Influence of radiotherapy for the first tumor on aggressiveness of contralateral breast cancer

Authors

  • Maria E.C. Sandberg,

    Corresponding author
    1. Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden
    • Department of Medical Epidemiology and Biostatistics, Karolinska Institute, Box 281, 171 77 Stockholm, Sweden
    Search for more papers by this author
    • Tel: +46 8 52483985

  • Sara Alkner,

    1. Department of Oncology, Clinical Sciences, Lund University, Lund, Sweden
    2. Clinic of Oncology, Skåne University Hospital, Malmö, Sweden
    Search for more papers by this author
  • Mikael Hartman,

    1. Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden
    2. Department of Epidemiology and Public Health and, National University of Singapore, Singapore
    3. Department of Surgery, National University of Singapore, Singapore
    4. Center for Molecular Epidemiology, National University of Singapore, Singapore
    Search for more papers by this author
  • Sandra Eloranta,

    1. Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden
    Search for more papers by this author
  • Lisa Rydén,

    1. Department of Surgery, Clinical Sciences, Lund University, Lund, Sweden
    2. Clinic of Surgery, Skåne University Hospital, Malmö, Sweden
    Search for more papers by this author
  • Alexander Ploner,

    1. Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden
    Search for more papers by this author
  • Hans-Olov Adami,

    1. Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden
    2. Department of Epidemiology, Harvard School of Public Health, Boston, MA
    Search for more papers by this author
  • Per Hall,

    1. Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden
    Search for more papers by this author
  • Kamila Czene

    1. Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden
    Search for more papers by this author

Abstract

We aimed to investigate if characteristics of contralateral breast cancer (CBC) are influenced by adjuvant radiotherapy for the first breast cancer. Using information from population-based registers and medical records, we analyzed two cohorts comprising all women with CBC diagnosed >3 months after their first cancer (809 patients in Stockholm 1976–2005 and 750 patients in South Sweden 1977–2005). We used Poisson regression to calculate risk of distant metastasis after CBC, comparing patients treated and not treated with radiotherapy for the first cancer. Logistic regression was used to estimate odds ratio (OR) of more aggressive tumor characteristics in the second cancer, compared to the first. For patients with CBC in Stockholm with <5 years between the cancers radiotherapy for the first cancer conferred a nearly doubled risk of distant metastasis [incidence rate ratio (IRR) = 1.91; 95% confidence interval (CI): 1.27–2.88], compared to those not treated with radiotherapy. This was replicated in the South Swedish cohort [IRR = 2.12 (95% CI: 1.40–3.23)]. In Stockholm, we found an increased odds that, following radiotherapy, a second cancer was of more advanced TNM-stage [OR 2.16 (95% CI 1.13–4.11)] and higher histological grade [OR = 2.00 (95% CI 1.08–3.72)] compared to the first, for patients with CBC with <5 years between the cancers. No effect on any of the investigated outcomes was seen for patients diagnosed with CBC >5 years from the first cancer. In conclusion, patients diagnosed with CBC within 5 years had worse prognosis and more aggressive tumor characteristics of the second cancer, if they had received radiotherapy for their first cancer, compared to no radiotherapy.

Approximately 10–15% of all patients with breast cancer will be diagnosed with contralateral breast cancer (CBC) during the first 20 years after initial diagnosis.1, 2 The prognosis of women with CBC has been shown to be worse than for patients with unilateral disease.3, 4 Several studies have shown that CBC prognosis is highly dependent on time between first and second cancer3, 5; women diagnosed with CBC within 5 years from the first cancer have a particularly bad prognosis. Prognosticators for CBC has not been studied in detail but seem to include tumor characteristics of the first and second cancer, age at diagnosis of first cancer and time interval between the first and second cancer.6, 7

Radiotherapy is well known to protect against local recurrence and improve breast cancer-specific survival.8, 9 In Sweden, approximately 60% of all patients with breast cancer receive adjuvant radiotherapy.10 The common dose of 48–50 Gy following surgery11 have been very similar from 1980s until the end of the study, and somewhat lower during the 1970s (2945 Gy),12 such a dose entails scattered radiation to the contralateral breast, estimated to 23 Gy13–15 or about 5% of the dose to the treated breast.16 Because such doses may be carcinogenic, the risk of CBC following adjuvant radiotherapy for breast cancer has been studied for several decades with inconclusive results.2, 9, 13, 15 However, no studies have investigated if prognosis after the second cancer and/or malignant features of the second cancer is associated with adjuvant radiotherapy for the first cancer.

