Diagnostic accuracy of needle-localized open breast biopsy for impalpable breast disease

BACKGROUND

Needle-localized open breast biopsy (NLBB) is considered the gold standard procedure for the diagnosis of impalpable breast disease. In an observational follow-up study the sensitivity and negative predictive value of this procedure was determined in a clinical population with long-term follow-up.

METHODS

Some 199 consecutive patients with a benign histological diagnosis on NLBB were followed for the occurrence of breast cancer, using information from the Dutch National Morbid-Anatomical Record Department. Based on a review of mammograms and histological slides, an expert panel decided whether the carcinomas detected during follow-up were newly developed, or were present already at the time of the NLBB.

RESULTS

After a median follow-up of 60.5 months, seven carcinomas were detected. At panel review, six appeared to have been missed by NLBB. The sensitivity of NLBB was 99 per cent after 2 years of follow-up, but dropped to 96 per cent after 5 years. Similarly, the negative predictive value dropped from 99 per cent after 2 years to 94 per cent after 5 years of follow-up.

CONCLUSION

NLBB is an accurate diagnostic procedure for the evaluation of impalpable breast disease. However, with longer follow-up the accuracy becomes lower than generally reported.

Diagnostic accuracy of large-core needle biopsy for nonpalpable breast disease: a meta-analysis

For the evaluation of non-palpable lesions of the breast, image-guided large-core needle biopsies are increasingly replacing needle-localized open breast biopsies. In this study, the diagnostic accuracy of this minimally invasive technique was evaluated by reviewing the available literature. Five cohort studies were included in a meta-analysis. Sensitivity rate, histological agreement between needle biopsy and subsequent surgery or long-term mammographic follow-up and clinical consequences for different disease prevalences were assessed. The sensitivity rate of large-core needle biopsy for the diagnosis of breast cancer was high (97%). The reclassified agreement rate between core biopsy and subsequent surgical biopsy or long-term mammographic follow-up was also high (94%). In case of 20% breast cancer prevalence among women referred after screening (as in the US), the risk of breast cancer despite benign large-core needle biopsy result is less than 1%. In European countries, however, prevalence of breast cancer among referred women is 60-70%. This would result in a risk of breast cancer despite benign large-core needle biopsy result of 4-6%. The results of this meta-analysis indicate that the image guided large-core needle biopsy is a promising alternative for the needle localized breast biopsy. However, additional research is needed to explore the limiting factors of the technique. Without such detailed knowledge, a benign histological diagnosis on large-core needle biopsy in countries with high prevalence of malignancy among referred women should be interpreted with caution.

Hypertension and breast cancer risk in a 19-year follow-up study (the DOM cohort). Diagnostic investigation into mammarian cancer

BACKGROUND

To investigate whether hypertension and the use of anti-hypertensive drugs are associated with breast cancer risk.

METHODS

This was a prospective study of 11, 011 women living in Utrecht, the Netherlands, aged 50-65 years at enrolment in a breast cancer screening project (DOM cohort). Women attended screening rounds between 1974 and 1985 at which blood pressure was measured and information on drug use and breast cancer risk factors was ascertained. Since 1974 (median follow-up time 19 years), information on breast cancer occurrence and death has been registered. Hypertension was defined as a systolic blood pressure > 160 mmHg or a diastolic blood pressure > 95 mmHg or current use of drugs for the indication hypertension. Cox's regression analysis was used to investigate the association between hypertension (treated or untreated) and subsequent breast cancer risk. Analyses were adjusted for age, body mass index, height, parity, familial breast cancer, smoking and oral contraceptive use.

RESULTS

A total of 523 women were diagnosed with breast cancer. Hypertensive women experienced a statistically significant increased breast cancer risk of 23% (age-adjusted hazard ratio (HRa) = 1.23; 95% confidence interval (CI) 1.01 -1.49). After adjustment for all confounders, the increase was 14% (HR = 1.14; 95% CI 0.93-1.40). The decline in risk was mainly attributable to the effect of BMI. The risk was similar in treated (HR = 1.22; 95% CI 0.91-1.63) and untreated hypertensive women (HR = 1.13; 95% CI 0.91-1.40).

CONCLUSION

These results do not support an association between hypertension and breast cancer, and if there is a link, it is likely to be positive and relatively small in size (+14%). This relation, if present, is not attributable to anti-hypertensive drugs, since the relation is also present in non-drug users.

