Obesity is a well-established risk factor for cardiovascular disease (CVD) incidence of stress-related clinical manifestations (1). Most cardiovascular prevention guidelines recommend weight stability maintenance and weight loss in those subjects with obesity, but many patients fail to do so, resulting in weight cycling (2, 3). Stress and anxiety exposure can also have a significant long-term impact on the alignment to gain and maintain weight due to inadequate compensatory mechanisms or maladaptive processes, which may lead to rapid and frequent weight “yo-yo” fluctuations (4–6). Actually, studies on overweight and obese individuals reported that only 20% of participants were able to maintain long-term weight loss (7).

Body-weight fluctuation, which is also named body-weight variability or weight cycling, refers to repeated weight loss and successive regain (8). Several years ago, the first evidence that body-weight variability could be related to a high risk of CVD arose from the Framingham cohort (9). Later, several studies evaluated the association of body-weight variability with CVD outcomes and deaths among the general population and patients with type 2 diabetes (T2D) or previous CVD (10–17). However, while some epidemiological studies have shown that body-weight fluctuations may be associated with metabolic disorders and complications, resulting in negative health consequences (9, 18, 19), other cohort studies have failed to confirm these findings or even get opposite results concerning mild-weight fluctuations (13, 14, 20–22). These investigations have typically focused on the variability of weight, an imprecise measure of adiposity, and did not assess the variability of markers of overall obesity such as body mass index (BMI), which captures both fat and lean mass composition (11), and neither assessed the time-rated velocity of these fluctuations and variability (11, 23). Moreover, there is not a single definition or measurement concerning weight cycling matters (8), and to our knowledge, the involvement of intensive BMI fluctuations in affecting the association with cardiovascular stress-related morbidities has not been previously examined. Therefore, the aim of the present study concerning weight cycling was to evaluate the association between BMI fluctuation and BMI fluctuation rate with incident CVD in a Caucasian European cohort.

Data from 1,725 women and 2,587 men with a mean ( ± SD) age at baseline of 53.7 ± 13.3 were screened. After 40,323.32 person-years of follow-up, there were 436 incident cases of CVD. Age, BMI, and the prevalence of hypertension were more likely to increase across quartiles of BMI fluctuation and BMI fluctuation rate ( Tables 1A , 1B ). Furthermore, this trend was found associated with higher frequencies of sedentary behavior. The values for TG and TyG index increased, and the level of HDL-C decreased in proportion to both BMI fluctuation and BMI fluctuation rate quartiles ( Tables 1A , 1B ). The proportion of women and antiaggregation therapy increases across the quartiles of BMI fluctuation ( Table 1A ).

We found a slightly significant (p < 0.05) increased risk of CVD among the quartiles of BMI fluctuation, with an HR of 1.34 (95% CI: 1.00-1.81) for those participants in the fourth quartile of BMI fluctuation ( Table 2 ). However, in the multivariate-adjusted model, a higher and significant and progressive increase in the risk of CVD with BMI fluctuation rate was found in the third and fourth quartiles as compared with the bottom quartile: HR of 1.82 (95% CI: 1.33-2.50) and HR of 3.71 (95% CI: 2.71-5.07), respectively (p for trend <0.001) ( Table 2 ).

The stratified analysis by baseline BMI, sex, and age groups showed a heightening risk of CVD in all age subgroups, with an HR of 3.7 (95% CI: 1.4-9.2), 4.3 (95% CI: 2.6-7.3), and 2.3 (95% CI: 1.3-3.9) for the highest quartile of BMI fluctuation, of the age groups of <50, 50-65, and >65 years old, respectively ( Figure 2 ). We also observed a higher risk in men among those of the third and fourth quartiles, with an HR of 2.1 (95% CI: 1.4-3.2) and 3.8 (95% CI: 2.5-5.6), respectively, and a higher risk in overweight and obese participants of the fourth quartile, with an HR of 2.0 (95% CI:1.3-2.9) and 1.8 (95% CI:1.0-3.3) in that order ( Figure 2 ). Remarkably, BMI fluctuation was not associated with CVD in women and normal-weight participants ( Figure 2 ), but this was just a trend.

The potential involvement of age, sex, and baseline BMI on the relation of quartiles of BMI fluctuation rate to incident CVD was also addressed ( Figure 2 ). Thus, the incidence of CVD increased in proportion to the quartiles of BMI fluctuation rate in all the groups of the stratified analysis. Interestingly, the HRs for developing a CVD were 5.6 (95% CI: 3.0-10.6) and 6.5 (95% CI: 4.3-9.7) for women and men in the fourth quartile, respectively. Noteworthy, the risk of a CVD was larger in overweight/normal-weight than obese participants. Across the age subgroups, the greater risk of CVD was found in the youngest participants, subgroups age <50 and 50-65 years old, reaching an HR of 6.4 (95% CI: 2.9-13.9) and 4.0 (95% CI: 2.6-6.3) for those participants of the fourth quartile of BMI fluctuation rate, respectively ( Figure 2 ).

