An investigation of breast cancer risk factors in Cyprus: a case control study

Breast cancer is the most frequent malignancy in the female population of Europe and North America where an estimated 1 in 9 women are at risk of developing the disease [1]. Numerous epidemiological studies over the last three decades have revealed a number of risk factors associated with breast cancer [2–4]. In addition to genetic and reproductive factors, breast cancer risk displays wide ethnic and geographic variation [5]. Within Europe the incidence varies by almost two-fold. It is highest in Northern European countries, with an estimated 84.6 cases per 100,000 population with intermediate rates recorded in Southern Europe and lowest rates occurring in Eastern Europe with 42.6 cases per 100,000 population [6]. In addition to the geographic variation, influences on incidence rates, have also been attributed to differences in the use of mammography, diet, physical activity, body size, alcohol consumption and socioeconomic and reproductive factors [7]. The discovery of highly penetrant breast cancer susceptibility genes such as the BRCA genes in the mid 1990's [8, 9] emphasized the importance of genetic factors, but it is currently believed that environmental factors are of greater significance [10].

Cyprus is the third largest island in the Mediterranean Sea with a Greek-Cypriot population of 749,200 as recorded in July 2004, by the statistical service of the Republic of Cyprus. Data from the National Cancer Registry established in 1998 by the Ministry of Health show an average incidence of 400 female breast cancer cases per year [11]. This corresponds to an age-standardized incidence rate of breast cancer of 60 per 100,000 population [11] which is comparable to the rates seen in other Southern European countries, a region where a moderate breast cancer risk is usually recorded [6]. Cyprus is a member of the European Union since 2004, and has a long history of practicing a Mediterranean lifestyle. However the lifestyle is rapidly changing towards a western type of life, so it is of interest to study the pattern as well as the influence of breast cancer risk factors in our population.

Indeed, there are no data as yet available about breast cancer risk factors in the Cypriot female population. The main aims of this report are (a) to describe the background levels of established and recognized breast cancer risk factors among Cypriot women and (b) to assess the strength of their association with breast cancer risk. In order to achieve this we carried out a case-control study, involving a breast cancer patient group and a control group of healthy Cypriot women attending mammography clinics as part of the National screening programme. A number of risk factors that modulate breast cancer such as socioeconomic status, reproductive events, family history and lifestyle factors were investigated.

Interestingly, previous published work by our group has shown that the Cypriot population exhibits some unique genetic features as revealed by the identification of novel mutations in the BRCA1 and BRCA2 genes [12, 13]. Moreover, this is the first study into breast cancer epidemiology among the Cypriot female population and will function as a baseline, of future efforts into combining person-based data, with the molecular/genetic data in order to study gene-environment interactions in our population.

A strong inverse association was observed between breast cancer risk and number of children. Compared to nulliparous women, crude odds ratios were 0.90, 0.84 and 0.68 in women with one, two and three children respectively; p-value for trend < 0.01. However, as expected, there was a strong association between number of children and likelihood of breastfeeding. Nearly half of women with one child reported not having breastfed at all while only 10% of them breastfed for longer than 6 months. In contrast, amongst women with 3 or more children, only 15% did not breastfeed while as many as 38% reported breastfeeding for at least one year. The Kendall's tau-b was 0.42, p-value for independence < 0.001, indicating a relatively strong and statistically significant correlation between breastfeeding duration and number of children. Considering every possible pair among the participants, this non-parametric measure is calculated on the basis of the number of occasions where the ranking of both breastfeeding duration and number of children are in the expected direction (i.e. a higher rank in one variable is accompanied by a higher rank in the other) as compared to the number of discordant pairs. The strong univariable association observed with number of children attenuated once breastfeeding was controlled for in the multivariable logistic analysis while not much change was observed in the case of breastfeeding. This cannot be simply a result of collinearity as the confidence intervals also remained largely unaffected, more likely indicating that the observed association with number of children was confounded by the effect of breastfeeding.

In fact, this is also supported from the results of the stratified analysis where the estimate for the odds ratio for a one unit increase in the number of children was 0.92 (95%CI 0.80-1.05) among women who did not breastfeed and 0.97 (95%CI 0.82-1.15) among women who breastfed. The Mantel-Haenszel estimate controlling for breastfeeding is 0.94 (95% CI 0.85, 1.04) and test of homogeneity of ORs = 0.61, indicating no evidence of effect modification in the relationship of breast cancer and number of children by breastfeeding status. Further adjusting for other variables revealed no association with parity while interestingly wide confidence intervals were observed. On the other hand, the effect estimate of breastfeeding (and the confidence intervals for that matter) practically remained unchanged at 0.72 (95%CI 0.58, 0.90).

