Association between Cardiovascular Risk Factors and Long-Term All-Cause Mortality in Thyroid Cancer Survivors: An Israeli Population-Based Study

Elena Izkhakov (  elenaiz@tlvmc.gov.il ) Tel Aviv Sourasky Medical Center https://orcid.org/0000-0002-4742-878X Lital Keinan-Boker University of Haifa School of Public Health Micha Barchana University of Haifa School of Public Health Yacov Shacham Tel Aviv Sourasky Medical Center Iris Yaish Tel Aviv Sourasky Medical Center Narin N. Carmel Neiderman Tel Aviv Sourasky Medical Center Dan M. Fliss Tel Aviv Sourasky Medical Center Naftali Stern Tel Aviv Sourasky Medical Center Joseph Meyerovitch Clalit Health Services


Introduction
Thyroid cancer (TC) usually carries an excellent prognosis. Data from the United State Surveillance, Epidemiology, and End Results (SEER) program showed a case fatality rate from TC as being around 0.5 deaths per 100,000 TC survivors. 1 However, a recent analysis of SEER data showed increasing rates of mortality and incidence of TC in the United States from 1974 to 2013, particularly for advanced-stage papillary TC. 2 This is in contrast with an analysis of worldwide data through 2012, which showed a declining mortality from TC in parallel with an increasing incidence. 3 In Israel as well, nationwide trends show an increased incidence of TC and a modest increase in the 5-year relative survival during the last 3 decades. 4 A number of high-risk and pathological conditions have been reported among TC survivors compared to healthy controls, such as an increased prevalence of obesity and diabetes, 5 aortic stiffness, 6 and isolated left ventricular diastolic dysfunction. 7 Data from the SEER program showed heart disease to be the cause of death for 34% of the non-cancer mortality among TC survivors. 8 Data from our recently published Israeli population-based study showed a 26% increase in cardiovascular and cerebrovascular morbidity among thyroid cancer survivors compared to matched controls, who had neither thyroid disease nor any type of cancer. 9 The risks of all-cause and cardiovascular mortality of differentiated TC (DTC) survivors in The Netherlands, independent of cardiovascular risk factors, were found to be increased by 4.4-fold and by 3.3-fold, respectively, compared to sex-and age-matched controls from the general population in the same region during a median follow-up of 8.5 years. 10 The increased morbidity [5][6][7] and the increasing incidence of the disease worldwide 3,11,12 lends special importance to understanding the elements that comprise the mortality risk of TC survivors. The aim of the current study, therefore, was to investigate the long-term all-cause mortality of a large cohort of Israeli TC survivors compared to individuals without either thyroid disease or a diagnosis of cancer who were members of the same healthcare system, while controlling for known cardiovascular risk factors at the beginning of the study and those diagnosed during a 15-year follow-up.

Study Design
This large historical cohort study (as previously described elsewhere) 9 is based on the computerized data of the Clalit Health Services (CHS), the largest healthcare fund in Israel and a provider of healthcare to more than 4.3 million Israeli residents (52.3% of the total Israeli population). The CHS database comprises demographic and medical data, including diagnoses, laboratory tests, drug prescriptions and purchases, and date (but not cause) of death.

Study Population and Follow-up
The TC survivors (TC group) included individuals diagnosed with TC between January 1, 2001 and December 31, 2014, who underwent thyroidectomy and received radioactive iodine treatment and levothyroxine therapy. The non-TC group was comprised of CHS members who had neither thyroid disease nor any type of cancer during the same time period. The non-TC group was matched to the TC group by sex and age (± 2 years) at a ratio of 4:1. The follow-up of the TC survivors started on the date of TC diagnosis, and the follow-up of the matched non-TC individuals started in the same year of diagnosis as the TC survivors to whom they were matched. The follow-up of both groups ended on June 30, 2016 or on the date of death, whichever occurred earlier.

Methods
The Medical Ethics Committee of the CHS provided approval to conduct this study. Signed informed consent from the participants was waived since the study was based on existing databases. The study inclusion criteria were membership in the CHS throughout the entire study period (January 1, 2001 to June 30, 2016) and age ≥ 18 years. The exclusion criteria were any other primary cancer prior to study entry, with the exception of squamous or basal cell carcinoma of the skin, and advanced renal failure (creatinine > 1.5 mg/dL), the latter because of its impact on the clinical decisions regarding the application of radioactive iodine treatment.