Early age at menopause and breast cancer: are leaner women more protected? A prospective analysis of the Dutch DOM cohort

To investigate the relationship between age at menopause, body mass index, and breast cancer risk, we used data from a prospective cohort study (DOM cohort) in the Netherlands. Participants in this breast cancer-screening project included 10,591 women living in Utrecht, aged 49-66 years at enrolment. During a median follow-up period of 19 years, women attended screening rounds at which anthropometric measurements were taken and questions were asked about menopausal status, age at menopause, medication use and other risk factors for breast cancer. Cox regression analysis was used to investigate the relationship between age at menopause and subsequent breast cancer risk. Breast cancer incidence decreased with an earlier age at menopause. Women with a menopausal age of 44 years or younger had a 34% lower risk of breast cancer, than women with a menopausal age over 54 years (hazard ratio is 0.66 (95% confidence interval 0.43-0.91)). The annual hazard of breast cancer incidence decreased by 2.6% per year reduction in age at menopause. The protective effect of an early age at menopause was stronger for women with a low body mass index (< or = 27 kg/m2; reduction of 44%) than for women with a high body mass index (>27 kg/m2; reduction of 24%), although this difference was not statistically significant (P for interaction = 0.58). This difference was most pronounced in women who had ever smoked. Adjustment for known breast cancer risk factors did not alter the crude risk estimates significantly. In conclusion, this study provides evidence of the protective effect of lower age at menopause on subsequent breast cancer risk. This protective effect may be even stronger in leaner women.

Seventeen-year evaluation of breast cancer screening: the DOM project, The Netherlands. Diagnostisch Onderzoek (investigation) Mammacarcinoom

The DOM project is a non-randomized population-based breast cancer screening programme in Utrecht which started in 1974-75. The 17-year effect has been evaluated by a case-control study of breast cancer deaths during the period 1975-92 in women living in the city of Utrecht, born between 1911 and 1925, whose breast cancers were diagnosed after the initiation of the DOM project. Controls (three for each case) were defined as women having the same year of birth as the case, living in the city of Utrecht at the time the case died, and having had the opportunity of screening in the DOM project. Screening in the period 1975-92 indicated a breast cancer mortality reduction of 46% (odds ratio of 0.54, 95% confidence interval 0.37-0.79). The strongest protective effect was found at a screening interval of 2 years or less (mortality reduction of 62%, odds ratio of 0.38), and for the highest number of screens (mortality reduction of 68%, odds ratio of 0.32 for more than four screens). Exclusion of breast cancer deaths that occurred within 1 year of diagnosis, to allow for 'lead-time' bias, gave an odds ratio of 0.61. Early diagnosis of breast cancer by screening reduces breast cancer mortality in the long term. Bias due to the study design may slightly overestimate the protective effect. A screening programme with a 2-yearly, or smaller, interval between successive screens will improve the protection of screening.

Hypertension, antihypertensive drugs, and mortality from cancer among women

OBJECTIVES

To investigate whether hypertension and the use of antihypertensive drugs are associated with mortality from cancer among women.

DESIGN

A prospective study of 11075 women living in Utrecht, the Netherlands, aged 50-65 years at enrolment in a breast cancer screening project (DOM cohort). Women attended screening rounds between 1974 and 1985, during which blood pressure measurements were taken and information on drug use and smoking was ascertained. Since 1974 (median follow-up time 19 years) information on cause of death has been obtained from the patient's general practitioner. Hypertension was defined as a systolic blood pressure > 160 mmHg, a diastolic blood pressure > 95 mmHg, or current use of antihypertensive drugs. Cox regression analysis was used to investigate the association between hypertension (treated and untreated) and total and site-specific mortalities from cancer. The influences of systolic and diastolic blood pressures per 10 mmHg increase for women not using antihypertensive drugs were evaluated. Analyses were adjusted for age, smoking, and body mass index.

RESULTS

In total, 704 women died of cancer and 1633 women left the study area. Hypertensive women had a greater than normal (not statistically significant) total risk of mortality from cancer [hazard ratio (HR) 1.10, 95% confidence interval (CI) 0.93-1.31]. Risks for treated and untreated hypertensive women were similar. Cancer-site-specific analyses revealed a statistically significantly lower mortality from gastrointestinal cancers among untreated hypertensive women (HR 0.64, 95% CI 0.44-0.93), whereas mortality from lung cancer was more prevalent among these women (HR 2.50, 95% CI 1.37-4.59). Mortality from lymphatic and hematopoietic cancers for drug-treated hypertensive women was greater than normal (HR 2.11, 95% CI 1.04-4.28), as was mortality from cancers of the uterus, cervix, and ovary (HR 1.80, 95% CI 1.00-3.26).

CONCLUSION

These results support the hypothesis that, if there is a link between blood pressure and cancer, it is likely to be positive and relatively small (+10%); and applies also to nondrug-treated women. The relation may apply for some types of cancer, but not for others.