None or only weak associations were observed between breast cancer risk and other risk factors investigated, for instance BMI, age at first pregnancy, smoking, exercise or use of oral contraceptives. Surprisingly, more controls (28.8%) than cases (15.5%) reported taking hormone replacement therapy (HRT). The inverse association persisted in the fully adjusted model, and in fact risk of breast cancer appeared to be inversely related to the duration of HRT use. Adjusting for number of children in the multivariable models, did not affect any of the estimates. Indicatively, the odds ratio for family history now becomes 1.62 (95%CI 1.22, 2.16), while the estimates were 0.83 (95% 0.62, 1.09) and 0.67 (95% 0.47, 0.94) in women who reached menarche between the ages of 12-14 and 15 or later respectively. Lastly, inferences were also unaffected when breastfeeding was expressed in terms of duration in the multivariable models (instead of the binary classification presented in the table).

The aetiology of breast cancer is still poorly understood and known breast cancer risk factors explain only a small proportion of cases. Epidemiological studies conducted in different populations have identified a spectrum of well established and probable risk factors for breast cancer [15]. These include age, socioeconomic status, reproductive events, breastfeeding, family history and lifestyle among others. However, most epidemiological breast cancer studies involve subjects living in North America and Western Europe, regions which represent only a fraction of the global population. Therefore there is a need to study breast cancer epidemiology in populations in less-well studied regions of the world, in order to gain a better understanding of breast cancer aetiology [15]. It is noted that this is the first epidemiological study into breast cancer risk factors in the Cypriot population.

The main aim of this study was to assess the strength of associations between some recognized risk factors and breast cancer among Cypriot women. We should note that while the factors investigated here (other than BMI) were all self-reported, there is no reason to believe that any recall should be substantially differential, not least due to the standardised interview procedure by which information was collected from both groups of participants. A total of 1109 Cypriot women diagnosed with breast cancer between years 1999-2005 were recruited. It should be noted that the incidence of breast cancer in Cyprus is now around 400 cases per year. The cases in this study represents as many as 50% of the total number of cases diagnosed between 1999-2005. As many as 90% of all breast cancer cases in Cyprus are registered, receive treatment and are followed by the Bank of Cyprus Oncology Centre in Nicosia. This centralized point of access, ensures that the study participants, are to a large extent representative of all cases on the island, in terms of their socio-demographic characteristics. It should be noted that the recruitment of cases was based on the list of scheduled appointments during the study period and not on a random selection of all diagnosed cases.

The 1177 controls were recruited amongst women who participated in the National breast cancer screening programme with the use of mammography, to a large extent the same population that would give rise to the cases. Since no matching was employed, unavoidable (small) differences in the age distribution between the cases and the controls (mainly due to the under-representation of younger and older age-groups amongst the mammography screening users) were dealt with by adjusting for age in multivariable analyses. The Cypriot National screening programme for breast cancer in all women aged 50-69 was officially introduced in 2006, after being pilot tested at a small-scale in 2003. Recent data from the Cyprus Ministry of Health suggest that more than half of the women invited respond to the call. Even so this does not imply that the sample of controls are representative of the general female population on the island, in terms of the background socio-economic characteristics and levels of potential risk factors for breast cancer in the general population.

In terms of the basic demographic characteristics of the two groups, no major differences were detected in the levels of education or marital status. However, well established risk factors for breast cancer identified in other populations, such as family history of breast cancer, age at menarche and breastfeeding exhibited the strongest associations with breast cancer risk among Cypriot women [4, 5, 10]. Unlike Britain and other European countries, there are no formal occupation-based socio-economic classification systems in Cyprus. In the absence of such socio-economic indicators or information on family income, educational status was used as a proxy. Controlling for the possible confounding effect of educational status in multivariable models did not affect inferences about major risk factors of breast cancer.

Healthy volunteers who participate in epidemiological studies tend to have a higher educational level, compared to the rest of the population [16]. In general, no major differences in the level of education between cases and controls were observed in this study. Nevertheless, it is possible that women who attend the screening programme are likely to be more educated, thus masking any such differences. Furthermore, a study of risk factors of breast cancer in Iran has shown that marital status may have an impact on the incidence of breast cancer in Iranian women. It was observed that women who never married were at higher risk for breast cancer [17]. However, marital status by itself may not be a determining factor for modifying breast cancer risk. Since there is a strong interaction between marital status and parity, the increased breast cancer risk associated with single women may possibly be due to nulliparity.

Many studies have examined the association between body mass index (BMI) and breast cancer incidence. It was observed that in postmenopausal women, obesity (BMI > 30 kg/m²) was associated with about 50% increase in breast cancer risk when compared with lean women (BMI 20 kg/m²). This association was not observed in premenopausal women. In contrast, in some studies it was observed that during the premenopausal years, breast cancer risk was slightly lower in obese women [18]. In our study, no information on menopause status at the time of diagnosis was available. By comparing BMI at the time of interview between cases and controls, no statistically significant differences were observed between the two groups. The relationship between BMI and breast cancer risk is complex and is better assessed in prospective studies where further anthropometric data are accurately collected at baseline and at regular intervals afterwards. In this context, no conclusive results regarding the association between BMI and breast cancer incidence in the Cypriot population can be drawn on the basis of this study.

