The increased risk of cardiovascular disease (CVD) in breast cancer survivors compared to women without a cancer history is related to the cardiotoxic effects of breast cancer treatment and overlapping risk factors of breast cancer and CVD (1–3). After breast cancer diagnosis, the majority gain weight, which continues, especially in premenopausal women, into breast cancer survivorship (4, 5). Even in the absence of weight gain, changes in body composition consisting of gain in adipose tissue without a gain in or with loss of lean tissue have been observed (6). Chemotherapy also causes alternations in skeletal muscle and creates a predisposition to muscle atrophy and weakness (7). As a result, CVD burden associated with post-diagnosis weight gain has been observed to as long as 5 years in Asian patients (8). Of equal importance, excessive body fat in cancer survivors has been shown to affect quality of life and disease-free survival (9).

Prior research among breast cancer survivors have focused on mortality and body mass index (BMI) or waist circumference (10, 11). In terms of the association between change in BMI and the risk of CVD or CVD mortality, there were contradicting results (12). For example, there was no association between a change in BMI and the risk of CVD in short-term breast cancer survivors. The significance of examining body composition over BMI was further shown in a relatively recent study on adiposity and CVD (13). In this study, the increased risk of CVD was also observed in normal-weight breast cancer survivors with greater visceral adiposity. Research on changes in body composition, distinguishing fat mass and muscle mass, and the risk of CVD in young adults has been carried out, however, little is known on changes in body composition and the risk of CVD in breast cancer survivors (14).

Various anthropometric and imaging indices of obesity and its relationship with CVD risk have been summarized (15). In the current study of 40,095 breast cancer survivors without prior CVD, we utilized validated prediction equations with anthropometric data and health habits to examine the association between changes in body composition and the risk of subsequent CVD.

Table 1 shows the characteristics of the study population, based on change in percentage of predicted lean body mass. Mean ages(standard deviation) for Low to Low, Low to High, High to Low, and High to High groups were 59.99 (9.18), 58.86 (8.71), 56.38 (8.56), and 55.25 (8.10), respectively. The majority of subjects were never-smokers with rare alcohol consumption and physical activity. In addition, the majority of subjects received chemotherapy, radiation therapy, or hormone therapy.

Cardiovascular risks of the study population were observed, based on change (prior to and 5 years after diagnosis of breast cancer) in pLBMP, pASMP, and pBFMP (Table 2). Compared to those who continued to have low pLBMP and pASMP, those with persistently high pLBMP and pASMP had lower risks of CVD (aHR, 0.68 [95% CI, 0.53–0.87] and aHR, 0.60 [95% CI, 0.44–0.81], respectively). In contrast, both the High to Low and High to High pBFMP groups had higher risks of CVD (aHR, 1.44 [95% CI, 0.98–2.10] and aHR, 1.48 [95% CI, 1.15–1.89], respectively), compared to those who maintained a low pBFMP. Notably, those with increased (a low to high change) pBFMP had higher risk of CVD (aHR 1.51, [95% CI 0.99–2.31]. This pattern was also evident in Models 2, 3, and when High and Low groups were classified based on mean pLBMP, pASMP, and pBFMP of first and second health checkups (Suplementary Table S1).

Results from the stratified analysis on the association of change in predicted body composition with CVD according to subgroups of age, CCI, and treatment pattern are shown in Table 3. In comparison to the Low to Low group, subjects in the High to High group, aged 60 or above, with less comorbidities, and who had received hormone therapy, had significantly lower (for pLBMP and pASMP) or higher risk (for pBFMP) of CVD. These risks were not affected by subgroups of age, CCI, or treatment pattern (p for interaction >0.05).

Figure 2 depicts the increased or decreased levels (changes) in metabolic factors, namely, that of blood pressure, fasting serum glucose, and total cholesterol, in relation to predicted body composition. Pertaining to changes in pLBMP and pASMP, the Low to High group showed decreased sBP, dBP, total cholesterol, and FSG, compared to the Low to Low group. Pertaining to change in pBFMP, the Low to High group showed increased sBP, dBP, total cholesterol, and FSG (aMean, 2.20 [95% CI, 1.15∼3.25]; aMean, 1.10 [95% CI, 0.37–1.82]; aMean, 1.71 [95% CI, −1.12∼4.53]; and aMean, 4.39 [95% CI, 3.05–5.73]. This pattern was also evident when comparing within those who initially had high pLBMP, pASMP, or pBFMP. Those who maintained high pLBMP and pASMP had decreased metabolic markers compared to those who changed from high to low pLBMP and pASMP; those who maintained high pBFMP had increased metabolic markers compared to those who changed from high to low pBFMP. In the case of change in total cholesterol, total cholesterol decreased less in the High to High group of pLBMP and pASMP (aMean, −0.64 [95% CI, −3.12∼1.84] and aMean, −0.59 [95% CI, −3.10∼1.91], respectively), compared to that of the Low to Low group while total cholesterol decreased more in the High to High group of pBFMP (aMean, −4.72 [95% CI, −7.20∼−2.23), compared to that of the Low to Low group.