Breast-cancer incidence in relation to height, weight and body-fat distribution in the Dutch "DOM" cohort

In a cohort of 11,663 participants in a breast-cancer screening program, height, weight, waist circumference and hip circumference were measured, and information about menstrual and reproductive history was obtained by questionnaire. After exclusion of 83 women with unclear menopausal status, the subjects were divided into 3 sub-cohorts: 5,891 women who were pre-menopausal at the time of data collection, 3,521 women who had entered the study after natural menopause, and 2068 women who had been hysterectomized and/or ovariectomized. After a median follow-up of 10.6 years, 147, 76 and 52 incident cases of breast cancer were detected in the 3 respective sub-cohorts. No statistically significant association was found in any of the sub-cohorts between breast-cancer risk and height, weight, body-mass index (BMI) or hip circumference. In the sub-cohort of women with natural menopause, however, risk of breast cancer was positively and significantly associated with the ratio of waist-to-hip circumferences (WHR) (RR = 2.63 for upper vs. lower quartile), and this association did not change after adjustment for variations in disease risk related to body height and weight. Although similar to observations in other cohort studies showing positive associations between obesity and breast-cancer risk in post-menopausal women, our results are different, in that WHR and not BMI appears to be the more specific indicator of breast-cancer risk.

Underreporting of energy, protein and potassium intake in relation to body mass index

BACKGROUND

Differential underreporting of dietary intake by subgroups of body mass index (BMI) will confound associations between dietary intake and BMI-related diseases. We estimated the magnitude of BMI-related underreporting for energy, protein, and potassium intake for the Dutch cohorts of the European Prospective Investigation into Cancer and Nutrition (EPIC).

METHODS

The study population consisted of 134 Dutch men and women, aged 21-71 years, who participated in a pilot of EPIC. Ratios of reported dietary intakes to biomarkers were used as measures for underreporting. Dietary intake was assessed by a food frequency questionnaire (FFQ) and repeated 24-hour dietary recalls. Biomarker for energy intake was calculated basal metabolic rate; for protein and potassium intake the biomarker was 24-hour urinary nitrogen and potassium excretion, respectively. The measures of underreporting were linearly regressed on BMI (in kg/m2).

RESULTS

Significant negative regression coefficients were observed when regressing energy ratio on BMI with adjustment for physical activity (FFQ: beta = -0.04 for men, beta = -0.02 for women; 24-hour recalls: beta = -0.03 for men, beta = -0.04 for women). In men, a significant negative regression coefficient (beta = -0.03) was observed when regressing protein ratio on BMI; for the recalls however only after adjustment for age and education (beta = -0.02). In women, negative regression coefficients were also obtained, but for the FFQ only after exclusion of dieting women (both FFQ and 24-hour recalls: beta = -0.02). According to the recalls, but not the FFQ, a significant negative regression coefficient (beta = -0.02) was observed among women when regressing potassium ratio on BMI.

CONCLUSIONS

In this Dutch population, BMI-dependent underreporting of 20-25% over the observed range of BMI is present for protein and energy, Further study on BMI-dependent underreporting of dietary intake in EPIC cohorts is warranted.

Methodological aspects of physical activity assessment in epidemiological studies

A range of epidemiological studies conducted over the past decades has produced strong support for the view that lack of physical activity is associated with increased risks of several chronic diseases, such as cardiovascular disease, diabetes mellitus, cancer, etc. Much is still unknown about the type and degree of activity that is required to influence the risk of specific diseases. Furthermore, physical activity can act as a confounder in relationships between other exposure variables (e.g. diet) and disease. Thus, the measurement of physical activity in epidemiological studies is of great importance. The questionnaire is the most frequently used method in epidemiological research. Before using a questionnaire on a large scale, validity and reproducibility should be assessed in a representative population. Some practical and methodological aspects of physical activity validation studies are described, together with the possible implications of the results.

Physical activity and cardiovascular disease risk profile in women

In a population of 4,576 Dutch women aged 49-70 years who participated in the European Prospective Investigation into Cancer and Nutrition (EPIC) in 1993-1995, the relation between physical activity and the presence of cardiovascular disease risk indicators was assessed cross-sectionally. Physical activity was determined from a self-administered questionnaire, while blood pressure, heart rate, body mass index, waist/hip ratio, and waist circumference were measured at the study center. Mean risk indicator levels were calculated for different activity categories. Blood pressure was most clearly associated with time spent in sports (mean systolic blood pressure, adjusted for age, level of education, and smoking, 128.9 mmHg in the highest sports tertile, and 132.1 mmHg in the lowest sports tertile; mean diastolic blood pressure, 77.8 mmHg and 79.0 mmHg, respectively). Body mass index, waist/hip ratio, and waist circumference showed an inverse relation with cycling, gardening, do-it-yourself-activities, and sports. In this population, leisure-time activity was inversely related to cardiovascular disease risk indicators, but work activity and housework were not. The authors conclude that if investigators wish to measure physical activity in women over age 50 years with the aim to identify high- and low-risk groups for cardiovascular disease, they should consider not only housework activity, but also leisure-time activities such as cycling, sports, and do-it-yourself activities.