We aimed to investigate the hypothesis that radiotherapy for the first cancer enhances the malignant features of the second cancer in the increasingly large group of patients with CBC. To achieve this aim, we analyzed two large, population-based cohorts of patients with CBC for which we had accessed all medical records comparing all patients treated with radiotherapy to all patients not treated. Features of the contralateral cancer were measured in three ways: the risk of distant metastasis after CBC, the odds that the second cancer had more advanced TNM-stage and the odds that the second cancer had higher histological grade, compared to the first.

Methods

Study population

The population-based Stockholm Breast Cancer Register records all incident breast cancers diagnosed in the Stockholm health-care region (a catchment area of 1.5 million inhabitants) since 1976 and contains information on date of birth, date of diagnosis of both tumors, TNM-stage and location of the tumors.

From the Stockholm Breast Cancer Register, we selected all patients with CBC diagnosed more than 3 months after the first invasive breast cancer. Participants had to be female and diagnosed with both cancers within the Stockholm health care region between the years 1976 and 2005. Both cancers had to be diagnosed in stage I–III with no other recorded malignancy before diagnosis of the second cancer. We excluded patients with stage IV breast cancers or patients with other prior malignancy to minimize the risk that metastatic disease was misclassified as a primary contralateral cancer. Altogether 936 patients fulfilled these inclusion criteria. Follow-up continued until the date of distant metastasis, the date of death or December 31, 2006, whichever came first.

We reviewed the oncological medical records and the pathology reports from 99% of all eligible patients. We extracted additional information that was not available in the register (information on histological grade and adjuvant therapy for both cancers). For 109 patients (12%), the review of medical record and pathology reports revealed that the inclusion criteria were not met, and they were excluded from the cohort; the majority of these, 77 (8%) had distant metastasis before diagnosis of the second breast cancer. For ten cases (1%), the medical record could not be retrieved and they were also excluded. Finally, we also excluded eight patients due to unknown treatment for first breast cancer. The final study cohort consisted of 809 patients with CBC.

To validate our results, we analyzed a second cohort of patients with CBC. The South Swedish cohort included all women in the Southern health care region of Sweden (a catchment area of 1.7 million inhabitants) diagnosed with the second primary breast cancer between 1977 and 2005. These patients were identified from the South Swedish Breast Cancer Register, and the surgical and oncological medical records as well as the pathology reports were reviewed, according to the same criteria as the Stockholm cohort. However, in the South Swedish cohort, information on histological grade was available for less than one-third of the patients, and no information was available on tumor stage. The final cohort consisted of 750 patients with information on treatment of their first breast cancer, age at first diagnosis, calendar period of first diagnosis and date of diagnosis of distant metastasis, death or date of end of follow-up.

Analyses

Our analytical approach was to investigate the risk of distant metastasis after CBC and the odds that the second cancer had more advanced TNM-stage or higher histological grade compared to the first (Fig. 1). All patients with CBC treated with adjuvant radiotherapy were compared to all patients with CBC not treated with adjuvant radiotherapy, regardless of systemic adjuvant treatment such as endocrine- or chemotherapy. In our secondary analyses, we compared all patients treated with adjuvant radiotherapy only to patients not receiving any adjuvant therapy. Consistently, we calculated our estimates for CBCs overall and then separately for CBCs occurring within and beyond 5 years from the first cancer, by including interaction terms.

Figure 1.

Illustrating the analytic approach, showing the time line of the study participants, with the two possible outcomes (distant metastasis or no distant metastasis). Poisson regression was used to calculate risk of metastasis and logistic regression was used to calculate the odds of a more advanced TNM-stage at the second cancer compared to the first, and the odds of a higher histological grade at the second cancer than the first.

