F. E. Alexander

Edinburgh trial of screening for breast cancer : mortality at seven years

Between 1979 and 1981, 45,130 women in Edinburgh aged 45-64 were entered into a randomised trial of breast cancer screening by mammography and clinical examination. The initial attendance rate was 61% but this varied according to age and socioeconomic status and decreased over succeeding years. The cancer detection rate was 6.2 per 1000 women attending at the first visit; the rate fell to around 3 per 1000 in the years when mammography was routinely repeated and to around 1 per 1000 at the intervening visits with clinical examination alone as the screening method. After 7 years of follow-up the mortality reduction achieved was 17% (relative risk = 0.83, 95% CI 0.58-1.18), which was not statistically significant, even when corrected for socioeconomic status. In women aged 50 years and over a mortality reduction of 20% was achieved.

Community lifestyle characteristics and risk of acute lymphoblastic leukaemia in children

High rates of leukaemia in children and young people have been associated with features of community isolation and population growth. Incidence data collected by two specialist registries were used to compare incidence rates at ward level with relevant ward characteristics derived from routine census and Ordnance Survey data for England and Wales. An excess risk of childhood acute lymphoblastic leukaemia (ALL) was found for wards which are farthest from large urban centres. The excess was greatest for wards of higher socioeconomic status and for children aged 1-7 years (the childhood peak), for which a two-fold excess was seen. These findings in general support the hypothesis that childhood leukaemia has an infectious aetiology.

The Edinburgh randomised trial of screening for breast cancer: description of method.

Edinburgh was selected as one of the centres in the UK Seven-year Trial of Breast Screening of women aged 45-65 which began in 1979. Subsequently, our study was extended to a randomised trial with its own control population within the city. Half the practices were randomly allocated for screening, giving a cluster sampling of women. The total number in the trial is 65,000. Women with previously diagnosed breast cancer are excluded. Women allocated for screening are invited to the clinic and screened according to the procedures specified in the U.K. protocol, having clinical examination every year and mammography on alternate years. The two modalities of screening are assessed independently and the role of nurses is being evaluated. Breast cancer incidence is monitored by pathology register and the local cancer registry office and deaths from the General Register office. Long-term follow-up will be obtained through flagging at NHS Central Register. To determine the value of screening, standard statistical methods will be used to compare breast cancer mortality rates in the whole of the screening population with that of the controls. This trial has a power of 83% of detecting a reduction in mortality of 35% after 7 years of follow-up and a power of 95% of detecting a similar reduction at 10 years (alpha = 0.05, one-sided test).

COMPARATIVE PATHOLOGY OF PREVALENT AND INCIDENT CANCERS DETECTED BY BREAST SCREENING: Edinburgh Breast Screening Project

In the Edinburgh Breast Screening Project 210 cancers were detected from commencement in 1979 up to December, 1984. By this time the full initial cohort had completed at least 3 visits and a proportion had attended for up to 5 visits, so pathological characteristics for prevalent and incident cancers could be compared. The main differences are in distribution of histological type of cancer, detection of occult invasive disease, and lymph-node positivity among incident tumours. Only the first of these was statistically significant. This evaluation shows that cancer detection by screening in Edinburgh conforms with screening theory, in which detection of good prognosis tumours is favoured at the prevalence screens, and faster growing, aggressive tumours are found at the incidence screens. Qualitative histopathology may provide a better measure than standard quantitative judgments of size and lymph node status to compare the varieties of cancer detected by screening programmes and to understand the biology of the disease.

Space-time clustering of childhood acute lymphoblastic leukaemia: indirect evidence for a transmissible agent.

Despite numerous anecdotal reports of small clusters of cases of childhood leukaemia, formal statistical analyses have yielded equivocal results (Linet, 1985). Incidence data from the UK national childrens tumour registry (CCRG) for 1968-1983 have recently become available for small area analyses by location at diagnosis (OPCS, 1991). Extensive analyses using a variety of methodologies have shown consistent, though weak, evidence of the occurrence of both spatial clustering and space-time interactions. Results from one of these analyses (Alexander, 1991) are now extended to test specific prior hypotheses generated by an independent case-control study (Alexander et. al., 1992). These suggested that transmission of a specific, though unknown, agent (Z) plays some role in the development of childhood acute lymphoblastic leukaemia (ALL) with the times when children are susceptible to infection differing by age-of-onset and hence subtype of ALL. For cases with older onset (aged 5 years and over) it was suggested that persistent infection may have been established in utero or early infancy and, now, formal testing of appropriate space-time interactions provide indirect confirmation of this (P = 0.0002). More recent exposure to Z may contribute to ALL in the childhood peak years (Alexander et. al., 1992) but the confirmation provided here is less strong (P = 0.05). The results afford new impetus to a search for a transmissible aetiologic agent or agents; these need not be rare and the results should not be interpreted as evidence for direct case to case transmission.

