Ten-year cardiovascular risk among cancer survivors: The National Health and Nutrition Examination Survey

Background Cancer survivors have a higher risk of developing and dying from cardiovascular disease (CVD) compared to the general population. We sought to determine whether 10-year risk of atherosclerotic CVD (ASCVD) is elevated among those with vs. without a cancer history in a nationally representative U.S. sample. Methods Participants aged 40–79 years with no CVD history were included from the 2007–2016 National Health and Nutrition Examination Survey. Cancer history was self-reported and 10-year risk of ASCVD was estimated using Pooled Cohort Equations. We used logistic regression to estimate associations between cancer history and odds of elevated (≥7.5%) vs. low (<7.5%) 10-year ASCVD risk. An interaction between age and cancer history was examined. Results A total of 15,095 participants were included (mean age = 55.2 years) with 12.3% (n = 1,604) reporting a cancer history. Individuals with vs. without a cancer history had increased odds of elevated 10-year ASCVD risk (OR = 3.42, 95% CI: 2.51–4.66). Specifically, those with bladder/kidney, prostate, colorectal, lung, melanoma, or testicular cancer had a 2.72–10.47 higher odds of elevated 10-year ASCVD risk. Additionally, age was an effect modifier: a cancer history was associated with 1.24 (95% CI: 1.19–4.21) times higher odds of elevated 10-year ASCVD risk among those aged 60–69, but not with other age groups. Conclusions Adults with a history of self-reported cancer had higher 10-year ASCVD risk. ASCVD risk assessment and clinical surveillance of cardiovascular health following a cancer diagnosis could potentially reduce disease burden and prolong survival, especially for patients with specific cancers and high ASCVD risk.

Introduction oversampling of underrepresented population subgroups. The National Center for Health Statistics Research Ethics Review Board approved and documented informed consent from all participants [12]. All data were completely anonymized and de-identified before access and analysis.
The current study included adults aged 40-79 years old without a history of CVD. The presence of CVD was self-reported and based on positive affirmation of any of the following conditions: "Has a doctor or other health professional ever told you that you had congestive heart failure/coronary heart disease/angina/heart attack (also called myocardial infarction)/stroke?" Participants were considered to have a positive history of cancer if they answered "yes" to the question "Have you ever been told by a doctor or other health professional that you had cancer or a malignancy of any kind?" Participants who refused or had missing data on cancer status were not included in this analysis. Type of cancer was self-reported and categorized as breast, bladder or kidney, prostate, colorectal, other gastrointestinal (GI, including esophagus, gallbladder, liver, pancreas, and stomach), cervical, ovary, uterus, lung, melanoma, hematologic, thyroid, testis, and other (including bone, brain, nervous system, soft tissue, and more than three cancers).

Data collection
Demographic characteristics including age, sex, race, marital status, education, and income to poverty ratio were self-reported using a standardized questionnaire. BMI was calculated by study technicians using standard measures of height (meters) and weight (kg), and we classified BMI <25.0 kg/m 2 , 25.0-29.9 kg/m 2 , and �30 kg/m 2 as underweight/normal weight, overweight, and obese, respectively. Daily activities and leisure time activities were measured based on the Global Physical Activity Questionnaire. Physical activity was quantified using the selfreported frequency of vigorous and moderate recreation activities (at least 10 minutes continuously) in a typical week. Participants who reported 0, 1-4, or �5 days per week of physical activity were classified as sedentary, physically inactive, and physically active, respectively [13,14]. A dietary interview was conducted to measure detailed dietary intake information for each participant. Dietary intake was assessed according to the Life's Simple 7 Healthy Diet metric [15,16]. Specifically, dietary intake in five components was evaluated: �4.5 cups fruit/ vegetable per day; � three 1-oz whole grain per day; <1,500 mg of sodium per day; � two 3.5-oz servings of fish per week; and <450 calories from sugared drinks per week. Participants with 0-1 components or 2-5 components were classified as poor diet or intermediate/ideal diet, respectively, due to few participants meeting ideal diet criteria. A depression score was calculated using the nine-item Patient Health Questionnaire (PHQ-9) to determine the frequency of depressive symptoms over the past two weeks [17].