We used Poisson regression to estimate the incidence rate ratio (IRR) as a measure of risk, of distant metastasis following the second cancer, comparing patients treated to those not treated with adjuvant radiotherapy. The risk estimates were adjusted for calendar period (in 5-year categories) and age at diagnosis (in 10-year categories) of the first cancer, as well as histological grade and TNM-stage of both cancers, adjuvant therapy of the second cancer (radiotherapy, chemotherapy and endocrine therapy and combinations thereof) and time interval between the first and second cancer (latency time; ≤5 years and >5 years). The underlying time scale was time since the second cancer. We further included an interaction term in the model, allowing the effect of adjuvant radiotherapy to vary over latency time.

Logistic regression was used to estimate the odds ratio (OR) of a more aggressive second cancer, adjusted for calendar period of first diagnosis (in 10-year categories), age at first diagnosis (in 10-year categories) and latency time, contrasting radiotherapy-treated patients to non-treated patients. Our outcomes of interest were more advanced TNM-stage at the second cancer, compared to the first and higher histological grade at the second cancer, compared to the first. If the first cancer was diagnosed in stage III, it is not possible to have a more advanced stage of the second cancer. To accommodate this, we also defined our outcome as being diagnosed with the highest stage at both cancers. The same definitions were used for the analysis of histological grade.

All data preparation and analyses were done using SAS Statistical Package 9.2 and Stata 10.1. The studies were approved by the ethical committees at the Karolinska Institutet and Lund University, Sweden. The funding sources of this study had no part in study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.

Results

Descriptive data

Mean age at first diagnosis among the 809 patients with CBC in the Stockholm cohort was 56.8 years, which was very similar to the 750 patients in the South Swedish cohort; 57.8 years (Table 1). The average follow-up time after second breast cancer diagnosis was 6.4 years (5.6 years for patients treated with radiotherapy and 7.3 years for patients without radiotherapy treatment) in the Stockholm cohort and 6.2 years (6.0 years for radiotherapy treated patients and 6.4 years for untreated patients) in the South Swedish cohort. Overall, approximately half of the patients with CBC received adjuvant radiotherapy (52% in Stockholm and 62% in the south of Sweden). Distant metastasis after the second breast cancer diagnosis were diagnosed in 203 patients (25%) and 226 patients (30%) in the Stockholm and South Swedish cohort respectively.

Table 1. Characteristics of the patients in the Stockholm cohort and South Swedish cohort, showing the proportion exposed adjuvant radiotherapy for the first cancer, and the total cohort
inline image

Risk of distant metastasis

In the Stockholm cohort, the risk of distant metastasis following the second breast cancer diagnosis was 43% higher [IRR 1.43; 95% confidence interval (CI): 1.041.95] among patients who received radiotherapy as part of their first adjuvant treatment, compared to patients who did not receive radiotherapy (Table 2). This excess risk decreased with follow-up time as indicated by a significant interaction (p = 0.02) between radiotherapy and latency time between the first and the second breast cancer. When the CBC was diagnosed within 5 years, radiotherapy increased the risk of distant metastasis by 91% [IRR 1.91 (95% CI: 1.272.88)], while no excess risk was seen when the time interval exceeded 5 years. The risk estimates did not change substantially when we restricted adjustments to calendar period, age at first diagnosis and latency time (Table 2). Our findings were confirmed when analyzing the south Sweden cohort (Table 2). Further, the Stockholm cohort and south Sweden cohort were combined and the IRR of distant metastasis for patients treated with radiotherapy compared to those not treated was 1.42 (95% CI: 1.161.74) (data not shown in Table). Among the patients with latency time <5 years, the IRR was 1.92 (95% CI: 1.462.53), and for patients with latency time >5 years, the IRR was 0.96 (95% CI: 0.711.30). This combined analysis was adjusted for the factors for which we had information in both cohorts (age, calendar period and latency time).

Table 2. Incidence rate ratio of distant metastasis after CBC
inline image

Finally, we restricted both the Stockholm cohort and the south Sweden cohort to patients treated with radiotherapy only and patients with no adjuvant therapy, we found a similarly increased risk of distant metastasis for the radiotherapy treated patients, and the effect was again confined to the patients with short latency time (Table 2). In the Stockholm cohort, where we had information on histology, we also restricted the cohort to patients with ductal histology at the second cancer and found a similar pattern; IRR of distant metastasis for patients with latency time <5 years: 1.81 (95% CI: 1.033.17) and for patients with latency time >5 years: 0.96 (95% CI: 0.541.74) (data not shown in Table).

