Breast cancer is the most common type of cancer and the major cause of cancer-related mortality among women worldwide [1]. Many patients are diagnosed with this disease each year and are often treated with surgery followed by adjuvant radiation therapy [2]. The use of radiation therapy has increased dramatically since randomised trials in the 1980s showed equivalent outcomes for patients treated with breast conserving surgery and radiation therapy and those treated with modified radical mastectomy [3–5]. With advances in early diagnosis and treatment, breast cancer is becoming an increasingly survivable disease resulting in a large population of long-term survivors. Recent trials have shown an overall survival benefit in favour of adjuvant radiotherapy for breast cancer [6–8]. Nevertheless, there is clear evidence for the association between radiation exposure and cancer, especially from epidemiological studies of survivors of the atomic bombings in Japan [9, 10], as well as from various studies of medically-exposed groups [11]. In particular, irradiation of surrounding tissues during breast radiotherapy can cause second cancers to develop within these tissues [12, 13]. The second malignancy refers to a new primary cancer in a person who has survived an earlier cancer. The probability of a radiation induced second malignancy after radiotherapy is a topic that has been widely discussed [14–20]. While the benefit of radiotherapy should outweigh the risks of developing subsequent cancers, it is important to evaluate the long-term consequences of breast cancer treatment.

The aim of this cohort study was to evaluate the incidence of second primary cancers in a group of patients treated with ionising radiation therapy for breast cancer in comparison to patients not treated with ionising radiation therapy. The study involved the analysis of clinical records for female breast cancer patients treated at the University of Florence (Italy) with radiotherapy and/or chemotherapy and/or hormonal therapy from 1965 to 1994 and who were subsequently followed-up [21]. In this paper, we evaluate the effect that radiotherapy may have had on the subsequent risk of second malignancies, including the possible influence of age at treatment and menopausal status.

Data were collected on 5,248 patients with breast cancer who were submitted to radiotherapy, chemotherapy, hormonal therapy or no additional therapy at the University of Florence from June 1965 to December 1994. All of the patients had received surgery for breast cancer. A treatment schedule of 2 Gy/day, 5 days/week, for a total dose up to 60 Gy was used. Of these patients, 51.0% received radiotherapy and, for 74.0% of the patients in this group, the exposed volume was the breast only, compared with breast and lymphatic drainage in 26.0% (in 17.5% the mammary chain and supraclavicular nodes were irradiated and in 8.5% the chest wall was irradiated). The radiation source was a tele-cobalt unit (dose rate 0.80 to 1.20 Gy/minute) or linear accelerator. Different irradiation techniques with a 6 MV X-rays linear accelerator were used in later years, namely: (a) two tangential fields; or (b) two tangential fields plus direct irradiation with an electron beam of 12 MeV. Values calculated on 80 treated patients showed that, in case (a), 12% of the total dose was absorbed at a distance of 3 cm from the irradiated volume and that this value decreased progressively to 2.4% at 12 cm. In case (b), the absorbed dose was 3.8% of the total at 3 cm and 1.4% at 12 cm.

Follow-up for second malignancies was mainly conducted through direct contact with the patients at regular visits at out-patient clinics. On some visits information on health status was also verified by accompanying relatives. For this study, the cohort was assembled in 1996 and data were collected retrospectively.

The Cox proportional hazards model for ungrouped survival data [22] was used to estimate the relative risk for second cancers after radiotherapy treatment and to evaluate how the risk varied according to other factors. In particular, the hazard function at a given time was modelled as λ(t) = λ ₀ (t)•exp[βx], where λ ₀ (t) is the baseline hazard function, x represents one or more explanatory variables, and βx is the logarithm of the relative risk. Parameter estimation and significance tests were carried out using the Epicure software [23]. In particular, Wald-type confidence intervals were computed for parameter estimates, whilst significance tests were based on χ² approximations to the distribution of the log likelihood ratio. Some patients received chemotherapy and/or hormonal therapy in both radiotherapy and non-radiotherapy groups; therefore, the relative risks of second cancers due to radiotherapy were reported both unadjusted and adjusted for chemotherapy and hormonal therapy to check if there were any confounding effects from these therapies.