Estimating 10-year ASCVD risk
We implemented the Pooled Cohort Equations (http://tools.acc.org/ASCVD-Risk-Estimator/) to estimate 10-year ASCVD risk [11]. The total score was calculated based on participants' age at completing NHANES survey, high-density lipoprotein cholesterol (HDL-C), total cholesterol (TC), systolic blood pressure, smoking status, and diabetes, stratified by gender and race [11]. HDL-C, TC, and systolic blood pressure were measured by study technicians during the physical examination [12]. Diabetes was determined based on self-reported medical conditions or medication use for diabetes. Smoking status was defined as current vs. not current smoker (including both former and never smokers). Participants with a PCE �7.5% were considered to have elevated 10-year ASCVD risk while PCE <7.5% was considered as low 10-year ASCVD risk, consistent with prior literature [11].

Statistical analysis
All analyses incorporated the NHANES sample weights and accounted for the complex sample survey design using standard methods [18]. Continuous variables were presented as weighted means ± standard error, and categorical variables were presented as weighted frequencies. We used ANOVA and chi-square tests to compare continuous and categorical variables by cancer status (no cancer history vs. positive cancer history), respectively. Unconditional logistic regression was used to estimates unadjusted and adjusted odds ratios (ORs) and 95% confidence intervals (CIs) for the association of cancer status and elevated 10-year ASCVD risk. Adjusted models included BMI, marital status, education level, and income poverty ratio, physical activity, dietary intake, and depression to control for potential confounding. Unconditional logistic regression was used to compare each site-specific cancer to no cancer history in relation to elevated (vs. low) 10-year ASCVD risk. For breast, cervical, ovarian, and uterine cancer, we estimated the OR comparing female participants with no history of cancer. Similarly, for prostate and testicular cancer, we estimated ORs comparing male participants with no history of cancer.
To examine whether the association of cancer status and elevated 10-year ASCVD risk differed by age group, a stratified analysis was conducted by age group (age 40-59 vs. age 60-79). The interaction between cancer status and age group was included in each model and tested using the adjusted Wald test. Since CVD and cancer share certain risk factors and we lacked information on the timing of the CVD diagnosis, additional sensitivity analyses were conducted to include participants with a history of CVD to reduce the potential for selection bias. Statistical significance for the interaction was evaluated as P<0.10 and for all other analyses was P<0.05 [19]. All statistical analyses were completed using Stata MP Version 16.1 (Stata-Corp, College Station, TX) in December 2020.

Results
We identified 15,095 adults aged 40 to 79 years with no CVD history and non-missing information regarding cancer status and risk factors used to calculate the Pooled Cohort Equations risk estimate. Weighted mean age at the NHANES examination was 55.2 years old, 53.0% were females, 71.7% were non-Hispanic White, 9.9% were non-Hispanic Black, and 18.4% selfreported other race (including Mexican American, Other Hispanic, Asian, Other or multiracial). About 70.3% of the population was married, 32.4% had a college or higher education, and 18.3% had <138% income to poverty ratio (Table 1).
In our study population, 24.8% of participants were classified as having elevated 10-year ASCVD risk (PCE �7.5%). Among those with a cancer history, 35.1% as compared with 23.4% of individuals with no cancer history were classified as having elevated 10-year ASCVD risk ( Table 2). Mean estimated 10-year ASCVD risk for individuals with vs. individuals without a cancer history was 8.3 ± 0.4% and 5.1 ± 0.1%, respectively. Comparing individual ASCVD risk factors by cancer status revealed older age, higher systolic blood pressure, and personal history of diabetes (all P<0.001) among those with vs. those without a cancer history. Table 3 shows associations between cancer status and odds of elevated 10-year ASCVD risk. In the unadjusted model, individuals with a cancer history had two-fold increased odds of elevated 10-year ASCVD risk (OR = 3.00, 95% CI: 2.39-3.77) compared to those with no cancer history. After controlling for BMI, marital status, education level, income to poverty ratio, dietary intake, physical activity, and depression score, individuals with a cancer history had 2.4 times increased odds of elevated 10-year ASCVD risk compared to those without a positive cancer history (OR = 3.42, 95% CI: 2.51-4.66). Participants who reported a diagnosis of bladder or kidney, prostate, colorectal, lung, melanoma, testicular, and other cancers, as well as those who reported don't know of their cancer type had increased odds of elevated 10-year ASCVD risk, compared to those with no cancer history (Fig 1). After adjusting for potential confounders, compared to those with no cancer history, participants with a history of testicular cancer had the highest odds of elevated 10-year ASCVD risk (OR = 11.47, 95% CI: 1.13-116.51), followed by prostate (OR = 9.45, 95% CI: We observed a significant interaction between age and cancer status in relation to odds of elevated 10-year ASCVD risk, and therefore present ORs stratified by age groups (Table 4). While the odds of elevated 10-year ASCVD risk did not differ among 40-49, 50-59, and 70-79 year-olds, compared to those without cancer diagnosis, those who had a positive cancer diagnosis had increased odds of elevated 10-year ASCVD risk (OR = 2.05, 95% CI = 1.47-2.85) among 60-69 year-olds (P interaction<0.001 , Fig 2.).
In the sensitivity analyses that included individuals with a history of CVD, 15,285 (86.6%) participants self-reported no cancer history while 2,006 (13.4%) self-reported a positive cancer history. Mean estimated PCE was 6.2 ± 0.1% overall and 12.9 ± 0.3% for individuals with a history of CVD (S1 Table in S1 File). Among individuals who had CVD, 27.3% were classified as

