped that this would reduce the psychological sequelae of breast cancer this is not always so, although reconstruction can improve an otherwise miserable existence for some women. The increasing trend towards breast conservation should reduce the number of women requiring reconstruction after total mastectomy: if reconstruction is required after breast conservation therapy there must be doubts as to whether the original surgical operation was appropriate.

The two main questions relating to breast reconstruction are the timing and the technique. Breast reconstruction can be performed either at the time of initial surgery or delayed for, say, 1 to 2 years after diagnosis. The obvious advantage of immediate reconstruction is that women do not undergo a period of loss of the breast. Furthermore, only one surgical procedure is required. It must be remembered, however, that a reconstructed breast generally compares unfavourably with the original, and that immediate reconstruction can result in unfavourable comparison by the patient. It might also be argued that initial reconstruction, and its possible postoperative problems, should take second place to the actual treatment of the cancer. One of the author's personal fears is that if immediate reconstruction takes place and rapid local or distant recurrence occurs the patient may believe that surgical management of the cancer has taken second place to reconstruction. Finally, reconstruction may delay the early diagnosis of local recurrence, but data on this subject are lacking.

The advantage of delayed reconstruction is that the patient will compare, usually more favourably, the reconstruction with the mastectomy. Adjuvant therapy with chemotherapy or irradiation will have finished and the patient may be more psychologically attuned to her disease. Many women, however, defer or refuse delayed reconstruction on the basis that they and their partner have become quite used to the missing breast and that they have had enough medical treatment. This in part explains the author's experience that many younger women with children to look after tend to refuse reconstruction, whereas there is greater acceptance in women aged over 50.

There are numerous methods of reconstruction: no one method is perfect. Appropriate counselling is of particular importance before embarking on any reconstructive surgery. The patient should see photographs of both good and bad results to ensure a realistic approach. If possible she should talk to women who have previously undergone reconstruction. It is essential to stress the reconstruction will provide a ‘mound’ but that it will not feel or move like the original. Further counselling is required about nipple reconstruction, as this aspect is often forgotten.

The reader is advised to consult specialized plastic surgical texts on methods of reconstruction. However, for the sake of completeness some of the more widely used techniques are briefly discussed.

This is the easiest technique to perform, although skin necrosis and implant extrusion may occur, especially after irradiation and if a subcutaneous position is used. Most surgeons prefer to place the implant in a subpectoral position; the subcutaneous technique is mentioned only to be condemned as it frequently causes skin necrosis, extrusion, and a poor cosmetic result (Fig. 14) 834.

When inserting the implant subpectorally care must be taken to ensure that the prosthesis is in an appropriate position. Since there is a tendency for such implants to migrate superiorly it is necessary to divide the tight fascial bands between the insertion of the pectoralis muscle and the rectus abdominus to allow appropriate positioning.

Two main cosmetic problems occur with this technique. The first is capsule formation, which is unsightly and also imparts a very hard texture to the breast. Capsule formation is less common if ‘textured’ implants are used. If excessive, the capsule can be surgically fractured or excised to improve appearance and to soften the reconstructed breast.

The second cosmetic problem with simple implants is lack of ptosis which occurs with the normal breast, especially with increasing age. The use of tissue expansion techniques can, to some extent, overcome this problem (see below), especially if performed in conjunction with one of the various plastic surgical procedures which enhance ptosis.

Most implants in current use are made of silicone gel. This material has the advantages that it is inert and has a relatively natural feel. Recent publicity regarding risk of cancer and autoimmune disease have been exaggerated. The alternative type of implant using saline gives a rather soft, striated and ‘flabby’ result. Furthermore leakage of saline can occur in the long term.

A major difficulty relating to this and other types of reconstruction is the fact that a nipple has to be recreated (see below). This problem can be avoided by performing a subcutaneous mastectomy, although many oncologists feel this is an inappropriate cancer operation because 2 to 10 per cent of the original breast tissue may be left behind following this technique.

Simple implants have limitations with respect to lack of natural ptosis and complications in irradiated skin. Both these problems can, to some extent, be overcome by gradual tissue expansion. Traditionally, this has been achieved by a tissue expander followed by insertion of a permanent prosthesis. The introduction of a combined tissue expander/permanent prosthesis has greatly simplified matters. The Becker expander/mammary prosthesis consists of an outer lumen filled with silicone gel and an inner chamber connected to a removable, self-sealing, side port (Fig. 15) 835. This inner chamber is filled with saline and is gradually expanded to the desired volume over a number of weeks. The expansion eventually provided is substantially greater than the volume of the opposite breast. Once expansion is complete the saline chamber is aspirated to allow creation of a reconstructed breast of similar volume to that on the opposite side, while at the same time providing some ptosis. The self-sealing injection port can then be removed under local anaesthetic.

The most commonly adopted technique has been the pedicled latissimus dorsi myocutaneous flap, because of its proximity, good blood supply, and ease of creation (Fig. 16) 836. The bulk of muscle transposed, however, is often quite small and the technique frequently has to be combined with an implant. Some authorities report good results after immediate reconstruction with this technique, although it leaves a scar on the back and is often complicated by seroma formation.

The most recent innovation has been the rectus abdominus myocutaneous flap. A longitudinal or transverse incision can be used. In the former a pedicle graft is swung on its axis to provide tissue bulk on the chest wall. Most authorities now prefer to use a transverse rectus abdominus flap in which a horizontal, elliptical, subumbilical incision is used for access (Fig. 17) 837. Because of the rather precarious blood supply to this region many surgeons now recommend a free transverse rectus abdominus flap with microvascular anastomosis between epigastric and thoracodorsal vessels. There is a graft failure rate of about 8 per cent with this technique.

The cosmetic results of the free transverse rectus abdominus flap are such that they should be regarded as the ‘gold standard’ by which other methods of reconstruction are judged. Unfortunately, it is a time-consuming procedure and requires specific expertise with microvascular anastomosis. It is also relatively contraindicated in patients who are very obese or very thin, in those who have a pre-existing lower abdominal scar, in those who are heavy smokers, and in those who are above age 55.

All the above forms of reconstruction may require reduction mastopexy in the contralateral breast if equality is to be achieved.

For many women the nipple is the most important part of the breast, yet the majority of techniques described above fail to provide appropriate nipple reconstruction. Only subcutaneous mastectomy routinely preserves the nipple–areola complex; the disadvantages of this method are described above. Several methods of nipple reconstruction are available, using either any excessive skin on the reconstructed breast or a free graft from a distant site, such as the groin or thigh. An alternative approach is to use an adhesive prosthesis.

A difficulty with nipple reconstruction relates to pigmentation. If the reconstructed nipple is pale in comparison to that on the opposite side then simple tattooing may provide an appropriate effect.

The management of the axillary nodes is a controversial issue, and practice varies widely from centre to centre. However, the management of axillary nodes should provide certain well-defined end points. First it should act as a guide to staging and prognosis. Second, it dictates the need for adjuvant therapy with irradiation, chemotherapy or hormonal therapy. Third, it must provide good local control of disease in the axilla as untreatable axillary metastases can cause much morbidity.

Despite the importance of disease in the axillary nodes, however, there is no evidence that good therapeutic control of the axilla correlates with survival.

The major argument relating to management of axillary nodes concerns the extent of their surgical excision and the impact that this may have on staging, prognosis, and the subsequent control of axillary disease. The extent of surgical excision of the axillary contents can range from that of non-intervention to routine block clearance of all nodes in all patients. All approaches have their supporters and detractors.

Failure to assess axillary nodes has been rightly condemned. This approach has unfortunately been relatively common and has resulted in subsequent inability to prescribe rational local and systemic adjuvant therapy. Radiotherapy then has to be given blindly resulting in the needless irradiation of disease-free nodes. Similarly, it is impossible to provide a rational programme of systemic hormonal or chemotherapy if nodal status is not known.

Recently, however, there has been a resurgence of this rather negative approach to axillary nodes. This has resulted from appreciation that nodal control fails to correlate with prognosis and the wider use of adjuvant therapy. Supporters of non-intervention claim that if all patients receive systemic adjuvant therapy and careful monitoring, axillary nodal relapse can be treated as it occurs clinically. In fact, such axillary nodal relapse is relatively uncommon—a further point in favour of this approach. Whether this view will become more widespread is unclear.

Axillary nodal sampling, as opposed to full surgical clearance, has been widely practised in the United Kingdom although, like the policy of non-intervention, it has been criticized. A number of studies have failed to demonstrate adequate node retrieval using this technique, and staging is claimed not to be as accurate as with clearance. There are also doubts as to whether radiotherapy is as good as clearance in controlling axillary nodal disease.