In summary, having a persistently high predicted LBMP or ASMP among breast cancer survivors was associated with lower CVD risk than a persistently low predicted LBMP or ASMP, respectively. Conversely, persistently high predicted BFMP among breast cancer survivors was associated with higher CVD risk than persistently low predicted BFMP. Additionally, those with a low to high change in pBFMP had a higher risk of CVD than those with persistently low pBFMP. Changes in body composition were accompanied by changes in metabolic markers. To the best of the authors' knowledge, this is the first study to demonstrate the CVD risk with changes in body composition, by distinguishing muscle mass and fat mass as predicted LBMP, ASMP, and BFMP in breast cancer survivors.

Our findings are consistent with a previous study that examined adiposity from CT scans taken near diagnosis and subsequent CVD risk in breast cancer survivors (13). In this previous study, visceral and intramuscular adiposity were associated with increased CVD incidence, though only adiposity was examined and at one-time point. Similarly, central obesity was associated with higher all-cause and breast cancer–specific mortality among Black breast cancer survivors (31); CVD nor CVD-related mortality was observed in this study. In a cross-sectional study of anthracycline chemotherapy for breast cancer patients, greater thigh muscle fatty infiltration was associated with impaired oxygen extraction, which is a predictor of CVD morbidity and mortality (32). According to a study that observed for change in BMI and waist circumference between diagnosis and 24 months post-diagnosis in early-stage breast cancer patients, weight change was not associated with risk of CVD, while any elevation in waist circumference was associated with increased risk of CVD (12). As noted by the author, BMI and weight change do not accurately represent body composition, underscoring the significance of examining body composition and changes in body composition.

Although a low to high change in pBFMP showed a higher risk of CVD than those with persistently low pBFMP, so did that of a high to low change in pBFMP (both were not significant). Having a high body fat mass percentage before initial breast cancer diagnosis or at any timepoint in cancer survivorship may be decisive. Therefore, current results should be interpreted with discretion, and further examination should be carried on CVD risk according to changes in body fat mass percentage respective to muscle mass percentage as well as changes in metabolic markers.

The associations of changes in predicted LBMP, ASMP, and BFMP with CVD risk were statistically significant in the older or healthier individuals, which is consistent with findings from a previous study on young adults (14). Furthermore, these changes were significant in breast cancer survivors who had a history of hormone therapy. Although cardiovascular effects vary based on the type and combination of hormone therapy, hormone therapy is largely associated with an increased risk of stroke, coronary heart disease, and notably, venous thromboembolism (33, 34). Therefore, those who received hormone therapy, may benefit, in terms of CVD prevention, from changes in LBMP, ASMP, and BFMP.

The metabolic changes associated with changes in body composition have been studied before (25). Similar findings in breast cancer survivors were observed in our study. Increased or persistently high (compared to a high to low change in) predicted LBMP and ASMP were associated with decreased blood pressure, fasting serum glucose, and total cholesterol. Conversely, increased or persistently high (compared to a high to low change in) predicted BFMP was associated with increased blood pressure, fasting serum glucose, and total cholesterol. Indeed, adipose tissue inflammation includes insulin resistance, alterations in lipid metabolism, and blood pressure regulation, favoring endothelial dysfunction and atherogenesis (15). Skeletal muscle tissue, another endocrine organ that produces myokines, is also involved in the complex network related to metabolic functions (35). Myokines such as irisin and fibroblast growth factor-21 (FGF-21) are induced by physical exercise and increase insulin sensitivity and in the case of FGF-21, acts on lipolysis; apelin not only has an anti-inflammatory role but also controls cardiac muscles and blood pressure. However, further research is needed to investigate the joint effects of adiposity, muscle mass, and the crosstalk between respective cytokines (36).

Our study is not without limitations. Prediction equations were used to measure body composition, which may be imperfect for patients with cancer and difficult to apply to other ethnicities. However, a previous large validation study in the same ethnic group was conducted and showed high predictive values. In addition, in previous studies using the same equations, similar changes in body composition (increase in body fat mass and decrease in appendicular skeletal mass or lean body mass) were associated with an increased risk of metabolic syndrome and CVD in young adults (14, 25). Utilizing prediction questions may not only be less costly but also overcome the limitations of applying independent anthropometric indices (cut-offs). Furthermore, the study was not inclusive of cardiovascular biomarkers associated with adiposity (37). Therefore, future exploration on delineating associated immunometabolic pathways may provide insight on the progression from changes in body composition to CVD in breast cancer survivors. A major strength of our study is the large study population, considering that it was limited to 5-year breast cancer survivors. Changes in predicted body composition were observed as percentages, which may be clinically more meaningful, and were inclusive of both adiposity and muscle mass. Though it is unknown whether positive changes in body composition may reverse CVD risk, body composition is not unmodifiable, with lifestyle modification including physical activity and adequate nutrition; adhering to a healthy lifestyle showed to be particularly beneficial for those with a high genetic risk for CVD (38). In conclusion, among breast cancer survivors, persistently high muscle mass, represented as predicted LBMP or ASMP, was associated with lower CVD risk. Preventing an increase in fat mass may be beneficial in preventing CVD in breast cancer survivors, as a part of cancer survivorship.