PLOS ONE
Ten-year cardiovascular risk among cancer survivors having elevated 10-year ASCVD risk. The association between cancer status and elevated 10-year ASCVD risk based on Pooled Cohort Equations did not change, and the interaction between age and cancer status in relation to odds of elevated 10-year ASCVD risk remained (S2 Table in S1 File). Compared to the observed association in terms of specific cancer types in the main analysis, we did not observe the higher odds of elevated 10-year ASCVD risk in lung cancer (OR = 2.59, 95% CI: 0.98-6.88), but observed lower odds of elevated 10-year ASCVD risk in thyroid cancer (OR = 0.11, 95% CI: 0.01-0.86) in the model including individuals with a CVD history (S3 Table in S1 File).

Discussion
In this population-based, cross-sectional study, we observed higher estimated 10-year ASCVD risk among individuals with a cancer history compared to those with no cancer history. Specifically, those with a cancer history had two-fold increased odds of elevated 10-year ASCVD risk compared to those with no cancer history. This association varied by cancer type, demonstrating important subgroups who could potentially benefit from CVD prevention interventions. Further, the relationship between cancer history and elevated 10-year ASCVD risk was modified by age, with a statistically significant increased odds of elevated 10-year ASCVD risk observed in the age group of 60-69 year-olds. Although not statistically significant, we observed 3.8 times increased odds of elevated 10-year ASCVD risk for those aged 40-49. Our results have implications for the growing number of U.S. cancer survivors, who are living longer and are at risk of dying from non-cancer-related causes.
Our findings are consistent with prior literature documenting an association of cancer history with increased CVD incidence and mortality, including adult survivors of childhood cancer [10, [20][21][22][23][24]. Increased CVD risks have been observed among breast, lung, prostate, and other cancer survivors from large cohort studies conducted in the United Kingdom, the Netherlands, and the U.S. [8,9,25]. Additionally, population-based studies provide some evidence that cancer survivors have a higher burden of subclinical CVD, such as elevated high-sensitivity cardiac troponin T, a marker of subclinical myocardial damage and increased CVD risk, compared to study participants without a cancer history [26].
Our study adds to the current body of literature by examining a composite measure of ASCVD risk-the Pooled Cohort Equations-which has been recommended by the American College of Cardiology/American Heart Association Task Force to predict 10-year risk for first ASCVD event. Few published reports have used the Pooled Cohort Equations to characterize 10-year ASCVD risk among cancer survivors, and no study has compared the 10-year ASCVD risk between those with vs. without a cancer history. In an NHANES study, comparing individuals with cancer history (n = 987) to those without a history of cancer (n = 10,184), elevated 10-year ASCVD was associated with 71% increased risk of cancer-specific mortality [27]. However, this study did not report the estimated Pooled Cohort Equations or the percentage of participants with elevated 10-year ASCVD risk according to cancer history. Several studies have utilized the Framingham risk score (FRS) to compare CVD risk between individuals with and without a cancer diagnosis and reported mixed findings [28]. For example, in a study of 1,222 Korean cancer survivors and 5,196 non-cancer controls, cancer survivors had a higher FRS than those without a cancer history, in line with our study. The average 10-year probability of CVD in relation to the cancer type was significantly higher in patients with hepatic, colon, lung, breast, and gastric cancer [29]. On the other hand, FRS was not significantly higher among those with breast [30], testicular [31], childhood cancer [32], or ovarian cancer [33] compared with controls. The current analysis, which demonstrated increased odds of elevated 10-year ASCVD risk among participants with a history of bladder or kidney, prostate, colorectal, lung, melanoma, testicular, and other cancer, suggests that ASCVD risk estimation may be more clinically relevant for individuals diagnosed with specific cancers.