A major problem relating to sampling is that it is poorly defined. There is no agreement as to the extent or boundaries of axillary surgery when sampling, although it is important to emphasize the necessity of removing enlarged or palpably hard nodes.

The supporters of axillary node sampling have answered all the above criticisms. Firstly, those studies which failed to demonstrate adequate nodal retrieval used inappropriate anatomic landmarks to define the axillary glands. Thus, the extent of dissection had been demarcated by landmarks such as the axillary tail of the breast or the intercostal–brachial nerve. When the axillary fat has been entered using appropriate anatomical dissection, sampling provides as good a qualitative assessment of the axilla as clearance. It must be admitted, however, that sampling fails to provide as much prognostic information in terms of a full nodal assessment, which may be performed after complete dissection of the nodes.

The supporters of sampling also claim that, if done properly, node-positive patients can be specifically targeted for adjuvant radiation therapy, whereas a full axillary dissection needlessly overtreats the 50 per cent of patients with negative nodes. The available data indicate that adjuvant radiotherapy provides as good overall control in the axilla as clearance. Such control does not, however, correlate with survival.

The whole question relating to sampling depends on how well it is done. When inappropriately performed it is rightly condemned. However, if done properly it provides qualitative data as good as those obtained from clearance, allows appropriate patients to be treated with radiotherapy, and provides a rational basis for systemic treatment. Such an opinion has been confirmed by Steele and his colleagues, who showed that surgical clearance provided no further qualitative data once adequate sampling had been performed.

Many of the criticisms relating to sampling are answered by axillary clearance. Block dissection of axillary nodes accurately stages the axilla, both qualitatively and quantitatively, and has the advantage that it provides a good mechanism for tumour control. The main criticism of clearance, however, is that the 50 per cent of patients with negative nodes do not require this treatment. There are also doubts as to the efficacy of clearance obtained with differing operative techniques. In some centres the pectoralis minor muscle is routinely removed or divided; in others it is merely retracted. This may explain the wide variation in nodal count between centres undertaking axillary clearance. The technical aspects of clearance may be compounded when it has to be performed through a separate axillary incision distant from the primary tumour site: in these circumstances access to the apex of the axilla may be limited.

Critics of clearance also suggest it may be followed by a higher incidence of arm lymphoedema than is sampling. However, as long as axillary clearance is not combined with radiotherapy there is little evidence that long-term morbidity after this more radical procedure is any worse than following sampling with adjuvant X-ray therapy.

Overall, there is little to choose between well-performed axillary sampling, with adjuvant radiotherapy treatment for patients with diseased nodes, and axillary clearance. Axillary clearance has the disadvantage that 50 per cent of patients undergo unnecessary dissection of their axilla, whereas radiotherapy is both time-consuming and expensive postoperatively. In the author's unit reliance is generally placed on axillary sampling followed with adjuvant radiotherapy for node-positive patients. However, if there is bulk nodal disease with metastases greater than 2cm in diameter we prefer full surgical axillary clearance.

Systemic metastatic disease in patients with breast cancer is a serious development, with a median survival after diagnosis of only about 14 months. Such metastases are particularly distressing when they occur in patients originally thought to have a relatively good prognosis by virtue of tumour size and negative axillary nodes. Earlier expectations of improved survival with either more radical surgery or adjuvant radiotherapy have not been realized. The systemic component of breast cancer, as described above, indicates that either the patient is cured by local treatment or that death occurs from metastatic disease at about the same time that she would have died without local intervention. This is because according to the non-Halstedian theory of breast cancer metastases are established prior to diagnosis and are outside the scope of even the most extensive local treatment.

The limitations of local treatment provided the initial rationale for the use of adjuvant systemic therapy at the time of, or following, initial surgical treatment of breast cancer. Such an approach was first adopted for breast cancer in 1948, when oophorectomy was suggested as an adjuvant treatment. Early studies produced conflicting results but indicated the need for a wider appraisal of the systemic approach which gradually developed over a 30-year period as more active chemotherapeutic and hormonal agents became available. As a result of numerous studies adjuvant systemic therapy is now regarded as an integral party of the primary treatment of breast cancer.

The first large-scale studies evaluated single agent chemotherapy such as thiotepa, phenylalanine mustard, and cyclophosphamide. After a 12-year study Nissen-Meyer and his colleagues demonstrated a significant survival advantage in patients receiving a brief perioperative course of cyclophosphamide compared to untreated controls. Other experience with single agents was also encouraging, especially in patients with positive nodes, although there was some scepticism because of stratification within the trials and variations in treatment methods. Subsequent studies have evaluated a large number of chemotherapeutic regimens using a variety of agents and varying durations of treatment. A combination of cyclophosphamide, 5-fluorouracil, and methotrexate has been most widely used, because of its known activity in patients with overt metastatic disease and its relative ease of administration.

The initial study evaluating this combination therapy in breast cancer showed a 30 per cent reduction in mortality in patients receiving 12 cycles of treatment, but also showed that the effect was most clearly apparent in premenopausal women with 1 to 3 positive nodes. These results were regarded as a major advance in treatment of breast cancer, although other similar studies subsequently failed to confirm these findings. This discrepancy was compounded by occasional studies that even showed a worse prognosis in patients receiving adjuvant cyclosphosphamide, 5-fluorouracil, and methotrexate, although the majority of investigators demonstrated some benefit in selected groups of patients.

Because of the conflicting results obtained by various trials of adjuvant chemotherapy there was initially very little agreement as to the value of this form of treatment in patients with breast cancer. The matter was finally settled by adopting the statistical technique of meta-analysis of all randomized trials evaluating adjuvant chemotherapy for breast cancer. The results of this analysis were initially published in 1985 and 1988 but were subsequently updated in 1992 when 10-year monitoring was available.

More than 30 trials evaluating adjuvant polychemotherapy are now available for study, with a follow-up of up to 10 years. Meta-analysis has demonstrated a 25 per cent reduction in annual mortality in women aged under 50. This translates to an absolute reduction in 10-year mortality of about 10 per cent in node-positive patients, but only 4 per cent in those who are node-negative. Single agent chemotherapy does not have the same efficacy. The effect of chemotherapy in women under the age of 50 is almost completely manifest within 5 years of initial treatment, but continues for at least 10 years.

Combination polychemotherapy is also effective in women aged over 50, although not to the same degree as in younger patients. Meta-analysis demonstrates a 12 per cent reduction in annual mortality. The absolute survival difference is only 5 per cent for node-positive patients and even less for those who are node-negative.

As a result of the meta-analysis adjuvant polychemotherapy with cyclophosphamide, 5-fluorouracil, and methotrexate is now recommended for all node-positive women aged less than 50. A six-cycle course of treatment is now regarded as standard. Centres with a more aggressive treatment policy are also recommending this treatment for node-negative women aged less than 50, especially if there are other adverse factors such as large tumour size, poor differentiation, and oestrogen receptor-negative status. At the present time such chemotherapy has not been demonstrated to be associated with significant survival improvement. However, it is associated with prolonged time to initial relapse and it is argued that when monitoring has been continued for long enough a survival advantage will be truly demonstrated.

The 1992 overview analysis also demonstrated an adjuvant effect of cyclophosphamide, 5-fluorouracil, and methotrexate in node-positive women aged over 50. This effect was not apparent in the 1985 study and may relate to improved techniques now available for administering chemotherapy to older individuals. Some units are giving this combination chemotherapy to some node-positive patients over the age of 50, such as those who are oestrogen receptor-negative, as long as their overall performance status will allow this potentially toxic treatment.

A common side-effect of adjuvant chemotherapy in premenopausal patients is cessation of menses. Some authorities have therefore suggested that chemotherapy causes a pharmacological castration and ovarian ablation. Support for this concept is demonstrated in the most recent overview analysis of the positive effect of adjuvant oophorectomy in premenopausal women. However, even if the adjuvant effect of polychemotherapy is created by pharmacological ovarian ablation, this treatment also has an effect in node-positive postmenopausal patients. It is therefore likely that cylophosphamide, 5-fluorouracil, and methotrexate has a direct chemotherapeutic action in younger women as well as by potentially causing ovarian failure.

It is clear that adjuvant chemotherapy will be used more extensively in patients with breast cancer. Its use is likely to be extended to women less than 50 years old with node-negative disease. It is also likely to be used increasingly in patients over this age who have a poor prognosis and for whom adjuvant hormonal therapy may be inappropriate because of factors such as oestrogen receptor negativity.

