The frequency of breast cancer screening: results from the UKCCCR Randomised Trial

https://doi.org/10.1016/S0959-8049(01)00397-5Get rights and content

Abstract

The optimal frequency of breast cancer screening has been a subject of debate since the inception of the UK National Breast Screening Programme (NHSBSP). This paper reports on the only randomised trial directly comparing different screening intervals. 99 389 women aged 50–62 years who had been invited to a prevalent screen were randomly allocated after the scheduled prevalent screen date to the study arm (invited to three further annual screens), or to the control arm (invited to the standard single screen 3 years later). 37 530 women in the study arm and 38 492 in the control arm had attended the prevalent screen. The endpoint was predicted breast cancer deaths. The prediction was based on both the Nottingham Prognostic Index (NPI) and a similar method derived from survival data from a series of tumours in the Swedish Two-County screening trial (2CS). Both indices were based on the size, lymph node status and histological grade of the invasive tumours diagnosed in the two arms of the trial. The pathology of the cases diagnosed was subject to review by two pathologists using standard criteria. The tumours diagnosed in the study arm were significantly smaller than those diagnosed in the control arm (P=0.05). The relative risk of death from breast cancer for the annual compared with the 3-yearly group was 0.95 (95% Confidence Interval (CI): 0.83–1.07, P=0.4) using the NPI and 0.89 (95% CI: 0.77–1.03, P=0.09) using the 2CS. Shortening of the screening interval in this age group is predicted to have a relatively small effect on breast cancer mortality. Improvements to the screening programme would be targeted more productively on areas other than the screening interval, such as improving the screening quality.

Introduction

In 1988, population screening for breast cancer was introduced nationally in the UK for women aged 50–64 years in the National Health Service Breast Screening Programme (NHSBSP). The interval between screens was chosen to be 3 years. There are no other trials comparing screening intervals.

The United Kingdom Co-ordinating Committee on Cancer Research (UKCCCR) trial of Breast Screening Frequency was established to investigate whether more frequent screening is advantageous. The trial randomised women to invitation to annual mammographic screening or to the routine NHSBSP first incident screen at 3 years following invitation to a prevalent screen.

A trial comparing two different mammographic screening regimes has to overcome two essential obstacles. First, because both arms are receiving some form of screening, breast cancer deaths will be rare overall. Secondly, such deaths occur over the long term. Thus, a trial with breast cancer death as the end-point would need very large numbers of subjects and very long follow-up. Thus, although the ideal endpoint of such a trial would be mortality from breast cancer, this would involve waiting for at least a further 10 years after the closure of the trial and randomising approximately 200 000 women in order to have adequate power [1].

This trial was therefore, designed to use prognostic features of the tumours diagnosed to predict subsequent mortality from breast cancer [1]. The predicted mortality would be a valid outcome provided that the treatment under trial influenced the true mortality largely via its effect on the prognostic features (which is the case in a trial of diagnostic method) [2], and the prediction of the true mortality was accurate [3]. We therefore designed the trial using as an end-point the predicted deaths from the tumour size, node status and histological grade of the tumours diagnosed. These were combined into the Nottingham Prognostic Index (NPI) [4], which has been shown to predict breast cancer survival in various tumour series, including one of approximately 10 000 cancers 5, 6. The tumour size, node status and grade have also been shown to predict the mortality reduction from screening in various randomised trials of screening [7]. The result based on predicted deaths is available earlier than that based on observed, and has greater statistical power, due to being based on a larger number of events [1].

This paper reports the main endpoint, the predicted survival in the two groups. A later publication will carry full details of the analysis by centre, of attendance and non-attendance at the prevalent screen, of subsequent attendance rates following the prevalence screen, recall rates and numbers of diagnostic operations. There will be further reports from the radiological subgroup on the analysis of ‘missed’ cancers, particularly at the prevalence screen, on the number of views at each visit and other matters, and from the pathology review group comparing features of the tumours detected in the two groups. A study of tumour biology and growth rates and an economic evaluation will also follow.