Pathology characteristics that optimize outcome prediction of a breast screening trial

The ability of pathology characteristics to predict outcome was tested with the 1029 cancers accumulated in the Edinburgh Randomized Trial of breast screening after 14 years follow-up. The majority (55.7%) were in the screening arm, which also had more operable cases (81.3% vs 62.2%); the reduction in the proportion of inoperable breast cancers in a UK female population invited to mammographic screening is a notable effect of the trial. In the 691 operable invasive cases the size, histological type, grade, node status and node number group individually showed highly significant (P< 0.001) association with survival. In multivariate analysis the Nottingham Prognostic Index (NPI) derived from these features showed highly significant association with survival (P< 0.001). However, when first adjusted for NPI, combined addition of pathological size in 6 categories and histological type as special or not had an independent association with survival that was statistically firmly based (P< 0.001). For operable breast cancer the gains are in smaller sizes, better histological features, and higher proportion node negative. The weighting factors applied to pathology indicators of survival in the NPI are not optimal for a population included in a trial of screening. In particular, a linear trend of the index with pathological size is not appropriate. Inclusion of histological type as special or not improves the index further.

A specialist leukaemia/lymphoma registry in the UK. Part 2: Clustering of Hodgkin's disease.

Part 1 describes the epidemiology of Hodgkins disease occurring in those parts of the United Kingdom which are included in the Leukaemia Research Fund data collection survey. A total of 1,023 cases diagnosed between 1984 and 1986 were available for analysis. At county and district levels there was little heterogeneity in the distribution of cases. However, at the electoral ward level there were real differences for the younger age group (0-34). In this paper methods of investigation which are not dependent on census boundaries are applied and the presence of localised spatial clustering is confirmed. There is some evidence that the pattern of clustering relates to the nodular sclerosing subtype. These results are related to hypotheses of an infectious aetiology.

A specialist leukaemia/lymphoma registry in the UK. Part 1: Incidence and geographical distribution of Hodgkin's disease. Leukaemia Research Fund Data Collection Study Group.

This paper describes the epidemiology of Hodgkins disease occurring in parts of the United Kingdom between 1984 and 1986. The cases were carefully diagnosed and the data rigorously cross-checked as part of the larger Leukaemia Research Fund Data Collection Survey of all lymphoid and haematogenous malignancies. The age-specific rates show the lack of an older adult second peak. Spatial variation is examined in some detail. At county and district levels there is little heterogeneity in the distribution of cases. However, at the electoral ward level there were real differences for the younger age group (0-34).

Epidemiology of Non-Hodgkin's Lymphoma in Parts of England and Wales (1984-1988).

The descriptive epidemiology of non-Hodgkins lymphoma (NHL) (1984-1988) in parts of England and Wales is described by low grade (LG) and high grade (HG) subtypes. Major differences in age specific incidence are evident particularly in those under 35 years where LG-NHL is virtually absent in childhood and rare in young adulthood. Over the 5 year period of registration of cases significant increases in incidence were noted for males with both HG and LG disease. The geographical analyses demonstrate clear epidemiological differences between LG-NHL and HG-NHL. The SMRs for 22 English and Welsh counties are not correlated for LG and HG-NHL disease, whose distributions are significantly different. Regression analyses at the electoral ward level show a strong association between LG-NHL and high-socio economic status and there is a weaker link between HG-NHL and residence in an urban area. Formal comparison of LG and HG-NHL shows significant differences for the effect of socio-economic grouping but not for urban-rural status. There is strong evidence to show that major differences exist in patterns of LG and HG-NHL for age specific incidence, geographical distribution and area risk factors. Interpretation of these epidemiological features are highly suggestive of differing aetiologies.