Patients with a very poor prognosis, such as those with a large number of diseased nodes, may in future be considered for even more aggressive polychemotherapy such as provided by cyclophosphamide, Adriamycin, and 5-fluorouracil.

A further area of interest relates to ‘neo-adjuvant’ chemotherapy. Polychemotherapy prior to surgery may downstage the primary tumour and has a theoretical advantage of allowing early systemic treatment to potentially poor prognosis patients. However, it has yet to be determined whether this approach improves survival.

Hormonal manipulation is associated with almost as much controversy as adjuvant chemotherapy. However, its relative ease of administration and lesser toxicity has resulted in its more widespread application. Meta-analysis has now demonstrated very definite indications for its use.

Early studies evaluating hormonal manipulation adopted adjuvant oophorectomy in premenopausal women: these failed to show consistently any survival advantage. The introduction of tamoxifen produced an easily administered, relatively non-toxic form of oestrogen blockade, and this was, therefore, the natural agent to assess as an adjuvant treatment. Trials again produced conflicting results. Overview analysis, however, has subsequently demonstrated an overall reduction in annual odds of death of 17 per cent. The effect is greatest in patients age over 50 with positive nodes, in whom there is a 20 per cent reduction in annual mortality. This translates to an 8 per cent reduction in 10-year mortality. There is a small, yet still significant, effect in node-negative individuals. The effect of tamoxifen is less apparent in women aged under 50, whether node-negative or node-positive, with an overall delay in development of recurrence of 12 per cent. However, even in this group the results are described as ‘non-significantly favourable’.

As is the case for adjuvant chemotherapy the effect of tamoxifen is found within the first 5 years but continues for at least 10 years after initial diagnosis. An important effect of adjuvant tamoxifen was the demonstration in the overview analysis that it reduces the risk of contralateral breast cancer by 39 per cent. This has obvious implications with respect to prophylaxis in high-risk groups of patients.

The role of adjuvant tamoxifen is now well established in women over the age of 50. The duration of treatment is, however, unclear: recommendations vary between 2 years, 5 years, and continued provision until relapse. Meta-analysis has demonstrated the best results for adjuvant tamoxifen when used for at least 2 years. Early fears relating to possible increased mortality from myocardial infarction and morbidity from osteoporosis have not been realized. It appears that, as far as cardiac and bony tissues are concerned, the oestrogenic effects of tamoxifen outweigh its anti-oestrogenic influence.

Despite meta-analysis the role of adjuvant tamoxifen remains a controversial issue. A number of authorities have expressed doubts as to its efficacy in oestrogen receptor-negative patients, although both overview analysis and some individual studies, contrary to expectations, have shown benefit in this particular subgroup. Similarly, some studies have shown a beneficial effect in women under the age of 50, although this has not been fully confirmed by overview analysis. Tamoxifen is not without side-effects, particularly in premenopausal patients, about 30 per cent of whom experience menopausal symptoms, nausea, weight gain, vaginal dryness, or discharge.

The most surprising finding of the 1992 overview analysis was a 24 per cent reduction in mortality in node-positive patients aged less than 50 following oophorectomy. The effect was as great as that of adjuvant chemotherapy and, as discussed above, gave credence to the concept that the cyclophosphamide, 5-fluorouracil, methotrexate regimen acts by providing a pharmacological oophorectomy. It is also likely that this drug treatment has inherent chemotherapeutic activity. It is as yet uncertain whether oophorectomy, rather than chemotherapy, should be used as an adjuvant in premenopausal breast cancer. It is also unclear whether oophorectomy, if used as an adjuvant treatment, should be performed surgically or whether reliance can be placed on either irradiation or administration of luteinizing hormone releasing hormone antagonists such as Zoladex to provide this effect. If performed surgically it is probable that oophorectomy could be performed using laparoscopic techniques.

The following recommendations are based on the 1992 overview analysis. However, they are only guidelines and there will be variation in policy from centre to centre.

1.Six cycles of polychemotherapy with the cyclophosphamide, 5-fluorouracil, methotrexate regimen. All node-positive women aged less than 50. Possibly of value in node-negative women aged less than 50 and node positive patients over 50 if it can be tolerated in the latter group. This regimen in node-positive patients over age 50 is most likely to be of benefit if there are doubts as to the efficacy of adjuvant tamoxifen because of oestrogen receptor negativity etc.

2.A 2-year course of tamoxifen. All women over aged 50, particularly if node positive. The effect on women under 50 is small but probably valid especially if it is felt that adjuvant chemotherapy is not warranted.

Although there is a temptation to combine adjuvant chemotherapy with tamoxifen, there is no evidence that their effects are additive. Indeed, there is experimental evidence that tamoxifen inhibits the action of chemotherapy. Randomized trials are currently in progress evaluating combination chemotherapy with tamoxifen given either concurrently or sequentially.

The majority of patients with Paget's disease are elderly, although the author has seen one patient presenting with this condition in her twenties. The clinical presentation of Paget's disease with an itching, vesicular eruption involving the nipple is confirmed by biopsy and the subsequent demonstration of large cells with pale cytoplasm (Paget's cells) in the epidermis (Fig. 18) 838. The origin and nature of Paget's cells are unknown, although they reflect an underlying ductal carcinoma. The ductal cancer may be either in situ or invasive and may or may not be associated with a mass. There is sometimes quite extensive disease within the breast.

The differential diagnosis is with eczema, which is usually bilateral and associated with disease elsewhere. Paget's disease of the nipple should also be distinguished from a frank tumour that is simply eroding the skin of the nipple and areola.

Paget's disease of the nipple has traditionally been treated by mastectomy with lower axillary clearance and is thought to have an excellent prognosis. If the cancer is of the in-situ type this is certainly true. Patients with more extensive disease associated with invasive cancer have a relatively worse prognosis.

Because so many patients with Paget's disease are elderly there has been a recent trend to treat this condition by wide local excision of the nipple and surrounding tissues. However, this may not be satisfactory treatment if positive margins result. Furthermore, excision of the nipple, especially in small breasts, may lead to a rather unusual cosmetic result. Treatment with tamoxifen or simple irradiation therapy alone has not been adequately evaluated.

Breast cancer is a common condition in very elderly women. Some, but not all, authorities claim that tumours in this age group are slowly growing, are associated with a high oestrogen receptor status, and have good prognosis. Treatment with the anti-oestrogen tamoxifen has therefore been advocated: randomized controlled trials have shown this to be as effective, in terms of survival, as surgery. Some authorities, however, have found an unacceptable incidence of local failure after this treatment and recommend mastectomy. Surgical excision is therefore still a good alternative and can be supplemented by adjuvant tamoxifen therapy.

The conservative approach using tamoxifen alone should be reserved for the very frail. It should also be remembered that not all elderly patients have good prognosis tumours. Presentation with axillary nodal involvement and the subsequent development of visceral metastases is common.

Breast cancer during pregnancy and lactation is perhaps the subject par excellence in which opinion is based on anecdotal rather than proven scientific facts. It is fortunately rare, accounting for less than 1 per cent of all breast cancer cases. However, it is this rarity, along with difficulty in evaluating the pregnant or lactating breast, which makes diagnosis difficult and delayed treatment common.

The traditional view of breast cancer during pregnancy and lactation was that it was an inflammatory type of tumour with cutaneous erythema from intradermal lymphatic involvement (mastitis carcinomatosis). However, although inflammatory cancers do occur in pregnancy they are more common in other age groups. The majority of tumours seen in pregnant women are similar to those occurring in the non-pregnant state.

Those who believe breast cancer in pregnancy to be of an inflammatory type have also suggested that these tumours have a very bad prognosis. Anecdotal evidence implies that the prognosis is poor, although proper scientific appraisal of the available literature on the subject fails to confirm this belief totally.

Breast cancer occurring during pregnancy should be treated according to standard principles. Each case must be judged from its own merits regarding the indications or otherwise for breast conservation, adjuvant irradiation, and chemotherapy.

Previous recommendations regarding termination of pregnancy have been revised in the light of lack of evidence that it improves survival. However, as irradiation and adjuvant chemotherapy are contraindicated during pregnancy then a first trimester termination would be indicated for patients with locally advanced disease in whom these adjuvant treatments are necessary. The pregnancy can be continued, if the patient wishes, in those with apparently local diseases. The treatment of choice would be mastectomy and axillary clearance, as this would avoid need for adjuvant irradiation. If nodes were found to be positive then consideration would have to be given to termination of the pregnancy and provision of adjuvant chemotherapy.

If cancer is detected in the last trimester than surgical treatment can be provided using normal criteria. Adjuvant chemotherapy or irradiation can be delayed until after delivery.