The magnitude of the associations found across the quartile of BMI fluctuation rate and the outcome of a CVD was larger than that displayed by using BMI fluctuation ( Figure 2 ). The differences between the mean of BMI fluctuation and BMI fluctuation rate according to baseline BMI, sex, and age subgroups were statistically significant ( Figure 3 ).

The main results of the present study were consistent with a time-limited scenario of 5 years that we included in the sensitivity analysis ( Supplemental Table 1 ). An association between the risk of CVD among the third and fourth quartiles of BMI fluctuation and BMI fluctuation rate was found, with a robustness OR when using the BMI fluctuation rate (OR of 1.59 and 1.95 vs. OR of 2.19 and 3.45, for the third and fourth quartiles, respectively) as reported ( Supplemental Table 1 ).

This research study showed that BMI fluctuation rate is a stronger predictor than just BMI fluctuation in the development of cardiovascular-related events. This approach is apparently the first analysis specifically focused to evaluate BMI fluctuation rate and cardiovascular outcomes, despite that many investigations have found that body-weight fluctuations may be associated with metabolic disorders and complications (23).

The pathophysiological mechanisms behind this increased risk in CVD in higher BMI fluctuation are under study, and different hypotheses have been proposed (3). Studies on animals and humans revealed the involvement of weight fluctuations in adipose tissue inflammation, adipose hypertrophy, and oxidative stress, which are conditions that stimulate the development of insulin resistance (32–35) and lead to cardiovascular complications (12). This cyclical oscillation could also be associated with abdominal and visceral fat accumulation (36), negatively affecting insulin levels and causing inflammation (37, 38). Therefore, weight cycling and the “yo-yo” pattern lead to insulin resistance, adipocyte inflammation, and metabolism disorders, resulting in increased oxidative stress and endothelial dysfunction, leading to atherosclerosis (39, 40). Epigenetic pathways are in many cases adaptive mechanisms to internal situations (7). It has been widely studied that the expression of candidate genes related to obesity and cardiovascular diseases is associated with DNA methylation and histone acetylation, consequential in the changes in cellular phenotypes and cardiac function (41). Weight “yo-yo” fluctuations may enhance epigenetic adaptation to unbalanced nutrition, altering gene expression and constitution (7). In fact, several epigenetic alterations were discovered in response to weight loss interventions, and genome-wide methylation analysis of adipose tissues after 6 months of endurance training demonstrated changes in DNA methylation of 63 genes involved in obesity (42). These genes participate in metabolic pathways, cytoskeletal organization, cell adhesion, and cell signaling (42). Transcriptomic studies in patients subjected to weight cycling showed the promoted expression of genes involved in the activation of pathways related to the formation of fibrin clot, cardiomyopathy, lipids, and cell surface interaction at the vascular wall (43–46). Weight regain results in rapid adipose tissue growth and hyperplasia, with an increase of cytokines, such as leptin, which acts on the limbic system by stimulating dopamine uptake, creating a feeling of fullness (47). However, these adipokines induce the production of reactive oxygen species as a process of mitochondrial oxidative phosphorylation, engendering oxidative stress (47). The stability of weight often is not achieved, resulting in weight fluctuations and the beginning of a new weight cycling (7). Interestingly, these pathways were only affected after weight regain (43). The central nervous system with homeostatic control centers within the hypothalamus and hindbrain appears to be the key for the maintenance of body weight (7). Cortisol is involved in stress response through the sympathoadrenal system and the hypothalamic–pituitary–adrenocortical axis (48–50). These regulatory mechanisms have been critical for the survival of individuals and species but may become maladaptive when stress is intense or chronic and may bring on rapid weight fluctuations (4, 51, 52). Also, some researchers reported that patients with weight fluctuations had significantly lower general well-being, minor eating self-efficacy, and greater stress than weight maintenance patients, regardless of body weight (53). Additionally, other investigators found that individuals who reported stressful life events had significantly higher odds of having metabolic syndrome (54). On the other hand, it has been shown that patients who regained weight after losing weight, whom they called “relapsers,” had greater levels of serum leptin and insulin than weight maintainers (55). Ultimately, the accumulation of stressful life events has been associated with insulin resistance, obesity, raised triglycerides and oxidized LDL-C, and cardiovascular stress-related morbidities (4–6, 56).

Several epidemiological studies have described the association between weight fluctuations and CVD, especially in patients with T2D or a previous CVD (8, 9, 13–18). However, other cohort studies have failed to confirm these findings or even found opposite outcomes and trends (13, 14, 20–22, 57, 58). A meta-analysis, with 25 studies involving more than 400,000 participants, reported no effect in more than 60% of both total participants and events (18). The inconsistency in some results may be due to several factors, such as the heterogeneity of the measurement of weight fluctuation, baseline BMI, and age. In this context, a strong association between body-weight fluctuations and the risk of cardiovascular events among participants with coronary artery disease has been described (11). However, no data on baseline BMI were reported, which is a limitation, and rapid body-weight fluctuation should be considered in further studies concerning weight “yo-yo” outcomes (23).