Study Variables
The following data were collected for the TC and non-TC groups: demographic characteristics, smoking status, anthropometric characteristics, mean blood pressure, pulse rate, and laboratory tests that included levels of glucose and creatinine, lipid pro le, and thyroid function tests at a date closest to study entry, and any diagnosed pathological conditions at study entry and at the end of the follow-up.
Diabetes mellitus was de ned as having at least one of the following upon study entry: a diagnosis of diabetes mellitus recorded in the CHS database, 2 plasma glucose measurements > 125 mg/dL, a random plasma glucose measurement > 199 mg/dL, or a record of hypoglycemic medications. Hypertension was de ned as having at least one of the following upon study entry: a diagnosis of hypertension recorded in the CHS registry, 3 or more measurements of systolic blood pressure > 140 mm Hg or diastolic blood pressure > 90 mm Hg, or a record of medications for hypertension. Dyslipidemia was de ned as having at least one of the following upon study entry: a diagnosis of hyperlipidemia recorded in the CHS registry, at least 2 plasma low density lipoprotein (LDL) cholesterol measurements > 160 mg/dL, a triglyceride level > 150 mg/dL, or a high-density lipoprotein cholesterol level < 40 mg/dL for males or < 50 mg/dL for females, or a record of hypolipidemic medications. Cardiovascular disease was de ned by the International Statistical Classi cation of Diseases and Related Health Problems (ICD) code as having at least one of the following diagnoses recorded in the CHS database upon study entry: ischemic heart disease, acute myocardial infarction, percutaneous transluminal coronary angioplasty, or coronary artery bypass graft. Cerebrovascular disease was de ned by ICD code as having at least one of the following diagnoses recorded in the CHS database code upon study entry: transient ischemic attack, cerebrovascular accident, carotid artery stenosis/occlusion, or carotid endarterectomy. The prevalence of diabetes mellitus, hypertension, dyslipidemia, and cardiovascular and cerebrovascular diseases at the end of follow-up was assessed according to the same de nitions as those for the beginning of follow-up.

Statistical Analyses
Baseline patient characteristics are presented as means and standard deviations for continuous variables, and as frequencies and percentages for categorical variables. Independent t-tests and Chisquare tests were used to compare between the study groups for continuous and categorical variables, respectively. Unadjusted and adjusted Cox proportional hazard models were performed to evaluate the hazard ratios (HRs) and 95% con dence intervals (95%CIs) for death. Kaplan-Meier-based adjusted survival curves for mortality are provided. All-cause mortality risk was compared between the TC and non-TC groups by gender and age (≤ 44, 45-54, 55-64, 65-74, and ≥ 75 years) at study entry. Unadjusted and adjusted Cox proportional hazard models were performed to evaluate the HRs and 95%CIs for longterm mortality by a minimum latency period of 2 years. Data were analyzed with SPSS software version 23.0. (SPSS Inc. Headquarters, 233 S. Wacker Drive, 11th oor Chicago, Illinois 60606, USA). A 2-sided P value less than 0.05 was considered statistically signi cant.

Results
Baseline Characteristics of the TC and Non-TC Groups The TC group was comprised of 5,677 TC survivors (mean age 50 ± 16 years) of whom 21.4% were males. The non-TC group was comprised of 23,962 members of CHS (mean age 47 ± 15 years) of whom 20.5% were males. Although the study sample was matched for age and sex, exclusion of individuals with renal failure slightly changed the distributions (Table 1). 9 At baseline, the anthropometric characteristics, mean blood pressure, and pulse rate were similar between the study and control groups (Table 1). 9 A higher proportion of individuals in the TC group had hypertension compared to the non-TC group (24.7% vs. 19.3%, respectively, P < 0.001) and dyslipidemia (32.9% vs. 28.5%, P < 0.001), and a lower proportion had cardiovascular and cerebrovascular diseases (1.4% vs. 4.9% and 0.5% vs. 1.7%, P < 0.001 for both) ( Table 2). 9 Table 1 Baseline characteristics of the thyroid cancer survivors (n = 5,677) and the non-thyroid cancer individuals (n = 23,962).   Bold indicates signi cant.
The mortality events that occurred during the rst 2 years of follow-up were excluded in order to allow a minimal latency period and to assess long-term mortality. The subsequent HR for mortality in the TC group adjusted for prevalent cardiovascular risk factors (age, sex, hypertension, diabetes mellitus, dyslipidemia, and current smoking) and for prevalent atherosclerotic cardiovascular and cerebrovascular disease at follow-up onset remained signi cantly increased (HR 1.59, 95%CI: 1.40-1.80).