For tumours occurring in the second trimester the choices of management are more difficult. A full discussion with the patient and partner is required and a treatment policy formulated which takes into account not only the state of the disease but also the patient's wish regarding the pregnancy.

Breast cancers occurring during lactation present special problems. Lactation should be suppressed with bromocriptine and treatment of the cancer based on standard principles.

One question asked by younger women previously treated for breast cancer is what the effect of a subsequent pregnancy may be on their well-being. Many authorities recommend that after the diagnosis of breast cancer pregnancy should be avoided for 2 and preferably 5 years. Such advice is again based on somewhat anecdotal evidence and is not widely supported in the literature. However, it is important to give honest advice to patients with a poor prognosis so that the consequences of an untimely pregnancy can be avoided.

Locally advanced tumours are defined as those more than 5cm in diameter or those with fixed axillary nodes in which there is no evidence of distant metastases. The management of these cancers has been somewhat disappointing because of poor local tumour control and the high incidence of subsequent metastatic disease, resulting in an overall survival rate of 20 per cent or less.

The traditional treatment of locally advanced breast cancer, radical mastectomy, has produced disappointing results, with up to 40 per cent of patients developing local–regional recurrence within 2 years and low overall survival rates. As a result there has been enthusiasm for treatment with high-dose radiotherapy. Some studies, using doses in excess of 60Gy, have obtained similar results to those achieved with mastectomy. Local recurrence rates could be reduced if the tumour was surgically debulked or if an even higher radiation dose was provided by an interstitial implant.

There is probably little to choose between mastectomy and primary irradiation in the management of locally advanced breast cancer. A combination of the two may produce improved local control, although this has yet to be demonstrated in prospective trials.

Recently there has been enthusiasm for adjuvant chemotherapy in patients with locally advanced tumours. Multiagent chemotherapy has been shown to improve disease-free survival in patients with locally advanced disease. A number of authorities now recommend neo-adjuvant chemotherapy to ‘downstage’ the tumour prior to treatment with surgery or irradiation. Not only does the main tumour shrink, but chemotherapy is delivered to potential micrometastases soon after diagnosis. However, no data are currently available which indicate whether this approach yields benefit in terms of either local control or overall survival.

Ductal carcinoma in situ accounts for only about 5 per cent of all symptomatic breast cancers. It presents either with a mass or with nipple discharge, although it is occasionally a chance finding on biopsy for other reasons. In screening programmes, however, the incidence of ductal carcinoma in situ may be as high as 20 per cent and its relative importance has therefore increased over the last 5 years.

The natural history of ductal carcinoma in situ has been discussed. Most available data relate to symptomatic patients rather than those detected by screening, but it appears that the risk of developing invasive cancer in the former group is about 30 to 50 per cent over a 15-year period. The comedo variant of the tumour has a greater malignant potential than the cribriform, papillary, solid, and micropapillary types. Microinvasion is difficult to evaluate. Recent evidence suggests that this feature does not necessarily imply increased risk of frank malignant changes and may, in fact, be invariably present if searched for assiduously.

Further evidence relating to the malignant potential of ductal carcinoma in situ is available from studies which have found that this cancer occurs in the contralateral breast in up to 50 per cent of women previously treated for breast cancer. However, the risk of developing a frank invasive cancer is only about 20 per cent at 20 years after the diagnosis of the original primary tumour.

One unexplained feature of ductal carcinoma in situ is its high expression of the C- erb2 oncogene. This is present in approximately 80 per cent of all such tumours, but in only 40 per cent of invasive tumours.

The treatment of ductal carcinoma in situ is controversial. When it is extensive in the breast, mastectomy is usually recommended. Such treatment provides local control and survival rates of almost 100 per cent. Treatment of smaller tumours, such as those detected by screening, is the subject of controversy. While some surgeons still prefer mastectomy, others recommend breast conservation and wide excision of the tumour. Adjuvant radiotherapy may reduce the risk of subsequent relapse; the same might also be true for adjuvant tamoxifen although there is no clear evidence to confirm this.

Despite the low incidence of axillary node disease in women with ductal carcinoma in situ it is always necessary to stage the axilla appropriately by limited dissection of its contents. A major drawback of performing local excision is the risk of inadequate resection and subsequent positive margins. Mammographic appearances do not provide a good guide to the extent of ductal carcinoma in situ: even a 2-cm margin of apparently normal breast tissue is associated with an unacceptable inadequate resection rate in screen-detected cancers. The supporters of local excision for ductal carcinoma in situ claim that even if margins of resection are positive for tumour a salvage mastectomy can be performed if relapse occurs. The flaw in this argument is that approximately 50 per cent of patients who develop relapse after wide excision will have invasive cancer. Although the impact of recurrence on survival is unknown, it is clearly a dubious principle to leave a patient with positive margins after local excision for ductal carcinoma in situ. Under these circumstances the author always recommends mastectomy with or without reconstruction.

This has generated less interest than ductal carcinoma in situ in recent years because of the importance of the latter in screening programmes. Furthermore, its malignant potential is less than that of ductal carcinoma: only about 25 per cent of patients demonstrating invasion within 20 years of diagnosis. The risk of developing an invasive cancer is as great in the contralateral breast as in the side in which the original diagnosis was made. Lobular carcinoma in situ is now regarded, therefore, as a mark of cancer risk rather than a true premalignant tumour. It does not express the C- erb2 oncogene.

There are no clinical or radiological features of lobular carcinoma in situ, which is a chance histological finding following biopsy for some other condition. Recommendations for treatment vary from simple monitoring to mastectomy with contralateral mirror image biopsy because of its predisposition to bilaterality. The author suggests that wide excision should be performed. If the specimen obtained shows only a limited amount of fully excised tumour, routine observation is sufficient. If disease is extensive, and particularly if there are other risk factors for breast cancer, mastectomy with or without reconstruction must be considered. There is little merit in the contralateral biopsy.

Breast cancer is a common condition with a long natural history and there is evidence, perhaps controversial, that treatment at an earlier stage improves prognosis. In theory, therefore, breast cancer lends itself to earlier detection by mammography. The guidelines for successful screening have been laid down by the World Health Organization, and most of these criteria are fulfilled by breast screening programmes (Table 7) 278.

Unfortunately, there is no clear consensus as to the treatment of breast cancer, particularly for in situ subtypes of the disease. Mammography can be applied to populations of women, although it is technically sophisticated and may create anxiety. Whether breast cancer screening is truly cost beneficial in terms of overall health care has yet to be determined.

There are four main studies evaluating breast cancer screening: the Health Insurance Plan (HIP) study from New York, two studies from Sweden (the Two Counties and the Malmo studies), and one study from the United Kingdom. In addition there are case-control studies from Holland and Italy, and interim results from other centres such as those from Canada.

These studies provide conflicting results. The HIP study demonstrated a 30 per cent reduction in mortality in the screened group of patients although this was not detected until after 7 years. This study has been criticized in various ways, although it was a remarkable feat when it is considered that it was initiated in the early 1960s, when mammographic techniques were rudimentary. This may explain the small impact provided by mammography in detecting cancers in the screened group. Two hundred and ninety-nine cancers were detected in the study group; 167 of these were interval cancers in that they became clinically apparent between screening and 88 were clinically obvious at the time of screening. In only 44 patients was a mammographic abnormality found which was not clinically apparent.

The Two Counties Study from Sweden was based on randomization by geographic area and there are a number of statistical considerations which may be criticized. However, on balance, it is a large, well constructed study which provides good data in favour of screening.

The second Swedish study from Malmo failed to show a significant mortality difference between the screened and unscreened populations, although this was a relatively small trial and a true effect may have been missed. The United Kingdom study also failed to demonstrate a mortality difference between the screened and unscreened populations at initial observation times, although after 8 years a significant difference did become apparent.

Analysis of the above studies in conjunction with case-control studies and interim results from other trials indicates that screening may reduce the mortality of breast cancer by about 30 per cent. The effect is most obvious in women aged over 50. Few data support screening in women younger than 50, although a significant reduction in mortality did occur in this age group after prolonged observation in the HIP study.

National breast cancer screening programmes have now been initiated in a number of countries, particularly in northern Europe. The age range of women screened and the screening interval depends on the programme concerned. Generally, however, women between 45 and 65 are screened approximately every 2 years. In the United Kingdom, screening is undertaken over the age range 50 to 65, every 3 years. These criteria have been established to provide the maximum reduction in mortality for minimum cost. Initial screening is generally provided by a single oblique mammographic view; if an abnormality is found the patient is recalled for clinical examination and two-plane mammographic evaluation. Some programmes do not rely on clinical palpation as an initial mode of screening because of lack of evidence as to its efficacy.