Some interesting results arose from the subgroup analyses in our study. Thus, the stronger association using BMI time-rated fluctuations rather than BMI fluctuations and the risk of developing a CVD according to baseline BMI, sex, and age groups was consistent in all subgroups. In addition, we found a weaker association using BMI fluctuation, particularly in women and in the analysis stratified by baseline BMI. The highest quartile of BMI fluctuation was associated with the risk of CVD in all age subgroups, men, and overweight and obese patients. These results are consistent with other previous observations (14, 15, 18). Interestingly, the risk of a CVD was lower in the fourth quartile of the BMI fluctuation rate in obese patients, compared with the HR found for normal-weight and overweight participants. This outcome could be due to the baseline increased risk associated with obesity, the “obesity paradox” reported in previous studies: obese individuals may have developed a CVD because they were obese or because of previous “yo-yo” weight fluctuations (3, 59). Thus, these results should be examined with caution. Another possible explanation is the potential bias due to related weight loss. Weight fluctuations in obese individuals probably would have a major trend toward losing weight and reducing cardiovascular risk. This bias may serve to increase morbidity risk in the non-obese groups, creating the appearance that obesity is protective (59). Moreover, the statistically significant differences between the mean of BMI fluctuation according to baseline BMI, sex, and age subgroups shown in our study suggest that weight fluctuations should be addressed independently or stratified by these groups. However, further studies are warranted to confirm these associations.

BMI was applied to evaluate fluctuation and fluctuation rate according to a better clinical approach and to be able to compare patients with different heights (9). The gain of any kilograms should be in the context of the height of a patient. The use of BMI to examine weight fluctuations may allow us to compare different groups of patients and to open up in other populations or cohorts, despite limitations in some cases (18).

Interestingly, BMI time-rated fluctuation is not just a stronger predictor to develop a CVD, as shown in our study, but it has a clinical and practical message for the management of patients. Therefore, weight maintenance is recommended to minimize weight fluctuations and avoid rapid BMI cycling. Indeed, it appears that the speed or velocity of the fluctuation acts as a primary determinant. A key point for patients is to be aware of losing or gaining weight in short periods of time, and the accumulation of these rapid “yo-yo” weight fluctuations would increase their cardiovascular risk and complications.

The several limitations of this study also warrant consideration. We did not record lifestyle habits such as nutrition and energy intake. Although we have not adjusted for this possible confounding factor, we used other additional variables to adjust, such as BMI or cholesterol levels, which are indirectly related to nutritional habits. Future studies are needed to confirm the specific relationship between other CVD stress-related diseases and T2D with BMI fluctuation rate. The missing values of lifestyle behavior were imputed. Despite that, statistical significance was maintained. Also, the lack of information on the use of lipid-lowering therapy and antidiabetic and weight-loss drugs may have influenced the results. However, if this bias explained the results, then the expected change in estimates would be toward the null. The use of BMI fluctuation rate in a Cox proportional hazard analysis, which addresses the risk over time, may lead to overvaluing of our results. However, we conducted a logistic regression model after 5 years of free disease period for all the participants to exclude the possible time interaction in the Cox model, and consistent results were found. Also, the use of BMI fluctuation rate (|(kg/m²)/year|) could not be intuitive in clinical practice, but our data upraise the idea of the use of weight time-rated fluctuation as a stronger predictor rather than weight fluctuations. Finally, residual confounding by unmeasured factors is always a possibility in observational research, and causal relations cannot be definitely established but can be a consequence. Nevertheless, the main known risk factors for CVD in the adjustment were included, which could act as markers of a general healthier lifestyle.

In conclusion, BMI fluctuation rate (|(kg/m²)/year|) was associated with a significantly increased risk of CVD in this clinical cohort. The magnitude of the risk was stronger than using BMI fluctuations (|kg/m²|). This finding may suggest that the time-rated weight fluctuation is a more important determinant to increase the risk of CVD than the weight fluctuation itself. Overall, clinical interventions for avoiding quick weight cycling are needed to prevent future cardiovascular events related to stress comorbidities and associated complications within precision medicine strategies (28).

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Ethics Committee of the Universidad de Navarra (30/2015). Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.

LS-I contributed to the generation of the database, performed the data analysis, and drafted the manuscript. DN-G contributed to the generation of the database and data research, extracted data, performed the data analysis, and drafted the manuscript. DM-U contributed to the discussion and the design and intellectual revision of the manuscript. JP reviewed the data analysis and contributed to the discussion, data interpretation, and design and intellectual revision of the manuscript. AF-M conceived the plan of analysis, performed the data analysis, and contributed to the design, data analysis and interpretation, and intellectual revision of the manuscript. JM contributed to the design, data analysis and interpretation, and intellectual revision of the manuscript. All authors made substantial contributions to the conception and design and analysis and interpretation of data, drafted or revised the article critically for important intellectual content, and approved the final version to be published.