The researchers carrying out the study had no identifiable details of patients forwarded to them and, therefore, ethical approval was not required under Italian laws when the project was initiated in 1996.

In this study, we have used a clinical records-based cohort to analyse the effects of radiotherapy for breast cancer on the incidence of subsequent second cancers. All patients in this cohort received surgery for breast cancer. The advantage of this approach is to minimise any possible systematic difference between the irradiated and non-irradiated groups by reducing variability in the analysis. Such an approach has been used in other epidemiological studies of similar nature, as reported [17].

In our analysis, the relative risk for all second cancers appeared to be increased five or more years after radiotherapy compared to those non-irradiated. This is in agreement with the epidemiological evidence for a latency period of several years between exposure to radiation and a raised risk of second cancers [20, 24–26]. The increased relative risk for all second cancers was observed in the 50 to 64 and 65+ years age-at-treatment groups (Table 4). Since these groups largely consist of post-menopausal patients, this finding may indicate an association with menopausal status. Raised relative risks were indeed observed among patients who were post-menopausal or underwent ovariectomy and, in the case of post-menopausal patients, were statistically significant. For all second cancers other than contralateral breast cancer and leukaemia, the risk variation by age-at-treatment was slightly different (Table 8). Here the increased risk was seen principally in the 50 to 64 year age-at-treatment group. Nevertheless, the raised risk was still linked with patients who had post-menopausal status or underwent ovariectomy after adjustment for chemotherapy and hormonal therapy, for example, RR = 2.16 (95% CI 1.22, 3.84) for post-menopausal patients; RR = 1.87 (95% CI 0.75, 4.66) for patients who underwent ovariectomy and RR = 1.05 (95% CI 0.52, 2.12) for pre-menopausal patients. These findings are generally in agreement with other studies that reported older age was associated with an increased risk of a non-breast second malignancy [19] and that most of the secondary lung cancers occurred more than five years after radiotherapy and in women who were >50 years at the time of their breast cancer diagnosis [18].

Increased risk of leukaemia can arise two to five years after exposure to radiation [27]. In our analysis, there was suggestion of a raised incidence of leukaemia among radiotherapy patients in the period two or more years after radiotherapy. There were seven cases in the radiotherapy group compared with only one case in non-radiotherapy group, with a relative risk of 6.67 (95% CI 0.76, 58.00) after adjustment for chemotherapy and hormonal therapy. The raised risk was not statistically significant, reflecting the small number of cases in this cohort. An increased risk of leukaemia has also been reported in previous epidemiological studies of breast cancer patients treated with radiation [28]. This raised risk might be associated with regional radiation therapy that includes an internal mammary node field, which may expose the thoracic spine to a relatively high radiation dose [13]. Due to the small number of leukaemia cases in this cohort, and there is no information available on leukaemia subtypes, analyses cannot be performed regarding leukaemia subtypes after radiotherapy in this study.

Raised risks of breast cancer have been reported in various studies of women exposed to radiation; for example, Japanese atomic bomb survivors [9, 10, 29], female tuberculosis patients who received multiple fluoroscopies [30, 31], and female patients who received radiotherapy for various benign conditions [32]. The risk was also seen in women who had direct breast exposure prior to the age of 30 years [31–35]. However, the causes of contralateral breast cancer amongst breast cancer patients given radiotherapy are less obvious. In the Early Breast Cancer Trialists' Collaborative Group report which evaluated the effects of radiotherapy, a significantly increased risk of contralateral breast cancer was found [5]. However, in another large case-control study from Denmark, there was no significant raised risk of contralateral breast cancer among women who received radiotherapy [36]. A more recent large-scale study included 13,472 women also failed to show an increased risk of contralateral breast cancer for those received radiotherapy [37]. In some studies, it was reported that the increased risks of contralateral breast cancer were most likely observed within the first year following diagnosis of the primary breast cancer [38], or associated with patients with more advanced stage disease [39–41]. Since some patients selected for post-mastectomy radiation have a poorer prognosis than other patients, there may well be bias in estimates of the risk of contralateral breast cancer that can be attributed to radiotherapy.