Section snippets

Patients and methods

Five screening units (Cardiff, Epping, Gateshead, Newcastle and Nottingham) participated in the trial, between 1989 and 1996. The primary end-point of the trial was the expected deaths from all invasive tumours diagnosed from immediately after the prevalence screen up to and including the final trial screen 3 years later. The primary aim of the trial was to establish, for women who had attended a prevalent screen and for whom breast cancer had not been diagnosed, the relative benefits of

Results

In the study group, among those women who had attended the prevalence screen, overall attendance rates at the first, second and third annual screens were 78, 78 and 81%, respectively. In the Control group, attendance at the 3-year screen was 85% among those who had attended the prevalence screen.

A total of 643 cancers was diagnosed during the period of the trial. Of these, 535 (443 invasive) were diagnosed in women who attended the prevalent screen, a rate of 2.35 per thousand per annum. In the

Discussion

In this trial, annual screening conferred an increase over 3-yearly screening of 13% in invasive cancers detected and of 60% in screen-detected cancers, with an increased total cancer detection rate and a reduction in the size of the tumours, both of which were of borderline statistical significance. Despite these, there was no statistically significant further reduction of predicted mortality from breast cancer for annual as opposed to triennial screening. The small amelioration in prognostic

Acknowledgements

The Medical Research Council, the Cancer Research Campaign, the Imperial Cancer Research Fund and the Department of Health funded this project. We thank the staff of the five screening centres, the surgeons, radiologists, pathologists, radiographers, technical and clerical staff involved in the diagnosis, histopathology and treatment of the cancer cases, and the former project managers at the Medical Research Council (MRC) Biostatistics Unit. We also thank the women who participated in the

References (13)

  • R.W Blamey

    The design and clinical use of the Nottingham Prognostic Index in breast cancer

    The Breast

    (1996)
  • N.E Day et al.

    Trial design based on surrogate end points-application to comparison of different breast screening frequencies

    J. R. Stat. Soc. A

    (1996)
  • R.L Prentice

    Surrogate endpoints in clinical trialsdefinition and operational criteria

    Stat. Med.

    (1989)
  • C.B Begg et al.

    On the use of surrogate end points in randomized trials

    J. R. Stat. Soc. A

    (2000)
  • J.H Todd et al.

    Confirmation of a prognostic index in primary breast cancer

    Br. J. Cancer

    (1987)
  • I Balslev et al.

    The Nottingham Prognostic Index applied to 9,149 patients from the studies of the Danish Breast Cancer Cooperative Group (DBCG)

    Breast Cancer Res. Treat.

    (1994)
There are more references available in the full text version of this article.

Cited by (87)

  • Does the Combination of Phone, Email and Text-Based Reminders Improve No-show Rates for Patients in Breast Imaging?

    2023, Current Problems in Diagnostic Radiology
    Citation Excerpt :

    Screening mammography is one of the most cost-effective ways to reduce breast cancer-related deaths. However, the benefit of screening is heavily dependent on patient compliance to routine exams.5 Up to two thirds of screening-detected cancers are localized (DCIS or T1N0 tumors), whereas clinically detected cancers are more likely to be locally aggressive and are usually node positive.20

  • Mammography screening among the elderly: A research challenge

    2015, American Journal of Medicine
    Citation Excerpt :

    In contrast, the present results pertain to 48 months of screening coverage for both annual and biennial screening. The latter study29 compared annual with triennial screening. In a comment to that study, Andersson30 expressed concerns about the possibility of beta error and suggested that greater clarity might have been achieved had more baseline data been available.

  • Mammography screening in Nigeria - A critical comparison to other countries

    2015, Radiography
    Citation Excerpt :

    The reason for the three years screening interval might be to minimise the cumulative radiation dose women are exposed to during screening, and therefore the possibility of developing radiation induced cancer.20 The Breast Screening Frequency Trial Group21 gave evidence that the effectiveness of a screening program is not about the frequency of screening but the effectiveness of the process to detect breast cancer when it is present. The screening program in Nigeria has recently been encouraging asymptomatic women to screen biennially.

  • Breast cancer screening in Austria: Key figures, age limits, screening intervals and evidence

    2015, Zeitschrift fur Evidenz, Fortbildung und Qualitat im Gesundheitswesen
View all citing articles on Scopus
1

The group members are listed at end of paper.

View full text