Breast cancer screening has its critics. The reduction in mortality of 30 per cent translates to an improvement in survival of approximately 10 per cent. If these relative percentages are converted to absolute figures the number of patients in a given population whose lives may be extended by screening is small. Furthermore, the results of the studies on screening may have been influenced by bias. Lead-time bias is the time by which thediagnosis is advanced through screening: only the HIP study considered this factor. Lag-time bias is the tendency of screening programmes to detect tumours of low biological potential. Other biases, such as selection bias for inclusion in screening programmes and over-diagnosis bias (the diagnosis of an abnormality which might not otherwise have been diagnosed in the lifetime of the patient) should also be considered.

Critics of screening claim that its expense and relatively small effect on survival are not cost-effective. They claim that any improvement in survival may result from increased diagnosis of biologically favourable tumours such as ductal carcinoma in situ and those tumours of special type. In routine clinical practice ductal carcinoma in situ accounts for only 5 per cent of all cancers detected; in the screened population it accounts for about 20 per cent of all tumours diagnosed. Similar figures pertain to ‘special’ tumours with a good prognosis, such as tubular and medullary cancers. Whether these criticisms are justified cannot be addressed at the present time.

After the initial screen, in which 20 to 30 per cent of cancers can be clinically detected, screening is therefore particularly directed to the detection and treatment of small tumours with a favourable prognosis, and ductal carcinoma in situ. Lymph node metastases in this population should therefore be relatively uncommon: they occur in only about 15 per cent of patients as compared to 45 per cent in the unscreened population. However, as has been discussed above, there are doubts as to the ideal treatment of these screen detected cancers, especially ductal carcinoma in situ. Diffuse, extensive ductal carcinoma in situ is usually treated by mastectomy with or without reconstruction. Localised disease can be treated by wide excision; the role of adjuvant radiotherapy or tamoxifen after this treatment is, however, open to speculation and is currently the subject of a randomized trial in the United Kingdom. There are also doubts as to whether adjuvant radiotherapy is required following wide excision of small favourable tumours. This is also the subject of a randomized trial.

If breast cancer screening is to be successful it must be centrally controlled and funded, performed by specially trained personnel, and be associated with a quality assurance programme. Adequate public information is also of prime importance. If screening is performed in less than 70 per cent of the target population its impact is likely to be small. If performed properly with due attention to quality assurance, screening is likely to be of benefit to women in late middle-age.

The follow-up of women treated for breast cancer is generally somewhat haphazard with varying intervals of review, no protocol for investigation, and an arbitrary time of discharge—usually 5 years. This confusion is compounded by lack of evidence that prognosis is improved with an intensive, investigation—orientated, hospital observation compared to that of the primary care physician.

Breast cancer patients should receive emotional and psychological support as well as physical evaluation. The latter is directed towards the detection of local regional recurrence as well as distant metastases, although it should be appreciated that two-thirds of recurrences are symptomatic and noted by the patients between hospital visits.

The evaluation of local regional recurrence should include detection of new primary breast cancers. Mammography is therefore recommended every 1 to 2 years after initial diagnosis. Local recurrence in the breast or in regional lymph nodes can be treated or at least palliated with further surgery, irradiation, or chemotherapy.

Investigation for early, asymptomatic distant metastases is much less efficacious. Bony metastases are the most frequent type of relapse, and some authorities have recommended bone scintigraphy at yearly intervals, or less often. The yield of such investigations is low, the expense is considerable, and there is little evidence that detection and treatment of such asymptomatic recurrence improves the prognosis over that obtained when treatment delayed until the patient complains of symptoms. Furthermore, false-positive investigations provide a constant source of uncertainty. The argument for routine tests such as chest radiographs and liver scanning is even less convincing than that for bone scintigraphy.

In our own Unit we see patients at 3-monthly intervals for the first year. Stage I patients are then seen 6-monthly for a year and yearly thereafter. Stage II and III patients are seen 3-monthly for 2 years, 6-monthly for the third year, and then yearly. This is somewhat arbitrary. The time of discharge is also empirical, but for research purposes we monitor patients for 10 years, although the accumulation of large numbers of patients may well limit this ideal with the passage of time.

The arbitrary description of the prognosis of breast cancer as a survival rate after an indeterminate amount of time is somewhat naïve. First, the natural history of breast cancer is uncertain and survival, even after the development of metastases, may be prolonged without treatment. Second, recurrence and subsequent death may occur as long as 25 years after the original diagnosis and, finally, if survival is prolonged death from other causes must be taken into account. Despite these limitations the analysis of survival demonstrates that about 80 per cent of patients with Stage I disease are alive after 5 years, whereas at 10 years this figure falls to 50 per cent. For Stage II disease survival rates are 50 per cent and 35 per cent, respectively. Overall survival, including those with Stage III and Stage IV disease at presentation, is likely to be less than 25 per cent after 10 years.

Critics of this method of survival analysis recommend an alternative method (Fig. 19) 839. Curve A describes the survival characteristics in an age-matched population without breast cancer. Curve D is that of the population with breast cancer. Initially the mortality from breast cancer is excessive and curve D diverges from A. With the passage of time deaths from breast cancer become less common and the two curves become more parallel. True parallelism indicates that the risk of death in the population indicated by curve D is no greater than that of A and the group, as a whole, can be considered cured. Such analysis with long-term follow-up, however, shows an increased death rate from breast cancer even 30 years after initial diagnosis, although some of this long-term mortality is due to death from second breast cancers and the cardiopulmonary effects of irradiation.

Curve D describes the survival of two distinct groups. Patients in curve B have an intrinsically good prognosis. Curve C, however, represents the poor prognosis group. Unless their outlook can be improved it may be that the overall survival of symptomatic breast cancer patients will remain unaltered.

Prognostic factors have important implications for both the breast cancer patient and for her medical attendant. Not only do they act as a guide to overall prognosis, but they may also determine the need for adjuvant treatment. The latter is of current interest in that it may define a subgroup of patients with apparently favourable node-negative tumours who nonetheless have a poor prognosis and to whom adjuvant therapy may be specifically targeted.

The major arguments relating to prognostic factors are their inter-relationships and relative importance. The relative importance of clinical as opposed to pathological factors is also controversial: there is inherent inaccuracy in clinical assessment of features such as tumour size and axillary nodal involvement. Overall clinical stage is, therefore, only a relatively poor prognostic indicator, and is less important than specific pathological and certain biochemical parameters. Other clinical factors, such as patient age, menopausal status, obstetric history, and site of tumour in the breast have not been shown to be of much prognostic importance.

Prognostic factors can be determined by a multivariant analysis of the various clinical and pathological features of a group of patients with breast cancer. There is some discrepancy relating to the relative importance of prognostic features between studies although most show that axillary nodal disease, tumour size, and differentiation are the most important (Table 8) 279.

The question of interdependence is difficult to address. An obvious example is tumour differentiation, which may be regarded purely as an expression of the inherent biology of the tumour. Thus a poorly differentiated tumour has unfavourable biological characteristics and is likely to be large at presentation, to be associated with axillary nodal involvement and negative receptor status, and to have a poor prognosis. On the other hand, many studies have shown that tumour differentiation is an independent prognostic variable in its own right and acts as a determinant independent of factors such as tumour size and nodal disease.

Nearly all studies show that the presence of axillary lymph node metastases is the most important prognostic determinant. Prognosis is a function of the number of lymph node metastases, although other factors such as level of disease, extranodal spread, or size of tumour metastases may be of some minor significance. Thus, isolated tumour cells seen in axillary nodes using histochemical techniques represent the most minor degree of nodal involvement. The significance of these features is unclear. A single micrometastasis less than 2 mm in diameter is associated with an overall prognosis similar to that of node-negative disease. Disease affecting one node is also associated with a prognosis similar to that of node-negative disease, although after 10 to 12 years such spread may be associated with a worsening prognosis. Disease affecting up to three lymph nodes classifies the patient in a relatively favourable subgroup of stage II. Those with four to 12 nodes affected form an intermediate group; the prognosis is decidedly poor if more than 12 lymph nodes exhibit metastatic disease.

Tumour size is the second most important prognostic determinant. Although tumour size and lymph node metastasis are closely related they are also independent of each other in terms of prognosis. The influence of tumour size is most obvious when lymph nodes contain metastases and is probably of greatest importance in the first 5 years after the initial diagnosis.

The grading of invasive breast cancer has been regarded as notoriously difficult because of varying interpretation by pathologists. Some recent experience indicates that specialists in this field can, however, obtain a remarkable degree of agreement. Grade is classified as I, II, or III on the basis of tubule formation, nuclear pleomorphism, and mitotic rate. Mitotic rate is the most powerful factor.