Prediction of peripheral fracture risk by quantitative microdensitometry

BACKGROUND

The purpose of the study was to investigate whether quantitative microdensitometry (QMD) could be used for screening purposes to identify a population with a high risk for peripheral osteoporotic fractures.

METHODS

In 1984 bone mineral density (BMD) measurements were made on the index finger of 612 women. Repeated BMD measurements were obtained in 1989 in 440 women. Cox proportional hazards models were used to obtain adjusted estimates of the hazard ratio of a fracture according to BMD or bone loss. Receiver operating characteristic curves were constructed and the areas under the curves (AUC) were compared.

RESULTS

Thirty-five women experienced peripheral osteoporotic fractures between the first and the second measurement. Women in the three lowest quartiles of bone density were 1.4-1.8 times (diaphyseal site) and 2.4-2.8 times (metaphyseal site) as likely to experience a fracture as those in the highest quartile. Women in the quartile with the highest bone loss had a risk of 6.9 (diaphyseal site) and 7.5 (metaphyseal site) times higher than women in the lowest quartile. The discriminative power of the measurement as a single test was moderate (AUC 63%); two successive measurements, made with an interval of 5 years to measure bone loss, increased the discriminating power (AUC 74%; P < 0.05).

CONCLUSIONS

Phalangeal BMD and bone loss, as measured by QMD, are important risk factors for peripheral osteoporotic fractures, but cannot be used as an efficient screening instrument for selecting women with the highest fracture risk.

Estimation of reproducibility and relative validity of the questions included in the EPIC Physical Activity Questionnaire

BACKGROUND

The EPIC core questionnaire on lifestyle contains a number of questions on physical activity designed to rank subjects according to level of physical activity (short PA questionnaire). These questions are based on a more extensive questionnaire designed to measure absolute total energy expenditure (extensive PA questionnaire), that was validated in a pilot study preceding EPIC. Reproducibility and relative validity of the short PA questionnaire were estimated by selecting, from the pilot study data, the answers to a number of questions from the extensive questionnaire that resembled those actually included in the short version.

METHODS

The population of the pilot study consisted of 126 men and women aged between 20 and 70 years. Reproducibility was estimated by administering the extensive questionnaire three times: at baseline, and after 5 and 11 months. In order to determine the relative validity of the extensive questionnaire, a 3-day activity diary, repeated four times, was used as the reference method.

RESULTS

Over the study period (13 months), mean absolute energy expenditure, estimated from the questions included in the short questionnaire, was fairly constant in men but not in women. REPRODUCIBILITY: Spearman correlation coefficients ranged from 0.47 to 0.89 in men, and from 0.49 to 0.81 in women. RELATIVE VALIDITY: Spearman correlation coefficients between the short questionnaire and the diary were between 0.32 and 0.81 for men, and between 0.28 and 0.72 for women.

CONCLUSIONS

The questions selected for the short questionnaire are not suitable for estimating energy expenditure at an absolute level. Reproducibility and relative validity of the ranking of subjects seemed satisfactory and comparable to the extensive questionnaire. The results imply that the short questionnaire is suitable for ranking subjects in the EPIC study.

Relative validity and repeatability of a new questionnaire on physical activity

BACKGROUND

A physical activity questionnaire was developed with the aim to estimate usual individual daily energy expenditure. The questionnaire focused on the number of hours usually spent on various activities. In a pilot study it was tested for repeatability and validity in a population of 126 Dutch adults (64 men, 62 women).

METHODS

For assessment of repeatability the questionnaire was administered three times during the course of 1 year. A four-times-repeated 3-day activity diary was used as a reference instrument to evaluate validity.

RESULTS

Differences in mean energy expenditure among repeated administrations of the questionnaire were small and not significant. Spearman's test-retest correlation coefficients for total energy expenditure for men were 0.76 [95% confidence interval (CI) 0.63-0.85] and 0.70 (95% CI 0.54-0.81) at 5 and 11 months, respectively, and for women were 0.58 (95% CI 0.36-0.74) and 0.71 (95% CI 0.54-0.82). There was a significant trend showing increasing mean diary energy expenditure for successive tertiles of questionnaire energy expenditure. The correlation between the questionnaire and the diary was 0.66 (95% CI 0.49-0.78) for men and 0.43 (95% CI 0.18-0.63) for women.

CONCLUSIONS

It was concluded that this questionnaire is a useful tool for estimating energy expenditure in epidemiological studies.

