Radiotherapy did not increase thyroid cancer
risk among
women with breast cancer: A nationwide
population-based
cohort study
Li-Min Sun1*, Cheng-Li Lin2,3, Ji-An Liang4,5*, Wen-Sheng Huang6 and Chia-Hung Kao4,7
Breast cancer is the most frequently diagnosed malignancy in women worldwide; for example, an estimated 1.67 million new cancer cases were diagnosed in 2012.1 Breast cancer has
also been the most common type of cancer diagnosed among women in Taiwan since 1996. According to statistics from the Department of Health, Executive Yuan, Taiwan, the
breast cancer incidence rate increased by 19.6% from 2008 to 2011.2 The age-adjusted incidence rate of breast cancer in
Taiwan has increased steadily and reached 74.63 new cases (including 64.28 invasive carcinoma cases and 10.35 carcinoma in situ cases) per 100,000 people in 2011.3 Moreover,
the thyroid cancer incidence rate has also increased gradually in Taiwan. The age-adjusted incidence rate of thyroid cancer
among Taiwanese women was 8.74 per 100,000 women in 2000–2006,4 reaching 13.45 per 100,000 women in 2011 and
ranking the sixth most common cancer among Taiwanese women.3
Because of successful cancer screening, earlier detection, advanced diagnostic tools, more effective treatments, and an aging population, the proportion of long-term survivors of breast cancer is rising. Consequently, monitoring for breast cancer survivors has become crucial, not only for controlling the disease, but also for managing cancer or treatmentrelated health problems.5 Previous studies have suggested
6–10 although no really excess subsequent thyroid cancer
development have also been reported.11,12 Both breast cancer
and thyroid cancer are hormone-related cancers that involve a unique mechanism of carcinogenesis. Endogenous and exogenous hormones drive cell proliferation, and thus, the risk for accumulating random genetic errors increases.13 This
phenomenon may partially explain to the observation of a
higher incidence of thyroid cancer among breast cancer survivors than among the general population. However,
researchers have considered that treatment-related factors, especially that of radiotherapy (RT), may be involved in this possible relationship.6,10,14
In this study, we used data from a Taiwan nationwide
database to clarify this relationship. The first aim of this retrospective cohort study was to evaluate whether breast cancer
survivors have an increased incidence of thyroid cancer. Our second aim was to determine whether RT plays a role in the increased risk of developing thyroid cancer.
Methods Data source
In March 1995, National Health Insurance (NHI), a compulsory and universal program, was implemented in Taiwan,
covering approximately 99% of the country’s 23.75 million residents.15 The National Health Research Institutes established
and maintains the NHI Research Database (NHIRD),
releasing data annually to the public for research purposes. The NHIRD contains medical information, including that for inpatient and outpatient care facilities, drug prescriptions,
sex, date of birth, dates of visits or hospitalizations, and diagnoses coded in the format of the International Classification
of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM). For this study, we used a subset of the NHIRD containing
health care data from files in the Registry for Catastrophic Illness Patient Database (RCIPD), Longitudinal Health Insurance
Database 2000 (LHID2000), and Registry of Beneficiaries. Personal identification information of the insureds was scrambled
to for privacy protection. This study was approved by the Ethics Review Board of China Medical University
(CMU-REC-101-012).
Sampled participants
This retrospective cohort study used data extracted from the RCIPD and LHID2000 for the period 2000–2008. We
selected women aged 20 years or older, newly diagnosed with breast cancer (ICD-9-CM code 174) during 2000–2008 from the RCIPD as the breast cancer cohort, which was divided into two groups according to RT history. The index date for each patient receiving RT was the first treatment day of RT. The index date of each patient receiving no RT was randomly assigned according to a patient receiving RT. Finally,
we extracted 55,318 patients with breast cancer who had no history of thyroid cancer (ICD-9-CM code 193) or other cancer (ICD-9-CM codes 140-208) before the index date. Among these, 28,187 patients received RT and 27,131 did not. Three non-breast cancer controls for each breast cancer case were frequency-matched with the breast cancer cohort according to age group with an interval of 5 years and the year of index date and were subject to the same exclusion criteria. A total of 165,954 controls comprised the comparison control cohort. The person-years of the follow-up were estimated for the study participants beginning from the index date until the date of thyroid cancer diagnosis, censoring because of failure to follow-up, withdrawal from the insurance program, or December 31, 2011.