In our analysis, no raised relative risk for contralateral breast cancer was observed during the period five or more years after exposure. This is in agreement with previous epidemiological studies [13, 36, 37, 42]. Since our analyses excluded the first five years following treatment, we believe that this has excluded any effect of metastatic disease in the opposite breast. However, our result is based on a small number of cases, which makes it difficult to detect any raised risk.

Other solid cancers have also been reported to link with radiotherapy following breast cancer [14, 43, 44]. In our analyses, the relative risk for all second cancer excluding leukaemia and contralateral breast cancer appeared to be increased five or more years after radiotherapy. A statistically significantly increased risk was observed in the 50 to 64 year age-at-treatment group. This may indicate an association with menopausal status.

With a fairly small total number of second cancers in this cohort, the excess risk associated with radiotherapy was small over the period of follow-up. Since many of the women were still alive at the end of follow-up, the possibility of raised risks continuing several decades over radiotherapy cannot be ruled out and - based on other studies [10] - would be expected.

The study of second cancers involves special problems not inherent to most epidemiological investigations. For a second primary cancer to be classified correctly, a metastatic lesion or local recurrence of the original primary cancer must be ruled out. Other treatments may affect subsequent cancer risks for certain organs, for example, the ovary or uterus might be surgically removed. Direct comparison between the irradiated and non-irradiated groups is not without problems. The non-exposed women generally have an earlier stage disease than do women treated with radiation, and the expected number of second cancers among long term survivors is often too small, as in our analysis, to permit meaningful direct comparisons on a site-by-site basis.

It can also be argued that some of the second cancers might occur due to factors other than radiotherapy, such as a family history of breast cancer. It was estimated that up to seven percent of breast cancer cases are estimated to be due to breast cancer susceptibility genes (for example, BRCA1, BRCA2, p53 and PTEN) [45]. Although we do not have information on family history of breast cancer, a previous study [46] reported that there were increased risks of second cancers of the oesophagus, stomach, ovary, non-Hodgkin's lymphoma and leukaemia in women with a family history of breast cancer compared to those without such a history. There might also be a joint effect of radiotherapy and cigarette smoking on second cancer such as lung cancer. It was reported [43] that radiotherapy increased the risk of second primary lung cancer especially among ever smokers. Although we cannot adjust for cigarette smoking in our analysis, the effects of radiotherapy on lung cancer are evident in our cohort since all four lung cancer cases were occurred in the radiotherapy group (Table 2). This is consistent with previous studies on lung cancer risk after radiotherapy treatment [47, 48]. The region of treatment may also affect the risk of the second cancer. No correlation was found with the region treated with radiation but there was a trend for a higher risk of second malignancy when the internal mammary nodes were treated [49]. Our study supports this finding since 26% of radiotherapy patients in this cohort had the mammary chain and supraclavicular nodes or chest wall irradiated.

It appears to be that the patients used radiotherapy along with chemotherapy or/and hormonal therapy had less risks of developing second cancers compared with those patients received radiotherapy alone. However, we do not have detailed information about the staging of breast cancer at the time of diagnosis and what criteria were based on for the prescription of different therapies. Therefore, it is difficult to explain the reasons why the use of chemo- and/or hormonal therapy combined with radiotherapy might lead to a lower risk of second cancer compared to the use of radiotherapy alone. However, it is interesting to note that women who only received chemotherapy or hormonal therapy did not have raised risk of second cancer, although numbers of cases may be too small to detect any raised risk in these subset analyses.

This study indicated a raised risk of second malignancies associated with radiotherapy for breast cancer. The relative risk for second cancers appeared to be highest for the 50 to 65 years age-at-treatment group during the period five or more years after radiotherapy and this might be related to the patients' menopausal status. Among specific types of cancer, there were raised risks for leukaemia and for all cancers combined except leukaemia and contralateral breast cancer. The risk of contralateral breast cancer was not raised during the period five or more years after radiotherapy.