Histopathology and growth rate of interval breast carcinoma. Characterization of different subgroups

BACKGROUND

Interval breast cancers are defined as carcinomas occurring within 2 years after a negative screening Distinction has to made between cancers existent at the time of screening but missed for some reason, and fast-growing incident cancers. This is important because the natural history and the implications for the treatment of the patient might differ.

METHODS

Radiologic and histopathologic characteristics were assembled for 104 interval cancers diagnosed within the DOM project, the Utrecht program for the early detection of breast cancer. At a mammographic review for 27 cases, signs of malignant or benign tumor were found (missed cases). For 77 cases no radiologic signs were present on review. Twenty of these cases had a mitotic rate of > 3 and a high tumor growth rate (mean doubling time: 51 days). This combination seemed implausible, therefore, it was thought hypothesized that these tumors were most likely present, although radiologically invisible (masked), at the time of screening. The remaining cases (n = 57) were classified as true interval cancers and further divided into 14 fast-growing cases (mitotic rate > or = 3/high-power field [HPF]) and 43 cases with an intermediate growth rate (mitotic rate < 3/HPF).

RESULTS

Factors associated with the masking of tumors were the histologic tumor type, absence of microscopic calcifications, and presence of dense breast tissue. Fast-growing tumors were characterized by a young patient age, absence of microscopic calcifications, and a high percentage of regional lymph node positive tumors. The 5-year survival probability was 100% for missed cases, 70% for masked cases, 80% for cases with an intermediate growth rate, and 54% for fast-growing cases.

CONCLUSIONS

It is possible to separate interval breast cancers in true interval cases and cases (most likely) existent at the time of screening. Part of this last group is invisible by mammography (masked).

Repeatability and relative validity of two physical activity questionnaires in elderly women

In a population of 33 women aged 51-71 yr, two physical activity questionnaires (a modified Baecke questionnaire and the pre-EPIC questionnaire) were tested for repeatability (5 and 11 months). For assessment of relative validity, the questionnaires were compared with a physical activity diary (12 d) as the main reference method, a single 24-h Caltrac accelerometer score, and energy intake estimated from a 24-h dietary recall repeated 12 times. Repeatability of the questionnaires at 5 months was 0.82 (Baecke) and 0.42 (pre-EPIC), at 11 months 0.73 and 0.60, respectively. Correlation with the activity diary was 0.51 for the Baecke and 0.64 for the pre-EPIC questionnaire. Correlation with the Caltrac was 0.22 for both questionnaires, and with energy intake -0.21 and 0.43, respectively. Factor analysis suggested that the questionnaires and the diary measured a common aspect of activity, which could be interpreted as "reported normal physical activity." Women who reported sweating or breathlessness during the past week scored significantly higher only on the Baecke questionnaire. It was concluded that the two questionnaires seemed to be able to rank older women according to physical activity in epidemiological studies.

Age specific sensitivity and sojourn time in a breast cancer screening programme (DOM) in The Netherlands: a comparison of different methods

STUDY OBJECTIVE

To estimate age dependent sensitivity and sojourn time in a breast cancer screening programme by different methods.

POPULATION AND METHODS

The study population comprised women participating in the DOM project--the Utrecht screening programme for the early detection of breast cancer. Breast cancer screening prevalence data and incidence rates after a negative screen were used to estimate age specific sensitivity and mean sojourn time by different methods.

MAIN RESULTS

Maximum likelihood estimates of the mean sojourn time varied from one year for women aged 40-49 years to three years for women over the age of 54. Sensitivity was calculated by two different methods. Both pointed to a high sensitivity (around 100%) in the age groups 40-49 and over 55 years. For women aged 50-54, the sensitivity varied from 63% to 100%, depending on the method used and the value of the baseline incidence rate.

CONCLUSIONS

Different methods of estimating sensitivity pointed at an acceptable level in women over and under 50 years of age. Sojourn time, and thus the tumour growth rate, seemed to be age dependent. This could mean that the until now disappointing screening results in women under 50 years of age are not so much a result of low sensitivity as of a relatively high tumour growth rate in younger women.

Survival in interval breast cancer in the DOM screening programme

The study describes breast cancer survival of 75 interval cancer cases (cancer occurring within 2 years of a negative screen) detected in women who participated in the DOM screening programme. After mammographic revision, this group was divided into 17 so-called 'missed' cases and 58 'true' interval cases. Ten year survival of these 58 'true' interval cases was 58%, which was not significantly different from that of 219 cancers detected in a non-screened, control group of women, diagnosed with breast cancer before the start of screening (63%; log rank chi 2 test, P = 0.98). Results remained essentially the same after correction for age at diagnosis, tumour size, axillary status and year of diagnosis. Ten year survival of 'true' interval cancers (58%) was slightly worse than that of 'missed' cases (67%; log rank chi 2 test: P = 0.38). This difference could largely be explained by differences in tumour size and axillary status. We conclude that there was no important difference in survival between 'true' interval cancers and non-screened historical controls. This could mean that either this subgroup of interval cancers does not constitute an excess of rapidly growing tumours, or if it does, that a fast growth rate is not associated with an exceptionally poor prognosis.