Comorbidities, medications and treatments
The baseline comorbidities and medications for each participant were assessed, including coronary artery disease (CAD;
ICD-9-CM codes 410-414), asthma (ICD-9-CM code 493), chronic obstructive pulmonary disease (COPD; ICD-9-CM codes 491, 492, and 496), stroke (ICD-9-CM code 430-438), alcohol-related illness (ICD-9-CM 291, 303, 305, 571.0–571.3, 790.3, A215 and V11.3), hypertension (ICD-9-CM codes 401-405) and diabetes (ICD-9-CM code 250) and medications
including steroids, thiazide diuretics, and statins were determined.
In addition, we considered breast cancer-related treatments, including aromatase inhibitors, tamoxifen, operations
Statistical analysis
We described and compared the distributions of age, comorbidity, and medication between the cohorts and breast cancer
patients (receiving or not receiving RT) by using v2 tests for
categorical variables and the Student’s t test for continuous variables. The Kaplan–Meier curves for the cumulative incidence of thyroid cancer in the three cohorts were compared
using the log-rank test. The incidence of thyroid cancer in the three cohorts (control group, breast cancer patients receiving no RT, and breast cancer patients receiving RT)
was calculated in the follow-up period. Univariate and multivariate Cox proportional hazards regression analysis was used
to estimate the risk of developing thyroid cancer associated with RT. Hazard ratios (HRs) and 95% confidence intervals (CIs) were estimated for the Cox models. Further analyses were performed to assess whether the association with thyroid cancer varied according to patient age at the diagnosis
of breast cancer and the length of the follow-up period. In addition, we analyzed the joint effects between the RT status and breast cancer-related treatments. All analyses were performed using SAS statistical software for Windows (Version
9.3; SAS Institute, Cary, NC), and the significance level was set at 0.05.
Results
Table 1 presents the baseline demographic factors, comorbidities and medications of the patients in the three cohorts.
Among the patients with breast cancer, 62.9% were aged 20– 54 years. Compared with the control cohort (mean age 52.4 years), the breast cancer patients without RT had a higher mean age (54.8 years) and the breast cancer patients with RT had a lower mean age (50.8 years). Patients with breast cancer
had a higher prevalence of COPD and diabetes and use of steroids, thiazide diuretics, and statins than did those in
the control cohort. Compared with breast cancer patients receiving no RT, breast cancer patients receiving RT tended to have a higher proportion of treatment that included aromatase inhibitors, tamoxifen, operations on the breast and
After a 12-year follow-up, the cumulative incidence of
thyroid cancer estimated according to the Kaplan-Meier analysis revealed significant differences among the three cohorts
(Figure 1, p<0.001). The cumulative incidence of thyroid cancer was much higher in patients with breast cancer who received or did not receive RT than in those in the control cohort. The overall incidence of thyroid cancer was higher in the breast cancer cohort than that in the control cohort (4.26 vs. 2.16 per 10,000 person-y), with an adjusted HR (aHR) of 1.98 (95% CI51.60–2.44) (Table 2). Among the patients with breast cancer, those receiving and those not receiving RT exhibited a significantly higher risk of thyroid cancer compared with those in the control cohort (aHR52.17, 95% CI51.66–2.85; aHR51.82, 95% CI51.39–2.38, respectively). The overall incidence of thyroid cancer was not significantly higher in the breast cancer patients receiving RT
compared with those receiving no RT (4.77 vs. 3.85 per 10,000 person-y; aHR51.28, 95% CI50.90–1.83).