Doses of scattered radiation are higher to the medial than to the lateral part of the contralateral breast.13 In the Stockholm cohort, we had valid information on location (medial/lateral) of the second breast cancer in a subset of 322 (40%) of the patients with second breast cancers occurring within 5 years. This information allowed us to analyze malignant features of the second cancer by location. The excess risk of distant metastases following radiotherapy was higher among women whose second cancer was located in the medial [IRR: 3.03 (95% CI: 1.108.32)] than in the lateral [IRR: 1.46 (95% CI: 0.703.02)] part of the breast. This analysis was adjusted for calendar period and age at first diagnosis and latency time between the cancers.

In all analyses above, we have investigated the effect of adjuvant radiotherapy for the first breast cancer. To assess possible bias by indication, we also calculated the IRR of distant metastasis after radiotherapy for the second breast cancer. We saw no evidence of a risk increase for the radiotherapy-treated patients; indeed, for patients with latency time <5years, we saw a protective effect: [IRR: 0.39 (95% CI: 0.240.63)]. This analysis was repeated for patients without systemic adjuvant therapy and neither among these patients did we see any risk increase.

Odds of worse tumor characteristics

The overall odds for higher stage of the second cancer following adjuvant radiotherapy was not significantly higher in the radiotherapy-treated group compared to the nontreated [OR = 1.36 (95% CI: 0.882.09); Table 3], but we observed evidence of interaction with latency time (p = 0.05). Hence, patients with CBC diagnosed within 5 years from the first cancer were at a twofold higher odds of being diagnosed in a more advanced stage, if they had previously received radiotherapy [OR 2.16 (95% CI: 1.134.11)], while we found no such association when the latency time was longer than 5 years. A similar pattern emerged when we compared histological grade of the first and the second cancer (Table 3); again with a significant interaction between latency time and radiotherapy (p = 0.02), and a statistically significantly increased odds [OR = 2.00 (95% CI: 1.083.72)] for treated vs. non-treated women diagnosed with CBC within 5 years. This association was even stronger both for more advanced stage [OR = 2.48 (95% CI: 1.317.58)] and higher grade [OR = 3.15 (95% CI: 0.956.49)] in analyses restricted to women who received no systemic adjuvant therapy (Table 3).

Table 3. Odds of more aggressive tumor characteristics (TNM-stage/histological grade) at second cancer than at first cancer
inline image

Discussion

In two large, independent cohorts of patients with CBC, we observed an increased risk of distant metastasis for patients with CBC treated with adjuvant radiotherapy for their first cancer, compared to patients with CBC not treated with adjuvant radiotherapy. This association was strongest among women diagnosed with the second cancer within 5 years from the first; for these women the risk was almost twofold. In the same group, we also observed twofold increased odds for the second cancer to be of more advanced stage and higher grade, compared to the first, for the treated patients.

Confounding by indication is a common problem in observational studies that investigate treatment effect. Such bias arise when cancers of more aggressive characteristics are treated more intensely, thereby feigning an association between the treatment and a worse prognosis. The decision to give radiotherapy is based not only on the aggressiveness of the cancer but also on the surgical technique, age of the patient and the patients' wishes. Still, to accommodate this potential bias, we have adjusted our analysis of distant metastasis for TNM-stage and histological grade of the two cancers. Our analysis of TNM-stage and histological grade compares the tumor characteristics between the first and the second cancer of the same patient, to avoid the potential bias by indication that would lead to exposed patients overall having higher stage/worse grade compared to unexposed patients. Because the original comparison groups consist of a combination of different therapies, we also restricted our analyses to patients who received no systemic adjuvant therapy, and this did not change our findings. Our final approach to investigate bias by indication was to analyze the risk of distant metastasis in relation to radiotherapy for the second breast cancer. If more aggressive cancers are selected for radiotherapy and also have an increased risk of distant metastasis, the same mechanisms should be in play for both cancers. Reassuringly, we found no evidence of increased risk of distant metastasis after radiotherapy for the second breast cancer. Further, the potential bias by patient characteristics like socioeconomic status and comorbidity would, if present, give rise to an association in opposite direction of what we observe. All in all, we do not believe that bias by indication could be the sole explanation for the findings presented in this study.