[Mortality rate in lung cancer decreasing in men and increasing in women]

OBJECTIVE

To determine the trend in lung cancer mortality rates among men and women in the Netherlands during the period 1951 to 1992.

DESIGN

Descriptive research.

METHOD

In a previous paper the lung cancer mortality rates of men and women during the period 1951 to 1982 were analysed using the Netherlands Central Bureau of Statistics (CBS) population data and the CBS mortality data. The mortality rates were directly standardized for age using as a standard the Dutch sum population of 1951 to 1982. The current study describes the lung cancer mortality for men and women until 1992 using exactly the same methods.

RESULTS

The direct standardized mortality rates for men had decreased since 1987. For women, on the contrary, these rates had increased further. The age-specific mortality rates for men have decreased since the calendar period 1983-1987 and since the birth cohort 1916-1920. The age-specific mortality rates for women have increased until the latest calendar period 1988-1992 and until the latest birth cohort 1931-1935.

CONCLUSIONS

Lung cancer mortality in men has peaked, in women it is still rising.

Validity and repeatability of a modified Baecke questionnaire on physical activity

BACKGROUND

In a pilot study for the European Prospective Investigation into Cancer and Nutrition (EPIC) a modification of the Baecke questionnaire on physical activity was tested for repeatability and relative validity in a population of 134 men and women aged 20-70 years.

METHODS

For the assessment of repeatability Pearson's correlation coefficients and percentages of agreement after classification in tertiles were computed between administrations of the questionnaire at baseline, and after 5 and 11 months. Relative validity was determined by comparing the questionnaire to a four times repeated 3-day activity diary.

RESULTS

Repeatability after 5 and 11 months was good, with test-retest correlation coefficients between 0.65 and 0.89 for main sections of the questionnaire. The percentages of agreement, exceeding chance (Cohen's kappa) were 57% and 56% for men (at 5 and 11 months respectively) and 41% and 46% for women. The correlations with the diaries were 0.56 in men and 0.44 in women. Agreement apart from chance between classification in tertiles for both methods was 35% for men and 10% for women.

CONCLUSIONS

These data show that repeatability is good and relative validity as compared to an activity diary is moderate but well within the range of values found in other studies. The questionnaire is more valid in men than in women.

Age at menarche and breast cancer risk in nulliparous women

Early age at menarche is a known risk factor for breast cancer, some inconsistency in the literature not withstanding. Relative risks for an early menarcheal onset as compared to a late onset vary from 1.0 to 1.9. To avoid (residual) confounding by parity-related factors a case-control study was conducted among nulliparous women. 135 cases and 540 controls were selected from two population-based screening projects for breast cancer in The Netherlands. Four controls were matched to each case for year of birth, for screening-centre, and for number of screening examinations. Women with a menarcheal age of 10 or 11 years showed a 2.2 times (95% confidence interval: 1.2-4.0) higher risk for breast cancer compared to women who had their first menstrual period at 12 years of age or older.

Effect on breast cancer screening response in The Netherlands of inviting women for an additional scientific investigation

STUDY OBJECTIVE

The study aimed to determine whether asking women to undertake an additional scientific study would deter them from attending screening for breast cancer.

DESIGN

A randomised study was conducted in all women aged 50-70 years who were eligible for breast cancer screening and living in the city of Utrecht. A total of 1863 women were invited for mammography only and 1863 women were invited to participate in the European Prospective Investigation into Cancer and Nutrition (EPIC) in addition to the mammography.

SUBJECTS

The study population comprised a random sample of 3726, 15% of the female population of Utrecht aged 50-70 years.

MAIN RESULTS

The attendance rate for breast cancer screening was 53%, irrespective of the invitation to participate in the additional scientific study.

CONCLUSIONS

Asking women to attend for an investigation in addition to the routine screening procedure for breast cancer did not affect the overall response to screening.