Stratifying by age, we observed that younger (20–54 years) patients with breast cancer exhibited a significantly higher risk of thyroid cancer than those in the comparison control cohort did, regardless of whether they received RT (Table 3). When we focused on patients with breast cancer, in all age groups, no patients receiving RT exhibited a significantly increased risk of thyroid cancer compared with patients receiving no RT. Table 4 presents the comparisons of the risks of thyroid cancer according to stratification of the
follow-up duration. Within a 5-year follow-up period, compared with control participants, both groups of patients with
breast cancer had significantly higher risks of thyroid cancer. The breast cancer cohort of patients receiving RT followed for >5 years was associated with a marginally significantly higher risk of thyroid cancer (aHR51.79, 95% CI51.03–
3.11) compared with the control cohort. Furthermore, compared with the breast cancer patients receiving no RT and
with no aromatase inhibitor use, the breast cancer patients receiving RT and with aromatase inhibitor had a much lower
(Table 5). Among the patients with breast cancer, compared with those receiving no RT or operations on the breast, patients with operations on the breast and receiving no RT had a 0.47-fold lower risk of developing thyroid cancer (95% CI, 0.27–0.82), followed by patients with receiving both RT and operations on the breast (aHR50.54, 95% CI50.31–
0.95). The breast cancer patients receiving RT and no chemotherapy had a higher risk of thyroid cancer (aHR51.93, 95%
CI51.09–3.40) compared with patients receiving no RT and no chemotherapy. Discussion
This study involved using a comprehensive national database to investigate the risk of thyroid cancer among patients with an antecedent diagnosis of breast cancer. We found that Taiwanese women with breast cancer exhibited a significantly
higher risk of subsequent thyroid cancer, a risk that seems to be unrelated to RT. Stratified analysis by age revealed a significantly higher risk of developing thyroid cancer only in the
younger breast cancer patient group, and joint effects
between RT and different types of treatments revealed various patterns of thyroid cancer risk.
The breast and the thyroid are hormone responsive organs that are closely related to changes in endocrine function and glandular disease.16 The coincidence of thyroid disorder and
breast cancer has long been a subject of debate. Benign thyroid diseases are suggested to be related to breast cancer,17–19
and women treated for thyroid cancer have exhibited an increased incidence of subsequent breast cancer compared with women in the general population.20,21 Nevertheless, previous
studies investigating the possible association between
breast cancer and subsequent thyroid cancer have determined inconclusive results.6–12 Our study reveals that breast cancer
survivors, especially for younger patients (20–54 years) and with a follow-up duration shorter than 5 years more likely
have an increased risk of thyroid cancer, results that are consistent with those in a recent report.10 Kuo used the Surveillance,
Epidemiology, and End Results (SEER) nine database
to identify the number of patients with a diagnosis of breast and/or thyroid cancer in 1973–2011 and found that breast
cancer survivors had an increased risk of developing thyroid cancer, especially within 5 years of breast cancer diagnosis and at a younger age when diagnosed.10 Lal et al. also used
SEER data covering 1975–2008 to evaluate the risk of subsequent primary thyroid cancer after another malignancy. In
addition to noting significantly increased risks within 60 months after breast cancer diagnosis, they found that the increased risks persisted 60–119 months after breast cancer diagnosis (standardized incidence ratio 1.16).22 However,
Galper et al. used an earlier version of the SEER database to evaluate second non-breast malignancies after conservative surgery and RT treatment for early-stage breast cancer. They found only one case of thyroid cancer development among 1884 breast cancer survivors.11
In addition, radiation exposure is a proven risk factor for thyroid cancer,23–26 enhancing the concern of RT-induced
thyroid cancer in patients with breast cancer, because the thyroid gland is adjacent to the breast; the risk is increased especially when the lymphatics in the supraclavicular area are covered in the RT field. Our data revealed that patients in both breast cancer groups had significantly higher risks for thyroid cancer development, even though RT increased thyroid cancer risk among breast cancer patients nonsignificantly. Similar findings have been reported.27,28 For example,
Grantzau and Overgaard conducted a systematic review and meta-analysis on 762,468 patients to assess the risk of second non-breast cancer after receiving RT treatment for breast cancer. They found that there was no significant association between RT and second thyroid cancer.27 Furthermore, a
population-based study involved using the SEER database to derive data for 194,798 women with breast cancer and found that the risk of RT-associated thyroid carcinoma after initial RT treatment for breast carcinoma undetectably low.28 By
contrast, Lal et al. found that an increased second primary thyroid cancer risk after breast cancer was mostly observed in nonirradiated patients.22
The joint effects of RT and various types of treatments among patients with breast cancer showed various patterns.