Strengths of this study include that the main results were replicated in two independent populations and that it was conducted in a country with a unified healthcare system and a population-based cancer register, which enabled close to complete case identification and very good retrieval of medical records (99% in the Stockholm cohort and 83% in the south Sweden cohort).

Women with CBC are commonly reported to have a worse survival than women with unilateral breast cancer.3, 17–20 A few studies indicated that the latency time of the metachronous CBC had an impact.3, 5, 19 In the present study, we show that the prognosis after CBC is worsened by radiotherapy for the first cancer, especially for CBCs with short latency time (Table 2). This association is independent of TNM-stage and histological grade, because our analysis is adjusted for those factors. Tumor stage of the first and second cancer has been well established to be associated with overall prognosis in patients with CBC.6, 20, 21 Progesterone receptor status, size of the second tumor and histological grade of both cancers has also been related to prognosis.21 In our analysis, we have demonstrated a twofold higher risk for a radiotherapy-treated patient, compared to a non-radiotherapy treated, to have higher stage and grade of the second cancer than of the first among patients whose second cancer occurred within 5 years from the first (Table 3).

The surgical and radiation techniques potentially affect the radiation dose to the contralateral breast. We lack information on the individual dose to the contralateral breast, which precludes an analysis of a possible dose–response relationship. However, if radiotherapy affects the risk of distant metastases, we would expect this risk to be more pronounced for women having second tumors occurring in the medial part of the opposite breast, because the radiation dose is higher to that part, which is closer to the treated breast,13 and we indeed found indications of this.

As our findings are confined to cancers arising within 5 years from the first cancer, radiotherapy does not appear to induce new cancers—since this process is likely to take much longer time22 but rather to affect biologic features of subclinical cancers or premalignant lesions in the contralateral breast. Interestingly, studies on random biopsies of the opposite breast of patients with unilateral breast cancer showed that 6–7% of the patients had invasive or in situ malignancies and an additional 9% had premalignant lesions.23, 24 In other words, approximately 15% of all patients with breast cancer have lesions in the contralateral breast with the possibility to progress into CBC. We hypothesize that if such lesions are exposed to a carcinogenic stimulus such as low-dose ionizing radiation, this can enhance rate of mutations and thereby accelerate tumor progression and growth. Further, radiation might affect the prognosis of already existing cancer/premalignant lesions via stiffening the extracellular matrix. Radiation has been shown to have this effect, and it is well known that rigidity of the extracellular matrix promotes cell migration and tumorigenesis.25, 26 Furthermore, premalignant cells transplanted into mice with irradiated extracellular matrix grow into more and larger tumors compared to premalignant cells transplanted into non-irradiated mice.25

In light of the current discussion on prophylactic mammary irradiation,27 this study illustrates an important aspect. Prophylactic mammary irradiation was first suggested in 2007 by Brenner et al.27, 28 to lower the risk of second breast cancer. It is well known that radiation can both cure cancer by killing cancer cells and initiate cancer, dependent on the dosage given.29 In addition, our study suggests that if a cancer/premalignant lesion is already present, but undetected, low-dose radiation will on average, result in a more malignant phenotype.

In conclusion, patients who are treated with adjuvant radiotherapy for their first breast cancer and who are diagnosed with CBC within 5 years seem to have a more aggressive second cancer. The clinical implication of this finding is not to refrain from adjuvant radiotherapy, but to minimize radiation to the contralateral breast (possibly by using accelerated partial breast irradiation30 when appropriate) and to take given therapy into account in the management of the second cancer. A more intense treatment than indicated by the tumor characteristics of the second cancer might be called for. In the view of prophylactic mammary irradiation as an approach to decrease the risk of second breast cancer, it will be important to assess the validity of our findings in other populations and evaluate how the (potentially) decreased risk of second breast cancer relates to the (potentially) increased risk of metastasis, should a second breast cancer anyway occur.

Acknowledgements

The authors acknowledge the Regional Oncological Center in Stockholm and the Stockholm Breast Cancer Group for access to the Stockholm Breast Cancer Registry, the South-Swedish Oncological Centre and the South-Swedish Breast Cancer Group for access to the South Swedish Breast Cancer Register and the helpful staff at the surgical and oncological clinics and medical archives. Kamila Czene was financed by the Swedish Cancer Society grant no: 5128-B07-01PAF.

Ancillary