The epidemiological profile of women with an interval cancer in the DOM screening programme

Risk factors for breast cancer were compared in 107 women with interval breast cancer (cancers occurring within 2 years after a negative screen) and 258 women with breast cancer detected at 1st screening. All women (aged 40-67) were screened in the DOM project (the Utrecht programme for the early detection of breast cancer). Women with an interval cancer reported more often a history of benign breast disease (OR 4.66, 95% C.I. 2.08-10.41) and an artificial menopause (OR 4.07; 95% C.I. 1.74-9.55) than women with a screen detected cancer. Women with an interval cancer were taller than women with a screen detected cancer; this was seen most clearly in women with an artificial menopause (chi 2 for trend = 5.88; p = 0.02) and to a lesser extent in premenopausal women (chi 2 for trend = 1.70; p = 0.19). Premenopausal women with an interval cancer were heavier than women with a screen detected cancer (chi 2 for trend = 4.66; p = 0.03); whereas natural postmenopausal women with an interval cancer were leaner than women with a screen detected cancer (chi 2 for trend = 1.57; p = 0.21). All analyses were done while correcting for age and selected other risk factors for breast cancer. These results suggest that the epidemiological profile of pre- and post-menopausal women with an interval cancer differs from that of women with a screen detected cancer, which might imply a different natural history of these two types of breast tumours.

The contribution of routine follow-up mammography to an early detection of asynchronous contralateral breast cancer

The role of routine mammography was assessed in the early detection of asynchronous contralateral breast cancer (ACBC). The breast cancer patient populations of two cities, Nijmegen and Eindhoven, The Netherlands, which were subjected to a well-defined follow-up program, were compared. The program consisted of regular physical examination and annual mammography in Nijmegen and physical examination only in Eindhoven. From 1975 until 1987, 24 ACBC patients were detected within a group of 880 breast cancer patients in Nijmegen (3%) and, from 1971 until 1984, 14 ACBC patients within a group of 411 patients in Eindhoven (3%). In Nijmegen, eight of the 23 evaluable patients (35%) had a contralateral tumor with a histologic size smaller than 10 mm or an in situ carcinoma, compared with one of the 14 of the Eindhoven patients (7%), whereas 18 of the 24 (75%) versus eight of the 14 patients (57%) were node-negative. Thus annual mammography is very likely a contribution in the early detection of contralateral breast cancer as compared with follow-up by regular physical examination only.

Breast cancer screening in women over age 50. A critical appraisal

Breast cancer screening with mammography is strongly advocated by some and rejected by others. Discussions in the medical literature focus on the beneficial and unfavourable effects of breast cancer screening. Among the latter are the suggested excess mortality rate of screen-detected breast cancer patients, the high number of unnecessary biopsies, the possibility of overdiagnosis and subsequent unnecessary treatment, and the occurrence of interval breast cancer cases. It is concluded that screening can to a certain extent prevent women from dying of breast cancer but some negative effects are unavoidable. However, adequate training of the screening team can minimize the number of unnecessary biopsies and the occurrence of interval cancers. Research is needed to evaluate the natural course of so-called 'minimal' lesions, and to optimize screening tests. The search for new (imaging) techniques is needed to detect lobular cancers in dense breasts.

[A survey of the epidemiology of breast cancer in The Netherlands]

Breast cancer is responsible for the main part of the total cancer incidence (+/- 30%) and cancer mortality (+/- 20%) among Dutch women. Due to aging of the Dutch population alone, the absolute numbers of breast cancer cases and deaths will increase considerably. Risk factors for breast cancer give no clues for primary prevention. By means of periodic screening patients are diagnosed at an earlier stage of disease, with a better prognosis. Even without a screening programme symptomatic women are diagnosed at earlier stages of disease. This paper gives a review of trends in breast cancer incidence and mortality in The Netherlands, as well as a risk profile of breast cancer patients and the relation between disease stage and survival. Its purpose is to obtain a better insight into the intended effects of the national breast cancer screening programme.

[Observed effects of the mass screening program for breast cancer in Nijmegen in 1975-1986]

Since January 1975, the start of the screening programme for breast cancer in the city of Nijmegen, The Netherlands, over 14 years have elapsed. Data up to the end of 1986 have been processed and are presented in this article. The younger the birth cohort, the higher the attendance rate. In the course of the screening programme, attendance rates decline. Referral rates are highest in the first screening round and for the eldest birth cohort. Detection rates increased during the most recent screening rounds, after an initial decrease during the earliest rounds. They are higher for the eldest birth cohort also. The relative number of interval carcinomas is highest among younger women (as related to the screening-detected carcinomas).

Evaluation of overdiagnosis of breast cancer in screening with mammography: results of the Nijmegen programme

After 12 years of screening for breast cancer in Nijmegen (1975-86), during which period six mammographic examination rounds were carried out, the extent of overdiagnosis was evaluated. Overdiagnosis is defined as a histologically established diagnosis of invasive or intraductal breast cancer that would never have developed into a clinically manifest tumour during the patient's normal life expectancy if no screening examination had been carried out. The whole 12-year period shows an excess of 11% of breast cancer cases in Nijmegen, compared with the neighbouring city of Arnhem, where no mass screening was performed. The incidence of breast cancers in Nijmegen in the period 1975-78 is higher, compared with the incidence rates in Arnhem; the rate ratio is 1.30. For the time-intervals 1979-82 and 1983-86 the rate ratios are 1.03 and 1.01 respectively with (0.89; 1.18) and (0.86; 1.16) as 95% confidence intervals. This leads to the conclusion that there is no evidence that screening programmes using modern mammography constitute a significant risk for overdiagnosis of breast cancers.