Characteristics of the Individuals Who Died
Compared to the individuals in the non-TC group who died, the TC survivors who died had a higher prevalence of hypertension (14.6% vs. 10.3%, respectively, P = 0.002), dyslipidemia (11.4% vs. 7.2%, P < 0.001), and cardiovascular disease (33.6% vs. 22.3%, P < 0.001) at the end of the 15-year follow-up period. During the study period, second primary tumors among the TC survivors who died were not more common than primary tumors among the individuals in the non-TC group who died. Strati cation of the TC patient group by selected cardiovascular risk factors (hypertension, dyslipidemia, diabetes mellitus, and cardiovascular and cerebrovascular disease) at the end of the follow-up revealed a direct association between the number of those risk factors and mortality risk (Table 4).

Discussion
Overall all-cause mortality risk This large Israeli population-based historical cohort study demonstrated a higher all-cause mortality rate among TC survivors than among individuals without any thyroid disease or any type of cancer from the same population, matched by sex and age. The difference remained statistically signi cant after adjusting for age, sex, baseline cardiovascular risk factors, and cardiovascular and cerebrovascular diseases at study onset. The novel ndings of this study are the association between cardiovascular risk factors at the end of follow-up and the higher mortality among Israeli TC survivors compared to non-TC individuals. In contrast, there was no association between mortality and the presence of a second primary malignancy in the former group and the presence of a primary malignancy in the latter group at the end of follow-up.
The majority of the TC survivors in the current cohort were female (79%). Strati cation by sex showed that the mortality HR for male TC survivors was higher than that of females. Similarly, analysis of the SEER data revealed a higher mortality rate among male TC survivors than among females. 2 Moreover, during the last three decades, age-adjusted mortality attributed to TC was reported to remain stable among Israeli men and to decrease among Israeli women. 4 Surprisingly, female gender and a diagnosis of dyslipidemia at the beginning of the follow-up for the entire cohort were associated with a lower risk for mortality. This nding may be explained at least in part by statin treatment.

Mortality Risk by Age
There was a signi cant and pronounced age-related increased mortality risk for TC survivors compared to the non-TC group, particularly from age 55 years onwards. Our ndings concur with a recent study that found 55 years to be a valid cutoff age for risk assessment in TC survivors. 13 Another recent publication concluded that the increasing age risk should be considered along a continuum. 14 The SEER study also reported that the 5-year and 10-year probability of death increased with age among TC survivors. 8 The lower relative mortality risk among younger TC survivors is supported by the results of a German study that disclosed no reduced life expectancy among DTC survivors who were under 45 years of age at DTC diagnosis and with tumor-node-metastasis (TNM) stages I, II, or III compared to the general population. 15