About 20 per cent of cases are classified as well-differentiated (Grade I) although there is some variation between centres. Forty per cent of patients are classified as intermediate (Grade II), and the remainder demonstrate poor differentiation (Grade III). Investigations such as those of the Cancer Research Campaign and the Nottingham Tenovus Breast Cancer Study group clearly demonstrate a relationship between grade and prognosis, which is independent of other variables.

A difficult problem relating to histological grade is its correlation with the morphological type of the tumour, as there is a relationship between tumour type and prognosis. Special types of breast cancer such as pure tubular, mucinous, and invasive cribriform carcinoma have an excellent prognosis while infiltrating lobular and medullary carcinoma appear to have an intermediate prognosis between the first group and the common infiltrating ductal carcinoma of no special type. It could be argued that histological grading is appropriate only to invasive ductal cancer of no special type, but this would result in important prognostic information being unavailable for up to 25 per cent of patients.

The grading of classic lobular cancer is also difficult because of the nature of the tumour cells and lack of histological pattern in this particular type of cancer.

The importance of axillary nodal disease, tumour size, and differentiation has resulted in a number of successful attempts to define a ‘prognostic index’ using this information.

It has been known for many years that certain patients with breast cancer respond to endocrine therapy and have a more indolent form of the disease. Such endocrine sensitive tumours can be identified by measurement of oestrogen or progesterone receptors within the cancer itself. The oestrogen receptor is a cytosol protein that can be identified using a variety of methods. Previously, the most common method of detection involved the incubation of the supernatant fluid of a tissue homogenate with radiolabelled oestrogen. The unbound label is then removed by dextran-coated charcoal or sucrose density centrifugation and an assay performed. More recent methods use monoclonal antibodies against the receptor; such methods require smaller specimens although they are difficult to perform.

When cancer cells with intact and functional oestrogen receptors are exposed to oestrogen the latter is transported to the site of mRNA production in the nucleus. The result of this interaction includes the formation of additional oestrogen and progesterone receptors and other growth factors that help dictate cell characteristics.

The presence of oestrogen or progesterone receptors within the cell has been associated with an improved short-term prognosis in some, but not all, studies. Although they provide some prognostic information they are somewhat weak indicators of overall prognosis and not as important as tumour size, nodal status, or degree of differentiation.

Any improvement in prognosis of oestrogen receptor-positive tumours may be due to improved disease-free survival, improved survival after the first demonstration of metastases, or both. There is conflicting evidence as to whether oestrogen receptor status provides the differential information for premenopausal as opposed to postmenopausal women and for node-negative as opposed to node-positive disease.

The relative importance of oestrogen and progesterone receptors is difficult to assess. Some studies indicate that the presence of progesterone receptor is an inferior prognostic indicator compared to oestrogen receptor. On the other hand some authorities have demonstrated progesterone receptor to be as good a prognostic indicator as node status or tumour size. If oestrogen and progesterone receptors are both present the prognostic information is likely to be more powerful than if only one of these factors is present.

Other histological features such as vascular or lymphatic invasion in the breast, perineural invasion, tumour necrosis, or mucin production may be of prognostic importance although their clinical significance is open to doubt.

The identification of high-risk patients and their potential for targeting this specific adjuvant therapy has resulted in the recognition of other factors which may have prognostic importance. Areas of current investigation are summarized below.

ErbB&sub2; oncogene

Cathepsin D

Collagenase activity

p53

S-phase fraction

CD44 glycoprotein

Time will tell if any of these current areas of research have any bearing on long-term prognosis of patients with breast cancer.

Recurrence of breast cancer is of ominous significance. It is incurable, with a median survival of about 15 months although there is a wide range lasting from only a few weeks to many years. The role of treatment in metastatic disease is two-fold: to alleviate symptoms and to prolong survival. Despite a plethora of new drugs and therapeutic options there is little evidence that overall survival has been improved once metastases have occurred, although the duration and quality of life in patients responding to therapy have improved.

The management of metastases demands a multimodality approach by surgeons, radiotherapists, and medical oncologists. The role of careful diagnostic radiology in assessing the extent of disease cannot be over-emphasized: appropriate counselling is particularly important, especially for mothers with young families, whose overall outlook must be frankly discussed.

Once a patient has presented with recurrence it is necessary to determine the full extent of metastatic spread. This requires full and careful examination of the site of the original primary tumour and its regional lymph nodes. A full series of staging investigations, including chest radiography, abdominal ultrasound, and bone scintigraphy are also necessary to determine the presence of other sites of occult metastatic disease. Brain scanning is not recommended: the likelihood of this investigation detecting asymptomatic recurrence is small.

There is little evidence that tumour markers such as carcino-embryonic antigen have any significant role in assessment of metastatic breast cancer or the evaluation of response to treatment.

A full series of staging investigations helps to determine the appropriate therapy. It also allows a more accurate appraisal of response to treatment.

The most important prognostic indicator in metastatic disease is disease-free interval after diagnosis of the primary tumour. The median survival of women who relapse in a single site within 1 year of diagnosis is about 11 months; those who experience a disease-free interval of more than 5 years have a survival of up to 40 months.

The site of metastatic disease is also important: prognosis is more optimistic in patients with local regional or bony metastases than in those with recurrence in the liver or central nervous system. The number of metastatic sites is also a factor: survival of patients with three sites of metastases is only 50 per cent of those with one affected site.

Some prognostic information may be obtained from knowledge of oestrogen and progesterone receptor status, although this is not as important as the factors discussed above. However, such information may be important in determining treatment: patients positive for such receptors are more likely to respond to endocrine manipulation.

The treatment of metastatic disease at individual sites is discussed below. However, as the majority of patients with metastases can be regarded as suffering from widespread systemic disease, the use of systemic therapy is nearly always appropriate. Breast cancer is relatively unusual in being responsive not only to radiotherapy and chemotherapy but also to various endocrine manipulations.

The mechanism of response to endocrine therapy is unclear. The simplistic explanation is that breast cancer requires oestrogen for growth and that inhibition of this hormone's action will produce a response. This explanation, however, fails to explain why a given breast tumour can respond to both oestrogen withdrawal and addition. It also fails to explain how breast cancer may respond to other therapeutic manipulations involving progestogens, androgens, and corticosteroids, when there is little evidence that they are involved in tumour growth.

Despite the difficulties in understanding the mechanisms of endocrine therapy it appears that its effect is mediated by receptors, especially those for oestrogen. The overall response to endocrine therapy in breast cancer is about 30 per cent. However, it may reach 60 per cent in those positive for oestrogen receptors, or be as low as 10 per cent in those negative for such receptors. The higher the level of oestrogen receptors the greater the likelihood of response. If the level of progesterone receptors is also elevated the response rate is likely to be increased.

A wide range of endocrine therapies is available (Table 9) 280. Some are pharmacological agents; others require surgical manipulation. The latter were frequently adopted up to the 1960s, but in the last 20 years oral agents have been used with increasing frequency. Apart from their ease of administration, pharmacological agents have a much broader application, as many of the surgical options were limited to premenopausal patients only.

In general there is little evidence that response rates to or survival benefit from one type of endocrine therapy are superior to those of any other, although androgens have fallen into disfavour because of their lower activity and side-effects. The choice of endocrine therapy is therefore based on ease of administration and freedom from side-effects. As a result the anti-oestrogen, tamoxifen, has become the treatment of choice for first-line endocrine therapy. Younger patients in particular, however, are prone to suffer hot flushes and weight gain with this treatment. Randomized trials have shown tamoxifen to be as effective as oophorectomy, aminoglutethimide, and progestogens.

Progestogens are as active as tamoxifen as a first-line agent. They are also a good second-line treatment after failure of response to tamoxifen in both premenopausal and postmenopausal patients. Side-effects, particularly weight gain, fluid retention, and nausea can be a problem.

Aminoglutethimide is as active as tamoxifen as a first-line treatment and, in some studies, has shown to be particularly effective against bone metastases. Dosage should be slowly increased but at the normal therapeutic dose (250 mg four times daily) supplemental hydrocortisone (30 mg daily) and even mineralocorticoid is required. Cushinoid and addisonian states can therefore accompany aminoglutethamide therapy. Other side-effects, such as rashes, nausea, and drowsiness somewhat limit its use, although it can be useful in premenopausal patients after failure of tamoxifen and progestogens.

Several authorities have recently described success with luteinizing hormone-releasing hormone agonists such as Zoladex. It is likely that this group of drugs will be used more often in premenopausal patients with metastatic breast cancer.