The occurrence of interval cancers in the Nijmegen screening programme

Since January 1975 a population-based screening programme for the early detection of breast cancer has been carried out in the city of Nijmegen. During five interscreening periods of 2 years each a total of 158 so-called interval cancers were diagnosed. Careful revision of all screening and diagnostic mammograms was executed. Of all interval cancers 26% were 'missed' at the previous screening examination (due to technical or observer error), 16% were radiographically occult at the time of diagnosis and 58% were 'true' interval cancers. Interval cancers were regarded as 'true' when an obvious lesion was observed on the diagnostic mammogram while no suspect signs were seen on the previous screening mammogram. The prevalence of 'missed' cancers did not decline in the course of the screening programme. Radiographically occult tumours were localised, mostly in Wolfe's P2/DY breast parenchyma (83%), 33% were lobular invasive and 25% ductal non-invasive. 'True' interval cancer cases (58%) showed the same overall survival as control breast cancer patients, diagnosed in a non-screening situation. Shortening the screening interval would reduce interval cancer rates and probably further decrease breast cancer mortality in a screened population. However, from the present series of interval cancers 63% would not have been prevented by an annual screening examination. As regards women under age 50 annual screening would still leave 66% of all interval cancers in this age group undetected. Probably more benefit will be gained by searching for new imaging techniques to reduce numbers of 'missed' cancers and to detect lobular invasive and ductal non-invasive cancers in dense breast parenchyma.

Screening for breast cancer in Nijmegen. Report of 6 screening rounds, 1975-1986

A population-based screening programme for breast cancer was initiated in Nijmegen in 1975 with mammography as the only screening procedure. Up to January 1987, 6 screening rounds were carried out with a 2-year screening interval. Rates of attendance, referral, biopsy and detection were calculated and numbers of interval cancers are presented in order to give a clear view of what repeated screening can accomplish in a population. At the first screening round the attendance rate was 87% for women under age 50 and 83% for women aged 50-64. For women aged 65 or over the initial attendance rate was 40%. Rates of attendance declined in subsequent years. Detection rates were highest for elderly women at their first examination: 9.5 per 1,000 screened women. Corresponding rates were 5.6 and 2.3 per 1,000 for women aged 50-64 and below 50 respectively. The positive predictive value for referral was, on average, 20% for women under age 50 and 50% for elderly women, although a sharp increase was seen in the last 2 screening examinations for all age-groups. Predictive values for biopsy were higher: 30% on average for women aged under 50 and 60%-70% for elderly women, again with a sharp increase in the last 2 screening rounds. Interval cancer rates, calculated as the number of cancers occurring within 2 years among negatively screened women at risk, showed no particular trend and varied between 0.9-1.3 per 1,000 woman-years after each screening round. Compared to screen-detected cancers, interval cancers occurred more frequently in younger women. In women under age 50, the ratio between screen-detected and interval cancer was about 1:1, while it was about 2:1 for elderly women.

The predictive value of positive test results in screening for breast cancer by mammography in the Nijmegen programme

After 10 years of screening for breast cancer by mammography in Nijmegen, the predictive value of positive screening results (PV+) was evaluated. The percentage of women with breast cancer in the group of referred women (PV+) for women under age 50 was 16-26%, regardless of the number of screening examinations they had. The percentage of women with breast cancer in the group of women who were biopsied was 25-40%, regardless of the number of examinations. For women aged 50 and over the predictive value was 34-57% and 58-90% respectively. It was further evaluated whether characteristics such as age, Quetelet index, parity, and Wolfe-classification could be used to increase the PV+ in women who were identified as positive by mammography. A logistic regression model analysis showed that true-positive and false-positive cases differ significantly only in terms of age and breast complaints. Although the model had a good fit, it could not be used to distinguish false-positive from true-positive test results.

Mammographic breast pattern and the risk of breast cancer

Within the Nijmegen breast cancer screening programme a study was made of the relation between mammographic breast features according to Wolfe's classification and breast cancer risk. The time of follow-up was 6 years. The Nijmegen study challenged the strong relationship previously suggested. In comparison with women of N1 and P1 breast type, the relative risk for women with P2 and DY breast type was estimated at 0.7, with 95% confidence limits of 0.2 and 2.4, after adjustment for the major risk factors. The results could not be explained by intra-observer variability or left-right difference.