Factors Contributing to Increased Mortality among TC Survivors
The interpretation of our results should take into account the various factors that may contribute to increased mortality in TC survivors. First, despite the excellent prognosis and the low and decreasing fatality rates, the small proportion of TC survivors who die of their disease naturally increases the general mortality rate in TC survivors, particularly in the long term during which recurrence is a factor. In a Dutch study, 10 progression or recurrence of TC was the cause of mortality for 39% of TC survivors who died during a mean follow-up of 8.5 years. An increased risk of second primary cancers, compared to the general population, may also increase overall mortality. Such risk has been documented in Israel 16 and elsewhere. 17 Although a second primary cancer also worsens the prognosis of TC survivors, 18 a second primary cancer was not more common among the TC survivors who died than a de novo primary cancer among the individuals who died in the non-TC group in the current study. However, analyses of the current ndings revealed a direct association between the number of cardiovascular risk factors and mortality risk among the TC patients.
Increased mortality due to non-cancer causes must also be considered in relation to TC. SEER data showed that the risk of dying from a cause other than the primary disease in TC survivors is nearly 2-fold than that of dying from TC. 8 In that study, the 10-year probabilities of death from TC, from other cancers, and from non-cancer causes were 3.0%, 2.0% and 3.9%, respectively. In the current study, increased mortality persisted after controlling for baseline cardiovascular risk factors and morbidity, in addition to age and sex. At the time of TC diagnosis, the anthropometric and clinical characteristics of the TC survivors were very similar to those of the non-cancer individuals, with the exception of hypertension, which was slightly more prevalent among the TC survivors. Similarly, a Dutch population-based study reported a higher than expected prevalence of hypertension among TC survivors. 19 In the current study, the prevalence of diabetes mellitus at baseline was similar between the TC group and the non-TC group.
This concurs with a recently published Israeli study that was based on a nationwide cohort and which found no association between diabetes mellitus and TC. 20 Nevertheless, diabetes mellitus may still affect TC prognosis. For example, TC survivors with type 2 diabetes mellitus and DTC were found to be more likely to have an advanced TNM stage at the time of diagnosis as well as increased disease-speci c mortality. 21 The prevalence of hypertension, dyslipidemia, and cardiovascular disease at the end of the follow-up were higher for the individuals in the TC group who died during the follow-up period of the current study than among those in the non-TC group who died. These differences between the groups at study closure were greater than those recorded at baseline.
The treatment of TC may affect the mortality risk in a number of ways. The majority of the Israeli TC survivors were treated according to a standard of care consisting of thyroidectomy, radioactive ablation, and thyroid hormone suppression treatment. Several studies have shown a greater risk of second primary cancer among TC survivors treated with radioactive ablation than among those who did not undergo such treatment. 17,18 This increased risk was shown to also prevail among low-risk TC survivors 22 and only when the cumulative radioactive iodine dose was ≥ 37.0 GBq. 23 The mortality risk was also higher for the 5-15% of TC survivors who become refractory to radioactive iodine therapy. 24,25 Thyroid hormone suppression is a component of TC treatment, both for those who respond and for those who are refractory to radioactive iodine therapy. Associations have been reported among thyroid hormone suppression treatment and atrial brillation, 26 impaired small and large artery elasticity, 27 increased left ventricular mass, 27 abnormalities of heart morphology related to impaired exercise performance, 28 a prothrombotic condition, 29 and myocardial strain. 30 In the absence of data on thyroid hormone treatment, we do not know the degree to which subclinical hyperthyroidism was achieved in the TC survivors, or the proportion of individuals in the TC group that may have had endogenous subclinical hyperthyroidism.
However, subclinical hyperthyroidism not in the setting of TC has been shown to be associated with atrial brillation 31,32 worse physical capacity, 33 increased risk of heart failure, and cardiovascular morbidity. 32,34 Moreover, endogenous subclinical hyperthyroidism was associated with increased risks of total and cardiovascular mortality in a pooled analysis from 10 cohorts of individuals not treated with thyroxine. 31 A Danish population-based study found increased heart failure and increased cardiovascular and all-cause mortality among patients with subclinical hyperthyroidism compared to euthyroid individuals. 35 In contrast with these reports, a recently published small study with up to a 9-year follow-up showed no impairment in cardiac function and structure among individuals who received thyroid hormone suppression treatment. 36 However, the patients' thyroid-stimulating hormone (TSH) levels in that study were < 0.1 mU/L in the intermediate-risk-of-recurrence group and < 0.3 mU/L in the lower-riskof-recurrence group, which is higher than the recommendations for TSH suppression. Having no data on TSH levels during the follow-up in our cohort, we were unable to compile the effects of treatment on mortality.
Health-related quality of life (HRQoL) is an additional factor that may affect health and long-term mortality in cancer survivors. A recent publication reported that 14-17 years after diagnosis, almost half of DTC survivors who lled in a HRQoL questionnaire expressed anxiety about disease recurrence, and that this negatively impacted their HRQoL. 37 Poor HRQoL has been found to be related to all-cause mortality in various populations. 38,39 Stage at diagnosis 2,15 and genetic variance 40 have been shown to affect mortality. The current study did not distinguish between types of TC. DTC, which comprised the vast majority of TCs in our population, confers considerably better prognosis than do the less common types of TC. 1 The strengths of this study are the long-term follow-up ndings of a large population-based cohort, which are based on the computerized data le of the largest healthcare fund in Israel (covering around 52.3% of the total Israeli population). They take into account baseline pathological conditions and those diagnosed during the follow-up period of the study (including dyslipidemia, hypertension, diabetes mellitus, cardiovascular and cerebrovascular diseases, and new malignancies). A signi cant limitation of this study is the lack of data on the cause of death, as well as on other variables, such as the histological variant of the TC, the stage of the TC, the doses of radioactive iodine treatment, and the TSH levels during the follow-up period.
In summary, the nding of a higher all-cause mortality risk in TC survivors compared to the general population, despite the excellent prognosis and the decreasing fatality rates for TC, is of concern. In the current study, a higher prevalence of hypertension, dyslipidemia, and cardiovascular disease at the end of the follow-up period was associated with mortality among the group of Israeli TC survivors. Moreover, we found a direct association between the number of cardiovascular risk factors at the end of the follow-up and the mortality risk among TC survivors. As such, our ndings suggest that cardiovascular risk factors may predict higher mortality in TC survivors compared to non-TC individuals. We therefore recommend a

Declarations
Ethics approval and consent to participate: The Helsinki Medical Ethics Committee of the Clalit Health Services provided approval to conduct this study (approval # 0040-15-COM) and waived signed informed consent since the study was based on existing databases.

Consent for publication:
Not applicable.
Availability of data and materials: The data that support the ndings of this study are available from database of Clalit Health Services, but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of Clalit Health Services. Figure 1