The success of the various oral agents has reduced enthusiasm for surgical endocrine therapy: adrenalectomy and hypophysectomy have almost disappeared from use. Some authorities still recommend oophorectomy with enthusiasm, and recent data relating to its potential role as an adjuvant in premenopausal Stage II disease suggests that its future may be assured. Whether oophorectomy will be replaced by the the newly introduced luteinizing hormone-releasing hormone agonists remains to be seen. The suggestion that a combination of endocrine therapies may improve survival has not been demonstrated in randomized studies, although higher response rates may be seen.

Tamoxifen is, therefore, the ideal first-line endocrine therapy, with an overall response rate of about 30 per cent (up to 60 per cent in those positive for oestrogen receptors). Failure to respond is an indication for abandoning endocrine therapy. Relapse following an initial response can be treated with a progesterone, although the response rate to such second-line treatment is only about half that seen with first-line therapy. Subsequent treatment failure may be treated by aminoglutethimide in premenopausal women and aminoglutethimide or oestrogen in postmenopausal patients, although response rates continue to diminish. Oestrogens must be used with care in patients with bone metastases because of the risk of hypercalcaemia. If oophorectomy has a role it is probably as a second-line treatment after a good response to tamoxifen in premenopausal patients. Androgens are now rarely used in patients with metastatic breast cancer.

A characteristic, but relatively unusual, feature of response to hormonal therapy is ‘flare’, the mechanism of which is unclear. It is characterized by an exacerbation of symptoms associated with erythema and swelling after the initiation of endocrine treatment. The symptoms resolve within 1 month.

About 60 per cent of patients with previously untreated metastatic breast cancer respond to appropriate chemotherapeutic protocols, although complete resolution of disease is seen in fewer than 20 per cent. The median duration of response is about 1 year, although this depends on the site of metastatic disease: better responses are seen in soft tissue and bone than in the liver or central nervous system. The time of response is also variable, depending on the site of disease. Cutaneous or lymphatic metastases may respond within 3 to 6 weeks; response may not be seen until after 4 or 5 months if metastases are in bone.

Overall response rates to the more commonly used chemotherapeutic agents used in breast cancer are shown in Table 10 281. The most active agents are anthracyclines such as Adriamycin. In some, but not all, studies Adriamycin alone has produced a response rate approaching 60 per cent; in other investigations it has been shown to be as active as combination chemotherapy with cyclophosphamide, 5-fluorouracil, and methotrexate.

Vincristine has been widely used in combination therapy for metastatic breast cancer but no advantage over regimens in which it is lacking has been demonstrated. Its advantage is that, unlike other commonly used agents, it is not myelosuppressive.

Mitomycin C demonstrates substantial activity against metastatic breast cancer and has even been shown to induce responses in tumours resistant to Adriamycin and cyclophosphamide. It is most commonly used in combination with methotrexate and mitozantrone (MMM) as a second-line treatment.

Combination chemotherapy has enhanced activity, albeit at the expense of a higher incidence of side-effects. The most commonly used regimen is the combination of cyclophosphamide, methotrexate, and 5-fluorouracil (CMF), which has an overall response rate of up to 60 per cent. The addition of vincristine and prednisone to this combination does not significantly enhance the response obtained. However, the substitution of methotrexate with Adriamycin (CAF) is associated with an increased response rate of prolonged duration, although with more side-effects.

Adriamycin, either alone or in combination with cyclophosphamide and 5-fluorouracil (AF), has been used mainly as a second-line treatment following failure of initial chemotherapy, or in the management of younger patients with a poor prognosis, such as those with liver metastases. Its toxicity precludes its use in older patients.

A more recently adopted combination is based on mitomycin C, methotrexate, and mitozantrone (MMM). This has potential value both as a primary management of metastatic breast cancer and a second-line treatment.

There has been a natural tendency to combine chemotherapy with endocrine treatment. Although response rates and duration of response may be enhanced the only overall benefit is small and as a result such combinations are not recommended. In theory at least the endocrine treatment may slow down the cell cycle, thus reducing the efficacy of the chemotherapy.

Recurrence in the remaining breast tissue after partial mastectomy or on the chest wall after total mastectomy has been discussed above. Local recurrence must be distinguished from regional recurrence: the latter relates to recurrent disease in the adjacent lymphatic field. Regional recurrence is always of clinical importance and implies a worsening of prognosis. However, it is particularly important if it occurs in a treated rather than an untreated field. The management of regional recurrence depends on the site of relapse. If it is in the internal mammary or supraclavicular areas radiation can be used, followed by endocrine or cytotoxic therapy as appropriate. Surgery has little part to play in the treatment of relapse at these sites.

Treatment of axillary relapse is by full surgical clearance, if this has not already been performed. Clearance of a previously irradiated axilla is likely to cause arm lymphoedema. Relapse occurring after clearance is treated by radiotherapy or systemic treatment.

The liver is a common site of metastatic spread of breast cancer. Such metastases are characteristically multiple and surgical resection has no role in their management. Resection is not even indicated in the unusual situation of a single hepatic metastasis. The prognosis of hepatic metastases is poor, and aggressive therapy is indicated. Hepatic metastases do not respond well to hormone therapy. In young patients with good performance status, combination chemotherapy, including Adriamycin, is appropriate as first-line therapy. Older patients may only be suitable for endocrine therapy, although this is not likely to be so efficacious.

Peritoneal metastases may cause acites or intestinal obstruction. Such metastatic spread is said to be particularly common in patients with lobular carcinoma. Systemic combination chemotherapy is required if it can be tolerated. Surgical intervention should be avoided if possible; it may be required to relieve obstruction, or for insertion of a peritoneal-systemic shunt if ascites becomes intractable.

Occasionally, isolated metastases are seen in the ovary or the adrenal gland. If the retroperitoneum is involved, however, it is usually a manifestation of widespread carcinomatosis including the peritoneal cavity. Renal failure due to obstructive uropathy is a common cause of death in these patients.

Metastatic disease to the lungs is often multiple and not amenable to surgical resection or radiotherapy. Systemic combination chemotherapy is indicated, although endocrine treatment is more appropriate for older poor-performance patients. Metastatic spread causing lymphangitis can cause distressing dyspnoea. The diagnosis can be difficult in its early stages because it produces few changes on chest radiographs. Combination chemotherapy should be instituted if possible, with the addition of steroids for symptomatic relief.

Widespread pleural metastases may cause an effusion; if recurrent and symptomatic they should be tapped. Pleurodesis is indicated if pleural effusions are persistent.

Bony metastases are the most common form of secondary spread of breast cancer. These are usually at multiple sites, although it is not unusual for a solitary tumour metastasis to cause symptoms. Pain and pathological fracture are the most common forms of presentation.

A short course of radiotherapy will relieve the symptoms of bony metastases. Systemic therapy is indicated at the same time because of the high risk of developing recurrences at multiple sites, which may not be clinically or radiologically apparent. Endocrine therapy with tamoxifen is appropriate first-line treatment, especially in older patients; failure to respond is an indication for chemotherapy. Oestrogen therapy is contraindicated in patients with widespread bony metastases because of the risk of hypercalcaemia; conversely aminoglutethimide has been shown to be particularly active.

Pathological fractures require appropriate orthopaedic evaluation and surgical fixation where necessary. Widespread bony metastases, even if asymptomatic, may cause hypercalcaemia, with symptoms of fatigue, weakness, vomiting, and eventually coma with renal failure. Initial treatment involves rehydration with forced diuresis and administration of steroids to enhance calcium excretion in the urine. Resistant cases can be treated with either mithramycin or one of the diphosphonate drugs; both reduce serum calcium by inhibiting osteoclastic bone reabsorption.

Metastases may affect the brain itself, the cord, the meninges, or the epidural space, where they may cause cord compression. The symptoms resulting from such spread are therefore very varied, and central nervous system metastases must be considered in any patient with neurological signs or symptoms. Their treatment relies on radiation therapy. Surgery has a relatively small role, except for patients with cord compression: excision of brain metastases has no advantage over radiation therapy alone. Systemic therapy has not been widely used as most agents fail to cross the blood–brain barrier, although recent experience suggests that combination chemotherapy with steroids may have some activity in these patients.

The significance of occult bone marrow metastases is unclear. If symptomatic they are often associated with bony deposits. Treatment is complex because of potential leucopenia and thrombocytopenia resulting from both marrow replacement and the effects of chemotherapy. The choice of endocrine or chemotherapy is made on the basis of performance status, age, and ease of adminstration.

Acute presentation to the ophthalmologist with sudden painless blindness is not uncommon and may be due to a choroidal metastasis. Treatment is with local radiotherapy.