Patients receiving both RT and aromatase inhibitor treatment had a significantly lower risk than did those receiving neither treatment. Patients with surgery and receiving or not receiving RT also had significantly reduced risks compared with
those receiving neither treatment. By contrast, patients receiving RT and no chemotherapy showed a significantly higher
risk than did those receiving neither treatment. The intricate mechanisms underlying the possible interactions between RT and other treatments affecting thyroid cancer development remain undetermined.
Plausible explanations for the observed link between breast cancer and subsequent thyroid cancer remain undetermined.
Although the link can be explained partly by a continued surveillance bias, genetic predispositions, environmental modifications,
general cancer susceptibility, and a possible therapy
effect,7,8 these factors alone are inadequate for explaining the
entire relationship.22 We concern the possible surveillance
bias when incidence of thyroid cancer is observed in this study. As both thyroid and breast cancers are often treated by endocrine surgeons, it is expected that breast cancer patients undergo more frequent and more detailed examinations of thyroid compared with controls. To verify this, we further performed analyses to determine whether surveillance bias exists. We found that the percentages of screening procedures of thyroid (ultrasound of head and neck and thyroid scan) among breast cancer group and control group were 1.05% and 3.18%, respectively, and control group had a significantly higher rate than the breast cancer group (p<0.001). In addition, diagnostic procedures for thyroid and parathyroid glands
(tissue confirmation) among breast cancer group and control group were 0.043 and 0.042%, respectively, and the difference is not significantly (p50.95). Thus, surveillance bias is
unlikely to explain the findings.
To the best of our knowledge, this study is the first nationwide population-based study conducted in Asia to focus on subsequent thyroid cancer risk among women with breast cancer. The strengths of our study are its populationbased design, generalizability of findings, and a low loss
follow-up rate in the longitudinal design, including study and control cohorts. In addition, NHIRD covers a highly representative sample of Taiwan’s general population because the
reimbursement policy is universal and operated by a single- buyer, the government in Taiwan. All insurance claims
should be scrutinized by medical reimbursement specialists and peer review. However, several intrinsic limitations caused by the database should be acknowledged. First, family history and genetic information were unavailable from the NHIRD. Several inherited conditions have been linked to different types of thyroid cancer,29,30 and mutations in the CHEK2
gene, which is involved in DNA damage repair, cause an increased predilection for multiple cancers, including those involving colon, breast, thyroid, prostate, and kidney tumors31; however, we could not adjust for these possible
confounders. Second, the NHIRD does not include information regarding patient lifestyle or behavior, so adjusting for
health behavior-related factors such as smoking and alcohol consumption was impossible. Certain unhealthy habits (e.g., smoking or consuming unhealthy food) have been determined to increase the risk of certain types of cancer.32 Third,
breast cancer or thyroid cancer histology information was unavailable in the NHIRD. Kuo mentioned that women who had breast cancer followed by thyroid cancer were more
likely to have had invasive ductal carcinoma, and breast cancer survivors who developed thyroid cancer were more likely to have an aggressive type of thyroid cancer10; however, we
could not verify this claim. Fourth, the data of RT dose for breast cancer patients were not available in the NHIRD, so we cannot evaluate the possible dose dependent effect of RT and subsequent thyroid cancer risk. Finally, the evidence derived from a retrospective cohort study is generally lower methodological quality than that from the prospectively
randomized trial because a retrospective cohort study is subject to have many biases due to lack of the necessary adjustments or possibly unmeasured or unknown confounding
factors. Despite the limitations of the administrative data, the data regarding breast cancer and thyroid cancer diagnoses
and breast cancer treatments in this study were highly reliable.
In conclusion, we found a significantly higher risk of thyroid cancer in Taiwanese women with breast cancer, particularly those who were younger, compared with women in the general population. Considering the relatively young median age at diagnosis of breast cancer in Taiwan,33 this risk may
become a topic of concern. Although the underlying mechanisms remain to be explored, RT seems to be uninvolved in
this relationship. Before confirmatory evidence is provided by further comprehensive studies, younger breast cancer survivors should receive a thyroid gland check up, especially