Age was an important modifier of the association between cancer history and 10-year risk of future ASCVD. Increasing age is strongly associated with an increased risk of developing both cancer and CVD [7]. In line with this notion, compared to those with no cancer history, we observed that those with a self-reported cancer diagnosed between the ages of 60 and 69 had significantly increased odds of elevated 10-year ASCVD risk (OR = 2.24), whereas in other age groups, the odds of elevated 10-year ASCVD risk did not differ according to cancer history. However, we observed a greater relative magnitude of ASCVD risk in the younger age group (40-49 years), which could be explained by low ASCVD risk in the general population of younger adults. However, as adults age, their ASCVD risk increases, resulting in a smaller relative magnitude of difference between older adults with a cancer history vs. those without a cancer history. This aligns with differences in CVD mortality by age groups observed in previous studies. A recent analysis including 3.2 million cancer patients demonstrated CVD mortality in cancer survivors compared with the general population gradually decreased with increasing age at cancer diagnosis [34]. Further, this study observed higher heart disease mortality among younger cancer patients compared with similarly aged individuals in the general population. Similarly, Zaorsky et al. identified 7.5 million cancer patients from the nationally representative data from the Surveillance, Epidemiology, and End Results and found younger age of cancer diagnosis was associated with a higher standardized mortality ratio of stroke [35]. Our findings, which focus on estimated 10-year ASCVD risk as opposed to CVD mortality, demonstrate that the cancer diagnosis adds to the CVD burden, particularly among younger individuals. This could serve as a forewarning to those diagnosed with cancer in their 40's. Understanding the interaction between age of cancer survivors and ASCVD risk can assist in developing CVD prevention interventions for younger populations, particularly those who are at high risk for ASCVD.
Shared etiologic factors underlying CVD and cancer may contribute to the higher 10-year ASCVD risk we observed among those with a cancer history [36][37][38]. Indeed, older age and diabetes were more common among those with a cancer history compared to those without a history. We also adjusted for additional ASCVD risk factors not included in the Pooled Cohort Equations model, suggesting the cancer diagnosis adds to the 10-year ASCVD risk independent of CVD risk factors. As we were not able to incorporate temporality of the CVD diagnosis in relation to the cancer diagnosis in our analyses (due to lack of information on the timing of CVD diagnosis), we conducted sensitivity analyses that included individuals with a history of CVD to reduce potential selection bias. We observed similar associations of age, cancer status, and elevated 10-year ASCVD risk, which suggests our findings are unbiased.
The other major mechanism linking cancer with subsequent CVD is attributed to use of certain cancer treatments. Radiation directed at the chest may cause the development of coronary artery disease or blockages [39,40]. Chemotherapeutic agents, such as anthracyclinebased regimens for breast cancer [41], androgen deprivation therapy for prostate cancer [42], and immune-checkpoint inhibitors for melanoma, non-small cell lung cancer, and renal cell cancer [43], can induce cardiac toxicities including vascular compromise, cardiac structural problems, and cardiac dysfunction [44][45][46][47][48][49]. Practice guidelines call for the evaluation and monitoring of CVD risk factors among cancer survivors for early detection and management of long-term toxic effects from cancer treatment [42,[50][51][52][53]. Given the increased CVD burden among cancer survivors, new CVD risk calculators that incorporate factors salient to cancer survivors (e.g., use of chemotherapy/radiation, age at cancer diagnosis, etc.), should be devised to fully account for the increased burden in this population.
Limitations of this study include the cross-sectional study design, potential for recall bias, and lack of information of cancer treatment, clinical data (e.g. menopausal status), pertinent tumor characteristics, and timing of each comorbidity. Future studies would benefit from adding treatment type to broaden our understanding of the cardiotoxic profile of cancer treatment and allowing for risk-based surveillance and monitoring for CVD. Finally, we had relatively low numbers of individuals with each specific cancer type, limiting statistical power. Apart from the large, nationally representative sample, which strengthens the generalizability of our results, other strengths include adjustment for potential confounders (e.g. socioeconomic factors, dietary intake, and physical activity) and use of standardized measures of height and weight along with laboratory-based values of lipids and total cholesterol, which allowed us to derive accurate estimates of ASCVD risk.
Our findings suggest that a cancer history is positively associated with increased 10-year risk of ASCVD. As the number of cancer survivors continues to grow annually and considering that 65% of cancer survivors will be alive five years after their diagnosis [3], it is vitally important that we adapt our current screening tools to accommodate the excess CVD risk that a cancer diagnosis might contribute.
Supporting information S1 File. Sensitivity analyses that include individuals with a history of CVD. (DOCX)