Other ocular symptoms may be produced as a result of metastatic disease to the bony orbit, the orbital contents, the optic nerve, or to the optic pathways within the brain itself.

Psychological sequelae following treatment of breast cancer

Women suffer psychiatric morbidity for at least 1 year after the diagnosis of breast cancer. Approximately 20 per cent of women demonstrate a significant departure from their normal mood 12 months after diagnosis compared with only 8 per cent of women who have undergone biopsy for benign disease.

The major symptoms relate to a combination of depression and anxiety, along with associated sleep disturbance, impaired concentration, fatigue, and various somatic symptoms such as headache and palpitations. About one in 10 of these patients have symptoms severe enough to warrant inpatient treatment.

Preoperative counselling and appropriate support in both the perioperative and postoperative periods is mandatory in the management of breast cancer. Specially trained health care professionals, including breast care nurses, social workers, and clinical psychologists should provide an integrated support network in conjunction with nursing and medical personnel. This is of particular importance if radiotherapy or chemotherapy are to be considered. The routine provision of such support results in improved quality of life; it may even favourably influence survival. Physiotherapy while in hospital reduces symptoms due to arm stiffness and may well speed rehabilitation. It is important to ensure that patients undergoing mastectomy without reconstruction are also introduced to surgical appliance specialists so that a prosthesis may be fitted while in hospital.

There is some controversy over the impact of a formal programme of psychological support in patients with breast cancer. Some controlled studies have failed to demonstrate a reduction in psychiatric morbidity following perioperative counselling. There is even some evidence that excessive counselling may lead to ‘sensitization’ of the patient and increased anxiety or depression, at least in the short term.

One area of particular interest relates to the psychological sequelae of mastectomy itself. It was hoped that partial mastectomy, with breast conservation, would reduce psychological morbidity. However, this does not appear to be the case: there is a similar incidence of psychological problems after both total and partial mastectomy. Fear of mastectomy is inextricably linked to that of the cancer itself, and variations in treatment have relatively little impact in reducing psychological morbidity. However, it must be remembered that mastectomy, as opposed to breast conservation, may devastate the lives of a substantial proportion of women.

Breast cancer is nearly always a sporadic event; a true genetic factor is present in less than 10 per cent of cases. The majority of individuals with risk factors therefore do not develop breast cancer and the majority of patients with the disease are not at excessive risk. This makes prevention of breast cancer difficult.

Demographic studies show breast cancer to be particularly a disease of the Western world, and perhaps a result of the life-style adopted by those individuals in whom there is a high incidence. It is clearly difficult to change life-style factors such as age of having children, diet, and use of exogenous hormones, and attempts to reduce the incidence of breast cancer by influencing these factors may fail.

Two areas of prevention require special mention. It has been suggested that the low incidence of breast cancer in Japan may be secondary to the very low fat diet eaten in that country. In theory, reducting the amount of fat eaten by populations at risk may reduce the cancer incidence. However, it is very difficult for people of Western culture to reduce the amount of fat in their diet to that of endogenous Japanese.

Studies on adjuvant tamoxifen have shown a significant reduction in the incidence of contralateral breast cancer. Prophylactic chemosuppression with tamoxifen, especially for ‘high-risk’ patients, has therefore been suggested, and this is the subject of a number of trials currently in progress. An alternative method of reducing the incidence of breast cancer might be that of chemoprevention with agents such as the retinoids or selenium.

Occasionally very high-risk patients will ask their medical attendant for prophylactic mastectomy, with or without reconstruction, to avoid breast cancer. There is undoubtedly the occasional case where this is justified but if possible this rather aggressive approach should be avoided.

Atkins H, Hayward JL, Klugman DJ, Wayte AB. Treatment of early breast cancer: a report after ten years of a clinical trial. Br Med J 1972; 2: 423–429.

Baum M. The curability of breast cancer. Br Med J 1976; 1: 439–442.

Brinkley D, Haybittle JL. Long term survival of women with breast cancer. Lancet 1984; 1: 1118.

Chu KC, Smart CR, Tarone RE. Analysis of breast cancer mortality and stage distribution by age for the Health Insurance Plan Clinical Trial. J Natl Cancer Inst 1988; 80: 1125–1132.

Cuzick J, et al. Overview of randomised trials comparing radical mastectomy without radiotherapy against simple mastectomy with radiotherapy in breast cancer. Cancer Treat Rep 1987; 71: 7–14.

Cuzick J, et al. Overview of randomised trials of post-operative adjuvant radiotherapy in breast cancer. Cancer Treat Rep 1987; 71: 15–29.

Early Breast Cancer Trialists' Collaborative group. Effects of adjuvant tamoxifen and of cytotoxic therapy on mortality in early breast cancer. N Engl J Med 1988; 319: 1681–1692.

Early Breast Cancer Trialists' Collaborative Group. Systemic treatment of early breast cancer by hormonal, cytotoxic, or immune therapy. Lancet 1992; 339: 1–15.

Early Breast Cancer Trialists Collaborative Group. Systemic treatment of early breast cancer by hormonal, cytotoxic, or immune therapy. Lancet 1982; 33: 71–85.

Fentiman I. Ductal carcinoma in situ. Br Med J 1992; 304: 1261–1262.

Fentiman IS, Mansel RE. The axilla: a no-go zone. Lancet 1991; 337: 221–223.

Fisher ER, et al. Pathologic findings from the NSABP (Protocol No.4). V. Significance of axillary nodal micro- and macrometastases. Cancer 1978; 42: 2032–8.

Fisher ER, Sass R, Fisher B. Pathologic findings from the National Surgical Adjuvant Project for Breast Cancers (Protocol No.4). Cancer 1984; 53: 712–23.

Fisher B, et al. Eight year results of a randomised clinical trial comparing total mastectomy and lumpectomy with or without irradiation in the treatment of breast cancer. N Engl J Med 1989; 320: 822–828.

Haagensen CD, et al. Treatment of early mammary carcinoma: a co-operative international study. Ann Surg 1969; 170: 875–78.

Haagensen CD, Lane N, Lattes R, Bodian C. Lobular neoplasia (so-called lobular carcinoma in situ) of the breast. Cancer 1978; 42: 737–769.

Harris JR, Lippman ME, Veronesi U, Willett W. Medical Progress. Breast Cancer. N Engl J Med 1992; 327: 319–327.

Hayward JL. The Guy's trial of treatments of early breast cancer. World J Surg 1977; 1: 314–316.

Holland R, Veling SHJ, Mravunac M, Hendriks JHCL. Histologic multifocality of Tis, T1–2 breast carcinomas. Cancer 1985; 56: 979–990.

McGuire WL. Adjuvant therapy of node negative breast cancer. DeVita VT. Breast cancer therapy: exercising all our options. N Engl J Med 1989; 320: 525–529.

Page DL, Anderson TJ. Diagnostic Histopathology of the Breast. Edinburgh: Churchill Livingstone 1987.

Roberts MM, et al. Edinburgh trial of screening for breast cancer: mortality at seven years. Lancet 1990; 335: 241–246.

Schnitt SJ, Silen W, Sadowsky NL, Connolly JL, Harris JR. Ductal carcinoma in situ (intraductal carcinoma) of the breast. N Engl J Med 1988; 318: 898–903.

Shottenfeld D, et al. Ten year results of the treatment of primary operable breast carcinoma. Cancer 1976; 38: 1001–1007.

Smith I. Adjuvant tamoxifen for early breast cancer. Br J Cancer 1988; 57: 527–528.

Steele RJC, et al. The efficacy of lower axillary sampling in obtaining lymph node status in breast cancer; a controlled randomised trial. Br J Surg 1985; 72: 368–369.

Tabar L, et al. Reduction in mortality from breast cancer after mass screening with mammography. Lancet 1985; i: 829–832.

Tabar L, Fagerberg G, Duffy S, Day N. The Swedish Two County Trial of mammographic screening for breast cancer: Recent results and calculation of benefit. J Epidemiol Community Health 1989; 43: 107–14.

UK Trial of early detection of breast cancer group. First results on mortality reduction in the UK Trial of early detection of breast cancer. Lancet 1988; ii: 411–416.

Veronesi U, et al. Breast conservation is the treatment of choice in small breast cancer: long term results of a randomised trial. Eur J Cancer 1990; 26: 668–670.

Wilson RE, Donegan WL, Mettlin C, Smart CR, Murphy GP. The 1982 National Survey of Carcinoma of the Breast in the United States by the American College of Surgeons. Surg Gynecol Obstet 1984; 159: 309– 318.