Family history of breast cancer is one of the most well established, widely accepted risk factors for this disease. Having one first-degree relative with breast cancer approximately doubles a woman's risk for developing breast cancer. The risk is elevated significantly by increasing the number of affected relatives [19, 20]. Our findings are in accordance with other published studies and suggest that a positive family history of breast cancer is one of the most significant risk factors for this disease in Cyprus. In fact, if women with positive family history of cancer are more likely to attend mammography clinics for screening, the observed association may actually be an underestimation.

Breast cancer risk is associated with several reproductive factors. It is well established that breast cancer risk increases with early age at menarche [21–23]. This association is consistent with the hypothesis that breast cancer risk is related to the extent of breast mitotic activity. This activity is driven by estrogen and progesterone exposure during the luteal phase of the menstrual cycle [24], which determines the probability of tumorigenic somatic events [25]. Therefore, an early age at menarche increases the period during which the breast is mitotically active and subsequently increases breast cancer risk. Similarly to previous investigators, we observed that an early age at menarche is associated with an elevated risk of breast cancer in our population.

The effect of breastfeeding on breast cancer risk has been controversial indicating either no association, or a weak protective effect against breast cancer [26]. Studies in countries with a long duration of breastfeeding have reported substantial protective effects, whereas a number of studies in Western populations failed to detect an association, possibly due to the low prevalence of prolonged breastfeeding [26]. A meta-analysis of breastfeeding and breast cancer risk, by the Collaborative Group on Hormonal Factors in Breast Cancer, showed that increasing duration of breastfeeding confers a protective effect on breast cancer risk, over and above that already known to be afforded by parity itself [27]. Breastfeeding is hypothesized to reduce the risk of breast cancer primarily through two mechanisms, differentiation of breast tissue [28] and reduction of the lifetime number of ovulatory cycles [29]. The results of our study suggest an inverse association between breastfeeding and breast cancer risk, even though a dose-response relationship with self-reported duration of breastfeeding was not observed, possibly a result of misclassification. This finding is consistent with the results of the large collaborative study showing breastfeeding to be protective for breast cancer [27]. Breastfeeding is one of the few potentially modifiable factors that can reduce breast cancer risk. Further investigations are warranted in order to understand the underlying mechanisms of the protective effect of breastfeeding and how protection might be conferred. Once these are determined, interventions, which would mimic breastfeeding, could be developed for the benefit of women who have never breastfed [27].

Surprisingly, we observed an inverse association between hormone replacement therapy (HRT) and breast cancer risk. Results from randomized controlled trials and from observational studies have shown that use of HRT increases breast cancer risk. Furthermore, the effect of HRT on breast cancer risk depends also on the different combinations of HRT as well as on the duration of usage [30–34]. Overall, studies have concluded that oestrogen-progestagen combinations increase the risk of breast cancer, and that the risk was elevated, in women who were treated for at least 5 years [32, 34]. The above results should be interpreted with caution due to the difficulties experienced by Cypriot women in recalling information regarding type and exact duration of HRT use. However, there is no reason to believe that such recall was differential among cases and controls. Unlike other studies [34, 35], the vast majority of women in our sample reported never having used HRT while only a small percentage of women in our sample reported use for periods longer than 5 years (e.g. 7.8% among healthy controls). Furthermore, while in prospective studies one would expect to find an association between HRT use and breast cancer risk, in our retrospective study it is not surprising to observe an inverse association. It is very likely though that cases, once diagnosed with breast cancer, were not prescribed with HRT which would explain the reverse association seen in our study.

Only a weak association was observed between lifestyle risk factors such as exercise and smoking. It should be noted though that these parameters refer to current status only, as no information for past habits, duration or intensity of smoking were available. As a result, any association is hard to interpret due to problems with directionality of effect e.g. associations with exercise may be more likely to reflect that women diagnosed with breast cancer are less likely to currently exercise, rather than a causal relationship between lack of exercise and breast cancer.

The strongest associations with breast cancer risk in the Cypriot population were observed with family history of breast cancer, early menarche and breastfeeding. It is noteworthy that the protective effect of breastfeeding appeared stronger than the effect of pregnancy alone. Overall, the findings of our study corroborate with the results of previous investigations on descriptive epidemiology of risk factors for breast cancer. Nevertheless, this study presents the first report for breast cancer risk factors in Cypriot women and provides the first scientific evidence using National data, for executing more targeted campaigns of prevention and early diagnosis in the Cypriot population.